1. The option that expired

For the past few years, anyone working on getting drugs into the human brain has held a free option. The option was to wait.

Not wait on Arrowhead specifically. Wait on the question. Could a drug injected under the skin cross into the brain and shut off a gene, or was that going to stay a monkey result forever? While the answer was unknown, every company with a brain program could proceed at its own pace, because a result that might have changed the whole field might equally have led nowhere.

On October 14 that option expired, because the question got answered. I wrote most of what follows in the days after, and it has had two weeks to be tested against what the rest of the field did next.

Who was holding that option, what it cost them when it expired, and the part almost nobody said out loud at the time. Some of this is now a matter of record rather than prediction, and I have marked where.

The question was open, so waiting was free. On October 14 the question closed.

2. What Arrowhead actually won

Arrowhead is the first company to show that a drug given as an injection under the skin can sharply reduce the protein output of a gene in the human central nervous system. Healthy volunteers given 12.8 milligrams per kilogram on Day 1 and Day 15 had a mean 73 percent reduction in spinal-fluid tau at Day 105, which is ninety days after the last injection. The measurement is tau protein in spinal fluid, which is the standard readout in this field and an indirect one. The cohorts are eight people each, six on drug and two on placebo, so that 73 percent is the average of six. The interim safety profile was encouraging: most reported events were mild, with no serious treatment-emergent adverse events, no discontinuations attributed to those events, and no persistent anemia or neurobehavioral adverse events.

It matters which part is first. Carrying a drug into the human brain by binding a receptor on the barrier was done before, and done well enough to win an approval. The receptor route had taken an enzyme and antibodies across. What it had never taken was a gene-silencing drug given under the skin, with knockdown to show for it. That is a real first and it is worth exactly what it says.

Worth being blunt about what the result does not show. Spinal fluid tau is an indirect readout, not a measurement of what happened inside brain tissue. The cohorts carried six people on drug. Regional distribution in a human brain, which cell types the receptor actually reaches, and whether the pathological forms of tau fall along with the total are all untested in people. Durability to Day 105 supports the idea of quarterly dosing and does not establish a maintenance regimen, nor what happens after years of repeated injections.

The human data strongly support gene silencing in the central nervous system. They do not yet show clinical benefit. Closing that gap takes more than a biomarker and more than one trial.

3. Everyone else arrives by appointment

The receptor Arrowhead uses to get into the brain is the transferrin receptor, which normally ferries iron across the barrier. Hijacking it is not a new idea, and Arrowhead is not the only company doing it.

Figure 1. Compiled from prescribing information, company disclosures and trial registrations through October 2026. The comparison concerns delivery route and treatment burden, not relative efficacy or direct competition across indications.

Denali has an approved medicine that uses that receptor to carry a replacement enzyme into the brains of children with Hunter syndrome. The FDA granted it accelerated approval in March on the strength of a reduction in a spinal-fluid disease marker, which means confirmation of clinical benefit is still outstanding. It is meaningful human evidence that this receptor can deliver a payload into the brain at a dose that matters, and it arrived before today.

Denali also has a tau oligonucleotide using the same receptor. DNL628 dosed its first patient in March, is given intravenously, and expects human biomarker data in the first half of 2027. Roche has an antibody shuttle on the same receptor in Phase 3 for Alzheimer’s.

Everyone else reaches the brain by the older road, and there are a great many of them. Seventeen programs are in the clinic by lumbar puncture, against three subcutaneous ones that all run on Arrowhead chemistry. Biogen’s tau drug goes into the spinal canal, and so does Novartis’s, which has been recruiting a Phase 3 in progressive supranuclear palsy since June. One of Arrowhead’s own partnered programs goes that way too, SRP-1004 against ataxin-2, because the route is a choice made target by target rather than a doctrine. Alnylam goes there five times. Every clinical program in its central nervous system pipeline arrives by lumbar puncture: a tau siRNA called ALN-5288 dosing Alzheimer’s patients since late 2025, brezosiran in Huntington’s disease with a pivotal study planned for 2027, mivelsiran against the amyloid precursor protein in Phase 2 in two indications, ALN-SOD in ALS dosed twice a year, and ALN-SNCA in early Parkinson’s disease.

That last one is worth pausing on. It is the cleanest comparison in the field.

Alnylam’s ALN-SNCA and Arrowhead’s ARO-SNCA silence the same gene, alpha-synuclein, for the same disease. Same modality, same target, same biology. Alnylam delivers it through a needle in the spine and is dosing Parkinson’s patients now. Arrowhead designed its competing program for delivery under the skin and licensed it to Novartis in late summer 2025 for two hundred million dollars up front, more than a year before its platform had human validation. ARO-SNCA is still preclinical.

Two companies, one gene, two roads. One of those programs has already helped anchor a two-hundred-million-dollar upfront deal. Which road produces the better medicine is still unknown.

That comparison tells you nothing about which drug will work. Alnylam is in patients and Arrowhead is not, and ARO-MAPT is human evidence for the platform rather than for ARO-SNCA. It tells you this instead: the route was a choice, both companies made it deliberately, and only one of them now has human evidence about what that kind of choice is worth.

Denali’s approved drug is an intravenous infusion, once a week, running about four hours each time. That is a morning in a chair with a line in your arm, fifty-two times a year. Its tau program is infused. Roche’s antibody is infused. The two tau programs in the spinal canal need a lumbar puncture each time.

Arrowhead’s is a shot, and the durability data suggests it may be given four times a year. That is a difference in kind rather than in count, and the case against reading it as decisive comes later in this note.

None of that diminishes what Arrowhead reported, and the route is why. The sentence I would defend is that Arrowhead was first to show gene silencing in the human brain from a subcutaneous dose, not that Arrowhead is the only company that can reach the brain. Others got there first. They got there by appointment.

The receptor was proven before October 14. What was not proven is that you can use it with a syringe instead of an infusion chair.

4. The installed base

Treating the human brain by acting on RNA is not new. Three drugs already do it, all approved, and all delivered through a needle between the vertebrae.

What they share is the route, and the route is not a choice. These molecules are large and carry a strong negative charge, and they do not cross from the blood into the brain. Going in through the spine is not the preferred way. It is the only way that worked.

The Spinraza label spells out what that costs the patient. Five milliliters of spinal fluid are withdrawn. The drug is injected into the space around the spinal cord over one to three minutes. Loading doses first, then maintenance every four months, for life. The patients are often children.

The scale is real. Spinraza was approved at the end of 2016, more than fourteen thousand patients have been dosed with it worldwide since, and it brought in 374 million dollars in the first quarter of this year alone.

Fourteen thousand people have taken a drug that works, and the price of taking it is a needle in the spine every four months, for as long as they need it.

Now the part an investor should care about. Spinraza was the established treatment when a daily pill and a one-time gene therapy gave patients other choices, and its sales have come down from their peak since. Not every dollar of that decline belongs to the lumbar puncture, and the argument does not need it to.

The drug did not stop working. The market acquired a choice.

That is what every intrathecal program has to think about now. The question is no longer whether patients will tolerate a needle in the spine. They demonstrably will. The question is what happens when they might not have to.

5. An open line is open for both sides

The chess idea that organizes all of this is simple.

When a file or a diagonal opens in a chess game, it does not open for one player. The pawns come off and the line is simply there, available to whoever has a piece ready to occupy it. Players spend whole middlegames maneuvering to be the one with a rook pointed down the file at the moment it opens, because the opening of the line is not the achievement. Being positioned on it is.

Arrowhead opened a line on October 14. It proved, in people, that the systemic route into the brain carries a gene-silencing payload deep enough and long enough to matter. That proof cost years of work and an enormous amount of money to produce.

Every company holding a receptor-based delivery platform and a nucleic acid payload woke up on October 15 with better information than it had the day before, and none of them paid for it. One important scientific uncertainty, the one that kept boards cautious about those programs, just diminished for all of them at once.

The experiment de-risked a route, not every molecule that might someday travel it.

The proof was expensive to produce. It is free to read. That asymmetry is what to think about.

6. Why Arrowhead is still better placed than that sounds

The open-line problem is real and it is not fatal, because being positioned on a line is different from knowing the line exists.

Several advantages separate it from a company that merely knows the route works. It has the only human data showing gene silencing in the brain from this route, and data is what moves regulators and partners. It manufactures its own material, which matters in a field where chemistry, process control and reproducibility decide whether a promising construct ever becomes a scalable medicine. A shot instead of an infusion is a difference patients feel and payers eventually price. It also has two rare tauopathies already in regulatory conversations, which gives it a development path beyond Alzheimer’s, though Novartis is further along clinically in one of them.

A large partner had also committed to this platform in the brain before anyone could be sure it worked. Novartis signed a deal in late August of last year carrying a two hundred million dollar upfront payment for the alpha-synuclein program and further collaboration rights, more than a year before the human validation arrived.

The deeper advantage is less visible. A delivery platform is not one invention. It is a receptor binder, a linker, a payload chemistry, a dosing schedule, and a manufacturing process that can make it at scale without ruining it.

The binder choice alone helps decide what the product can become. Arrowhead built its construct around a receptor-binding antibody fragment rather than a whole antibody or a fusion protein. That architecture may help with a subcutaneous route, though size alone does not explain it: payload, formulation, dose and tolerability all bear on how a drug can be given. The point is narrower and more useful. Arrowhead did not find a different way to administer an existing drug. It engineered a construct meant to make systemic dosing practical.

Knowing the route works tells a competitor little of that. It tells them the destination exists.

The result is free to read and the patents are free to read too. Arrowhead has an issued patent covering specified receptor-binding antibodies and their conjugates, including the linker chemistry, plus pending applications reaching for the fragment format and the cap that stabilizes the construct.

Reading them gives nobody the right to practice them. That is what a claim is for. It is also not a wall: issued claims protect particular constructs, pending ones protect little until they issue, and a competitor can build outside them. What the estate does not do is close the route. Denali, Roche and others have their own binders and have had them for years. The protection sits around Arrowhead’s construct, not around the idea of going in this way.

Arrowhead now has clinical proof of delivery into a fifth organ system, after the liver, the lung, skeletal muscle and fat, and says it expects to reach a new cell type every eighteen to twenty-four months. On the call Chris Anzalone described the brain platform as feeling like a company’s worth of potential value on its own. That is management’s assessment rather than an independent valuation, and it captures the scale of what a buyer would be evaluating.

The lead Arrowhead holds today is technological. Whether it becomes clinical, regulatory and commercial is the work ahead.

7. Arrowhead was holding the option too

Here is what the coverage missed.

Arrowhead said publicly that if early ARO-MAPT data were encouraging, it would substantially expand its central nervous system pipeline beginning at the end of this year. That was sensible to say when the data might not arrive. That morning it became a plan with numbers on it: two clinical trial applications by year end or early next, at least three new wholly owned brain candidates into the clinic within twelve months, and another partnered candidate behind them in 2027.

Expanding a CNS pipeline means naming targets, running toxicology, building supply, and spending money, in the same window the company is commercializing Redemplo for familial chylomicronemia syndrome and preparing a supplemental application to take the same drug into severe hypertriglyceridemia.

The scientists answered their question. The finance department gets the next one.

That cuts the other way too. Every program Arrowhead adds costs money. Every potential buyer inherits a decision about whether that program deserves more.

Arrowhead is not starting from an empty treasury. At the end of June it reported roughly 1.6 billion dollars across cash, restricted cash and available-for-sale securities. The question is not how to finance the first few brain programs. It is which ones deserve the capital, how aggressively to advance them, and which ones Arrowhead should fund itself rather than partner.

That is what success looks like in this business, and I am not complaining about it. I mention it because a reader who treats the readout as pure upside will be surprised by the next earnings call. The clock that started on October 14 started for everyone, and Arrowhead is standing on the same board.

The good news and the obligation arrived in the same press release.

8. What waiting costs now

Large pharmaceutical companies can still wait. No part of a healthy-volunteer readout forces a purchase, and it puts no clock on a corporate development team.

What changed is not whether they can wait. It is what waiting costs.

Take the central nervous system expansion Arrowhead has said it intends to pursue on encouraging data. Its patent filings describe a world of possible brain targets extending well past tau, across diseases where a protein needs lowering and nobody has been able to reach it. Those disclosures are not a funded development plan. They show how broadly the company is thinking, and the decisions ahead are about which of them deserve real money.

Figure 2. Compiled from Arrowhead’s October 14 presentation and webinar remarks, and from Denali’s 2026 disclosures.

Three new wholly owned brain programs are meant to enter the clinic inside a year. Two applications go in by year end or early next. Regulators are already in conversation about frontotemporal dementia and progressive supranuclear palsy, with a further one to two quarters of back and forth on trial design and studies aiming to begin in 2027. Patient data arrives that year too.

A buyer looking at Arrowhead a year from now is looking at a different company. More programs in the clinic. More capital committed. Some risk retired, some newly exposed. Whether that company is worth more or less turns on what gets proved along the way. Every favorable result retires another piece of scientific risk, and a buyer who waits for that evidence may find itself negotiating over a more valuable asset. The next result could also disappoint. That is the bargain: certainty has a value, and so does the chance to buy before certainty arrives.

The best disclosed reference point is the Novartis agreement. In the late summer of last year it committed two hundred million dollars up front for exclusive rights to Arrowhead’s alpha-synuclein program and the right to select further targets, against as much as two billion in milestones. That is not a clean price for one drug, and it should not be read as one. What it tells you is that a large pharmaceutical company was willing to commit serious capital before the central question had an answer.

That question has now been answered. Whether the next agreement commands better economics is unknown. What is not unknown is that Arrowhead walks into those conversations holding evidence it did not have when Novartis signed.

Waiting buys certainty and may cost opportunity. Moving early buys opportunity and carries the risk patient data will eventually expose. Nobody gets both the early price and the late evidence.

It is worth saying plainly when waiting stays the right call. Patient data is still ahead and could disappoint. Competing architectures are advancing on their own timelines, and a buyer that waits may find a second supplier rather than one. Arrowhead’s own spending raises what the asset costs. A large company that sits still through 2027 is not being slow. It is pricing an option that still has time on it.

There is a quieter consequence to waiting. Every program Arrowhead licenses changes what a future buyer can acquire. Exclusive development rights leave the company, while cash, milestones and royalty interests stay. Whether any given deal raises or lowers what the company is worth depends on its terms, and a buyer prices the retained economics rather than pretending they are gone. Novartis has already selected additional targets under its Arrowhead collaboration, so the portfolio available to a future buyer is changing before the next major readout.

Against that, these companies have shown how they would rather handle it. In the past two weeks Novartis committed 575 million dollars up front to license from a Chinese biotech, and Roche put up to a hundred million into an obesity program from Suzhou. Those are substantial commitments. Neither is an acquisition. Both show that a large company can buy exposure to promising science without buying the company that produced it, and the readout does not take that option away.

What may disappear is the chance to negotiate on the terms that existed before human validation. The next agreement will turn on the data, on who else is interested, and on which rights Arrowhead still controls. Nobody knows that price yet.

The option to wait still exists. Nobody gets it for free anymore.

That was the argument when this was written. Fifteen days later the largest company in RNAi made it for me, without being asked to.

On its third-quarter call, Alnylam was asked directly whether it would pursue a receptor-based approach for the central nervous system. The answer defended the spinal canal on terms worth reading closely: the company does not see intrathecal delivery as a barrier in this disease, with the infrequent administration it anticipates. Then it said that work is underway internally to find a brain shuttle that can deliver peripherally, and that an update would come in the not so distant future.

Put the two halves side by side. The defense is specific to Huntington’s disease, where patients face a fatal diagnosis and no approved disease-modifying treatment exists. The shuttle work is a platform investment, and a platform need not be limited to one indication.

A company that believes the spinal canal is the destination does not fund a way around it.

Alnylam did not say when that work began, and there is no basis for attributing it to October 14. The program may have been running for years. The two positions are not in conflict either. They are what sober portfolio management looks like when the route you have works today and another route may change the economics later. Arrowhead did not necessarily start Alnylam’s clock. It made the clock visible.

9. Who cannot wait at all

The buyers have a price problem. The programs have a different one, and theirs is harder to solve with money.

Novartis sits on both sides of the argument, which is the rarest position in this field. It holds exclusive rights to Arrowhead’s alpha-synuclein program and further collaboration targets, all built on subcutaneous delivery. It is also running its own tau silencer into the spinal canal, in a Phase 3 that has been enrolling since June, targeting three hundred patients, with primary completion expected in 2029. That is the same disease Arrowhead just told regulators it wants to enter, with a drug that would not require a lumbar puncture.

Different drugs, different proteins, but competing answers to the same question about how RNA medicine gets into the brain. Novartis is paying for both. That is an interesting place to be standing when the evidence starts to move.

The people who genuinely cannot wait are not the buyers. They are the programs that now have to explain themselves against a potentially less burdensome alternative.

October 14 did not make the existing drugs worse. It made the alternatives harder to ignore. Every future conversation about an intrathecal tau drug, with a regulator, a payer, a trial site or a patient choosing between two studies, now contains a question that did not exist on Monday. Asked about this directly, Boxer said the lumbar puncture is already a barrier to enrolling patients in trials, and sometimes a barrier to a patient accepting a drug after it is approved. For inherited tauopathy families scattered across countries with thinner health systems, he said a spinal route may simply put the treatment out of reach. There is a version of this that is a shot. That question does not go away, and it gets louder with every data release from the other side.

The most interesting clinical question of the day came from Adam Boxer, the UCSF neurologist Arrowhead invited to present. Boxer knows this field from the inside, including research on Novartis’s own intrathecal tau program and disclosed paid consulting for Novartis. His question was not whether tau can be lowered. It was about what might be limiting the benefit when it is.

Boxer pointed to neurological events reported during treatment with Biogen’s drug, including confusion and falls, most often in the week following an injection, and raised the possibility that they offset some of the benefit. What caused them was left open. The procedure itself, through inflammation or otherwise? The drug? The biology of removing tau? Some combination?

If the spinal route contributes materially to those events, it may be more than a burden on patients. It may limit the benefit a drug can safely deliver.

The two readings point in opposite directions. If the route contributed materially, a subcutaneous drug offers more than convenience. It may deliver the same biology without part of the tolerability cost. If the trouble came from lowering tau itself, Arrowhead meets the same ceiling, and changing the needle does not change the target.

Boxer said plainly that the data cannot separate those explanations. Neither can I. These drugs differ in far more than the needle, in chemistry, exposure, dose and the patients who got them, so comparing separate trials cannot isolate the route. What has changed is that the question can now be put to two very different ways of reaching the same target.

Two slides later Arrowhead’s chief medical officer reported no neurobehavioral adverse events, using the same word Boxer had used. Healthy volunteers over a few months is not the same test as patients over eighteen, and nobody should pretend otherwise.

Ionis shows this better than anybody. The company has built the deepest neurology franchise in this modality, and every medicine in it arrives through a needle between the vertebrae. All three approved drugs discussed above originated with Ionis. Zanvastro, their zilganersen for Alexander disease, was approved in September 2026 and became the first neurology medicine Ionis launched on its own. Obudanersen for Angelman syndrome and ulefnersen for FUS-ALS are both in Phase 3. The tau drug with Biogen is theirs as well.

Three approved medicines and fourteen investigational programs, every one of them into the spinal canal. Ionis is not standing still about it, though, and this is where the story turns.

Months before October 14, Ionis had already disclosed preclinical work on systemic delivery, including a collaboration with ABL Bio on a brain shuttle that binds the insulin-like growth factor 1 receptor rather than the transferrin receptor Arrowhead uses. Its own researchers also posted preclinical work on transferrin-receptor shuttles carrying both siRNA and antisense across the barrier after intravenous and subcutaneous dosing in animals. In monkeys, a tau-targeted construct of theirs reached better than seventy percent knockdown of MAPT across several brain regions after repeated intravenous dosing. That work is a preprint rather than a peer-reviewed paper, and none of it is human.

None of that is human validation, and none of it is a subcutaneous product. It establishes a different point. Arrowhead did not prove to the rest of the field that a second road was worth looking for. Some of them were already well down it.

A franchise built on one road is a strength. Building the second one before somebody else occupies it is how a company keeps that strength.

Which puts one line in their own annual filing in an interesting light.

Their own annual filing says they are not aware of any other medicine in clinical development for Alexander disease. That is very likely correct. Clinical competition is a snapshot, though, and the systemic delivery work now running across the industry shows how fast the picture outside that snapshot can move. The absence of a competing drug today says little about the delivery technology that might support one in a few years.

The route also keeps producing medicines. In September, Ionis and Otsuka reported positive Phase 3 results for ulefnersen in FUS-ALS, hitting a primary endpoint that combined functional impairment and survival. The spinal canal did not prevent that. Whatever systemic delivery eventually offers, convenience alone does not displace a drug that demonstrably helps people.

I should say what kind of pressure this is and is not. No trial has to be redesigned on this account, and a healthy-volunteer result does not establish a new regulatory comparator or standard of care. The pressure is on the next conversation rather than the current protocol.

The buyers can wait. The programs cannot. Sometimes both sit inside the same company, and the clock runs at different speeds in different rooms.

The strongest answer from the other side came nine days later, and it should be stated at full strength. Alnylam reported that brezosiran, delivered by lumbar puncture, lowered mutant huntingtin in spinal fluid by 60 to 64 percent at three months, with a maximum individual reduction of 83 percent, durability supporting twice-yearly dosing, and none of the three findings that ended an earlier huntingtin program. No rise in neurofilament, no inflammatory signal, no ventricular enlargement. A pivotal study follows in 2027.

In patients. For a fatal disease with no approved treatment. That is more clinically relevant evidence, because the knockdown happened in patients rather than in healthy volunteers. It is not yet evidence of clinical benefit, and it is not a basis for ranking the two drugs.

Convenience is an argument about who will take a medicine. It is not an argument about whether the medicine works.

10. The assumption inside the price

Every company running a gene-silencing drug into the spinal canal now holds an asset whose value rests on an assumption. The assumption is that the route does not matter, or that the systemic alternative will not reproduce in patients. That is a perfectly respectable position. It is also now a testable one, and it has a date attached, because Arrowhead’s patient cohorts report in 2027.

An investor holding those companies is pricing that assumption whether or not anyone has stated it. It pays to ask what has to stay true for each of these routes to hold its value.

Three assumptions hold the intrathecal position up. Each one is now testable. How much of that risk the market has already priced is a separate question.

Novartis is where the gap shows most plainly. Its Phase 3 runs to a primary completion in 2029 while it holds a license to the other route, and the assumption inside that is specific: that the first program keeps its value even if the second answer wins.

It may well. The trial may read out positive, the drug may help people, and the burden may prove irrelevant in a disease that kills within a decade. Those are the assumptions, and they are reasonable. They are also what an investor should be tracking rather than taking on faith.

The case overreaches without the other side of it, so here it is. No company owes the market an opinion on a competitor’s healthy-volunteer data. Forty-one volunteers is not a patient population, a biomarker is not an outcome, and Biogen has clinical signal in Alzheimer’s patients where Arrowhead has spinal-fluid tau in healthy people. A board that reads October 14 and concludes the route is a detail is not being dishonest. It is making a judgment.

The three, plainly. That systemic delivery will not reproduce in patients. That a lumbar puncture will not materially constrain adoption where an alternative exists. That efficacy and safety advantages will be large enough to outweigh the procedure if it does. The first begins to face a human test in 2027. The others take longer, and by the time they are answered a great deal of capital will already be committed.

The reason this matters is timing. Programs on a losing route do not get discontinued on the day the route loses. They get discontinued years later, after more capital and more patients enrolled, and the investors who exit worst are the ones who never knew which assumption they were holding.

An assumption with a date on it is not a risk until the date arrives. It is a risk the whole time.

11. What would make me wrong

The proof is narrower than the argument. Healthy volunteers are not patients, one target is not a platform, and the headline average rests on six people. The argument above assumes the route generalizes, and that has not been shown.

The competitor is closer than people think. Denali is also advancing DNL422, a systemically delivered oligonucleotide against alpha-synuclein, in IND-enabling work, so four companies now pursue that target across four delivery strategies, counting Ionis’s ION464 in multiple system atrophy. Ionis has disclosed preclinical work on systemic oligonucleotide delivery with ABL Bio using an IGF1R-targeted shuttle, and separate preclinical work of its own on transferrin-receptor shuttles.

The narrower version of that worry is the likelier one. DNL628 is expected to report human biomarker data in the first half of 2027, while Arrowhead expects patient data during that year. The order and strength of those two results could do a lot to shape how investors judge the platforms. If it comes in competitive on knockdown and tolerability, Arrowhead’s lead gets harder to define by the biomarker alone. Route, duration, brain distribution and the cost of giving the drug all move toward the middle of the comparison. A similar number in spinal fluid does not mean two drugs behave alike in the brain, and it does not mean patients experience them alike.

The open line may also be narrower than I think. Knowing a route works is a long way from working it. If Arrowhead’s specific chemistry, linker and manufacturing turn out to be the hard part, the open-line worry in Section 5 matters much less than I have made it sound.

The price argument can fail. Large companies have waited through stronger signals than this one, and no part of this forces a decision. A company spending hard on expansion can also get cheaper rather than dearer, if the market decides the spending is undisciplined. Appreciation is not urgency, and buyers have watched assets appreciate and sat still before.

More tau lowering may not be required, and nobody knows how much tau a person can spare. Boxer described early results for the Novartis drug in PSP where a far more modest reduction in spinal fluid tau came with encouraging neurofilament findings. Those are patients rather than volunteers, at different doses and times, in small numbers, so the percentages do not line up. The question does. Tau does real work in a neuron and ARO-MAPT lowers all six forms of it. Boxer called the therapeutic window the key question in the field right now, and said plainly that the Biogen results are producing questions nobody can answer yet.

Convenience is not the same as dosing frequency. Biogen studied its drug at intervals as long as twenty-four weeks. A less invasive procedure four times a year is not obviously better than a more invasive one twice.

A shot is not automatically a simple product. The dose that validated the route was substantial, and commercial development still has to solve formulation concentration, injection volume, manufacturing cost and the minimum dose that does the job. A quarterly injection that takes several syringes or carries a heavy cost of goods gives back part of the advantage the route appears to offer. The 6.4 mg/kg cohort reached a 69 percent mean reduction against 73 percent at 12.8, on different schedules, so the figures are not a controlled comparison. They are a hint that the top dose may not be the product.

The question is not how much tau Arrowhead can lower at the highest dose tested. It is how little drug it takes to matter.

The receptor may reach less of the brain than the regional data suggests. Bret Bostwick, who co-invented CNS-targeted RNA therapeutics at Alnylam, has said publicly that transferrin receptor shuttles will solve delivery for only a subset of the brain, because some cells worth targeting do not carry the receptor. Twelve regions is not every cell type inside them. If he is right, the route opens part of the brain rather than the whole of it, and the companies now looking at CD98, LRP1 and IGF1R are looking for the right reason.

The target is still unproven, though not untested. Biogen’s tau-lowering drug reduced tau on brain scans in patients and produced encouraging cognitive signals, and still missed its primary endpoint. That is more human evidence than Arrowhead has on whether lowering tau helps, and it remains unsettled. Tau genetics links strongly to the primary tauopathies and much less to Alzheimer’s, which is the reverse of where the money has gone. Failure in Alzheimer’s alone would not settle the question for MAPT-related frontotemporal dementia or progressive supranuclear palsy. If lowering tau fails across those too, this drug loses its rationale. The delivery system could still hold real value against other targets.

12. The file is open

A line opens once. After that the position is simply different, and every player on the board adjusts to a world where that line exists.

Arrowhead did the opening. It put in the years, took the risk, and produced a result that told an entire field what it did not know the day before. That deserves more credit than it will get.

One reader saw it immediately. Within an hour of the webinar, Bret Bostwick, who had co-invented CNS-targeted RNA therapeutics at Alnylam, called the durability profile best in class, named biodistribution rather than convenience as the true differentiator, and wrote that subcutaneous blood-brain-barrier-penetrating siRNA was officially validated in the clinic. A transformative milestone, he said, for the entire neurodegeneration field. He reached that judgment in sixty minutes.

One of the field’s best-placed readers called it transformative. The stock closed up 0.41 percent and the week moved on.

In the days that followed, the stock even traded below where it had closed before the data. It has since recovered. Neither movement settles the scientific argument. It is a reminder that proving a thesis and being paid for proving it are separate events, and that the second one runs on its own schedule.

The file does not belong to the player who opened it, however. It belongs to whoever has a rook on it.

Arrowhead has the only human data on this route, its own oligonucleotide manufacturing, and two rare diseases already in front of regulators. That is a reasonable claim to the file. It is not a permanent one.

The next twelve months come down to one question for Arrowhead: whether it occupies the line it opened, or simply shows everyone where it was. That is the smaller question, though, and it belongs to one company.

The larger one belongs to everybody else. A field spent thirty years treating the spinal canal as the price of reaching the brain. On October 14 that stopped being the only price. Nobody has to move. Everybody now has to explain why they are not moving, and the first company asked that question in public answered it by saying a brain shuttle is under research and an update is coming.

The file is open. Everyone can see it. The arguing from here is about who gets to stand on it.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. The characterization of what is and is not in the published literature, the probability judgments, and the chess analogy are the author’s own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The problem nobody solved for thirty years

siRNA has a delivery problem, and that problem is the whole problem. The molecule works. Put it inside the right cell and it will silence the gene you aimed at, reliably, for months. Getting it inside the right cell is where the field has spent its entire existence.

Chris Anzalone described the sequence in a BioCentury interview this month, and it is worth laying out, because the public conversation treats delivery as a single step.

Figure 1. The three problems that have to be solved together before any new cell type opens to siRNA.

Three problems, in order. You need a receptor that sits on the cell you want, that internalizes, and that recycles fast enough to carry repeated doses. Then you need a ligand that binds it and survives being conjugated to an oligonucleotide. Then, once inside, the payload has to leave the endosome and reach the silencing machinery, because a cell is not a bag of water and concentration in a tissue is not activity in a cell.

Solve the first two and fail the third, and you have a drug that arrives everywhere and silences none of it.

The liver came first because the biology cooperated. Hepatocytes already carried a high-capacity receptor for clearing circulating glycoproteins, one that internalized and recycled efficiently. A sugar called GalNAc gave the field a way to use machinery nature had already installed. Even with that head start the field spent the better part of a decade getting from idea to approved medicine.

The liver is opening theory. The moves are published, the position at the end is known, and anybody can play it. Past the liver the book runs out.

Every organ after the liver has been found rather than followed, and the brain is the hardest of them.

That is what made October 14 worth attention. Arrowhead reported that ARO-MAPT, given under the skin on Days 1 and 15, lowered spinal fluid total tau by a mean of 73 percent at Day 105, ninety days after the second dose, with a maximum individual reduction of 85 percent. No spinal injection to deliver the drug, and no infusion into the central nervous system. Lumbar punctures were still needed to draw the samples.

That is not proof that Alzheimer’s can be treated. It is narrower and prior to it: evidence that a subcutaneously delivered siRNA can engage a target in the human central nervous system. A working delivery system is not a working medicine. Without the first there is no second.

2. What Alnylam said when somebody asked

At a roundtable this morning on Brezosiran, its intrathecally administered siRNA for Huntington’s disease (also known as ALN-HTT02), Alnylam was asked directly whether dosing into the spinal canal is a commercial problem and what the company is doing about brain shuttles.

Christi Kelsey, Senior Vice President, Value & Pipeline Commercialization at Alnylam, took the first half and gave a good answer. In Huntington’s, efficacy drives treatment decisions rather than convenience, and specialists have indicated that dosing into the spinal canal twice a year is acceptable. In a fatal disease with no approved disease-modifying treatment, she is right.

Toby Ferguson, Senior Vice President, Neuroscience Therapeutic Area Head at Alnylam, took the shuttle question. He confirmed that Alnylam is working on a number of brain shuttle platforms across its CNS portfolio, said the company would talk about those in the future when it has data to share, and framed the choice between a shuttle and an intrathecal approach as a trade-off to be weighed target by target. Then he said the present focus is establishing proof of concept with brezosiran, after which Alnylam would be well positioned to optimize from there.

Plural platforms. No publicly disclosed shuttle results. No announced timeline. A company that would very much like to tell you about this if it had a result to show.

Read that against who is saying it. Alnylam has worked on CNS delivery for years and produced the C16 conjugate, which is a real piece of chemistry and the reason brezosiran reaches the brain at all. They have the people and the money. If a subcutaneous route into the brain were solved, or nearly solved, I would expect them to be presenting it rather than deferring. That is an inference about disclosure behavior rather than a fact about their laboratory.

Alnylam is not arguing that the spinal canal is the right route for every disease of the brain. It is arguing that intrathecal dosing is workable where patients will tolerate a great deal, and building a different route for the rest.

3. What each route has actually been shown to reach

Claims about brain delivery are simple to make and hard to check. I published the map below separately last week on X; it collects what has been measured, region by region.

Figure 2. Regional evidence map. Tau pathology by disease against measured knockdown by route, compiled from company presentations and published releases.

One caveat before the numbers. The regional knockdown percentages all come from monkey studies, and those studies differ in target, formulation, dose and measurement method. The pathology columns describe human disease. These are not interchangeable datasets. The human ARO-MAPT result is spinal fluid tau alone, so regional uniformity in a person remains an inference from the animal work rather than an observation.

Two findings matter. Arrowhead’s subcutaneous construct reached all twelve regions it was tested in, between roughly 63 and 83 percent on charts presented at TIDES, RNA Leaders and the CNS Delivery Summit, with no depth-dependent decline across the regions measured. Its intrathecal comparator reached 86 percent in thoracic spinal cord, 68 percent in temporal cortex, 23 percent in putamen and no measurable knockdown in caudate.

That second result is a comparison between two Arrowhead formulations, not a verdict on intrathecal delivery as such, and the distinction matters. Alnylam’s own intrathecal chemistry does reach the striatum. Roughly half of huntingtin protein cleared from caudate and putamen in monkeys three months after a single dose, with cortex above eighty percent. The route is not incapable. It is uneven, and the unevenness runs the wrong way for diseases that live in deep structures.

One limitation applies to Arrowhead as much as to anybody. Spinal fluid is an imperfect guide to what is happening inside brain tissue. The animal data suggest the regional advantage. The human data establish that the delivery system is pharmacologically active. How completely those two overlap is the open question, and it is the same question every company in this paper has to answer.

4. The honest case for the spinal canal

Intrathecal delivery bypasses the blood-brain barrier by putting the drug straight into cerebrospinal fluid. It still has to penetrate brain tissue from there, and distribution is not uniform, but it avoids the problem of carrying a payload across the barrier from the bloodstream. Lower doses, less systemic exposure, and no requirement that the drug survive a trip through the circulation. Alnylam reports no dose-dependent rise in spinal fluid neurofilament and no ventricular enlargement, which were among the safety concerns that troubled earlier huntingtin programs, and up to sixty percent mean reduction of mutant protein in its Phase 1b.

That matters more than it sounds, because Alnylam’s own framing of the field’s history is that exposure-related tolerability prevented anyone from exploring deep lowering. Earlier programs ran into rising neurofilament as doses climbed. That is one ceiling earlier programs ran into, and Alnylam’s argument is that its chemistry has raised it.

Systemic delivery carries a cost that intrathecal delivery does not. A drug circulating through the body can silence its target in tissues no one aimed at. Toby Ferguson named this directly as part of the trade-off and he was right to. Arrowhead’s answer so far is healthy-volunteer safety data showing no serious treatment-emergent events and no persistent anemia. Encouraging, and not evidence of long-term safety under chronic dosing.

The strongest argument for the spinal canal is not Alnylam’s, though. It is Biogen’s.

Diranersen is an intrathecal antisense drug against the same target Arrowhead is chasing, and in July it reported Phase 2 results in early Alzheimer’s. Spinal fluid tau down 50 to 65 percent. Tau on brain scans down in every region evaluated. Twenty-six percent slowing on the clinical dementia scale at the 60 milligram dose, with larger effects on the cognitive measures.

The trial missed its primary endpoint, which was dose response, and showed no separation on daily function. That miss is worth sitting with. Higher doses did not produce greater slowing, even though tau fell substantially across the dose groups. Perhaps there is a plateau. Perhaps distribution and exposure matter more than the percentage suppressed. Perhaps the clinical differences are less reliable than they look, coming as they do from secondary analyses after a failed primary. Nobody knows. What the result does establish is that deeper biomarker lowering cannot be assumed to carry greater clinical value, and 73 percent should not be read as automatically better than 60. The efficacy question is open, and Biogen intends to advance into registrational development regardless. Set the two side by side, though, and the comparison is uncomfortable. A drug delivered by lumbar puncture has produced clinical signal in patients. Arrowhead has produced a biomarker in healthy volunteers.

Convenience is an argument about who will take a medicine. It is not an argument about whether the medicine works.

Twice a year into the spinal canal is tolerable in a fatal disease. It is a different proposition in a disease with more than seven million patients in this country alone and a community neurologists managing most of them.

5. The counterexample worth taking seriously

Receptor-mediated transport into the human brain is no longer theoretical. Denali has an approved medicine that crosses using the transferrin receptor. The approval was accelerated and rests on a spinal fluid biomarker, with confirmatory evidence still being collected, which is worth noting before anybody treats it as settled.

What that validates is the crossing, and more than the crossing. The enzyme is taken into cells and trafficked to the lysosome, where it does its work. That is a real intracellular destination. An siRNA needs a different one: it has to escape the endosome and reach the silencing machinery in the cytoplasm. That is the third problem on the list above, and it is where most of this difficulty lives. Denali proved a door opens. It did not prove an siRNA can get from the doorway to the room it needs to reach.

6. Where everyone actually stands

Five positions, and not one belongs to a company that has decided the problem is solved.

Arrowhead. ARO-MAPT is the wholly owned brain program in the clinic. Two more are there through Sarepta, one intrathecal against ataxin-2 and one subcutaneous against huntingtin. The company has said it intends to put at least four more CNS candidates into the clinic by the end of 2027. It also holds the only clinical evidence yet produced that a shot under the skin can lower a target in the human central nervous system.

Alnylam. Brezosiran in the clinic by the spinal canal, real chemistry behind it, and brain shuttle platforms in the laboratory with no publicly disclosed data and no date attached.

Novartis. NIO752, an intrathecal tau drug, is in Phase 3 for progressive supranuclear palsy, a primary tauopathy rather than Alzheimer’s. Separately, Novartis paid Arrowhead to license a subcutaneous program against alpha-synuclein. One company, both routes.

Denali. An approved medicine shows receptor-mediated delivery into the brain works for an enzyme payload. DNL628 now tests whether the same transport concept can carry a tau-lowering oligonucleotide. It is given intravenously rather than under the skin, so the burden is an infusion rather than a shot, with human biomarker data expected in the first half of 2027.

