Newsletter · · Ashutosh Agarwal

Doctors Pump the Brakes on the Alzheimer's Blood Test - The Neuro & Alzheimer's Pipeline - Week of August 23, 2026

Two practicing memory specialists urged caution on the Alzheimer's blood test while a prominent skeptic attacked the anti-amyloid drugs' cost and benefit, sharpening the diagnostics-versus-efficacy debate for the week of August 17 to 23, 2026. The same trials get framed as slowing decline by about 30 percent or as half a point on an 18-point scale, and that gap is the whole bull-bear case.

The Neuro & Alzheimer's Pipeline

Week of August 23, 2026: Doctors Pump the Brakes on the Alzheimer's Blood Test

This week the Alzheimer's debate turned on two fronts at once: whether to order the blood test, and whether the drugs it feeds are worth their cost and risk. Four episodes touched the disease, and together they flipped the newsletter's running story on its head.

Last week the signal was a family doctor in a safety-net clinic pushing to order the blood test, the first real crack in the "doctors won't order it" wall. This week two practicing memory specialists pushed back the other way, and a prominent skeptic went on a popular wellness podcast and attacked the drugs at the end of the funnel. The demand-side debate this newsletter keeps circling did not advance so much as it got louder on both sides. Here is what was actually said, by whom, and why it matters for the numbers.

(Quick vocabulary, used throughout: amyloid and tau are the two toxic proteins that build up in an Alzheimer's brain and are the target of today's approved drugs. pTau-217 and pTau-181 are forms of tau you can now measure: pTau-217 from an ordinary blood draw, and both from spinal fluid. A-beta-42/40 ratio is an amyloid measure; a "ratio" test combines two markers and tends to be more accurate than a single one. MCI = mild cognitive impairment, the early stage before full dementia. Anti-amyloid therapy is the drug class, Lilly's Kisunla (donanemab) and Eisai/Biogen's Leqembi (lecanemab), that clears amyloid and modestly slows decline; both are infusions and both carry a brain-swelling/bleeding risk called ARIA. PET and CSF are the two long-established, FDA-approved ways to confirm amyloid: a brain scan and a spinal-fluid draw. Centiloid / SUVR is the number a PET scan spits out to grade how much amyloid is in the brain. TREM2 is a receptor on the brain's immune cells that some drugmakers are now trying to switch on. CME = certified continuing-medical-education content doctors listen to for credit: teaching, not company news, and sometimes funded by a drugmaker, which matters for how you weigh it.)

TL;DR

  • The diagnostics debate swung back toward caution. A Cedars-Sinai memory specialist warned this week against ordering the Alzheimer's blood test outside of patients who already have symptoms, the near-opposite of last week's family doctor who wanted to order it routinely. His worry: false alarms, anxiety, and even damage to a patient's long-term-care insurability. Symptomatic-only keeps winning.
  • The drug-efficacy fight reopened, loudly. Neurologist and best-selling author David Perlmutter used a popular longevity podcast to hammer the anti-amyloid drugs: ">20%" brain bleeds/swelling, "$40,000 a year," and a benefit he framed as "one half a point on an 18-point scale." Same trials a Lilly-funded teaching program described as slowing decline "by about 30%." Both numbers are real; they're the same result described two different ways, and the gap is the whole bull-bear argument in one sentence.
  • The only stock discussed on its own merits was Lilly, and only as a chart. On a stock-technicals show the hosts called Lilly the top big-pharma name, with the driver being weight-loss growth; Alzheimer's and Kisunla were not part of that conversation. Lilly's Alzheimer's exposure surfaced only inside the diagnostics debate, as the funder of a teaching program and the maker of donanemab.

