Independent contribution of ATN biomarkers for progression of amnestic mild cognitive impairment to AD on top of neuropsychological tests

Abstract Background ATN biomarkers (amyloid beta [A], pathologic tau [T], and neurodegeneration [N]) have been known for their predictive value of Alzheimer’s disease (AD). Here, we aimed to investigate their independent contributions to predicting progression of amnestic mild cognitive impairment (aMCI) to AD on top of neuropsychological tests. Method We included 496 aMCI patients with neuropsychological tests and ATN biomarkers from cerebrospinal fluid (CSF) examinations, MRI and FDG‐PET at baseline. The involved ATN biomarkers included CSF‐Aβ, CSF‐Ptau, CSF‐Tau, FDG‐PET and an MRI‐based Alzheimer’s disease resemblance atrophy index (AD‐RAI) calculated by AccuBrain IV2.0. All the demographic and neuropsychological features (Table 1) were entered into the cox model with backward elimination to predict time to AD (subscores of FAQ and ADAS‐cog also entered as candidate predictors) using training cohort (n = 330, 34.8% developed AD within 3.60±2.46 years). The resulting model was treated as the basic model. The enriched models with individual and multiple ATN biomarkers were further compared with the basic model regarding the improvement in C‐index (measuring prediction performance of time to AD) and time‐dependent AUC (t‐AUC, measuring prediction performance of time‐specific progression) in validation cohort (n = 166, 35.5% developed AD within 3.66±2.52 years). Result The features that survived backward elimination for the basic model were shown in Table 2. Binary form of A/T/N variables were preferred for Aβ/Tau, while continuous form preferred for Ptau/FDG‐PET/AD‐RAI (Table 3). Compared with the basic model, all the enriched models with individual A/T/N biomarkers (except for Basic+Tau) presented higher C‐index in the training cohort (p<0.05), while only Basic+(AD‐RAI) and Basic+(FDG‐PET) presented higher C‐index in the validation cohort (Table 4). Basic+(FDG‐PET) presented highest t‐AUC (0.861∼0.924) among the models with individual A/T/N biomarkers (Figure 1A). The enriched model with multiple ATN biomarkers (Aβ/Ptau/FDG‐PET/AD‐RAI) presented highest C‐index in both training and validation cohort, which also showed higher t‐AUCs (generally ≥0.90 through year 3∼7) than the models with individual A/T/N biomarkers in trend (Figure 1B). These findings indicated that multiple ATN biomarkers should be used for the optimal model (Figure 2). Conclusion ATN biomarkers, when combined in use, presented most contribution to predicting aMCI‐to‐AD progression independent of demographic characteristics and neuropsychological tests.