Biogen. Intrathecal tau lowering demonstrated in patients, regional reductions on tau imaging, and clinical signals that are encouraging and not conclusive. A failed primary endpoint and a stated intention to go registrational regardless. The most advanced direct comparator anybody has.

Four of the five are building or buying a way past the spinal canal. Biogen is the exception, and Biogen is the one with clinical signal. One company has shown the subcutaneous route working in a person.

7. What would change my mind

The obvious one. Alnylam presents shuttle data. They have the chemistry group and the capital, and the day they show a subcutaneous construct reaching a primate brain, the technical part of the gap starts closing. It would not close the rest of it. The distance between discovering a delivery system and developing a medicine with it is measured in clinical studies rather than laboratory announcements.

The quieter one. Arrowhead’s patient cohorts read out in 2027 and the knockdown comes in materially below what healthy volunteers showed. A diseased brain may differ in barrier integrity and in distribution, and Arrowhead has not yet reported whether the same depth and durability reproduce in patients.

The longer-term one. ARO-MAPT turns out to be the exception rather than the rule. The chemistry works for tau and proves hard to reproduce against unrelated brain targets. Endosomal escape is not a property of the delivery vehicle alone; payload, sequence and conjugate chemistry all bear on it, so success against one target does not automatically carry to the next. That would leave Arrowhead with a promising drug rather than the reusable platform the argument here depends on, and the next distinct CNS target to reach human knockdown is the proof point.

The one that matters most. Arrowhead reproduces deep tau lowering in patients and the clinical benefit proves no better than what an intrathecal competitor achieves. Delivery can be a competitive advantage without being a clinical one, and the market would then have to decide what the easier route is worth on its own.

The nearest one. Denali's DNL628 produces deep, durable tau lowering in people after intravenous dosing. That would not erase a lead in subcutaneous siRNA, and it would show that a second delivery platform reaches the same destination, which is a different kind of problem than a competitor catching up on the same approach.

8. The part that is actually hard

Two of the largest pharmaceutical companies in the world ran RNAi delivery programs and sold them after deciding not to continue the work internally. Arrowhead bought both. That is a fact about the difficulty of this problem more than about anyone’s foresight, and I have written about the strategic reading of it in my other papers.

What the last month has added is a view of where everyone else is standing.

The field is not arguing about whether to reach the brain without a spinal needle. It is arguing about who gets there first.

Alnylam has intrathecal data and shuttle platforms it has not yet shown. Novartis is paying for both routes. Denali has crossed the barrier with a different payload and is now trying it intravenously with an oligonucleotide. Biogen has a spinal-dosed drug with promising clinical evidence.

Arrowhead has what none of them has publicly demonstrated with subcutaneous siRNA: substantial target engagement in the human central nervous system.

That result is not a medicine. It is a key that turned once, in a lock no one else has opened.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper.

The three-step delivery description is as given by Chris Anzalone in a BioCentury interview of October 16, 2026. Alnylam’s remarks and the tolerability framing are from its brezosiran roundtable of October 2026, where the slide on technical challenges in the field refers to tominersen and WVE-003 rather than to brezosiran; the brain shuttle comments are Toby Ferguson’s and the intrathecal commercial remarks are Christi Kelsey’s. Brezosiran primate and Phase 1b figures are from that same presentation. Arrowhead regional knockdown figures are from its TIDES USA 2025, RNA Leaders USA 2025 and CNS Delivery Summit presentations and the ARO-MAPT release of October 14, 2026, and are read from published charts rather than tabulated values. Denali’s approved medicine uses transferrin-receptor-mediated transport of an enzyme and is not an oligonucleotide. DNL628 is an antisense oligonucleotide on an engineered Fc fragment, given intravenously, with first patient dosed in March 2026. The regional map was published separately by the author.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. A gate and a check are different events

In June and July I published four papers arguing that Novartis, Lilly and Roche might each be pushed into buying Arrowhead. The companion note put the mechanism in a sentence: they would be pushed not because they wanted to, but because the cost of letting a rival buy it first can be worse than the cost of buying it themselves.

That mechanism needs two separate events, and I have been careless about keeping them apart.

The first is a gate. A late-2026 readout had to show human pharmacodynamic activity consistent with the delivery platform reaching the brain. The companion note was explicit that a clear failure would knock out the most valuable part of the platform and deflate the urgency. The gate was a survival test, not a starting gun.

The second is a check, in the chess sense. Part Two introduced Lilly as the player who can simply outspend the others, and described its entry as the move that puts the whole board in check. The cascade begins when somebody moves. Until then the others are watching each other.

On Wednesday the gate opened. Mean reduction of 73 percent in spinal-fluid tau, measured ninety days after the last injection, from a shot under the skin. Six treated volunteers in that cohort, dosed twice where the lower cohorts were dosed three times.

Nobody has played the check.

More precisely: nobody has played a check the market can see. Whether somebody has made a private approach is not information I have, for reasons the sixth section gets into. The first check could also be a credible proposal rather than a signed agreement, and the forecast later uses the narrower, publicly observable standard.

A first bid would start the test of the game theory rather than prove it. The other two could bid, license around the new owner, build alternatives, or do nothing at all. The original papers asserted that losing Arrowhead would be too expensive to tolerate. The first bid is when that assertion finally becomes observable.

The gate was a test the thesis had to survive. The check is a move somebody has yet to make.

Figure 1. The Forced Move scientific gate, Arrowhead’s later clinical benchmark, and the results announced October 14, 2026.

The distinction decides what a quiet week means, though it must not become a way of never being wrong. If the first bid is what starts the cascade, a quiet week cannot show whether the cascade works. It does say how willing the potential first movers have been to begin it. The science cleared its gate. The acquisition mechanism is untested, and the one timing argument those papers did make, that a well-placed buyer would rather act before a positive readout, has not played out as I expected.

The science itself I covered in The Bishop Moved, and the market’s failure to reprice it upward in The Liver Got Paid. This note is about the board.

2. Where I was wrong

Three corrections first.

The bar was never mine. The 50 to 60 percent figure everyone quoted was Arrowhead’s, not mine. Hamilton named it on the August 4 earnings call, ten weeks before the readout, and it happens to match the range diranersen had reached. The result came in at 73, with clear dose-related activity, though the top two groups sit close together and were dosed on different schedules.

I missed a piece on the board. Novartis registered a Phase 3 of its own intrathecal tau drug in progressive supranuclear palsy in March, and started it on June 3. My first paper went out on June 22, nineteen days later, and the last on July 8. None of the four mentions the trial. It sat on a public registry for three months before I published a word.

I left a row off it too. The payoff grid had three outcomes: Novartis wins, Lilly wins, Roche wins. An independent Arrowhead appeared in that work as the baseline against which competitive losses were measured, but never as a strategic outcome anyone could choose. No row asked what happens if all three go on licensing, building alternatives or simply waiting. That is where the board sits today.

3. The contract that turned down the engine

The companion note built every number on two values. The floor is the denial value, the competitive damage a company avoids purely by keeping the most threatening rival from winning. The ceiling is that denial value plus the full worth of owning Arrowhead outright.

Denial is the engine. Take it away and a forced move stops being forced.

Which brings me to a document I had not read in June. Section 2.10.2 of the Arrowhead and Novartis agreement is titled Arrowhead Change of Control, and it anticipates exactly the scenario those papers were about. If somebody buys Arrowhead, Novartis may disband or restructure the joint committees and restrict what it shares. The clause addresses separating competing programs and walling off confidential information.

Read it for what it does to the floor. If a rival wins Arrowhead, Novartis keeps the programs it licensed, limits what the new owner learns, and carries on.

Be careful about how far that goes. The license removes one source of competitive exposure: Novartis cannot lose what it already controls merely because somebody else buys the company. It could still lose access to future programs and control of the unlicensed platform. The contract narrows the exposure. It does not tell us by how much, and recomputing the floor is work I have not done. The question is not whether the clause removes Novartis’s reason to bid. It is how much of the original denial value rested on exposure the agreement already protects. Until I separate that from the worth of future programs and control of the platform, I cannot say how much of the floor survives.

Two boundaries on all of this. The protection runs to Novartis alone. Neither Roche nor Lilly benefits from a clause they are not party to, though the rights Novartis already holds shape what either of them would be buying. It also covers the programs Novartis licensed, not the brain pipeline Arrowhead is building around them. The company intends to put three more wholly owned candidates into the clinic inside a year, which leaves substantial brain opportunity outside the existing licenses.

Denial is the engine of the whole thesis. For one of the three, the contract turned the engine down.

There is a further distinction the original numbers never settled, and it may be the more important one. The damage a company would suffer if a rival bought Arrowhead is not the same as the chance the rival tries. A very large problem conditional on somebody moving is not automatically a very large reason to move first. The model estimated the cost of losing. It did not establish how likely anybody was to start the contest.

Part One called Novartis the partner who knows too much, and argued that proximity creates urgency. The contract suggests proximity also bought insulation. The best-informed player is also the one holding contractual protection against part of the mechanism that was supposed to compel it.

It cuts the other way too, which is the more interesting half. The clause also contemplates an acquirer continuing its own competing programs, subject to those safeguards. The parties wrote down a sale without writing it off. Parts of the filed agreement are redacted, so this establishes what was anticipated rather than what anybody intends.

Which leaves Novartis with more than one move. It can run the spinal-canal program toward its 2029 completion and hold the license as a hedge against its own route being obsoleted. It can develop what it has already licensed, with whatever target-selection and substitution rights the agreement still allows. It can also bid for the rest of Arrowhead. The first two preserve choices already made. Only the third is a contest for control.

4. What the readout did, and what it did not

It may have raised the price Arrowhead would accept. The July note said a successful first readout from the brain program would lift the platform’s standalone value, and with it the price below which Arrowhead refuses to sell. That is not a new observation and I should stop presenting it as one. It was in my own work three months ago. Whether the board actually moved that number is not visible from outside.

It may also have strengthened Arrowhead’s standalone position. On Wednesday its chief executive committed to at least three new wholly owned brain programs entering the clinic within twelve months, two applications filed by year end or early next, and another partnered candidate behind them. He described the delivery platform as feeling like a company’s worth of value on its own. That is a man describing what he intends to build.

What it did not produce was a sustained repricing. Arrowhead reached 60.53 during Wednesday’s session, more than four percent above Tuesday’s close, and finished the day at 58.37. Thursday brought a rally of more than two percent in XBI, the biotechnology ETF, and thirteen cents for Arrowhead. By Friday it closed at 55.90, roughly four percent below where it stood before the readout.

I cannot tell you why. Expectations already in the price, doubt about clinical benefit, the cost of the expansion, a bad tape, or plain disagreement. The trading does not say. Any buyer still watching Arrowhead now has more information than it had on Monday and, for the moment, a cheaper quoted price. Whether the board raised the number it would accept is not visible from out here. Those are different prices set by different people.

Success can raise the acquisition price while lowering the incentive to sell.

Which suggests the odds do not rise steadily with the science. Early proof pulls buyers in by retiring uncertainty. More proof raises the price, the bargaining power and the minimum the board will accept. The window may be widest after the technology becomes credible and before independence becomes too valuable to give up.

5. Putting a number on it

A reader is entitled to ask how likely I actually think the check is. Describing a position is not an answer. Here are numbers through the end of 2027, when Arrowhead expects its first patient data and plans to move its two rare-disease brain programs into clinical development.

An acquisition here means a publicly announced definitive agreement transferring control of Arrowhead, signed by December 31, 2027. A license, a minority stake or the sale of individual programs does not count. The numbers that follow are judgments rather than calculations, and I have put the drivers beside each one so you can disagree with an input instead of an output.

Figure 2. Author’s estimates through December 2027. Not derived from a model.

Independent, 50 percent. My largest single outcome, for five reasons. No acquisition was announced in the four months before the readout, the window in which the Novartis paper argued a well-placed buyer would get a lower entry price. Chris Anzalone spent Wednesday morning committing to three new brain programs rather than a process. Licensing captures much of what a buyer wants without paying a premium. Novartis is protected from being shut out of what it holds. The next major patient data are expected in 2027, and those results could materially change the calculation.

Roche, 15 percent. The cleanest fit on the evidence I can see, which is a judgment rather than a ranking. Its shuttle reaches the brain by infusion, Ocrevus faces biosimilars in 2028 and 2029, and it holds no publicly disclosed contractual position in Arrowhead’s brain platform. It also put a hundred million dollars into Rona the day before Arrowhead reported, which shows continued appetite for external RNAi without being evidence it intends to buy anybody.

Novartis, 15 percent. Best informed and contractually protected, which is the paradox of the three. It has a cliff of its own that the June papers underweighted: Cosentyx, at 6.7 billion dollars of 2025 sales, faces estimated US patent expiration in 2029, the same window as Roche’s Ocrevus. Two of these three are staring into the same window. The change-of-control clause protects what it holds. Buying Arrowhead would mean paying for a platform that could eventually challenge the delivery route its own Phase 3 depends on. A defensive bidder rather than an initiator.

Lilly, 15 percent. Level with the other two, and I had this too low. Its chief executive called tau a tough target eight days before the readout, but skepticism about tau is not skepticism about a delivery platform. Lilly has repeatedly backed several approaches to a problem rather than committing to one, and a buyer who hedges across technologies may find ownership attractive precisely because the winning approach is still unknown. That makes Lilly a plausible owner of the technology, not evidence it intends to become one.

Its own exposure is real and differently shaped. Trulicity produced about 4.3 billion dollars in 2025 against a US compound patent expiring in 2027, and Lilly recognized roughly 3.4 billion from its Jardiance collaboration, whose US patent runs to 2029. Mounjaro alone produced nearly 23 billion with protection into 2036. Lilly has the deepest balance sheet in the room, a large new franchise, and less immediate need than its peers to buy a replacement revenue stream.

Figure 3. Capacity ranges as published in the Forced Move series, June and July 2026, on an enterprise value basis.

One more piece of history belongs here, because it involves two of the three. Arrowhead bought Roche’s entire RNAi business in 2011 and Novartis’s entire RNAi business in 2014. Chris Anzalone’s account of why is blunt: they did not believe the field could solve delivery.

Two of the companies in this note have owned this technology before and sold it, having concluded it would not work. The thesis is that they now have to buy it back at a very large multiple of what they let it go for.

That cuts both ways and both belong on the record. These boards can be wrong about RNAi. They are also willing to walk away from it.

One more piece of context applies to all three, and it cuts against me. Roche and Novartis are not alone. Several large franchises face patent and biosimilar exposure in the closing years of this decade, which widens the market for external innovation generally.

That does not help Arrowhead as much as it first appears. A company losing exclusivity in 2029 needs earnings, not merely interesting technology. An early-stage brain platform may be among the most valuable assets it could own for the decade after, and it does not fill the gap when the patents go. The same cliff that raises an acquisition budget can point that budget somewhere else entirely.

The patent cliff creates buyers. It does not select their targets.

That constraint sits under all three numbers. Each of these companies has reason to buy before the decade turns. None has an obvious reason to buy this.

The equal odds are not a claim that the three face equal pressure. They are an admission that I cannot reliably turn very different incentives into different probabilities. The last five points belong to somebody I have not modeled, which Merck KGaA demonstrated is not a hypothetical.

Fifty percent is the least precisely grounded number here. The science makes Arrowhead strategically interesting and several companies have reason to consider it. Neither establishes that any one of them bids. I cannot produce a reliable base rate fitted to these circumstances and will not pretend otherwise. The estimate reflects my judgment of the combined incentives, not evidence that a transaction is already under discussion. It is a forecast and it can be wrong.

One more boundary. These concern whether control changes hands, not what somebody pays for it. The valuations in the original papers are separate estimates, and the October readout validated none of them.

A word about those valuations, since the range in the companion note starts above the largest biotech acquisition ever completed and runs to roughly double it. Bristol Myers Squibb paid about 74 billion dollars for Celgene. Roche paid 47 billion for Genentech. Pfizer paid 43 billion for Seagen and its entire antibody-drug-conjugate platform. Against that record, 90 to 150 billion dollars is off the top of the chart, and anybody checking the number against precedent will find it there.

The defense is not that Arrowhead is special. It is that the number was never a price forecast. A ceiling is what a buyer could rationally pay in a contested situation, denial value plus the worth of owning the asset outright. What somebody would actually pay is a different question and a lower number.

One feature of the asset is genuinely different, though it is a category point rather than a superlative. A normal biotech acquisition prices a pipeline: molecules, indications, launch dates, probabilities. Arrowhead would be bought for the capability that produces pipelines, across seven cell types it can now address, five of them already in the clinic. Nobody has sold a validated multi-tissue delivery engine before, which is why there is no comparable transaction to anchor against.

A number with no precedent is not a high number. It is an unanchored one.

That cuts both ways, and the second way is the one shareholders should sit with. An unanchored number can land far below where I modeled it as easily as above.

A willing buyer is also not enough. A transaction needs Arrowhead’s board and an acquirer to agree on a number, and that may be the largest uncertainty in the whole exercise. It is the one variable none of the outside analysis can see.

If Arrowhead is still independent in January 2028, the original papers will have described pressure without producing the move they contemplated. That would not validate the 50 percent. One outcome never validates a probability.

6. Why you would not see it coming

I argued in The Liver Got Paid that the models pricing Arrowhead have no field for what it proved. This section asks a narrower question a reader put to me: what would acquisition interest look like from the outside, if it existed? The honest answer is that it might look like silence, and the reason is worth understanding.

A preliminary approach generally triggers no duty to announce it. Discussions with a prospective buyer and its advisers can stay confidential under appropriate confidentiality arrangements. Regulation FD does not impose a blanket twenty-four-hour deadline on them, and a preliminary approach is not one of the enumerated Form 8-K events. Nasdaq expects prompt disclosure of material information but allows delay for confidential negotiations, which is exactly what that carve-out exists for.

Talks can therefore sit in the dark lawfully until they are signed. A leak or unusual trading can force the issue sooner, at which point the company may have to respond on a day it did not choose. Neither is guaranteed to happen. The Silence deal announced on Monday, which the next section takes up, is one illustration: no leak I saw, no unusual tape, just a finished agreement.

You may see it in the tape. You may see no sign of it at all until an agreement is signed.

None of that is legal advice and I no longer practice. It explains why an absence of news is not evidence of an absence of interest, and why anyone waiting for a press release to tell them the auction opened has the sequence backwards.

7. Three sessions later, somebody moved

This note was finished last week. On Monday morning Merck KGaA announced an agreement, dated the day before, to acquire Silence Therapeutics, expected to close in the first quarter of next year, for $1.1 billion in enterprise value, $1.3 billion in equity, all cash, at a 130 percent premium to Friday’s close and 82 percent over the thirty-day average.

Silence is a liver company. The drug driving the deal silences a gene expressed almost exclusively in liver cells, in a rare blood cancer, and it has finished Phase 2 with a positive result. One principal value driver, with other pipeline behind it including a Phase 3-ready program against Lp(a). No approved medicines. No commercial organization.

Three details in the announcement matter more than the price.

It was competitive. Silence’s chairman called it the result of a competitive sale process. The announcement does not say how many bidders there were, or who, and leaves it there.

The buyer was outside my framework. I modeled Novartis, Lilly and Roche as potential buyers of Arrowhead. Merck KGaA was not among them, though it had bought SpringWorks the year before. Paying $1.1 billion for Silence does not establish that Merck would pay ten or a hundred times that for Arrowhead. It does establish what my framework never examined: the set of companies willing to own RNAi reaches past the three I named.

I did not see it coming. No warning, no leak I caught, no unusual tape I noticed. Three trading sessions after I argued you would not see one until it was done, Silence supplied the illustration. The negotiations were invisible to me. The agreement was not.

Now set the two companies beside each other, because the comparison is uncomfortable. Silence: a Phase 3-ready rare disease candidate, a second advanced cardiovascular program, an RNAi platform, no approved medicines. Enterprise value $1.1 billion. Arrowhead on Friday: an approved medicine on the market, positive Phase 3 data in hand for a label expansion, human data in five organ systems, two partnered brain programs with three more promised inside a year, the first reported clinical demonstration of substantial spinal-fluid tau lowering after a subcutaneous dose, and roughly $1.6 billion in cash and securities. Equity value near $7.9 billion at Friday’s close. Enterprise value around $7.5 billion once the cash, the convertible notes and the royalty obligations are netted, roughly seven times what Merck agreed to pay for Silence. That figure is approximate: the price is Friday’s, the balance sheet is June 30, and the treatment of the royalty obligations moves it either way.

That is not a claim that Arrowhead should be worth some multiple of Silence. The companies are not comparable and I will not pretend otherwise. It is an observation about what a buyer pays once it has decided, and about how little the public market appears to have rewarded what happened last Wednesday.

Merck put $1.1 billion of enterprise value on a liver RNAi business. It did not buy a license. It bought the company.

8. The best argument against all of this

There is a better objection than the one I used to lead with, and it is not that management sounds enthusiastic about going it alone. Executives say that whether or not a deal is possible.

The real objection is economic. The platform may be worth more to Arrowhead as a licensing and commercial business than any buyer is currently willing to pay to own it outright. Novartis showed in August of last year that substantial access can be bought without buying the company. Sarepta showed the same with the muscle programs. A buyer acquires the unpartnered pipeline, the research engine and the right to decide who gets access next. It does not acquire the ability to unwind rights already granted.

There is a point where renting stops being cheaper. One program can be licensed. The next ten bring upfronts, milestones, royalties, exclusivity limits and the standing risk that somebody else takes the target you wanted. Ownership wins when controlling the engine is worth more than the flexibility of buying what it produces. For these three companies, the public evidence does not establish that Arrowhead has crossed that line.

A further possibility is worth holding, and it is not bearish. There may be many buyers for Arrowhead’s technology and no buyer for Arrowhead at a price its owners would take. Those outcomes are not contradictory. They are what happens when the parts are worth different amounts to different companies and no single buyer can capture all of them. A platform can become indispensable without the company becoming an acquisition, and a decade of licensing revenue is not necessarily the worse outcome for anybody holding the stock.

Arrowhead does not look like a company preparing to sell either, and the expansion Chris Anzalone announced on Wednesday is the evidence. That counts for a little. It is the weakest of the three arguments in this section.

The recent behavior of the buyers points the same way. In the past two weeks Novartis committed $575 million up front to license from a Chinese biotech and Roche put up to a hundred million into an obesity program from Suzhou. Neither is an acquisition. Both are companies choosing access over ownership, which is the choice Novartis already made with this very platform fourteen months ago.

The papers themselves were careful on this point in a way the title was not. A forced move describes a position in which every alternative is worse. It does not promise a timetable. Players sit in difficult positions for a long time, and the position does not care.

What would argue the other way. Three results would shift the argument from licensing back toward ownership. Patient data showing clinical benefit rather than biomarker movement, because a working drug is harder to rent. A second brain target reaching the same knockdown, because that converts a result into a capability. Patent claims giving Arrowhead real leverage over a rival’s freedom to operate would strengthen the ownership case as well, because control of exclusion rights can be worth more than access to the technology. None has yet been established to a degree that would settle the ownership question.

9. What would make me wrong

The result is thinner than the headline. The 73 percent is the mean for six treated volunteers in an eight-person cohort, at the Day 105 assessment. Patient data are not expected until 2027, and healthy people are not sick people.

One brain target is not a brain franchise. ARO-MAPT has shown substantial activity on one biomarker against one target. Whether the same approach produces comparable knockdown against other brain targets is unresolved in humans. If the chemistry turns out to be target-specific, what I have been calling a company’s worth of value may be one exceptional drug with a much narrower delivery story. Arrowhead disputes this, and its claim is narrower and stronger than the one I just described. Chris Anzalone’s position is that once delivery into a cell type is validated, further targets in that same cell type follow, because the cell does not care which sequence it receives. Reaching a new cell type is the hard part; a second gene in a cell type already reached is not. By that account the brain is open and a second brain target should work. It is still a claim nobody has tested.

A competitor may close the gap before any buyer moves. Denali already has an approved medicine using receptor-mediated transport across the blood-brain barrier, and a separate tau program in patients reading out in the first half of next year. If that reads well, Arrowhead keeps its scientific first and loses the assumption that first means best placed.

The price argument can fail in both directions. Large companies have sat through stronger signals than this one. A company spending hard on expansion can also get cheaper rather than dearer, if the market decides the spending lacks discipline. Rising value is not the same as urgency.

The target is still unproven. Lowering tau has not been shown to change the course of anyone’s disease. Biogen reduced tau on brain scans, produced encouraging cognitive signals, and still missed its primary endpoint. If the mechanism fails, the delivery platform keeps real value against other targets and this particular drug does not.

The one that would retire the thesis. There is a way to be right about the platform and wrong about the rest of it. If patient data strengthen, if other brain targets validate the route, and the large companies keep licensing instead of buying, then what I got wrong was not the science. It was the assumption that anybody needs to own all of this. The technology can be exactly what I thought, and the argument for buying the company can still fail.

10. The clock runs while you think

In chess the clock does not stop for difficulty. A player in a hard position can think for forty minutes, and at the end of forty minutes the position is exactly what it was, minus forty minutes.

I said in June that three companies would be forced to move. The more precise claim was that the first buyer would put the others in check. Whether they answer is the part the model has yet to prove.

Three tests, then. The science passed the first. The second is whether somebody makes a credible bid. The third is whether that bid forces the competitive response my original model anticipated. Only the first has been completed.

The gate I named opened, inside the window I pointed at, and the result came through it clear of every hurdle anyone had named. The position that condition was supposed to create now exists. Three companies are sitting in it. One of them has a registrational trial running on the alternative route. None of them has to act at all.

A forced move is not a prediction that somebody moves. It is a description of what is left when they do.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. The characterization of what is and is not in the published literature, the probability judgments, and the chess analogy are the author’s own.

The Forced Move papers were published on June 22, June 29 and July 6, 2026, with the companion note on July 8. ARO-MAPT results are from Arrowhead's announcement and investor webinar of October 14, 2026. Arrowhead's acquisitions of the Roche and Novartis RNAi businesses in 2011 and 2014, and the platform and pipeline figures in Section 5, are as described by Chris Anzalone in a BioCentury interview of October 16, 2026. The Arrowhead and Novartis agreement was entered into on August 29, 2025; the change-of-control provision is at Section 2.10.2 and the target-selection procedures at Section 3.1.2, and portions of the filed agreement are redacted. Novartis's Phase 3 study of NIO752 is registered as NCT07498426.

Patent expiry years are estimates and vary by territory; sales figures are as reported for 2025 and, where a product is partnered, reflect the named company's recognized revenue rather than worldwide product sales. Arrowhead's enterprise value is approximate, derived from its October 16 closing price and its cash and securities position as of June 30, 2026. Merck KGaA's agreement to acquire Silence Therapeutics was announced on October 19, 2026, and the Novartis and Roche licensing transactions on October 2 and October 13. Diranersen results are from Biogen's announcements of May and July 2026. The probabilities in Section 5 are the author's own estimates and not the output of any model.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The claim I am going to try to prove

Arrowhead rose 24 cents on Wednesday, four tenths of one percent, on the day it reported the first clinical evidence that a shot under the skin can silence a gene in the human central nervous system.

One point has to be made immediately, because leaving it out would be dishonest. Wednesday was the worst day for American stocks since July. A hot inflation print sent the ten-year Treasury above 5 percent, the S&P 500 fell 1.11 percent, and the Nasdaq fell 1.88. Against those, a gain of 0.41 percent looks respectable.

Biotechnology told a different story. The sector index finished the day essentially flat. Measured against its own industry rather than against technology stocks having a bad afternoon, Arrowhead’s outperformance was under half a percentage point.

One more detail from Wednesday belongs here. Arrowhead traded as high as 60.53 that session, more than four percent above the previous close. By the bell almost all of it was gone. The question is not whether investors noticed. It is why so little of what they did survived the afternoon.

The next day gave everyone more time. The webinar had run at 2:30 that afternoon, so Thursday was the first full session with the presentation, the dose curve and the scientific explanation all public. Biotechnology rallied more than two percent. Arrowhead added thirteen cents.

Across the two sessions the sector gained about two percent and Arrowhead gained less than one. The gap widened after the explanation rather than closing.

Hold that number. Under half a point on the day, and further behind by Thursday. The rest of this note is about what the same market has paid for comparable evidence on other days.

A lot of people found that insulting. Rather than simply agree, here is a falsifiable version of the complaint. The market has repriced companies violently on exactly this kind of evidence before. Twice in the last five years it moved roughly forty percent in a single session on an interim Phase 1 readout with single-digit patient numbers and no clinical outcome at all.

Both times, the organ was the liver.

2. Four firsts, four tapes

Figure 1. Compiled from company announcements, press coverage at the time, and the trading record for each date.

Intellia. In June 2021 Intellia reported that six patients had received a gene-editing drug through an intravenous infusion, and that the protein it targets had fallen by a mean of 87 percent at day 28. No clinical benefit was measured. The company called it proof of concept for its delivery platform. The shares closed up more than 55 percent. The forty in this note’s title is the figure the wires carried that morning, and the conservative one.

Verve. In November 2023 Verve reported the first base editing performed inside a human body. Nine patients evaluable, cholesterol down between 39 and 55 percent at the higher doses. One participant had a heart attack the investigators judged possibly related to treatment. The shares closed down about 41 percent.

Sarepta. In March of this year Sarepta reported the first clinical data on two siRNA drugs for muscle disease: drug concentrations in muscle tissue, suppression of a panel of genes downstream of the target in one program, a placebo-adjusted reduction of the target gene in the other, and favorable tolerability. The shares closed up about 35 percent.

Arrowhead. On Wednesday Arrowhead reported that six treated volunteers had a mean 73 percent reduction in spinal-fluid tau, measured ninety days after the last injection. That is pharmacodynamic evidence of silencing in the central nervous system rather than a measurement of gene expression in brain tissue, and the distinction is worth keeping. No serious treatment-emergent events, no discontinuations attributed to them, no persistent anemia. The shares rose 24 cents.

The market is not indifferent to this category of news. It priced three of these in double digits and the fourth at four tenths of one percent.

3. The March case is Arrowhead’s own technology

Start with Sarepta, because it is the most useful of the four comparisons.

SRP-1001 and SRP-1003 are Arrowhead drugs. Arrowhead’s own proxy statement lists them under partnered programs and gives their former names: ARO-DUX4 and ARO-DM1. The delivery chemistry that carries them into muscle is Arrowhead’s. Sarepta licensed the platform. The trade press covering the readout said so in the headlines, calling them Arrowhead’s siRNA assets and RNAi drugs from Arrowhead.

What Sarepta reported on March 25 was early, and it was not trivial. Drug concentrations in muscle well above a competitor’s, a dose-dependent rise, suppression of the gene panel downstream of DUX4 of roughly 90 percent against placebo in the FSHD program, and a 50 percent placebo-adjusted reduction of the target gene in the myotonic dystrophy program, which the company itself called limited given the numbers involved. No clinical outcomes in either.

Sarepta closed up about 35 percent.

In March the market added about a third to a company’s value for results built on Arrowhead’s technology.

Arrowhead moved too. It rose eight percent that same day. Biotechnology was strong, up about 3.7 percent, so call it four points of sector-relative gain. Not all of it belongs to Sarepta’s news and I cannot separate out what does. The point is simpler than attribution.

Investors had no difficulty repricing Arrowhead in March on evidence about its licensed delivery chemistry working in muscle. Seven months later, on evidence about its own delivery chemistry working in the brain, the sector-relative gain was under half a point. The same stock, the same engineering group, roughly a tenth of the response.

That needs one qualification. Sarepta holds the development and commercial rights, so most of the direct economics of those two programs belong to it, with Arrowhead retaining milestones and royalties. The market had a rational reason to reward Sarepta. What is harder to see is how much it assigned to the layer underneath, which was evidence that a targeted delivery chemistry works in people, and that chemistry is Arrowhead’s.

This is the hardest of the comparisons for a skeptic to wave off. It was this year, in this market, in this sector. The data was in patients rather than healthy volunteers, and it was thinner than Arrowhead’s. It was also not tau, so whatever the market thinks of that target cannot explain it.

4. The Verve case closes the other door

If the only comparison were Intellia, somebody would reasonably say that 2021 was a different market, that money was free and gene editing was a story stock, and that none of it carries over.

Verve answers that. November 2023 was not a bull market in biotech. The move was also violent in the other direction, which is the point. The market read an interim Phase 1 biomarker readout in nine patients, found one safety signal in it, and took forty percent off the company inside a session.

A market that can do that has not lost the ability to price early platform data. It was paying close attention in 2023. It paid close attention in 2021. On Wednesday its reaction was barely visible.

5. Eight examples are not a distribution. They are a rebuttal

A reasonable reader at this point says I have picked three days out of five years and arranged them to suit myself. That is fair, and these eight cannot establish how the market normally prices Phase 1 data. No such claim is being made. What they establish is narrower: investors have repeatedly assigned large changes in company value to early genetic-medicine evidence, sometimes with no clinical outcome, sometimes without the company reporting a result of its own. Whether the market knows how to react is not in dispute.

Figure 2. Single-session share moves following interim Phase 1 data releases in genetic medicine. Compiled from company announcements and press coverage at the time.

Wave is the extreme case. In October 2024 it reported the first therapeutic RNA editing ever performed in a human being, in two patients, on a single dose. The stock closed up about 74 percent. Eleven months later it reported more data from the same Phase 1b/2a study, the result came in below what investors wanted, and the stock fell about 17 percent.

One detail about that October makes it the closest parallel to Arrowhead of any case here. WVE-006 was already licensed to GSK, which held the global rights, with Wave keeping milestones and royalties. The market repriced the originating company by three quarters anyway. The clinical questions were not equivalent, and the protein Wave restored has a better established link to its disease than tau reduction has to slowing dementia. The licensing structure is what makes the comparison useful. Investors were willing to pay the originating technology company even when a partner held the global rights.

The September 2025 readout is the one worth sitting with, though, because of what happened to everybody else that day. Beam rose 15 percent and Korro rose 17 percent. Neither company reported a result that day. They moved because a competitor’s result changed where the value in a shared approach might end up, which is not quite the same as validating the approach.

That is the market this note is about. It will reprice a company by double digits because a rival published. It repriced Arrowhead by four tenths of one percent for the first gene silencing ever achieved in a human brain.

6. Where Arrowhead’s data is stronger, and where it is weaker

Honesty here cuts both ways, so take the weaker side first.

Against Arrowhead. Intellia and Verve dosed people who had the disease. Arrowhead dosed healthy volunteers. That is a real difference and I will not argue it away. A biomarker in a sick person carries information about treating that person which the same biomarker in a healthy one does not.

The reason for the difference is worth knowing, though. Exposing a healthy person to a potentially permanent change in their genome, with no prospect of benefit to them, is a risk-benefit calculation that rarely survives review, so those companies had little choice but to study patients. Silencing is reversible and a shot under the skin is routine, which is how Arrowhead got 41 volunteers through a study in the time it takes to enroll a handful of patients. What makes the data look weaker is downstream of what makes the drug interesting.

There is a further point against my own concession. For validating delivery specifically, rather than efficacy, healthy volunteers may offer the more controlled test. Sick brains vary. Disease stage, atrophy, comorbidity and concomitant medication all add noise to a measurement whose whole purpose is to establish whether the drug arrived. A clean dose-response in people whose biology is not confounded tells you more about the vehicle than a messier signal in patients would.

Then a qualification to the qualification. Spinal-fluid tau is an indirect measure, and nobody biopsied a brain. A diseased brain may differ in barrier integrity, in tau biology and in how the drug distributes. The healthy-volunteer result makes the delivery case stronger. The patient cohorts will show how much of it survives contact with the disease.

For Arrowhead. Now the stronger side.

Intellia reported its headline number at day 28. Verve used a time-weighted average from day 28. Wave saw a signal by day 15 and carried it to day 57 on one dose. Arrowhead’s 73 percent was measured ninety days after the last injection.

Ninety days does not simply beat twenty-eight, and it would be sloppy to claim it does. Intellia was editing DNA, aiming at a permanent change, so its short observation window was never the point. Arrowhead is silencing RNA, which wears off. The significance of day 105 is not that the number is larger. It is that substantial suppression persisted a quarter after the last injection from a drug that makes no permanent change at all, which is what an infrequent dosing schedule would require. That is a potential commercial advantage, provided efficacy and repeat dosing hold up.

Arrowhead showed rising mean suppression across three dose levels, 40, 69 and 73 percent, though the top two are close together and the highest cohort received two injections where the others received three. Verve’s safety signal cost it forty-one percent. Arrowhead’s interim safety observations were encouraging, including on the specific risk that attaches to its delivery receptor.

Then the organ.

7. The liver is the cooperative one

The next paragraph is the argument.

Lipid nanoparticles, the delivery vehicle in both of those gene-editing programs, reach the liver with the biology helping rather than resisting. Put them in a vein and the liver takes them up through a receptor it already uses for cholesterol. Getting that to work took real engineering and years of it. The point is that the organ was cooperating.

The brain is the opposite. Plenty of small molecules cross its barrier, which is why anesthetics and antidepressants exist. What that barrier excludes is large charged molecules, and an siRNA construct is exactly that. Oligonucleotide drugs that reached the human brain before this week generally required direct administration, most often a needle between the vertebrae.

Two companies demonstrated they could reach the organ where the biology cooperates, and the market repriced them by tens of points each.

8. What actually changed

Not the market’s appetite for platform data, as Verve shows. Not the quality of the evidence, which is stronger on durability and safety and weaker on population. What may have changed is how the company gets valued.

Intellia in 2021 and Verve in 2023 were primarily platform companies. Each booked some collaboration revenue, and neither had an approved product generating commercial sales. The platform dominated the investment case. Validate it and every number in the model moves, because there was little else in the model to anchor them.

Arrowhead’s chief executive, Chris Anzalone, called this a therapeutic holy grail on the call, and had the sense to flag that the term is overused before he used it. He is not wrong about the difficulty. Delivering an oligonucleotide into the brain from a shot under the skin is a problem the field has failed at for thirty years.

The trouble is that difficulty is not an input. A model has fields for market size and launch year and none for how long a problem went unsolved.

Which produces an odd result. An easy problem gets solved on the way to a product, so its value lands inside somebody’s forecast. A genuinely hard one gets solved before the business that would use it exists, so no forecast has been built. The harder the problem, the less a spreadsheet has to say about having solved it.

Chess engines are useful here. They used to judge a position by adding up terms a programmer could name: material, king safety, pawn structure, mobility, each weighted in advance by somebody who decided what mattered. Then in 2017 a program called AlphaZero, which had been given the rules and left to teach itself by playing, began handing over material for positions the old arithmetic had no term for, and winning. The arithmetic was not wrong. It was counting pieces at a moment when the position had become about what the pieces could reach.

