Fig. 7
From: Clinical and biological relevance of glial fibrillary acidic protein in Alzheimer’s disease
Classification of amyloid status by blood-based markers. a Correlation between AV45 composite SUVR and blood-based markers. Stepwise logistic regression showed that combining GFAP and ptau-181 together is the optimal model to predict brain amyloid status (amyloid positive versus amyloid negative). The receiver operating characteristic curves were shown for GFAP, ptau-181, and NfL separately or GFAP + ptau-181 (red). The analysis was carried out across all participants ((b)) or among cognitively impaired participants ((c)). The area under the curve (AUC) values and 95% confidence intervals were marked in the figure. Adding NfL as an additional predictor to the optimal model has a marginal improvement of AUC less than 0.01