Volatomic profiling of blood identifies environmental chemicals associated with metabolic dysfunction-associated steatotic liver disease in lean US adults.

Abstract

Background: Metabolic dysfunction-associated steatotic liver disease (MASLD) affects lean individuals, but the contribution of environmental exposures remains poorly understood. This study investigated whether blood volatomic profiling of volatile organic compounds (VOCs) identifies environmental chemicals associated with MASLD and advanced fibrosis in lean adults using explainable machine learning.

Methods: We performed a cross-sectional analysis of lean adults from the NHANES 2017-2020 survey with whole-blood measurements of nine VOCs. XGBoost decision trees were used to evaluate associations between VOC profiles, MASLD, and advanced fibrosis. Shapley additive explanations (SHAP) quantified the contribution of individual VOCs and characterized exposure-response patterns.

Results: Among 860 lean adults, 9.0% had MASLD and 1.6% had advanced fibrosis. The MASLD model demonstrated modest discrimination (area under the receiver-operating characteristic curve [AUC-ROC] 0.65), whereas the advanced fibrosis model performed poorly (AUC-ROC 0.24) because of limited events. Chloroform was the strongest VOC associated with both MASLD and advanced fibrosis, followed by o-xylene, bromodichloromethane, benzene, and furan. SHAP analyses demonstrated nonlinear exposure-response relationships and high sensitivity to small variations in low-level VOC concentrations.

Conclusion: Blood volatomic profiling identified distinct VOC signatures associated with MASLD in lean adults, with chloroform as the dominant contributor. Exploratory findings also suggested potential associations with advanced fibrosis that warrant validation in larger studies.