A discounted cash flow model is an evaluation function. Somebody decided in advance which terms it contains.

A blank field is not an estimate of zero. It is a decision not to estimate.

Put plainly for anyone who skipped the chess: this is a limit of the instrument, not a judgment rendered by it. Discounted cash flow can be extended to value uncertain platforms, and real-options methods exist precisely for this. The question is whether the published models reviewed here do any of that.

Arrowhead has an approved drug, a Phase 3 readout, and 1.6 billion dollars. It is priced on discounted cash flow now, anchored on a cardiovascular franchise. Five of the commercial assumptions such a model runs on are market size, penetration, price, probability of success and launch timing.

A healthy-volunteer biomarker, against a target no regulator has ever approved a drug for, supplies too little to populate those inputs with any confidence.

9. A drug, and a method for making drugs

A drug is a bet on one molecule, one disease, one market. A delivery platform is a bet on the ability to place more of those bets.

If ARO-MAPT fails in Alzheimer’s, one therapeutic hypothesis failed. It does not undo the evidence that this construct crosses into the brain and lowers its target there. If the same chemistry carries a different payload, the consequences are not confined to tau.

One experiment changed the odds on experiments nobody has run yet. That is what separates a platform from a drug.

Though it is not one key. The chemistry carrying siRNA into muscle for the Sarepta programs is an integrin-targeting ligand. The construct going into the brain uses an antibody fragment against a different receptor entirely. These are not the same key.

That does not weaken the argument. It changes what the argument is about. Arrowhead is not demonstrating that it found a universal delivery molecule. It is demonstrating that its approach to engineering tissue-specific delivery keeps producing solutions for tissues that did not have one.

A key that opens one door is useful. A method for cutting keys to doors nobody could open before is a different asset.

A platform also creates choices, not only programs. Arrowhead does not have to pursue every target it can now reach. It can select, abandon, partner or keep, and revise as the biology comes in. Maintaining the capability costs money every year, so flexibility is not free and does not by itself imply a positive number. That flexibility may carry economic value even so, and a forecast built around individual drugs has no place to put it.

Opening a file does not require the rook to move. It gives the rook moves it did not have before.

Liver, then lung, then muscle, then fat, now brain. None of these opened the same way. The lung program used an inhaled route. The fat program showed suppression directly in tissue. The brain program gives pharmacodynamic evidence after a subcutaneous shot, with nobody biopsying a brain. Different problems, different routes, different grades of evidence.

Which is the point rather than a weakness in it. The common achievement is not that every organ has fallen. It is that the same group keeps finding ways to make siRNA work in tissues that had no way. None of that guarantees the next door opens. The list of doors already opened is what is being priced here, or not priced.

A portfolio of unrelated drugs is a set of separate bets. A portfolio built on shared delivery engineering has a common layer beneath it. Every new target still carries its own biology, but none starts from zero on whether the drug can get there.

There are two levels to this and they behave differently. Within the brain, programs using related machinery share technical risk directly. Across organs, what is shared is broader and looser: the engineering capability that keeps producing tissue-specific solutions.

The correlation cuts both ways at each level. A problem with the brain receptor would damage several brain programs together, and would say little about muscle, which uses a different mechanism entirely. One platform does not mean one risk. It means related capabilities whose risks overlap in some places and not others.

Correlation is not a gift. It is a property.

It also means the cause of any future failure matters. If tau turns out to be the wrong target, the delivery evidence survives more or less intact. If the construct proves toxic over time, or distributes worse in sick brains than in healthy ones, the platform takes the damage with the drug.

This is not speculation about targets Arrowhead might one day find. On Wednesday’s call Chris Anzalone said the company already has one wholly owned and one partnered brain program in the clinic, expects another partnered candidate there in 2027, and intends to put at least three more wholly owned candidates into the clinic inside twelve months. He described the platform as feeling like a company’s worth of value on its own.

A reader can discount all of that. Transferability to other targets is unproven. The doses are high. Manufacturing, long-term receptor safety and real clinical convenience are all open. One target is not a platform. Each of those argues for a discount, possibly a brutal one, and together they may fully explain why an analyst declines to put a figure on it today. That is a different claim from the capability being worth zero, and the narrower one is the only claim this note makes.

In chess, opening a file changes what every piece that can use it is worth. Arrowhead may have opened one. The models are still counting pieces.

10. At least three analysts said it out loud

Three banks published on Arrowhead this week, and these are only the notes I could read. All three are positive on the stock. None of the three identifies a material contribution from the brain program to its published target.

Figure 3. Compiled from the three published research notes available to the author, all dated October 14, 2026. Other banks cover Arrowhead and their models were not reviewed.

UBS is the most explicit, and the most useful, because it states the mechanism rather than hiding it. ARO-MAPT is not included in the valuation and represents upside. That is not a judgment that the program is worthless. It is a statement that the model has no field for it.

Stifel reaches the same place in plainer language, writing that there is no credit here and that severe hypertriglyceridemia remains what matters for the stock. Morgan Stanley’s risks to the target, up and down, are cardiovascular and obesity. The brain appears on neither list.

A caution about all of this. One limitation has to be stated plainly. Published analyst targets are not the market. They are evidence of how several influential participants are framing the problem, and no more than that. Private models, discretionary positioning and speculative premium could all carry CNS value that never appears in a price target, and the trading record cannot tell us how much of this was already discounted.

What the three notes do establish is narrower. Substantial new clinical evidence arrived, and explicit value for it is hard to locate in the models I could read.

11. What these numbers are actually made of

One more item from the UBS note, because it reveals what price targets are made of.

UBS raised its target from 63 dollars to 109, a jump of 73 percent, on assumption of coverage. A new analyst took the pen. The note is built on the cardiovascular franchise and explicitly excludes the brain, so whatever moved the number by three quarters, it was not the data that landed that morning.

That is not a criticism of the analyst, whose model is more thorough than most. It is a reminder of what a price target is: one person’s spreadsheet, which changes when the person does.

12. What would make me wrong

The populations differ. Intellia and Verve dosed patients. Arrowhead dosed healthy people. If the market simply weights that difference more heavily than I do, the whole comparison softens.

Perhaps the market expected it. Prices move on the distance between a result and what was already believed, not on how important the result is. Arrowhead had announced the readout in advance, booked a webinar, and discussed a benchmark of 50 to 60 percent beforehand. Seventy-three cleared it. Whether it cleared what investors had already priced is a different question, and management’s stated objective is not the same as the market’s expectation. I cannot rule this out, and it is the best argument against the whole of this note. It does raise a second question. If the market already expected systemic silencing in the brain to work, that expectation should be visible somewhere in what the brain is worth. In the three models I could read, it is not.

Not every biomarker is equal. Cholesterol is a validated treatment target with decades of outcome data behind it. Tau is biologically compelling and therapeutically unresolved. A 73 percent reduction in spinal-fluid tau establishes pharmacology, not a reduction in anybody’s risk of dementia, and a market discounting the commercial program on that basis is making a defensible judgment rather than an error.

Which leaves the question this note keeps circling. If tau turns out to be the wrong target, does the ability to deliver gene silencing into the human brain become worthless? Those are two risks, not one, and no rule requires them to be priced together.

Tau has a history. Biogen’s diranersen lowered tau substantially, reduced tau on brain scans, produced encouraging cognitive signals, and still missed its primary dose-response endpoint. An older Biogen antibody against extracellular tau failed outright. A market that has watched this target disappoint may be pricing target risk rather than ignoring delivery, in which case it is making a judgment rather than an omission.

The stock was already broken. Arrowhead traded above 95 dollars inside the last year and sits near 58 now. A stock already down by a third on commercial worries may be too preoccupied to listen, which would be a different failure from the one I am describing.

Size matters. These companies were smaller than Arrowhead is now, and a move of forty percent on an eight and a half billion dollar company is a different ask than on a two billion dollar one.

They may simply be right. The last one is the serious one. The models may be right. A Phase 1 biomarker in six people, against a target with no approval precedent, may genuinely deserve a zero until there is more. Analysts have declined to credit assets before and been correct to wait.

Here is how to tell which of these is right. Those are three different explanations and they imply three different tests. If the market is waiting for patients, the Alzheimer’s and tauopathy data in 2027 should move it. If it will not credit the platform until a second target works, the first non-tau brain readout is the one to watch. If the constraint is commercial visibility rather than science, the next informative event may not be a readout at all. It may be a transaction, because a licensing deal requires somebody other than Arrowhead to put a number on this. Upfront cash, the breadth of target access and where the milestones sit would say more than another biomarker. A single-asset license and broad platform rights are very different signals. The next eighteen months may begin to separate them.

13. Twenty-four cents

The liver is the organ where the biology cooperates. Two companies showed they could edit a gene in it, in a handful of patients, with a biomarker and no clinical outcome, and the market repriced them by forty points in one direction or the other. In March it paid a third company roughly a third of its value for reaching muscle, using chemistry Arrowhead invented and licensed out.

The brain is the organ that a hundred years of pharmacology has failed to reach without a needle in the spine. Arrowhead reached it with a shot, in six people, with a biomarker and no clinical outcome, and held the effect for three months.

Twenty-four cents.

The market did not decide the brain was worth less than the liver. It never got as far as deciding.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. The characterization of what is and is not in the published literature, the probability judgments, and the chess analogy are the author’s own.

Intellia’s interim Phase 1 results for NTLA-2001 were announced on June 26, 2021, covering six patients across two dose cohorts, with an 87 percent mean reduction in serum transthyretin at the higher dose by day 28, and were published in the New England Journal of Medicine. The share move is as reported at the time. Verve’s interim results for VERVE-101 were presented at the American Heart Association Scientific Sessions on November 12, 2023, covering ten dosed and nine evaluable patients, with time-averaged LDL-C reductions of 39 and 48 percent in the two patients at 0.45 milligrams per kilogram, and a serious cardiac event in one participant judged possibly treatment-related. The share move is as reported contemporaneously. Sarepta’s first clinical results for SRP-1001 and SRP-1003 were announced on March 25, 2026.

Arrowhead’s proxy statement lists both as partnered programs under their former designations, ARO-DUX4 and ARO-DM1, and Sarepta’s siRNA platform is licensed from Arrowhead. That share move is also as reported at the time. Sector-relative comparisons use XBI, the SPDR S&P Biotech ETF, as the benchmark, on closing prices. The additional moves shown in Section 5 are drawn from company announcements and press coverage at the time of each release: Wave’s first-in-human RNA editing result of October 2024, from the ongoing Phase 1b/2a RestorAATion-2 study, and its multi-dose update from the same study on September 3, 2025, together with the same-day moves in Beam Therapeutics and Korro Bio, neither of which reported results that day.

ARO-MAPT results are from Arrowhead’s announcement and investor webinar of October 14, 2026. Price targets, model contents and the change in the UBS target on assumption of coverage are from research notes published by UBS, Morgan Stanley and Stifel dated October 14, 2026. Arrowhead’s cash and securities position is from its quarterly filing for the period ended June 30, 2026.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. It moved

Three days ago I published a note arguing that the tau percentage arriving this morning would carry less information than people would read into it. The argument still holds, and I will come back to it. The first point to make, though, is the plain one.

ARO-MAPT reduced tau in the spinal fluid of healthy volunteers by a mean of 73 percent, with one individual reaching 85 percent. The drug was given as an injection under the skin. Arrowhead describes this as the first clinical validation of a platform delivering siRNA across the blood-brain barrier after a subcutaneous dose. Gene silencing has reached the human brain before, but by going in directly: antisense drugs injected into the spinal canal, and gene therapies infused into brain tissue through the skull.

Arrowhead did it with a shot.

Consider what that route makes possible. A spinal injection needs a procedure room, a trained hand, and a patient willing to come back for it. That constraint shapes what follows from it: who enrolls in the trial, how long the trial runs, which diseases are worth attempting, how many people ever get the drug. Remove it and the economics and practical reach of many brain programs could change.

It also changes what is worth trying. A disease that could never justify repeated spinal taps becomes approachable. Some patients for whom repeated spinal procedures are impractical become reachable. Treating people earlier, before the damage accumulates, stops being a logistical fantasy. None of that is proven this morning. All of it becomes possible to attempt.

The bishop moved. The long diagonal is open.

2. The number almost nobody will read

Everyone will quote the 73 percent. The number sitting underneath it is 105.

Day 105 is when the measurement was taken. The final dose in that cohort was given on Day 15. The reduction was therefore measured ninety days after the last drug entered the body, and it was still 73 percent. That is not a reading taken near the dosing period. Whatever the peak was, the release does not say.

Be clear about what that does and does not establish. One measurement at Day 105 shows suppression at Day 105. It does not show that the effect held at 73 percent the whole way, and it does not say how long it lasts beyond that point. The time course is missing.

What it does establish is durability at a point three months removed from the last dose, which is a credible basis for investigating dosing once a quarter. Against a procedure in a hospital, that is a different kind of medicine.

The headline is 73 percent. The number that matters is 105.

3. What I said to watch, and what arrived

On Sunday I listed six readings that would carry more information than the headline percentage. Here is the honest grading.

Figure 1. The six readings published October 11 and the results reported October 14. Source: Arrowhead’s announcement of October 14, 2026.

Three came back well. Three need detail the release did not provide, and all of them belong in the questions when Arrowhead hosts its webinar this afternoon. Neurofilament light, a marker of neuroaxonal injury, is not mentioned at all, in a safety section that otherwise reports no persistent anemia and no neurobehavioral events. No cohort sizes are given anywhere either, which means the 85 percent is one person and the means rest on groups of unstated size.

Neither omission is evidence of a problem, but both limit what the data can establish. It helps to separate two kinds of limit here. Some are built into the study: the volunteers were not selected for Alzheimer’s-related tau pathology or for decline, and neither pathology nor clinical benefit was assessed. No announcement can fix that. Others are simply absent from a short release and could be answered this afternoon. A reader should know which numbers are load bearing and which are not.

4. The ladder separated. Is it flattening?

The dose groups separated. At least numerically. That is where the interesting questions begin.

Figure 2. Mean reduction in spinal-fluid total tau at Day 105. Source: Arrowhead’s announcement of October 14, 2026.

The means are ordered in the right direction, and that is encouraging. Whether they describe a clean dose-response curve is another question.

Then do the math the release does not do for you. The top cohort received two doses. The two below it received three. The cumulative amount given was therefore 6.3, 19.2 and 25.6 milligrams per kilogram. The per-injection dose doubled from the middle group to the top, but the total drug administered rose by only a third.

That changes the reading. Tripling the total administered dose coincided with 29 additional percentage points of mean tau reduction. Raising it by another third coincided with four. The pattern still points toward diminishing returns, but it is a weaker claim than the headline doses suggest, and it is not a ceiling.

Administered dose is also not the same as exposure, which depends on uptake, distribution and clearance and cannot be read off a dosing schedule. To know whether the curve has genuinely flattened you would need cohort sizes, the spread of individual responses, and measured exposure rather than inferred. None of that is in the release. Anyone modeling 85 or 90 percent from the next cohort should want those numbers first, and so should anyone declaring a ceiling.

That cuts the other way too. The group with the fewest injections produced the deepest reduction at Day 105, which does not prove the shorter schedule was better, since the per-injection dose was also higher. It does make a question worth asking: how many loading doses does this drug actually need? Every injection removed from a regimen removes one more burden on the patient.

The top cohort got two doses. The ones below it got three. That is not a detail, it is the comparison.

5. How it compares

Biogen’s diranersen, the antisense drug that reduced tau on brain scans and showed signals of slower decline on secondary measures in early Alzheimer’s this year, reported spinal-fluid tau reductions of 50 to 65 percent across its doses. That drug is delivered into the spinal canal.

Arrowhead reported 73 percent through a shot under the skin.

That comparison does not establish superiority, and I would not use it that way. Different trials, different populations, different schedules, different measurement days. Healthy volunteers are not patients. Biogen also ran for eighteen months, and its best clinical result came from a dose given every six months, which is a less frequent schedule than the quarterly dosing Arrowhead is investigating.

There is a further asymmetry in what was measured, and it does not favor Arrowhead. Biogen’s tau reductions were accompanied by declining tau PET signal in patients with Alzheimer’s disease. Arrowhead has a spinal-fluid number in healthy volunteers.

The delivery distinction is real, though, and it is the part that does not depend on the numbers lining up. Biogen needed an injection into the spinal canal to reach substantial tau lowering. Arrowhead has now reached substantial tau lowering with a subcutaneous injection.

The comparison worth watching is not seventy-three against sixty-five. It is what a patient has to go through to get the drug.

6. Where the drug goes

There is a second difference between the two routes, and it may matter more than convenience does.

A drug injected into the spinal canal reaches the brain by diffusing inward from the fluid around it. That works well for structures near the surface and progressively worse the deeper you go. In primate studies spanning dozens of regions, accumulation in deep structures ran more than an order of magnitude below the cortex. Autopsy tissue from patients treated with an approved intrathecal drug showed the same gradient, with substantially higher concentrations in the lumbar spinal cord than in the brainstem, though measurable drug reached both.

Arrowhead ran the comparison itself and showed it at a conference in May. In monkeys given the drug into the spinal canal, relative concentrations were 21.6 in temporal cortex, 11.0 in thoracic spinal cord, and 1.0 in both caudate and putamen. That is a twenty-two fold spread from the best-served region to the worst. Under the skin the same four regions ran 3.6, 1.0, 2.0 and 3.3, a spread of under four fold. Confine it to the three brain regions and the subcutaneous spread falls under two fold, which is the comparison the company’s own slide draws.

Read that carefully, because it is a claim about evenness rather than depth. The company reported 70 to 80 percent knockdown across all twelve regions it measured, including the pons, the medulla, the thalamus and the basal ganglia, with up to 85 percent in some cortical areas. Staining showed the effect reaching neurons, astrocytes, microglia and oligodendrocytes in the brainstem. The argument is not that the shot silences harder. It is that in these monkeys it silenced across every region measured, including deep structures the spinal route reached less efficiently.

Twenty-two fold variation by spinal injection. Under four fold by a shot. The gap in evenness is the argument, not the gap in depth.

Four qualifications matter here. The two arms used different schedules, a single 15 milligram dose into the spinal canal against three weekly doses of 3 milligrams per kilogram under the skin, so this is a comparison of routes as each was run rather than of optimized regimens against each other. The concentrations were normalized within each route, so the absolute levels are not directly comparable. The work is also Arrowhead’s own, presented at a conference rather than peer reviewed, while the intrathecal literature it argues against is published.

The fourth is a detail in the timing. Both groups were sampled on Day 29, but that was twenty-eight days after the single spinal dose and only fourteen days after the last of the three subcutaneous ones. The two arms had different amounts of time to clear before anyone measured them. The separate knockdown comparison was run at a matched interval after the final dose in both groups, which makes the gene-silencing result the more informative of the two.

One more counterweight, and it is the strongest objection to this section. An independent primate study published in 2025, using an antisense drug against a different gene entirely, found measurable target engagement across neuronal cell types in deep brain regions even where intrathecal drug accumulation sat near the limit of detection. Its authors concluded that intrathecal drugs remain relevant to deep-brain diseases. Lower concentration is not the same as no activity, and a reader should not take the gradient to mean the other route cannot work down there. The claim worth defending is that uniformity may be better, not that only one route arrives.

Whether any of this holds in people is unanswered. Spinal-fluid tau is a bulk measure, and this morning’s release reports no regional detail at all.

Hold it next to what comes next, though. Progressive supranuclear palsy attacks deep structures, including the basal ganglia, the subthalamic nucleus and the brainstem, several of which sit far from the surfaces a spinal injection works from. That is unlikely to be a coincidence in the company’s plans.

7. Two diseases, named, with regulators already involved

In September I wrote a note called The Long Diagonal arguing that Arrowhead’s fastest road to approval in the brain might run through rarer diseases rather than through Alzheimer’s. The evidence was circumstantial: a talk title, a biomarker on a slide, two neurologists on a conference agenda.

This morning’s release says the company has initiated regulatory discussions in preparation for planned studies in MAPT-associated frontotemporal dementia and progressive supranuclear palsy.

Not considering. Not evaluating. Initiated discussions, in preparation for planned studies, in both diseases.

That is further along than I guessed, and it arrived just over two weeks later. Neither disease has an approved disease-modifying therapy. Both have small and biologically well-defined populations, and in neurofilament light a progression marker that Adam Boxer, the neurologist presenting for Arrowhead this morning, spent years establishing in exactly these diseases. None of that means approval is close, or that these routes are necessarily faster. It means the company is already building a development path that does not run only through Alzheimer’s.

I argued the fastest road ran through the rarer diseases. The company has already started the conversation.

8. What this actually proves

Three days ago I drew a distinction between three questions that get collapsed into one: did the drug reach its target, did it change the pathology, did it help the patient. This morning gives compelling evidence for the first, though the measurement is indirect, and leaves the other two exactly where they were.

Tau pathology and disease modification were not assessed in these volunteers, by design. There was no Alzheimer’s decline being measured, and no evidence that the study looked at reversing existing pathology. The release says so plainly, which is to the company’s credit. Patient enrollment continues, with further data updates expected in 2027.

No part of today tells you that lowering tau helps anyone with Alzheimer’s. That remains an open question, and the strongest evidence for it is still Biogen’s, not Arrowhead’s.

What today tells you is different, and it may be worth more. A company has demonstrated that a subcutaneous injection can produce substantial target suppression in the human central nervous system, still measurable three months after the final dose. The tau thesis could be wrong in every particular and that would still be true.

For an investor the practical consequence is that two different assets moved this morning, and from here they need not move together. One is ARO-MAPT, whose value still turns on patient data, years of safety, and regulatory outcomes in three diseases. The other is the delivery system, which is a reusable capability if it generalizes beyond this target, and that has not been shown yet.

There is a company commitment attached to this. On its spring earnings call Arrowhead said that if early ARO-MAPT data were encouraging, readers should expect a substantial expansion of its central nervous system pipeline beginning at the end of 2026. The data arrived this morning. Whether that expansion materializes, and which targets it names, is the first concrete test of whether the platform argument is real or a hopeful reading of one result.

Which one survives depends on how the drug fails, if it fails. If tau is adequately suppressed in the neurons that matter and Alzheimer’s progresses anyway, the delivery system did its job and the failure belongs to the therapeutic hypothesis. If the trouble is that the drug reached too few neurons, or reached them unevenly, or the chemistry carrying it causes harm, the platform is implicated directly. Those are different failures and they should produce different valuations.

There is a second level to this, and it decides how much the platform is worth. The regional evidence earlier covers one drug against one gene in monkeys. The harder question is whether the same chemistry carries a different gene, into a different cell type, in a different disease. Crossing the barrier is one achievement. Showing that the same delivery chemistry can reliably silence other targets, in the cell populations a given disease damages, is another. That validation is still ahead. The first has real support this morning. The second is where the platform’s value gets settled.

The drug could fail and the delivery system still succeed. Today was the delivery system.

9. What would make me wrong

This is not the disease. Older brains, damaged brains, brains with tangles in them may handle this drug differently. Every assumption in this note rests on a population that does not have the disease.

The measurement is indirect. Spinal-fluid tau is a proxy for what is happening in brain tissue. It is a good proxy and it is the one the field uses, but it is not a measurement of drug concentration in a neuron, and it is not a measurement of tangles.

The numbers come from unstated denominators. No cohort sizes were given. The 85 percent is one individual. The release also gives no placebo-adjusted effects and no confidence intervals. The percentage is clear. The precision around it is not. Small groups can produce impressive means, and the next sample can move those means in either direction. If cohort sizes and variance eventually show that the gap between 69 and 73 percent could readily be explained by sampling variability, the top of that ladder may be one step rather than two.

Safety is early, and good so far. The disclosed safety is encouraging: no serious treatment-emergent adverse events, no deaths, no treatment-emergent adverse events leading to discontinuation, most events mild, and no persistent anemia or neurobehavioral events reported. That is a good first look. It is also one interim look at small groups of healthy people over a few months, which cannot establish long-term tolerability in older patients with disease.

The ceiling may already be here. The curve appears to bend near the top of the reported range, though the differing schedules complicate that reading. If suppression is approaching saturation near 70 percent, more drug may buy progressively less. The present data cannot establish where any ceiling sits, or whether a different schedule would move it.

The drug does not distinguish normal tau from pathological tau. ARO-MAPT targets MAPT expression and is designed to reduce production of all six tau isoforms. It does not pick out the aggregated form, and pathological tau is not a separate isoform in any case, since any of the normal ones can misfold. Normal tau does work in healthy neurons, in microtubule regulation and axonal transport. Whether deep suppression can be sustained safely over years is an open question about the target itself, not only about the trial.

The delivery route has its own exposure. Arrowhead’s platform crosses the barrier using a ligand that binds transferrin receptor 1, and the company’s own structural work shows it binding at a site away from where the body’s own transferrin docks. That receptor sits on tissues outside the brain, including developing red blood cells. Other programs using this receptor have had to watch blood counts and iron handling, which is a concern carried over from their experience rather than a finding reported here. This morning’s release reports no persistent anemia, which is the right result. One interim look in healthy volunteers does not settle it.

The biggest one. Clinical benefit from lowering tau remains unconfirmed. Biogen showed slowing on several secondary measures and still missed its primary endpoint. The mechanism is unproven in the way that matters.

10. The line it opens

In chess the bishop is the piece beginners undervalue. It moves only on its own color, it cannot jump over an obstacle, and in a closed position it can sit behind its own pawns for twenty moves without doing useful work. Players learn to prefer the knight, which hops over whatever is in its way.

Then the position opens, and the bishop reaches across the whole board in one move.

Arrowhead has spent years extending its chemistry past the liver, into the lung, into fat, into muscle. The blood-brain barrier was the one position that stayed closed. This morning it opened.

What matters about a long diagonal is not the square the bishop lands on. It is every square that suddenly comes within reach. Tau is one target. The delivery system that carried it is the line.

A bishop is worth very little in a closed position. The whole game changes when the position opens.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

All ARO-MAPT results, including the dose levels, the Day 105 measurements, the safety summary, and the statement regarding regulatory discussions in frontotemporal dementia and progressive supranuclear palsy, are from Arrowhead’s announcement of October 14, 2026. Trial design details are from the ClinicalTrials.gov registration, NCT07221344. The diranersen comparison figures are from Biogen’s announcement of May 14, 2026 and its presentation at the Alzheimer’s Association International Conference on July 14, 2026, and are drawn from a different trial in a different population. The six readings graded in Section 3 were published in Ninety-Eight Percent of Nothing on October 11, 2026. The argument regarding rarer tauopathies was published in The Long Diagonal on September 29, 2026. Cumulative dose figures in Section 4 are calculated from the per-injection doses and schedules in that announcement.

The intrathecal and subcutaneous primate comparison in Section 6, including the regional concentration figures, the twelve-region knockdown range, the cell-type staining and the transferrin receptor mechanism, is from Arrowhead’s TIDES USA 2026 presentation, Systemic RNAi Targeting MAPT, available in the investor section of the company’s site. The counterweight finding of deep-brain target engagement despite low intrathecal accumulation is from Frei and colleagues, Cell Type Distribution of Intrathecal Antisense Oligonucleotide Activity in Deep Brain Regions of Non-Human Primates, 2025. The intrathecal distribution gradient is also described in published primate pharmacology atlases and in autopsy analyses of patients treated with an approved intrathecal antisense drug. No regional distribution data for ARO-MAPT in humans has been reported. The statement regarding planned expansion of the central nervous system pipeline is from Arrowhead’s quarterly earnings call in the spring of 2026. The chess interpretation is the author’s own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

Updated. This note has been revised twice since publication, both times because readers supplied better information than I had. Section 3 now leads with the explanation given by Sacha Schermerhorn, founder and chief executive of Babylon Biosciences, of why the 98 percent figure was close to guaranteed. Section 7 has been corrected: Biogen did publish per-regimen clinical results and a tau range for the Phase 2, which I had treated as unavailable. The correction strengthens the argument rather than weakening it, and the section says so.

1. The number everyone is waiting for

On Wednesday morning, Arrowhead will report the first human data for ARO-MAPT, its brain drug. Somewhere in the presentation there will be a percentage: how much tau the drug removed from the spinal fluid of healthy volunteers. Management has said publicly it wants to see somewhere between 50 and 60 percent.

Within an hour, that number will be on every message board and in every analyst note, and it will be treated the way a test score is treated. Higher is better. A 60 is a triumph, a 45 is a disappointment, and the stock will move accordingly.

I want to make a narrower argument than that. Tau drugs have produced two impressive numbers in the past five years. Both were real and both were earned. In each case the inference the field drew from the number turned out to be wrong, and it was wrong in a different way each time.

This note is about those two mistakes, and what they should change about how you read Wednesday.

Three questions sit behind any drug like this, and they are easy to run together:

1. Did it reach its target?

2. Did it change the pathology?

3. Did it help the patient?

Each of the two trials below answered the first one cleanly and then came apart at a different point. Wednesday’s number speaks to the first question alone. That is not a retreat from what I have written before, which argued that evidence of central nervous system pharmacological activity after a subcutaneous dose would be a major event for this company. I still think so. The narrower point is that one percentage cannot answer three questions.

A bigger number on Wednesday is not a better result. It is a different result, and the difference has to be argued for.

2. The first number: ninety-eight percent

A few years ago, Biogen ran a trial of a drug called gosuranemab in progressive supranuclear palsy, a brutal disease in which tau builds up in the parts of the brain that control movement and balance. The drug was an antibody. It circulated in the fluid around the brain and grabbed loose pieces of tau floating there.

It worked. In the published results, unbound tau fragments in spinal fluid fell by 98 percent. In a field where getting any drug into the brain is the central problem, that is an extraordinary measure of target engagement. The drug found its target and removed essentially all of it.

The disease did not notice. On the rating scale used to track how fast patients decline, the treated group worsened by 10.4 points over a year. The placebo group worsened by 10.6. The p value was 0.85, meaning the difference was well within what chance alone would produce. With nearly 500 participants and almost identical outcomes between the groups, the trial provided no persuasive evidence of clinical benefit. That is not the same as proving the true effect was exactly zero, and it does not need to be. A drug cannot be developed on an effect that large a study cannot see.

The lead investigator published that conclusion himself, and his wording was careful. Neutralizing tau at that particular end of the protein, he wrote, does not translate into clinical benefit. Not that tau is the wrong target. That one way of reaching it fails.

Ninety-eight percent target engagement. A rating scale that moved from 10.6 to 10.4. In most fields, that ends the target.

3. Why it failed, which almost nobody writes about

Start with the number, because it is less impressive than it looks. Sacha Schermerhorn pointed this out after publication and it is the cleanest account of the 98 percent I have seen.

It turns on one word in the trial report. The test measured unbound tau, meaning tau floating free, with no drug attached to it. The drug’s entire job was to attach itself to tau.

The test therefore could not see the tau the drug had grabbed. That tau had not left the body. It simply had a drug stuck to it, and the test only counted the free ones. Give a drug that binds tau, then count tau that is not bound, and the number falls. That is the drug working, not the disease retreating.

It is tighter than even that. The antibody grabs tau at one specific spot near the end of the protein, residues 15 to 22, and the test was looking at that same spot. Drug and ruler were pointed at the same small stretch of the same molecule. A large drop was close to guaranteed before anyone enrolled a patient.

The Alzheimer’s trial shows what happens when you measure a different way. At week 76, the unbound tau figure was down 93 percent. Two other tau measurements, ones that do not care whether a drug is attached, were down 23.8 percent and 15.5 percent. Same patients, same drug, same day. Only the ruler changed.

This is Goodhart’s Law inside a clinical trial. A measure that becomes a target stops being a good measure.

The test counted only the tau floating free. The drug’s job was to grab hold of it.

A cleaner number would not have saved it either, and the reason is in the tissue.

Three patients who received gosuranemab were examined after they died. The antibody had clearly acted. Tau had accumulated inside the disposal compartments of cells wrapped around blood vessels, exactly where a protein ends up after an antibody tags it for removal. The machinery worked.

What had not changed was the pathology. Compared with untreated patients, tangle burden was unchanged in the regions examined. Three brains is a small window, and the finding is an absence of observed reduction rather than proof that no tangle anywhere was cleared. The drug cleaned out one pool of tau and the tangles stayed.

Here is the part that makes it almost painful. In progressive supranuclear palsy, unlike in Alzheimer’s, total tau in spinal fluid is not elevated. The protein that causes the disease sits inside neurons and inside the support cells around them rather than floating in the fluid. The various tau measurements are not interchangeable, so this is a statement about the pool the antibody could reach, not about every form of tau in the body.

The antibody therefore cleared almost all of a pool that turned out to be too small, or too far from the pathology, to change the course of the disease. That is not proof that loose tau plays no part in progressive supranuclear palsy. It is proof that removing it was not enough. The drug was fishing in a nearly empty pond, and it caught almost every fish in it.

The Alzheimer’s trial closes the argument. There the extracellular pool the drug targets is genuinely present, and spinal-fluid tau is elevated. The drug engaged its target again. It failed on the clinical endpoint, failed on every secondary endpoint, and left tau PET unchanged after a year and a half. The company ended the program.

The argument therefore does not depend on a quirk of progressive supranuclear palsy. Clearing loose tau outside the cell did not change the pathology in either disease, including the one where there was plenty of it to clear.

Ninety-eight percent of one pool of tau. No measurable reduction in tangles, in either disease.

4. The obvious fix, and why it is harder than it looks

The natural response is to build a better antibody, one that goes after the misfolded tau rather than the loose fragments. The chief executive of Eli Lilly made exactly this point in an interview last week: tau is essential, he said, and what you really want is the misfolded kind.

That instinct is right, and the execution runs into two walls that have little to do with effort.

The first wall is that a shape is not a sequence. Misfolded tau is not a different protein. It is the same protein, from the same gene, with the same sequence of amino acids, folded into a different three-dimensional shape. Misfolding happens after the protein is built. There is no misfolded tau gene and no misfolded tau message, so a drug that works at the genetic level has no target to aim at. A shape cannot be silenced.

The second wall is geography. Over the past decade, researchers in Cambridge imaged tau filaments pulled from the brains of people who died of different tauopathies. Alzheimer’s folds tau one way. Pick’s disease folds it another. Progressive supranuclear palsy, corticobasal degeneration and the dementia seen in former boxers and footballers each have their own fold. The folds are so distinctive that the field now classifies these diseases by structure rather than by symptoms.

As one of those researchers put it, different folds mean different chemical surfaces. A drug precise enough to recognize the misfolded shape in Alzheimer's has no guarantee of recognizing the misfolded shape in progressive supranuclear palsy, because it is a different surface. Some binding sites may turn out to be shared across diseases. The point is that cross-disease reach becomes a question to prove rather than a property you get for free.

Which produces a result nobody expects. The more precisely you aim at the bad tau, the harder it becomes to claim the rest of the board. Precision is usually free. Here it has to be paid for one disease at a time.

The more precisely you aim at the bad tau, the harder it becomes to claim more than one disease.

5. The other approach, and why it is cruder on purpose

The alternative gives up on telling good tau from bad. It lowers the production of all of it.

The logic rests on how tau pathology spreads. Misfolded tau does not sit still. It recruits normal tau and templates it into the same bad shape, and the pathology travels from region to region along the brain’s wiring. Lower the supply of normal tau and you starve the process of raw material.

If a fire spreads by catching nearby fuel, you can fight the fire or take away the fuel. Fighting the fire requires telling one shape from another. Removing the fuel does not.

It also works in the right place. These drugs act inside the neuron, on the message before the protein exists. That is the same compartment the tangles occupy. The antibodies were working primarily outside the cell, while much of the established pathology sits inside it.

Because every tauopathy is built from the same raw material, one drug has a claim on all of them. The crude approach keeps the whole board.

6. The second number: tangles, reduced

This is where the story gets genuinely good, and then genuinely complicated.

Biogen and Ionis built a drug called diranersen that does exactly this. It lowers the production of all tau. Because it is an antisense drug with no way of reaching the brain on its own, it has to be injected into the spinal canal.

In May, the Phase 2 study read out. Tau PET signal, which images aggregated tau in a living brain, fell in every dose group. The Alzheimer’s Association described this as the first Phase 2 evidence that a treatment can remove tangles from a living human brain, though a drop in imaging signal and the physical disappearance of individual tangles are related measurements rather than the same one. Decline slowed on multiple cognitive measures. Ninety-four percent of participants chose to continue into the extension study.

The trial also missed its primary endpoint.

7. How it missed, which is the whole point of this note

The primary analysis was designed to detect a dose-response relationship in clinical outcomes, not simply to determine whether any treatment group differed from placebo. It used a statistical method that compares the observed results against several pre-specified shapes a dose-response curve might take, measured on a standard dementia rating scale at week 76.

The trial did not establish that relationship. What it did instead is more interesting, which is why the miss is worth a section of its own.

Three regimens were tested over eighteen months: 60 milligrams every six months, 115 milligrams every six months, and 115 milligrams every three months. Against placebo, on the dementia rating scale, decline slowed by 26 percent, 14 percent and 9 percent respectively. On a cognitive test, 42, 32 and 29 percent. The lowest dose produced the largest effect on five of six measures, and most of its differences reached nominal significance. Biogen’s own language is that higher doses were not associated with greater slowing of decline.

Now put the biomarker beside it. Across all three regimens, spinal-fluid total tau fell by 50 to 65 percent. The company reported that as a range across the doses rather than a figure for each, so the exact pairing is not public. What is public is the shape: the biomarker moved within a narrow band while the clinical benefit nearly tripled in the opposite direction.

Tau fell across a band of fifteen points. Clinical benefit tripled as the dose came down.

That is not a study that could not test the relationship. That is a study that tested it and found it running backwards.

I want to be careful about what that does and does not prove, because the careless version of this is wrong. This is one Phase 2 with three arms and sixty patients in the group that did best. Most of its clinical differences were nominally significant rather than formally so, meaning they were not corrected for testing several endpoints at once. A larger trial could show a perfectly orderly relationship, and the inverse pattern could be an artifact of how the arms happened to fall out.

What it establishes is narrower and harder to argue with: across the doses tested, higher doses were not associated with greater clinical benefit. The relationship everyone assumes exists was examined directly in the only randomized trial of this mechanism, and the data came back pointing the other way.

Figure 1. Two biomarker successes with different limits.

8. What this changes about Wednesday

Put the two together. The first number was 98 percent, and it bought no benefit because clearing that pool was not enough to alter the course of the disease. The second number showed tau pathology genuinely falling on brain scans and decline slowing at every dose, and the clinical benefit still ran backwards against dose. Why is unsettled. The biomarker moved within a narrow band while the clinical effect did not.

