# Dietary Branched-Chain Amino Acids and Hyper-LDL-Cholesterolemia: A Case–Control Study Using Interpretable Machine-Learning Models in Chinese Children and Adolescents

## TL;DR

This study finds that higher intake of branched-chain amino acids in the diet is linked to increased risk of high LDL cholesterol in Chinese children and adolescents.

## Contribution

The study is the first to use interpretable machine learning and causal mediation analysis to link dietary BCAAs to hyper-LDL-cholesterolemia in children.

## Key findings

- Each 1g/day increase in isoleucine, leucine, and valine intake raises hyper-LDL-cholesterolemia risk by 30%, 11%, and 16%, respectively.
- LDL-C mediates the effect of BCAAs on cardiovascular disease, with a significant indirect effect confirmed via Mendelian randomization.

## Abstract

Background: Plasma branched-chain amino acid (BCAA) concentrations are positively associated with low-density lipoprotein cholesterol (LDL-C) levels. However, the relationship between dietary branched-chain amino acids and hyper-LDL-cholesterolemia is unclear in children and adolescents. Methods: This study explored the correlation between BCAAs and hyper-LDL-cholesterolemia risk through propensity score matching and conditional logistic regression. Machine learning based on LightGBM indicated the important role of BCAAs in the prediction of hyper-LDL-cholesterolemia. To examine the dose–response relationship, Restricted Cubic Splines (RCS) and receiver operating characteristic curves (ROC) were employed. The causal link between BCAA and cardiovascular disease (CVD) was explored via mediation Mendelian randomization. Results: For every 1 g/day increment in the intake of isoleucine, leucine, and valine, there was a corresponding 30%, 11%, and 16% rise in the risk of hyper-LDL-cholesterolemia, respectively. The optimal cut-off values stood at 5.53, 6.40, and 4.18 g/day, respectively. Utilizing the inverse variance weighted method for estimation revealed that the total effect of BCAA on CVD was OR = 1.06 (95% CI: 1.02~1.11), with p = 0.005. The indirect effect, mediated by LDL-C, was OR = 1.02 (95% CI: 1.00~1.02), with p = 0.026. The direct effect was noted at OR = 1.05 (95% CI: 1.01~1.09), with p = 0.017. Conclusions: Dietary BCAAs are positively correlated with hyper-LDL-cholesterolemia in children and adolescents. LDL-C serve as a mediator of CVD caused by BCAAs.