International Journal of Infectious Diseases (Apr 2025)

Identification and evaluation of blood transcriptional biomarker for tuberculosis screening

  • Siqi Zhang,
  • Cheng Bei,
  • Meng Li,
  • Jianfeng Zeng,
  • Liangzi Yang,
  • Tantan Ren,
  • Guofang Deng,
  • Ruimin Hong,
  • Juanjia Cai,
  • Dan Li,
  • Chuan Wang,
  • Peng Xu,
  • Howard Takiff,
  • Shuihua Lu,
  • Peize Zhang,
  • Qian Gao

DOI
https://doi.org/10.1016/j.ijid.2025.107838
Journal volume & issue
Vol. 153
p. 107838

Abstract

Read online

Objectives: Non-sputum-based screening methods for active case finding are a priority for ending tuberculosis. We sought to identify and evaluate blood transcriptional biomarkers suitable for tuberculosis screening. Methods: We integrated five blood RNA-seq datasets from global tuberculosis patients and identified genes that are differentially expressed between tuberculosis patients and healthy controls, using resampling and exhaustive testing. Three candidate biomarker combinations were identified from seven microarray datasets and small-scale clinical samples. The performance of these combinations for screening was evaluated in a cohort of close contacts of pulmonary tuberculosis (PTB) patients, and the results compared with Xpert HR. Results: We identified three 3-gene biomarker combinations, each containing two upregulated genes (FCGR1A, BATF2, or GBP5) and one downregulated gene (KLF2), and used these combinations to screen 352 close contacts of PTB. The biomarker combinations distinguished confirmed PTB patients from other participants with AUCs ranging from 0.848 to 0.870. With specificity fixed at 70%, all three combinations showed sensitivities of 87.5%. In a cohort of 205 presumptive pulmonary tuberculosis patients, the AUCs for distinguishing confirmed tuberculosis patients from other diseases ranged from 0.784 to 0.806. At 70% specificity, sensitivities were 75.9-81.5%, and were significantly higher with larger sputum bacterial loads. The performances of the three combinations for tuberculosis screening or diagnosis were comparable to Xpert HR. Conclusion: The three transcriptomic biomarkers identified in this study performed well for tuberculosis screening, nearly meeting the minimum WHO benchmarks for a triage test and showed potential utility in the development of new screening tools.

Keywords