The Left Tail Is Gone: SHASTA Delivers and the Bill for Speed Is $215 Million

Results vs. Consensus

Two of the three headline lines in this quarter are misleading on their face, and in opposite directions. Revenue beat consensus by 70% because it contains a one-time upfront payment that no model carried. Earnings per share missed by eight to thirteen cents because of an income allocation to a minority shareholder in a Chinese subsidiary that has nothing to do with the operating business. The only line that says what it appears to say is the expense line, and it says the cost base is at a record.

FQ3 2026 Scorecard

Year-over-Year Comparison

Read this table from the bottom up and it tells a cleaner story than reading it from the top. The pre-tax loss before the minority allocation is $178.905M against $179.089M a year ago. Twelve months of pipeline expansion, a commercial launch from zero, an entire European regulatory campaign and a doubling of the field organisation produced a loss line that did not move. That is because revenue rose $47.5M and operating expenses rose $52.0M, an almost exact offset, and because $4.7M of improvement below the line closed the remaining gap.

The expense composition is where the year's decisions are visible. Selling, general and administrative expense rose 52.3%, and within it salaries rose 77% and professional and outside services rose 57%. That is the commercial organisation being built. Research and development rose 22.1%, and within it candidate costs rose 33% while discovery costs fell 13%. That is money moving out of the laboratory and into late-stage clinical execution, which is what a company with two Phase III readouts and a Phase III enrolment completion in the same quarter should look like.

Sequential Comparison

The sequential table is the one that matters, and it is unflattering. Three months ago we wrote that a 3.7% sequential decline in operating expenses was the single most encouraging number in the release, and we carried $173M of R&D and $42M of SG&A forward on the strength of it. Both moved the other way and by more than the prior decline. Operating expenses added $30.3M in a single quarter, the largest sequential dollar move since the June 2025 quarter, and the operating loss widened by $28.8M.

The prior-quarter comparison also shows the other income line reversing. The March quarter carried a $19.0M non-cash gain on the Visirna transfer of in-process research and development to Bisirna, which we said at the time produced the entire apparent earnings beat. That gain does not repeat, and the equity-method loss on the resulting stake does, at $2.470M this quarter. Total other expense of $(8.812)M against $3.695M of income is a $12.5M sequential swing, of which $19.0M is the missing gain partly offset by better interest income.

Revenue assessment

Revenue of $75.253M breaks into $26.0M from the Sarepta collaboration, $0.4M from the Sarepta clinical supply agreement, $20.232M from the Novartis recognition schedule, $1.241M from the Sanofi arrangement covering Greater China, $25.0M from Madrigal, and $2.385M of REDEMPLO commercial sales. The last figure is a derived number and the CFO explained the derivation on the call rather than printing it, which is a partial improvement on four quarters of refusing to supply it at all.

"As previously mentioned, we are not intending to headline specific REDEMPLO product sales numbers until they become a meaningful driver to our financials. That said, commercial revenue can be derived from our disclosures as a difference between total revenue and collaboration revenue and represented approximately $2.4 million for the quarter."
— Daniel Apel, Chief Financial Officer

The problem is what happened underneath the beat. Last quarter's clean base of roughly $64M is now $50.3M, and the difference is almost entirely Sarepta. That stream fell from $41.7M to $26.4M, a 36.7% decline, with the clinical supply component collapsing from $4.3M to $0.4M. The 10-Q confirms why: the cumulative catch-up adjustment for Sarepta's step-in right, which was constrained and deferred when we wrote about it in May, was finalised and recorded in this quarter. It is not sized anywhere in the filing or in the CFO's remarks, so the reader is left to infer it from a $15.3M sequential hole in a single counterparty line.

What replaced it is a one-time payment. Madrigal's $25.0M upfront was recognized in full on completion of the licence and technology transfer, which is legitimate accounting and genuinely good news about the partnering model. It is also precisely the kind of discrete event that made this revenue line unforecastable through 2025, and its arrival in the same quarter that the recurring base fell 21% means the headline beat carries almost no information about run-rate.

Cost assessment

Research and development of $198.2M is a record and rose 14.4% sequentially. Candidate costs of $126.6M carried the increase, up 33% year over year on what the filing describes as progression of the pipeline into and through clinical trials and higher outsourced manufacturing. The CFO said the increase was expected internally, which is a defence of the forecast rather than of the level.

"In line with our forecast, this also contributed to the pickup in expenses when compared to fiscal quarter 2."
— Daniel Apel, Chief Financial Officer

The forward statement is the part that changed. In May the CFO said roughly two-thirds of fiscal-year-to-date clinical trial spend was the plozasiran Phase III programme and that it should moderate once the studies read out in the summer, which we translated into a step-down beginning in the September quarter. The studies have now read out. The moderation has been rescheduled.

"Accordingly, we expect costs associated with active execution of those studies to begin to moderate down over time, beginning in fiscal 2027. At the same time, we will continue to invest in regulatory activities, commercial supply readiness for potential sHTG launch and advancement of our broader pipeline."
— Daniel Apel, Chief Financial Officer

Fiscal 2027 begins on October 1, 2026, so the slip is one quarter and not a year. The larger issue is the level from which the moderation starts. We modelled a step-down from $173M. It now starts from $198M, and the second sentence of that quotation adds three new spending commitments on top. Selling, general and administrative expense tells the same story with less ambiguity: $47.1M, up 12.9% sequentially, with the headquarters lease doubled from approximately 49,576 to approximately 98,444 square feet on a seven-year-and-eight-month term, and a further wave of field personnel onboarded the week before the call.

EPS assessment

Reported diluted loss per share of $(1.36) on 143.378M weighted-average shares, against consensus in a band from $(1.23) to $(1.28). The gap is not operational. Net loss including the noncontrolling interest was $(178.916)M, which on the same share count is $(1.25), inside the consensus band and marginally better than the most widely cited figure. The $0.11 difference between the two is the $15.364M of income allocated to the Visirna minority, and it is the second consecutive quarter in which an unexplained Visirna-related item has moved the reported earnings line by more than a dime.

