Predicting diagnostic conversion from mild cognitive impairment to Alzheimer's disease: A Bayesian hierarchical model approach using ADNI patient data

BackgroundThere is still need for a better understanding of which specific follow-up medical assessments might offer greater predictive value for diagnostic conversion from mild cognitive impairment (MCI) to Alzheimer's disease (AD).ObjectiveTo examine the longitudinal predictive importance of follow-up medical assessments to detect diagnostic conversion from MCI to AD.MethodsA sample of 572 participants from the ADNI database with valid data at baseline medical visit were included. Bayesian hierarchical models were employed to investigate longitudinal predictors of diagnostic conversion in a 36-month medical follow-up cohort, for measures of cognitive function, psychopathological symptoms, and demographical data. An additional 48-month medical follow-up cohort was considered to investigate the predictive importance of cerebrospinal fluid biomarkers (Aβ₄₂/Aβ₄₀ ratio) for diagnostic conversion.ResultsMini-mental State Examination (MMSE) (β = -2.6; 95% HDI: [-3.6--1.5]) and Clinical Dementia Rating scale Sum of Boxes (CDR-SB) (β = 5.6; 95% HDI: [4.3-7.0]) can predict diagnostic conversion from MCI to AD over a 36-month medical follow-up, with CDR-SB showing the greatest predictive importance in all Bayesian models. Higher scores on CDR-SB were associated with increased risk for a diagnosis conversion, approximately 30% greater probability at 24-month follow-up, and > 50% greater probability at 36-month follow-up.ConclusionsThe CDR-SB provides a reliable cognitive assessment to detect diagnostic conversion from MCI to AD over a period of 36 months, which is key to help clinicians screening for early diagnosis of AD using affordable non-invasive procedures.