Newsletter · · Ashutosh Agarwal

The Tau Drug Where the Low Dose Won - The Neuro & Alzheimer's Pipeline - Week of August 2, 2026

A synthesis of what clinical and investor podcasts said about the neuro and Alzheimer's pipeline for the week of July 27 to August 2, 2026, including Biogen's antisense tau drug shrinking established tangles while its lowest dose helped most, a formal guideline red light on screening healthy people with pTau-217, and the first patient dosed in a combination trial.

The Neuro & Alzheimer's Pipeline

Week of July 27 to August 2, 2026: The Tau Drug Where the Low Dose Won

Last week we got the box score from the field's big annual meeting in London. This week we got the play-by-play, a long, unusually clear-headed recap on a medical-journal podcast that finally put real numbers on the two stories everyone's been arguing about: the new tau drug that broke a scientific record while missing its own test, and the cheap blood test that keeps failing the one exam that matters, the one where someone agrees to pay for it. We also heard, for the first time, that the "just combine the drugs" idea everyone keeps talking about has stopped being talk: the first patient in a real combination trial has been dosed. And a man with early Alzheimer's walked us through exactly why he picked one drug over the other, the kind of ground-truth the sales models rarely capture.

(Quick vocabulary, used throughout: amyloid and tau are the two toxic proteins that pile up in an Alzheimer's brain, amyloid clumps between nerve cells, tau tangles inside them. pTau-217 is a form of tau you can now measure from a simple blood draw. PET scan is a brain scan that lights up amyloid or tau directly, and is the current gold standard. APOE4 is the most common gene that raises Alzheimer's risk. ARIA is the brain swelling-and-bleeding side effect of the anti-amyloid drugs. CDR-SB is the main yardstick trials use to score how fast someone is declining, lower is better. An antisense drug is one that switches off a gene's instructions before the harmful protein ever gets built.)

TL;DR

  • The tau drug's result is genuinely strange, and the strangeness is the story. Biogen's antisense tau drug did something no tau drug had ever done: it made existing tangles shrink on brain scans. But its Phase 2 trial missed, and in a twist, the lowest dose helped thinking the most (26% slowing) while the highest dose helped least (9%). A UCSF expert floated the reason bears should sit with: you can lower tau too much, because the brain actually needs some of it. That's a second independent expert making the same worrying point we flagged last week.
  • The blood-test "screen everyone" dream just got a formal red light. A big new study showed pTau-217 predicts who will decline over the next decade, but the doctors' own updated guideline, refreshed at the same meeting, still says don't give the test to healthy, symptom-free people outside of research. Accuracy was never the problem. Permission and payment are.
  • Combination therapy stopped being a slide and became a trial. The first patient has been dosed in an NIH-funded study testing an anti-amyloid drug plus an anti-tau vaccine, together. That's the whole field's endgame taking its first real step, and it needs more diagnostics, not fewer.

What's new

The tau drug broke a record and failed its own test in the same breath. The single most important thing this week was a granular walk-through of Biogen's antisense tau drug diranersin (BIIB080, the program Biogen runs with Ionis), on the JAMA Medical News recap of the conference (July 31). Dr. Gil Rabinovici of UCSF (expert, an academic neurologist not tied to the company) laid out the Phase 2 CELIA study in roughly 400 early-stage patients. The headline was supposed to be a clean "more drug, more benefit." It wasn't: "though all doses showed a benefit compared to placebo, it was actually the low dose that had the greatest benefit, about 26% of slowing on change on the CDR-SB, whereas the changes with the mid and high dose were 14 and 9% respectively." So the trial technically missed, there was no dose-response, and the drug you'd expect to work best worked worst.

And yet the biology was jaw-dropping. The drug lowered tau in spinal fluid by up to 60%, and, the record-breaker, "tangles that had already formed and were being imaged by PET at baseline seemed to disappear or at least reduce in intensity... We thought the tangles were there to stay, but these results suggest that if you lower tau, they may be more dynamic." No tau drug had ever reversed established tangles before.

"It may be that you can lower tau too much... if you lower tau a little bit, you can have benefits, but if you lower it too much, you might start to interfere with some of the normal functions and that may actually have a deleterious effect." Dr. Gil Rabinovici, JAMA Medical News

Why it moves numbers: this is the same "therapeutic window" worry a BioCentury editor raised last week, now voiced independently by a second, unaffiliated academic, and backed by the actual dose-by-dose numbers. When two experts who don't work together look at the data and land on "less may be more," that's no longer a one-off hot take; it's a design problem the whole broad-tau-lowering field has to solve. One more sobering detail Rabinovici flagged: a dose-dependent "confusional state" side effect (worse at higher doses), and the drug is delivered by spinal tap, every 24 weeks at the winning low dose, which is at least infrequent. Phase 3 is going ahead anyway.