The direction reversed. In the March quarter the minority absorbed $4.840M of loss, which flattered attributable earnings. In this quarter it took $15.364M of income, which penalises them. Noncontrolling-interest equity moved from $(15.061)M at March 31 to $512K at June 30 on that allocation. Nothing in the press release, the 10-Q or the call explains what Visirna earned to generate it, in a quarter when the Sanofi line that runs through that entity contributed $1.241M of consolidated revenue.

Interest expense of $24.562M again did essentially no cash damage. Across the nine months, $69.489M of the $70.913M of interest expense is accretion on the royalty-monetisation liability and the credit facility, and the supplemental cash-flow disclosure records interest paid of nil. Interest income of $16.935M covers 69% of the reported interest expense and all of the cash portion. Anyone treating Arrowhead's reported interest burden as a call on cash is double-counting.

Revenue Composition

Arrowhead reports a single revenue line and operates as a single segment. The economically meaningful decomposition is by counterparty, because each stream has a different mechanic, a different durability and a different reason to disappear. The table reconciles to the income-statement total.

Component figures are as disclosed in the 10-Q and on the call and are individually rounded; the total is the income-statement figure.

Assessment: two thirds of this quarter's revenue came from two counterparties whose contributions are moving in opposite directions for structural reasons. Novartis is the closest thing to an annuity on this income statement, running at roughly $20M a quarter with $126.3M still sitting in deferred revenue against preclinical obligations. Sarepta was that, and is now something less: the step-in permanently reduces the reimbursable research cost that flows through revenue, and the catch-up that trued it up has been taken. A modeller should carry Sarepta at the new $26M level rather than averaging the two quarters, and should treat the Madrigal line as absent from every future period until another deal is signed.

Commercial Performance

REDEMPLO has been approved in the United States since November 18, 2025 for familial chylomicronemia syndrome, a population of roughly 6,500 in the United States. This was the third commercial quarter. The launch is working. The disclosure set describing it changed shape again.

Assessment: the arithmetic finally closes far enough to be useful, and what it shows is a funnel with a leak that management still will not measure. REDEMPLO is dosed once every three months, so the $45,000 annual list price corresponds to $11,250 per quarterly dose at 100% of list. Product revenue of $2.385M is therefore about 212 recognized doses at full list, roughly 265 at a 20% gross-to-net deduction and roughly 303 at 30%. Call it 210 to 300 paid doses in the quarter. Against a stated run rate of 20 to 30 new prescriptions a week, which is 260 to 390 new prescriptions written over a thirteen-week quarter on top of the roughly 180 patients already shipped as of May 7, the conversion from prescriptions written to doses paid remains well under half.

The two rate statements also do not sit together. Management says prescription volume more than doubled over the course of the quarter and, in the same call, that the weekly run rate is approximately 20 to 30 and consistent with what was communicated previously. A rate that more than doubles inside a quarter cannot end that quarter at the same run rate it was described at three months earlier. Either the doubling refers to cumulative volume, in which case it is the mechanical consequence of a flat run rate applied to a small base, or the run rate did rise and the 20-to-30 band is stale. The press release uses "roughly doubled" where the call uses "more than doubled," which does not help.

The disclosure set rotated rather than extended. Last quarter produced a cumulative prescription count, a shipped-patient count, a class-naive share and a switch share. This quarter produced none of them and introduced a prescriber count instead. That a "more than 400" figure appears in both quarters against different denominators, prescriptions in May and prescribers in August, is the kind of coincidence that makes a launch look more continuous than the disclosures actually support.

Key Topics & Management Commentary

Overall Management Tone: This was the most confident call of the four we have covered, and the confidence was earned rather than asserted: the prepared remarks led with a completed pivotal programme and a purchased regulatory accelerant rather than with a promise. The posture shifted from the relaxed, forward-looking stance of May to something closer to operational, with the commercial lead delivering a five-part structured update and the chief medical officer declining, repeatedly and without apology, to characterise embargoed data. Where management was least convincing was on the cost base, where a forward moderation was rescheduled inside a single sentence that also added three new spending commitments.

1. The readout that decided the thesis

SHASTA-3 and SHASTA-4 read out on July 22, thirteen days before the print, and they are the reason this report carries a different rating from the last three. Both studies met the primary endpoint of triglyceride reduction against placebo and every prespecified secondary endpoint. Median triglyceride reductions from baseline at month twelve were 79% and 81%, against approximately 27% for placebo. The prespecified pooled analysis produced a statistically significant 78% reduction in cumulative acute pancreatitis events across the broad severe-hypertriglyceridemia population, and a 100% reduction in the subgroup with triglycerides above 880 mg/dL and a prior history of acute pancreatitis.

"Both studies met their primary endpoint and every prespecified secondary endpoint, a clean sweep across 2 pivotal trials. Median triglyceride reductions from baseline were 79% and 81% in SHASTA-3 and SHASTA-4, respectively. These results were deep, durable and remarkably consistent across both studies."
— Dr. Christopher Anzalone, President and Chief Executive Officer

The safety description was equally unqualified: no new safety signals, no clinically meaningful adverse changes in routine laboratory measures, no clinically meaningful adverse changes in liver enzymes, no hypersensitivity cases, no thrombocytopenia signal, and in a prespecified MRI subgroup no statistically significant difference in mean liver fat content against placebo. The last of those pre-empts the question that a competing mechanism has made investors ask of every triglyceride-lowering agent.

We flagged in May that the pivotal pool adjudicates pancreatitis on modified rather than strict Atlanta criteria, which broke the read-across from the clean two-year open-label extension that most of the bull case rested on. That concern is now moot in the only way that matters. The broader adjudication raised event counts in both arms and the effect survived it. The endpoint did not need the extension's read-across; it produced its own.

Assessment: this is the single most important fact in three quarters of coverage of this company, and it is not ambiguous. The bear scenario we carried at 25% probability, in which no pancreatitis claim is available and plozasiran meets a cheaper first mover on triglyceride lowering alone, is substantially closed. What remains open is whether the FDA grants a pancreatitis-risk-reduction claim in the label, which is a regulatory question and not a data question.

