Neurology · Medical Literature
Donanemab treatment effect by baseline tau burden and disease severity: Observations from the TRAILBLAZER-ALZ 2 trial.
Abstract
Clinical trials indicate that disease-modifying therapies can slow clinical decline in Alzheimer's disease (AD), with earlier initiation associated with greater slowing. In the TRAILBLAZER-ALZ 2 trial, the treatment effect of donanemab on the Clinical Dementia Rating Scale Sum of Boxes (CDR-SB) score was assessed across disease stages defined by baseline tau PET, plasma P-tau217 levels, or predicted disease progression. Donanemab-mediated slowing of disease progression occurred across baseline tau PET and plasma P-tau217 levels. Participants with lower baseline tau PET and P-tau217 showed greater slowing with donanemab versus placebo. Modeling CDR-SB scores indicated that earlier treatment (at the 25th percentile of baseline Predicted disease progression) delayed disease progression by 60% over 76 weeks, compared to 33% and 17% at the 50th and 75th percentiles. Donanemab benefited participants with early symptomatic AD across clinical and pathological severities, with the greatest slowing in those treated earlier. gov Identifier: NCT04437511.
Clinical relevance
In early symptomatic Alzheimer's disease, does the magnitude of donanemab's treatment effect on CDR-SB-measured disease progression vary as a function of baseline tau burden (tau PET, plasma P-tau217) and predicted disease progression stage? __SNAPSHOT__ {"studyDesign":"Secondary analysis (prespecified and post hoc) of a Phase 3 randomized, double-blind, placebo-controlled trial (TRAILBLAZER-ALZ 2; NCT04437511), using Bayesian disease progression modeling of CDR-SB scores stratified by continuous and categorical baseline biomarker indices","setting":"Multicenter; 277 sites across 8 countries (international)","sampleSize":"n=1,736 total randomized (donanemab: n=860, placebo: n=876); 1,182 with low/medium tau pathology and 552 with high tau pathology","population":"Adults aged 60–85 years with early symptomatic AD (MCI or mild AD dementia), MMSE 20–28, confirmed amyloid pathology (≥37 Centiloids on 18F-florbetapir or 18F-florbetaben PET), and confirmed tau pathology on 18F-flortaucipir PET; stratified at enrollment by low/medium vs. high tau burden","intervention":"Donanemab 700 mg IV every 4 weeks for the first 3 doses, then 1,400 mg IV every 4 weeks for up to 72 weeks; participants were switched to blinded placebo upon meeting prespecified amyloid clearance criteria","comparator":"Matched placebo IV infusion every 4 weeks","primaryOutcome":"For this secondary analysis: change from baseline in CDR-SB score over 76 weeks, assessed across subgroups defined by baseline tau PET SUVR, plasma P-tau217 levels, and a model-derived 'Predicted disease progression' (PDP) score representing continuous disease stage","followUp":"76 weeks (18 months)","fundingSource":"Industry (Eli Lilly and Company)"}
Key findings
Across the full enrolled cohort (n=1,736), donanemab-mediated slowing of CDR-SB progression was observed across the continuum of baseline tau PET and plasma P-tau217 levels, with no subgroup demonstrating absence of benefit. Participants with lower baseline tau PET and lower plasma P-tau217 demonstrated greater treatment effect magnitude versus placebo compared with those with higher tau burden. The CDR-SB between-group difference in the low/medium tau primary analysis population was –0.67 (95% CI: –0.95 to –0.40), representing 36.0% slowing of progression; in the combined population (all tau strata), CDR-SB and iADRS showed 29% and 22% slowing of decline respectively (p<0.001 for all). Using a Bayesian disease progression model applied to continuous baseline PDP scores, modeling indicated that treatment initiated at the 25th percentile of baseline PDP (earliest disease stage represented) delayed progression by 60% over 76 weeks, compared with 33% at the 50th percentile and 17% at the 75th percentile (absolute effect sizes and confidence intervals for the PDP-modeled estimates are not reported in the abstract). Plasma P-tau217 tracked tau PET–defined disease stage closely, with similar associations to treatment response across both biomarkers. Amyloid-related imaging abnormalities (ARIA) represent the primary adverse event class: ARIA-E (edema/effusions) occurred in 24.4% of donanemab-treated vs. 1.9% of placebo-treated participants in the placebo-controlled trials; ARIA-H (microhemorrhages/hemosiderin deposition) occurred in 31.3% vs. 13.0%, respectively. Serious ARIA-E was reported in 1.5% and symptomatic ARIA-E in 5.8% of donanemab-treated participants. ARIA risk was independently associated with APOE ε4 allele number, baseline microhemorrhage burden, cortical superficial siderosis, higher amyloid load, and elevated mean arterial pressure. NNT and NNH are not reported in the abstract and cannot be calculated from available data.