The blood test predicts the future, and the guidelines still say don't use it on the healthy. A large study published in JAMA alongside the meeting followed about 2,700 people with normal thinking and sorted them by their pTau-217 blood levels. The punchline, again from Rabinovici on JAMA Medical News: "about 24% of people who had high levels and 38% of people who had a very high level had developed clinically significant impairment within five years." The test clearly sees trouble coming. But he was blunt about what to do with that: "it is really very premature to recommend that cognitively normal older adults get these blood tests, even though they are currently available. There is actually a clinical practice guideline around these blood tests that was updated at this meeting and continues to recommend against obtaining plasma pTau-217... in cognitively normal people outside of research studies." His reasoning: the study populations were highly educated, high-income, not very diverse, so a group-level prediction doesn't translate to a reliable answer for the individual in your office. Why it moves numbers: LabCorp (LH) and Quest (DGX) already sell this test, and the fattest imagined market is mass screening of the "worried well." This week the field's own guideline-writers put a formal red light on exactly that use. Last week the bottleneck was "who pays?" This week it's "the experts are telling doctors not to order it yet." Both point the same direction: the near-term blood-test market is the symptomatic patient, not the healthy 60-year-old who's nervous about their memory.

Combination therapy just took its first real step. For months, "the future is combinations, amyloid plus tau, like cancer" has been a slide at every conference. This week it became a dosed patient. Rabinovici on JAMA Medical News: "At UCSF, we just randomized the first patient for an NIH-funded study called the Alzheimer-tau platform... this clinical trial is testing whether donanemab, an anti-amyloid treatment, an anti-tau vaccine, or a combination of the two can have a benefit." Why it matters: this is Lilly's Kisunla (donanemab) being tested head-to-head and in combination with a tau shot, under public funding. If combinations work, every patient needs both an amyloid readout and a tau readout, which makes the diagnostics complex bigger, not smaller, and makes the reimbursement fight above even more consequential. He also made a pointed aside that the trial is NIH-funded and that "continued federal funding of research in the United States... has made a huge difference," a quiet flag that the pipeline's early, unglamorous science leans on government money.

A patient explained, in plain terms, why he chose Leqembi over Kisunla. The most useful demand-side color came from an actual patient. On Being Patient (July 28), Scott Redfern (a patient, lived experience, the ultimate operator), diagnosed at 61 with younger-onset Alzheimer's, described being offered both drugs and picking Biogen and Eisai's Leqembi (lecanemab): "I chose to go with lecanemab because of the prolonged treatment that you can go on to not only clear the plaque out of your brain, but also prevent those toxic oligomers in the future from... continuing to build up the plaque." That is, almost verbatim, Eisai's pitch for staying on the drug long-term, and it's the exact axis on which Leqembi (ongoing dosing) and Lilly's Kisunla (dose until the plaque clears, then stop) compete. Why it moves numbers: patients are hearing and repeating the "keep treating to stop it coming back" argument, which favors a longer, stickier Leqembi revenue tail if it holds. Just as telling was his diagnostic journey: a perfect 30/30 on the office memory test, then "the newer blood test, the PTAO-217, which came back inconclusive," and only then an amyloid PET scan that confirmed it. A real-world reminder that the blood test is a triage tool, not the final word; the expensive scan still closes the case.

The debate

Does under-the-skin dosing plus cheap blood tests plus a widening pipeline finally turn anti-amyloid (and now anti-tau) into a multi-billion-dollar franchise, or do modest benefit, brain-bleed risk, and diagnostic bottlenecks keep uptake weak while the tau bets stay unproven?

Bull: The science keeps clearing bars nobody had cleared. A tau drug just reversed established tangles on brain scans, a genuine first, and posted 26% slowing at its best dose, right alongside the marketed amyloid drugs. The blood test is so good it can now forecast decline a decade out. And the field's consensus endgame, combinations, isn't a theory anymore; the first patient is dosed. A patient on the podcast is choosing the longer-treatment drug and repeating the durability pitch unprompted. Every one of these expands the measure-then-treat market.