2. The $215 million voucher, and the arithmetic behind it

On July 31 Arrowhead agreed to buy a rare pediatric disease priority review voucher for $215.0M, closing expected in the fiscal fourth quarter after Hart-Scott-Rodino clearance and funded from existing cash and investments. Priority review shortens the FDA's action goal from ten months to six. The company intends to use it on the plozasiran severe-hypertriglyceridemia supplemental application, which it still expects to submit before the end of 2026.

The price was disclosed both in the 10-Q and, unusually for this management team, aloud on the call. So was a return claim.

"Under the terms of the APA, we will pay the current holder $215 million at closing, which we expect to occur in our fiscal fourth quarter following HSR Clearance. According to our projections, should we gain approval in sHTG, the increase in present value of REDEMPLO simply as a result of shifting our launch aspirations and uptake curve forward by 4 months, provides a greater than 3x return on the PRV investment."
— Daniel Apel, Chief Financial Officer

A greater-than-threefold return on $215M implies more than $645M of present-value creation from a four-month shift. No supporting arithmetic was offered and none was requested. The claim is not implausible, because what moves forward is the whole uptake curve rather than four months of terminal revenue, and on a franchise management now sizes at $3B to $4B of annual peak sales the whole-curve shift is worth far more than four months of early-launch sales. It is nonetheless an unaudited internal model output presented as a fact, in a quarter when the company spent more on a regulatory instrument than it lost from operations.

Assessment: we think the purchase is correct and we think the framing is loose. The strategic logic is straightforward: the competing product has a first-mover advantage in this class, and four months is a meaningful share of the window in which that advantage compounds. The commercial lead confirmed the resulting launch target, which is the more useful disclosure than the return multiple. The cost of the option is real, immediate and equal to roughly 14% of quarter-end cash resources.

3. The price holds, and the data is now the defence

The 25% list-price reduction taken in the March quarter, from $60,000 to $45,000, was the decision we were most equivocal about in May. We accepted the stated motive, which was to pre-empt a payer step-through requirement in severe hypertriglyceridemia, and noted that it bought a contingent benefit with a certain cost. The contingency has now resolved, and management was asked directly whether the data changes the price.

"If the question is, do we intend to move that price now that we have these data, the answer is no. We believe this is the right price for this drug. We think that there is a real reason to price this at a slight premium to our competitor, given what we see as a better safety profile, a better reduction in triglycerides from baseline, a simpler approach with quarterly dosing rather than monthly dosing, a lack of need to -- we believe a lack of need to follow liver enzymes because we just haven't seen those issues and a simple 25-milligram dose for all patients rather than having to titrate up."
— Dr. Christopher Anzalone, President and Chief Executive Officer

The commercial lead framed the same position as a structural one, describing the $45,000 figure as sitting under a single unified pricing model that will carry across indications. The premium to the competitor is 12.5% and management is not proposing to widen it on the strength of a pivotal win.

Assessment: this resolves the price cut in the bulls' favour, and it does so at a cost. The option premium paid in March turned out to be in the money, and holding the price with a statistically significant pancreatitis result in hand is a stronger negotiating position than cutting into a payer conversation later. The cost is that the entire severe-hypertriglyceridemia opportunity is now permanently underwritten at $45,000 rather than $60,000, which is a 25% haircut to every revenue-per-patient assumption in a market management sizes above one million high-risk US patients. Both facts are true; the first is worth more than the second because a step-through requirement in the larger indication would have been worth far more than 25%.

4. The cost base went the other way

Operating expenses of $245.3M are a record and rose 14.1% sequentially, reversing the first decline in four quarters that we treated as the most encouraging line in the March release. Research and development of $198.2M rose 14.4%, driven by candidate costs of $126.6M. Selling, general and administrative expense of $47.1M rose 12.9%, with salaries up 77% and professional and outside services up 57% year over year.

Two commitments underneath the number deserve to be read together. The headquarters lease in Pasadena was amended in April to roughly double the leased footprint, from approximately 49,576 to approximately 98,444 square feet, on a term of seven years and eight months, generating $13.4M of new right-of-use assets. And the field organisation took another step function up in the week before the call.

"On our commercial infrastructure, our field organization continues to scale in a deliberate sequenced way sized for both the current FCS opportunity and the future sHTG opportunity as it unfolds. Our commercial team's tenure and productivity continue to build, and we're seeing that reflected in the prescription and payer metrics I just walked through. Importantly, if the launch timing for sHTG is accelerated as we expect, we will be ready."
— Andy Davis, Senior Vice President and Head of the Global Cardiometabolic Franchise

Assessment: the level is defensible and the guidance around it is not. Spending ahead of an approval you have good reason to expect is what a company with a de-risked pivotal readout should do, and a doubled headquarters plus a doubled call-point universe is a coherent response to a market that goes from 5,000 to over 20,000 physician targets. What is not defensible is that the moderation promised for the September quarter in May has become "beginning in fiscal 2027" in August, from a base $25M a quarter higher, in the same paragraph that adds regulatory activity, commercial supply readiness and broader pipeline advancement as new claims on the line. Investors were given a step-down and have received a step-up with a later step-down attached.

5. The recurring revenue line fell while the headline beat

Consensus of $44.17M was set for a quarter in which the Sarepta step-in would reduce reimbursable research revenue. The print of $75.253M is 70% above it. Almost the entire excess is a payment nobody modelled and nobody will model again.

Strip the $25.0M Madrigal upfront, and revenue is $50.253M against the roughly $63.7M of underlying base we identified in May. Sarepta fell from $41.7M to $26.4M as the cumulative catch-up landed. Sanofi fell from $10.5M to $1.241M because the March quarter carried a $10.0M Chinese approval milestone. Novartis was flat at $20.232M and product revenue more than doubled to $2.385M. The net is a recurring base 21% lower than a quarter ago.

