Abstract
Background
The 2024 Lancet Commission report identified 14 modifiable risk factors associated with dementia prevention that, if addressed, could reduce dementia incidence by as much as 45%. Little attention has been paid to simultaneous examination of all 14 risk factors in relation to Alzheimer's disease (AD) blood‐based biomarkers or mild cognitive impairment (MCI).
Method
We operationalized the 14 risk factors in 936 men in the Vietnam Era Twin Study of Aging at average age 62 when all risk measures were available: education, hearing, vision, depressive symptoms, social engagement, physical exercise, traumatic brain injury (TBI), glucose, low density lipoprotein (LDL), body mass index (BMI), blood pressure, smoking, alcohol consumption, and air pollution (particulate matter 2.5: PM2.5). Outcomes at average age 68 comprised plasma biomarkers [Aβ42/40, pTau231, neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP)] and MCI diagnosis. To address co‐linearity and non‐linearity among predictors, analyses were conducted with Random Forest using grid search and nested cross‐validation, with and without adjustment for age and APOE status (ε4+; ε4‐). Outcomes were examined separately.
Result
Risk factors accounted for 0% (ptau231, AB42/40), 7% (NfL) and 8% (GFAP) of the biomarker variance. The AUC for predicting MCI by age 68 was 0.62. The 3 strongest predictors for NfL were fasting glucose, education, and BMI; for GFAP: BMI, education, and PM2.5. For MCI, predictors in descending order of importance were: blood pressure, PM2.5, education, LDL, BMI, glucose, depressive symptoms, TBI, alcohol consumption, exercise, social engagement, hearing, acuity, age, and smoking. When added to the models, APOE was the least important contributor to plasma NfL, GFAP, and MCI.
Conclusion
Correlated Lancet Commission risk factors when considered together, accounted for relatively little of the variance in plasma NfL, GFAP, or MCI in men across 6 years from late midlife to early old age. The modifiable risk factors contributed more strongly than APOE and/or age. Top predictors for biomarkers and MCI were similar (glucose, BMI, PM2.5, education) illustrating the importance of modifiable health and environmental factors. These results highlight the complex health and lifestyle etiologies of AD‐related plasma biomarkers and may signal areas of opportunity for early intervention.