Association between Dried Fruit Intake and DNA Methylation: A Multivariable Mendelian Randomization Analysis

Objectives: Observational studies have reported associations between dried fruit intake and DNA methylation(DNAm). However, inherent flaws in observational study designs make them susceptible to confounding and reverse causality bias. Consequently, it is unclear whether a causal association exists. In the present study, we aimed to investigate the causal associations between dried fruit intake and DNAm.

Methods: We performed two-sample Mendelian randomization (MR) using the IEU Open GWAS database aggregated data. Forty-three single nucleotide polymorphisms (SNPs) associated with dried fruit intake as instrumental variables (IVs) were selected as exposure. DNAm outcomes include Gran (estimated granulocyte proportions); AgeAccelGrim(GrimAge acceleration); Hannum (Hannum age acceleration); IEAA(Intrinsic epigenetic age acceleration), AgeAccelPheno( PhenoAge acceleration), and DNAmPAIadjAge (DNAm-estimated plasminogen activator inhibitor-1 levels). We used the MR pleiotropy residual sum and outlier test (MRPRESSO) and Radial-MR test to identify any level of multi-effect outliers and assessed the causal effect estimates(after removing outliers). The primary causal effects were estimated using inverse-variance weighted (IVW) method and undertook sensitivity analyses using MR methods robust to horizontal pleiotropy.The direct effects of dried fruit intake on DNAm were estimated using multivariable mendelian randomization (MVMR).

Results: Leveraging two-sample MR analysis, we observed statistically significant associations between dried fruit intake with a lower AgeAccelGrim(β=-1.365, 95% confidence intervals [CI] -2.266 to -0.464, PIVW=2.985×10-3) and AgeAccelPheno (β= -1.933, 95% CI -3.068 to -0.798, PIVW=8.371×10-4). By contrast, the effects level on Gran (β=0.008, PIVW=0.430), Hannum(β=-0.430, PIVW=0.357), IEAA(β=-0.184, PIVW=0.700), and DNAmPAIadjAge (β=-1.861, PIVW=0.093) were not statistically significant. MVMR results adjusting for the potential effects of confounders showed that the causal relationship between dried fruit intake and AgeAccelGrim(β= -1.315, 95% CI -2.373 to -0.258, PIVW=1.480×10-2) and AgeAccelPheno(β= -1.595, 95% CI -2.987 to -0.202, PIVW=2.483×10-2) persisted. No significant horizontal polymorphism was found in the sensitivity analysis.

Conclusion: Our MR study suggested that increased dried fruit intake is associated with slower AgeAccelGrim and AgeAccelPheno. It can providing a promising avenue for exploring the beneficial effects of dried fruit intake on lifespan extension.