Centiloid cut-offs for amyloid PET: a grey zone between 11 and 26

Design

Individual participant data meta-analysis of 53 studies

Population

49,227 amyloid PET scans with 5 tracers, 15 countries; mean age 71

Primary outcome

Centiloid positivity cut-offs, data-driven and visual-read based

Effect

Single cut-off 18 CL (95% CI 16 to 19); negative <11, positive >26; visual-read cut-off 27, kappa 0.86

An individual participant data meta-analysis pooled Centiloid values from 49,227 amyloid PET scans in 53 studies from 15 countries, using five tracers. Mean age was 71 and 62% were cognitively impaired. Published in JAMA on 6 October 2026.

A data-driven approach found a single positivity cut-off of 18 Centiloids (95% CI 16 to 19). A double-cutoff approach found scans could be confidently called negative below 11 (9 to 13) and confidently positive above 26 (24 to 28). Against visual reads in 35,045 scans, Centiloids predicted visual positivity well (kappa 0.86), with the best-matching cut-off at 27.

Heterogeneity between studies was high, reflecting different tracers, scanners and populations, so local calibration still matters. But the work gives reporting departments a defensible framework: a quantitative value alongside the visual read, with explicit caution for values between 11 and 26, which is where decisions about amyloid-targeting treatment are hardest. The framework applies wherever amyloid PET is read, including centres not yet offering those treatments.

• Report a Centiloid value alongside the visual read where software allows
• Treat values below about 11 as confidently negative and above about 26 as confidently positive
• Flag values between 11 and 26 as indeterminate and correlate with clinical findings
• Validate thresholds locally for the tracer and scanner in use
• Discuss indeterminate scans with the referring neurologist before therapy decisions

Why it matters

Eligibility for amyloid-targeting treatment can hinge on a number near a threshold that varied between centres.

Don't overread it

Heterogeneity was very high; these pooled cut-offs need local validation and do not by themselves define treatment eligibility.

The statistics, in plain English

A kappa of 0.86 means Centiloid values and visual reads agreed very well beyond chance. An I² of 95% or more means the individual studies differed a lot, so the pooled cut-off is an average across varied settings.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.