Longitudinal Blood DNA Methylation Changes During Weight-Loss Intervention and Dementia Progression Risk

Weight loss improves metabolic health, but the DNA methylation (DNAm) changes induced by the lifestyle intervention and their relevance to dementia outcomes remain unclear. We studied longitudinal blood DNAm changes during an 18-month weight-loss intervention in the CENTRAL clinical trial and evaluated their relevance to dementia progression in an external cohort.

We analyzed paired baseline and 18-month blood DNAm data from 103 male CENTRAL participants, including 47 in diet-only and 56 in diet plus physical activity groups. CpG-level methylation M-value changes between baseline and follow up were tested using linear mixed-effects models adjusted for dietary group, age, smoking score, leukocyte proportions, and random participant effects. Differentially methylated regions were identified using comb-p, and pathway enrichment was assessed using methylGSA software. Weight-loss-associated CpGs and regions were compared with cardiometabolic and dementia epigenome-wide association studies (EWAS) findings. In 117 Alzheimer's Disease Neuroimaging Initiative participants with repeated DNAm and diagnostic follow-up, we tested whether DNAm changes at baseline resembling the reverse of the CENTRAL weight-loss profile predicted dementia progression.

At a 5% false discovery rate, 51 CpGs and three differentially methylated regions (DMRs) were identified in the diet-only analysis, and 49 CpGs and one DMR were identified in the diet plus physical activity analysis, that were significantly associated with the weight loss interventions. Enriched pathways included DNA double-strand break response and ATM-mediated phosphorylation in the diet-only analysis, as well as energy metabolism and insulin secretion in the diet plus physical activity analysis. Weight-loss-associated DNAm signals overlapped with cardiometabolic, inflammatory, and dementia-related EWAS findings in expected directions. In ADNI, 28 participants progressed clinically. Between-visit DNAm changes at baseline in converters showed higher correlation scores with the reversed diet-only DNAm profile than non-converters (mean correlation score, 0.131 vs. 0.037; P-value = 0.026), and higher correlation scores were associated with increased progression risk in a Cox model adjusted for age, sex, APOE ε4, baseline diagnosis, education, and smoking history (hazard ratio = 1.49 per standard deviation, P-value = 0.041). The diet plus physical activity profile showed a similar but weaker association.

Weight-loss intervention was associated with blood DNAm changes enriched in genomic maintenance and metabolic pathways. External EWAS comparisons and ADNI validation suggest that weight-loss-responsive DNAm profiles may capture biological processes connecting lifestyle-related metabolic change with dementia progression risk.