Distinct serum metabolic profiles with supportive diagnostic value in differentiating tuberculosis and Mycobacterium avium complex pulmonary disease

Abstract

Background: Pulmonary infectious diseases caused by Mycobacterium species, including Mycobacterium tuberculosis and Mycobacterium avium complex (MAC), remain significant public health threats. However, current gold-standard diagnostics are time-consuming and have limited ability to differentiate these clinically similar presentations. This study investigated serum metabolic distinctions between tuberculosis (TB) and MAC pulmonary disease (MAC-PD) to identify biomarkers with supportive diagnostic value for differential diagnosis. Methods: We performed LC/MS-based metabolic profiling of 181 serum samples from TB and MAC-PD patients. The study cohort was subsequently divided into a training set (TB, n = 30; MAC-PD, n = 30) and a validation set (TB, n = 51; MAC-PD, n = 70). Results: Five key metabolites were identified, including four sphingoid base lipids that were decreased in TB compared with MAC-PD, and 2-hydroxyglutaric acid (2-HG), which was increased. Logistic regression using this five-metabolite panel achieved strong discriminatory performance, with an area under the curve of 0.988 (95 % CI: 0.970-1.00 0) in the training set and 0.997 (95 % CI: 0.991-1.00 0) in the validation set. Consistent performance across multiple machine learning models reinforces the stability and supportive diagnostic value of the five-metabolite panel. Conclusions: This study provides a novel approach for the differential diagnosis of two major mycobacterial pulmonary diseases. The identified metabolites, particularly alterations in sphingoid base lipids and 2-HG, demonstrated robust discriminative potential. These findings support their potential role as biomarkers in clinical practice, enabling earlier and more accurate differentiation of TB and MAC-PD. (c) 2026 The Author(s). Published by Elsevier Ltd on behalf of King Saud Bin Abdulaziz University for Health Sciences. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).