Assessment: the beat is real accounting and poor information. Three months ago the useful thing about the March quarter was that nothing episodic dominated the income statement, which is what made a base observable for the first time. That property lasted exactly one quarter. A model built on this print should carry Novartis at $20M, Sarepta at the reset $26M, Sanofi near $1M, product revenue growing off $2.4M, and nothing at all for Madrigal. That is roughly $50M with an upward slope from the commercial line only.

6. Sarepta's step-in lands, and is still not sized

The clearest open question from the March quarter was how large the Sarepta step-in adjustment would be and which direction it would go. The 10-Q answers the timing and declines the magnitude. It states that the scope of the modification, including the transition terms, was finalised during the fiscal third quarter and that the related cumulative catch-up adjustment to revenue was recorded in that period. No dollar figure appears in the filing. The CFO described the Sarepta line as coming mainly from ongoing recognition of initial consideration and reimbursement of clinical and manufacturing expenses, without mentioning the modification at all.

Assessment: a $15.3M sequential decline in the largest revenue stream, arising from a contract modification that the company disclosed one quarter earlier and finalised in this one, is the sort of item that belongs in the CFO's prepared remarks. Its absence there is the third consecutive quarter in which the most consequential accounting fact in the period appeared only in the filing. The underlying business change is modest and arguably positive, since Sarepta assuming trial responsibility removes cost as well as reimbursement revenue. The disclosure practice is the problem, not the economics.

7. Peak sales get a number for the first time

In May we listed the absence of any peak-sales or penetration framing for severe hypertriglyceridemia as the most conspicuous omission on the call: management had sized the population above one million high-risk US patients and set the price at $45,000, and had never multiplied the two. That changed in the first five minutes of this call.

"We have an approved product and positive pivotal data that we believe supports a potentially much larger indication that we think could drive peak sales in the $3 billion to $4 billion per year range."
— Dr. Christopher Anzalone, President and Chief Executive Officer

Asked to split it geographically, the CEO said the overwhelming majority is in the United States, and separately declined to forecast anything ex-US given uncertainty around most-favoured-nation pricing policy. He also volunteered a ramp caveat that a promotional management team would have left out.

"However, this is going to be a relatively slow ramp because this is a brand-new market. We are in the education business. And so it's going to take a bit of time for us to get the word out because, look, the world is not used to looking at TGs closely because there has been -- in part at least because there have been no good ways to really reduce triglycerides until now."
— Dr. Christopher Anzalone, President and Chief Executive Officer

Assessment: the number is management's and it is the low end of it we are prepared to underwrite. A $3B peak on a $45,000 net-of-nothing price implies roughly 67,000 patients on therapy at peak, which against a stated addressable population above one million is under 7% penetration and is not aggressive. The credibility of the framing is helped considerably by the fact that the same executive volunteered the slow-ramp caveat in the same session, and by the refusal to put a number on ex-US revenue rather than filling the gap with an assumption.

8. Building for a launch that has not been approved

The commercial lead gave the most specific forward operational disclosure of the call, and it is the one that dates the investment case.

"While I won't go into the details of the size of our field force for sHTG, I would tell you that we'll be moving from effectively addressing over 5,000 HCP targets to a world will be addressing over 20,000 HCP targets across both the specialists that I've mentioned previously and also potentially those primary care physicians who act like specialists. So we would anticipate the final onboarding of the optimization of our field force to happen before the end of the year as we prepare for a potential accelerated launch in sHTG in the second quarter of next year."
— Andy Davis, Senior Vice President and Head of the Global Cardiometabolic Franchise

A calendar second-quarter 2027 launch is materially earlier than the second-half-2027 approval we carried in May, and it is the operational expression of the voucher. It also implies that the entire commercial build has to be complete by December, two to three quarters before any revenue from the larger indication arrives.

Assessment: this is the timeline that should replace the one in every model, and it is the reason the selling and administrative line will not moderate when the research line does. The company is committing to carry a field force sized for 20,000 targets through at least two quarters in which it can only sell into a 6,500-patient orphan indication. That is the right decision if the label arrives on schedule and an expensive one if it slips.

9. The balance sheet after the voucher

Cash and investments closed at $1,566.9M against $1,784.1M three months earlier. Those two figures sit on different presentations: the FQ3 8-K moves cash held at the Visirna variable interest entity out of total cash resources, while the FQ2 figure includes it. Adding back the $35.1M of VIE cash held at June 30 puts the quarter's reduction at $182.1M on a like-for-like basis, in a quarter with no financing event of size. Set against $700.0M of convertible principal, the $181.4M carrying value of the credit facility and the $392.5M royalty-monetisation liability, net cash is $293.0M. The voucher has not yet been paid; pro forma for the $215.0M, net cash is approximately $78.0M.

Three months ago the same calculation produced roughly $500M and we retired the funding bear point to contained on the strength of it. The company also sold 367,000 shares under the at-the-market programme during the quarter for $27.2M of net proceeds, bringing fiscal-year-to-date at-the-market issuance to $76.1M gross. The CFO's closing remark on the balance sheet did not mention either the at-the-market sales or the pro forma position.

"As a concluding remark, we believe that our strong balance sheet provides significant financial flexibility to support ongoing clinical development, current and future commercialization activities and our long-term strategic priorities."
— Daniel Apel, Chief Financial Officer

Assessment: the gross figure is genuinely strong and the net figure is thinner than it was. $1.57B of resources against a cash operating cost base near $224M a quarter is roughly seven quarters of gross runway before any partner receipts, and partner receipts in the nine months to June totalled $475.0M across the $200.0M Novartis upfront, the $200.0M second Sarepta milestone, the $50.0M first Sarepta annual-fee instalment and the $25.0M Madrigal upfront. But the quarter consumed $182M like-for-like, the voucher takes another $215M, the commercial organisation is still scaling into an unapproved indication, and the company is quietly using its at-the-market programme at prices in the seventies. This is not a solvency question. It is a question of whether the next financing happens from strength after a label, or from necessity before one.

