Abstract
Electronic nose (e-nose) devices may be used to identify volatile organic compounds (VOCs) in exhaled breath. VOCs generated via metabolic processes are candidate biomarkers of (patho)physiological pathways. We explored the feasibility of using an e-nose to generate human "breathprints" at high altitude. Furthermore, we explored the hypothesis that pathophysiological processes involved in the development of acute mountain sickness (AMS) would manifest as altered VOC profiles. Breath analysis was performed on Sherpa and lowlander trekkers at high altitude (3500 m). The Lake Louise Scoring (LLS) system was used to diagnose AMS. Raw data were reduced by principal component (PC) analysis (PCA). Cross validated linear discriminant analysis (CV-LDA) and receiver-operating characteristic area under curve (ROC-AUC) assessed discriminative function. Breathprints suitable for analysis were obtained from 58% (37/64) of samples. PCA showed significant differences between breathprints from participants with, and without, AMS; CV-LDA showed correct classification of 83.8%, ROC-AUC 0.86; PC 1 correlated with AMS severity. There were significant differences between breathprints of participants who remained AMS negative and those whom later developed AMS (CV-LDA 68.8%, ROC-AUC 0.76). PCA demonstrated discrimination between Sherpas and lowlanders (CV-LDA 89.2%, ROC-AUC 0.936). This study demonstrated the feasibility of breath analysis for VOCs using an e-nose at high altitude. Furthermore, it provided proof-of-concept data supporting e-nose utility as an objective tool in the prediction and diagnosis of AMS. E-nose technology may have substantial utility both in altitude medicine and under other circumstances where (mal)adaptation to hypoxia may be important (e.g., critically ill patients).