Do anti-amyloid monoclonal antibodies such as lecanemab and donanemab produce clinically meaningful cognitive benefit in early Alzheimer's disease, and how do the reported benefits compare with the ra

Bottom line. In the pivotal trials, lecanemab and donanemab slowed cognitive and functional decline by small absolute amounts that fell below established minimal clinically important differences, while raising amyloid-related imaging abnormality rates severalfold above placebo.

Do anti-amyloid monoclonal antibodies such as lecanemab and donanemab produce clinically meaningful cognitive benefit in early Alzheimer's disease?

No — the benefit is statistically significant but small in absolute terms and, according to the sources' own appraisals, below minimal clinically important difference (MCID) benchmarks. In TRAILBLAZER-ALZ 2, the 76-week iADRS difference was 2.92 points in the combined population and 3.25 points in the low/medium-tau subgroup (iADRS change −10.2 vs −13.1 and −6.02 vs −9.27, respectively), with CDR-SB differences of −0.7 and −0.67 (Direct; PMID: 37459141). The published MCIDs — approximately 1.0–1.6 points on CDR-SB and 5–9 points on iADRS — exceed these group-level differences, and an independent critical analysis concluded the average changes of ~0.5 CDR-SB and ~3 iADRS points are 'imperceptible to patients, their families, and their physicians' (Direct; PMID: 38951040). The 2026 Cochrane review, pooling 17 phase 3 trials, graded the 18-month class effects as trivial (CDR-SB ~0.29 points; ADAS-Cog ~0.85 points, SMDs of 0.11–0.12) and 'not clinically meaningful,' also downgrading certainty for possible functional unblinding bias (Direct; PMID: 41985900). Notably, regulators disagreed: the FDA granted traditional approval to lecanemab and approved donanemab in 2024, the EMA and UK MHRA licensed lecanemab, but NICE rejected NHS reimbursement of both drugs, the Dutch Zorginstituut refused lecanemab reimbursement for absent net benefit, and Germany's IQWIG concluded lecanemab offers no added benefit (Direct; PMID: 41985900).

How do the reported benefits compare with the rates of amyloid-related imaging abnormalities (ARIA)?

The benefit–harm comparison centers on a sub-point scale benefit set against ARIA rates that are severalfold higher on drug than placebo, and far higher than the absolute benefit. Key figures:

Pooled across the class, Cochrane estimated any ARIA-E risk ratio of 10.02 versus placebo (about 107 additional events per 1,000) against a mean CDR-SB benefit of ~0.29 points (Direct; PMID: 41985900). Most ARIA is asymptomatic and transient — donanemab's ARIA-E resolved radiographically in 96.0% of episodes (median 58 days), and 58.3% of first events occurred by the third infusion (Direct; PMID: 40063015); lecanemab ARIA-E occurred within 3 months in 71% and 6 months in 92% of cases, resolving by 120 days in 81% (Direct; PMID: 38730496). However, 72% of donanemab serious ARIA occurred within the first three infusions, and fatal outcomes were documented with both drugs, including deaths after thrombolytic administration for strokelike symptoms that proved to be ARIA (Direct; PMID: 40063015; Direct; PMID: 38730496). The Cochrane review's bottom line: given absolute benefit well below MCID, high ARIA risk, and heavy monitoring burden, the class 'does not offer a favourable benefit to risk balance' (Direct; PMID: 41985900).

Magnitude of benefit in the pivotal trials

Both approved antibodies met their primary endpoints with consistent, statistically significant slowing of decline, but cognition worsened in both arms of both trials. In TRAILBLAZER-ALZ 2, 23 of 24 gated outcomes were statistically significant, with the low/medium-tau subgroup (68% of randomized participants) showing larger differences than the combined population — 3.25 versus 2.92 iADRS points — consistent with greater benefit at earlier tau stages; importantly, benefit in patients without neocortical tau uptake (excluded by trial entry criteria) has not been definitively established (Direct; PMID: 37459141; Direct; PMID: 40155270). An independent reanalysis framed the same data against scale ranges: the lecanemab effect equals 2.5% of the full CDR-SB range (or 15.6% normalized to patients' remaining cognitive capacity), and the donanemab effect 2.4% of the iADRS range (or 8.5%) — figures the authors argue must accompany the 27–35% relative-slowing headlines, which divide a small absolute difference by placebo decline and implicitly assume continued linear divergence that remains unproven beyond 18 months (Direct; PMID: 38951040). Both pivotal trials were sponsor-funded (Eisai/Biogen for CLARITY-AD; Eli Lilly for TRAILBLAZER-ALZ 2), and the relative-slowing percentages originate from those sponsor analyses; two papers from one sponsor about that sponsor's product constitute a single source of evidence rather than independent corroboration (Derived; PMID: 41985900).

