Rabinovici, Gil D.

Abstract Background Ten years after the original publication, the Centiloid framework is now broadly used to harmonize amyloid‐PET quantification, facilitate data sharing and comparison across cohorts, and even assist visual interpretation in clinical settings. We evaluated the global implementation of Centiloids by comparing their distribution and corresponding positivity thresholds across cohorts. Methods We gathered data from publicly available cohorts and reached out to investigators across the world to collect cross‐sectional Centiloids, demographic and clinical information, and visual read data. Gaussian mixture models (GMM, k=2) were fitted to Centiloid values for each cohort and cutoffs were calculated as mean + 2SD of the lower Gaussian. When visual reads were available, we determined Centiloid cutoffs that maximized correspondence with visual reads (Cohen’s kappa). Data was combined across cohorts using random effects meta‐analyses. Results As of January 2025, we included 37 cohorts ( n = 41,678 participants) with heterogeneous pipelines, radiotracers, and clinical and demographic characteristics (Table‐1). The low Gaussian peaks ranged from ‐9 to 10CL; the meta‐analysis identified a common peak at 1CL. The second peaks were more heterogeneous (range=38‐102CL; meta‐analysis outcome=64CL). Across cohorts, the proportion of cognitively unimpaired versus impaired participants impacted the position of both peaks, with better separation in cohorts enriched in impaired individuals (Figure‐1C). The meta‐analysis indicated a GMM‐based cutoff of 19CL (95%CI: 16‐21CL, Figure‐1B); subgroup analyses showed no evidence of significant effect between single versus multicenter settings (17 versus 20CL, p = 0.30), MRI‐based or PET‐only processing (18 versus 19CL, p = 0.88), and no evidence of difference across radiotracers (Flutemetamol: 16CL; PIB: 17CL, Flutafuranol: 18CL, Florbetaben: 19CL, Florbetapir: 20CL, p = 0.78). In a subset of 29,496 participants with visual reads available, binary visual reads corresponded well to Centiloids (common kappa=0.86, Figure‐2A). The visual read‐based cutoff of 24CL (95%CI: 21‐27CL, Figure‐2B) maximized correspondence between visual read and quantification and was slightly higher than the GMM‐based cutoff. All meta‐analysis models showed high non‐random heterogeneity (I 2 >80%) across studies, suggesting non‐random differences in peaks and cutoffs. Conclusions Meta‐analysis‐based cutoffs align well with thresholds from the existing literature. High heterogeneity among studies underscores the need to investigate contributing factors, raising concerns about applying common cutoffs.

Abstract Background PET visual reads for multi‐center studies are conducted centrally by one or a few experts. In CLARiTI, a broader network of raters will perform the reads, necessitating methods to ensure their accuracy and reliability. Method All ADRC sites ( N = 37) will prospectively acquire amyloid PET scans for 800 cognitively unimpaired and 1200 cognitively impaired participants. Each site can use one of four amyloid PET tracers ([ 18 F]florbetaben, [ 18 F]florbetapir, [ 11 C]PIB, [ 18 F]NAV4694). PET scans will be visually interpreted by ten PET neuroimaging experts located across eight ADRCs, using their preferred hardware, software, and file format (DICOM, NIFTI). Calibration requires readers to perform blinded, independent visual interpretation of 180 amyloid PET scans (30 unique and 15 duplicate scans per tracer), previously read and selected by an expert (Table 1). An even mix of visually non‐elevated and elevated scans were selected, with quantitation spanning ‐26‐123 Centiloids (Figure 1), and 15% were visually borderline (similar proportion to ADNI4), with quantitation spanning 4‐48 Centiloids (mean 21). Kruskal‐Wallis H test showed no significant difference in the distribution of Centiloid values for scans in each tracer group, Χ 2 (3)=2.133, p = 0.545. Result Calibration scans were read by a second expert not serving in CLARiTI. There was 97% concordance and almost perfect agreement between readers (κ=0.932, p <0.001; Figure 2). Discordantly read scans were visually borderline elevated ( N = 2), borderline non‐elevated ( N = 1), and negative ( N = 1), with quantitation of 8‐21 Centiloids (mean 14). Intra‐rater reads were 98% concordant and had almost perfect agreement (κ=0.966, p <0.001). These findings hold promise for scaling‐up visual reads in CLARiTI, and we expect ≥80% concordance and substantial to almost perfect agreement (κ≥0.7) for inter/intra‐rater reliability. Conclusion Calibration is essential for accurate/reliable amyloid PET visual reads in large multi‐center studies with multiple raters. Through the calibration process, we will develop methods to improve the adjudication of borderline/challenging scans and formalize methods for visual reads of amyloid PET tracers that are investigational (i.e., [ 11 C]PIB, [ 18 F]NAV4694). Involving multiple raters across different sites, who use diverse hardware/software/file format combinations, will mirror the natural heterogeneity present in real‐world clinical practice. This calibration set can also serve as a standardized training tool for other studies incorporating amyloid PET visual reads.