On Wednesday a third number arrives. It will measure how much tau left the spinal fluid of healthy volunteers who do not have the disease, do not have tangles, and will not be followed for cognitive decline.

Figure 2. Compiled from the published trial results described above. The ARO-MAPT column describes what the study can and cannot measure, not a prediction.

Set the three programs side by side and the shape of the problem is visible. Two of them have been tried in people and neither settled the question. The third changes one variable, and it is the one nobody has solved.

I am not saying the number is meaningless. It is the first meaningful test of whether a shot under the skin can produce tau lowering in the human central nervous system, and that is a genuinely large question nobody has answered. It is the right question for this stage of development.

I am saying the number answers the first of the three questions and leaves the other two alone. It tells you a shot under the skin can lower tau in the human central nervous system. It does not tell you how widely the drug reached the neurons that matter, how much pathology would change, or whether a patient would be better off. A 60 would demonstrate greater tau suppression than a 50, and that is a real pharmacological result. It would not establish greater clinical benefit.

One corollary is worth stating plainly, because it cuts against the way this will be covered. ARO-MAPT could fail as an Alzheimer’s drug and still matter enormously to Arrowhead. If a shot under the skin produces real pharmacological activity in the human central nervous system, that result belongs to the delivery system rather than to tau, and the delivery system is what the company has spent a decade building. Whether it carries to other brain targets would still have to be shown, one target at a time. The competition sharpens the point.

Novartis has an antisense drug against the same gene, NIO752. In a Phase 1 trial in progressive supranuclear palsy, spinal-fluid neurofilament light held steady in treated patients while it climbed on placebo. That was presented at a conference rather than published, and I have not seen whether it was a prespecified endpoint or an exploratory one. A Phase 3 is planned, with a motor rating scale as the primary measure. It is delivered into the spinal canal. The point is that the two questions can be answered differently, and most coverage will treat them as one.

The target is not what separates these programs. The needle is.

The drug could fail and the delivery system still succeed. Most coverage on Wednesday will treat those as the same result.

9. Six readings worth more than the percentage

If the headline number carries less information than it appears to, the rest of the presentation carries more. Here is what I will be reading.

Figure 3. Author’s list, drawn from the trial registration, the primate data Arrowhead has published, and the two trials described above.

The study day matters more than most people realize. Tau is a long-lived protein and the pool turns over slowly. In Arrowhead’s primate work, knockdown took three to four months to reach its floor, though an animal timeline is a hypothesis about humans rather than a schedule. If Wednesday’s sample was drawn before the effect reached its maximum, the reported figure understates where the drug was heading. A 50 percent reduction on a curve still descending may be more informative than a 60 percent reduction that has already flattened.

The dose question is the direct lesson from the Phase 2 above, and it is worth setting out the possibilities before the data arrives rather than improvising at eleven in the morning.

The strong version. Cohorts separate on dose, the reduction holds at a later study day, and safety is clean. That is strong evidence the delivery system works, and the best case available from a study of this design.

The ambiguous version, which will be misread. Cohorts converge on a similar reduction regardless of dose, with clean safety. That looks like a disappointment and may not be one. It resembles the pattern the antisense data showed, where the biomarker moved within a narrow band across a wide dose range, and it would still establish pharmacological activity in the central nervous system from a subcutaneous dose.

The bad version. A small reduction, or one that fades by the later sampling points. That raises a specific question about whether the primate results translate, or about whether the dosing schedule is right, and it is the version that would actually matter.

The middle case is the one to prepare for. It will look like a miss and may be the same ceiling the antisense drug approached.

10. Why the speakers tell you more than the slides

One more observation, and it is the kind that has served this newsletter well before.

Arrowhead did not book a generic expert for Wednesday. It booked Adam Boxer of the University of California, San Francisco. Boxer ran the trial described in section two. He published its failure and wrote the sentence concluding that neutralizing tau at the N-terminus does not work. He also leads the research program that established spinal-fluid neurofilament light as a measure of how fast these diseases progress, and he runs two government-funded platform trials where the next generation of tau drugs will be tested.

When the first results from the antisense approach appeared, Boxer called it a landmark study of human tau therapeutics, the first to target tau at the genetic level. The man who published the obituary for one approach called the other one a landmark.

His talk on Wednesday is titled Silencing MAPT in Alzheimer’s Disease and other Tauopathies. The last three words are the ones to notice.

In November, the full scientific data goes to a different conference, presented by Catherine Mummery of University College London. Mummery was first author on the first human study of the antisense drug and presented its Phase 2 results in July.

Between them, these two people ran the defining tau-lowering programs in the world. Their participation says a good deal about how seriously the field takes this drug. It does not establish that the drug works, that a particular indication is coming, or that any specific plan exists. Credentials are evidence of standing, not of outcome. Still, a company that books these two speakers has made a choice, and the choice is legible.

Arrowhead did not book a generic expert. It booked the man who buried the alternative.

11. What would make me wrong

The lowest-dose finding may not replicate. The dose-response result came from one Phase 2 with three arms. Statistical noise explains a great deal in studies that size, and a larger trial could show a perfectly orderly relationship between dose and benefit. If it does, the central argument here weakens considerably.

Antibodies are not dead. The first generation of antibodies aimed at the wrong end of the protein. A second generation aimed at the middle, and one of them slowed the rate of tau accumulation on brain scans by roughly a third to nearly 60 percent, with cognitive signals in some subgroups. It still missed its primary endpoint, and so did a competitor, but the selective approach is not finished. It may have needed a better target, not a different philosophy.

The crude approach has not won either. No tau-lowering drug has yet produced an unambiguous clinical win. Diranersen slowed decline and missed its primary. That is a result worth having, and it is not the same as proving that lowering tau helps people.

The seeding model could be wrong. The argument in section five depends on misfolded tau recruiting normal tau. If pathology propagates some other way, lowering the supply accomplishes little and only clearance matters.

Somebody may solve the selectivity problem. Targeted degraders, which hand misfolded proteins to the cell’s own disposal machinery, could eventually crack conformational selectivity from the other direction. They are early and the brain-penetration problem is real, but the geography argument in section four assumes nobody cracks it.

Lowering tau is not free. Reducing a protein the brain normally makes, by half, for years, is untested. Animals missing tau entirely do reasonably well, which is reassuring, but the data are not perfectly consistent and no human has lived decades with less of it.

Receptor-mediated delivery carries its own questions. Carrying a drug into the brain by binding a receptor on the blood-brain barrier raises questions outside the brain, because those receptors are rarely confined to it. Depending on which receptor a program uses and how the molecule is built, effects on blood counts and iron handling have been a live concern for others working in this area. This note does not establish that the same applies here. A clean tau number alongside a blood-count signal would be a different story than it first appears.

12. The third number

There is a habit in chess worth borrowing here. Beginners count material. A queen is nine, a rook five, a bishop three, and the position gets judged by adding. Stronger players know the count is only the beginning. They read where the pieces sit, whether the king is safe, which files are open, who moves first. Counting is easy and reading is hard, and reading is what decides the game. A player can be a piece up and completely lost.

Ninety-eight percent was a material count. The position never changed.

The field has been handed two impressive tau numbers in five years. One cleared almost all of a pool, and clearing it was not enough to alter the course of the disease. The other lowered tau pathology in living brains and slowed decline across every dose, and then produced its largest clinical effect at its smallest dose.

Both drugs did what they promised. Both numbers were real. Neither number carried the information the field read into it, and in each case the reason was only obvious afterward.

On Wednesday, a percentage will appear on a slide and a great many people will decide what it means before lunch. The honest position is that if the number shows meaningful tau lowering, it establishes pharmacological activity in the human central nervous system from a shot under the skin, and that alone. If it does not, it raises a specific question about translation or dosing. Either way, that is the question Arrowhead has spent years and a great deal of money trying to answer, and nobody has answered it before.

The rest is for next year, when the patient cohorts can measure tau on a brain scan, track cognition, watch the nerve-damage marker, and run long enough for any of them to move. None of that is available from healthy volunteers, by design.

Two numbers, two wrong inferences. The third tells you whether a shot under the skin can produce meaningful tau lowering in the human central nervous system, and no more than that. It is still the only question worth asking this week.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. The characterization of what is and is not in the published literature, the probability judgments, and the chess analogy are the author’s own.

The gosuranemab results, including the 98 percent reduction in unbound N-terminal tau and the PSP Rating Scale outcome, are from the Phase 2 trial published in Nature Medicine in 2021. The postmortem findings in three treated patients are from a separate 2021 neuropathology report. The observation that spinal-fluid tau is not elevated in progressive supranuclear palsy appears in published reviews of tau-targeted therapy.

The cryo-EM filament structures and the structure-based classification of tauopathies are from work published by the Medical Research Council Laboratory of Molecular Biology between 2017 and 2021. Diranersen results are from Biogen’s topline announcement of May 14, 2026, the detailed presentation at the Alzheimer’s Association International Conference on July 14, 2026, and the Alzheimer’s Association summary of that presentation. Biogen reported mean spinal-fluid total tau reductions of 50 to 65 percent across all studied doses in the Phase 2, and reported that range across the doses rather than a value for each. The per-regimen clinical figures are from the same announcement.

The earlier Phase 1b tau PET findings are from Biogen’s 2023 conference disclosures and the subsequent publication in Nature Aging. Bepranemab results are from UCB’s Phase 2 presentations in 2024 and 2025. Adam Boxer’s role, publications, and research programs are from his university profile, the published trial, his comments to Alzforum, and public trial registrations.

The NIO752 Phase 1 neurofilament result and Phase 3 plan are from Alzforum’s conference coverage of the program in 2025, which reproduced the neurofilament figure courtesy of Gunter Hoglinger. I have not located a peer-reviewed publication of that result. The ARO-MAPT trial design and endpoints are from its ClinicalTrials.gov registration. The gosuranemab binding site at tau residues 15 to 22, and the week 76 values for unbound tau, phosphorylated tau and total tau from the Alzheimer’s trial, were supplied by Sacha Schermerhorn of Babylon Biosciences and are credited in Section 3. The interpretation throughout is the author’s own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. A week before the brain data

On October 7, Arrowhead announced that the FDA had granted plozasiran entry into a program called STAR, for its coming filing in severe hypertriglyceridemia. Entry means the filing qualifies for a six-month priority review instead of a ten-month standard one. The agency also says it will generally aim to act at least a month earlier than that, though it calls this a target rather than a promise. The filing is on track for the end of this year.

Most readers will file this under good housekeeping and move on to October 14. I want to spend a few pages on it, because three facts sit underneath it that the headline leaves out.

The first is that almost nobody uses this program. The second is that Arrowhead had already paid for the speed it just received for free. The third is that the skill on display here is the same skill the company is about to need for a far harder ask.

One clarification before the pull quote, because the easy version of this story is wrong. Plozasiran already held breakthrough designation in this indication, which points toward a six-month review on its own. The company bought a voucher over the summer to be certain it had that option. STAR then delivered the same six-month clock without spending the voucher, and added two advantages the designation does not: a review that starts on an incomplete package, and a stated aim of acting a month inside the priority date. The two hundred and fifteen million dollars is the unspent voucher. The months are a smaller and separate gain, and Section 8 argues about how small.

The company paid two hundred and fifteen million dollars for a six-month review. Nine weeks later it qualified for one anyway.

2. What STAR actually is

The name is Split Real Time Application Review. The FDA created it in October 2022 under the funding agreement that governs how the agency reviews drugs, and it runs through September 2027.

The mechanics are the interesting part. Normally a company assembles its entire application and sends it in, and the review clock starts when the package is complete. Under STAR, the company sends the application in two pieces. Part one contains the whole application except the final clinical study reports and the integrated summaries. Part two, with those documents, follows roughly two months later. The agency starts reading while the company is still finishing.

The idea is borrowed from a program the cancer division has used for years. Read the parts as they arrive rather than waiting for the whole.

One eligibility rule deserves its own sentence, because almost nobody will notice it. To enter, a sponsor must attest that no foreign manufacturing site is involved in the supplement, and the agency verifies that attestation when part one arrives. If a foreign site turns up, the application loses its place. The reason is practical: a foreign inspection takes time the split review was meant to save.

Figure 1. Review times from the FDA's program description and Arrowhead's announcement. The five-month bar is a target rather than a commitment.

3. The lane nobody drove in

Here is where the story stops being routine. The FDA published an assessment of its own pilot earlier this year, and the language is unusually blunt for a regulator describing its own work.

Here are the numbers, published by the agency in January. Across the first three years, through September 2025, the FDA received six requests to enter the program. All six went to the drug center. None went to the biologics center. One sponsor withdrew. The other five were denied. Zero were accepted.

The agency’s own summary of that record: it received no applications in the pilot program as of the end of the 2025 fiscal year. It had committed to holding a public workshop on the program by March 2026, and it cancelled the workshop, on the grounds that it had no experience to discuss. It also ran a separate experiment inviting original applications to use the same split approach, and received no formal requests at all.

Read that again. The agency built a faster lane, kept it open for three years, advertised it in webinars and at industry conferences, and had to cancel the meeting it had promised to hold about the lane, because nobody had driven in it.

Arrowhead’s grant came a year after that assessment window closed, so I cannot call it the first entry ever. The agency has not published a tally for the year since. What I can say is that the pilot spent its first three years at zero, and that this is the only public entry I can find since.

The reasons for the five denials are listed too, and the third one matters for what follows. Applications were turned away for failing the substantial improvement bar, for not addressing a serious unmet need, or because some part of the proposed application would have required a longer review anyway, which would have defeated the purpose of splitting it.

4. Why an empty fast lane stays empty

The obvious question is why a company would decline free speed. The answer is in the mechanics, and it is the most useful idea in this note.

To file in two parts, a company has to finish the whole application except the final study reports two to three months earlier than it otherwise would. The data have to be clean. The manufacturing section has to be done. The proposed label has to be written. The safety summaries have to be nearly complete. All of it has to be good enough to put in front of a reviewer who will start forming opinions immediately, before the company has had a chance to polish the final narrative.

Most organizations cannot do that, and the ones that can often do not want to. A filing is usually a scramble that ends the night before submission. A program that demands the scramble finish a quarter early is not free at all. It is paid for in organizational discipline, months in advance, by a team that has to work differently from how most teams work.

The foreign site rule belongs in the same column, and it screens on a constraint a company cannot fix in a quarter. A great deal of the world’s drug substance is made outside the United States. A sponsor in that position cannot opt in at all, however good the trial was.

Arrowhead built a drug substance plant in Verona, Wisconsin, announced at two hundred to two hundred and fifty million dollars, years before this program mattered to anybody. The rule does not require that plant. It requires that the supplement involve no foreign site, which any domestic manufacturer would satisfy. I have not confirmed which site supplies this filing, so I will not claim the factory bought the entry. What I will say is that a company making drug substance in the United States has one fewer problem to solve here.

The agency’s third reason for denial says the rest of it. A package that would have needed a longer review anyway could not be rescued by splitting it.

The price of admission is not the FDA’s goodwill. It is being finished early.

There is a second implication, and it is worth more than the four months. To hand a regulator the whole application minus the final study reports, a company has to have the trial finished, the database locked and the results settled well before the filing date. The readiness is not a claim about the schedule. It is evidence about it. A company nowhere near a year-end filing would not have been in a position to ask.

The toll on the fast lane is paid in preparation, which is why the lane is empty.

5. The increment

Chess clocks used to be simple. Each player got a fixed allowance, and when it ran out, the game was over regardless of the position on the board.

Modern clocks add an increment. After every move, a few seconds are added back to your clock. Play promptly and you can finish a long game with more time than you started. Dither, and the increment never catches up with what you are burning. The increment does not care how much money you have. It only responds to how you play.

That is the right way to read this announcement. Arrowhead did not buy the time it gained. It was handed time back for having the application far enough along that the agency could start reading it in pieces. Section 8 argues about how many months that is really worth. The mechanism is the same either way.

Preparation is the only currency the clock recognizes.

6. The voucher it no longer needs

At the end of July, Arrowhead agreed to buy a priority review voucher for $215 million, and announced it on August 4. These are transferable tickets that let a company move one application to the front of the queue. The government hands them out as a reward for developing drugs in neglected areas, and they can be sold on.

The company said at the time exactly what the voucher was for: the plozasiran filing in severe hypertriglyceridemia, expected before the end of the year. Nine weeks later, plozasiran qualified through STAR for the same six-month path, and the voucher was no longer needed for the application it had been bought to accelerate. Chris Anzalone said as much in this week’s announcement, noting that the company can now redeploy it or sell it.

That is not a rounding error. Voucher prices have roughly doubled in a few years, and 2026 sales have run between about 180 and 210 million dollars. On the same day as the Arrowhead announcement, Ultragenyx agreed to sell one for 210 million.

The asset is therefore worth roughly what Arrowhead paid for it in July, give or take the spread on a market that prints every other month. The $215 million is not sunk. It is parked.

A voucher can be spent once, and only on a marketing application. The brain program is years away from one, so nobody should read this as a voucher being saved for ARO-MAPT. What Arrowhead holds is simpler: an asset it can turn into cash at any time, or keep for a filing that has not been written yet.

The choice between those two looks different after the fourteenth than it does today. If the company has just shown it can reach the human brain, it has more use for optionality. If it has not, it may prefer the money.

7. The sentence in the criteria

There is one more line worth reading closely. To grant entry, the FDA must determine that clinical evidence from adequate and well-controlled studies indicates the drug may demonstrate substantial improvement on a clinically relevant endpoint over available therapies.

Available therapies in severe hypertriglyceridemia include fibrates, prescription fish oils, and now Ionis’s olezarsen, which reached the indication first. Arrowhead arrived second.

I want to be careful about how much weight that carries. It is a threshold for entering a program, not a finding of superiority. The FDA also states plainly that acceptance guarantees neither the expedited action nor an approval. What entry does secure is the six-month clock. The agency has also said that real-world evidence cannot be used to clear this bar, so what persuaded it came from the controlled trials.

Still, a regulator looked at the plozasiran data and concluded it may substantially improve on what is already available. In the indication where this company got there second, that is the most interesting sentence in the release.

The wording is worth one more beat. The same phrase, substantial improvement over available therapies, is the standard behind breakthrough designation, which plozasiran already holds in this indication. The agency used it once before olezarsen was approved for severe hypertriglyceridemia, and has now used it again, for entry, with that drug on the market. The bar is a criterion for entry and not a ranking. It was applied the second time with a rival APOC-III drug already carrying the label.

The bar was applied a second time, with the competitor already approved.

8. Why this is worth less than it looks

Three reasons, and they belong before the bullish part rather than after it.

A buyer will not pay for a regulatory team. Regulatory affairs is the one capability the large drugmakers have in abundance. Lilly, Roche and Novartis employ hundreds of specialists who do only this work and have dealt with the agency for decades. In an acquisition, a regulatory department is among the most redundant assets on the books. Nobody pays a premium for one. Note the scope of that objection, though. It applies only if the company is sold. If Arrowhead stays independent, the capability is not redundant to anybody, and the next section is about what it is worth in that case.

The criterion judges the data. A superb regulatory group cannot get a mediocre drug into STAR. The criterion is about what the trials showed. Most of this signal is about plozasiran, not about the people who filed the paperwork.

The gain may be smaller than ten minus six. A published study of supplemental approvals found the median review already ran about six months. Plozasiran also already held breakthrough designation in this indication, which points toward priority review on its own. The step from ten months to six may overstate what STAR adds on top of what this drug would likely have received anyway. The genuinely incremental piece is the expedited action target and the earlier start.

9. Why it is worth more than it looks

The tempting conclusion is that Arrowhead has a good relationship with the FDA. That is the lazy version of this argument, and I do not think it is the right one. Agencies do not grant clocks out of affection.

The valuable asset is not a relationship. It is a position, and the difference matters because one of them survives a transaction and the other may not.

A regulatory team can leave, be reorganized, or be made redundant. A regulatory position attaches to the drug. Designations, agreed endpoints, an agency that has already read and accepted a package, a clock that is already shorter: those travel with the asset through any deal, and they convert directly into earlier revenue. Four or five months of a launch is a number that goes in a model.

Figure 2. Compiled from Arrowhead’s announcements.

For a platform company the pattern compounds, and that is the part the market has no good way to price. Each successful interaction leaves behind pieces that get used again: manufacturing experience, validated assays, an agency that already understands how this chemistry behaves, and a working sense of what the reviewers will ask next. Scientific precedent in one tissue does not transfer to another, and nobody should pretend it does. The organizational learning does transfer. A company that intends to file drugs in the liver, the lung, fat, muscle and the brain gets that benefit on every one of those filings, not once.

Count what plozasiran has collected. On the designation side: American orphan status, fast track, breakthrough designation in familial chylomicronemia syndrome, European orphan status, and breakthrough designation again in severe hypertriglyceridemia. On the approval side: the United States, the European Union, China, Australia and Canada. STAR is the newest entry on that list rather than the final item in a tidy series. The number is less interesting than the run. None of these came from a single good quarter.

What if nobody buys the company? All of which assumes a sale. Suppose there is not one, which is what the company says it expects. Then the capability stops being redundant to anyone and starts belonging entirely to Arrowhead’s own shareholders, which is a better outcome, not a worse one.

Start with the competitor, and be honest about the size of the gap. Ionis won the severe hypertriglyceridemia label on June 24 of this year and had the drug available in July. Arrowhead files at year end. Even on a six-month clock, an answer lands around the middle of next year. The lead is roughly a year, and most of it is already banked before Arrowhead’s file reaches the agency. Four or five months off the review does not close that. It shortens the tail of it.

Shortening the tail still counts. In a market where an incumbent builds prescriber habit month by month, every month the follower is absent is a month of habit it has to break later. Time to the label is not a soft benefit. It can become share, at the margin, which is where share in these markets is usually won.

Then consider the pipeline. Arrowhead owns key programs outright, including in obesity and liver disease, and its wider pipeline spans the liver, the lung, fat, muscle and the brain. Each of those filings gets the benefit of whatever this team has learned, and none of it is shared with a partner. An acquirer would absorb the capability into a department that already has one. Independence keeps it concentrated in a company where it is scarce.

The clock survives a sale. The team only survives if there is not one.

10. The harder ask

Here is why I think this matters more than a routine press release should, and triglycerides have no part in it.

In a note two weeks ago I argued that Arrowhead’s brain drug may find its fastest path through rare tauopathies rather than Alzheimer’s, and that the route runs through a blood marker of nerve damage called NfL. That argument requires a difficult concession. It needs the FDA to accept a biomarker it has never accepted for a tau disease, in a tiny population, as evidence that a drug is reasonably likely to help.

That is not a filing. That is a negotiation, conducted over years, with a review division weighing an argument no regulator has accepted before. Evidentiary standards do not bend for favored sponsors, and nobody should suggest they do. A company whose last several submissions were clean, early and accurate does walk into that conversation with regulatory credibility already established.

A STAR entry proves no part of that. It is one small piece of evidence that this is the kind of sponsor that shows up with the work done. Seven days before a readout that could make the hard question live, I find that worth noticing.

Finishing a package early and persuading a division to accept a new biomarker are different muscles. This week tested the first one.

11. What would make me wrong

The filing could slip. A filing is on schedule until it is not. Every argument here assumes the sNDA goes in by year end.

Entry is not approval. The FDA says in its own materials that entry guarantees neither an expedited review nor an approval. The clock can be shortened and the answer can still be no.

The lane may be less empty than it looks. If several other companies turn out to have entered quietly, the scarcity argument weakens considerably, and this becomes a competent filing rather than an unusual one.

Good paperwork is not good science. Regulatory fluency and scientific judgment are separate skills. A company can be excellent at filings and still be wrong about which drugs to file.

A faster label is not a faster launch. None of this changes the commercial ramp. Chris called the expansion a slow burn, and a shorter review moves the starting line without changing the slope of what follows.

Time to label may not convert to share. Prescribers switch patients when a drug is better, not only when it is first, and Ionis has most of its lead banked already.

The expedited action may not happen. The earlier action is a target the agency aims at, not a goal date it owes. If the review simply lands on the six-month date, the part STAR uniquely added is the earlier start of reading, not a shorter clock, and the whole gain is smaller than this note implies.

12. Seconds, not dollars

An increment cannot be purchased. You can buy a better computer, a bigger team, a stronger opening repertoire. You cannot buy the seconds that come back after a good move. Those are earned by playing well, one move at a time, and they accumulate quietly until someone looks at the clock late in the game and notices that one player has more time than the other.

Notice also when the work gets done. Nobody at Arrowhead knew in the spring that the FDA would grant this. The application was being finished early anyway, by people who had no way of knowing it would pay. That is the uncomfortable part of an increment. You cannot decide to be ready once the opportunity shows up. The seconds you collect were bought months earlier, in rooms where nobody was watching.

Arrowhead bought a voucher over the summer to be sure of a six-month review. In October it qualified for one anyway, after the drug cleared the bar and the work was far enough along to be read in pieces. The voucher never got used, and the earlier start came on top.

The data arrives on October 14. Whatever the number says, Arrowhead walks into that position with months off one regulatory clock and two hundred and fifteen million dollars of unspent acceleration sitting in a drawer.

You cannot buy an increment. You pay for it in advance, with work nobody watches, and collect it on a clock nobody checks.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

The STAR entry, review timelines, filing schedule, voucher status and Chris Anzalone’s quotation are from Arrowhead’s press release of October 7, 2026. The program’s launch date, two-part submission mechanics, entry criteria, exclusion of real-world evidence, and the statement that acceptance guarantees neither expedited review nor approval are from the FDA’s STAR program materials and from Regulatory Focus reporting of the FDA’s program description.

The entry request counts, the denial reasons, the cancelled public workshop, the separate pilot for original applications and the agency’s inability to assess the program are from the FDA’s published STAR Pilot Program Assessment, covering October 2022 through September 30, 2025 and posted in January 2026. The eligibility criterion barring foreign manufacturing sites, and the attestation verified at the part one submission, are from the FDA’s STAR program webpage. The Verona facility figures are from Arrowhead’s own announcements and from state economic development reporting at the time of construction.

The olezarsen approval date and United States availability are from Ionis’s announcement of June 24, 2026. The voucher purchase price of $215 million, the July 31, 2026 agreement date and the stated intention to use it for the severe hypertriglyceridemia filing are from Arrowhead’s announcement of August 4, 2026 and the subsequent events note to its quarterly report for the period ended June 30, 2026. Other voucher prices are from 2026 transaction announcements, including Ultragenyx’s agreement announced October 7, 2026, and from Endpoints News reporting on the voucher market. The median review time for supplemental approvals is from a 2023 analysis published in JAMA Network Open. The designations and approvals listed are from Arrowhead’s announcements. The chess analogy and the interpretation throughout are the author’s own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The second wave

A new siRNA company seems to show up every month. One Chinese brokerage counted more than thirty small nucleic acid deals worldwide in 2025, worth more than thirty-six billion dollars, and Chinese biotechs have become one of the main sources of them. Argo Biopharma of Shanghai struck a pact with Novartis with a headline value of $5.2 billion.

SanegeneBio, founded in 2021 with labs in Boston and China, signed with Lilly for up to $1.2 billion, then with Genentech in February for $200 million upfront plus milestones of up to $1.5 billion. Big pharma, which once walked away from this field, is writing checks to companies that did not exist five years ago. Call them the second wave: the siRNA companies founded since about 2020, many of them Chinese, building on the published playbook. Avidity and Dyne, which appear later, came earlier and are not part of it.

Readers keep asking me whether these newcomers threaten Arrowhead and Alnylam. The honest answer has two halves. In the liver, the newcomers are real competitors, and I will explain why. Beyond the liver, and in the brain above all, they are about to play a game whose middlegame nobody has written down.

A quick word on what these companies make. An siRNA is a short strand of genetic material that tells a cell to stop making one protein. Much of the core chemistry of the strand itself is well understood. The hard part has always been delivery: getting the strand into the right cells, in the right tissue, in enough quantity to matter, without doing harm along the way.

The newcomers can buy or hire the answers to yesterday’s problem. The next problem has no answer key.

The gap between knowledge that can be copied and knowledge that has to be earned is the whole argument of this paper.

2. The first wave

This has happened before, and how it ended is the most useful lesson in this paper.

Between 2006 and 2010, the biggest companies in medicine decided siRNA was the future and bought in. Merck paid $1.1 billion for Sirna Therapeutics in 2006. Roche paid Alnylam $331 million in 2007. Novartis partnered with Alnylam. Pfizer, Abbott and others built programs of their own.

Then they left, one after another. Roche shut down its RNAi research in 2010 after spending roughly half a billion dollars on it over three years. Novartis declined to extend its Alnylam partnership the same year, triggering layoffs at Alnylam. Pfizer closed its 100-person RNAi unit in early 2011, and Abbott quit too. Merck sold Sirna to Alnylam in 2014 for $175 million, about one-sixth of what it had paid, explaining that it wanted less emphasis on platform technologies. Novartis exited the field that same year, citing ongoing challenges with formulation and delivery, and in 2015 sold what it had kept to a small company in Pasadena called Arrowhead.

Figure 1. Sources: company filings and contemporaneous reports. Dates are the public announcements.

Here is the part that matters most. The first wave did not fail somewhere exotic. It failed at the liver. Early trials were confounded by inflammation from the fatty particles used to carry the drugs, and it was hard to even prove the effects came from silencing the target. The science was getting better fast. Pfizer’s own former chief scientist for its RNAi unit said liver knockdown improved about a hundredfold between 2008 and 2011. The companies left anyway, because the improvement was not arriving on the clock their boards were watching.

Alnylam’s president at the time called big pharma a miserable barometer of new platforms, and pointed out that it had made the same mistake with antibodies a generation earlier. He was right. By 2019, Novartis, Regeneron and Roche were all buying back in.

The last time everyone entered this field, the entrants were the richest companies in medicine. Nearly all of them left, and the problem that beat them was the liver.

3. What it took to survive

The two survivors did not win because they never failed. They won because they failed at the worst possible moment and kept going.

In October 2016, Alnylam halted revusiran, its second most advanced drug, after 18 patients died in a 206-person trial and the data showed more deaths on the drug than on placebo. The stock fell 48.5 percent in a day. A month earlier, it had stopped another program after liver enzyme rises in healthy volunteers.

Seven weeks later, Arrowhead discontinued all three of its clinical programs. Its intravenous delivery vehicle had caused deaths in a primate toxicology study, at doses higher than those planned for people, and regulators made clear the questions would take a year and a half or more to answer. The stock fell more than 65 percent in a day.

Both companies came back. Alnylam’s first drug was approved in 2018, sixteen years after the company was founded. Arrowhead already had a new system, given by a shot under the skin, working in animals at lower doses with good safety margins. That system, the one it uses today, reached the clinic within about two years.

What each company carried out of 2016 was not a drug. It was a record: which sequences caused trouble, which chemical changes made the difference, what the monkeys showed before the people did, and how to read early warning signs the next time. Fragments of that record show up in papers and patents. The complete failure history does not, and it is the main asset a newcomer cannot buy.

The survivors are not the ones who never failed. They are the ones who failed in 2016 and kept the notes.

4. In the book: the liver is open

Chess players call published opening theory the book. Anyone can study it, and strong players know it cold. A decade ago the liver was out of book. Today it is in it.

The reason is a sugar tag called GalNAc, which steers an siRNA to liver cells, plus a set of chemical changes that let the drug survive in the body for months. Alnylam worked that out over years of trial and error. The methods have been published for a long time, the core patents are aging, and the rest of the field has learned them. A well-funded newcomer with experienced hires can now build a competent liver drug far faster than the pioneers could. The deal flow shows it. The Argo programs Novartis licensed are liver-delivered drugs for heart and metabolic disease, and some are already in mid-stage trials.

The clearest American example went public in late September. ADARx was founded in San Diego in December 2019 and is run by Arrowhead veterans. Its chief executive, Zhen Li, worked on siRNA chemistry at Merck and later ran chemistry and nonclinical development at Arrowhead. Its chief technology officer, Rui Zhu, was a key inventor on the programs that became Arrowhead's Redemplo.

The company's name comes from ADAR, the enzyme behind RNA editing, and in its early years it described a platform for silencing, degrading and editing RNA. By the time of its IPO, its filing described a next-generation siRNA company, and its three drugs in human trials all silence genes in the liver. It raised $446 million in an upsized offering, after AbbVie paid $335 million upfront for options on its next drugs. Its programs beyond the liver are still preclinical, and its brain program is not expected to enter a first trial until late 2027 or early 2028.

The company has not said why its emphasis changed, and I will not guess. The record shows what reached the clinic first. Veterans carry the book with them, and what they build first is what is already in it.

The pattern holds at the very top of the field. John Maraganore, who built Alnylam, co-founded City Therapeutics in 2024 to take RNAi beyond the liver. Its first two drugs, one already in human trials and one about to start, both work through the liver.

I want to be straight forward about what that means for Arrowhead, because it is not comfortable. The company’s approved drug and most of its near-term revenue come from the liver. Its nearest competitor there today is Ionis’s Tryngolza, not a Chinese copy, and Chinese liver drugs are still years from an American or European label. When they arrive, they will press on prices and on what Arrowhead can charge partners for liver programs. That pressure is coming. It is not here yet, and any paper that pretends it will never come is not worth your time..

The liver was the moat. It is becoming the commons. The moat did not disappear. It moved.

Notice the shape of the history, though. The first wave failed at the liver because the liver was out of book. The newcomers can succeed at the liver because they arrived after the book was written. Beyond the liver, they face a problem structurally similar to the one Merck and Roche faced in 2008.

The important point, then, is not that Arrowhead and Alnylam still own the liver. Increasingly, they do not. It is also not that every tissue beyond the liver is equally unknown. The board now has five parts. The liver is becoming a commons.

Muscle already has an opening, written by someone other than Arrowhead or Alnylam. Fat cells have an early opening, first sketched by Arrowhead. The lung has a more developed one, also Arrowhead’s, reached by inhaler. The brain has no book yet for anyone, the incumbents included, and neither do the eye or the heart. The advantage has moved toward the parts of the board where the field has not standardized the answer: getting into new tissues, knowing early when a program is going wrong, and having the history to prove a chemistry is safe.

Patients, the barrier that grows after success. This one does not fade as the book gets written. It grows. Once a drug is approved for a disease, the next one has a harder time finding patients. Doctors and ethics boards are reluctant to give a placebo to patients who could take a drug that works, so later trials usually test the new drug on top of the approved one, or against it. That takes more patients, a bigger effect and more time. In rare diseases, the few centers that see these patients are often already running the first company’s trials. The chemistry of the liver is a commons. The patients are not.

This barrier does not favor incumbents. It favors whoever is first in a disease, and Arrowhead knows it from the losing side, following Ionis’s Tryngolza into severe hypertriglyceridemia. For the second wave, it compounds the problem. Many of the liver genes it is going after, such as PCSK9, APOC3 and Lp(a), already have approved drugs or late-stage rivals. A newcomer can build a competent liver drug quickly and still spend years filling the trial that proves it. China offers a partial way around, with large numbers of untreated patients and fast trial startup, but American regulators have been reluctant to approve drugs on Chinese data alone, so the global trial still has to be run.

The same logic runs in reverse in the brain. No drug is approved to slow PSP or inherited frontotemporal dementia. Whoever gets there first will not just win the first approval. It will make every trial after it harder to fill.

5. Out of the book: five barriers

The second wave is not entering the field to stay in the liver. Its ambitions extend to the brain, muscle, fat and lung, because those are among the major remaining frontiers. Those tissues are not equally open, though, and the incumbents are not only Arrowhead and Alnylam. Ionis has decades of experience with antisense, a cousin technology, in many of the same tissues, and Novo Nordisk owns Dicerna.

Muscle already has an opening. Avidity, now part of Novartis, delivered siRNA into human muscle with an antibody. Dyne uses a similar antibody to carry exon-skipping drugs into muscle, and the FDA accepted its first application in July, with a decision due January 21, 2027. A newcomer that reaches muscle in the next few years will mostly be learning an opening Avidity and Dyne have already demonstrated.

Fat needs a definition. The word does two jobs, so it needs pinning down. Some obesity drugs, including Arrowhead’s INHBE program, silence a gene in the liver and change how the body stores fat. Those are liver drugs, and they stay in the book. Others deliver the drug into fat cells themselves, and in this paper, fat means that second kind. Here an early opening exists, and Arrowhead sketched it.

In January, it reported that its ALK7 drug cut the target gene in fat cells by a mean of 88 percent at the top dose, in four people at week 8, with a 14.1 percent placebo-adjusted drop in visceral fat after one dose. The company called it the first RNAi drug to silence a fat-cell gene in humans. It is small and early, a first page rather than a book. It shows the drug can reach the fat cell, and it has not yet shown how long the effect lasts. SiranBio’s SA030, licensed to GSK in May, is the first second-wave attempt to follow.

The lung has a short book. Arrowhead wrote this one too, by inhaler rather than injection, and it is further along than fat. Its inhaled drug ARO-RAGE was tested in a randomized trial in healthy volunteers and people with asthma, published in Nature Medicine in September. A single dose cut its target protein by a mean maximum of about 90 percent in fluid washed from the lungs, with silencing still deep weeks later, and asthma patients showed a matching drop in the blood. One company, one target, inhaled rather than injected, and not yet copied. That is closer to a short book than a first page.

The brain has no book yet. Here everyone is out of book, the incumbents included. Alnylam’s most advanced brain drug is still given by spinal injection, and Arrowhead’s shot under the skin gets its first human test on October 14. The same is true further out. Arrowhead lists the eye and heart muscle cells as preclinical delivery targets, though an unannounced eye drug, given by a shot under the skin, has been enrolling on the public trial registry since June. Nobody has shown either tissue reached from a shot under the skin in people yet.

The five barriers below apply most to the brain. They still slow anyone trying to copy the newer openings in fat and lung, where only one company has human data so far. They are not the same height, so I have listed them from tallest to shortest.

Figure 2. Author’s summary of what published science and patents teach a newcomer, and what they only partly reveal.

Failure memory, the tallest. A newcomer can hire veterans, and the best ones have, as ADARx shows. What it cannot hire is the shared record of what did not work, because that record lives in the group, not in any one person. I made this argument at length in The Seconds, and the second wave is its real-world test. Artificial intelligence does not automatically close the gap. The models learn from failure data, and much of that data is exactly what the incumbents never published. I should be honest that this is also the barrier I can least prove. Nobody outside a company can inspect its failure file, and the difference will only show when a new chemistry runs into a 2016style surprise.

Safety history. Regulators get comfortable with a chemistry slowly, one trial at a time. Arrowhead and Alnylam each have years of human safety data across thousands of people, counting their partners’ large trials. They also learned some lessons the hard way, as 2016 showed. A newcomer introducing a materially new chemistry starts much closer to zero, and the clock runs in years.

