What Centiloid changes after anti‑amyloid therapy have been observed in trials, and how do they map to clinical benefit?
Executive summary
Clinical trials of anti‑amyloid monoclonal antibodies have repeatedly shown large Centiloid (CL) reductions—often taking mean group values from high baseline CL into ranges below the conventional positivity threshold (~30 CL)—and trial-level analyses link deeper, faster amyloid lowering (particularly to roughly 15–25 CL or below) with measurable but modest slowing of clinical decline (for example, lecanemab’s amyloid substudy mean ~23 CL at 18 months and an associated 0.45 CDR‑SB benefit) [1] [2]. Experts and consortium guidance converge on pragmatic cut‑points (CL <10 likely negative; CL >30 conservative positive) and propose treatment initiation and monitoring thresholds in the 24–30 CL neighborhood, while also warning of measurement uncertainty, heterogeneity across drugs, and unresolved causal chains to clinical outcomes [3] [4] [5].
1. What Centiloid shifts have been observed in pivotal trials and cohorts
Phase‑2/3 programs have documented substantial absolute Centiloid reductions: for example, the lecanemab amyloid PET substudy reported a mean amyloid level of 22.99 CL at 18 months in the treated group—below the ~30 CL positivity threshold—while other programs (aducanumab, donanemab, gantenerumab, gantenerumab variants) produced varying time‑courses and magnitudes of CL decline summarized across trials [1] [4]. Sponsors and early phase studies report rapid clearance in some newer agents—Roche’s trontinemab reduced most participants below a 24 CL threshold within 28 weeks in a phase Ib/IIa cohort and many to very low CLs (<11) [6]—and FDA labeling notes Centiloid rises after treatment cessation at ≈2.6 CL/year on average, underscoring that reductions are time‑ and dose‑dependent [7].
2. Thresholds experts and consortia use to interpret Centiloid
International working groups and the AMYPAD consortium recommend pragmatic thresholds for clinical context: CL <10 reliably excludes Aβ pathology, CL >30 is a conservative positivity cut‑off, and a “gray zone” lies between—experts have suggested a 24–30 CL band as an implementable point at which to consider anti‑amyloid therapy in early symptomatic patients (MCI/mild dementia) [3] [4] [5]. These consensus thresholds inform both entry criteria in trials and the interpretation of post‑treatment PET, though authors caution about measurement uncertainty (typically 3–8.3 CL across the scale) and pipeline differences [8].
3. How Centiloid change maps to clinical benefit in trials
Multiple analyses converge: trials that achieved greater and faster amyloid lowering—often bringing mean treated-arm Centiloids below ~25 CL—showed statistically significant but clinically modest slowing on endpoints such as CDR‑SB (for example, lecanemab’s 0.45‑point CDR‑SB difference at 18 months, roughly a 27% slowing) and meta‑analytic work suggests a moderate correlation between Centiloid change and end‑of‑study CDR‑SB differences [2] [9] [10]. Conversely, trials that failed to reduce CL sufficiently (or more slowly) showed null clinical outcomes, illustrating an apparent dose–response: “sufficient” reduction—often operationalized near or below 15–25 CL—appears associated with detectable clinical slowing [9] [11] [2].
4. Important caveats and biological context
The relationship is probabilistic and mediated by biology: evidence links Aβ levels above ~25 CL with downstream tau accumulation and clinical impairment, suggesting an interplay where late or insufficient Aβ removal may not arrest tau‑driven progression [2]. Measurement limitations (inter‑pipeline variability, 3–8 CL uncertainty) and imaging vs pathology mismatches mean individual CL values must be interpreted cautiously [8]. Imaging effects such as amyloid‑removal‑related pseudo‑atrophy on MRI after antibody therapy add another confounder to interpreting structural outcomes alongside CL change [12].
5. Practical implications and remaining uncertainties
Operationally, trials and emerging clinical pathways use Centiloid both to select patients (often using 10–30 CL cut‑offs) and to monitor treatment response; achieving end‑of‑trial CLs below ~25 (and ideally below the conservative 30‑CL positivity line) is associated with the trials that reported benefit, with earlier and larger reductions predicting stronger outcomes [4] [5] [9]. However, causality is not perfectly settled: the magnitude of clinical benefit remains modest at group level, individual responses vary, risk (ARIA) and durability (CL rebound ≈2.6 CL/year off therapy) are concerns, and further work is required to define patient‑level predictive thresholds and optimal treatment durations [7] [2] [8].