This individual participant data (IPD) meta-analysis pooled amyloid PET scans from 49,227 participants across 53 studies (15 countries) to derive robust Centiloid positivity cutoffs using Gaussian mixture models (GMMs) and correspondence with binary visual reads.
The data-driven GMM approach yielded a single positivity cutoff of **18 Centiloids** (95% CI, 16–19). A double-cutoff approach set high-confidence negativity below **11 CL** (95% CI, 9–13) and high-confidence positivity above **26 CL** (95% CI, 24–28). Scans matched visual reads best at **27 Centiloids** (Cohen κ = 0.86; 95% CI, 0.83–0.89). Scans in the **11–26 CL gray zone** warrant caution.
- Very high heterogeneity across studies (I² = 95–97%), limiting precision of pooled cutoffs. - Cross-sectional design; longitudinal amyloid accumulation trajectories are not captured. - The gray zone (11–26 CL) lacks clear clinical guidance, and optimal management in this range remains undefined.
For amyloid PET interpretation, use **<11 Centiloids** as high-confidence negative and **>26 Centiloids** as high-confidence positive; scans in the 11–26 CL range should prompt clinical judgment rather than automatic classification. The 27 CL cutoff best aligns with expert visual reads and may be the most pragmatic threshold when making eligibility decisions for amyloid-targeting therapies.
Explore related topics