Clinical meaningfulness: MCID benchmarks and regulatory divergence

Against published MCID thresholds — CDR-SB ~1.0 point for MCI and ~1.6 for mild dementia; iADRS 5 for MCI and 9 for mild dementia — the group-level differences of ~0.45–0.7 CDR-SB and ~2.9–3.3 iADRS points fall short, as the TRAILBLAZER-ALZ 2 authors' own MCID definitions and the eNeuro reanalysis both note (Direct; PMID: 38951040). Cochrane reached the same verdict with pooled data: absolute differences of ~0.29 CDR-SB points and ~0.85 ADAS-Cog points, SMDs of 0.11–0.12 (trivial), with certainty downgraded for risk of functional unblinding — because infusion reactions and ARIA were far more frequent in treated arms, participants and informants could guess allocation, potentially biasing informant-based scales like CDR-SB toward apparent benefit (Direct; PMID: 41985900). Yet regulatory bodies split: the FDA granted traditional approval to lecanemab and approved donanemab in July 2024; the EMA and UK MHRA licensed lecanemab; meanwhile NICE rejected NHS reimbursement of both drugs because benefits were too small to justify cost, the Dutch Zorginstituut refused lecanemab for absent net benefit, and Germany's IQWIG concluded no added benefit (Direct; PMID: 41985900). The donanemab appropriate-use workgroup — an independent academic body — explicitly states its recommendations 'do not provide endorsement for treatment' and left efficacy meaningfulness unjudged, reflecting this unresolved debate (Direct; PMID: 40155270).

ARIA incidence, natural history, and serious outcomes

ARIA is the dominant treatment-related harm, though most events are asymptomatic and transient. In the donanemab pooled placebo-controlled analysis (984 treated, 999 placebo; Lilly-sponsored), any ARIA occurred in 37.0% versus 14.2% on placebo; ARIA-E in 24.4% versus 1.9% (5.8% vs 0.1% symptomatic); ARIA-H in 31.3% versus 13.0%; serious ARIA-E in 1.5%; and macrohemorrhage in 0.3% versus 0.2%, with ARIA-related discontinuation in 5.3% (Direct; PMID: 40063015). Three deaths among 984 donanemab-treated placebo-controlled participants and one placebo death were considered treatment related (Direct; PMID: 37459141), and additional ARIA-related deaths occurred in the open-label addendum and long-term extension, including one after tenecteplase for presumed stroke that proved to be severe ARIA-E (Direct; PMID: 40155270). For lecanemab, Core rates were ARIA-E 12.6% versus 1.7% placebo and ARIA-H 16.9% versus 8.9%, with symptomatic ARIA-E at 2.8%; serious ARIA-E adverse events occurred in 0.8% (Direct; PMID: 38730496). Timing concentrates early: 58.3% of donanemab first ARIA-E events occurred by the third infusion and 72% of serious ARIA within the first three infusions (Direct; PMID: 40063015), while lecanemab ARIA-E occurred within 3 months in 71% and 6 months in 92% of cases (Direct; PMID: 38730496). Resolution is the rule — 96.0% of 578 donanemab ARIA-E episodes resolved radiographically (median 58 days), and 68.6% of rechallenged patients had no recurrence (Direct; PMID: 40063015) — and lecanemab ARIA-E resolved by 120 days in 81% of cases (Direct; PMID: 38730496). APOE ε4 allele number is the strongest risk factor for both drugs (ARIA-E in 41.7–42.1% of donanemab homozygotes versus 11.0–14.8% of noncarriers; 32.6% versus 5.4% for lecanemab), alongside baseline microhemorrhages, superficial siderosis, elevated mean arterial pressure, and higher amyloid burden; antihypertensive use was associated with reduced risk, and antithrombotic use was not associated with increased ARIA frequency, though trials were underpowered for macrohemorrhage (Direct; PMID: 40063015; Direct; PMID: 38730496). Practically, thrombolytics are contraindicated without MRI excluding ARIA, and the donanemab AUR recommends against anticoagulants, while the lecanemab cohort showed a 2.7% intracerebral hemorrhage rate among patients on anticoagulants (Direct; PMID: 40155270; Direct; PMID: 38730496).

