Anti-Amyloid Alzheimer's Drugs Show Trivial Cognitive Benefit With Real Brain Bleed Risk

Alzheimer's disease affects tens of millions worldwide, and the amyloid hypothesis — that clearing beta-amyloid plaques from the brain will slow cognitive decline — has driven billions in drug development. This Cochrane systematic review, published April 2026, is the most comprehensive independent synthesis to date of the clinical evidence for amyloid-beta-targeting monoclonal antibodies (Aβ-mAbs), covering nine agents across 17 randomized controlled trials with 20,342 participants. The mean participant age ranged from 70 to 74 years, and all studies were industry-funded. Seven antibodies were assessed: aducanumab (3 trials), bapineuzumab (4), crenezumab (2), donanemab (1), gantenerumab (4), lecanemab (1), and solanezumab (2).

For the primary efficacy outcome of cognitive function measured by the ADAS-Cog scale at 18 months, the pooled standardized mean difference was −0.11 (95% CI −0.16 to −0.06; 13 studies, 9,895 participants; moderate certainty). This translates to a reduction of just 0.85 points on the ADAS-Cog — a scale where a minimally clinically important difference is generally considered to be 3–4 points. The authors classify this as a trivial or null effect. For dementia severity measured by CDR-SB, the SMD was −0.12 (95% CI −0.24 to 0.00; 9 studies, 8,053 participants; low certainty), corresponding to 0.29 points — again, well below clinical significance thresholds.

Functional ability results were mixed. On the ADCS-ADL scale, the SMD was 0.09 (95% CI 0.03 to 0.16; 3 studies, 3,478 participants; moderate certainty), classified as trivial. On the ADCS-iADL scale (instrumental activities of daily living), a single lecanemab trial showed SMD 0.21 (95% CI 0.10 to 0.32; 1,252 participants; low certainty), and the ADCS-ADL-MCI scale showed SMD 0.23 (95% CI 0.12 to 0.33; 4 studies, 2,802 participants; low certainty) — both classified as small effects. These modest functional signals came from individual trials rather than pooled data, limiting their interpretability.

On the safety side, amyloid-related imaging abnormalities — edema (ARIA-E) — were substantially more common in treated patients. The absolute risk difference was 107 more cases per 1,000 (95% CI 77 to 148 more; 11 studies, 13,595 participants; moderate certainty). Symptomatic ARIA-E added 29 more cases per 1,000 (95% CI 22 to 38; moderate certainty). For ARIA-H (hemorrhage), three studies showed highly heterogeneous results (I² = 81%), preventing pooled analysis. Symptomatic brain hemorrhage data came from a single study (ARD 4 more per 1,000; 95% CI −1 to 31; moderate certainty). Critically, serious adverse events (ARD 6 more per 1,000; 95% CI −10 to 26; high certainty) and all-cause mortality (ARD 2 more per 1,000; 95% CI −3 to 11; high certainty) were not significantly increased.

The authors raise important methodological concerns. Risk of bias was rated low for safety outcomes but flagged as having 'some concerns' for efficacy outcomes, primarily due to functional unblinding — participants and investigators may correctly guess treatment assignment because of visible ARIA side effects, potentially inflating perceived benefit. The review also notes that most studies did not separately report symptomatic versus asymptomatic ARIA, making it difficult to fully characterize the safety burden. The authors' overarching conclusion is stark: successful amyloid clearance does not appear to produce clinically meaningful cognitive or functional benefit, and future Alzheimer's disease-modifying research should prioritize alternative mechanisms such as tau pathology, neuroinflammation, or synaptic preservation.