What's New

A Cedars-Sinai memory specialist says: don't order the blood test unless the patient already has symptoms, and beware the direct-to-consumer version. On Brain Talk (Being Patient) ("What to Remember When You Are Forgetting," Aug 18), Dr. Zaldy Tan (OPERATOR/INSIDER, a practicing memory-disorders specialist at Cedars-Sinai) was blunt: "the recommendation is only test this in people who are symptomatic, who are showing signs of memory loss." He warned specifically about tests ordered "through a precision medicine company or certain primary doctors who may not be familiar with the result," which he said "may result in more confusion and anxiety than any real thing," because "right now, we do not have a way to treat someone who has biomarker positivity but no symptom of the disease." He even flagged a financial hazard: a positive result "may have implications... [for] your potential insurability for your long-term care insurance, health insurance." Why it moves numbers: this is the exact opposite emphasis from last week's safety-net family doctor who said "start ordering biomarkers." The two specialists who spoke this week both drew the line at symptomatic patients, which caps the near-term testing market at the sick, not the worried-well, and pushes back directly on the "blood test walks into primary care" thesis from seven days ago.

Tan reframed a positive amyloid test as a risk marker, not a diagnosis: the cholesterol analogy. Same episode: he stressed that a positive result "means that you have the Alzheimer's pathology. It doesn't mean you have Alzheimer's disease," comparing it to high cholesterol, a heart-attack risk factor, not a heart attack. He cited the 90-plus study and the Nun study, where people had amyloid in their brains yet "never had a sign of Alzheimer's." For a listener already diagnosed with MCI, though, he was constructive: get an amyloid PET scan and a structural MRI, rule out other causes, and "talk about what your treatment options are because now we have monoclonal antibodies," naming lecanemab (Leqembi) and donanemab (Kisunla). Why it matters: it sharpens the funnel the whole space depends on: the drugs are for symptomatic MCI/mild patients confirmed by scan or spinal fluid, not for a healthy 55-year-old who bought a blood test out of curiosity.

A Lilly-funded teaching program laid out the confirm-the-diagnosis toolkit, and quietly reminded everyone the drugs slow decline "by about 30%." On Keeping Current CME ("Alzheimer's Disease Precision Diagnostics, Part 2: Cerebrospinal Fluid Biomarkers," Aug 20), Dr. Ana Pereira (OPERATOR/INSIDER, a cognitive neurologist at Mount Sinai; the program was "supported by an independent educational grant from" Eli Lilly, disclosed up front) walked through spinal-fluid testing. Her useful takeaways: spinal-fluid biomarkers have "over 90% concordance with amyloid PET," the "ratio of CSF analytes rather than A-beta-42 alone is the most accurate," and for tracking a drug's effect, PET is best; she described a patient who started at "98" centiloids of amyloid and, after therapy, "now [is] zero... basically entirely removed the amyloid from the brain." On the drugs themselves she said lecanemab and donanemab "very robustly remove amyloid" and it's "predicted to slow progression by about 30%." Why it matters: hold that "30%" next to Perlmutter's "half a point on an 18-point scale" below: they are the same clinical result, and the distance between them is exactly what the market can't agree on.

A prominent skeptic used a popular longevity podcast to attack the anti-amyloid drugs head-on. On the Extend Podcast with Darshan Shah, MD (Ep. 192, "Dr. Perlmutter: How to Protect Your Brain," Aug 20), Dr. David Perlmutter (PUNDIT, a neurologist and best-selling author, and a vocal anti-amyloid critic; note he also plugs supplements and sits on a product company's board in this same episode, so weight him as an opinionated advocate, not a neutral source) argued the approved drugs deliver "one half a point on an 18-point Alzheimer's scale... at great cost and great risk," with "more than 20% of these people get brain hemorrhages and or brain swelling," an "IV every two weeks at a cost of $40,000 a year that Medicare is covering." He said he "had this discussion with the director of Medicare and Medicaid for the United States a couple of weeks ago," and, his claim, which I can't independently verify from the podcast, that the Leqembi advisory panel voted "10 said no" to approval with one yes and one abstention before the FDA cleared it anyway. He contrasted this with a small Ornish/Tanzi lifestyle study ("57 people... 20 weeks") in which he said "70% of these people not only stabilized, but also many of them improved." Why it matters: this is the bear case on the drugs (not just the diagnostics) getting a mainstream, large-audience megaphone, the kind of narrative that shapes patient demand and reimbursement politics even when the speaker has an axe to grind.

