Exhaled Breath Biomarkers Reflect the Inflammasome and Lipidome Changes in Ischemic Heart Disease: A Study Using Machine Learning Models and Network Analysis.
Basheer Abdullah Marzoog, Peter Chomakhidze, Daria Gognieva, Artemiy Silantyev, Alexander Suvorov, Anastasia Stroeva, Malika Mustafina, Alina Yur'evna Fedorova, Abram Syrkin, Philipp Kopylov
{"title":"Exhaled Breath Biomarkers Reflect the Inflammasome and Lipidome Changes in Ischemic Heart Disease: A Study Using Machine Learning Models and Network Analysis.","authors":"Basheer Abdullah Marzoog, Peter Chomakhidze, Daria Gognieva, Artemiy Silantyev, Alexander Suvorov, Anastasia Stroeva, Malika Mustafina, Alina Yur'evna Fedorova, Abram Syrkin, Philipp Kopylov","doi":"10.12997/jla.2025.14.3.350","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>To define relationships between lipidomics, inflammasome, and exhaled volatile organic compounds (VOCs) in ischemic heart disease (IHD) and develop a VOC-based diagnostic machine learning model for non-invasive diagnosis.</p><p><strong>Methods: </strong>A single-center prospective study involved 80 participants between 27 Oct 2023 and 11 Jun 2024: 31 with stress-computed tomography (CT) myocardial-perfusion-confirmed IHD and 49 perfusion-negative controls. All underwent stress CT perfusion, bicycle-ergometry, and breath collection at rest, peak exercise, and 3-minute recovery into a PTR-TOF-MS-1000. Lipid measurements were made (total, high-density lipoprotein [HDL]-, low-density lipoprotein [LDL]-, very LDL-cholesterol, triglycerides, apolipoprotein B [ApoB], lipoprotein-a) and inflammatory biomarkers (interleukin-6, C-reactive protein). LASSO regression mapped VOC-biomarker associations. An XGBoost classifier integrating VOCs, lipidome, inflammasome, and lipid-lowering therapy status was evaluated with cross-validated Youden index.</p><p><strong>Results: </strong>Controls showed minimal biomarker-VOC relationships. Patients exhibited significant lipid-VOC correlations, including HDL-C with m/z 49.995 (r=0.31) and an inverse correlation between total cholesterol and m/z 94.053 (r=-0.35). Key discriminative VOCs were 2-ethyl-2,5-dihydro-4,5-dimethylthiazole, HO3PS2, CH8N3P, and m/z 49.995. Exercise revealed dynamic ApoB and LDL interactions exclusive to IHD. Inflammasome had limited direct VOC links; IL-6 inversely correlated with total cholesterol in IHD, while CRP aligned with HDL in controls. The final model achieved: AUC 0.931 (95% confidence interval [CI], 0.869-0.978), sensitivity 0.613 (95% CI, 0.435-0.793), specificity 1.000 (95% CI, 1.000-1.000), NPV 0.803 (95% CI, 0.692-0.903), PPV 1.000 (95% CI, 1.000-1.000).</p><p><strong>Conclusion: </strong>Exhaled VOC patterns reflect lipid dysregulation in IHD. Combined with lipid and inflammatory data, VOCs enable high-accuracy, non-invasive IHD discrimination, supporting breathomics as a promising diagnostic adjunct.</p><p><strong>Trial registration: </strong>ClinicalTrials.gov Identifier: NCT06181799.</p>","PeriodicalId":16284,"journal":{"name":"Journal of Lipid and Atherosclerosis","volume":"14 3","pages":"350-371"},"PeriodicalIF":0.0000,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12488795/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Lipid and Atherosclerosis","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.12997/jla.2025.14.3.350","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/7/8 0:00:00","PubModel":"Epub","JCR":"Q2","JCRName":"Medicine","Score":null,"Total":0}
引用次数: 0
Abstract
Objective: To define relationships between lipidomics, inflammasome, and exhaled volatile organic compounds (VOCs) in ischemic heart disease (IHD) and develop a VOC-based diagnostic machine learning model for non-invasive diagnosis.
Methods: A single-center prospective study involved 80 participants between 27 Oct 2023 and 11 Jun 2024: 31 with stress-computed tomography (CT) myocardial-perfusion-confirmed IHD and 49 perfusion-negative controls. All underwent stress CT perfusion, bicycle-ergometry, and breath collection at rest, peak exercise, and 3-minute recovery into a PTR-TOF-MS-1000. Lipid measurements were made (total, high-density lipoprotein [HDL]-, low-density lipoprotein [LDL]-, very LDL-cholesterol, triglycerides, apolipoprotein B [ApoB], lipoprotein-a) and inflammatory biomarkers (interleukin-6, C-reactive protein). LASSO regression mapped VOC-biomarker associations. An XGBoost classifier integrating VOCs, lipidome, inflammasome, and lipid-lowering therapy status was evaluated with cross-validated Youden index.
Results: Controls showed minimal biomarker-VOC relationships. Patients exhibited significant lipid-VOC correlations, including HDL-C with m/z 49.995 (r=0.31) and an inverse correlation between total cholesterol and m/z 94.053 (r=-0.35). Key discriminative VOCs were 2-ethyl-2,5-dihydro-4,5-dimethylthiazole, HO3PS2, CH8N3P, and m/z 49.995. Exercise revealed dynamic ApoB and LDL interactions exclusive to IHD. Inflammasome had limited direct VOC links; IL-6 inversely correlated with total cholesterol in IHD, while CRP aligned with HDL in controls. The final model achieved: AUC 0.931 (95% confidence interval [CI], 0.869-0.978), sensitivity 0.613 (95% CI, 0.435-0.793), specificity 1.000 (95% CI, 1.000-1.000), NPV 0.803 (95% CI, 0.692-0.903), PPV 1.000 (95% CI, 1.000-1.000).
Conclusion: Exhaled VOC patterns reflect lipid dysregulation in IHD. Combined with lipid and inflammatory data, VOCs enable high-accuracy, non-invasive IHD discrimination, supporting breathomics as a promising diagnostic adjunct.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