10. The Greater China vehicle keeps moving cash and equity outward

Arrowhead's ownership of Visirna fell from 66.25% to 56.38% on January 2 when option holders exercised 14 million options. In March, Visirna paid $94.8M of a declared $100.0M dividend, of which $38.4M went to minority shareholders. This quarter the entity allocated $15.364M of income to those same minority holders, taking noncontrolling-interest equity from $(15.1)M to $0.5M and taking $0.11 off reported earnings per share. Cash held at the variable interest entity fell from $137.8M at fiscal year end to $35.1M at June 30. The nine-month cash-flow statement records $21.6M of income taxes paid against $0.1M in the comparable prior period.

Assessment: none of this is discussed anywhere outside the filing, and the pattern is now four quarters old. A subsidiary that holds the rights to up to $255.0M of remaining Chinese development milestones has been diluted by ten percentage points, has distributed nearly a hundred million dollars of cash with 41% of it leaving the group, and is now generating income allocations that move consolidated earnings per share by more than a dime with no stated cause. We do not think this is sinister. We think it is unexplained, and unexplained items that move reported earnings are exactly what a reader is entitled to have management address.

11. The pipeline behind plozasiran

The quarter's non-plozasiran progress was substantive. ARO-INHBE showed a 44% placebo-adjusted reduction in liver fat in a small subgroup of obese patients with elevated baseline liver fat receiving at least 200mg as monotherapy, against a 30% threshold the chief medical officer described as the level generally thought to translate into histologic and potentially clinical benefit; a Phase 2b protocol has been submitted to regulators. The zodasiran YOSEMITE Phase III completed enrolment in mid-July at 70 patients against a 60-patient design, with data in the second half of 2027. Madrigal took ARO-PNPLA3 on a $25.0M upfront with up to $975.0M of milestones and tiered royalties to the mid-teens. Management expects 23 clinical candidates by year end, eleven wholly owned and twelve partnered.

Two first-in-human readouts land in September: ARO-DIMER-PA, the first dual-functional molecule silencing both APOC3 and PCSK9, and ARO-MAPT, the first candidate on the subcutaneous central-nervous-system delivery platform.

"This is a potentially exciting data set, not only because of the potential of ARO-MAPT against Alzheimer's disease and other [indiscernible] but also because we think it could provide the first clinical proof of concept that we are able to address brain targets with RNAi using a simple subcutaneously administered conjugate."
— Dr. Christopher Anzalone, President and Chief Executive Officer

Assessment: the platform pillar advanced on evidence rather than on assertion for the second consecutive quarter, which is what separates this company from most single-asset commercial-stage biotechs. The September readouts are option value rather than thesis: a positive ARO-MAPT result validates a delivery platform across a preclinical CNS portfolio including partnered programmes, and a negative one costs nothing that is currently in the price. We would not pay for either in advance.

Guidance & Outlook

Arrowhead has never issued quantitative annual revenue or expense guidance and did not begin here. What it does issue is dated commitments, and the useful exercise is to grade the ones made in May and record the ones made in August.

New commitments made on this call

Implied path: the fiscal fourth quarter should carry operating expenses at or above $245M, a $215.0M cash outflow for the voucher, a Sarepta line near the reset $26M, Novartis near $20M, no Madrigal, and product revenue growing off $2.385M. That points to a September quarter with revenue near $50M and a cash reduction of $400M or more including the voucher, before any partner milestone.

Guidance style: Arrowhead does not guide numbers and does guide dates, and the dates have been reliable on clinical execution and unreliable on cost. Every clinical milestone promised in May either landed on time or landed early. The single financial forward statement made in May did not.

Analyst Q&A Highlights

Twelve analysts asked questions. The session had an unusual shape: almost nobody asked about the quarter. There was not a single question about the record operating expense line, the Sarepta catch-up, the noncontrolling-interest allocation or the balance sheet. The questions divided between what the embargoed data will show at the end of August and how the commercial organisation converts a pivotal win into revenue, and management answered the second set fully and the first set not at all.

What the detailed data will show, and what it will not

The opening question of the call combined the regulatory timeline with a direct probe at the shape of the efficacy result, referencing an existing debate over whether the reported median triglyceride reduction flatters the mean. Management answered the regulatory half and declined the data half, citing the conference embargo. The same refusal was given to three further questioners later in the session.

Q: "And separately, can you talk about expectations for the ESC late-breaker data? There's been some debate on median versus mean TG reduction magnitude."
— Maurice Raycroft, Jefferies

A: "I'll answer the second question, the ESC data, you'll have to wait and see at ESC that we really are under embargo until the conference, so we can't discuss any details around the study. In terms of rate limiters for the sNDA, we do plan to have a pre-sNDA meeting with FDA and then subsequent to our discussions with the agency would file the submission."
— Dr. James Hamilton, Chief Medical Officer and Head of R&D

Assessment: the embargo is real and the refusal is procedurally correct, but it leaves a twenty-six-day window in which the most important dataset in the company's history is described only by its own topline release. Four separate analysts probed the same soft spot, which tells you the topline left gaps that a careful reader noticed. The specific gap is that efficacy was reported as a median, and the market has not been shown the mean or the distribution behind it.

The size of the MRI subgroup behind the liver-fat statement

The most precise question of the call sought the denominator behind the prespecified imaging analysis that management used to close down the liver-fat concern in its prepared remarks. The answer was the embargo again.

Q: "maybe can you remind us what proportion of all patients in SHASTA-3 and SHASTA-4 actually received an MRI at baseline in year 1? Just trying to understand what's the sample size here and how meaningful that analysis will be."
— Luca Issi, RBC

A: "Initial question, like I said before, we're under embargo. I can't really give any details around the SHASTA-3 or SHASTA-4 study, but we'll present all of that at ESC. For SHASTA-5, we don't have any plans to terminate that study right now. The plan would be to continue that and maybe get a better idea of what our label is going to look like before we make any decisions to stop the study."
— Dr. James Hamilton, Chief Medical Officer and Head of R&D

Assessment: the question is the artifact here, not the answer. Management asserted in prepared remarks that a prespecified imaging subgroup showed no statistically significant difference in mean liver fat, and then declined to say how many patients that subgroup contained. A null result in a small subgroup and a null result in a large one are different pieces of evidence. The second half of the answer is the more useful disclosure: the dedicated pancreatitis study continues, and the decision to stop it is now explicitly tied to what the label looks like rather than to the pooled analysis succeeding, which is a change from the framing given in May.

