Reviews/Anti Amyloid Antibodies/Amyloid-related imaging abnormalities requiring MRI monitoring
Safety · AMAB-SAF-01 · v2
This claim is about a known safety effect of a class of Alzheimer's drugs. When these antibodies strip amyloid out of the brain, some patients develop brain swelling or small bleeds that show up on MRI scans, a phenomenon called ARIA. Several large trials that compared the drugs against placebo all found far more brain-swelling events in treated patients: donanemab 24.4% versus 1.9% [2], bapineuzumab 15.8% versus 0.8% [6], and lecanemab 6.2% versus 1.4% [3]. These events tend to appear in the first few months and are more common in people carrying a particular genetic risk variant [1][2]. Because ARIA can only be seen on a scan and one trial explicitly concludes it requires safety monitoring [2], the panel treated the claim as well supported, while noting that the brain-bleed side of ARIA is muddier because untreated patients also show bleeds.
Three placebo-controlled antibody programmes independently show a large treatment excess of the edema form of ARIA: donanemab 24.4% versus placebo 1.9% [2], bapineuzumab 15.8% versus 0.8% [6], and lecanemab 6.2% versus 1.4% [3], with lecanemab phase 2 reporting ARIA-E incidence under 10% [1]. Onset is early, with 58.3% of donanemab ARIA-E events by the third infusion, and risk rises with APOE ε4 allele number [2], a gradient echoed by the correlation with serum concentration and higher incidence in ApoE4 homozygotes in [1]. In a bapineuzumab biomarker cohort, 154 of 1,512 participants developed incident ARIA-E [5]. [2] concludes that ARIA 'requires safety monitoring', and ARIA is an imaging-defined event detected on MRI. The microhemorrhage form is less clean: placebo ARIA-H is appreciable (13.0% in [2]) and in the Asian cohort lecanemab ARIA-H (14.4%) sits below placebo (16.2%) [3]. The contested point: The division is not over whether the drugs cause ARIA-E but over the microhemorrhage limb: placebo arms carry substantial ARIA-H (13.0% in [2]; 16.2% in [3], exceeding lecanemab's 14.4%), so ARIA-H is partly background cerebral amyloid angiopathy rather than a pure drug effect. The 'requiring MRI monitoring' limb is also a clinical recommendation stated in [2] rather than a comparative trial endpoint.
All 20 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Anti Amyloid Antibodies, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The ARIA-H (microhemorrhage) limb is substantially confounded by background cerebral amyloid angiopathy: placebo ARIA-H is high (13.0% in [2]) and in [3] lecanemab ARIA-H (14.4%) is actually below placebo (16.2%). The clean treatment-caused signal is ARIA-E, not ARIA-H.
[4] CAD106 is an active Aβ vaccine, not an administered monoclonal antibody, so it is off the exact wording 'anti-amyloid antibodies' and should not be counted as direct support.
Where it remains uncertain. The microhemorrhage limb is confounded by background amyloid angiopathy, as placebo ARIA-H is high (13.0% [2]) and even exceeds lecanemab ARIA-H in one cohort (16.2% versus 14.4% [3]).
Aleatory
ARIA-E occurrence varies by individual with APOE ε4 allele number, baseline microhemorrhages and cortical superficial siderosis [2], so who develops it is only partly predictable.
Epistemic
Because placebo arms carry substantial ARIA-H (13.0% [2]; 16.2% [3]), the share of microhemorrhages genuinely caused by the antibody rather than background amyloid angiopathy is not resolved by this corpus.
Inferential
The 'requiring MRI monitoring' conclusion is inferred chiefly from [2]'s stated conclusion plus ARIA being imaging-defined, not from a dedicated endpoint across the corpus.
The load-bearing papers from a corpus of 20, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Lecanemab ARIA-E incidence, early onset, mostly asymptomatic.
“In the phase 2 core and OLE, there was a low incidence of ARIA-E (<10%), with <3% symptomatic cases. ARIA-E was generally asymptomatic, mild-to-moderate in severity, and occurred early (<3 months).”
Panel auditInternal validity: medium · scope of inference: Early Alzheimer's disease patients treated with lecanemab 10 mg/kg; ARIA frequency, severity and timing.Key limitations: no placebo ARIA rate quoted · no sample size in abstract · single antibody
Placebo-controlled causation for both ARIA limbs; ARIA-E excess (24.4% vs 1.9%) is unambiguous.
“ARIA is an adverse event associated with donanemab treatment that requires safety monitoring.”
Panel auditInternal validity: high · scope of inference: Early symptomatic Alzheimer's disease aged 60-85 treated with donanemab; ARIA frequency, severity, timing and risk factors.Key limitations: no p values in abstract for the frequency contrast · single antibody
Lecanemab ARIA-E excess vs placebo; also shows ARIA-H not above placebo in this cohort (scope caveat).
“The most common adverse events of special interest were ARIA-H (lecanemab:14.4 %; placebo:16.2 %), ARIA-E (lecanemab:6.2 %; placebo:1.4 %), and infusion-related reactions (lecanemab:12.3 %; placebo:1.4 %).”
Panel auditInternal validity: medium · scope of inference: Asian-region early Alzheimer's disease patients on lecanemab; ARIA rates relative to overall Clarity AD.Key limitations: small subgroup (n=294) · ARIA-H not higher than placebo · no significance test
ARIA occurred in six cases, all of them strong serological (antibody) responders, linking induced anti-Aβ antibody response to ARIA.
Panel auditInternal validity: medium · scope of inference: Mild Alzheimer's disease patients given active Aβ vaccine CAD106; tolerability and ARIA among antibody responders.Key limitations: active immunotherapy not passive antibody · very few ARIA events · small sample
4 of 20 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Multiple placebo-controlled RCTs across different antibodies give a robust, dose- and time-coherent causal signal for ARIA-E, with the monitoring requirement directly asserted, warranting the Established tier short of the ceiling.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Monitoring-versus-no-monitoring outcome study. The 'requiring MRI monitoring' limb rests on [2]'s stated conclusion, not a comparative endpoint, so the patient benefit of scheduled MRI is unquantified in this corpus. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A prospective study comparing scheduled MRI surveillance against symptom-triggered imaging, measuring severe ARIA detection, hospitalisation and permanent sequelae.
Open question: Does scheduled MRI monitoring reduce severe ARIA outcomes compared with imaging only on symptoms? Why it is still open: [2] states ARIA requires safety monitoring but no included primary trial compares monitoring strategies against each other.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19