Monkeys, tall but changing. Delivery beyond the liver is hard to work out fully in mice. Blood-brain barrier crossing, muscle uptake and fat distribution differ between species, so for several tissues beyond the liver, and especially the brain, primate studies remain one of the most informative bridges between mice and people. Monkeys are scarce and expensive. In 2026 the price of a research-grade cynomolgus monkey in China reached about 200,000 yuan, roughly 30,000 dollars, close to its all-time high and more than double what it cost a year earlier.

For programs that lean heavily on primate testing, preclinical costs have run past twenty million yuan, with the monkeys as much as half of it. One Chinese brokerage projects a shortfall of fifteen to twenty thousand animals between 2026 and 2028. Well-funded companies lock in supply in advance. Biotechs with less cash have to choose which programs to delay.

It is not easier in the United States. China halted primate exports in 2020, and in 2021 the national primate research centers could not meet two-thirds of researcher requests. Arrowhead and Alnylam have more than a decade of primate data on their own chemistries. That history does not let them skip the studies for a new delivery handle.

Arrowhead’s brain drug still needed its own primate package. What the history buys is a shorter conversation with regulators about the backbone chemistry, and a trained eye for reading what the animals show. A newcomer gets neither, and has to queue for animals that cost far more than they used to. It is less a time machine than a head start. This barrier could also shrink as regulators push toward non-animal methods, a point I come back to in Section 8.

Patents, a middle barrier. The tricks for reaching new tissues are patented, and the incumbents hold many of them. A newcomer has to design around them or license them, and has to be ready for a fight if it guesses wrong. Alnylam has shown it will go to court over chemistry, suing Pfizer and Moderna in 2022 over lipid technology it says their COVID vaccines used. That case was about a different technology, though, and the patent estates that matter beyond the liver have not been tested in court against a newcomer. Determined competitors can sometimes design around them. I am asserting this barrier more than showing it, which is another reason it sits in the middle, and mapping those estates properly is a paper of its own.

Manufacturing, the shortest. This is the shortest barrier, and I want to be clear about why. Nobody needs their own factory to make clinical supply. Contract manufacturers exist, and some of the biggest are Chinese. That has turned from a convenience into a risk. The BIOSECURE Act became law in December 2025. In June 2026, WuXi AppTec, which makes drug substance for many American biotechs, including at least one oligonucleotide company, was added to the Pentagon’s list of Chinese military companies.

On August 7, a federal judge blocked the Pentagon from enforcing that designation while WuXi’s lawsuit proceeds, though a separate route onto the BIOSECURE list remains open and the first list is due by December. Existing contracts get a five-year grace period, and the law restricts federal contracts and funding rather than drug sales. Legally, it is a complication rather than a wall, and a shorter one than the June headlines suggested. The telling part is what happened anyway. Within weeks of the June listing, customers canceled projects or moved programs to other suppliers. Supply chains move faster than courts.

Arrowhead built its own 160,000-square-foot plant in Verona, Wisconsin, for about 250 million dollars, designed for commercial-scale production of its drugs. That is an advantage in control, supply and insulation from politics, not knowledge nobody else has. It matters most at scale, and the biggest-volume programs, such as liver-acting obesity drugs, sit in the part of the board that is becoming a commons. The plant protects the commoditizing business. It is not the brain moat.

A monkey costs thirty thousand dollars. The trained eye that reads the study does not come with the animal.

6. What the odds actually look like

I want to be careful here, because the easy version of this argument is wrong. Across all of drug development, experienced sponsors do only somewhat better than new ones. One study of trials across this century found the top twenty pharmaceutical companies took drugs from first human trials to approval 9.2 to 9.8 percent of the time, against 8.0 to 9.0 percent for biotechs. Sponsor track record does matter, and the BIO study of the last decade found it significantly tied to late-stage success, but it is a nudge, not a cliff.

The broad statistics, then, do not justify assuming the newcomers will fail. This field’s own history supplies a more specific warning. The first wave had more money, more chemists and more monkeys than any newcomer today, and nearly all of it left. The two survivors nearly died. That is a small sample from a different era, and I am not going to pretend it is a statistical base rate. It is a warning: when delivery moves beyond what is already solved, money and talent have not been enough to make development predictable.

The time it takes is just as telling. Alnylam was founded in 2002 and got its first approval in 2018. Even now, after more than twenty years, its most advanced brain drug is still given by spinal injection. If the company that wrote most of the liver book is still working out the brain, a five-year-old company should not be expected to do it on schedule.

The exception came before the second wave, and it proves the point. Avidity, founded in 2012, used an antibody rather than a sugar to carry siRNA into muscle. It announced the first delivery of siRNA into human muscle in December 2022, a decade after it started. In October 2025, Novartis agreed to buy the company for twelve billion dollars, a forty-six percent premium. The lesson cuts both ways. A newcomer can open a new tissue, given ten years and a book of its own. When one does, the prize is the whole company. That is the real claim of this paper. It is not that newcomers cannot succeed beyond the liver. It is that they cannot copy that success on the timetable the liver allowed.

The Avidity sale also sets up a choice for any company that wants into a tissue nobody else has opened. It can wait a decade for someone to write the book, or buy the company that already has.

Notice, too, how pharma itself is pricing the second wave. Most of every headline value is paid only if the drugs work. Argo’s 5.2 billion dollar Novartis pact came with 160 million dollars upfront. A 160 million dollar upfront payment against a 5.2 billion dollar headline looks less like full conviction than a relatively inexpensive option on future success. GSK’s deal for SiranBio’s fat-cell drug is starker still: 55 million dollars upfront against about a billion in milestones.

Here is how I would set the odds. In the liver, newcomers can win, and some will. They can be cheaper, and that is part of what pharma is buying. In the brain, the one tissue where nobody has an opening for a drug given outside the spine, they face most or all of these barriers at once. I expect most of those programs to arrive years later than planned, cost more than budgeted, and run into safety questions the incumbents have more practice reading. Whether their drugs will be worse is unknowable today, because no second-wave company has yet established a durable human opening outside the liver. That they will often be slower and more expensive is the part I would bet on.

The newcomers will be judged where their own book ends. For many of them, that is still the liver.

7. What to watch

This argument is testable, and the evidence will arrive over the next few years. Here is what I will be watching.

Copies of the existing openings, and the brain. Muscle data from a second-wave company would mostly show that the Avidity and Dyne opening can be learned. Fat-cell or lung data would show the same about Arrowhead’s openings. SiranBio’s ALK7 program for GSK is the first such test, and its Phase 1 data could arrive soon. Note how long each copy takes, because that is the real measure of a head start. The dataset that matters most is different: silencing in the human brain from a drug given outside the spine, by anyone other than Arrowhead. Nobody has shown it yet.

How much is paid upfront. If upfront payments start to rise as a share of headline deal values, pharma is gaining confidence in the newcomers. If they stay small, it is still buying options.

Monkey prices. Prices rising further means the barrier is getting taller. Prices falling, or new supply coming online, means it is shrinking.

The BIOSECURE list. Whether WuXi AppTec and others land on the government’s first list of companies of concern, due by December, and how quickly newcomers move their supply.

The first liver approval and its price. The first Chinese-origin liver siRNA approved in the United States or Europe, and the price it launches at, will tell you how much pressure Arrowhead’s liver franchise faces.

October 14. The first human data from Arrowhead’s brain drug, in healthy volunteers, arrives on October 14. It will show whether a shot under the skin gets the drug into the human brain and lowers its target. It will not show whether the drug helps anyone with Alzheimer’s. Patient data comes later.

8. What would make me wrong

Four results, in order of how much they would hurt the argument.

A newcomer reaches the brain in people, quickly. If a second-wave company shows lasting silencing in the human brain within the next two or three years, from a drug given outside the spine, especially with a pharma partner supplying the monkeys and the manufacturing, the barriers were lower than I have made them. That is the result that would hurt.

Monkeys stop mattering. The FDA published a roadmap in 2025 to reduce reliance on animal testing in preclinical safety studies. If lab-grown tissue models and computer simulations become acceptable substitutes for primate work faster than I expect, the monkey barrier shrinks, and so does the value of a decade of primate data.

The liver erodes faster than I think. If Chinese liver siRNAs reach approval and win share at prices Arrowhead cannot match, the liver section of this paper turns out to be understated, and the company’s near-term revenue is in more trouble than its platform story suggests.

A newcomer copies an existing opening, quickly. A second-wave company that matches Avidity and Dyne in human muscle, or Arrowhead in fat cells or the lung, within a few years would be mildly damaging. It would show that a written book can be learned fast, which shortens how long any head start lasts.

9. Out of book

When a chess player leaves published theory, the commentators say the game is out of book. From that moment every move has to be found at the board, on the clock, with no book to copy. Strong players prepare for this. They know the book, and they also know what happens after it ends, because they have been there before, lost, and written down what they learned.

In 2008 the liver was out of book, and the richest companies in the world could not find their way through it. The newcomers arrived after that book was written. In the liver they are already playing the middlegame, and pharma has paid for their seats. In muscle, one company wrote a book of its own and sold it for twelve billion dollars. In fat cells, Arrowhead has written the first page. In the lung, by inhaler, it has written a short book. In the brain, everyone is out of book, the incumbents included.

That is the honest shape of the advantage. It is not a locked position. It is a head start and a file of failures, in the part of the board where nobody has written a page. Arrowhead went out of book in 2016 without wanting to and has been playing at the board ever since. On October 14 it will show, for the first time in people, whether a shot under the skin can reach the brain. Whether that helps anyone with Alzheimer’s is a later question.

The opening is for sale now. The middlegame still has to be earned.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. The characterization of what is and is not in the published literature, the probability judgments, and the chess analogy are the author’s own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. What the agenda gave away

On September 29, Arrowhead set a date. On October 14 it will show the first human data for ARO-MAPT, its drug designed to lower tau in the brain from a simple subcutaneous shot under the skin. Almost everyone will be watching one number: how far tau falls in the spinal fluid.

I want to point at something else in the announcement, because it is easy to miss. The guest speaker is Adam Boxer of the University of California, San Francisco. He directs the university’s clinical trial programs for Alzheimer’s disease and for frontotemporal dementia, and he is one of the best-known investigators in a small family of diseases most investors have never heard of. His talk is titled Silencing MAPT in Alzheimer’s Disease and other Tauopathies. The press release describes the drug the same way.

That is the third signal in a matter of days. Arrowhead’s slides at an RNA conference on September 25 described the program as aimed at Alzheimer’s and related tauopathies. They also listed NfL, a marker of nerve damage measured in blood and spinal fluid, among the things the trial tracks. The first full scientific presentation of the data comes at a major Alzheimer’s meeting on November 19. Its listed first author is Catherine Mummery of University College London, who was also first author of the published Phase 1b trial of the leading rival tau drug.

Everyone will be watching how far tau falls. I would also watch where Arrowhead says the drug goes next.

2. The tau family

A tauopathy is any disease driven by a buildup of the protein tau inside brain cells. Alzheimer’s is the biggest one, but it is not the purest. In Alzheimer’s, tau shares the stage with amyloid, the sticky protein that the approved antibody drugs remove. That makes it hard to know how much of the disease a tau drug can fix on its own.

The rarer tauopathies are cleaner tests. Progressive supranuclear palsy, or PSP, affects roughly six people in every hundred thousand. It takes away balance, movement and control of the eyes, it moves relentlessly, and there is no approved treatment that slows it. Corticobasal degeneration, or CBD, is a close cousin. A substantial share of frontotemporal dementia, which usually strikes people between their mid-forties and mid-sixties, involves tau too, and a much smaller group is caused directly by inherited mutations in the MAPT gene. That is the same gene ARO-MAPT silences.

Starting small has another advantage, and it is about getting a clear answer. Alzheimer’s asks whether lowering tau can matter inside a disease that also involves amyloid, inflammation, damaged blood vessels and enormous differences from one patient to the next. Inherited MAPT disease asks a cleaner question. If the mutation drives the disease and the drug silences the gene, biology has fewer places to hide. Read purely from the biology, one logical order of attack runs from the clearest cause to the biggest market: inherited MAPT disease, then PSP and CBD, then Alzheimer’s.

Figure 1. Author’s summary. Prevalence for PSP from published reviews; the Alzheimer’s figure is from Arrowhead’s September 29 release.

3. Why PSP fits this particular drug

One detail in the PSP research jumped out at me. In PSP, much of the tau damage sits deep in the brain, in the basal ganglia, the brainstem and part of the cerebellum, with parts of the cortex affected as well. Those deep structures can be particularly hard to reach evenly with drugs delivered into the spinal fluid. They are also the regions Arrowhead keeps pointing to when it says its delivery reached deep brain structures consistently in animal studies.

If that delivery claim holds up in people, PSP may be the disease where it matters most.

A drug that reaches the deep brain is worth the most in the disease that lives there.

4. A door that has been opened before

There is also a regulatory reason to look at the small diseases first. In 2023 the FDA granted accelerated approval to tofersen, a gene-silencing drug for an inherited form of ALS, based on a drop in blood NfL. It did so even though the drug’s main trial did not show a significant clinical benefit. The agency’s outside advisers voted unanimously that lowering NfL was reasonably likely to predict benefit in that disease.

NfL has a history in PSP too. A 2016 study found that blood NfL is elevated in PSP and that patients with higher levels declined faster and lost more brain volume over a year. The senior author of that study was Adam Boxer.

Put those pieces side by side and a path appears. A rare, fast, untreated tauopathy. A drug designed to lower the protein at its source. A blood marker with a regulatory precedent and a track record in this exact disease. The inherited MAPT form of frontotemporal dementia is the closest parallel to tofersen of all: a small group of patients defined by a mutation in the very gene the drug turns down.

The side door into neurology has been opened once before. It was opened with NfL.

5. What would make me wrong

The number disappoints. Everything here depends on October 14. If tau does not fall far enough in healthy volunteers, none of the rest matters.

Tau drugs have failed in PSP before. Two antibodies aimed at tau, Biogen’s gosuranemab and AbbVie’s tilavonemab, failed large trials in PSP. Both grabbed tau floating outside cells. ARO-MAPT is designed to cut production of tau inside them, which is a different bet. The history is still a warning, and Boxer knows it firsthand, since he was an author on the gosuranemab trial. Tilavonemab showed it was hitting its target in the spinal fluid and did not help patients. Moving a biomarker is not the same as helping people.

No regulator has accepted NfL for a tauopathy. Tofersen was one disease, one mutation and one decision. The FDA could insist on clinical endpoints in PSP, and it would be within its rights to do so. That would slow the path, not close it. None of this requires the FDA to accept NfL as a stand-in for benefit. A fast-moving, biologically cleaner tauopathy could still give a much quicker test of whether deep tau silencing changes the disease.

Arrowhead may not go there. A talk title is not a trial. The company has not announced a PSP or frontotemporal dementia study, and it may choose to put everything into Alzheimer’s.

6. What to watch on October 14

Beyond the headline number, four things. The first three test the idea in this note. Whether management names another tauopathy as a next step. Whether NfL results appear, even as a safety check in healthy volunteers. Whether Boxer speaks about PSP or inherited frontotemporal dementia in any detail. Any one of those would make the tauopathy path harder to dismiss as coincidence.

The fourth is bigger than this note. The webinar opens with an overview of Arrowhead’s brain pipeline, and Chris Anzalone has already said that encouraging early ARO-MAPT data would bring a substantial expansion of that pipeline starting at the end of 2026. Watch whether management names new brain targets or puts specifics behind that promise. If the route into the brain works, tau is the first drug through the door, not the last.

7. The long diagonal

In chess, a fianchetto puts a bishop near the edge of the board, tucked in beside the king. Beginners think it looks passive. Stronger players know better. The bishop sits on the longest diagonal on the board and bears down on the center from the side, often with more force than a piece planted in the middle.

Alzheimer’s is the center of the board, the prize everyone wants. Arrowhead does not have to approach it head on. PSP and inherited frontotemporal dementia sit on the edge: small, overlooked and untreated. A drug that proves itself on tau there bears down on the biggest prize in neurology without walking straight into it. The diagonal runs from the cleanest biology to the largest market, not the other way around.

The fastest road to the center of the board is not always straight up the middle.

October 14 will tell us whether the bishop can move at all. The agenda already hints at which diagonal Arrowhead has in mind.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

The webinar date, agenda, speaker description, study design and the deep brain delivery language are from Arrowhead’s press release of September 29, 2026. The description of Arrowhead’s slides is from its presentation at the RNA Leaders USA Congress on September 25, 2026. The CTAD presentation of November 19, 2026 is from the published conference program. Tofersen’s accelerated approval and the advisory committee vote are from 2023 FDA and news reports. The NfL findings in PSP are from Rojas and colleagues, Annals of Clinical and Translational Neurology, 2016, with Adam Boxer as senior author.

PSP prevalence and the regions most affected are from published reviews. The gosuranemab and tilavonemab results are from their published Phase 2 trials. The interpretation of Arrowhead’s intentions and the chess analogy are the author’s own. Nothing here suggests that Dr. Boxer or Arrowhead has endorsed this reading.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. Start by giving them their due

I want to be clear about something before anything else, because notes like this usually get read by people looking for a villain.

Goldman understands this company well. Plozasiran is Arrowhead’s drug for dangerously high triglycerides. Goldman expects it to reach four and a half billion dollars a year at its peak in that use alone. That is nearly a billion above the average Wall Street estimate the note itself cites, and Goldman gives it a ninety percent chance of success. They credit it with deeper triglyceride lowering and a shot every three months instead of every month. They also credit it with a cleaner liver picture, no platelet problem, a broader label in Europe, and more room for error when a patient is late for a dose.

That is not a skeptic. That is most of the case a shareholder would make, written by somebody paid to be careful.

Their obesity section is good too. They understand that the kind of weight you lose matters more than the number on the scale. They also picked up the early data from small groups of patients. Adding Arrowhead’s drug to a leading weight-loss shot roughly doubled weight loss and tripled the loss of visceral fat, the harmful fat packed around the organs.

The rating is Neutral and the target is seventy-four dollars. Goldman’s previous published target, in January, was eighty-five. The pages I reviewed do not explain the difference.

The question worth asking is not whether Goldman missed something. It is how that many favorable observations produce a seventy-four dollar target.

2. The number I cannot get past

Here is how Goldman gets to seventy-four. It builds two estimates of what a share is worth and blends them.

The first is a discounted cash flow, the standard Wall Street method. You forecast the cash a company will earn over the years, then work out what that stream of cash is worth in today’s dollars. Goldman’s comes to seventy-two dollars a share.

The second is what the note calls a theoretical acquisition value, and it comes to seventy-seven. Goldman does not call that a takeover price, and I am not going to put words in their mouth. It is their estimate of what owning the whole company is worth, and it sits only about seven percent above what they think the company is worth on its own.

Weight the first at seventy percent and the second at thirty, and you get seventy-three and a half, rounded up to seventy-four.

Now put the seventy-seven next to where the stock itself has been trading.

Figure 1. The first two figures are from the Goldman note. The moving average and fifty-two week high are market data for September 24.

Seventy-seven dollars is below the stock’s fifty-day moving average. It is nineteen percent under the stock’s high for the past year. Whatever the number is meant to represent, that produces an odd result. The price put on owning the entire company sits below what ordinary investors were recently paying for a single share.

That does not make a sale at seventy-seven impossible. Stocks move, and boards weigh offers against wherever the share price sits that week. It does make seventy-seven a strange stand-in for what a buyer would pay. Getting a board to say yes usually takes a meaningful premium, and seven percent over Goldman’s own standalone value is unusually thin. It is hard to imagine a banker building a recommendation around a price the stock has already traded above.

The weighting is the part I keep coming back to. The acquisition estimate gets thirty percent of the weight and moves the final target by about a dollar and a half. Thirty percent of the weight, a dollar and a half of effect.

An alternative that changes almost nothing is not much of an alternative. It is the same forecast with a different ending bolted on.

3. What a buyer would actually be buying

Nobody acquires a company only for its 2030 revenue. They acquire it for what it will still be producing in 2040, and for the ability to keep producing it.

Arrowhead has advanced drugs into human trials across five tissue types by the company’s own account, with an eye program, ARO-033, now appearing on the public trial registry as a sixth and not yet announced. Its dimer program, ARO-DIMER-PA, put two separate gene-silencing instructions on one molecule and showed both worked in humans nine days before this note was published. It also has a drug designed to reach the brain from a shot under the skin. In primate studies it spread widely through the brain and kept lowering its target long after dosing, and the first human data is due any day.

Goldman’s acquisition value comes from taking the company’s estimated 2030 sales and multiplying by eleven. A single multiple like that does not break out any of the above. Whatever credit it gives the platform is buried inside the eleven rather than attached to anything you can point to. By 2030, some of the most interesting things a buyer would be paying for may still not be revenue. They may still be capabilities.

This is the one place in a research note where a bank openly tries to price what a company is worth to someone who wants to own it. The method it reaches for cannot isolate that value.

There is a deeper problem with running the two estimates side by side, and it takes a minute to explain. Any valuation model forecasts a company’s sales for a set number of years. After that it needs one last assumption to cover everything beyond the forecast, because the company does not stop existing when the spreadsheet ends. That last piece is called terminal value, and it often makes up a large share of the answer.

Goldman’s discounted cash flow handles it by assuming the business grows three percent a year, forever, once the forecast runs out, and the note says that three percent is where the platform’s value sits. The acquisition estimate handles it by putting a multiple on one year of sales. Both rest on the same sales forecast. They are not two different views of the company. They are one view with two different endings, and they land only five dollars apart.

A second estimate built on the same forecast cannot independently pick up value that never made it into that forecast in the first place.

Which brings me back to that three percent. A platform does not behave like three percent a year forever. It behaves like nothing, nothing, nothing, then a new tissue opens and an entire family of drugs becomes possible. Goldman is not refusing to value the platform. It is valuing the platform as terminal growth, and those are not the same.

4. The rows and the gaps

Goldman’s note includes a table listing the drugs it expects to earn money from and how much each might sell. Look at what gets counted and what does not.

Figure 2. Author’s reading of the note. Programs may appear elsewhere in exhibits not reproduced here.

I do not think leaving out a program with no human data is a scandal. That is ordinary practice and defensible on its own terms. What is harder to defend is the pattern once you stack this note against the others.

JPMorgan put no value on the dimer in its model. Morgan Stanley published sixty dollars a share for the brain drug and then left it out of its price target, writing in plain words that it does not model the program. Now the revenue table Goldman shows has no row for the brain drug, the dimer, or the delivery technology that produced both.

Three banks, one answer. The models are not disagreeing about the science. None of them has a row for it.

5. The partnered programs, and what Arrowhead actually owns

Drugs Arrowhead shares with partners are where a table like this gets tricky, and this one deserves a closer look than it usually gets.

The hepatitis B program is licensed to GSK. Arrowhead gets milestone payments along the way and a percentage of GSK’s sales, not the sales themselves. Fazirsiran works differently again. Arrowhead and Takeda split United States profits fifty-fifty and sell it together, and outside the United States Arrowhead gets a percentage of sales that steps up as sales grow. Neither is the same as selling a drug yourself, which is what Arrowhead does with plozasiran.

The table is headed peak sales, meaning each drug’s sales in its best year. It may well be showing total worldwide sales as a starting point, with Arrowhead’s actual share worked out elsewhere in a model I have not seen. If so, no error. My point is narrower. A table that puts those three side by side without marking which dollars actually reach Arrowhead is easy to read wrong, and plenty of people will.

Olpasiran is the case worth spending a paragraph on, because everybody assumes it belongs on a list of omissions and it does not. Olpasiran is Amgen’s heart drug, built on Arrowhead’s technology. Arrowhead sold its percentage of olpasiran sales to Royalty Pharma in 2022. What it has left is milestone payments. A peak sales line for olpasiran would not represent money Arrowhead collects on sales.

There is a second reason on top of that one. Olpasiran lowers Lp(a), and so does pelacarsen, a drug Novartis has been developing. Three weeks ago Novartis reported that pelacarsen, while lowering Lp(a), missed its main goal in a heart trial of more than eight thousand patients. That raised real uncertainty about whether lowering Lp(a) with a drug will translate into fewer heart attacks and strokes. An analyst who declines to build meaningful revenue from an Lp(a) drug into the model after that result is exercising judgment.

Which is the distinction this note turns on. Leaving something out because the money belongs to somebody else, or because the evidence just turned, is judgment. Leaving something out because there is nowhere to put it is architecture.

6. Where the model is not conservative at all

Here is the part I would want a reader to hold against everything above, because it cuts the other way.

Goldman gives Arrowhead’s pair of obesity drugs, counted together, a fifty-five percent chance of success. Both are in early human trials, Phase 1 and Phase 2. Historically, drugs at that stage have succeeded far less often than that, though obesity treatment has changed so much in the last few years that old success rates deserve some caution.

Take each drug in Goldman’s table, multiply its peak sales by the odds Goldman gives it, and the obesity pair comes to roughly thirty percent of the total. That is not the same as thirty percent of the price target, and I am not going to pretend otherwise. What it does show is that Goldman is not cautious across the board. A big share of the speculative value it is willing to count sits in two early obesity programs.

If you are going to accuse a model of being too cautious about the platform, you have to say where it is being generous. This is where.

Which sharpens the point rather than blunting it. The obesity drugs are already in Goldman’s revenue table. ARO-MAPT, the brain drug, is not.

7. What would make me wrong

Three things, and I would rather put them in writing now.

The platform never pays off. If Arrowhead keeps opening new tissues but those openings keep failing to produce drugs people will pay for, the value I think the model is missing was much smaller than I thought. I was pricing a capability that never turned into money.

Goldman adds it once the evidence is in. Suppose the brain data is good and Goldman’s next note adds a line for the brain drug, with odds and a market size. That would show its model can count platform value once there is human proof. That would weaken the stronger version of my argument. Goldman would not have been unable to value it. It would have been waiting for evidence first, which is a defensible place to stand.

Obesity delivers and the target holds up. If the obesity data due at year end justifies the fifty-five percent, the generous assumption I just flagged turns out to be a good call. The model would look better than I am making it sound.

8. Counting the pieces

Early chess engines counted material. A pawn was worth one, a knight three, a rook five. Add them up and you had the engine’s verdict on who was winning. Material was countable, so material was what got counted.

What those engines were famously bad at was position. A knight planted on a square nobody could ever chase it from, a clear lane for a rook, a wall of pawns that would decide the game forty moves later. Every strong human player knew those were worth something. Nobody could say precisely how much, so the engine scored them near zero and cheerfully traded a winning position for a pawn, because the pawn came with a number attached.

The position did not stop existing because the engine could not score it. The engine simply could not see it until it turned into a piece.

A research model counts drugs. Drugs have names and markets and odds. A delivery technology that might produce drugs nobody has named yet has no square on the board.

Goldman has not missed Arrowhead’s drugs. They have the drugs, in detail, and they are more positive on the lead product than the Wall Street average they cite. What the model cannot reach is the consequence of those drugs succeeding, because that consequence is not a product. It is a capability, and capabilities do not have rows.

The seventy-seven dollar acquisition value is the tell. Asked to put an acquisition value on the company, the model reached for the only tool it had, and came back with the same company at a rounding error above where it started.

The brain data is due any day. Whatever it says, the next note will tell you something useful about all of this, because either a new row appears or it does not.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

Figures attributed to Goldman Sachs are from its research note on Arrowhead dated September 24, 2026, as reviewed by the author. References to JPMorgan and Morgan Stanley are from their published notes of September 2026. Exhibit contents are the author’s reading of the pages reviewed and programs may appear in exhibits not examined here. Historical probability of success rates by development phase are drawn from published industry analyses. The characterization of research methodology, the interpretation of the terminal growth assumption, and the chess analogy are the author’s own and are not attributed to any analyst.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. A number that should not be possible

At an R&D day in 2023, Chris Anzalone said something I went back and listened to twice, because the number sounded wrong. In 2017 the company had optimized its TRiM platform for liver delivery and had not yet advanced a single TRiM candidate into the clinic. By the end of 2023, six years later, it expected to have advanced eighteen.

Eighteen drugs into human studies in six years. Call it one every four months, from a company with a fraction of the people and the money that the big firms in this field have.

That was three years ago and the count has kept climbing. Chris put the number at twenty-one or twenty-two candidates in clinical studies at a conference on September 16, and the company guides to two or three new clinical programs a year. The cumulative total since 2017 is higher still, since programs have entered while some of the original eighteen finished, stopped or went to partners. I have not tried to add it up.

The clock starts in 2017 for a reason, and skipping past it would make this story tidier than it deserves to be. Arrowhead’s first delivery approach was a different chemistry entirely, and it had candidates in human trials years earlier. Those programs were discontinued in late 2016 after a safety finding in a primate study. The company then rebuilt around the platform it has now.

The eighteen were therefore not built on an unbroken run. They were built after an approach that reached the clinic had to be abandoned, which is worth remembering when reading anything about accumulated knowledge. Some of what accumulates is the knowledge of what did not work.

The standard explanation is that RNAi is somehow easier. It is not. Every one of those eighteen went through the same work any drug goes through. What changed is how much of that work had to be done from scratch. The difference is not speed. It is repetition avoided.

2. What actually happens

Start with the part everybody assumes is hardest, which is finding the molecule in the first place.

Say you want to switch off a gene. A computer reads the message that gene puts out and lists every spot along it where a short piece of RNA could grab on. For an ordinary gene that is hundreds of spots.

Figure 1. Funnel from a published academic effort against PCSK9, shown to illustrate the shape of the process.

Then you start throwing them out. Most get discarded because the same sequence appears somewhere else in the genome, and a drug that switches off your target plus two genes you were rather fond of is not a drug. A handful actually get built and tested in cells, because making these drugs is not cheap. The best of those go into animals.

Then comes the real filter, which is less elegant than everything before it. You give rats far more drug than any person would ever receive, and then you go look at their livers.

3. The forty percent that is not what it looks like

Alnylam has published its experience with this step. Roughly forty percent of sequences fail the rat screen, typically with liver damage showing up as elevated enzymes and dead cells under a microscope.

Forty percent sounds like a disaster. It is not, and the reason is the best thing about this whole business.

They can routinely find both toxic and non-toxic sequences against the same gene. In those experiments the damage was not generally caused by silencing the intended target. It was sequence-dependent, and came from the particular sequence hitting something it should not.

Which means the failure costs you a sequence. It does not cost you the target. You go back to your list of ten and pick another one.

Sit with that for a second, because it is genuinely unusual. In most of drug development, a toxicology failure can end the program. The molecule is the program. Here the molecule is one of several shots at the same target, and what you actually care about, the gene, is untouched by the failure.

That one property changes the whole shape of the risk. You are not betting on a molecule and hoping. You are betting on a target and running a search, and searches you can run again.

4. What gets built fresh and what gets inherited

Here is the part that actually explains the four months.

Every one of these drugs passes through the same stations. Pick the target, design the sequences, screen for what else they might hit, animal work, toxicology, manufacturing, the regulatory filing. Nobody skips any of it.

What changes is how many of those stations you have to build yourself.

Figure 2. Author’s grouping of which development steps recur and which carry over.

The chemistry is the clearest example, and Alnylam has documented its own version of this more openly than anybody. Bare RNA falls apart in the bloodstream within minutes, so the molecule has to be armored first. Alnylam arrived at its pattern through what the literature describes as a comprehensive screening effort across both strands of the duplex, optimizing where to place each modification.

That work was enormous and it was done once. Every drug after it inherits the template. Nobody sits down and works out where the fluorines go all over again. Arrowhead runs different chemistry on a different platform, and the principle is the same. The hard optimization happens at the start and is then applied.

Same story with the piece that finds the tissue. A little cluster of sugar molecules called GalNAc latches onto a receptor that sits on liver cells and almost nowhere else. Working that out took years. The twentieth liver target does not require working it out again, because it is the same chemistry attached to a different sequence.

The platform is not a shortcut through the work. It is knowing which work does not need doing again.

Chris put it more bluntly at a conference this September, when somebody asked whether artificial intelligence lets a newcomer leapfrog a company like his. Arrowhead has made hundreds of thousands of RNAi molecules over the years, he said, and has learned a great deal about what makes some potent and some not. An engine is only as good as the data fed into it. Brute force, in his phrasing, is a hell of a thing.

There is a third category the figure above does not capture. Every so often the platform does not simply produce another drug. It acquires a capability, and the capability itself becomes inheritable.

ARO-DIMER-PA is the current example. Getting two silencing triggers onto one molecule and showing that both work in a person took years and had never been done. Arrowhead reported it on September 15. Chris said the same day that several more dimers are already in development and some should reach the clinic in 2027, including work on both of the obesity and liver targets the company has been running.

He was also unambiguous about whether it worked. The goal going in was LDL lowering in the range of the existing PCSK9 drugs and triglyceride lowering in the range of plozasiran.

His verdict on whether they got there was that they hit both in spades, which is not the language of a man hedging.

The first dimer was expensive. The ones behind it inherit what the first one paid to learn.

The output of the first program was therefore not just a drug. It was a method.

5. Nobody sets out to build a platform

Worth pausing on how a platform actually comes into existence, because the word platform suggests somebody drew it up first.

They did not. You have a drug to make. You solve the delivery problem because that drug needs solving, and the solution turns out to work for the next one. You settle the chemistry because this molecule falls apart in blood, and the pattern holds for everything after. Each piece gets built to finish a specific job, and only later does anybody notice it keeps getting reused.

A platform is not designed. It accumulates, and you find out you have one.

Which means it is a ratio rather than a status. Look back at the two columns. Five things built fresh, six inherited. Early on that split runs the other way and almost everything is bespoke, and you are a company with a drug. Somewhere along the line the right-hand column outgrows the left, and the label starts to mean something.

The efficiency lands somewhere people usually miss, too. It is not that any single drug moves faster, because Chris was not claiming that and neither am I. It is that the twentieth program asks far less of the organization than the first one did. Less bespoke chemistry, less novel safety interpretation, less arguing with a regulator about a class nobody has seen.

That is why twenty-odd programs can run at once. Not because each one is quick, but because each one costs the company less of its attention than the one before it.

A company where every program needs its own delivery system cannot carry twenty. It does not have twenty programs worth of bespoke effort to give.

Which raises the question nobody asks about platforms. If they are this useful, why does almost nobody have one?

Look at the shape of the cost. You pay for that right-hand column years before it starts paying you back, by running program after program at full bespoke price while the ratio is still pointed the wrong way. Most companies cannot survive that stretch. They run one asset, or two, and get bought or run out of money long before the inherited column outgrows the other one.

A platform only becomes cheap after you have paid for it many times over, and the paying comes first. That window is where most of them die.

Arrowhead nearly did. Nine years in, with candidates already in human trials, the whole delivery approach came apart and everything built on top of it went too. What makes the eighteen worth noticing is not only the pace. It is that they came after that.

It runs the other way too, and this part gets missed. The inherited column only grows while programs are moving through it. The first nineteen tell you how to read the twentieth. Only running the twentieth gives you something new to learn. Stop running programs and the accumulating stops with them, which makes all this a flow rather than a stock.

6. Which is why you cannot start with one

Turn that around and you get the problem facing anybody trying to enter this field now, and there are a great many of them. Chris counted roughly a dozen Chinese siRNA companies at one point, and said several are fast.

They can buy speed. What they cannot buy all at once is history. Every item in that right-hand column was paid for by somebody, and a company standing up its first program pays for all of them at once while also doing the work in the left-hand column.

The obvious objection is that the chemistry is published, and it is. Alnylam has been unusually open about where the modifications go and why. You can read it this afternoon.

What gets published is what worked. Nobody publishes the two hundred thousand molecules that did not, and those are most of the dataset.

Chris describes the accumulating end of this less delicately than I would. Fifteen years of banging their heads against the RNAi wall, in his phrase, and no substitute for history to get there.

The literature hands you the answer. It does not hand you the map of wrong answers that had to be eliminated to find it, or tell you which of your own ideas are already on that map.

I would not oversell this. A newcomer can rent manufacturing from a contract shop, license chemistry, hire people who have done it elsewhere, and patents expire on a schedule. What is hard to buy is the accumulated record of failure, and even that is a lead measured in years rather than a wall. Chris said as much himself, and the qualifier he attached to it is worth more than the claim.

7. Why the safety package moves faster

This is where the months usually disappear, so it is worth walking through.

For programs like these, reaching a person has typically required formal toxicology in two species, rats and monkeys, running about a month with a recovery period after. That does not go away.

What changes is what the findings mean. Read a published tox package for one of these drugs and the same phrase keeps turning up. The changes observed in kidney and liver were in line with previously described effects for this class. Known. Expected. Already characterized.

A regulator looking at a novel chemistry has to work out which findings matter. A regulator looking at the twentieth molecule from a characterized class is comparing against a picture they already have.

The same logic runs through the primate work. Published work suggests the mechanism producing months-long silencing is conserved across rodents, monkeys and people. That is why the monkey number tells you something useful about the human one, and it is why a company that has run the comparison twenty times has a better sense of what its primate data is worth.

8. Where the preparation runs out

None of this helps you with a new tissue, and that is where the argument stops.

Change the destination and most of the right-hand column moves back to the left. You need a new ligand, which means finding a receptor that is abundant on the cells you want and scarce everywhere else. You need to prove it carries a payload inside. You need new safety work, because the class effects you have characterized were characterized in a different organ.

Chris has put the cadence at a new cell type every eighteen to twenty-four months. Seven are addressable by the company's account, and he said in September that clinical programs were running in five. The public trial registry already shows a sixth, an eye program that the company has not yet announced. That is the number that matters for anyone thinking about how far this extends, and it is a very different number from four months.

9. An and company, not an or company

There is a wrinkle here that makes that cadence less orderly than it sounds. Arrowhead does not wait for the validating result before building what comes after it.

Chris has a phrase for this and he uses it often enough that it is plainly not improvised. Arrowhead is an and company rather than an or company.

An or company picks. An and company builds the next drug while the current one is still being tested, and finds out afterward whether it needed to.

He said it again in September, then went further. The follow-on brain programs are being built right now, without waiting to see the tau data, exactly the way the later dimers were built before the first one had reported anything.

Which means the clock on the next tissue starts before the current one is proven. The two clocks overlap rather than queue.

That is real compression, and it comes with a bill the company has already had to pay once. Developing on the assumption that a bet lands is why the pipeline fills quickly. It is also why the collapse of the first delivery approach in 2016 cost more than one program. Everything built on top of an assumption goes when the assumption does.

Four months buys you another drug. Two years or so buys you another organ, and the second one is the expensive purchase.

It is also why the brain readout carries the weight it does. Not because one Alzheimer’s drug is worth more than the others, but because opening a tissue is the slow, costly part of this business and the brain is among the hardest tissues to reach.