How large the field force becomes, and when

The commercial scale-up question produced the most consequential forward disclosure on the call, converting the voucher purchase into an operational date.

Q: "Early in the call, you talked about the sequencing scale up of your sales force. How big of a sales force do you envision that you'll need to reach? And how does that sequence look over the course of the next several months heading into the label expansion decision?"
— Brian Cheng, JPMorgan

A: "While I won't go into the details of the size of our field force for sHTG, I would tell you that we'll be moving from effectively addressing over 5,000 HCP targets to a world will be addressing over 20,000 HCP targets across both the specialists that I've mentioned previously and also potentially those primary care physicians who act like specialists."
— Andy Davis, Senior Vice President and Head of the Global Cardiometabolic Franchise

Assessment: a fourfold expansion of the call-point universe with the build complete before December and no revenue from the larger indication until the second calendar quarter of 2027 is the clearest statement yet of why the selling and administrative line will not moderate when the research line does. It also quantifies, for the first time, the commercial infrastructure that the whole severe-hypertriglyceridemia case has to pay for.

Whether the price moves now that the data exists

The pricing question was the one place where a pivotal win could plausibly have changed a prior decision. Management foreclosed it in two voices, first commercially and then from the chief executive without being asked twice.

Q: "Now that you have your data, how are you thinking about price differential versus Tryngolza?"
— Kaey Nakae, Chardan Capital Markets

A: "So I won't go into price details or contracting strategy only to say that you would be aware of the wholesale acquisition cost for REDEMPLO USD 45,000 per year. That does differ from our competitor, and we've communicated previously that we believe that premium is justified based on the product attributes of REDEMPLO across efficacy, safety and convenience."
— Andy Davis, Senior Vice President and Head of the Global Cardiometabolic Franchise

Assessment: the answer removes the last open question about the March price cut. A 12.5% premium held with a statistically significant pancreatitis reduction in hand is a stronger commercial position than a wider premium argued on triglyceride lowering alone, and it removes the risk that a payer step-through requirement appears in the larger indication. The permanent 25% reduction in revenue per patient stands, and it is now clearly a price paid for access rather than a concession extracted by payers.

The weekly prescription run rate, and the conversion question asked for the fifth time

A recurring line of questioning across four quarters has been how the growing prescription count converts into shipped, paid product. It was asked again, this time paired with a direct challenge to the run-rate arithmetic. Management gave the run rate, described it as unchanged, and answered the conversion question by describing the process rather than the result.

Q: "Maybe how should we think about the doubling of REDEMPLO prescriptions, I guess, in terms of weekly run rate, I think 30 per week maybe was the last disclosure you guys put out."
— Madison Wynne El-Saadi, B. Riley Securities

A: "Yes, we do see approximately 20 to 30 new prescriptions a week that has been the run rate and consistent with what we have communicated previously. Of course, our teams are very focused on also ensuring that those prescriptions find their way through the funnel ultimately to shipments to patients. And so our market access team is working incredibly hard to support our payer and our physicians and offices around compliantly navigating the prior authorization process and appeal process."
— Andy Davis, Senior Vice President and Head of the Global Cardiometabolic Franchise

Assessment: this is the most revealing exchange on the call and the least remarked upon. The question explicitly anchored on the 30-per-week figure given in May; the answer restated it as a 20-to-30 band while calling it consistent, which is a quiet reduction in the lower bound of a metric that has never been reconciled to revenue. More telling is the second half: management expects an inflection in how prescriptions filter through the funnel to patient shipments, which is an acknowledgement that the funnel currently leaks and that prior authorisation and appeals are where it leaks. Our own derivation, roughly 210 to 300 paid quarterly doses against 260 to 390 new prescriptions written in the period plus a carried base, is consistent with that. It is the fifth consecutive quarter in which conversion has been asked and not given.

How the peak-sales number splits geographically

Having volunteered a peak-sales range for the first time, management was immediately asked to decompose it, and gave a directional answer plus an unusually candid caveat about the part it will not forecast.

Q: "And maybe of that $3 billion to $4 billion range that you indicated for plozasiran, how much of that is U.S.-based versus international markets?"
— Joseph Thome, TD Cowen

A: "We'll be the 65th company to tell you that given the uncertainty around MFN, it's very difficult for us to -- at this point to know how these are going to be pursued in ex U.S. markets and what kind of revenue we're going to see outside the U.S. markets."
— Dr. Christopher Anzalone, President and Chief Executive Officer

Assessment: the refusal to invent an ex-US number improves rather than damages the credibility of the $3B to $4B range, because it means the range is not padded with a European assumption management cannot support. The practical consequence is that a company approved in five geographies and beginning German reimbursement within weeks is asking investors to underwrite an almost entirely domestic peak. Any model carrying material ex-US plozasiran revenue is ahead of the company's own framing.

What the September central-nervous-system readout will and will not answer

The first readout from the subcutaneous brain-delivery platform arrives in September, and the question sought to calibrate what a success looks like against a competitor's recently disclosed data. Management scoped the readout down before anyone could over-read it.

Q: "Just if you can remind us what top line data you might share with us in September? And what level of knockdown are you looking for?"
— Michael Ulz, Morgan Stanley

A: "So this will only be healthy volunteer data. We'll be discussing primarily safety and then total tau knockdown. There's not a lot of other biomarkers that we can measure in the healthies. So it's just a safety and pharmacodynamic dose range finding study in the healthies."
— Dr. James Hamilton, Chief Medical Officer and Head of R&D

Assessment: management set the bar at 50% to 60% total tau knockdown and framed the readout as safety and pharmacodynamics in healthy volunteers rather than as evidence of clinical effect. That is the correct expectation-setting and it is worth noting because the platform claim is larger than the asset claim: a positive result validates subcutaneous delivery across a preclinical central-nervous-system portfolio that includes partnered programmes. The asset is optional. The platform read is not.