Benefit versus harm: independent appraisals and residual bias concerns

Weighing the two sides, Cochrane's integrated conclusion is that the class effect on cognition and dementia severity is trivial to small and does not reach MCID, any ARIA-E risk rises roughly tenfold (RR 10.02; ~107 more events per 1,000), symptomatic ARIA-E adds ~29 events per 1,000, and resource burden (PET/CSF confirmation, serial MRI, IV infusions) is substantial — leading to the conclusion that the drugs 'do not offer a favourable benefit to risk balance' for MCI or mild dementia (Direct; PMID: 41985900). A statistical analysis of unblinding quantified how much of the observed effect could be bias: assuming all ARIA unblinded participants, fully explaining the CDR-SB effects would require therapeutic-insight effects of 3.7 points (95% CI 2.0–5.6) for lecanemab and 3.3 points (2.1–4.4) for donanemab — larger than plausible placebo/unblinding effects alone — so unblinding 'is unlikely to fully account for' the observed effects, but may explain a substantial share (Direct; PMID: 38380503). The sponsor-funded lecanemab safety paper reports that multiple sensitivity analyses showed ARIA occurrence did not adversely impact cognition or function, but this is a sponsor analysis of its own product and should be weighted accordingly (Direct; PMID: 38730496). A counterpoint: donanemab's long-term extension showed growing benefit through 3 years with early-start superiority on CDR-Global progression and only 2.4 Centiloids/year amyloid reaccumulation after treatment completion — but these analyses used external ADNI controls rather than internal randomization (Direct; PMID: 41330788).

Contradictions & open questions

Several tensions remain unresolved across the sources. First, regulators reached opposite conclusions from the same trial data — FDA traditional approval versus NICE/Zorginstituut/IQWIG reimbursement refusals — and the field has no settled answer on whether a 0.45–0.7-point CDR-SB difference is meaningful (Direct; PMID: 41985900). Second, the relative-slowing percentages (27–35%) assume continued linear divergence of treatment and placebo curves, which the 18-month blinded data cannot confirm; long-term extension data use external or open-label controls, so sustained divergence remains a hypothesis rather than demonstrated fact (Derived; PMID: 38951040; PMID: 41330788). Third, APOE ε4 homozygotes carry the highest ARIA risk (FDA boxed warning for donanemab) yet efficacy within this subgroup is not separately established — the donanemab AUR urges caution without excluding them, while tau-PET-negative patients excluded from TRAILBLAZER-ALZ 2 have undefined benefit (Direct; PMID: 40155270). Fourth, whether unblinding inflated the CDR-SB effect in caregiver-informed scales cannot be quantified precisely from published data, since the share of ARIA events that actually unblinded participants is unknown (Direct; PMID: 38380503).

Evidence gaps & limitations

The evidence base has well-defined boundaries. Blinded efficacy data span only 18–27 months in a disease lasting decades, with only 6 of the 17 Cochrane-included trials following participants beyond 18 months, so long-term divergence and durability remain extrapolations (Direct; PMID: 41985900). Trial populations were 75–90% white, with marked underrepresentation of Black and Hispanic participants, and estimates suggest only about 5–15% of community-dwelling people with MCI or early AD would meet trial eligibility — limiting external validity for real-world patients who are older and more comorbid (Direct; PMID: 41985900). Real-world lecanemab cohorts show ARIA and discontinuation rates in the same range as trials but with short follow-up (a Japanese all-case surveillance reported ARIA in 7.1% over ~6 months; a US specialty clinic reported 22% over 6.5 months), and observational designs cannot establish effectiveness (Direct; PMID: 41849958; Direct; PMID: 40354064). The donanemab ARIA analysis itself flags that race/ethnicity effects could not be assessed due to underrepresentation and that 76-week data cannot address longer-term outcomes in participants who experienced ARIA (Direct; PMID: 40063015). Finally, discontinuation rules (donanemab's amyloid-PET-based stopping criteria) lack head-to-head randomized comparison with continued dosing, and plasma biomarkers cannot yet guide stopping decisions (Direct; PMID: 40155270).

