Exhaled Breath Biomarkers Reflect the Inflammasome and Lipidome Changes in Ischemic Heart Disease: Using Machine Learning Models and Network Analysis

Ischemic heart disease (IHD) is a global health crisis exacerbated by diagnostic and preventive challenges. This study investigated associations between lipid profiles, inflammation biomarkers, and exhaled volatile organic compounds (VOCs) in IHD using machine learning. Eighty participants (31 IHD patients confirmed by computed tomography perfusion [CTP]; 49 controls) underwent CTP, breath analysis (PTR-TOF-MS-1000), and blood testing. LASSO regression with cross-validation analyzed links between VOCs, lipid parameters (HDL, LDL, ApoB, Lp(a), total cholesterol), and inflammation markers (IL-6, CRP). Controls showed minimal plasma biomarker-VOC correlations, whereas IHD patients demonstrated strong lipid-VOC relationships: HDL positively correlated with m/z 49.995 (r=0.31), and total cholesterol inversely with m/z 94.053 (r=−0.35). Key discriminative VOCs included 2-ethyl-2,5-dihydro-4,5-dimethylthiazole, HO3PS2, CH8N3P, and m/z 49,995.01251. IHD patients exhibited dynamic lipid-VOC interactions post-stress (e.g., ApoB, LDL), absent in controls. Inflammation markers had limited direct VOC ties but unique patterns: IL-6 inversely linked to total cholesterol in patients, contrasting with CRP/HDL correlations in controls. These findings highlight VOC-lipidome networks as potential non-invasive biomarkers for IHD-specific metabolic dysregulation. Integrating breathomics with machine learning may advance early diagnosis and personalized risk stratification, addressing critical gaps in IHD management.