The number that is the whole debate: "slow progression by about 30%" (Lilly-funded CME) and "one half a point on an 18-point scale" (skeptic) describe the same anti-amyloid trials. One is a relative percentage, the other an absolute score. Which one a doctor, a patient, or a payer believes is the number that decides how big this franchise gets.

The Debate

Does under-the-skin dosing + cheap blood tests + broader coverage finally turn anti-amyloid (and eventually 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 diagnostic machinery is real and maturing. A Mount Sinai neurologist confirmed spinal-fluid tests now agree with PET scans more than 90% of the time, that ratio-based tests are the accurate ones, and, vividly, that amyloid can be driven from 98 centiloids to zero on a follow-up scan. The drugs "very robustly remove amyloid" and slow decline "by about 30%." For confirmed MCI patients, even the cautious specialist (Tan) sends them straight toward monoclonal-antibody treatment. The infrastructure to diagnose and dose exists; it's spreading; and it works on the biology it targets.

Bear: The two people who actually diagnose these patients for a living both told listeners to slow down on testing. Tan won't order the blood test outside symptomatic patients, warns it can wreck a patient's insurability, and reminds everyone that a positive amyloid result "doesn't mean you have Alzheimer's disease." Perlmutter, loudly, to a big audience, reframed the drugs' benefit as trivial ("half a point") against a ">20%" brain-bleed rate and a $40,000 price tag. And "remove amyloid from the brain" is not the same as "make the patient better." That is the entire critique. The screening-the-healthy market still doesn't exist, no prevention drug is approved (Tan noted "at least three large clinical trials" are running but "not FDA approved yet"), and the tau bets remain untested in the clinic.

Net: last week the bulls got their first operator-side adoption signal. This week the bears got two operator voices pulling the other way plus a viral efficacy takedown. Nothing structural changed: no sales, no coverage decision, no pipeline read-out. The debate simply got more crowded, and the burden of proof is still on the same unreleased catalysts: hard Kisunla and Leqembi sales prints, a guideline that green-lights broader testing, and a clean Medicare coverage path.

Stocks in Play

Everything below is a read-through, or in Lilly's case chart talk, flagged as such rather than dressed up as fresh fundamentals.

  • LLY (Eli Lilly). The only real stock mention: on Stock Market Today With IBD (Aug 18), the hosts (PUNDIT/technical) called Lilly "the top big pharma stock out there" and "stock of the day," discussing chart levels around $1,200–$1,250 and a strategy of "buying stock in steps." Notably, the driver was weight-loss growth, "still some runway just on its weight loss side," and Alzheimer's/Kisunla was never mentioned. Separately, Lilly's name showed up twice this week in the diagnostics debate: as the disclosed funder of the CME program above, and as the maker of donanemab in both specialists' treatment discussions. Bull: the GLP-1 engine is doing the heavy lifting and the diagnostic funnel that feeds Kisunla keeps getting taught to more doctors. Bear: Kisunla is still a call option in the P&L, zero launch metrics for a third straight week, and this week a widely-heard critic attacked the class it belongs to. Next catalyst: actual Kisunla starts/infusion data (none this week). (IBD; Keeping Current CME; Brain Talk)
  • BIIB (Biogen) / Eisai, read-through only. Both specialists named lecanemab (Leqembi) as an available, appropriate treatment for confirmed MCI/mild patients; Perlmutter singled it out as his prime example of a drug he thinks shouldn't have been approved. Bull: it stays in the standard-of-care conversation for symptomatic patients. Bear: a mainstream podcast just told a large audience it's dangerous and barely works, and there was no ramp, persistence, or IQLIK sub-Q data this week. Next catalyst: Leqembi/IQLIK commercial numbers. (Brain Talk; Extend)
  • Blood-diagnostics complex (C2N, Roche, QTRX / Quanterix, Fujirebio, LH / LabCorp, DGX / Quest), read-through only, and this week it's a headwind. Bull: the pTau-217 and ratio tests are now standard talking points, and confirmed patients still get funneled to treatment. Bear: the specialist voice this week was caution: symptomatic-only, and an explicit warning against precision-medicine-company and primary-care ordering of exactly the kind that drives volume. That's a narrative step back from last week. Next catalyst: a guideline or Medicare coverage decision that green-lights broader testing. (Brain Talk; Keeping Current CME)
  • PET imaging (LNTH / Lantheus, GEHC / GE HealthCare), read-through, and a rare positive. The CME specialist made PET the preferred tool for monitoring a patient before and after anti-amyloid therapy (her 98-to-zero centiloid example), and Tan sends confirmed MCI patients for an amyloid PET scan. Bull: as more patients start therapy, PET's monitoring role is a recurring-scan story, and it's the objective gold standard. Bear: cheaper spinal-fluid and blood tests are explicitly framed as the efficient first line; PET is reserved for confirmation and monitoring. Next catalyst: none named this week. (Keeping Current CME; Brain Talk)
  • TREM2 drug developers, read-through only, and a rare pipeline data point. Perlmutter, discussing the brain's immune cells, noted that a company developing "TREM2 agonists or TREM2 targeting drugs... was just bought by Sanofi for $420 million," his words, offered as "not a stock recommendation" and not independently confirmed here. Bull: validation that big pharma is putting money into non-amyloid, immune-based mechanisms, the "there's more to the story than beta amyloid" thesis. Bear: early-stage, unproven in the clinic, and a passing mention on a wellness podcast, not a pipeline update. Next catalyst: none named. (Extend)

