Salt Added to Food and Body Mass Index: A Bidirectional Mendelian Randomization Study

Background and objective: Observational studies suggest that dietary sodium (salt) intake may be associated with body mass index (BMI). However, these findings may be biased by confounding and reverse causality. The present study aimed to apply a bidirectional Mendelian Randomization (MR) framework to determine the causal association between salt added to food (do not include salt used in cooking) and BMI by integrating summary-level genome-wide association study (GWAS) data. Methods: We performed two-sample MR analyses using summary statistics of GWAS. Inverse-variance weighted (IVW) method was used to analyze the effect of the preference of salt added to food on BMI. We used maximum likelihood estimation and random effect model as auxiliary verification. A bidirectional MR analysis with BMI as the exposure and salt added to food as the outcome was also performed. Results: We identified 74 single nucleotide polymorphisms (SNPs) that were genome-wide significant (P < 5x10^-8) for the preference of salt added to food in the UK Biobank (n = 462,630) and were investigated for their association with BMI in a meta-analysis of 322,154 European-descent individuals from GWAS and Metabochip studies. The IVW method estimate indicated that the preference of salt added to food was positively associated with BMI ({beta} = 0.1416, SE = 0.0576, P = 0.0139). Results from maximum likelihood estimation ({beta} = 0.1476, SE = 0.0363, P < 0.0001) and the random effect model ({beta} = 0.1411, SE = 0.0572, P = 0.0137) were consistent with the IVW. Bidirectional MR analyses suggested that BMI was not associated with the preference of salt added to food. Conclusion: Our results provided qualitative evidence supporting a causal relationship between salt intake and BMI.