Research notebook

What is the magnitude of cognitive/functional benefit of lecanemab in early Alzheimer's disease in the pivotal Phase 3 CLARITY-AD trial (CDR-SB and secondary outcomes)?

CLARITY-AD's benefit was statistically significant but arose without any cognitive improvement: cognition declined in both arms, and the treatment effect was smaller than that of the symptomatic drug donepezil; the same relative-rate framing was later applied to donanemab.

Status: verified • Confidence: high

• The eNeuro perspective reconstructs the CLARITY-AD data: over 18 months CDR-SB scores worsened by 1.21 points on lecanemab versus 1.66 points on placebo (baseline ~3.2), i.e., 0.45 points less decline, which the trial reported as a 27% slowing using the ratio (X−Y)/Y. (PMID 38951040, results)
• The Eisai-authored updated safety paper states that in Clarity AD, lecanemab's change from baseline on the primary CDR-SB outcome was less than placebo at 18 months, with all key secondary clinical outcomes supporting the primary outcome. (PMID 38730496, introduction)
• The 2026 Cochrane review includes CLARITY-AD (lecanemab 898 vs placebo 897, 18-month treatment) among the 17 phase 3 trials contributing to its pooled efficacy estimates, which showed small but statistically significant favorable differences for the second-generation antibodies. (PMID 41985900, discussion)
• The authors note that after 18 months of biweekly infusions the difference between antibody and placebo curves was only about half the difference seen with donepezil at 6 months, that cognition declined in both treatment and placebo arms, and that the relative 27% figure can look large while the absolute change (0.45 CDR-SB points) is imperceptible to patients. (PMID 38951040, discussion)
• Expressed against the full scale range, the lecanemab effect is 2.5% (0.45/18 CDR-SB points) or 15.6% when normalized to patients' remaining cognitive capacity — alternative metrics that the authors argue should accompany the 27% relative figure to avoid overstating benefit. (PMID 38951040, results)
• The Cochrane review notes that the phase 2 lecanemab trial showed a similar magnitude of change that was not statistically significant, and that the pivotal-trial effect sizes fall well below established minimal clinically important differences. (PMID 41985900, discussion)

What is the magnitude of cognitive/functional benefit of donanemab in early Alzheimer's disease in the pivotal Phase 3 TRAILBLAZER-ALZ 2 trial (iADRS and CDR-SB)?

Donanemab's trial population was enriched for earlier tau pathology, and greater benefit was seen in the low/medium-tau subgroup than in the combined population including high-tau participants; efficacy in patients without neocortical tau uptake (excluded from the trial) has not been definitively established.

Status: verified • Confidence: high

• In TRAILBLAZER-ALZ 2 the least-squares mean iADRS change at 76 weeks was −6.02 (donanemab) vs −9.27 (placebo), difference 3.25 (95% CI 1.88–4.62, P<.001) in the low/medium-tau population, and −10.2 vs −13.1, difference 2.92 (95% CI 1.51–4.33) in the combined population; 23 of 24 gated outcomes were statistically significant. (PMID 37459141, results)
• CDR-SB change at 76 weeks was −1.20 vs −1.88 (difference −0.67, P<.001) in the low/medium-tau population and −1.72 vs −2.42 (difference −0.7, P<.001) in the combined population. (PMID 37459141, results)
• The donanemab appropriate-use recommendations summarize TRAILBLAZER-ALZ 2 as showing 35% slowing of decline on iADRS in the low/medium-tau group and 22% in the combined population, with 36% and 29% slowing on CDR-SB respectively, and mean amyloid PET reduction of 88 centiloids versus placebo. (PMID 40155270, introduction)
• The eNeuro perspective contextualizes donanemab's 33% relative slowing: on the 144-point iADRS the absolute difference was 3.4 points (−10.2 vs −13.1 from a baseline of ~104), which is 2.4% of the full scale and about 8.5% when normalized to the relevant portion of the scale — far below the iADRS minimal clinically important difference. (PMID 38951040, discussion)
• TRAILBLAZER-ALZ 2 stratified participants by tau PET: 68% had low/medium tau and 32% high tau, with greater slowing of clinical decline in the low/medium-tau group across primary and secondary endpoints, suggesting greater benefit at earlier stages of tau deposition; tau PET is not required in practice, but clinical benefit in no/very-low-tau patients has not been definitively established. (PMID 40155270, introduction)
• The trial enrolled 1,736 participants with amyloid and tau pathology, allocating α=.04 to the low/medium-tau population outcomes and .01 to the combined population, with the low/medium-tau group showing larger numerical treatment differences on both iADRS (3.25 vs 2.92) and CDR-SB (0.67 vs 0.70 scaled differently). (PMID 37459141, results)