Chris is also clear about the shelf life of all this. Asked in September about the number of Chinese companies now working in RNAi, he said Arrowhead holds multi-year head starts across these capabilities, then added the qualifier that matters. Multi-year, not forever.

10. What the clock actually looks like

Before you look at the chart, be clear about where it comes from. Arrowhead has never published a stage-by-stage timeline and neither has anybody else here. This is my reconstruction. A few durations have published sources and are marked. The rest are my estimates, and any given program can look wildly different depending on how well understood the target already is.

Figure 3. Author’s reconstruction, not a company disclosure. Teal bars have published sources. Brown bars are the author’s estimates. Any individual program can look quite different.

Two things jump out.

The first is that almost everything before the candidate is chosen runs in sequence. You cannot rank sequences before you have settled on the gene. You cannot test in cells until you have synthesized. You cannot run the rat screen until you have something worth screening. That whole stretch is a queue, one thing after another.

The second is what happens after. Formal toxicology and manufacturing start at the same moment and run alongside each other, because both need the same thing, which is a chosen molecule. The dosing and recovery periods are published at about three months, and call it five by the time the analysis is done. One large contract manufacturer advertises around nine months to deliver the first material made to clinical standard.

Which means the critical path between a finished candidate and a first patient can run through the manufacturing plant rather than the toxicology lab.

Treat that the way you would treat any conclusion built on estimates. It follows from the durations above, and if those are materially wrong, so is it.

That reframes a decision I have written about before. Arrowhead built its own plant in Wisconsin rather than renting capacity. Read as real estate that is a cost line. Read against this chart it is the company taking ownership of a potential bottleneck, on a platform designed to keep twenty things moving at once.

Arrowhead’s own chief medical officer handed me the limit on all of this, and I would rather quote him against my argument than tiptoe around it. Asked in September whether artificial intelligence would speed things up, James Hamilton said there is no substitute for running the clinical trial. Shaving a few months off discovery is fine, he said, but you still have to grind through every stage of development.

He is right, and it puts a fence around this entire paper. Everything described here compresses the path to the clinic. None of it compresses the path through one.

There is a harder version of that point, and it is the one I would put to myself. Eighteen candidates into the clinic is a count of entries, not of finishes. Some of those programs are no longer running. Getting to the starting line quickly and cheaply does not make a drug more likely to work once it is there, and on that question Arrowhead gets graded the same way everybody else does.

What the platform changes is how many times you get to take the test, and how early you can sit down for it. Those are worth a great deal. They are not the same as passing.

A Phase 3 cardiovascular outcomes trial takes the years it takes whether the molecule arrived in eighteen months or sixty. What the platform buys is more shots on goal and an earlier start on each. Development time cannot be recovered later, so that earlier start compounds across a pipeline.

11. Opening preparation

Chess has a version of this, and it is the closest analogy I have found to what is actually going on.

Two grandmasters sit down across from each other. Twenty moves in, one has burned four minutes on the clock and the other has burned an hour. Same board, same position, same rules. One of them prepared this line at home, knows the critical branches, and has been playing from memory. The other has been calculating.

The prepared player is not smarter and is not thinking faster. They did the work earlier, once, and are now collecting on it.

What preparation actually buys is not memorized moves. It is knowing which of the thirty legal options are worth thinking about. The unprepared player burns an hour eliminating possibilities their opponent eliminated last year.

Arrowhead is not designing drugs faster than anyone else. It is arriving at move twenty with most of the clock still on it.

Eighteen drugs in six years is what that looks like from the outside. From the inside it is a company that already answered the questions everybody else is still sitting with, and now spends its time on the handful that are genuinely new each time.

Every one of those eighteen ran the same race. Most of them started somewhere near the finish.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper.

Several of the load-bearing facts here come directly from Arrowhead and from its executives speaking publicly. The eighteen clinical candidates through 2023, the count of twenty-one to twenty-two candidates in clinical studies as of September 16, 2026, the cadence for opening a new cell type, the count of seven addressable cell types with clinical programs in five, the description of the company as an and rather than an or and the practice of building follow-on programs before the validating data arrives, the multi-year head start and its limits, the hundreds of thousands of RNAi molecules made to date, and James Hamilton’s point that nothing substitutes for running the clinical trial are all Arrowhead’s own statements, from a 2023 R&D day and from public appearances in September 2026.

The ARO-DIMER-PA results, the assessment of them, and the timing of the dimers behind it are from the company release of September 15, 2026 and from Chris Anzalone speaking the same week.

The sixth clinical tissue is the ocular program ARO-033, taken from its public registration on ClinicalTrials.gov under NCT07662096, which lists Arrowhead as sponsor and describes subcutaneous administration. The company has not announced it. Everything stated here about ARO-033 comes from that public registration.

Several phrases in the text are Chris Anzalone's own language from those appearances rather than mine, and their repetition here is deliberate: brute force is a hell of a thing, banging their heads against the RNAi wall, no substitute for history, in spades, an and rather than an or, and multi-year, not forever.

What is not Arrowhead’s. The sequence funnel numbers come from a published academic program against PCSK9 and illustrate the shape of the process rather than any one company’s. The attrition rate, the toxicology findings and the chemistry descriptions are drawn from the published literature, much of it Alnylam’s. The manufacturing duration is one contract manufacturer’s advertised figure. The grouping of which development steps recur is the author’s own, and so is the development timeline, including the sequence of stages, which run in parallel and how long each takes. Arrowhead has not published a stage-by-stage timeline. None of that chart should be read as company guidance, as a description of any specific Arrowhead program, or as a forecast of how long any future program will take.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The scoreboard

Count the tissues where a company has an RNAi drug in a person, and the list is short.

Figure 1. Author’s reading of public pipelines as of September 2026. Spinal injection is marked separately because it is a different problem with a different burden on the patient.

Alnylam is the closest. It got its second tissue by a shot under the skin this year, putting a fat program into Phase 1 alongside its liver franchise, and its brain programs go in through the spine. Ionis works with antisense rather than RNAi, so it is not strictly on this board, but it is the nearest mature comparison and its picture looks the same.

Everybody else is in the liver or not yet anywhere, including the well-funded newcomers whose programs outside the liver have not reached a person yet. One company is in six. Liver for everything the liver does, muscle for a rare wasting disorder, fat for obesity, brain for Alzheimer’s, and the retina for macular degeneration, all five by an injection under the skin. Lung for cystic fibrosis and other pulmonary disease, which goes in by inhaler, straight to the airway, and never has to survive the bloodstream at all.

The sixth one has not been announced. ARO-033 sits on the public trial registry, recruiting since June for macular degeneration, first in healthy volunteers and then in patients. The registration states the route four separate times. It goes in as a shot under the skin, which means Arrowhead is trying to reach the retina from the arm, with the drug traveling through the bloodstream rather than being placed in the eye. Every approved drug for that disease goes in through a needle in the eye.

Chris Anzalone said in September that the company can address seven cell types with clinical programs in five. The registry entry had been updated the day before.

Six tissues against three, and on the harder route, the one where the drug has to travel through the blood and find its way, five against two. That is not a small gap and it is not a permanent one either. Alnylam is doing the work. What separates them right now is years.

2. Why everybody else stayed in the liver

The obvious question is why. If getting outside the liver is so valuable, and it is, why did the biggest and best-funded companies in the field not do it?

Because the liver was enough.

A sugar cluster called GalNAc grabs a receptor that sits on liver cells in enormous abundance and almost nowhere else. The pairing was not discovered by any of these companies. Academics worked out that the liver had this receptor and that the sugar would stick to it decades before anybody put RNA on the end of it. In that sense the handle was lying on the table.

Picking it up was a different matter. Getting from a known sugar to a drug that silences a gene for six months took years of chemistry, and Alnylam did most of that work. What matters is that it only had to happen once. The pattern got published, the patents eventually run out, and now anyone entering the field starts with the basic liver-delivery problem already solved for them.

There was also a scientific reason. The liver is forgiving. It filters blood, so anything you inject ends up passing through it. It has a receptor that recycles fast. It tolerates a lot of drug. Every other tissue is worse on at least one of those counts, and most are worse on all of them.

Staying in the liver was not a failure of ambition. It was the right call for anyone who already had a liver franchise. Which is precisely the point.

3. Why Arrowhead went looking anyway

Arrowhead did not have a liver franchise. In 2011 it had a Madison operation it had picked up from Roche for almost nothing, a delivery chemistry that had just been abandoned by its previous owner, and no approved drugs. Alnylam was years ahead in the liver and better funded.

Competing head-on in the liver meant fighting the incumbent on its home ground with fewer resources. Chris Anzalone said as much in September, looking back. Early on, he said, they decided that to extract real value from the technology they needed to get outside the liver, and that starting around 2010 they began building delivery for other tissues.

Worth being precise here, because the shorthand gets this wrong. Arrowhead never left the liver. It is still there, and the only drug it has on the market is a liver drug. What it did was refuse to be only there, at a moment when it had no approved product and every reason to concentrate on the one tissue that already worked.

The tissue record everybody admires now started as the second front opened by a company that could not win the first one on resources alone.

I want to be honest about what that means, because it is a better story than genius. Arrowhead went looking elsewhere because the good ground was already occupied. Necessity picked the direction. What happened after that is the part worth understanding.

4. What opening a tissue actually takes

Here is why the gap on the scoreboard is as wide as it is.

Figure 2. Simplified. Each step has its own literature and its own graveyard.

Opening a tissue is not one problem. It is six, in sequence, and an attempt can die at any of them.

You need a receptor that sits on the cells you want and almost nowhere else, and for most tissues that receptor is not obvious. You need a handle that grabs it, small enough to attach to RNA without wrecking either one. You need proof the receptor actually carries the whole package inside rather than just holding onto it at the surface. You need the package to escape the bubble it gets trapped in once inside, which is where most of what enters a cell stays. You need the silencing to last months rather than days. Then you need to find out where else that receptor lives, because it lives somewhere you did not choose, and whether that matters.

Six steps. No textbook covers all of them together, because no discipline does. The people who can do this learned by failing at each step in order, more than once.

That is the real reason the list is short. The chemistry to hit a target is something you can hire for. The ability to open a tissue is something almost nobody has, because the training does not exist and the only way to acquire it is to spend a decade at it.

5. The people who can do this

Which brings me to the part of this that does not appear in any model and that I think matters more than any single program.

A delivery scientist is not a chemist, though they have to know chemistry. Not a cell biologist, though the bubble that traps the drug inside a cell does not care what your degree says. Not a pharmacologist, though durability is a pharmacology question. It is all three at once, applied to a problem that most of the world has been content to leave alone because the liver was enough.

The number of people who can do that end to end is small. I would guess a few dozen, worldwide, and I am not confident of the second digit. There is no graduate program in it. There is no textbook. You become one by working on it for a long time alongside other people who are working on it, and a surprising number of the people who fit that description have spent their careers in one building in Wisconsin.

I wrote about that building in an earlier note. The short version is that these people came up at a Madison company called Mirus Bio, founded out of university research in the 1990s. Roche acquired Mirus Bio in 2008 and ran the site as part of its own RNAi effort. When Roche abandoned the field two years later, the operation went to Arrowhead. The group stayed together through both changes of owner because there was nowhere local to go. Two of Roche’s own site leaders were put back in charge within a month of Arrowhead taking over. Arrowhead brought aboard thirty-seven employees from the operation.

The platform everyone talks about as if it were a machine is a few dozen people who have been failing at the same six steps together for longer than most biotech companies have existed.

Here is why that word together matters more than it looks. A competitor could go out tomorrow and hire excellent delivery scientists from a dozen companies. It still would not get what Madison has, because a lot of what Madison has does not sit inside any one of them.

Failure makes information, and most of that information never becomes a paper or a patent. It becomes a conversation somebody remembers twelve years later. We tried that receptor. We tried that linker. Here is what happened, and here is why nobody has bothered since. One person remembers the handle that behaved strangely with a particular chemistry. Another remembers the toxicity signal from a receptor that looked perfect on paper. A third remembers the animal result that was going to be a program until somebody looked at where the drug actually ended up.

The clearest example is one the company would rather not be remembered for. In 2016 the delivery chemistry the team had inherited from Roche, which had already put candidates into human trials, came apart after a safety finding in a primate study. The programs were discontinued. Everything built on that chemistry went with them.

Within about two years the same people had a different delivery approach in the clinic. Every tissue on the scoreboard was opened with the one that came after the one that failed, by a team that knew exactly why the first one failed, in a level of detail no publication could capture.

You can hire a scientist. You cannot hire twenty years of shared memory one person at a time.

That makes this a stranger kind of moat than the usual one. Some of it sits in patents and molecules and databases, the way a normal moat does. Some of it sits between people, in the accumulated record of what was tried, what failed, and what was quietly set aside years ago for reasons never written down.

You cannot copy what you cannot see.

6. What six tissues in the clinic actually proves

Be careful here, because the scoreboard is easy to over-read.

Six tissues with a drug in human trials proves that the route into each one works well enough to dose a person. It does not prove that any drug in those tissues will work. Getting into muscle and fixing muscle are different problems, and the second one is not a delivery problem at all.

It also does not prove the six-step process is now routine. Chris puts a new tissue at every eighteen to twenty-four months. Brain and eye both reached the clinic this year, which is faster than that, and one fast year is not a trend. Each one is still a project with its own receptor to find and its own ways to fail.

What it does prove is narrower and still unusual. A group of people has solved the delivery problem in six tissues, five of them by the harder route through the blood. The closest competitor is in three tissues, two by that route. The second of the two arrived this year.

A drug proves a target. A tissue proves a team. Arrowhead has six of the second kind of proof, and the market only knows how to price the first.

There is a consequence of that which most models miss. Investors tend to value the tissues one at a time. Lung is worth something, muscle is worth something, the brain is worth something, add them up. If the argument here is right, that undercounts, because each tissue contains information about the next one.

Not about which receptor will work, or whether the biology will cooperate once you get there. About whether the people looking for that receptor know how to find one. That is what a platform is supposed to mean, if the word means anything. Success in one program changes the odds you should assign to a program that does not exist yet.

7. Where this can go wrong

Three ways, and I would take the first one seriously.

The people can leave. Everything above argues the asset is a team. Teams age, retire, and get recruited. The Madison group stayed together through two changes of owner partly because the labor market kept them there, and a very successful brain readout is exactly what would make that labor market noisier. If the delivery scientists are the platform, then retention belongs on the same page as the science. Most models treat it as boilerplate, the same key-person language every company files.

The next tissue may not open. Six tissues are now in the clinic. Whatever comes next remains unproven, and nobody publishes the attempts that did not.

The route can work and the drugs can still fail. This is the one the scoreboard hides. Six tissues is six opportunities for the biology to disappoint after the delivery succeeds. That would not invalidate the team. It would mean the team solved the part it controls and the disease did not cooperate.

8. The seconds

Chess used to have a custom that computers have mostly made obsolete.

In the era of adjournments and world championship matches that ran for months, a top player brought seconds. These were strong players in their own right, sometimes grandmasters, whose job was to prepare. They analyzed the opponent’s repertoire. They found the novelties. When a game was adjourned overnight with a sealed move in an envelope, the seconds stayed up until dawn working out the position while the champion slept.

The champion walked in the next morning with the winning line already in hand. The public saw the champion. The seconds sat in the back.

Every serious player knew the truth. The name on the trophy was the one who could afford the best seconds.

Chess engines, the programs that now play better than any person alive, ended most of that. Analysis that took a team a night takes a laptop a minute, and the second’s craft became one more job a machine does better.

There is no engine for opening a tissue. No program finds the receptor, builds the handle, and works out why the package will not get out of the bubble. The craft still lives in people, the way it did before the machines, and the number of people who have it is small in the way the number of world-class seconds was always small.

Arrowhead’s seconds have prepared more positions than anyone else in the field. It has had them since before it owned them. When the brain data lands in the next few weeks and everybody looks at the number, the number will belong to the company.

The position was prepared by people who have spent their careers in the back of the room, and who have now opened six doors before the rest of the field.

The champion gets the trophy. Somebody prepared the position.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper.

The tissue count is the author’s reading of public company pipelines and trial registries as of September 2026 and counts tissues with at least one RNAi program in human trials. The ocular program ARO-033 is taken from its public registration on ClinicalTrials.gov under NCT07662096, which lists Arrowhead as sponsor, describes subcutaneous administration in both the healthy volunteer and macular degeneration cohorts, and was last updated September 15, 2026. The company has not announced it. Everything stated here about ARO-033 comes from that public registration. Route of administration is marked because it matters: an injection that has to reach a tissue through the bloodstream is a different problem from an inhaled or spinal dose that is placed at or near the target.

Arrowhead’s pulmonary programs are inhaled by nebulizer, and the company has separately shown subcutaneous delivery to lung in animals. Statements attributed to Chris Anzalone are from public conference appearances in September 2026. The Madison history, including the 2008 Roche acquisition of Mirus Bio, the 2011 transfer of the operation to Arrowhead, and Arrowhead’s hiring of thirty-seven employees from the operation, is from contemporaneous Arrowhead announcements and is detailed in a prior BioBoyScout note. The six-step description of tissue delivery is a simplification and is the author’s own. The estimate of how many people can do this work end to end is a guess and is labeled as one. The chess custom described is historical and the analogy is the author’s.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. What Arrowhead reported this morning

On September 15, Arrowhead released interim topline results from the Phase 1/2a study of ARO-DIMER-PA. The numbers below are from that release and nothing else has been presented. Fuller detail goes to a medical congress later.

ARO-DIMER-PA is one molecule carrying two payloads. One switches off PCSK9, which governs how much LDL cholesterol the blood clears. The other switches off APOC3, which governs triglycerides. Both targets are reachable through hepatic delivery, and both have been silenced before, separately, by separate drugs.

Putting two independent triggers on one conjugate, and showing that both stay active in a person, is the novel part.

The obvious reading of this morning’s release is a convenience story. Two mechanisms, one injection, a large and undertreated market. That reading is correct and it is the smaller half of what happened.

2. The number to look at is the gap

Arrowhead reported mean maximal single-dose reductions of 72 percent for PCSK9 and 88 percent for APOC3.

A sixteen point spread sounds lopsided. The right question is not whether the two numbers match each other. It is whether each one is close to what a drug built for that job alone would have delivered.

Figure 1. Reductions as reported. Comparisons to other drugs are across different trials and populations and should be held loosely.

On that test both halves did their work. APOC3 fell 88 percent, which sits in the range a dedicated APOC3 drug produces. PCSK9 fell 72 percent. Downstream, LDL cholesterol fell 54 percent from a single dose, in the range of the roughly 50 percent LDL-C lowering reported with inclisiran on its established regimen. Nothing in those numbers suggests either trigger became pharmacologically irrelevant because it traveled with the other.

Whatever competition may exist between them, neither trigger was functionally crowded out. That was the thing worth finding out, and it is the thing the release does not spell out.

Had one looked strong while the other barely moved, Arrowhead would still have had a lipid drug. What it would not have had is clean evidence that two payloads can share the architecture without one becoming materially compromised. Future dimers will still be engineering projects. They no longer begin with the unanswered question of whether two functional triggers can coexist on the architecture at all.

3. What the lipid numbers say

Downstream of the gene silencing, the numbers a cardiologist cares about came out well.

LDL cholesterol down 54 percent. Triglycerides down 73 percent. Non-HDL cholesterol down 61 percent. ApoB down 50 percent.

James Hamilton pointed at the last one and he was right to. Because each atherogenic lipoprotein particle carries one ApoB molecule, ApoB estimates the total burden of those particles rather than the cholesterol sitting inside one class of them. In a patient whose problem is both high cholesterol and high triglycerides, treating one number can leave much of that burden intact. Fifty percent off ApoB from a single injection is the number that shows most clearly how broadly the treatment reduced that burden.

The comment from Steven Nissen in the release is worth reading for what it highlights. Even with intensive statins and a PCSK9 inhibitor, substantial risk remains in these patients, and the triglyceride-rich remnants may be part of what is left. That is a cardiologist saying the current standard of care does not finish the job.

4. Why one to two is the expensive jump

Here is why this matters beyond a lipid drug, and it comes down to where the difficulty sits.

Going from zero triggers to one was solved years ago and every approved siRNA drug does it. Going from one to two asks something genuinely new. Does a single conjugate carry two different payloads to the same cells. Do both get loaded into the silencing machinery. Does one crowd out the other. Does the molecule survive being twice as complicated.

Going from two to three still raises engineering questions. A third trigger could bring new competition for the silencing machinery, new potency or chemistry constraints, new manufacturing problems. It is a different kind of question now, though. Before this morning, the open issue was whether two triggers could share one architecture and both stay functional.

That first multiplexing boundary looks crossed. The next one has not been tested.

5. What this does to a pipeline

Many of the diseases medicine struggles with are not single-target problems. They are networks, with several genes or pathways contributing at once. Building one small molecule that precisely modulates several chosen targets is extraordinarily hard. Giving a patient a separate biologic for each is technically possible and quickly becomes an exercise in dosing, toxicity, manufacturing and cost.

Worth being precise about what that does and does not mean. This is not polygenic risk in the genetic sense, where hundreds of common variants each make small contributions and there is no practical way to address the whole distributed signal with a drug. It is the narrower case where a handful of known targets in the same tissue, reachable through the same delivery chemistry, each matter, and treating one leaves the others running.

RNAi offers a different possibility. Target recognition is encoded largely in sequence, and if multiple sequences can share one delivery architecture without losing activity, multiplexing starts to look less like combination therapy and more like molecular programming. Until this morning that proposition rested on chemistry and preclinical work. Now there is a human dataset behind it.

Which makes the interesting question not what ARO-DIMER-PA does for mixed hyperlipidemia. It is which combinations Arrowhead nominates next, and whether the patent estate already tells you.

It does, partly. US-20260176631-A1 is directed to hepatic delivery platforms carrying multiple RNAi agents on one conjugate. US-20250376688-A1 is directed to PCSK9. US-12365899-B2 covers APOC3. A further filing covers dual inhibition of both targets together, which the company lists in its annual report as its APOC3 and PCSK9 dimer group. Arrowhead did not discover this morning that dimers might work and then go looking for protection. The architecture was mapped first.

6. What got de-risked, and what did not

The biological question, whether both targets can be knocked down at the same time, looks answered. The architectural question, whether two triggers can share one conjugate and both stay active, now has human evidence behind it. The major product questions remain open: dose, durability, repeat dosing, safety at the intended regimen, and eventually whether any of it prevents a heart attack.

The biggest de-risking this morning was architectural, not commercial.

Four things, and they matter more than the ones being quoted.

No time points. Mean maximal reduction is the deepest number reached, not the number three months later. For a drug meant to be given quarterly, the trough matters more than the nadir and it is not in this release.

No dose attribution. Escalation completed through 400 mg. Whether 72 and 88 came from the same dose, or whether the best PCSK9 number and the best APOC3 number came from different cohorts, changes how you read the symmetry.

No numbers per cohort. The study enrolls up to 78 subjects across single and multiple dose portions. Early cohorts are small and these are means.

The multiple dose portion is still running. Which is where a quarterly regimen actually gets defined. A single-dose result tells you the architecture works. It does not tell you what the product looks like.

One thing the release does report and this note has otherwise passed over. The most common adverse events were injection site reactions and headaches, with no drug-related serious adverse events, and escalation completed through 400 mg. That is reassuring for this stage. A small single-dose dataset cannot establish the safety profile of the repeat-dose regimen the drug would actually be given on.

Additional detail goes to a medical congress, which is where these questions get answered.

7. What I would watch from here

Whether a second dimer gets nominated in the months ahead, and what it pairs. That will be the first test of whether Arrowhead treats this morning as a one-off product or a reusable capability.

Whether the durability supports quarterly dosing when the multiple-dose data arrives.

Then whether anybody else attempts it. Arrowhead now has a human result and a patent estate around the architecture. The interesting signal over the next year is how many other companies start talking about multi-target RNAi, because that tells you whether the field thinks this generalizes.

A drug that lowers two lipids is a product. A molecular architecture that can silence two independent targets at once is a capability.

Arrowhead reported the product this morning. The capability may end up worth more.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions. All ARO-DIMER-PA figures are from the company release of September 15, 2026, which reports interim single-dose topline results. Comparisons to inclisiran and to APOC3-targeting drugs are across separate trials with different populations and designs. Patent filings referenced are as published. The framing of the architectural problem is the author's own.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The trade nobody remembers

In July 2008 Roche bought a small Madison company called Mirus Bio for $125 million. Mirus had spun out of University of Wisconsin research in 1995 and had spent more than a decade on nucleic acid delivery, work that predates therapeutic RNAi entirely. That expertise turned out to be unusually valuable once RNAi arrived and the field ran into its defining problem, which was getting the molecule into the right cell.

The year before, Roche had paid Alnylam $331 million for access to the field. Add the milestones and the development spending and Roche had committed something close to half a billion dollars to becoming a leader in RNAi.

In 2010 they changed their minds and exited the field entirely.

Figure 1. Transaction figures as reported at the time. The 2026 market capitalization is approximate.

In October 2011 Arrowhead acquired Roche’s Madison operation. Not the technology alone, but an operating research site, its equipment, the intellectual property, the licenses from Tekmira and Alnylam, and a team of more than forty scientists.

Those figures are not comparable purchase prices and I do not want to present them as such. What they show is the scale of what Roche spent assembling a capability Arrowhead ended up inheriting.

Arrowhead paid no cash at closing. Roche took a promissory note of fifty thousand dollars and an equity stake of just under ten percent. Arrowhead subsequently recorded total purchase consideration for the transaction at roughly $5.27 million, comprising about $5.13 million in shares, the note, and a small amount assigned to contingent consideration.

That is not the same as paying nothing, and it is worth being precise. Roche also kept rights to negotiate for certain future products, milestone payments that trigger only after regulatory approval, and low single-digit royalties on some sales. The consideration was real. Much of what Roche stood to receive was contingent on success that had not happened yet, which meant Arrowhead committed very little cash at a moment when it had very little cash to commit.

Arrowhead’s own document explaining the deal said it more plainly than I could. “Fortunate to stand on these broad shoulders.”

2. The part that actually mattered

Everyone focuses on the price, which was remarkable. The more interesting question is why the team was still there to acquire.

Picture the same thing happening in Cambridge. Roche announces it is shutting its RNAi site. Forty scientists with deep delivery expertise hit the market on a Tuesday. By Friday they have offers from Alnylam, from Dicerna, from half a dozen startups, from whatever startup raised a Series A that month. The building gets subleased. The knowledge scatters across the ecosystem and stops being a platform.

Now picture it in Madison in 2011. Roche announces the same thing. Where does a delivery chemist go?

This is not a place without scientists. It is a place without competing employers, and those are entirely different problems.

There was almost nowhere local for an RNAi delivery team to go, and that may be the most important part of the story. Isolation did not make the team cheap. It made the team intact.

When Arrowhead arrived with stock and a plan, they were not recruiting forty individuals. They were acquiring a group that had worked together for years and would have been difficult to reassemble anywhere.

The handoff shows up in the personnel records. Within a month of closing, Arrowhead named David Lewis and David Rozema to run biology and chemistry in Madison. Both had held those functions under Roche, Lewis as site head and director of research, Rozema as director of delivery chemistry. The company also granted inducement options to thirty-seven selected new employees at the Madison facility. The intellectual property changed owners and a good deal of the operating organization crossed the bridge with it.

The conventional model treats talent density as the asset. Arrowhead’s history suggests the inverse can also hold. When the product is a platform rather than a molecule, talent stickiness is worth something too.

There is a way of framing that which I think generalizes beyond this company. A patent makes knowledge harder for a competitor to appropriate. Geography can make a team harder for a competitor to take apart.

The moat was never that the scientists could not leave. It was that leaving usually meant changing cities rather than changing badges, and friction is what lets an advantage last long enough to compound.

Geography did not create the capability. It gave the capability time to compound.

3. Why continuity is what compounds

A delivery platform is not a molecule and it is not a patent. It is accumulated knowledge about what works, what fails, and why, most of which never gets written down properly.

Which linker survives the bloodstream. What happens when you change the sugar. Why the third version worked and the second did not. That kind of understanding lives in the people who generated it, and it only compounds if those people keep working on the same problem together.

Boston is optimized for talent liquidity. You can staff a program in a month. You can also lose it in a month. That trade works perfectly well when the value sits in an asset. It works much less well when the value sits in accumulated organizational knowledge.

The Madison lineage was doing the reverse. Decades on the same underlying problem, across three different owners. Under Arrowhead alone, fifteen years of iteration carried that work from liver to lung, muscle, fat and brain, each step informed by the chemistry and delivery work that came before it.

For that kind of company, the ability to hire quickly matters far less than the difficulty of being raided.

4. The same distance, opposite value

Roche’s RNAi effort was spread across Madison, Kulmbach and Nutley inside a global company headquartered in Basel. Madison was not an afterthought, and it is worth correcting the easy version of this story. Roche itself described Madison and Kulmbach as centers of excellence for RNA therapeutics.

What distance did do was put the decision somewhere else. When Roche made a portfolio-level judgment to leave the field, the fate of the Madison organization was settled by people who were not in Wisconsin, and the sites in Germany and New Jersey went the same way.

Under Arrowhead the same geography produced the opposite effect. Madison went from being one node inside a global pharmaceutical company to being a core scientific operation inside a much smaller one, while the thin local labor market likely helped hold the group together long enough for that to happen.

One property, opposite effects, depending entirely on who controls the capital allocation.

5. Building outside the cluster

In December 2021 Arrowhead bought thirteen acres in the Verona Technology Park for just under three million dollars. On that land it built a GMP manufacturing plant and a lab and office building, together running to roughly three hundred thousand square feet. The investment was estimated at two hundred to two hundred fifty million at announcement. By the end of 2025, with the build-out substantially complete, Arrowhead reported costs incurred of approximately $298.5 million.

Figure 2. Figures from company announcements and the Wisconsin Economic Development Corporation.

Look at the land number again. Thirteen acres for under three million dollars. There is effectively no comparable transaction available in Cambridge or South San Francisco at anything resembling those economics.

The other half of the ledger is the direction the money flows. Verona authorized up to sixteen million in tax increment financing toward site improvements, repayable out of future tax increment. The state offered up to two and a half million in refundable credits contingent on job creation and capital spending. Neither is a check handed over at closing. Both are public money pointed at a campus Arrowhead owns.

In a place where everybody wants to be, you pay a premium to get in. In a place that wants you, they pay you to stay.

Arrowhead also owns the plant rather than renting capacity from a contract manufacturer. For a company whose platform advantage depends heavily on chemistry and delivery, having the people who make the molecule in the same organization as the people who design it is not a cost saving. It is a feedback loop.

6. The university nobody mentions

The weakest objection to Madison is that there is no talent there, and it is worth taking apart because it is simply wrong.

The University of Wisconsin has been doing serious nucleic acid science for most of a century. Har Gobind Khorana did the work there that helped crack how the genetic code turns RNA into protein, and won a Nobel for it. Howard Temin discovered reverse transcriptase there, which rewrote what everyone thought they knew about how genetic information moves, and won a Nobel for that. James Thomson derived the first human embryonic stem cell lines there.

None of that is a direct lineage into siRNA delivery and I would not pretend otherwise. What it establishes is narrower and sufficient. Madison is not a scientific wilderness, and it has not been one for a very long time.

The university also runs one of the oldest and most successful technology transfer operations in the country, the Wisconsin Alumni Research Foundation (WARF), founded in 1925. The blood thinner warfarin is named after it. Mirus Bio itself spun out of university research in 1995, which is how the delivery expertise that later became central to RNAi took root in Madison.

The state development agency put Arrowhead’s Madison research headcount at roughly two hundred ten in 2022, in a hundred and eleven thousand square foot facility, with intern and trainee pipelines into local institutions. Arrowhead has also disclosed research use of a primate colony housed at the Wisconsin National Primate Research Center, a university affiliate. That is not a company scraping for staff in a scientific vacuum. It is a company sitting inside an ecosystem.

Chris Anzalone talks about the place in the language of a relationship rather than a location decision. “A dedicated member of the biotech community in the greater Madison area” for more than a decade. “Strong local relationships.” A “productive and mutually beneficial relationship” with the local business community for many years.

That is the vocabulary of somebody who has been somewhere long enough for it to matter, which is the asset this whole paper is about.

7. The costs, which are real

None of which makes this free.

Arrowhead is a split company. The executive office is in Pasadena, the delivery chemistry and manufacturing in Wisconsin, more research in San Diego. That is a genuine organizational cost and anyone who has worked across time zones knows it.

Senior recruiting is harder. An accomplished executive who has built a life in the Bay Area will not casually move to Dane County, and Arrowhead has to pay up or wait longer to fill those roles.

The informal information flow is thinner too. In Cambridge you learn what a competitor is doing because somebody’s spouse works there. Arrowhead does not get that, which cuts both ways since nobody learns what Arrowhead is doing either.

8. What a buyer would actually be getting

This is where the argument starts mattering to anybody modeling a bid for this company.

Platform acquisitions carry a peculiar risk. The buyer can acquire every asset and still slowly lose the platform. A large company buys a small one for its capability, integrates it, moves reporting lines, folds the labs into an existing structure, and over a couple of years the people who actually held the capability in their heads take their payout and go. What is left is buildings, patents and an org chart.

Every acquirer of a platform company is really buying a bet that the people stay. The question is not what Arrowhead owns. It is what walks out the door on the Monday after closing.

In Cambridge that bet is a retention package, and retention packages have a known expiry. In Madison the bet is a fact about the map.

There is also an unusually clean natural experiment here, of a kind few platform companies can point to. The Madison organization has already been through one of the harsher versions of what an acquisition can do. Roche bought Mirus in 2008, ran the Madison organization for a little over two years, then reversed strategy and walked away from the field.

The transfer to Arrowhead did not happen until late 2011, and that gap may be the most telling part. The owner had already abandoned the field, and enough of the organization stayed coherent that somebody could still acquire it as an operating group. That is a change of ownership followed by abandonment, which is the failure mode buyers worry about, and the group came out the other side intact enough that Arrowhead retained dozens of them and put two of Roche’s own site leaders back in charge.

Most teams have never been tested. This one has been acquired, abandoned, transferred, and is still working on the same underlying problem in the same city nearly two decades on.

The corporate structure helps too, in a way that looks accidental and is worth noticing. Arrowhead keeps its executive office in Pasadena, its manufacturing and delivery chemistry in Wisconsin, and further research in San Diego. In almost any acquisition the synergies come from eliminating duplicate corporate functions, and those sit in California. A core part of what a buyer would want to preserve sits two thousand miles from the corporate functions it might want to cut.

You can take out a headquarters without anybody in a lab noticing. That is not true of a company where finance and chemistry share a cafeteria.

Which means the geography that looks inefficient in a standalone company becomes unusually efficient in an acquisition. A buyer would not need to relocate the Madison delivery organization or work out where manufacturing should live. Those questions have already been answered by the map, and the answer is Wisconsin.

Then there is the plant. Dedicated GMP capacity for this class of medicine is not easy to come by, and companies without their own remain dependent on outside manufacturers and their schedules. A buyer acquiring Arrowhead gets roughly a hundred and sixty thousand square feet of owned manufacturing built for exactly this chemistry, plus the process development group that knows how to run it. For anybody planning to launch more than one product, that is not a real estate line. It is control over a critical part of the launch schedule.

Roche had already written the playbook, which makes the Madison story stranger rather than simpler. When it acquired the rest of Genentech in 2009 it deliberately kept research and early development as an independent center in South San Francisco, explicitly to preserve the culture that had produced the pipeline. Roche understood the value of organizational continuity. A year later it simply concluded that RNAi itself no longer justified the investment, and Madison went with the decision.

The cheapest way to avoid breaking the factory is to buy one that is already standing somewhere you were never going to move it to.

9. The outpost

Chess has a square worth understanding here.

An outpost is a square deep in the opponent’s half where you can plant a knight and no enemy pawn can ever attack it. Put a knight there and it sits for the rest of the game, controlling squares, impossible to dislodge, growing more valuable with every piece that gets traded off.

The square itself is nothing special. It is a square. What makes it an outpost is that nothing can drive the piece away.

Madison is not a better place to do science than Cambridge. It is a place where a team that is already good is hard to dislodge, and over fifteen years of Arrowhead ownership that may have turned out to be the more valuable property.

The scientific lineage running through Madison can be traced through Arrowhead’s expansion out of the liver and into everything after it. The chemistry that reaches liver cells. The ligand that reaches the airway. A molecule given as a shot under the skin, now being tested in people for whether it can reach the central nervous system and silence tau.

Roche bought the team, funded it, and gave up on the field. Arrowhead took the square and never moved the piece.

Boston and South San Francisco are where you go to hire quickly. Nobody planned Madison as an alternative to either. It simply turned out that a company trying to accumulate one capability for the better part of two decades needed somewhere that would hold still.

Talent density is worth a great deal if you are building a drug. It is worth much less than continuity if you are building a factory that makes them.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. Transaction details are from contemporaneous reporting and from Arrowhead’s own disclosures and filings at the time of the 2011 acquisition. The consideration comprised a promissory note, restricted common stock, negotiation rights, post-approval milestones and royalties rather than cash at closing, recorded by the company at approximately $5.27 million. Verona campus figures reflect the single project announced in 2021, with costs incurred through December 2025 as reported by the company; incentive amounts are authorized maximums contingent on performance rather than payments received. Arrowhead also operates research facilities in San Diego and its corporate office in Pasadena. The characterization of why the Madison team remained available is the author’s interpretation and not a claim about any individual’s decisions.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. Three notes, one week

Morgan Stanley, Stifel and Bank of America all published fresh commentary on Arrowhead within days of the ESC presentations. All three are positive. All three carried price objectives set earlier in the summer, at $120, $104, and $100, and none of them moved.

I read all three looking for what they disagreed about. What struck me instead was what they had in common, and it is not a disagreement at all. It is a shared limit in the tool.

2. What each of them does

Morgan Stanley runs a single discounted cash flow. That means projecting the company’s future cash and discounting it back to today, in their case at 10 percent a year with 1 percent growth after that. There is no breakdown by drug and no disclosure of what odds are being assigned to any of them. The $120 is one number produced by one model, and a reader cannot see which assumption is carrying it.

Stifel says the quiet part out loud and then does it anyway. Condulis writes that the market gives the brain program, the dual-target molecule and the two obesity assets essentially no credit. He is right. He then builds a valuation that gives those same programs what he describes as highly risk-adjusted credit, which is a polite way of saying not much. He identifies the mispricing and then only partly corrects for it.

Bank of America shows its work, which is why its note is the most useful of the three. Gerberry breaks the valuation into pieces. The APOC3 franchise, meaning the approved drug plus the wider patient population it is applying to sell into, carries about half the entire company. The partnered liver program carries 9 percent. The two obesity programs carry 6 percent.

