Premium
Continuous and risk‐score‐based predictors of ATN Alzheimer's disease status among cognitively healthy individuals: Findings from the EPAD‐LCS study
Author(s) -
Calvin Catherine M,
de Boer Casper,
Raymont Vanessa,
Gallacher John,
Koychev Ivan G
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.041097
Subject(s) - medicine , dementia , framingham risk score , body mass index , family history , cohort , logistic regression , disease , alzheimer's disease , pathology
Background The Amyloid/Tau/Neurodegeneration (ATN) framework has been proposed as means of evidencing the biological state of Alzheimer’s disease (AD). Predicting ATN status in pre‐dementia individuals therefore provides an important opportunity for targeted recruitment into AD interventional studies. We investigated the extent to which ATN phenotype can be predicted by known AD risk factors as well as validated risk scores. Method We used baseline data from the European Prevention of Alzheimer’s Disease Longitudinal Cohort Study (v1500.0). 1181 participants with complete data (mean age 65±7 years, 57% female) were categorised into four ATN subgroups: no pathology, AD pathologic change, AD pathology, and, non‐AD pathology. Multinomial logistic regression tested composite risk scores (CAIDE, Framingham Vascular and Stroke scores) and the following individual risk exposures for association with ATN subgroups: age, sex, education, APOE e4, family history of dementia, systolic blood pressure, body mass index (BMI), high cholesterol, physical inactivity, smoking, hypertensive medication, prior cardiovascular disease, atrial fibrillation, and white matter lesion volume. ROC curve models were used to estimate the added value of predicting each pathology group according to significant risk factors and composite scores. Result 53.4% of the sample showed no brain pathology, 24.0% showed AD pathologic change, 14.1% showed non‐AD pathology, and, 8.5% had biological AD. In age and sex‐adjusted multinomial logistic regression there were significant group effects for age, APOE e4, family history of dementia, BMI, and white matter lesion volume. In ROC models, discriminative accuracy (AUC) in predicting AD pathologic change and AD pathology ( vs no pathology), was improved by 4% and 3% respectively after adding family history, BMI and white matter lesion volume to the most basic model (age, sex and APOE e 4). Risk composite scores did not add to the models. Conclusion Low BMI and high white matter lesion volume but not established risk scores associated with AD pathology; white matter lesions were also greater in the AD pathologic change group. Specific vascular risk factors, particularly white matter lesion volume, may be of additional value to ATN profiling (i.e. ATN(V) for biological AD in a pre‐dementia stage.