What are the incidence rates and clinical consequences of ARIA (edema and microhemorrhage/siderosis), including fatal and serious cases, for lecanemab and donanemab?

The clinical course of ARIA is usually benign: ARIA-E is predominantly asymptomatic, occurs early in treatment (lecanemab: 71% within 3 months, 92% within 6 months; donanemab: 58% of first events by the third infusion), resolves radiographically in most cases (donanemab: 96% of episodes, median 58 days), and ~69–71% of rechallenged patients did not have recurrence.

Status: verified • Confidence: high

• In the Clarity AD Core, ARIA-E occurred in 113/898 (12.6%) lecanemab vs 15/897 (1.7%) placebo, ARIA-H in 16.9% vs 8.9%, symptomatic ARIA-E in 2.8% and symptomatic ARIA-H in 1.2%; serious adverse events with ARIA-E were 0.8% and with ARIA-H 0.2%, and the 7 placebo and 6 lecanemab deaths in the Core were none considered related to study drug. (PMID 38730496, results)
• In the Core+OLE (1,612 lecanemab-treated), ARIA-E was 13.6% (3.3% symptomatic), ARIA-H 18.5% (1.7% symptomatic); ARIA-E was primarily mild-to-moderate radiographically (88.5%) and asymptomatic (96.7%), typically resolving within 4 months (81% by 120 days); 2 lecanemab deaths with concurrent intracerebral hemorrhage occurred in the OLE (1 after tPA, 1 on anticoagulants), and exposure-adjusted death rates were similar to Core placebo. (PMID 38730496, results)
• The Cochrane review pooled data showing amyloid antibodies increase ARIA-E risk about tenfold over placebo (RR 10.02, moderate certainty) while symptomatic ARIA-E remained uncommon (~29 additional events per 1,000), with intracerebral hemorrhage cases rare in individual trials. (PMID 41985900, discussion)
• Pooled across 984 donanemab-treated and 999 placebo participants: any ARIA 37.0% vs 14.2%; ARIA-E 24.4% vs 1.9% (symptomatic 5.8% vs 0.1%); ARIA-H 31.3% vs 13.0% (symptomatic 1.0% vs 0.3%); serious ARIA adverse events 1.6% (vs 0); macrohemorrhage 0.3% vs 0.2%; ARIA-related treatment discontinuation 5.3%. (PMID 40063015, results)
• The TRAILBLAZER-ALZ 2 primary report records ARIA-E in 205 participants (24.0%; 52 symptomatic) on donanemab vs 18 (2.1%; 0 symptomatic) on placebo, infusion-related reactions 8.7% vs 0.5%, and three deaths in the donanemab group and one in the placebo group considered treatment related. (PMID 37459141, results)
• Beyond the placebo-controlled trials, in the ongoing long-term extension an APOE4 heterozygote died of ARIA-E, and one participant died after receiving thrombolytics for strokelike symptoms that proved to be severe ARIA-E followed by intracranial hemorrhage; a participant with large baseline superficial siderosis developed fatal ARIA-H with cerebral hemorrhage. (PMID 40063015, results)
• The appropriate-use recommendations state 1.5% of ARIA-E and 0.4% of ARIA-H cases were serious including 3 ARIA-related deaths among 984 placebo-controlled-trial participants, plus 2 deaths in the 1,047-participant open-label addendum and 1 in TRAILBLAZER-ALZ 6; most serious and severe ARIA cases occurred within the first three months of treatment. (PMID 40155270, results)
• With lecanemab, ARIA-E generally occurred within the first 3 (71%) or 6 months (92%) of treatment and generally resolved within 4 months of detection (81% by 120 days), regardless of APOE4 status; multiple sensitivity analyses showed occurrence of ARIA did not adversely impact cognition or function on the primary endpoint models. (PMID 38730496, results)
• For donanemab, 58.3% of first ARIA-E events occurred by the third infusion; 96.0% of 578 ARIA-E episodes resolved radiographically with median resolution 58 days (range 13–350); among 315 rechallenged participants, 68.6% had no ARIA-E recurrence; ~80% of symptomatic cases in the placebo-controlled trials had symptom resolution within the study period. (PMID 40063015, results)
• Serious ARIA concentrated early: 72% of the 25 serious ARIA cases occurred within the first three monthly donanemab infusions, which motivated adding a 4-week MRI to the TRAILBLAZER-ALZ 2 protocol and titration of the first doses. (PMID 40063015, results)

