An Integrated Model Based on Gut Microbiota and APOE Genotype for Predicting Dementia Risk
Sehee Lee, Sun Hwa Hong, You Jin Nam, Yong Hyuk Cho, Sang Joon Son, Chang Hyung HongBackground: Dementia develops through the combined influence of genetic vulnerability, biological processes, and environmental exposures. The apolipoprotein E (APOE) ε4 allele is a well-known genetic contributor to dementia risk, and growing evidence links gut microbial alterations to cognitive decline and cerebrovascular-related pathology. Nevertheless, studies jointly evaluating genetic, microbiome, and clinical information remain relatively scarce. This study examined an integrated framework combining APOE genotype and gut microbiome data for cross-sectional dementia classification. Methods: We analyzed 292 participants representing three cognitive stages: subjective memory impairment (SMI), mild cognitive impairment, and dementia. Clinical variables, APOE genotype, and gut microbial metagenomic profiles were examined. Associations among genetic risk, Alzheimer’s disease pathology, and brain structural changes were assessed, and multivariable models were used to distinguish participants with dementia from those with SMI or MCI. Results: APOE ε4 carriage was most frequent among participants with dementia, while no ε4 carriers were observed in the SMI group. Gut microbial profiles differed according to the dementia-related genetic-risk category (mild vs. moderate-to-high). The fully integrated model showed a numerically higher cross-validated AUC than models constructed from fewer data domains. Streptococcus, Akkermansia, and Fusicatenibacter were more abundant in the moderate-to-high genetic-risk group; these taxon-level findings were exploratory and based on nominal p-values. Conclusions: The findings support an exploratory integrated framework for cross-sectional dementia classification based on genetic and gut microbiome information. Independent longitudinal and multicenter validation is required before the framework can be interpreted as predicting future dementia risk or supporting personalized clinical decisions.