Prediction of Alzheimer's Disease Dementia: Data from the GuidAge Prevention Trial

In therapeutic trials, it is crucial to identify Alzheimer's disease (AD) at its prodromal stage. We assessed the accuracy of the free and cued selective reminding test (FCSRT) compared to other cognitive tests to predict AD dementia in subjects with subjective cognitive decline or mild cognitive impairment. Subjects from the placebo group of the GuidAge trial over 70 years old and without clinical signs of dementia at baseline who completed the 5-year follow-up free of dementia (n = 840) or developed AD dementia (n = 73) were included in our study. Among all the tests, the sum of the 3 free recall of the FCSRT (FCSRT-FR) and the sum of free and cued recall (FCSRT-TR) yielded the best results to predict AD dementia occurrence (all p values < 0.05 for comparison of FCSRT-FR ROC and MMSE, CDRsb, and CVF ROCs). FCSRT-FR had an area under the ROC curve of 0.799 (95% CI 0.738-0.85) and the optimal cut-off was 20 (se 68.06%, sp 81.43%, PPV 23.90%, NPV 96,75%). Concerning FCSRT-TR, the AUC was 0.776 and the optimal cut-off was 42 (se 62.5%, sp 82.26%, PPV 23.20% and NPV 96.24%). This study sets the framework for implementing the FCSRT in clinical and therapeutic trials for efficient subject selection.