Everything else sits in one bucket worth about a quarter of the company, marked down on the assumption that each program has less than a 30 percent chance of working.

Figure 1. Composition as described in the published note. Percentages are approximate.

Look at what is inside that quarter. The brain platform, a systemically delivered route into the central nervous system with extensive primate data and its first human clinical readout arriving this month, with a patent listing thirty-four targets behind it. The dual-target molecule reading out this month. Two inhaled antiviral programs. Whatever else has not been announced.

More than twenty clinical programs spanning liver, lung, muscle, adipose and now the central nervous system, on a delivery platform with an unusually broad tissue footprint. Half the value sits in one drug and the brain program shares a quarter-slice with everything the company has not announced yet.

3. Why this happens, and it is not carelessness

These three notes do not all use the same method, and that is part of the point. What they share is a modeling convention: value what can be modeled, discount it for risk and for time, and leave out what cannot be modeled.

The most common implementation is called risk-adjusted net present value, where you value each drug separately. Morgan Stanley instead runs one company-wide cash flow. Different arithmetic, same convention underneath.

The idea is simple and it is sound.

You take each drug, estimate the cash it would generate if it works, multiply by the odds it works, and discount the result back to today. Add up all the drugs and you have a company. A cancer drug in Phase 1 might get 10 percent odds. An approved drug gets close to 100. It is disciplined, it makes you say out loud what you are assuming, and for most biotech companies it is exactly the right tool.

The trouble starts when a company’s biggest asset is not a drug at all.

Every discount in that calculation gets applied to something that already exists. A molecule with a name, a trial running, a date on a calendar. Conventional practice has no line for the thing that produced those molecules and will produce the next twenty. An analyst could create one. Nothing in the arithmetic forbids a line for the platform, for how productively it generates candidates, or for what people call optionality, meaning the value of being able to do something you have not committed to yet. It is simply that doing so requires assumptions that are hard to defend in print. The line therefore does not get created, and the ability to invent the next drug carries no explicit value at all.

As conventionally built, risk-adjusted valuation prices the inventory and not the factory. For most biotech companies that distinction does not matter, because there is no factory.

4. Undrilled acreage

There is an industry that built a language for exactly this problem decades ago, and I think the comparison is worth walking through.

Nobody values an oil company by adding up the cash from wells that are pumping today. That would be absurd. It would price a company sitting on an enormous untapped field exactly the same as one that has already pumped its last barrel.

The industry built categories instead:

1. Wells that are producing.

2. Reserves that have been found but not yet developed, where you know the oil is there and have not built the infrastructure.

3. Prospects, which are locations the seismic work has identified as worth drilling but where nobody has drilled.

4. Then undrilled acreage, which is land you hold and have not yet surveyed.

Each of those categories can carry value, and acreage itself regularly changes hands at a price per acre. Companies pay real money for the right to drill holes that may find nothing, because the option to find something is worth something before you know.

Now put Arrowhead into those buckets.

Figure 2. Categories adapted from oil and gas reserve reporting. The mapping is illustrative.

Plozasiran is producing. Approved in five geographies and generating commercial revenue. A valuation counts it, and it should.

The clinical pipeline is discovered but not yet developed. ARO-MAPT, the dual-target molecule, the two obesity programs, zodasiran. Real trials, real timelines, with delivery demonstrated clinically for the lipid and obesity programs and preclinically for the brain one. A valuation counts these too, heavily discounted, which is reasonable.

Then it stops counting.

Here is the part I think gets missed. Arrowhead holds a patent listing thirty-four specific targets its brain delivery system is designed to carry. Not a vague ambition to work in neurology. Thirty-four named genes, written into a filing, covering Huntington’s disease, three genetic forms of ALS, four inherited ataxias, prion disease, Parkinson’s, chronic pain and more.

That number, however, is easy to run away with. A list of targets in a patent tells you what a company thinks its technology could be pointed at. It does not tell you there are thirty-four drugs sitting in a drawer somewhere. Nobody has shown that each of those genes can actually be drugged, that a molecule against it works, or that the resulting medicine would sell.

What it does establish is that somebody has done the survey work. In oil terms these are mapped prospects: locations identified as worth drilling, written down with coordinates, sitting behind a delivery route whose first human data arrive this month. That is not raw land.

They should not be valued as thirty-four drugs. Treating thirty-four identified applications of a heavily validated preclinical delivery system as carrying no explicit value today is also an assumption, and it is the one conventional modeling effectively makes.

The patent estate is the map. The brain patent is the one I have read closely. It is not the only one of its kind, and once you look at the whole estate the structure is unmistakable. It maps onto the oil categories almost exactly.

Arrowhead reports roughly 643 issued patents and 833 pending applications worldwide, across more than a hundred patent families. For the purposes of this analogy, two recurring kinds of filing matter most.

One kind claims a route into a tissue. GalNAc sugar clusters for liver. αvβ6 integrin ligands for inhaled delivery to the lung. Antibodies against the transferrin receptor, TfR1, for the central nervous system, which is the door ARO-MAPT walks through. Then skeletal muscle delivery platforms and lipid conjugates for adipose tissue, where the published titles name the tissue rather than the chemistry.

Five tissues, five dedicated delivery families, each with its own filings. That is the acreage side of the portfolio. Arrowhead is staking out the routes it intends to protect.

The second kind is target-specific. Filings aimed at a particular gene, or a defined combination of them, appearing as individual programs emerge. APOC3, ANGPTL3, alpha-1 antitrypsin, PNPLA3, HSD17B13, Lp(a), complement C3, complement factor B, XDH, factor XII, MARC1, PCSK9, INHBE and more in the liver. Alpha-ENaC, beta-ENaC, MUC5AC, RAGE, MMP7 and TSLP in the lung. DUX4 in muscle. ALK7 in adipose. MAPT in the brain. Influenza A and coronavirus for the inhaled antivirals.

Better than thirty named genes with their own filings, across five tissues plus the antiviral work.

Somebody should push back here, because listing targets in a patent is what patent attorneys do. The marginal cost of adding another contemplated target to a broad disclosure is tiny next to the cost of actually developing it. Broad biotech target lists can reflect lawyering as much as development intent, and treating one as a pipeline would be foolish.

What makes this one different is the conversion record. Arrowhead has done the same thing in tissue after tissue. Liver now has more than a dozen targets with dedicated patent filings. Lung has six. Adipose has ALK7, where the company has already shown direct knockdown in human fat tissue, and there is no obvious reason a working route stops at one gene. Arrowhead does not stake out a route and then work a single target in it. It works the ground.

That is the difference between a list and a record. A list tells you what somebody thought to write down. A demonstrated history of turning names into programs makes it reasonable to expect that at least some of the remaining names will become programs too.

Those are the wells. The brain estate is where you can actually watch one get drilled.

Several of the thirty-four have already made that trip. SOD1 sits on the brain delivery target list and has its own dedicated filings. Worth stating that ARO-SOD1 was discontinued before the planned Phase 1 study enrolled, so this is not a success story. It still progressed from a name on the platform map to dedicated IP and a clinical-stage program, which is the conversion I am pointing at. Huntingtin has made the same trip, the gene behind Huntington’s disease. MAPT has as well, which is the one reading out this month. Prospects on a map became locations somebody decided were worth drilling, and the paperwork followed them across.

That is the machine, visible in the filings. Delivery patents stake out ground. Target patents get filed as the company works through it. Better than thirty genes now have filings of their own, and the brain list still holds most of its thirty-four.

The dual-target molecule reading out this month is visible in the estate too, though the filings did not arrive in a tidy conceptual order. APOC3 had its own target filings years ago. PCSK9 came later. A separate family covers hepatic delivery platforms carrying multiple RNAi agents on one conjugate. Then a dedicated filing for dual inhibition of both targets at once, which the company lists in its annual report as its APOC3 and PCSK9 dimer group. Which patent came first is not the point. The point is that the estate holds all three layers: the individual targets, the architecture for carrying more than one at a time, and eventually the specific combined drug.

A conventional valuation counts the wells. It does not count the ground, and it does not count the fact that the company keeps walking across it.

Nearly nineteen years of assembling all of it, and little of that capability appears explicitly in a conventional valuation, because there is no individual drug to attach a probability to.

5. The best argument against me

There is a serious case on the other side.

Platform value is exactly the sort of claim that gets abused. Every biotech with two molecules calls itself a platform company. Most of them are not. The graveyard is full of delivery technologies that worked once and never again, and an analyst who declined to pay for optionality was right far more often than wrong.

Analysts cannot model what does not exist. Refusing to assign value to an unnamed future drug is not a failure of imagination. It is discipline, and the alternative is a valuation that can be justified at any number you like.

Optionality is also unfalsifiable. If I say the platform is worth several billion dollars and you disagree, neither of us can settle it. If every unnamed future target can be invoked to justify today’s price, platform analysis stops being valuation and becomes storytelling. That is a genuinely bad property for a valuation input, and I understand entirely why a professional putting their name on a published number would rather leave it out.

6. What survives that

Three things.

The blind spot comes from the convention, not from a judgment that the platform is worthless. Three firms took different routes and arrived in the same neighborhood. That is not three analysts independently concluding the pipeline is worth little. It is a shared convention that tends to arrive there by construction, because standard practice gives the arithmetic nowhere to put platform optionality unless an analyst deliberately builds a separate category for it.

The analysts themselves know it. Stifel wrote that the market gives these programs no credit. That is not a stray observation. His own model then gives them some heavily risk-adjusted value, but only partly repairs the gap he had just identified. When somebody identifies a pricing failure and then only partly corrects for it in the same document, the constraint is more likely in the framework than in the analyst.

There is a third thing, and I hold it more loosely. Acquirers run these calculations too, so the difference is not that they use a different arithmetic. It is that a strategic buyer can pay for things a published asset-by-asset model has no room for, including competitive position, what the target denies a rival, and what the organization might produce next. When Roche bought the rest of Genentech it paid roughly $46.8 billion for the 44 percent it did not own, implying a total above $100 billion.

Genentech had enormous approved products by then, so I would not claim Roche was paying purely for the engine. What is more suggestive is what Roche chose to preserve after buying the rest. Genentech research and early development continued as an independent center inside Roche, explicitly to protect the culture and the research approach that had generated the pipeline. Roche bought the products. It also took unusual steps not to break the factory.

Everything described so far is public. A pharmaceutical company can pull the same patents, study the same conversion record and reach its own view about what the remaining ground is worth. The asymmetry does not require privileged information.

What differs is the burden of explanation. A buyer can debate a platform premium internally. An analyst who lets the same assumption materially move a published price target has to explain it to clients and live with it in the model months later.

The buyer still faces diligence, return hurdles, integration risk and internal approval, and a serious bidder may later see things public investors never do. None of that is needed to notice the optionality in the first place.

That is the strange part. The value does not have to be concealed to be hard to express. It can simply sit in a category that published research handles awkwardly and a private strategic judgment handles more comfortably.

7. What better would look like

Not a bigger number. A visible one.

If a note showed each asset separately with its own probability of success attached, a reader could argue with the pieces. If it carried a line, even a small one, for programs the company has demonstrated it can generate but has not yet named, a reader could argue with that too. The number might come out lower than $100. That would be fine.

Somebody is going to ask what a platform line would even be built out of, and that is a fair question. It does not require pretending thirty-four drugs already exist. It could start with things that are actually observable: how often this company has produced a clinical candidate, how long each one took, how reliably its delivery chemistry has carried from one target to the next inside a tissue, and what a program has historically been worth once it reaches the clinic.

Then discount all of it heavily, because the targets are unnamed and most of them will fail.

The point is not the number that comes out. The point is that zero is a number too, and for programs that have not yet been named and modeled, it is the one conventional practice uses by default.

The problem is not that the resulting values are too low. It is that when half the value sits in one drug and the entire brain franchise is inside a bucket labelled everything else, there is nothing specific to disagree with.

That standard should apply to my own work as much as anybody’s. When I put a number on Arrowhead in an acquisition context, I showed three possible buyers, what each could afford, what holds each of them back, and how a contested sale would end up setting the price. Anyone who thinks I am wrong can say which of those is wrong. That is the bar, and I would rather be visibly wrong than opaquely right.

8. Why this matters right now

Two readouts land this month. One asks whether a single molecule can switch off two genes at once. The other asks whether a shot under the skin can switch off a gene inside a living human brain.

Both of those sit inside the 25 percent bucket.

If they work, the thing that changed is not that two drugs got better odds. DIMER working tells you something about DIMER and something about whether Arrowhead can build molecules that hit two targets at once as a general matter. MAPT working tells you something about MAPT and something about whether a shot under the skin can produce meaningful target knockdown in the human central nervous system at all.

A drug readout updates the probability of one drug. A platform readout updates the probability distribution of drugs that do not exist yet. Conventional models naturally register the first and usually leave the second implicit.

The market finds out this month whether the land has oil under it. The question is whether the models treat that as information about two wells, or as information about the field.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions. Patent counts reflect the author's review of Arrowhead patent families, with each target family counted once regardless of jurisdiction or continuation. Valuation methodologies and composition figures are as described in published research notes from Morgan Stanley, Stifel and Bank of America following the ESC Congress, and percentages are approximate. Nothing here should be read as a criticism of any individual analyst's competence or integrity. The Genentech transaction value is as reported at the time.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. What happened

Novartis announced this afternoon that its big Lp(a) trial failed. Across 8,323 patients with established cardiovascular disease and high Lp(a), pelacarsen failed to reduce the combined risk of cardiovascular death, heart attack, stroke and urgent coronary procedures versus placebo.

Read the next sentence of their release carefully, because it is the whole story. Lower Lp(a) levels were achieved with pelacarsen.

The drug did exactly what it was designed to do. It lowered the thing. Lowering the thing did not improve outcomes in the overall study population.

2. Why this is a hypothesis failure and not a modality failure

The distinction matters and it is going to get blurred in the coverage over the next few days.

Pelacarsen is an antisense drug, which is a different chemistry from Arrowhead’s RNAi but aimed at the same job of silencing a gene’s message. Ionis discovered it and licensed it to Novartis. Nothing here says either chemistry is broken. The molecule got to the liver, reduced production of apolipoprotein(a) and kept Lp(a) low in thousands of people for years.

What failed is narrower than the target itself. HORIZON refuted the proposition that the depth and duration of Lp(a) lowering pelacarsen achieved, in this treated secondary prevention population, produces a detectable cardiovascular benefit. That is a specific claim and it is now dead. Whether Lp(a) is a viable target at all is a wider question this trial does not fully settle.

Twenty years of genetic evidence said Lp(a) causes cardiovascular disease. That evidence appears to have been right about causation and wrong about reversibility, at least in this population and at this level of lowering.

Genetics measures what happens to somebody who carries elevated Lp(a) from birth. HORIZON asked something different. It asked whether lowering it late, after atherosclerosis is already built and while statins and blood pressure drugs are doing their work, undoes enough of that accumulated history to change what happens next.

Those are not the same experiment, and causation does not guarantee reversibility. The industry has spent a long time treating them as though it did.

3. What it costs Arrowhead directly

Olpasiran is Arrowhead’s own Lp(a) drug. Arrowhead built it and licensed it to Amgen back in 2016, and Amgen is now running OCEAN(a)-Outcomes, which tests the same idea with Arrowhead’s chemistry instead of Ionis’s.

The economics are less exposed than they look, because of a decision made four years ago.

Figure 1. Amounts as disclosed in company filings and the November 2022 agreement.

In November 2022 Arrowhead sold its entire olpasiran royalty interest to Royalty Pharma for $250 million in cash, plus up to $160 million in milestones payable back to Arrowhead. The first of those, $50 million on completion of OCEAN Phase 3 enrollment, was collected in 2024.

That means $300 million has been received. The filings state plainly that Arrowhead is not obligated to repay it.

What is now in doubt is the remaining $110 million from Royalty Pharma, since both remaining triggers require an approved and selling drug, and much of the roughly $375 million in Amgen milestones still outstanding, particularly the regulatory and sales portions, which Arrowhead retained when it sold the royalty.

Arrowhead sold a royalty stream in 2022 for $250 million it never has to give back, four years before anybody knew whether the drug worked. Royalty Pharma bought the risk. This afternoon that looks like one of the better capital allocation decisions this management team has made.

I would not oversell it as foresight. Monetizing a partnered royalty to fund your own pipeline is a reasonable thing to do regardless of how the trial turns out. The trade was available to be made badly, though, and it was not made badly.

4. What happens to OCEAN(a)

Amgen almost certainly finishes. Enrollment closed in 2024, the trial runs to roughly seven thousand patients, it is event-driven, and the money is spent. Companies do not stop trials at this stage because a competitor missed.

The probability of success dropped hard this afternoon, though, and anyone telling you otherwise is selling something.

OCEAN(a) is not a carbon copy, and the differences deserve a fair hearing. It enrolled 7,297 patients at a higher Lp(a) threshold. Olpasiran produces substantially deeper suppression, above 95 percent at the higher doses in Phase 2 against roughly 80 percent for pelacarsen. Its primary composite is not identical to HORIZON. Novartis also wrote that the findings did not demonstrate reduced risk in the overall study population, which leaves the prespecified 90 mg/dL subgroup unaddressed.

Those differences are enough to keep the experiment worth running. They are not enough to pretend the prior probability has not moved. The surviving argument is that the effect needs sicker patients or deeper suppression than pelacarsen delivered, which is not absurd and is also what every program in this position says. I would want the subgroup data from the congress before giving it much weight.

5. The part that matters more than the money

Set the milestones aside. They are small against a twelve billion dollar company.

What happened this afternoon is the most expensive demonstration available that moving a biomarker is not the same as helping a patient.

Eight thousand three hundred and twenty-three people. Years of follow-up. A drug that unambiguously lowered the marker it was built to lower, in a target with two decades of human genetic evidence behind it, in a population where the marker is present in roughly one person in five. Then no demonstrated benefit.

Chess has a word for this. A refutation is a concrete demonstration that a line which looked perfectly sound actually loses. It does not mean the opening is bad or the pieces are wrong. It means that particular sequence has been worked out to the end and does not hold. Theory gets rewritten and everybody stops playing it.

This version of Lp(a) lowering was theory. Eight thousand patients was the refutation. The pieces are fine. That line is not.

6. What this should do to how you read September

Arrowhead reports first human data on ARO-MAPT at the end of this quarter or early next. The number everybody will trade on is tau reduction in cerebrospinal fluid.

That number tells you the drug got where it was going and reduced production of the protein it was built to target. That would demonstrate something RNAi has not previously established clinically, which is systemic delivery to the human central nervous system from a simple shot under the skin.

It is also exactly the kind of number pelacarsen produced.

I wrote earlier this week that a drug readout updates the probability of one drug while a platform readout updates the probability of drugs that do not exist yet. Both of those are still true. What this afternoon adds is the other half of the same idea. A biomarker readout establishes target engagement, not patient benefit, until somebody connects the two.

That connection is exactly what CELIA has begun to suggest for tau without yet establishing it. Diranersen lowered cerebrospinal tau by 50 to 65 percent and produced a clinical signal, and it missed its primary dose-response endpoint because higher doses did not produce greater benefit. For neurofilament in ALS, the FDA accepted a reduction as reasonably likely to predict clinical benefit, which is a regulatory judgment rather than a proven link. For Lp(a), Novartis has just spent eight thousand patients failing to establish that connection in the overall population.

If September delivers deep tau knockdown, that is a delivery result and it is worth a great deal to the platform. It is not, by itself, evidence that ARO-MAPT helps anybody with Alzheimer’s disease. Today is a reminder of how expensive it is to learn the difference.

7. Where that leaves things

For Arrowhead, a modest financial hit against milestones that were never in the base case, and a partnered program whose odds just got materially worse.

For Ionis, a harder afternoon. They lose the pelacarsen economics on top of already competing against plozasiran in severe hypertriglyceridemia.

For Novartis, worth watching. They face the largest patent expiry in their history and just lost a cardiovascular asset that was meant to help fill it. That does not reduce their need to buy something. It sharpens it.

For anybody holding this stock into a first-in-human CNS readout, a well-timed reminder that the number in the headline and the question that matters are not always the same number.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper. Lp(a)HORIZON details are from the Novartis release of September 4, 2026. Olpasiran deal terms are from the November 2022 Royalty Pharma agreement and Arrowhead's subsequent quarterly filings. Full Lp(a)HORIZON results have not been presented and the analysis here is based on topline disclosure only.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

The question I get more than any other about ARO-MAPT is not how much tau it will lower. It is how quickly this could reach patients, and whether it could qualify for accelerated approval.

This note is an attempt to answer it properly.

The conventional answer is grim. Alzheimer’s trials measure how fast a patient declines, and demonstrating a convincing difference generally takes many months to years, with eighteen to twenty-four months typical for pivotal programs. A program built that way puts approval somewhere in the middle of the next decade. Fast Track designation does not change that. It buys more meetings with the FDA and can make a program eligible for rolling review, meaning the application is submitted in pieces rather than all at once. It does not, by itself, turn a conventional efficacy trial into a shorter one.

There is a faster route, and it is not hypothetical. The FDA can approve a drug based on a biomarker that is reasonably likely to predict clinical benefit, with the trial proving it actually helps patients finished afterward. Two Alzheimer’s drugs have come through that door. A neurology drug almost nobody outside the field has heard of came through the same door.

The difference between how those cases went is the whole subject of this note, and it points at a number that is not currently being discussed.

1. A word on who actually decides

One thing has to be explained first, because both stories below turn on it.

When the FDA reviews a drug, it can convene an advisory committee. That is a panel of outside experts, clinicians, statisticians and often a patient representative, who read the application, hear the company and the agency present, and then vote in public on specific questions.

The vote is advice. The FDA is not bound by it and has gone against it before.

There is a second detail that matters more than it sounds. The agency writes the questions. It decides what the committee is asked to vote on and what stays off the agenda. Hold onto that.

2. How it goes wrong

On June 7, 2021, the FDA approved aducanumab for Alzheimer’s on the basis of reduced amyloid plaque.

Its own advisory committee had been unenthusiastic to the point of hostility. On the central question, whether it was reasonable to treat the one positive trial as primary evidence of effectiveness, the vote was 0 yes, 10 no, 1 uncertain. Its own statistical reviewers had rejected the application. Within days of the approval, three members of that committee resigned, one describing it as probably the worst drug approval decision in recent United States history.

The detail that matters most is procedural rather than scientific. None of the four questions put to the committee asked whether reducing amyloid plaque met the standard for accelerated approval. According to committee members writing afterward, the lead FDA scientist told them during the meeting that the agency was not using amyloid as a surrogate for efficacy. Seven months later the agency approved the drug under the accelerated approval pathway on the basis of amyloid reduction.

What followed was worse for the company than the controversy. The Centers for Medicare and Medicaid Services restricted coverage to patients enrolled in clinical trials. Major hospital systems declined to administer it. The drug was eventually withdrawn.

An approval that leaves the vast majority of eligible patients without coverage is not much of a commercial approval. The lesson of aducanumab is not that the surrogate was wrong. It is that a surrogate never argued in the open does not survive contact with the people who decide what gets reimbursed.

3. How it goes right

On April 25, 2023, the FDA approved tofersen for a genetic form of ALS under accelerated approval, based on a drop in a blood marker called neurofilament light chain, or NfL.

The Phase 3 trial had missed its primary endpoint, and not narrowly. The measure was a functional rating scale covering things like speech, swallowing, walking and breathing. Drug and placebo separated by 1.2 points on it, with a p-value of 0.97.

That last number deserves a translation. A p-value estimates how often a result this size would show up by chance if the drug did nothing. Below 0.05 is the conventional threshold for calling a result real. At 0.97, the trial was not close to the line. It was about as far from it as a result can get.

The advisory committee, meeting a month earlier, was asked two questions rather than one. On whether the drop in NfL was reasonably likely to predict clinical benefit, the vote was 9-0 in favor. On whether the clinical results were convincing evidence that the drug worked, the vote was 5-3 against, with one abstention.

Read that again. The committee rejected the clinical evidence and endorsed the biomarker, in two separate recorded votes, and the FDA approved on the biomarker. Nobody resigned. The review team’s own document notes that it identified no issues precluding accelerated approval.

The confirmatory trial, running in people who carry the mutation but have not yet developed symptoms, has a primary completion date in 2027 and runs into 2028.

4. The difference

Aducanumab and tofersen offer two sharply contrasting accelerated approval precedents, less than two years apart. One collapsed commercially. The other remains intact while its confirmatory trial runs. The distinction runs along two lines, and both matter for tau.

The first is what the biomarker measures. Amyloid plaque reduction tells you the pathology the drug was aimed at has changed, and the FDA does treat plaque reduction as reasonably likely to predict clinical benefit. It sits close to the therapy, in the same compartment the drug was designed to act on. NfL sits further out. Nerve cells release it when their fibers are damaged, so a falling NfL suggests the rate of neuroaxonal injury is coming down. It is not a measure of whether the drug hit its target. It is a measure of what is happening downstream of it.

Tofersen had a target engagement measure available. Its developers tracked the SOD1 protein in spinal fluid, which tells you directly whether the drug hit its target. That is not what the approval rested on.

The FDA did not approve tofersen because the randomized trial established clinical benefit. It did not. It approved tofersen because the evidence taken together supported NfL as reasonably likely to predict that it would.

There is a third case that belongs here, because it complicates the neat version of this story.

Lecanemab received accelerated approval in January 2023 on the same amyloid surrogate that had gone so badly for aducanumab. Six months later it converted to traditional approval, because its confirmatory Phase 3 demonstrated clinical benefit. Same biomarker, same pathway, opposite outcome.

Which means the lesson is not that pathology biomarkers fail and injury biomarkers succeed. The lesson is that a surrogate has to sit inside an evidentiary package that makes the clinical inference credible. Aducanumab had a contested surrogate, conflicting trials and a regulatory process that drew extraordinary criticism. Lecanemab had the same surrogate and a clean confirmatory result. Tofersen had a downstream marker, a coherent mechanistic story and a public committee endorsement.

The second difference is procedural. Tofersen’s surrogate was written into the questions, argued in public, and won a unanimous vote on the record. Aducanumab’s surrogate was never put to the committee as a question, and then became the basis for approval. One approval had a public record of expert endorsement behind it. The other did not. That does not mean a committee endorsement would have changed the reimbursement decision, and CMS grounded its restriction in insufficient evidence of improved health outcomes rather than in anything procedural. It does mean the approval entered the reimbursement debate without a public expert record validating the surrogate the FDA had ultimately relied on.

5. What CELIA just did to the tau surrogate

Which brings this to July, and to a result that has been read almost entirely as good news for tau. It is good news for the mechanism. It is bad news for the simplest version of the surrogate argument, and almost nobody has said so.

Diranersen, a drug that lowers tau, reported eighteen-month Phase 2 results in 416 patients with early Alzheimer’s. Total tau in spinal fluid fell 50 to 65 percent across all three dosing schedules. Tau imaging showed reductions across all doses while placebo rose. On the cognitive measures, the lowest dose arm slowed decline by 26 percent on one clinical scale, 42 percent on another, and 50 percent on a third. Worth noting immediately that this arm had 60 patients in it. Impressive percentages from small groups deserve to be read with the group size attached.

Caveats aside, that is the strongest randomized evidence yet that lowering tau might change the course of Alzheimer’s disease. The trial did not prove it, and it missed the endpoint it was built around. It is still the most encouraging thing this mechanism has produced, and it de-risks the biology for everyone working in it, including Arrowhead.

Now the problem. Higher doses produced greater tau reduction without producing greater clinical benefit. The cognitive measures favored the drug across all doses, but more tau lowering did not buy more slowing. The trial’s primary endpoint was an assessment of dose response, and on that measure it failed.

Sit with what that means to a regulator. A drug lowered tau robustly, in a dose-dependent way, confirmed by imaging. The strongest clinical signal came from the lowest-dose group, even though the higher doses produced greater tau reduction. Whether that reflects an optimal dosing window, a non-linear exposure relationship, a difference between the tau being measured and the tau doing damage, or simply noise in a 416-patient study, nobody yet knows.

If somebody asked the FDA today to accept the size of a spinal fluid tau reduction, by itself, as evidence that a drug will help patients, CELIA is the obvious counterexample, and there is currently no good answer to it.

Tau reduction, on this data, behaves like target engagement. It tells you the drug reached the brain and did its job. It does not, on its own, tell you the patient will be better off. That is one reason the tofersen precedent is particularly interesting for a tau-lowering program.

6. What Arrowhead would need

All of which produces a specific checklist rather than a general hope. Six things, roughly in order of how hard they are.

Deep tau reduction that lasts. The obvious one, and the one everybody is watching. It is necessary and nowhere near sufficient.

Two things make it harder than it sounds. A 50 percent reduction in healthy volunteers does not guarantee the same magnitude in somebody with established tau pathology, so the healthy volunteer number cannot simply be assumed to carry over. A reduction that fades between doses is also worth much less than one that holds, because a patient spends the back half of every dosing cycle drifting back toward where they started.

Figure 1: Where the evidence for each link currently comes from. Only the first is what September tests.

The NfL trajectory bending, and considerably more than that. This is the item nobody is discussing, it is the center of the whole case, and the tofersen file shows the bar is far higher than simply getting the number to move.

A surrogate is a measurement that stands in for the thing you actually care about. What anyone actually cares about in Alzheimer’s is whether the patient can still recognize their family in three years. You cannot wait years for that outcome every time you test a drug, so you look for something that moves earlier and reliably predicts it.

Tau reduction is not quite that, on the current evidence. It tells you the drug got in and did its job.

Chess has a word for the position tau is now in. A piece is pinned when it cannot usefully move, because moving it would expose something more valuable sitting behind it. The piece is still valuable. It simply cannot do the work by itself, and the reason has nothing to do with the piece.

That is where the tau number stands after CELIA. It is the figure management benchmarked, the figure September reports, and the figure the market will trade on. Sitting behind it is the ultimate question that matters, which is whether patients end up better off. Tau does not answer that question on its own, and diranersen just demonstrated it by lowering tau further without producing greater clinical benefit.

The consequence matters more than the metaphor. A bigger tau number does not fix this. Pushing harder on a pinned piece accomplishes nothing, because the problem was never how hard you pushed. What breaks a pin is bringing another piece to bear, and the pinned piece then becomes useful again.

NfL is the different piece. It is a marker of axonal injury and neurodegeneration, released when nerve fibers are damaged. When NfL falls, it suggests the rate of neuronal injury is falling. It is not a measure of whether the drug hit its target. It is a measure of what is happening downstream, in the disease itself. Tofersen’s accelerated approval centered on a blood-based biomarker, NfL, which is why the file matters so much here.

One caveat belongs up front. In SOD1-ALS, axons die quickly and in large numbers, so the NfL signal is loud and moves within months. Alzheimer’s is slower and the elevation is milder. The FDA accepting NfL in one disease does not commit it to accepting NfL in the other, and the agency will say so. Other markers have done it, amyloid in Alzheimer’s and dystrophin in muscular dystrophy among them. NfL is unusual because it measures downstream neuroaxonal injury rather than target engagement.

Now the complication, and it is a real one. The only reported human trial of a tau-lowering drug that I am aware of measuring NfL did not find what this argument needs.

In the Phase 1b study of diranersen, published in 2023, CSF NfL was tracked as an exploratory measure. Eight weeks after the last dose it showed no dose-responsive effect. NfL fell in the placebo group and in the lowest dose group, and rose slightly in the higher dose groups. The authors said plainly that the changes were not dose responsive and not concordant, and that assay variability and small sample size limited what could be read into them.

Anyone building a case on NfL has to deal with that, so here is the case.

The study enrolled 46 people across five arms. Individual dose groups had six to thirteen participants. It looked eight weeks past the last dose. That is a small, short experiment, and its own investigators said so.

More importantly, it may have been asking the wrong question. In somebody who already has Alzheimer’s, substantial neuronal and axonal injury has already occurred, and NfL is elevated because of it. A drug working upstream, at the level of tau production, would not be expected to reverse damage that has already happened. The relevant signal may therefore be a reduction in the rate of new injury rather than an immediate fall below baseline.

The right question is therefore probably not whether NfL falls. It is whether the climb slows.

Picture two lines over eighteen months, in illustrative indexed units rather than real values. On placebo, NfL runs 100, then 112, then 124, then 136. On drug, it runs 100, then 104, then 107, then 110. NfL never drops below where it started. That would still be one of the most important results this field has produced, because it would suggest that the rate of ongoing neuroaxonal injury had slowed.

One further caveat belongs here. NfL is not specific to Alzheimer’s disease. It rises with many kinds of neurological injury, climbs with age, and is affected by kidney function among other things. That raises the bar for using it as a surrogate, though a randomized design with repeated within-patient measurements helps mitigate those problems far better than a single absolute reading does.

Stated precisely, the question is whether the longitudinal slope differs from placebo, not whether the treated group’s absolute NfL value falls. Answering it requires a long trial with repeated measurements, which is exactly what the Phase 1b was not. The longitudinal experiment this hypothesis needs has not actually been reported.

One practical note follows from all of it. NfL can be measured in blood as well as spinal fluid. Plasma carries the practical advantage, because it can be drawn at ordinary study visits and repeated as often as the schedule allows, which lets investigators estimate an individual patient’s trajectory rather than relying on a handful of lumbar punctures. A program serious about this question collects plasma NfL longitudinally and treats CSF as complementary.

What the tofersen file actually required. Biogen’s advisory committee materials are public, and they are the clearest statement available of what the FDA wanted before accepting a novel surrogate. The agency’s own guidance, quoted in those materials, says a reasonably likely surrogate depends on two things: the biological plausibility of the relationship between the disease, the endpoint and the desired effect, and the empirical evidence supporting that relationship.

Biogen met the second requirement with roughly a hundred published citations spanning 1996 to 2023. That literature established, independently of any drug, that NfL is elevated in ALS above nearly every other neurodegenerative disease, that it distinguishes ALS from conditions that mimic it, that it correlates with the rate of decline across multiple separate cohorts, and that it predicts survival. Decades of work by people with no stake in the outcome.

They then did something more demanding than showing the marker moved. They built a model estimating how much clinical benefit each unit of NfL reduction bought. For every 10 pg/mL fall in plasma NfL at Week 16, the model predicted 0.77 points of preserved function on the ALS rating scale at Week 28, with a p-value of 0.0038, alongside parallel estimates for quality of life and for the risk of death.

The bar was not that the biomarker moved. It was a quantified relationship between how far it moved and how much the patient benefited, sitting on top of thirty years of independent literature.

There is a mirror-image piece of evidence in the same materials, and it is the kind of thing that makes a regulator believe a marker. A different antisense drug, aimed at a different ALS gene, raised CSF NfL by 37 percent. The patients receiving it did worse. A biomarker that moves the wrong way and drags outcomes with it is powerful confirmation that it tracks something real.

One process detail completes the picture. Biogen held a formal meeting with the FDA in April 2022 specifically to discuss the utility of NfL for accelerated approval, three months before filing. The surrogate was negotiated with the agency in advance and then ratified in public by the committee. It was not sprung on anybody.

Several markers agreeing, not one. The strongest version of this argument is not a single impressive number.

CELIA’s real problem was disagreement. Tau went down, imaging confirmed it, but the strongest clinical signal came from the lowest-dose group. When the biology and the clinical results point different directions, a regulator has no way to know which one to believe.

The fix is not a better single measurement. It is having tau, NfL, imaging and the cognitive scores all move the same way at the same time. Any one of them can be a fluke. Four of them agreeing is a story.

A dose-response that behaves. Regulators look for an exposure-response relationship that makes biological sense. It does not have to be a straight line, and more drug does not have to produce endlessly greater benefit, since real drugs plateau and therapeutic windows exist. What it does have to be is interpretable.

CELIA is hard to interpret on exactly that basis. Across the doses studied, increasing tau reduction did not translate into increasing clinical benefit. That single fact is the hardest thing in its dataset to explain and it is why the trial formally failed. A program that shows benefit rising with exposure has an argument available to it that diranersen currently does not.

The surrogate argued in public. The lesson from aducanumab, where the committee never voted on the surrogate the approval ultimately rested on, is not optional. It is also cheaper to follow than to skip.

For the strongest possible regulatory and reimbursement case, a tau surrogate should be written into the questions put to an advisory committee, debated openly, and endorsed on the record. The FDA is not obliged to convene a committee at all. The argument here is not that it cannot approve without one, but that after aducanumab, nobody should want a contested surrogate approval without transparent outside validation behind it. Approving first and explaining later is what happened in 2021, and the people watching most closely were not at the FDA at all.

They were at the Centers for Medicare and Medicaid Services, which decides whether Medicare pays. CMS is a separate agency with a separate process, and after aducanumab it restricted coverage to patients enrolled in clinical trials. An approved drug that Medicare will reimburse only under highly restrictive conditions, in a disease of the elderly, is not much of a commercial product. Winning the FDA argument and losing the CMS one is the worst outcome available.

The confirmatory trial already running. Accelerated approval is conditional. The company gets to sell the drug while it runs the study that proves the biomarker was telling the truth, and if that study fails, the approval can be withdrawn.

For years, companies collected the approval and let the confirmatory work drift. Congress has since given the FDA explicit authority to require that the confirmatory study be underway before accelerated approval is granted, and current agency policy strongly favors starting that work early rather than waiting until after approval. That pulls a great deal of expensive work forward, which is exactly the point.

The opportunity is not to replace tau with NfL. It is to connect them. Tau shows that ARO-MAPT is doing what it was designed to do. NfL could show that what it is doing matters to the disease.

7. What the current trial is not designed to do

Worth being clear about where things actually stand.

Public trial registries require a company to state, in advance, what a study is designed to measure. For the ongoing ARO-MAPT study, the single primary measure is side effects, tracked through day 270. The secondary measures are twelve readings of how the drug moves through the body, meaning how much gets absorbed, how long it lasts, how it clears, plus three routine safety checks on spinal fluid: total protein, glucose, and cell count.

Neither tau nor NfL appears in that list. That is a statement about what the study is formally designed to prove. It is not a statement that either biomarker is going unmeasured.

That is not a criticism. This is a first-in-human safety and pharmacokinetics study, and it is registered as one. Management has said the update will include tau knockdown in healthy volunteers, so tau is presumably being tracked as an exploratory measure. That is entirely normal, and exploratory biomarker measurements do not necessarily appear among a registry’s primary and secondary outcome measures.

The point is what it tells you about the number the market will trade on. It will arrive as an exploratory result, meaning a measurement taken out of interest rather than one the study was built to prove, from a trial designed to answer a different question, quite possibly in a small number of participants. It is the beginning of the regulatory argument, not evidence that can settle it.

