Use of amyloid-PET to determine cutpoints for CSF markers: A multicenter study

To define CSF β-amyloid 1-42 (Aβ42) cutpoints to detect cortical amyloid deposition as assessed by 11C-Pittsburgh compound B ([11C]PiB)-PET and to compare these calculated cutpoints with cutpoints currently used in clinical practice. We included 433 participants (57 controls, 99 with mild cognitive impairment, 195 with Alzheimer disease [AD] dementia, and 82 with non-AD dementia) from 5 European centers. We calculated for each center and for the pooled cohort CSF Aβ42 and Aβ42/tau ratio cutpoints for cortical amyloid deposition based on visual interpretation of [11C]PiB-PET images. Amyloid-PET-based calculated CSF Aβ42 cutpoints ranged from 521 to 616 pg/mL, whereas existing clinical-based cutpoints ranged from 400 to 550 pg/mL. Using the calculated cutpoint from the pooled sample (557 pg/mL), concordance between CSF Aβ42 and amyloid-PET was 84%. Similar concordance was found when using a dichotomized Aβ42/tau ratio. Exploratory analysis showed that participants with a positive amyloid-PET and normal CSF Aβ42 levels had higher CSF tau and phosphorylated tau levels and more often had mild cognitive impairment or AD dementia compared with participants who had negative amyloid-PET and abnormal CSF Aβ42 levels. Amyloid-PET-based CSF Aβ42 cutpoints were higher and tended to reduce intercenter variability compared with clinical-based cutpoints. Discordant participants with normal CSF Aβ42 and a positive amyloid-PET may be more likely to have AD-related amyloid pathology than participants with abnormal CSF Aβ42 and a negative amyloid-PET. This study provides Class II evidence that an amyloid-PET-based CSF Aβ42 cutpoint identifies individuals with amyloid deposition with a sensitivity of 87% and specificity of 80%.