Assessment of genetic and metabolite associations of branched chain amino acids with metabolic disease in the UK Biobank using Mendelian randomization.

Background: As the building blocks of proteins and precursors of many other important compounds, amino acids play a vital role in the biochemical processes needed to sustain life. The branched-chain amino acids (BCAAs) are unique in their structure and function, as they are metabolized in muscle tissue and play important roles in protein synthesis and energy production. However, despite their physiological importance, relatively little integrative research has been conducted into the direct relationships between this class of metabolites and their effect on risk for metabolic diseases.

Methods: Utilizing an integrative PheWAS approach using UK Biobank data, we were able to identify strong, high confidence, metabolite-disease correlations for the three BCAAs: leucine, isoleucine, and valine. Relationships were established through comparison of metabolite level-disease prevalence associations with polygenic scores for BCAAs, followed by Mendelian randomization analysis.

Results: All BCAAs studied demonstrated especially strong relationships with type II diabetes, and robust relationships with obesity, hypertension, sleep apnea, and chronic kidney disease. We illustrate this with a set of metabolite prevalence-disease risk plots that suggest differing potential for disease based on varying levels of branched-chain amino acid metabolites. Similar results are observed with polygenic scores for plasma BCAAs. Mendelian randomization shows positive effects of leucine and isoleucine on hypertension, and either reverse causality or no clear directional relationship for other associations, notably effects of obesity and type II diabetes on all three BCAAs, with limited or borderline evidence for other outcomes.

Conclusions: Overall, the results of our study highlight a relatively unexplored area of metabolite-disease associations and provide a blueprint for uncovering additional relationships using readily available biobank data.