Serum proteomics and machine learning unveil new diagnostic biomarkers for tuberculosis in adolescents and young adults.

Yu CHEN; Hongxiang XU; Yao TIAN; Qian HE; Xiaoyun ZHAO; Guobin ZHANG; Jianping XIE

Return

Serum proteomics and machine learning unveil new diagnostic biomarkers for tuberculosis in adolescents and young adults.

Author: Yu CHEN ¹ ; Hongxiang XU ² ; Yao TIAN ¹ ; Qian HE ³ ; Xiaoyun ZHAO ⁴ ; Guobin ZHANG ⁵ ; Jianping XIE ²
Author Information

1. Department of Tuberculosis, China Shenyang the Tenth People's Hospital (China Shenyang Chest Hospital), Shenyang 110044, Liaoning, China.
2. Institute of Modern Biomedical Research, School of Life Sciences, Southwest University, Chongqing 400715, China.
3. School of Public Health, China Medical University, Shenyang 110044, Liaoning, China.
4. School of Life Science and Biopharmaceutics, Shenyang Pharmaceutical University, Shenyang 110016, Liaoning, China.
5. Shenyang Disease Control Center, Shenyang 110623, Liaoning, China.
Publication Type:Journal Article
Keywords: apolipoprotein; diagnostic biomarkers; machine learning; serum proteomics; tuberculosis
MeSH: Humans; Proteomics/methods*; Biomarkers/blood*; Adolescent; Young Adult; Apolipoprotein A-I/blood*; Machine Learning; Tuberculosis/blood*; Proteome/analysis*; Male; Hemoglobins/analysis*; Female; Blood Proteins/analysis*; Adult
From: Chinese Journal of Biotechnology 2025;41(4):1478-1489
CountryChina
Language:Chinese
Abstract: Adolescents and young adults (AYAs) are one of the major populations susceptible to tuberculosis. However, little is known about the unique characteristics and diagnostic biomarkers of tuberculosis in this population. In this study, 81 AYAs were recruited, and the high-quality serum proteome of the AYAs with tuberculosis was profiled by quantitative proteomics. The data of serum proteomics indicated that the relative abundance of hemoglobin and apolipoprotein was significantly reduced in the patients with active tuberculosis (ATB). The pathway enrichment analysis showed that the downregulated proteins in the ATB group were mainly involved in the antioxidant and cell detoxification pathways, indicating extensive oxidative stress damage. Random forest (RF) and extreme gradient boosting (XGBoost) were employed to evaluate protein importance, which yielded a set of candidate proteins that can distinguish between ATB and non-ATB. The analysis with the support vector machine algorithm (recursive feature elimination) suggested that the combination of apolipoprotein A-I (APOA1), hemoglobin subunit beta (HBB), and hemoglobin subunit alpha-1 (HBA1) had the highest accuracy and sensitivity in diagnosing ATB. Meanwhile, the levels of hemoglobin (HGB) and albumin (ALB) can be used as blood biochemical indicators to evaluate changes in the protein levels of APOA1 and HBB. This study established the serum proteome landscape of AYAs with tuberculosis and identified new biomarkers for the diagnosis of tuberculosis in this population.