What risk factors (APOE e4 carrier status, anticoagulant use, baseline microhemorrhages) modulate ARIA incidence and severity, and what monitoring/management recommendations exist?

Monitoring and management recommendations center on early, frequent MRI surveillance — safety MRIs before the 2nd, 3rd, 4th and 7th donanemab infusions (a 4-week MRI reduced symptomatic ARIA-E risk by 36.3% in a post hoc analysis), enhanced vigilance during the first ~14 weeks for lecanemab, urgent MRI for any ARIA-suspect symptoms — with severity/symptom-based management (continue, suspend, or permanently discontinue dosing; high-dose glucocorticoids for severe ARIA), avoidance of thrombolytics, and exclusion of patients on anticoagulants.

Status: verified • Confidence: high

• In the donanemab pooled analysis, ARIA-E occurred in 41.7–42.1% of APOE4 homozygotes, 21.5–24.1% of heterozygotes, and 11.0–14.8% of noncarriers; ARIA-H in 50.0–53.6% of homozygotes vs 18.2–18.9% of noncarriers; serious, severe and symptomatic ARIA-E were also highest in homozygotes, and multivariable modeling gave APOE4 homozygotes an OR of 4.57 (95% CI 3.27–6.39) for ARIA-E. (PMID 40063015, results)
• In Clarity AD Core, lecanemab ARIA-E was 32.6% in APOE4 homozygotes, 10.9% in heterozygotes and 5.4% in noncarriers (placebo 3.8%/1.9%/0.3%), and the only baseline risk factors identified by logistic regression for ARIA-E were APOE4 genotype, baseline microhemorrhage, and white matter abnormalities. (PMID 38730496, results)
• Given the strong dose-dependent association between APOE4 allele number and ARIA risk, the FDA issued a boxed warning for donanemab in APOE4 homozygotes; the workgroup recommends APOE genotyping before treatment with risk discussion in shared decision-making, and does not exclude homozygotes but urges caution. (PMID 40155270, discussion)
• LASSO-selected multivariable analysis in 2,031 donanemab-treated participants identified six independent baseline predictors of ARIA-E: APOE4 status (homozygote OR 4.57; heterozygote OR 2.03), 2–4 baseline microhemorrhages (OR 2.53), baseline superficial siderosis (OR 2.18), mean arterial pressure ≥107 mmHg (OR 1.73), amyloid burden ≥108 CL (OR 1.31), and antihypertensive use as protective (OR 0.58); overall ARIA frequencies were similar with and without antithrombotic use, though the study was not powered for infrequent events like macrohemorrhage. (PMID 40063015, results)
• In the lecanemab Core+OLE, antiplatelet and anticoagulant medications did not increase the risk of ARIA-E or ARIA-H, but the rate of intracerebral hemorrhage with lecanemab was 2.7% (4/147) among patients on anticoagulants (0.5% overall vs 0.1% in Core placebo), and the label advises additional caution with anticoagulants or thrombolytic agents. (PMID 38730496, results)
• The donanemab AUR recommend excluding patients on anticoagulants until more safety data exist (anticoagulants should not be stopped just to start donanemab), allowing aspirin/antiplatelet monotherapy, requiring blood pressure control given elevated mean arterial pressure as a modifiable ARIA risk factor, and excluding patients with >4 microhemorrhages or any cortical superficial siderosis at baseline. (PMID 40155270, discussion)
• The donanemab AUR specify MRI monitoring before the 2nd, 3rd, 4th and 7th infusions (plus consideration of a scan before the 12th infusion in high-risk patients), urgent MRI for ARIA-suspect symptoms, a severity-by-symptom management grid (continue for mild asymptomatic ARIA, suspend for symptomatic/moderate ARIA, discontinue for severe ARIA or ≥10 new microhemorrhages/multiple siderosis), early high-dose glucocorticoids for severe ARIA, and a recommendation that intravenous thrombolysis not be given without an MRI excluding ARIA. (PMID 40155270, results)
• A 4-week MRI added to TRAILBLAZER-ALZ 2 reduced the risk of symptomatic ARIA-E by 36.3% (HR 0.64, 95% CI 0.41–0.98) with a statistically nonsignificant 23.9% reduction in serious ARIA; the authors conclude that enhanced clinical vigilance and increased monitoring early in donanemab treatment should be performed. (PMID 40063015, results)
• For lecanemab, the updated safety paper recommends enhanced clinical vigilance during the first 14 weeks per the approved prescribing information, notes that asymptomatic radiographically mild ARIA-E can be dosed through without interruption, and advises additional caution with anticoagulants and thrombolytics given observed intracerebral hemorrhages. (PMID 38730496, discussion)

