Critique: The Profit Motive Behind New Alzheimer’s Treatments

Source: jacobin
Authors: ByJeanne Lenzer
URL: https://jacobin.com/2026/05/alzheimers-pharma-testing-treatment-profit

What the article reports

The article argues that a pharmaceutical-industry-funded network — including the Alzheimer's Association, contract research organizations, paid advocates, and sympathetic legislators — is pushing widespread Alzheimer's blood testing and preventive drug treatment on a population that has little to gain and significant potential for harm. It profiles two patients harmed by Kisunla, documents financial ties between Alzheimer's Association funders and the drugs/tests being promoted, and cites studies showing the tests' low predictive value in cognitively normal people. It also scrutinizes the ASAP Act, legislation that would require Medicare to cover screening tests.

Factual accuracy — Mostly-solid

The article's specific numerical claims hold up to internal scrutiny and are sourced. The 18 percent statistic ("only 18 percent of those with a positive blood test developed dementia during a ten-year follow-up period"), the 37 percent brain-swelling rate from the Kisunla pivotal trial, the Lumipulse accuracy range ("89 to 98 percent accurate"), Redfield's $638,387 in Roche Diagnostics consulting payments, and the Alzheimer's Association's individual corporate donations (Eisai, Eli Lilly, Biogen, Novo Nordisk, Roche figures) are all specific enough to be falsifiable, which is a genuine craft strength.

Two claims warrant scrutiny. First, "less accurate than flipping a coin" — applied to cognitively normal people — is an authorial gloss, not a quoted finding; the cited data (18% developing dementia after a positive test over ten years) technically means the test has high false-positive rate for that outcome, but whether that constitutes worse than chance depends on baseline prevalence assumptions the article does not supply. Second, the article states Marty Makary "has now been ousted" as FDA commissioner; this claim, embedded parenthetically, is asserted without any source or dateline context and could not be verified from the article's internal evidence. If inaccurate or outdated, it would be an error of fact; as written it is an unsourced assertion dropped into parentheses.

The Stanford/FDA correlation finding ("little to no" correlation between amyloid removed and benefit) and the 2024 Lancet commission conclusion are attributed and specific, which is good practice.

Framing — Tendentious

1. "less accurate than flipping a coin" — This is an authorial interpretive claim presented as fact, not a quotation from any study or expert. A piece scoring 4-6 on neutrality typically has unattributed framing at precisely these hinge points.

2. "a 'breakthrough' new Alzheimer's drug" — The scare quotes around breakthrough signal editorial skepticism without attribution; the reader is told what to think about the FDA designation rather than being given the designation and the counterargument separately.

3. "disease creep" — Introduced as a named phenomenon but attributed to unnamed "critics," not a specific researcher or paper: "Critics suspect a profit motive behind the phenomenon." This places an interpretive frame in authorial voice while nominally attributing it to an anonymous collective.

4. "bad tests designed to make people eligible for bad drugs" — The closing quotation from Adriane Fugh-Berman is placed in the article's final line, a structural editorial choice that lets the piece end on its most polemical note. Sequencing of quotes is a framing tool; no competing summary quote closes alongside it.

5. "the Alzheimer's Association, a private institution...whose recommendations on the disease are often taken as the final word" — "Often taken as the final word" is an authorial characterization with no citation, positioned to pre-undermine the association before its first substantive appearance.

6. "And while Marty Makary...has now been ousted, there's little reason to suspect his replacement will be any less industry friendly" — This parenthetical editorializes about a future unnamed official with no supporting evidence; it is opinion embedded in a news frame without attribution.

Source balance

Ratio: ~7 critical : 2 supportive (and both supportive voices are undercut by conflict-of-interest framing immediately adjacent to their quotes). No independent academic researcher who supports the tests, no clinician who believes early detection has value, no patient who credits early diagnosis with benefit, no biostatistician who defends the predictive utility in higher-risk subpopulations, and no drug company spokesperson is quoted substantively. Artemis and Lilly declined comment; that is disclosed, but their absence still leaves the ratio sharply unbalanced. Redfield did not respond — also disclosed — but his only quoted material is from op-eds, not direct engagement with the piece's claims.

Omissions

1. Proponents' strongest scientific case. The article does not engage with the argument — made by some researchers — that early intervention in a progressive disease may require accepting imperfect biomarker accuracy, or that subgroup analysis (e.g., APOE ε4 carriers) might yield better positive predictive value. A reader cannot assess whether the critics' position is a consensus view or one side of a live scientific debate.

2. Prior-administration and regulatory history. The FDA cleared these tests; the article does not explain what evidentiary standard the FDA applied or why the agency found it sufficient. Nor does it note that the USPSTF process (mentioned in passing) is the normal gatekeeping mechanism — and what that process has found or is reviewing.

3. Base-rate context for the 18% figure. What is the background rate of dementia development in cognitively normal 60+ year-olds over ten years without a positive test? Without that denominator, the reader cannot assess whether 18% is actually worse than population risk, the same, or higher. The article implies it is low without supplying the comparison.

4. Clinical distinctions between drugs. Leqembi and Kisunla are treated interchangeably throughout. The article notes both cause brain swelling but does not distinguish their approval bases, trial outcomes, or risk profiles. Readers who may be on Leqembi for diagnosed early Alzheimer's cannot gauge whether the critique applies to their situation.

5. The case for disclosure rather than prohibition. The article's implicit conclusion is that screening and treatment expansion is harmful. It does not engage with the policy alternative of informed-consent reform, conflict-of-interest disclosure requirements, or restricted use criteria — positions that critics of the current regime hold that fall short of "don't test, don't treat."

What it does well

• Specific financial disclosures. The article names dollar figures, years, and companies: "from 2021 to 2024, Redfield was paid $638,387 for consulting services by Roche Diagnostics" — this is the investigative core and it is done rigorously. Readers can verify or challenge each figure.
• Conflict-of-interest transparency about its own sources. The piece proactively notes that Devi "does not receive funding from the pharmaceutical industry" and that Christmas "does not receive industry support" — a practice that strengthens credibility even while the overall balance is thin.
• Patient narrative grounded in medical documentation. "A CT scan of Barbara Feuerstein showing brain swelling" — the article includes imaging and clinical specifics (confusion, inability to write her name, hip fracture) that give texture to the adverse-event data rather than relying solely on numbers.
• Attribution of key studies to named journals and institutions. References to the "2024 Lancet standing commission report," the "American College of Physicians and the Annals of Internal Medicine" panel, and the Stanford/FDA review give readers a paper trail rather than vague gestures to "experts say."
• Disclosure of source refusals. Artemis's refusal to speak and Lilly's non-response are noted explicitly — "Artemis refused to grant an interview...stating that they have a policy against speaking with the media" — rather than silently omitted.

Rating