The 2023 headlines about two new Alzheimer’s drugs, lecanemab and donanemab, made it sound like the science was settled. Amyloid, a sticky protein that builds up in the brains of people with Alzheimer’s, could now be cleared by infusions. The 2026 evidence base says something more specific and more useful.
What two major 2026 papers say together
In April, the Cochrane Collaboration (the international gold standard for pooling medical evidence) published an analysis combining 17 randomized trials and 20,342 patients across every major anti-amyloid drug tested in mild cognitive impairment and mild Alzheimer’s (Nonino 2026). When they pooled all the data, the average effect on cognition was small, and the effect on everyday function was smaller. The risk of a side effect called ARIA (which can mean brain swelling or small bleeds that show up on MRI) went up.
The authors’ plain-language verdict: clearing amyloid from the brain does not appear to produce clinically meaningful benefit in the average patient, and future research should look at other mechanisms.
That is a big statement coming from Cochrane. It is consistent with the hypothesis that amyloid may sometimes be part of the brain’s protective response, and that addressing amyloid by itself is not sufficient to improve cognition.
Then there came a second paper in the same month.
A separate analysis of the Clarity AD trial (Perry 2026) looked specifically at the 85% of patients in the trial who did not carry two copies of a gene called APOE ε4. APOE ε4 is a common genetic variant that raises Alzheimer’s risk and also raises the risk of the ARIA side effect. In this group (the non-carriers and the people with just one copy), lecanemab produced a durable benefit at 18 months and, in an extension of the study, kept accumulating benefit out to 36 months. The people who started the drug later never caught up to those who started it earlier.
“Later starters never catch up” is the statistical signature of a drug that is affecting the disease, not just masking symptoms. If lecanemab were only boosting cognition temporarily, the late-starters would eventually look the same as the early-starters. That did not happen.
Both papers are true at the same time: the pooled average is underwhelming, while the genetics-defined subgroup treated early does better. To give a sense of magnitude, Clarity AD, the anti-amyloid trial with the best clinical signal, showed about a 0.5-point difference on CDR-SB (Clinical Dementia Rating Sum of Boxes). That translates to roughly 0.15 standard deviations of cognitive benefit. Most clinical research treats 0.3 standard deviations as a clinically important shift. The EVANTHEA preprint is showing more than 1 standard deviation of cognitive improvement. So the impression is that changing amyloid alone is not sufficient to significantly move the needle on cognition.
The donanemab piece and the policy question
A third paper, by Tang and colleagues (Tang 2026), showed that donanemab in mild cognitive impairment produces what should happen if the drug is actually working biologically: about 80% of patients cleared the amyloid, and the downstream markers (tau, another Alzheimer’s-related protein, and GFAP, a marker of brain inflammation) also moved in the right direction, alongside very modest cognitive improvement. That kind of convergent movement across multiple biomarkers is what a genuine disease-modifying therapy should produce.
A fourth paper, a policy piece by Rossini and Pappalettera (Rossini 2026), asked the question most neurologists were privately asking: should every patient with biomarker-positive mild cognitive impairment get one of these drugs? Their answer was no, because the side-effect risk is real, the population is huge, and a blanket policy doesn’t fit a precision-medicine disease.
What this means for you
The critical question that remains unanswered is under what circumstances should anti-amyloid drugs be combined with a complete, precision medicine approach. We have a couple of thoughts:
- APOE genetic testing should happen in all people at risk, because your genotype meaningfully changes both the benefit side and the risk side of the equation. Two copies of APOE ε4 (homozygotes) carry substantially more risk of the ARIA side effect. The FDA has currently black boxed this subgroup.
- A serious conversation about early treatment versus late needs to happen, because almost all of the data, including monoclonal antibodies shows that starting sooner before too much damage accumulates makes a difference. If your MoCA is above 19, this is a critical time to change the course of disease! You want to see improvement in cognition as well as improvement in anatomy, biomarkers and function.
- Biomarker confirmation is always a good idea, meaning a blood test called plasma p-tau217 (a test that picks up early Alzheimer’s biology from a regular blood draw), and often an amyloid PET scan to confirm that amyloid is actually driving the cognitive symptoms. Follow-up biomarkers at 6 and 12 months to confirm the drug is doing what it’s supposed to do biologically also makes sense. A patient whose numbers aren’t moving is telling you something a clinical exam won’t reveal for another year.
- Precision medicine should be tried first where possible as it can address the root causes. It is certainly possible that the precision medicine and anti-amyloid monoclonal antibodies may work best in combination for selected patients. This however, has not been tested. There are two subgroups where we could see this: patients who do not respond well to precision medicine alone, where amyloid burden may be high enough that both treatments are required, and patients with focal atrophy patterns such as primary progressive aphasia. At Rezilir, we have for the last decade practiced precision medicine for cognitive decline with an emphasis on addressing the root causes. In the last three years, we have been co-principal investigators in the EVANTHEA trial, which did not use monoclonal antibodies. In a later post, we will discuss potential options for research in this area.
The 2026 data on anti-amyloid monoclonal antibodies shows that the antibody removes a target. It does not remove the reason the target accumulated in the first place. That reason is usually five or six reasons: blood vessel health, metabolism, inflammation, hormones, infections, environmental exposures. Those reasons are what the rest of this series is about.
References — Part 1
Rossini PM, Pappalettera C. Should all MCI with Alzheimer’s biological diagnosis receive anti-amyloid therapy? Cell Death & Disease. 2026;17(1):212. doi: 10.1038/s41419-026-08456-z
Nonino F, Minozzi S, Sambati L, Del Giovane C, Baldin E, Bassi MC, De Santis C, Gonzalez-Lorenzo M, Vignatelli L, Filippini G, Richard E. Amyloid-beta-targeting monoclonal antibodies for people with mild cognitive impairment or mild dementia due to Alzheimer’s disease. Cochrane Database of Systematic Reviews. 2026;2026(4). doi: 10.1002/14651858.CD016297
Perry R, Kipps C, Soto Martín ME, Bozzali M, Logroscino G, Trafford S, Dhadda S, Kanekiyo M, Goodwin A, Hodgkinson M, Hersch S, Irizarry M, Kramer L, Froelich L. Lecanemab for treatment of individuals with early Alzheimer’s Disease (AD) who are apolipoprotein E ε4 (ApoE ε4) non-carriers or heterozygotes. The Journal of Prevention of Alzheimer’s Disease. 2026;13(4):100507. doi: 10.1016/j.tjpad.2026.100507
Tang L, Xu Y, Zhao H. Donanemab therapy in Alzheimer’s disease with mild cognitive impairment: Convergent amyloid, tau, and plasma biomarker normalization with cognitive improvement. The Journal of Prevention of Alzheimer’s Disease. 2026;13(5):100533. doi: 10.1016/j.tjpad.2026.100533