Practice implications
This secondary analysis of TRAILBLAZER-ALZ 2 establishes that donanemab's effect on CDR-SB is present across the full spectrum of tau pathology represented in the trial, but that the magnitude of benefit is meaningfully greater in participants treated at earlier pathological and clinical stages. The 60% modeled slowing at the lowest disease burden quartile versus 17% at the highest quartile provides a compelling biological rationale for earlier identification and treatment initiation. However, several important caveats apply: (1) Only an abstract is available for the 2026 sub-analysis paper (PMID 42273802); full statistical details, confidence intervals for the PDP-modeled estimates, and adjustment methods cannot be independently verified. (2) The PDP-based progression estimates derive from Bayesian disease progression modeling—a sophisticated but indirect approach—rather than from prespecified subgroup randomization; subgroup treatment-by-biomarker interaction testing results (formal tests of effect modification) are not reported in the abstract. (3) The parent trial enrolled only amyloid-and-tau PET–confirmed early symptomatic AD aged 60–85 with MMSE 20–28; findings cannot be extrapolated to asymptomatic/preclinical AD, moderate-to-severe dementia, younger patients, or populations without biomarker confirmation. (4) The trial required tau PET confirmation for enrollment, meaning the patient population was more narrowly defined than typical clinical practice, limiting generalizability to settings where tau PET is unavailable. (5) The observation that plasma P-tau217 tracks tau PET response associations is clinically significant and supports potential use of this more accessible biomarker for treatment-stage stratification, though this requires prospective validation. (6) The ARIA rate (24.4% ARIA-E, 31.3% ARIA-H, 5.8% symptomatic ARIA-E) is clinically meaningful and higher than observed with lecanemab, requiring structured MRI surveillance programs in any real-world implementation. (7) The trial was industry-funded, introducing potential for optimistic framing of subgroup analyses. __LIMITATIONS__ Only abstract available for the June 2026 sub-analysis; full statistical reporting (interaction p-values, confidence intervals for PDP-modeled estimates, covariate adjustment details) cannot be assessed.,PDP subgroup analyses are post hoc model-derived, not based on prespecified randomized subgroup allocation; formal treatment-by-biomarker interaction tests are not reported, limiting causal inference.,High tau participants (n=552) were enrolled to gather additional data but the trial was not powered for this subgroup; their individual-subgroup efficacy estimates carry substantial uncertainty.,Strict amyloid PET and tau PET inclusion criteria limit generalizability to community practice settings without PET infrastructure.,76-week follow-up is insufficient to characterize long-term durability, optimal retreatment thresholds after amyloid reaccumulation, or mortality outcomes.,Limited ethnic and racial diversity typical of large AD trials may restrict generalizability across populations.,The CDR-SB between-group differences (e.g., –0.67 points), while statistically significant, remain subject to ongoing debate regarding minimum clinically important differences on this scale.,Industry sponsorship (Eli Lilly) across all analyses introduces risk of selective reporting and optimistic framing of subgroup data.,ARIA rates (ARIA-E 24.4%, ARIA-H 31.3% in treated arm) represent a clinically significant safety burden that may constrain real-world uptake, particularly in APOE ε4 homozygotes. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ This secondary analysis of TRAILBLAZER-ALZ 2 demonstrates that donanemab provides CDR-SB benefit across the full baseline tau spectrum in early symptomatic AD, with markedly greater slowing (modeled 60% vs. 17% over 76 weeks) in patients treated at earlier pathological stages. These findings reinforce the clinical imperative to identify and treat eligible patients as early as possible using biomarker-confirmed staging, and support plasma P-tau217 as a practical surrogate for tau PET staging. Given the ARIA burden, industry sponsorship, and the post hoc modeling nature of the stratified estimates, these results should be considered practice-informing rather than independently practice-changing, pending prospective validation of biomarker-guided treatment initiation strategies.