Which also means the question worth asking is forward-looking. Once adequate target engagement is established, whether NfL is being collected now, and whether Arrowhead plans to carry it into the patient cohorts and the Phase 2 design, tells you more about the timeline to approval than the precise magnitude reported in September.

8. The faster door is not Alzheimer’s

There is a route to an accelerated approval that runs around the Alzheimer’s problem entirely, and Arrowhead’s own language already gestures at it. The program is described as being for Alzheimer’s disease and other tauopathies.

The other tauopathies may come first.

Progressive supranuclear palsy is a primary 4R tauopathy. Amyloid is not a defining pathology, unlike in Alzheimer’s disease. It moves fast and you can measure it. It is rare enough that trials run in the low hundreds instead of the thousands. There is no approved disease-modifying treatment. The protein the drug silences is the defining pathological protein in the disease, which makes it an unusually clean test of the mechanism.

Tofersen is the useful analogy here, though the analogy is regulatory rather than biological. A rare, fast-moving, well-defined neurodegenerative disease with major unmet need, approved on a biomarker after a Phase 3 that missed, is a far more applicable precedent for progressive supranuclear palsy than anything in the amyloid history. The disease biology, the biomarker kinetics and the trial endpoints are not interchangeable between the two, and the regulatory argument would still have to be built from scratch.

A tau drug that changes the course of a rare tauopathy could give the FDA a much stronger human precedent for accepting tau biology as part of the regulatory case in the common one. The rare disease is not the consolation prize. It is the argument.

9. The odds, and what would change them

Making a case for something is not the same as predicting it, so here are the odds as I actually see them.

Accelerated approval for ARO-MAPT in Alzheimer’s disease is unlikely. My estimate before CELIA was somewhere around ten to fifteen percent, and CELIA moves it in both directions at once, strengthening the mechanism while weakening the simplest version of the surrogate argument. I would leave it roughly where it was.

That number is not arbitrary, and it is worth showing the parts. Getting there requires deep and durable tau knockdown in patients, which I would put at better than even. It requires a measurable separation in the NfL trajectory, which I would put below even, given the limited human evidence to date rather than because the biology argues against it. It requires the FDA to accept NfL as a surrogate in Alzheimer’s specifically, having only ever accepted it in ALS, which I would put well below even. It requires the biomarkers and the cognitive data to point the same direction. It requires the confirmatory program to be underway early enough to satisfy the FDA’s accelerated approval expectations.

One distinction is worth drawing inside that list. The scientific probability and the regulatory probability are not the same thing, and the second is lower. Even a convincing NfL result would still leave Arrowhead having to establish that NfL is a valid surrogate in Alzheimer’s disease specifically, which is a separate argument from having produced good data.

The chain is long enough that even reasonably plausible probabilities at each step would produce a low-teens result overall. That is a conceptual probability tree rather than a calculated model, and it is offered as one. Any one of those links failing could close this particular route, which is why the number stays small even though several of the individual steps are more likely than not. It would not necessarily close every route, since a different biomarker package, unexpectedly strong clinical data, or a different indication could each open another.

A convincing NfL dataset in patients would move that number more than anything else, plausibly doubling it. That is a real if, given that the one reported tau-lowering trial I am aware of that measured NfL did not deliver it, and it is the single most important thing to watch over the next three years. Convergence across several markers and a clean dose-response would matter nearly as much.

An accelerated approval in progressive supranuclear palsy may be a materially better opportunity than the Alzheimer’s version, and could arrive years earlier if pursued in parallel from the start.

The reimbursement question sits behind all of it. The FDA approving a drug is not the same as patients getting it, and aducanumab is the reason that sentence has to appear in any honest note on this subject.

10. Where this leaves the September readout

The September number is a first step and will be treated as a verdict.

What it can establish is human central nervous system target engagement after subcutaneous dosing, meaning that systemic dosing produced a measurable pharmacodynamic effect within the central nervous system. If that had not worked, nothing else would have mattered. That is worth a great deal, and it is worth more to the platform than to this particular drug.

What it cannot establish is whether lowering tau helps a patient, because CELIA has just demonstrated that the two questions come apart.

The most credible path I can see to patients getting this drug early runs through a biomarker that was not in the headline, in a disease that is not Alzheimer’s, argued in front of a committee that has not yet been convened.

The case for early approval is real. It simply does not run through the number everyone will be looking at in September.

A Note on Supporting Independent Research

If this white paper has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,500 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this white paper accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this white paper; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This paper is provided for informational and analytical purposes only. It does not constitute investment advice, financial advice, legal advice, or a recommendation to buy, sell, or hold any security, and it is not a recommendation as to any corporate course of action. The author holds a long position in Arrowhead common stock. Past performance is not indicative of future results, and forward-looking analysis is inherently uncertain. The author and BioBoyScout are not registered investment advisors. The author assumes no obligation to update this paper.

About the Author

BioBoyScout is the publishing name for Robert Toczycki, an independent biotech investment research writer based in Chicago. The BioBoyScout series publishes institutional-grade analysis of structural dynamics in RNA-class therapeutics, with particular focus on Arrowhead Pharmaceuticals’ TRiM platform and the broader competitive landscape. Robert is a registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. The presentation nobody went to

Nine in the morning on Monday, in a session called Beyond statins, a physician from Peking Union Medical College Hospital in Beijing presented the first Phase 3 results anywhere in the world for zodasiran.

It appeared in the program under Visirna’s development code, VSA003, rather than zodasiran or ARO-ANG3, the name most Arrowhead investors know it by. Arrowhead mentioned it in one sentence on its investor call an hour later. As far as I can tell, almost nobody wrote about it.

The disease is homozygous familial hypercholesterolemia (HoFH). About as rare as diseases get, somewhere between one in 170,000 and one in 300,000 people.

Here is what these patients are dealing with. Your liver clears cholesterol out of your blood using a receptor on the surface of its cells. These patients inherit defects that leave that receptor pathway severely impaired, in some cases close to nonfunctional. Statins work largely by making the liver produce more of those receptors, so in this disease there is much less receptor function to work with. Untreated, patients develop heart disease as children.

The trial enrolled 46 people in China, 30 on drug and 16 on placebo. Average starting LDL cholesterol was around 390 milligrams per deciliter. A normal number is under 100.

2. The number, and who to measure it against

LDL cholesterol dropped 44.8 percent from where patients started, and 43.9 percentage points more than placebo, at six months. ANGPTL3, the protein the drug is designed to suppress, fell 89.4 percent from baseline, which is 86.2 points more than placebo.

Two comparisons make that number mean something, and neither is obvious from the press coverage.

The first is Arrowhead’s own earlier study. A Phase 2 called GATEWAY, run in a similar population, produced a 35.7 percent LDL reduction at this same dose. The China Phase 3 beat it by about nine points.

That should get your attention. It would have been reasonable to expect some fade in a larger, properly blinded Phase 3, particularly coming off an open-label Phase 2 where everybody knew who was getting the drug. GATEWAY was open-label. This one was blinded. The effect got bigger anyway.

The second is the drug this would compete with. Evinacumab is an approved antibody for this disease, it hits the same target, and in its pivotal trial it lowered LDL by 47.1 percent from baseline. It is given by intravenous infusion, once every four weeks.

Zodasiran came in at 44.8 percent on the same measure. Roughly two points apart. It is a shot under the skin, once every three months.

Figure 1. Both figures are reduction from baseline. They come from separate trials in different populations and are not directly comparable.

3. Why those two points are worth giving up

Thirteen infusions a year against four injections. That sounds like a convenience argument until you think about who is receiving them.

This disease gets diagnosed in childhood. Treatment never stops. Plenty of these patients are also on apheresis, which is a procedure that filters cholesterol out of the blood mechanically, and it means hours hooked to a machine every week or two on top of everything else. The average patient in this trial was in their thirties. Four of them were teenagers.

An evinacumab infusion needs a clinic, an appointment, a nurse, an IV line and most of a morning, every four weeks. A quarterly injection is a fundamentally smaller procedure. Even if every zodasiran dose were given in a clinic rather than at home, that is four brief visits a year instead of thirteen infusions. Spread across forty or fifty years of treatment, that stops being about convenience.

Chess has a name for this kind of trade. Giving up the exchange means handing your opponent a rook, the more valuable piece, in return for a bishop or a knight and a better position. On the scoresheet you have lost material. Tigran Petrosian built a world championship partly on these, and for years both computers and commentators undervalued them, because material is easy to count and position is not.

Zodasiran gives up roughly two points of cholesterol lowering. It gets back nine infusion visits a year, every year, for the rest of somebody’s life. Anyone scoring this on the efficacy number alone is counting material and ignoring the position.

Worth adding: two thirds of the patients in this trial were already on a PCSK9 inhibitor, and the drug worked on top of it.

4. What the slides left out

Two things are missing from the presentation, and I think both are informative rather than sinister.

The responder rate stops at 30 percent. The trial design slide says they planned to report the share of patients who cut LDL by at least 30 percent and by at least 50 percent. The results give you the 30 percent number, which was 83.3 percent of treated patients. The 50 percent number never appears. In a deck built to show a drug at its best, the omission suggests that number was not the headline.

Goal attainment is missing too. The presentation does not report how many patients reached a specific LDL target, and a group average cannot answer that question, because patients did not all start in the same place or respond by the same amount. What the average does tell you is how brutal this disease is. The treated group started around 383 milligrams per deciliter, and even a 45 percent cut leaves the average patient well above what guidelines want for an adult with this condition. Some individuals may well have reached goal. The slides do not say how many.

This should eventually get a real answer. The global Phase 3, called YOSEMITE, lists the proportion of patients getting below 100 milligrams per deciliter as a secondary objective. It is fully enrolled at 70 patients with completion expected around the middle of 2027.

5. The liver signal

Start with the context. Overall side-effect rates were essentially identical, 76.7 percent on drug against 75.0 percent on placebo, and there were no drug-related serious adverse events at all. Against that backdrop, one thing stands out.

Three patients on drug had liver-related adverse events flagged. None on placebo did.

Two were mild elevations in a liver enzyme called AST, three to five times the normal ceiling. One was moderate, above five times, and the investigator judged it probably unrelated to the drug. All three cleared up within six weeks. Total bilirubin stayed below twice the normal ceiling in every case, and that matters: a substantial bilirubin rise alongside those enzyme elevations would raise considerably more concern for clinically significant drug-induced liver injury.

The slide separately notes two moderate drug-related events in patients whose liver enzymes were already abnormal before they started. Both were still unresolved when the database closed.

Three events in thirty patients is a small number and a real one. I would watch it rather than worry about it, and I would want the imaging. An earlier zodasiran study measured liver fat directly by MRI and found it moved in the favorable direction. This trial did no imaging at all, and that is a gap I would want closed somewhere in the broader zodasiran safety package.

One detail says somebody was paying attention. Worsening blood sugar in diabetic patients was written into the protocol in advance as something to watch for specifically. Zero events, both arms.

There was one death, sudden cardiac death in a 61-year-old woman with several risk factors, judged unrelated to treatment. In a population carrying extreme lifetime cardiovascular risk, a sudden cardiac death is unfortunately not unexpected. Somebody will still quote it without the context.

6. The number worth watching

Lipoprotein(a) came in 28.9 percentage points lower than placebo, with a p-value of 0.0016. Lp(a) is set by your genes, barely responds to anything, and independently drives cardiovascular risk. A reduction that size from a drug not designed to touch it would be genuinely interesting.

I would hold it loosely. The placebo group started with much higher Lp(a) than the treatment group, 148.8 against 116.5, which is a real imbalance in a 46-person trial. The spread around the estimate is wide. The pivotal antibody data against this same target did not show anything close to this magnitude of Lp(a) lowering.

Worth watching in the global study. Not worth building anything on yet.

7. The pattern, which is the actual story

This is the second time in two days that Arrowhead presented a Phase 3 result resting on the same foundation.

Sunday: plozasiran matched or beat an approved competitor on triglyceride lowering and offered four injections a year against that competitor’s twelve.

Monday: zodasiran came within roughly two points of an approved competitor on LDL lowering and offered four injections a year against that competitor’s thirteen intravenous infusions.

Different disease, different target, different competitor, different continent. One sentence describes both. Comparable efficacy, radically better delivery.

That is what a platform looks like when it is working. Raw potency is not the common denominator here. Efficient delivery and durability are. Arrowhead is getting a molecule where it needs to go and producing gene silencing durable enough to dose on a schedule a patient can actually live with, which is the thing that has always been hardest in this field.

8. What it is actually worth

Very little directly. Quite a lot indirectly.

The direct commercial opportunity is not the point. China has roughly 5,000 patients with this condition, and only about four percent of them are on any cholesterol-lowering therapy at all. Greater China rights are licensed to Visirna, which is Arrowhead’s majority-owned regional vehicle rather than an unrelated partner, so the economics are not entirely someone else’s. Even so, nobody should be modeling meaningful revenue off a few thousand Chinese patients.

What it buys is a positive, blinded Phase 3, run separately by Visirna in China, on the same molecule in the same indication as the global YOSEMITE program. Zodasiran has had a complicated history, and in this disease specifically, until Monday it was an asset with encouraging open-label Phase 2 data and an open question about the liver. It is now an asset with a blinded Phase 3 that beat its own Phase 2.

That does not make zodasiran approved in Arrowhead’s major commercial markets. It changes what it is reasonable to assume about the trial that would.

9. The chair

Strip out the endpoints and it comes down to something simple.

Somebody with this disease will spend their life in treatment. The question Monday answered was not whether a new drug lowers cholesterol, because the approved antibody already does that, and slightly better. The question was how much of that life has to be spent sitting in a clinic.

Arrowhead did not present a better cholesterol drug on Monday. It presented one that is nearly as good and asks for a fraction of the patient’s life. That has been the argument all week.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Trial figures come from the presentation delivered at the ESC Congress on August 31, 2026. Comparisons to GATEWAY and to evinacumab are across separate trials in different populations and are directional only. Zodasiran is investigational and has not been approved by any regulatory authority for this indication. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

About the Author

Robert Toczycki is an independent analyst and registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

1. What was actually new

The headline numbers came out on July 22. Triglycerides down 79 and 81 percent in the two studies, pancreatitis events down 78 percent, both trials hitting every goal they had set. The market has had six weeks to think about that.

Sunday added what a press release cannot fit. Who the patients were. What happened to them month by month. Which side effects showed up and how often. The details behind the averages.

Buried in there is one chart that changes how you should read the whole program, and I have not seen anyone write about it yet.

2. The chart that matters

Patients got four injections, spaced three months apart, at Months 0, 3, 6 and 9. The trial then measured its main result at Month 12.

Sit with that timing for a moment. Month 12 is three months after the last shot. That is the point in the cycle when the drug has had the longest time to wear off, in the fourth cycle of the year, when any accumulating problem would have shown up.

It is, in other words, the point in the dosing cycle where you would expect the drug to look its weakest.

Every drug looks good right after you take it. The question that decides whether a once-every-three-months schedule actually works is what the patient looks like just before the next injection is due.

Here is the answer. At Month 10, triglycerides were down 84 percent in both studies. At Month 12, they were down 79 and 81 percent.

Roughly four points given back from Month 10 to Month 12, as patients reached the end of the three-month dosing interval in the fourth cycle of treatment. That is not meaningful wear-off. That is a remarkably flat curve through the end of the quarter.

That number does real work, because olezarsen is injected monthly. Twelve shots a year against four.

Before Sunday, the advantage of quarterly dosing was mostly a convenience argument, and a slightly hand-wavy one, since nobody had shown what happens at the end of the quarter. After Sunday, the durability needed to support that schedule is demonstrated. The drug holds at the exact point in the interval where meaningful waning should have been easiest to see.

3. Head to head with olezarsen

One fact has to sit underneath this entire comparison. Olezarsen was approved by the FDA for severe hypertriglyceridemia on June 24, six days ahead of its June 30 decision date. It is on the market for this condition right now, and any read of the SHASTA data that skips that is incomplete.

A caution goes with it. These two drugs have never been tested against each other. Everything below compares separate trials, run in somewhat different populations, and olezarsen measured its main result at six months where plozasiran measured at twelve. The gaps are large enough to be informative. They are not head-to-head evidence.

How do the two drugs actually stack up?

Triglyceride lowering: plozasiran, and it is the consistency that stands out. Olezarsen ran two pivotal trials as well. In the first, its two doses lowered triglycerides 63 and 72 percent more than placebo. In the second, the same two doses managed 49 and 55 percent. Plozasiran produced 79 and 81 percent across its two trials.

Look at what happened to olezarsen’s higher dose between studies: 72 percent, then 55 percent. That is a seventeen-point swing in the same drug at the same dose. Plozasiran moved two points. Whatever explains the difference, and trial populations and placebo performance can account for a good deal of it, the contrast is striking. Plozasiran’s replication across its two studies is unusually clean.

Getting patients to a safe number: plozasiran. Percentage reduction tells only part of the story. What a doctor actually wants to know is how many patients ended up below the danger line. At Month 12, 91 and 93 percent of plozasiran patients were under 500 mg/dL, the threshold that defines severe hypertriglyceridemia. Olezarsen reported 86 percent.

More striking is the normal range. Over half of plozasiran patients got below 150 mg/dL, which is a normal triglyceride level. In one of the two placebo groups, 2 patients out of 98 managed that. Two.

Pancreatitis: olezarsen looks slightly better on paper, and I would not make much of it. Both drugs cut pancreatitis attacks sharply. Olezarsen reduced them about 85 percent, plozasiran about 78 percent. The ranges around those two estimates overlap heavily, and olezarsen enrolled roughly 1,100 patients against Arrowhead’s 757, which buys you a tighter estimate regardless of how good the drug is. The honest read is that both drugs prevent pancreatitis and neither trial can tell you which does it better.

What plozasiran did produce is a striking number in the sickest patients. Among those who had already suffered a pancreatitis attack, close to 4 in 10 on placebo had another one within the year. On plozasiran it was roughly 1 in 25. That subgroup was small, 35 placebo patients against 60 treated, so the precise figure is soft. The direction is not.

There is one matched comparison worth having, in the sickest patients of all, those above 880 mg/dL with a prior attack. Ionis reported that four such patients would need treating for a year to prevent one attack. Arrowhead reported no events at all in that group. The caveat travels with it: the subgroup was small enough that the company did not publish the underlying numbers, so the direction is more informative than the figure.

Dosing: plozasiran, and Sunday is what turned it into an argument. Olezarsen is monthly. Plozasiran is quarterly, and the Month 10 to Month 12 data show the interval genuinely holds rather than merely being claimed. Four injections a year instead of twelve is a meaningful difference in a group of patients who are already managing a lot. Between 54 and 63 percent of the people in these trials were already diabetic, and roughly two thirds were taking two or more other cholesterol drugs before they enrolled. Every additional appointment is another chance to fall off treatment.

Liver enzymes: plozasiran. Olezarsen’s label lists liver enzyme elevations among its most common side effects. SHASTA showed no meaningful liver enzyme changes compared with placebo.

Figure 1. Every figure is as reported by each sponsor. The two programs were never compared directly.

4. Two places the data are thinner than they look

Two findings deserve more scrutiny than the coverage has given them. Both happen to sit exactly where Arrowhead would most want strength.

Liver fat. The press release says a planned MRI substudy found no meaningful increase in liver fat, with a p-value of 0.70. That reads as a clean result, and it has been repeated everywhere.

The substudy had 36 patients in it. Thirteen on placebo, 23 on plozasiran, at whichever trial sites happened to have the right scanner. Liver fat went up 1.0 percent on the drug and down 0.5 percent on placebo.

The p-value was 0.70, meaning the substudy did not detect a statistically significant difference between the groups. What that number cannot tell you is whether the study was large enough to confidently rule one out.

A p-value of 0.70 in 36 patients does not establish that there is no effect. It says the study did not find one, and with a sample that small, considerable uncertainty remains around the estimate.

This matters more than it might appear. Liver fat was supposed to be the clean differentiator against olezarsen, which showed increases of roughly 2 to 4 percent that grew with dose. Arrowhead’s own chief medical officer said in June that if plozasiran turned out to show something similar, both drugs would be in the same boat.

Sunday did not settle that question. It left it open, with a number that sounds like it settled it.

Blood sugar. Side effects related to worsening blood sugar control were reported in 14.3 percent of plozasiran patients against 8.7 percent on placebo. That is the first real imbalance this program has produced.

The defense is reasonable and it is in the data. Between 54 and 63 percent of enrolled patients were already diabetic, with average HbA1c, the standard measure of blood sugar control over the preceding months, between 6.4 and 6.6 percent. This is a population already at substantial risk of worsening blood sugar, which helps put the 8.7 percent placebo rate in context. When you look at the actual measured HbA1c rather than the reported side effects, it barely moved over twelve months in either group.

A side effect somebody wrote down and a lab value that actually changed are different things, and the lab values are reassuring. Even so, a 5.6-point imbalance in a population with this much baseline diabetes is conspicuous enough that regulators and clinicians are likely to ask about it.

One more figure belongs here, since this section exists to look at the uncomfortable parts. Three deaths occurred among plozasiran-treated patients, two cardiovascular and one from chronic myelomonocytic leukemia. Investigators attributed all three to pre-existing disease and judged them unrelated to treatment. Serious side effects overall were actually lower on drug than on placebo, 8.3 percent against 10 percent, and investigators found no treatment relationship for any of the three deaths. There is no evident mortality signal here, but the events are worth stating anyway.

5. What the call added

Management restated Sunday for most of the hour. The genuinely new material came from the independent discussant, and it did not point the way Arrowhead would have chosen.

Borge Nordestgaard presented unpublished registry data from Denmark covering 3.4 million people between 2008 and 2021, none of whom had cardiovascular disease at baseline. Within them he identified roughly 29,000 with severe hypertriglyceridemia. That group produced 400 cases of acute pancreatitis and 2,000 cardiovascular events. On an incidence basis, 2 per 1,000 patient-years against 18.

Five times as many cardiovascular events as pancreatitis attacks, and nine times the rate once you account for how long people were followed, in patients who had no cardiovascular disease when the clock started. His conclusion was not that the pancreatitis work is misdirected. It was that he would like to see studies going after both diseases, and that the long-term need in this population is cardiovascular.

Arrowhead put an independent expert on its own investor call, and he used the time to point out that these patients are roughly nine times more likely to suffer a cardiovascular event than the event the label is being built around.

That cuts in both directions. On the bearish side, if payers price plozasiran purely on pancreatitis prevention, the addressable market is narrower than headline patient counts suggest. Arrowhead’s own slide identifies roughly one million high-risk patients within the roughly three million Americans with the condition, and the initial commercial focus is that high-risk segment. Management nonetheless put a three to four billion dollar annual US opportunity on the same call.

On the bullish side, plozasiran reduced remnant cholesterol by 72 to 76 percent in these trials. If the drug ever runs a cardiovascular outcomes study and that reduction translates, the opportunity is a different order of magnitude than a pancreatitis label. That remains entirely hypothetical. SHASTA was not designed to demonstrate cardiovascular benefit and does not. Nordestgaard did not say otherwise. He simply pointed at where the events actually are.

What the glycemic number actually counted. An analyst asked what the imbalance actually consisted of, and Watts answered it. The 14.3 percent was not a single measurement. It pooled several loosely related endpoints, among them impaired glucose tolerance and the need to increase antidiabetic medication. On that basis he said it is difficult to make much of a five-point difference. He described the HbA1c change as very minor, not clinically significant, and something that resolves once treatment is intensified, adding that patients typically finish these studies with better control than they started with. Nordestgaard called the safety profile excellent against the risk these patients carry.

How the strongest numbers are constructed. The high-risk numbers arrive in three tiers, and the tiers matter. Across patients with triglycerides at or above 880, or above 500 with a prior attack, the number needed to treat is nine. Among those with a prior attack regardless of triglyceride level, it is three. In the narrowest group, above 880 with a prior attack, no events occurred at all. Watts called that last analysis exploratory and said plainly that the numbers are small. The direction is more reliable than any single figure until the publication lands.

The question management would not answer. The most useful exchange on the call came from Jefferies. Maury Raycroft said his back-calculation showed an imbalance among patients without a prior pancreatitis history, with three to four events in the plozasiran arm against none on placebo, and asked whether there was anything distinctive about those patients.

Management declined, citing the pending publication, and redirected to the point that the pooled endpoint was prespecified and hit statistical significance across the whole population. Both things are true. It is also true that a reader who wants to understand where the benefit concentrates, and where it does not, will have to wait for the manuscript.

A class effect, in his words. One more thing worth recording. Nordestgaard told the call that all five drugs in this class, across every study run so far, have produced roughly an 80 percent reduction in acute pancreatitis. His word was extremely consistent. That is the strongest available argument that neither company should be claiming an advantage on this endpoint, and it came from the independent discussant rather than from either sponsor.

The question nobody asked. Watts described the result as very reassuring, Arrowhead’s commercial lead described it as no liver fat elevation relative to placebo, and the size of the study went unmentioned by anyone on the call or in the questions that followed.

The switching question, answered by the wrong people. Bank of America asked the two physicians whether they saw any barrier to switching a patient from olezarsen to plozasiran. The answers were not the ones the switching argument wants.

Watts said switching happens in practice and that the quarterly interval is a patient preference matter, allowing that it leans in plozasiran’s favor. He also predicted the more likely response to inadequate control on olezarsen is a dose increase rather than a switch, while noting that the higher olezarsen dose carries its own problems.

Nordestgaard was blunter and less helpful to the thesis. Patients who get used to a drug and are having no trouble with it, he said, tend to stay on it, absent clearly more side effects or a real price difference. He then said what he had said twice already, which is that he would rather both companies spent their effort finding the enormous number of untreated patients than fighting over the ones already on therapy.

Management took the same line, framing severe hypertriglyceridemia as an untapped market with room for both drugs rather than a share contest. That is the diplomatic answer and it may also be the correct one. It is not, however, the answer that supports a thesis built on converting a competitor’s patients.

6. The longer route

Here is where this leaves things.

Arrowhead has the better drug on triglyceride lowering, on repeating that result across two trials, on getting patients into the safe range, on dosing schedule, and on liver enzymes. It has a comparable drug on preventing pancreatitis, which is the outcome that actually matters to a patient. It gives four injections a year where the competitor gives twelve.

What it does not have is permission to sell. Olezarsen has been approved for this condition since June. Arrowhead plans to file a supplemental application by the end of 2026, expanding the label on a drug the FDA has already approved rather than seeking a first approval. It will use a priority review voucher bought on August 4, which changes the FDA’s review goal from roughly ten months to six. Filing in December on that timeline points to a decision around the middle of 2027.

Call it a year of a competitor selling into a market before you can enter it. In a specialty field with a few thousand prescribing physicians, a year is enough time for habits to form.

That concentration cuts both ways, though, and the second edge is the one people forget. A small prescriber base lets Ionis establish habits quickly. It also means Arrowhead has a finite and identifiable audience to reach when it arrives, whether by converting patients already on therapy or, as management would prefer, by bringing untreated ones into it. This is not primary care, where changing behavior means reaching tens of thousands of physicians who have never heard of the disease.

Chess has a word for what happens next. A transposition is when two games reach the same position by different move orders. One player gets there down one path, the opponent down another, and once they both arrive, the order stops mattering. What is left on the board is the position.

The analogy is imperfect in a way worth naming. In chess both players arrive at the same moment. Here, Ionis got there a year early and has been using the time.

Pharmaceutical history offers plenty of reminders that arriving first is not the same as staying ahead. Lipitor was the fifth statin to market. Keytruda entered behind Opdivo. Neither case is proof of anything about this one, since market leadership turns on far more than which molecule performs better. What travels across all of them is that physicians can switch when the product difference is real. Nordestgaard’s answer on the call is a useful warning against assuming that dosing frequency alone will move a patient who is doing perfectly well. For a patient who is inadequately controlled, unhappy with monthly injections, or starting therapy for the first time, four injections a year instead of twelve gives Arrowhead an unusually simple conversation to have with a physician.

What Sunday did was hand Arrowhead the material for that conversation. What it could not do was change the order of arrival.

Ionis got there first. Arrowhead got there better. The position on the board is what settles it, and Arrowhead simply took the longer road to reach it.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. SHASTA-3 and SHASTA-4 figures come from Arrowhead's August 30, 2026 release and the slides presented at the ESC Congress Hot Line session the same day. Olezarsen figures come from the CORE and CORE2 results published in the New England Journal of Medicine and from sponsor disclosures. Comparisons across separate trials are directional only; the two programs measured their main results at different timepoints in somewhat different populations. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

About the Author

Robert Toczycki is an independent analyst and registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now

On August 24, Arrowhead confirmed that the twelve-month results from SHASTA-3 and SHASTA-4 will be presented in Munich on Sunday, August 30. The topline came out on July 22 and the stock has already digested it. A scheduling announcement is not usually worth writing about.

This one is, because two of the details in it were decided by the European Society of Cardiology rather than by the company, and both say something about how the field intends to receive this trial.

The market already has the headline. Sunday is about validation, and validation is awarded by people who do not work for Arrowhead.

Hot Line is not the same as late-breaking

The presentation is in a Hot Line session, in the Munich Auditorium, Hall B3, at 17:30 on Sunday. The session is numbered nine, which is a scheduling detail and nothing more. The two words that carry weight are Hot Line.

Congresses run a great deal of late-breaking science. Hot Line is the tier above it, and it runs alongside a separate and larger late-breaking science track. By the society’s own July announcement, this year’s Hot Line program spans twelve sessions and 59 trials. Arrowhead’s slot is in the Munich Auditorium, Hall B3, which is where the Hot Line sessions are held.

The society describes the selection in its own words. Professor Tomasz Guzik, who chairs the ESC Congress Program Committee, said this year brought a record number of late-breaking submissions, and that the committee, in his phrase, rigorously selected the studies with the greatest scientific quality and potential clinical impact.

That is an editorial judgment made by cardiologists reviewing submitted data. It is not a slot a sponsor can buy. The inference worth drawing is bounded but real: a committee that saw a record volume of submissions looked at what Arrowhead sent and put it on the main stage.

The discussant is the tell

Every Hot Line presentation is followed by an assigned discussant whose job is to place the trial in context for the field and say plainly whether it should change what physicians do.

The discussant here is Borge Nordestgaard of Copenhagen University Hospital.

That name deserves explanation for anyone who does not follow lipid research. For decades he has led the Copenhagen population studies. Much of the modern evidence that high triglycerides actually cause heart disease, rather than simply appearing alongside it, comes from that work. When guideline committees debate whether triglycerides are worth treating, his data are central to the discussion.

The society did not assign a drug-development specialist to discuss this trial. It assigned the researcher whose work established that lowering triglycerides should matter in the first place.

That is a choice about framing. It signals ESC is treating SHASTA as evidence bearing on the triglyceride hypothesis itself, not simply as a registration trial for a rare disease drug.

The presenter reinforces the same reading. Gerald Watts of the University of Western Australia has chaired international guideline work on inherited lipid disorders. Between the presenter and the discussant, the society has put two people with real influence over treatment guidelines on the same stage.

Worth noting separately: Arrowhead has both of them on its investor webcast the following morning at 8:00 Eastern, alongside management, taking questions. That does not mean the company knows what Nordestgaard will say. He is independent, and the point of a discussant is that his assessment is his own. What it does mean is that whatever he says, favorable or critical, investors will hear it directly rather than filtered through a press release.

What actually lands on Sunday

The July topline reported median triglyceride reductions of 79 and 81 percent in the two studies, a 78 percent reduction in acute pancreatitis events across the broad severe hypertriglyceridemia population, and a 100 percent reduction in the highest-risk subgroup. Both trials met their primary endpoint and all pre-specified secondary endpoints.

Sunday brings the full dataset behind those numbers, across approximately 750 randomized patients, and the specific thing worth watching is durability.

The primary endpoint is percent change in fasting triglycerides from baseline to Month 12. Patients received four doses, one every three months. Which means the twelve-month measurement is taken at the end of the fourth dosing cycle, when drug effect is at its weakest point before the next injection would be due.

Any drug can look good at peak effect. A quarterly regimen is only real if the effect is still where it needs to be at the end of the quarter, four cycles into a year of treatment. That is what this endpoint measures, and it is the number that determines whether quarterly dosing is a convenience claim or a clinical fact.

Two things not in the release

Liver fat. James Hamilton, Arrowhead’s chief medical officer, confirmed in June that liver fat data would accompany these results, and the competitive stakes are specific.

The rival drug is olezarsen, from Ionis. Both drugs switch off the same gene, APOC3, but they do it with different molecular machinery. Olezarsen is an antisense oligonucleotide. Plozasiran is an siRNA. Same target, different tools.

That distinction is the whole question. Olezarsen showed liver fat increases of roughly 2 to 4 percent that grew with dose. Plozasiran showed none at its commercial dose in mid-stage testing. If the effect comes from suppressing APOC3 itself, both drugs should eventually show it. If it is a property of the antisense chemistry, only olezarsen will. Hamilton has pointed out that another antisense drug against a different lipid target showed the same pattern, which argues for chemistry, and he has also said openly that it may be a combination of both, and that if plozasiran shows a similar increase the two drugs are in the same boat.

Sunday, across 750 patients, is where that question gets a real answer.

Pancreatitis events, case by case. Triglyceride reduction is the primary endpoint. Pancreatitis is the outcome that matters to a patient, and it was pre-specified and pooled across both studies rather than found afterward. It is also the strongest clinical claim in the program, and it rests on a process most investors never see.

A patient in one of these trials turns up at a hospital with abdominal pain. Is that an attack of pancreatitis, or something else? The answer decides whether the event counts toward a pre-specified clinical outcome, and the sponsor cannot be the one deciding. An independent committee of physicians therefore reviews each suspected case and rules on it, separately from the sponsor. The process is called adjudication, and the committee that performs it exists precisely because the company has an interest in the answer.

The full presentation should show that process, the individual events, and the statistical range around a reduction reported as 100 percent in the highest-risk group, where the total number of events is necessarily small. A 100 percent reduction on four events is a different fact from a 100 percent reduction on forty.

The presentation nobody is watching

Monday morning at 9:00 Munich time, in a session titled Beyond statins: the next wave of lipid-lowering therapies, a Phase 3 trial of zodasiran in Chinese adolescents and adults with homozygous familial hypercholesterolemia will be presented by Zhuang Tian of Peking Union Medical College Hospital.

It appears in the ESC program under the designation vsa003 rather than any name Arrowhead uses in its own materials, which is one reason it is easy to overlook.

Homozygous familial hypercholesterolemia is genuinely ultra-rare, with global prevalence estimated between one in 250,000 and one in 360,000. It is hard to treat for a specific reason. Statins and most cholesterol drugs work through a receptor on the surface of liver cells. That receptor pulls cholesterol out of the blood. These patients inherited two broken copies of that receptor, so the usual drugs have little to work with. Zodasiran switches off a different gene entirely, one that does not depend on that receptor functioning.

This is a second late-stage asset presenting late-stage data at the same congress, in a named session, and it has attracted almost no attention. Which is the smaller version of the same point this note began with. The information that moves a company is not always the information the market is looking at, and it is frequently sitting in a program listing that nobody reads.

What this week is

Nothing presented in Munich is new information in the sense that the market has not seen the headline. The topline was July 22.

What happens on Sunday is different from a data release. It is the moment a trial stops being a company announcement and becomes part of the medical literature, presented by a guideline author, assessed in public by the researcher whose own work created the question the drug was built to answer.

Approvals determine whether a drug can be sold. Congresses like this one determine whether physicians believe it should be.

Chess used to settle unfinished games the same way. When a game ran past its time limit it was adjourned, and in many competitions the result could be decided by adjudication, in which an appointed master examined the position and ruled on what it was worth. Both players had already made every move. Nothing on the board could change. What remained was a judge, appointed by the federation rather than chosen by either side, declaring in public what the position was worth.

Arrowhead has made its moves. The data are locked and the topline is five weeks old. One set of adjudicators has already done its work, quietly, ruling case by case on which hospital admissions counted. Munich supplies a different kind, who rules in public on what the whole thing means. Neither side picked either of them.

Regulators have already said plozasiran can be sold in familial chylomicronemia syndrome, the rarest form of this disease. Sunday is about the much larger population behind it, and whether cardiologists believe the drug belongs there. ESC has chosen who tells them.

A Note on Supporting Independent Research

If this note has been valuable to you, whether it shaped your thinking, validated your conviction, or simply saved you the time of doing this work yourself, a voluntary contribution is genuinely appreciated and directly funds the next paper.

For individual investors and readers

Any amount you feel reflects the value you received is welcome and meaningful. A contribution in the range of what you might pay for a single premium research report is a thoughtful gesture that makes a real difference.

For family offices, investment funds, hedge funds, and research platforms

This paper is the caliber of work that institutional research desks bill significant retainers to produce. If your team referenced it, distributed it internally, or used it to inform a position, a suggested contribution of $1,000 reflects the professional value of the analysis, though any amount is meaningful. Your support makes it possible to continue publishing at this level without a paywall that limits the reach of the ideas. If your organization requires an invoice to process a payment, please reach out directly at bioboyscout@gmail.com and one will be provided promptly.

There is no obligation and no expectation. This is purely a thank you for work that meant something to you.

Zelle: (847) 227-7909
PayPal: paypal.me/bioboyscout

Thank you for reading, and for being part of a community that takes this thesis seriously.

— Robert Toczycki | BioBoyScout

Important Risks, Disclosures, & Disclaimers

The author, Robert Toczycki (aka BioBoyScout), certifies that:

• all views expressed in this note accurately reflect his personal opinions about the topic discussed;

• he was not compensated in any form for producing this note; and

• he has not received and does not receive compensation from Arrowhead Pharmaceuticals.

This note is published by BioBoyScout and is intended for informational and educational purposes only. It does not constitute investment advice, a solicitation to buy or sell securities, or a guarantee of future results. The author holds a long position in Arrowhead common stock. Arrowhead Pharmaceuticals (ARWR) is a publicly traded company; investments in its shares involve material risks, including the risk of total loss. All financial projections, acquisition price estimates, and valuation analyses herein are hypothetical frameworks for analytical purposes and do not represent predictions of actual outcomes. Readers should conduct their own due diligence and consult a registered investment advisor before making investment decisions.

About the Author

Robert Toczycki is an independent analyst and registered US Patent Attorney with a JD, an Executive MBA completed at the top of his class, and a BS in Mathematics and Computer Science from the University of Illinois at Urbana-Champaign. He has a deep passion for financial analysis, particularly identifying valuation discrepancies and demonstrating them through rigorous, data-driven research and solid analytics.

Comments or questions: bioboyscout@gmail.com.

Copyright © 2026, BioBoyScout. All Rights Reserved.

Subscribe now