Bear: Look at what actually happened, not the press release. The tau drug missed, its low dose beat its high dose, and two independent experts now warn there may be a ceiling where removing more tau backfires, a fundamental design risk for the whole broad-tau approach, not a rounding error. The blood test's own guideline-writers just told doctors not to use it on healthy people, capping the screening dream from the medical side to match the payment problem from the money side. And a real patient's blood test came back inconclusive, sending him to the pricey scan anyway. Narrow eligibility plus an unpaid, not-yet-recommended front-door test plus drugs that are "good, not great" is not the shape of a runaway franchise.

Net: a substantive, information-rich week that, on balance, deepened the bull case on science and the bear case on access. The tau reversal is real and combinations are moving, that's the long-term bull story getting more concrete. But the near-term commercial engine still runs into the same wall from two directions now: payers won't fund the cheap test, and the experts won't recommend it for the mass market. The re-rating catalyst for this whole complex remains a policy decision, Medicare coverage plus a guideline that green-lights broader testing, not the next molecule.

Stocks in play

  • BIIB (Biogen): Bull: its antisense tau drug diranersin did something no rival ever managed, shrinking existing tangles on PET and lowering spinal-fluid tau up to 60%, with best-dose cognitive slowing (26% on CDR-SB) that rivals the marketed amyloid drugs; combination therapy, the field's endgame, is now in the clinic. Bear: the Phase 2 missed, the dose-response was backwards (low dose best, high dose worst), a dose-dependent confusional-state side effect showed up, delivery is via spinal tap, and a second independent expert has now endorsed the "you can lower tau too much" ceiling worry. Next catalyst: the Phase 3 design and dosing decision. (JAMA Medical News, July 31)
  • IONS (Ionis): Bull: Biogen's antisense partner on the platform behind the first-ever tau-tangle reversal, the technology is the asset, and it just showed it can clear tau deposits thought to be permanent. Bear: still one messy, missed Phase 2, no near-term revenue from it, and the same tau-window risk hangs over the whole program. Next catalyst: Phase 3 progression. (Ionis's role is prior context; this week's JAMA Medical News episode named only Biogen, July 31)
  • LLY (Lilly): Bull: its Kisunla (donanemab) is the anti-amyloid backbone in the first NIH-funded combination trial, a free, publicly funded shot at proving Lilly's drug plus a tau vaccine beats either alone; Lilly also makes the amyloid and tau PET tracers that Medicare fully covers. Bear: no fresh Kisunla launch metrics this week, and the patient color cut the other way, with a real patient choosing rival Leqembi specifically for its keep-treating model over Kisunla's stop-when-cleared approach. Next catalyst: the next Kisunla sales print; combination-trial readouts (years out). (JAMA Medical News, July 31; Being Patient, July 28)
  • BIIB and Eisai (Leqembi): Bull: a patient articulated the durability pitch unprompted, choosing Leqembi for ongoing treatment to keep plaque from rebuilding, the exact case for a longer, stickier revenue tail. Bear: it's one patient; the drug's real-world eligibility and persistence are still the open questions, and no sub-Q (IQLIK) ramp color surfaced this week. Next catalyst: Leqembi sales trajectory and sub-Q autoinjector uptake. (Being Patient, July 28)
  • DGX (Quest) and LH (LabCorp): Bull: both sell the pTau-217 test, which a major new study showed can predict who declines over 5 to 10 years (24% and 38% of high and very-high scorers within five years), real prognostic power. Bear: the field's updated clinical guideline still recommends against using it in healthy, symptom-free people outside research, capping the mass-screening market from the medical side just as reimbursement caps it from the money side; and a real patient's test came back inconclusive, sending him to PET anyway. Next catalyst: a guideline expansion and/or Medicare coverage for symptomatic use. (JAMA Medical News, July 31; Being Patient, July 28)
  • QTRX (Quanterix), C2N and Roche diagnostics, read-through only: Bull: the underlying pTau-217 technology keeps validating (now a decade-out risk predictor). Bear: no company-specific coverage this week, and the same "great test, no green light" wall applies to all of them; a competing modality (see below) is also emerging in the literature. Next catalyst: CMS coverage or a guideline change. (JAMA Medical News, July 31)
  • LNTH (Lantheus) and GEHC (GE HealthCare), read-through only: Bull: amyloid PET remains the confirmatory gold standard, as a patient's own journey (inconclusive blood test, then PET confirms) reminded us; PET stays reimbursed while the blood test does not and isn't yet recommended for screening. Bear: no direct PET-vendor commentary this week, the read is thematic, and the long-run substitution risk from cheaper blood tests is unchanged. Next catalyst: any CMS move on blood-biomarker coverage. (Being Patient, July 28)
  • Cognito Therapeutics (private, device watch, no ticker): Bull: on The Empowering Neurologist (July 28), Dr. David Perlmutter (an expert and skeptic, promoting a book) walked through the 40-hertz light-and-sound "gamma" approach pioneered at MIT, citing a randomized six-month trial of 76 people with mild-to-moderate Alzheimer's showing slower memory-test decline versus sham. A non-drug angle on the same disease. Bear: small trial, private company, a mechanism (targeting brain immune cells) that sits outside the mainstream amyloid and tau consensus and remains unproven at scale. Next catalyst: larger confirmatory data.

