Optimizing cutoffs for clinical interpretation of brain amyloid status using PET/MRI: a multisite study

BACKGROUND: Amyloid-β PET (Aβ-PET) imaging is playing an increasingly important role in the diagnosis and treatment of Alzheimer's Disease (AD). The Centiloid (CL) scale has been developed to standardize the measurements of Aβ PET imaging. Previous CL threshold settings were based on PET/CT, while our team's preliminary study found that CL values from PET/MRI are higher than those from PET/CT. Therefore, there is an urgent need to establish clinical interpretation cutoffs for Aβ status via PET/MRI to facilitate application in clinical practice. METHODS: The clinical performance of Aβ PET/MRI and cerebrospinal fluid biomarkers were evaluated in a multisite cohort of 720 participants. Aβ-PET scans were visually read and quantified using CL method. A two-cutoff approach identified thresholds maintaining > 90% sensitivity/specificity and ≤ 20% intermediate cases, selected by maximal Youden index. Group comparisons of cerebrospinal fluid biomarkers and cognition used generalized linear models adjusted for age and sex. RESULTS: The two-cutoff approach categorized Aβ-PET status as negative (CL ≤ 18.7), gray-zone (18.7-44.2), and positive (CL > 44.2). In Xuanwu cohort, it achieved an accuracy of 94.1%, a positive predictive value of 97.4%, a negative predictive value of 86.1%. In External cohort, the accuracy was 97.4%, positive predictive value 98.5%, negative predictive value 90.8%. Patients in the gray-zone group had lower cerebrospinal fluid(CSF) Aβ-42, Aβ-42/40 levels and a higher P-tau/Aβ-42 ratio. CONCLUSIONS: This study established optimal CL-cutoff, providing a clinically applicable framework for interpreting Aβ PET/MRI findings. Incorporating a two-cutoff system facilitates the tracking of the AD continuum and supports individualized therapeutic decision-making.