What They're NOT Saying

1. The size of the Sarepta step-in catch-up. The filing states that the modification scope was finalised in this quarter and that the cumulative catch-up adjustment to revenue was recorded in it. No dollar figure is given anywhere. The Sarepta stream fell $15.3M sequentially and the CFO described the line without mentioning the modification.
2. Why $15.364M of income was allocated to the noncontrolling interest. The allocation swung $20.2M sequentially, took $0.11 off reported earnings per share and moved noncontrolling-interest equity from $(15.1)M to $0.5M. It appears as a line in the press-release operating summary and is explained in neither the filing nor the call, in a quarter when the Sanofi arrangement that runs through that entity contributed $1.241M of consolidated revenue.
3. Prescription-to-paid-claim conversion. Fifth consecutive quarter without a figure, and this quarter the shipped-patient count that was offered for the first time in May was withdrawn rather than updated. The nearest thing to an answer was that the market-access team is working on prior authorisation and appeals and that an inflection is expected in how prescriptions move through the funnel.
4. Gross-to-net. Still refused as a matter of policy. The company will not say what proportion of a $45,000 list price it collects, which is the only remaining unknown that prevents the derived product revenue from being turned into a patient count.
5. The arithmetic behind the greater-than-threefold voucher return. A $215.0M purchase justified by an internal present-value calculation implying more than $645M of value creation, with no disclosure of the discount rate, the launch curve, the peak assumption or the probability of approval that produces it.
6. The mean triglyceride reduction, and the size of the imaging subgroup. Efficacy was reported as a median. Management used a prespecified imaging analysis in prepared remarks to close down the liver-fat question and then declined to say how many patients it contained. Both questions were asked directly and both were deferred to a conference twenty-six days after the print.
7. The doubled headquarters lease. The Pasadena footprint went from approximately 49,576 to approximately 98,444 square feet on a seven-year-and-eight-month term, generating $13.4M of new right-of-use assets. Disclosed only in the filing, in a quarter when the cost base was the weakest part of the release.
8. The at-the-market issuance. 367,000 shares sold in the quarter for $27.2M of net proceeds, and $76.1M gross fiscal-year-to-date. Not mentioned in the CFO's balance-sheet remarks, which closed on the significant financial flexibility the balance sheet provides.
9. $21.6M of cash income taxes paid. Against $0.1M in the comparable prior-year period. A real cash outflow at a company with no taxable income at the consolidated level, disclosed only in the supplemental cash-flow lines.
10. The Ionis patent litigation. Live in the Central District of California, seeking damages over plozasiran commercialisation, with no contingent liability recorded. Not mentioned on a call that discussed competitive positioning against the same counterparty by name.
11. What the fiscal fourth quarter looks like. Management confirmed the voucher closes in the fiscal fourth quarter and that costs moderate from fiscal 2027, which brackets a September quarter that carries a record cost base and a $215.0M outflow without a single forward figure attached to it.
12. Any operating expense ceiling. Three consecutive quarters of being asked, in one form or another, where this cost base settles. The answer remains that quarterly research and development will continue to be highly influenced by programme timing.

Market Reaction

• Pre-print setup: the shares closed at $89.59 on August 4, up 444.0% over the trailing twelve months from $16.47, up 6.7% over the trailing thirty days from $83.99, and up 34.9% year to date from $66.39. That close was the highest of the trailing 52 weeks, against a 52-week closing range of $15.93 to $89.59. The S&P 500 was up 13.0% year to date.
• The event was already banked: the SHASTA top-line was released before the open on July 22. The shares closed that session up 19.0% at $88.70 on 7.0 million shares, having traded as high as $95.49 intraday, 28.1% above the prior close. They then drifted to $84.65 by July 31 before rallying 4.2% into the print on August 4.
• Reaction session: the shares opened at $87.49 on August 5, a 2.3% downward gap, traded a range of $85.98 to $90.81, and closed at $86.91, down 3.0% or $2.68. The S&P 500 fell 0.2% on the same session, so the relative move was 2.8 points.
• Volume: 1.7 million shares against a 2.8 million thirty-day average, 0.6 times normal.

The volume is the whole story of this session. Compare it with the three prior prints. In February the stock barely moved, closing down 0.2% on 10.3 times average turnover as an offering settled, which was an ownership transfer with no change of view. In May it fell 6.7% on 1.5 times turnover, which was a repricing by holders who stayed. In August it fell 3.0% on six tenths of normal turnover, which is neither. It is the absence of a marginal buyer at a 52-week high thirteen days after the news that mattered, rather than a verdict on anything in the release.

That reading is supported by what the print actually contained for a marginal buyer. Everything genuinely new was either already public or unflattering. The SHASTA result was thirteen days old. The voucher and the Madrigal deal had both been announced. The European and Australian approvals had both been announced. What was new on August 4 was a record cost base, a rescheduled expense moderation, a recurring revenue line 21% lower and a headline earnings miss that most screens would have reported without the noncontrolling-interest adjustment. A stock at its 52-week high needed something in that list to work, and nothing in it did.

The setup matters as much as the content. This equity entered the print having risen 444.0% in twelve months and 23.2% since the last report. The July 22 session captured the entire binary in one move, and the fourteen sessions after it gave back a fifth of it and then took it back. By August 4 the shares were priced for a company that had already won, and a quarterly report was never going to add to that. A 3.0% fade on light volume is what a stock does when the news it was waiting for has already happened.

Street Perspective

Debate: does the label carry a pancreatitis claim?

Bull view: the pooled analysis was prespecified, it was statistically significant across the broad population, and the effect in the highest-risk subgroup was total. Acute pancreatitis is the clinical consequence that makes severe hypertriglyceridemia worth treating, and a sponsor that produces a significant reduction in a prespecified pooled analysis of two pivotal studies has the strongest possible basis to seek a claim. Breakthrough therapy designation in the indication is already in hand and the safety database now includes the MUIR-3 population.