Do the observed cognitive effects reach the threshold for 'clinical meaningfulness' as judged by patient/caregiver-relevant measures, health-economic modeling, and expert/regulatory appraisal (FDA approval, EMA decision)?

Regulators and payers diverged sharply on clinical meaningfulness: the FDA granted traditional approval to lecanemab (July 2023) and approval to donanemab (July 2024), and the EMA and UK MHRA licensed lecanemab, but NICE rejected NHS reimbursement of both drugs because benefits were too small to justify cost, the Dutch Zorginstituut refused lecanemab reimbursement for absent net benefit, and Germany's IQWIG concluded lecanemab provides no added benefit.

Status: verified • Confidence: high

• The eNeuro authors state that MCID values are 1.0 (MCI) and 1.6 (mild dementia) for CDR-SB and 5 (MCI) and 9 (mild dementia) for iADRS, so average changes of ~3 iADRS points and ~0.5 CDR-SB points 'represent differences imperceptible to patients, their families, and their physicians, no matter what percentage we use to describe the difference.' (PMID 38951040, discussion)
• The Cochrane review concludes that the pooled class effects on CDR-SB (~0.29 points) and ADAS-Cog (~0.85 points) at 18 months are 'considerably lower' than the anchor-based MCIDs (1–2 points CDR-SB; 2–4 points ADAS-Cog) and 'do not seem to be clinically meaningful,' rating effect sizes as trivial to small. (PMID 41985900, discussion)
• Expressed relative to scale ranges, the treatment effects are 2.5% (lecanemab CDR-SB) and 2.3–2.4% (donanemab iADRS), or 8.5–15.6% when normalized to remaining cognitive capacity — framing the authors propose alongside the 27–35% relative-slowing figures to contextualize magnitude. (PMID 38951040, discussion)
• The Cochrane review documents that the FDA, EMA and MHRA approved lecanemab (donanemab approved by FDA in 2024), while NICE issued negative opinions on NHS reimbursement of both lecanemab and donanemab because 'benefits are too small to justify the cost,' the Dutch Zorginstituut refused lecanemab reimbursement based on absence of net benefit, and IQWIG concluded lecanemab offers no benefit to patients. (PMID 41985900, full_text)
• The donanemab appropriate-use paper confirms FDA approval of donanemab in July 2024 for MCI/mild dementia with biomarker-confirmed amyloid pathology, based on TRAILBLAZER-ALZ 2 efficacy data, while explicitly stating that the AUR 'do not provide endorsement for treatment, evaluate meaningfulness related to efficacy or effectiveness, or consider costs and reimbursement.' (PMID 40155270, introduction)

What meta-analyses and independent critiques say about benefit-risk balance, effect sizes in context (e.g., slowing of decline percentages vs absolute CDR-SB changes), and remaining uncertainties (diversity, longer-term duration, dosing/discontinuation)?

