Predictors of bi-directional transitions from mild cognitive impairment in a diverse cohort

Trajectories following a diagnosis of mild cognitive impairment (MCI) are varied and may fluctuate over time. Among diverse ethnic and racial groups, social factors, medical comorbidities, and biases in assessment procedures may contribute to greater heterogeneity in the MCI diagnostic category and affect its prognostic significance for dementia. The study goal was to evaluate the frequency and variables associated with MCI transitions among non-Hispanic White (NHW) and Latinx older adults. Multistate Markov models characterized transitions across diagnostic states (cognitively unimpaired (CU), MCI, dementia) over ten years. Variables associated with transitions were assessed using hazard ratios (HR) and 95% confidence intervals (CIs). The study included 413 participants (58% female, mean age 72+8, 52% Latinx ethnicity). Following an MCI diagnosis, the likelihood of converting to dementia versus reverting to CU were equally probable. Older age, NHW ethnicity, APOE ε4, diabetes, lower BMI, and higher neuropsychiatric symptoms associated with elevated risk for dementia conversion, whereas younger age and lower neuropsychiatric symptoms associated with CU reversion. Above other factors, higher baseline serum glial fibrillary acidic protein (HR=1.762 (95% CI: 1.367-2.271)) and neurofilament light (HR=1.467 (95% CI: 1.152-1.69)) associated with increased dementia risk. Trajectories following an MCI diagnosis were highly variable with lower dementia conversion rates among Latinx relative to NHW adults, highlighting the need for strong diverse representation in research to capture the range of exposures shaping risk and resilience for cognitive decline. Well-validated blood-based biomarkers are likely to be instrumental for further improving personalized dementia risk predictions.