Read-throughs

  • Diagnostics (C2N, Roche, QTRX, LH, DGX): the demand-side pendulum swung back to caution. Two weeks ago a neurologist wouldn't order the blood test; last week a family doctor said order it routinely; this week a Cedars-Sinai specialist said don't, except in symptomatic patients, and warned against the direct-to-consumer and primary-care ordering that would grow volume fastest (Brain Talk). The through-line across all three weeks is now unmistakable: real-world testing volume is confined to symptomatic patients, and the screening-the-healthy market that would make this a mass diagnostics category still doesn't exist. The investable read is unchanged and, if anything, reinforced: near-term volume comes from work-ups of people who already have symptoms, gated by reimbursement rules.
  • The efficacy gap is the real risk to the whole chain. "Slows decline ~30%" versus "half a point on an 18-point scale" isn't two facts: it's one fact told by a friend and an enemy. Every downstream name (imaging, diagnostics, infusion, the drugmakers) ultimately depends on doctors, patients, and payers deciding the benefit is worth the ARIA risk and the cost. This week that argument got a louder skeptic (Extend) and a paid-but-credible defender (Keeping Current CME). Neither moved the trial data; both moved the narrative.
  • A non-amyloid, immune-based angle is quietly attracting capital. The Sanofi/TREM2 mention (Extend) is a small, unverified data point, but it fits a real theme: the smart money is increasingly interested in mechanisms beyond clearing plaque: the brain's own immune cells (microglia), TREM2, inflammation. If amyloid clearance keeps disappointing on the "does the patient get better" question, that's where the next franchise is likely to come from, and it's years away.

What Changed vs Last Week

Last week's story was a family doctor cracking the demand-side wall, ordering the blood test routinely in a safety-net clinic and putting a number on it (accuracy ~40% → 90%). This week the wall pushed back. A Cedars-Sinai memory specialist told listeners to not order the blood test outside symptomatic patients and to be wary of precision-medicine-company and primary-care testing, the near-mirror-image emphasis (Brain Talk). Reconciled, the last three weeks say the same thing more sharply than any single week did: order for symptoms, not for curiosity.

What's genuinely new: the debate about the drugs (not just the diagnostics) came back to the surface, and it did so on a large-audience wellness podcast rather than a specialist one. Perlmutter's "half a point / >20% bleeds / $40,000" framing (Extend) sitting one day apart from a Lilly-funded program's "slows progression by about 30%" (Keeping Current CME) is the cleanest side-by-side of the bull and bear efficacy cases we've had in weeks.