Major remaining uncertainties include: only 18–27 months of blinded efficacy data in a chronic progressive disease with no demonstrated long-term divergence; trial populations that were 75–90% white and selected (only ~5–15% of the general early-AD population would meet eligibility); unresolved optimal dosing/discontinuation (donanemab's amyloid-PET-based stopping rule lacks comparative randomized data); and safety external validity gaps since real-world patients are older, more comorbid, and monitored without trial guardrails.

Status: verified • Confidence: high

• Pooled class effects at 18 months: ADAS-Cog SMD −0.11 (trivial; absolute 0.85 points, moderate certainty); CDR-SB SMD −0.12 (absolute 0.29 points, low certainty); ADCS-ADL SMD 0.09; any ARIA-E RR 10.02 (absolute 107 more events/1,000, moderate certainty); symptomatic ARIA-E RR 52.49 (~29 more events/1,000); ARIA-H results heterogeneous across 3 trials (RR 2.31, 1.91, 0.85); SAEs RR 1.04 (high certainty) and mortality RR 1.17 (high certainty) — both statistically nonsignificant. (PMID 41985900, results)
• The review's implications-for-practice conclusion is that, 'given the small effect sizes, with absolute differences well below the MCID, the high risk of ARIA, and the substantial resource burden required for their administration and monitoring, monoclonal antibodies against amyloid-β do not offer a favourable benefit to risk balance for patients with MCI or mild dementia due to AD.' (PMID 41985900, conclusion)
• Because adverse events (infusion reactions) were much more frequent in treated arms, the review judged all studies at 'some concerns' for functional unblinding biasing informant-based efficacy outcomes, and downgraded the certainty of efficacy outcomes by one level accordingly, noting that unblinding could bias CDR-SB ratings toward more positive results. (PMID 41985900, methods)
• The eNeuro perspective argues the 27%/33–35% figures divide a small absolute difference by placebo decline, that this metric implicitly assumes continued linear divergence which is unproven beyond 18 months, that the mAbs' absolute effect is about half of donepezil's, and that NNH is ~3 for all toxic effects for both drugs versus an unestimable NNT — concluding 'caveat emptor' on risk/cost–benefit grounds. (PMID 38951040, discussion)
• The same critique quantifies that in CLARITY-AD 45% of participants had treatment-related adverse events with nearly 1 in 4 developing brain swelling/bleeding (severe bleeding 5 vs 1; three fatal cases across trials), and in TRAILBLAZER-ALZ 2 89% had treatment-related adverse events with over 1 in 3 developing brain swelling/bleeding (severe bleeding 7 vs 2). (PMID 38951040, results)
• The Cochrane review cites analyses quantifying the potential impact of unblinding (Wolters et al.), downgrades efficacy certainty for this risk-of-bias domain, and notes that given the trivial observed differences, any unblinding impact would further attenuate the true effect toward the null. (PMID 41985900, discussion)
• The Cochrane review notes mean trial-participant age of ~69.5–73.9 years versus >80 in real populations, that only 6 of 17 studies followed >18 months, that white participants exceeded 75–90% with marked under-representation of Black and Hispanic participants, that eligibility estimates range 5–15% of the AD population, and that long-term safety data and transparent symptomatic-ARIA reporting are needed. (PMID 41985900, discussion)
• The donanemab AUR highlight that amyloid-PET-based discontinuation (76.4% achieving clearance by 76 weeks, mean accumulation 2.8 CL/year afterward) is biologically rational but lacks head-to-head randomized data on continuing vs stopping, that plasma biomarkers cannot yet guide stopping, that real-world patients are older, more diverse, less educated and more comorbid than trial participants, and that clinicians should enroll treated patients in registries such as ALZNET to fill evidence gaps. (PMID 40155270, discussion)
• The donanemab ARIA analysis cautions that the impact of race/ethnicity on ARIA was limited by underrepresentation of minority populations, that data reflect only 76 weeks with longer-term outcomes in participants who experienced ARIA unknown, and that the open-label addendum suggested numerically lower ARIA rates (32.0% vs 37.0%) when treatment began at lower amyloid burden, supporting earlier initiation as a possible risk-reduction strategy. (PMID 40063015, results)