Bear view: a claim is a regulatory decision, not a statistical one. The pooling of two studies to power a secondary endpoint is exactly the design regulators scrutinise most closely, the adjudication used broader criteria than the open-label evidence the market has been quoting for two years, and detailed data has not been seen by anyone outside the company. Approval on triglyceride lowering with pancreatitis data confined to the clinical studies section of the label is a perfectly ordinary outcome and would leave the commercial case resting on a surrogate.

Our take: the bull case is stronger and the bear case is not empty. What the readout unambiguously did is remove the scenario in which plozasiran fails to differentiate on outcomes at all, which is the scenario that justified our Hold. Whether the differentiation appears in the indication statement or in section 14 changes the size of the opportunity, not its existence. We size the claim at slightly better than even and note that the market will know more on August 30 than it knows today, which is a reason to expect volatility rather than to wait.

Debate: is the cost base a build or a problem?

Bull view: a company with a completed pivotal programme, five approved geographies, a Phase III enrolment just closed and a fourfold expansion of its call-point universe ahead of it should be spending more, not less. Candidate costs rose because assets moved through the clinic, and commercial costs rose because a field force sized for one million patients cannot be hired the month before approval. The moderation is dated and the operating leverage is arithmetically enormous against a $3B peak.

Bear view: the specific commitment made three months ago was that research costs step down once the studies read out. The studies have read out and costs went up 14%. The moderation has moved to fiscal 2027 from a base $25M a quarter higher, and the same sentence added three new claims on the line. A company that has never given an expense ceiling and has now missed the only forward expense statement it ever made has not earned the benefit of the doubt on the next one.

Our take: the bears have the better argument on the forecast and the bulls have the better argument on the decision. Spending $245M a quarter to be ready for a second-quarter-2027 launch is correct capital allocation given what the July readout showed. Telling investors in May that the line steps down in September, and in August that it steps down some time in fiscal 2027, is a forecasting failure that should reduce the weight anyone puts on the next such statement. We now model the moderation as beginning in the December quarter at the earliest and being partly offset by commercial build through calendar 2027.

Debate: is the launch readable yet, three quarters in?

Bull view: product revenue more than doubled sequentially to $2.4M, prescription volume more than doubled inside the quarter, favourable policies are in place with the most significant US payers, the prescriber base is above 400 and the European label uniquely permits clinical diagnosis. For a rare disease with roughly 6,500 US patients and a historically poor diagnosis rate, that is a good third quarter, and the point of the FCS launch was always to build the machine for severe hypertriglyceridemia rather than to earn a return on 6,500 patients.

Bear view: the metric set changed again. The cumulative prescription count, the shipped-patient count, the class-naive share and the switch share all disappeared and a prescriber count arrived in their place. The weekly run rate was restated with a lower bound while being described as consistent. Conversion has been withheld for five quarters, gross-to-net is refused as policy, and $2.4M of product revenue against a $245M cost base means the commercial business is still a rounding error that cannot be audited from outside.

Our take: more readable than last quarter and still not readable enough. The derived product revenue figure is the single most useful commercial disclosure this company has produced, and the CFO explaining how to derive it is a genuine step forward from four quarters of silence. But a launch is only readable when the same metrics are reported in successive periods, and this one has now presented a different set three times running. Our own arithmetic implies 210 to 300 paid doses in the quarter against 260 to 390 prescriptions written on top of a carried base, and a conversion rate that low is either a durable feature of the payer pathway or a temporary one. Management's own answer, that it expects an inflection in how prescriptions filter through to shipments, implies the latter. We will believe it when a denominator survives two consecutive quarters.

Model Update Needed

Valuation: at the $86.91 reaction close, 141.135M shares outstanding plus 2.468M pre-funded warrants gives an economic market capitalisation of approximately $12.48B. Adding $700.0M of convertible principal, the $181.4M credit facility and the $392.5M royalty-monetisation liability, less $1,566.9M of cash and investments, gives an enterprise value near $12.19B. Against management's $3B to $4B peak-sales framing that is 3.0 to 4.1 times peak revenue. Against the $2.0B base case we carried in May it is 6.1 times. The re-rating that has taken this equity from $72.69 to $86.91 since the last report has therefore been accompanied by a larger upward revision to the denominator than to the numerator, which is why the multiple has compressed while the price has risen.

Fair-value range: $65 to $130, probability-weighted approximately $102, from $70 to $100 weighted near $79. The framework is unchanged in structure and repopulated for a resolved endpoint.

The probability-weighted value of approximately $102 is 17.9% above the $86.91 close. The scenario spread has narrowed relative to the weighted value, from $47 to $110 in May to $65 to $130 now, which is the mechanical consequence of a binary resolving. The reason the weighted value rose by more than the price did is that the removal of the $47 outcome is worth more than the July repricing captured, and that management's own peak-sales framing sits 50% to 100% above the base case we were carrying.

Thesis Scorecard Post-Earnings

Overall: the thesis is materially stronger and the company is materially more expensive to run. All four bull pillars were confirmed in the same quarter, which has not happened before in our coverage, and the confirmation of the first one is the event the previous three reports were waiting for. Against that, three bear points escalated and all three are financial rather than scientific: the cost base, the composition of revenue and the net liquidity position. That is the correct trade. A biotechnology company that resolves its pivotal binary and simultaneously discovers that execution is expensive has converted the risk it could not manage into the risk it can.

Action: buy. We were unwilling to underwrite a binary readout in May at $72.69 against a probability-weighted $79, and said we would revisit immediately after it in either direction. It resolved in the direction that removes the outcome carrying a quarter of our probability weight. At $86.91 against a probability-weighted $102, the risk-adjusted upside is 17.9% with a scenario spread that has narrowed by about a fifth relative to the weighted value. The specific risk we are accepting is the twenty-five days between this report and the detailed presentation at the end of August, during which the only description of the pivotal dataset is the company's own topline and management declined four times to characterise the distribution behind it. That risk is real, it is dated, and it is smaller than the one that has just been retired. We would add on any weakness into the conference and would revisit the position on the label decision rather than on the next quarterly print.