Diagnostic Performance of Plasma Biomarkers for A/T/N Image Classification in Alzheimer's disease

Abstract Background We aimed to validate the ability of Alzheimer’s disease plasma biomarkers to predict clinical diagnosis and the ability to predict image‐based A/T/N biomarkers positivity in Asian population. We also sought to define the optimal cut point for each biomarker. Method 232 participants were enrolled, all of whom underwent 18 F‐florbetaben (FBB), 18 F‐flortaucipir (FTP) PET, volumetric MRI, and neuropsychological assessments. Their plasma samples, collected at the time of the imaging scans and stored at ‐80 degrees Celsius, were tested for plasma p‐tau217/np‐tau217 ratio, p‐tau217, Aβ 42/40 ratio, and APS2 score at C2N Diagnostics. FBB and FTP positivity were determined based on the SUVR value through a Gaussian mixture model. The presence of hippocampal atrophy was defined as the mean minus 2 standard deviations of amyloid‐negative cognitive unimpaired participants. ROC analysis was conducted to assess the predictive ability and determine the optimal cut‐off point. Result The ability to predict FBB PET positivity was excellent for p‐tau217/np‐tau217 ratio, p‐tau217, and APS2 (AUC=0.964, 0.963, and 0.958, respectively) while Aβ 42/40 ratio showed lower performance with an AUC of 0.726. The Aβ 42/40 ratio did not have the ability to predict FTP positivity, but p‐tau217/np‐tau217 ratio, p‐tau217, and APS2 all exhibited excellent performance with AUC = 0.926, 0.937, and 0.907, respectively. The Aβ 42/40 ratio also did not have the ability to predict hippocampal atrophy, while p‐tau217/np‐tau217 ratio, p‐tau217, and APS2 demonstrated good performance (AUC=0.809, 0.815, and 0.811, respectively). Similarly, only the p‐tau217/np‐tau217 ratio, p‐tau217, and APS2 had the ability to differentiate clinical diagnosis while Aβ 42/40 ratio failed to distinguish. The AUC‐ROC for p‐tau217/np‐tau217, p‐tau217, and APS2 all showed no significant difference in DeLong’s test when predicting X Y Z. The optimal cutoffs for predicting amyloid, tau and neurodegeneration by p‐tau217/np‐tau217 ratio, p‐tau217 and APS2 showed increases following the progression of AD. Conclusion The plasma biomarker using p‐tau217 reflects the clinical status well on its own and is excellent at predicting the positivity of image‐based biomarkers and diagnostic performance. Among the image‐based A/T/N biomarkers, the predictive power was excellent in the order of A, T, and N. The p‐tau217 alone was not inferior to the p‐tau217/np‐tau217 ratio or APS2