Development of environmental risk score of the biomarker for chronic kidney disease in residents exposed to pollutant mixtures

Abstract

Listen

Translate article

BACKGROUND AND AIM: To develop an environmental risk score (ERS) of biomarkers for chronic kidney disease (CKD) in residents exposed to multipollutant. METHOS: Multipollutant with nine metals, four PAH metabolites, and four VOC metabolites measured in urine was constructed from the Forensic Research via Omics Markers (FROM) study in Korea (n=298). Beta-2-microglobulin (β2-MG), N-acetyl-β-D-glucosaminidase (NAG), and estimate glomerular filtration rate (eGFR) were used as CKD markers. Optimal models were selected among elastic net models (ENET), adaptive elastic net models (AENET), weighted quantile sum regression (WQS), Bayesian weighted quantile sum regression (BWQS), Bayesian kernel machine regression (BKMR), Bayesian additive regression tree (BART), and super learner (SL) by comparing mean squared error adjusted for sex and age. The selected models were stratified with a history of occupational chemical exposure (OCE) and additionally adjusted for urinary cotinine level. Variable importance (VI) was estimated to evaluate the associations between each pollutant and biomarkers of CKD. RESULTS: The performance of ERS based on BKMR was the best-fitting model and ERS based on SL or BART performed better than other models. The model with BKMR could not be used due to overfitting as the number of covariates increased. When stratified with a history of OCE, the most effective metabolites for β2-MG, NAG, and eGFR were Sb (VI=0.47), Hg (VI=0.13), and Pb (VI=0.34) in the non-OCE group, and V (VI=0.39), Hg (VI=1.19), and 1-ohp (VI=0.10) in the OCE group, respectively. CONCLUSION: This study suggests that ERS, especially the SL and BART model, stratified with a history of OCE were fitted when evaluating the association between multipollutant and CKD. KEYWORDS: Metals, multipollutant, environmental risk score, machine learning, kidney disease The Korea Environment Industry and Technology Institute through the Core Technology Development Project for Environmental Disease Prevention and Management, funded by the Korea Ministry of Environment (grant number 2021003320003).