Read-throughs

  • Blood diagnostics (DGX, LH, QTRX, C2N, Roche): accuracy proven, permission denied. The story keeps sharpening: the test can now forecast a decade of risk, but the guideline-writers explicitly won't recommend it for the healthy, and a real patient's test came back inconclusive. Near-term, these names live off symptomatic testing, not mass screening. Watch for a guideline change as closely as a CMS decision.
  • A new diagnostic competitor is stirring, circular RNA. On TT HealthWatch (July 31), hosts Elizabeth Tracey and Dr. Rick Lange (medical journalists) flagged a Nature Medicine study on a blood "circular RNA" signature, a 34-transcript pattern that, in a 1,221-person cohort, hit "almost 95%" accuracy, "higher than the predictive ability of plasma tau or amyloid PET," and started diverging "two to four years before clinical symptom onset." It's early, academic, and needs replication, but it's a reminder that pTau-217 may not be the last word in blood diagnostics, a slow long-run risk for anyone betting the category is a locked-in pTau-217 duopoly.
  • Symptom management (agitation), a real signal with a real warning. The JAMA recap also covered a not-yet-published AAIC study of a purified CBD/THC combo (2 mg THC, 100 mg CBD) in about 120 dementia patients with agitation: "84% of people were rated to be improved... versus only about 30% of the placebo group," an effect that held at 12 weeks. The catch was serious: "eight deaths in the active arm compared to three in the placebo arm... statistically significant," though none tied directly to the drug. Agitation at the end of life is a huge unmet need, but this is a caution flag, not an all-clear. No public pure-play here; it's a watch item for the neuro-symptom space.
  • The "beyond amyloid" paradigm crowd got louder, take with salt. The Empowering Neurologist (July 28) featured a full argument that Alzheimer's starts upstream of amyloid, in the brain's immune cells (microglia) flipping from protective to destructive, with lifestyle and metabolism as the levers, plus the 40-hertz device angle. And Science News Daily (July 27) covered an AI model ("Hickformer," from Carnegie Mellon, Pitt and UW in Science) mapping 3D-genome changes in Alzheimer's brain cells. Neither is investable this week, but both feed the long-running bear-of-the-whole-paradigm argument: if amyloid is downstream, the multi-billion anti-amyloid bet is aimed at the wrong target. Worth tracking as a source of tail risk to the entire complex.
  • Infusion centers and specialty pharmacy (OPCH): a small tell, a Leqembi patient described a two-hour infusion appointment as a routine part of his week ("I go in, I get my infusion, I come out and I go about my life"). As long as the marketed drugs are IV, the infusion-chair demand is real; the eventual shift to at-home sub-Q dosing is the slow headwind.

What changed vs last week

Last week was the conference box score; this week was the detailed film review, and it added real texture in three places.

On the tau drug, the data got specific and the bear brick got a second bricklayer. Last week we knew Biogen's tau drug had reversed the tau signal and that the market lopped about $2 billion off the stock. This week we got the dose-by-dose numbers (26%, 14%, 9% slowing, backwards) and, crucially, a second independent expert, an academic with no company tie, endorsing the exact "you can lower tau too much" ceiling worry that a BioCentury editor raised last week. When two unrelated experts converge on the same failure mode, it stops being a hot take and becomes a live design risk for the whole broad-tau field.

On diagnostics, the wall got a second layer. Last week the blood-test problem was "Medicare won't pay." This week the field's own guideline, refreshed at the meeting, added "and don't use it on healthy people yet anyway," even as a new study proved the test can predict a decade of risk. So the screening dream is now capped from both the payment side and the medical-recommendation side. That's a real, incremental negative for the mass-market blood-test thesis, and it keeps the near-term tailwind with PET.

On combinations, talk became action. Last week "combinations are the endgame" was consensus opinion. This week the first patient was dosed in an NIH-funded amyloid-plus-tau trial using Lilly's donanemab. Small step, but it's the first concrete one, and it reinforces that the diagnostics complex gets bigger if combinations win.