skip navigation bar
go content
go menu


Representative Research Publications

Metabolic Signatures of Humidifier Disinfectant-Associated Lung Injury 2025 > Representative Research Publications > Research Results Home

Metabolic Signatures of Humidifier Disinfectant-Associated Lung Injury

  • Environ. Health Perspect. / 2025. 5.
  • Jinwoo Kim (First author), Myung Hee Nam, Soo-Jong Hong (Corresponding author)

Research Summary

The South Korean humidifier disinfectant–associated lung injury (HDLI) case was one of the worst disasters involving household chemical products, resulting in over 5,800 casualties. Despite the strong association between lung injury and humidifier disinfectants, the underlying pathogenic mechanisms remain unclear.
We investigated pediatric patients wiith HDLI to identify key metabolic signatures. Using untargetted metabolomics, we observed significantly higher levels of oxidized lipids in comparison with healthy controls, with these levels negatively correlating with lung function. These metabolic signatures differentiated HDLI from other respiratory diseases in children, such as asthma and bronchiolitis obliterans. The 47 key metabolites identified in children were validated in an independent adult cohort. Furthermore, the classification performance of these key metabolites for HDLI achieved AUC values >0.95, which indicates a strong ability to differentiate between HDLI and healthy classes. The classification performance of these metabolic signatures for HDLI achieved an accuracy of 0.97, a precision of 0.95, an F1 score of 0.97,and are call of 1.00.

Related Figures

Fig. 1 Study population and metabolomic analysis process.Fig. 1 Study population and metabolomic analysis process.

Fig. 2 A. Spearman’s correlation analysis between key metabolites and lung function parameters in children diagnosed with HDLI. B. Evaluation of the classification model constructed by ML algorithms. Fig. 2 A. Spearman’s correlation analysis between key metabolites and lung function parameters in children diagnosed with HDLI. B. Evaluation of the classification model constructed by ML algorithms.

Our finding suggest a connection between HDLI and oxidative stress-induced lipid peroxidation. The oxidative stress signatures may serve as potential targets for biomarker development for HDLI, and the key metabolites could facilitate long term monitoring of HDLI.

Back  to List