The research, led by Zuoren Zhou and colleagues at The Affiliated Changsha Central Hospital of the University of South China, set out to answer a question that had not been systematically investigated before: whether the volume of MTB sequence reads returned by targeted sequencing could reliably differentiate bacteriological-positive from bacteriological-negative pulmonary tuberculosis. While targeted next-generation sequencing, or tNGS, has already shown substantial value in diagnosing pulmonary tuberculosis, particularly in cases that other methods fail to detect, its ability to gauge infectiousness had remained an open question.

The team initially evaluated 482 patients with suspected pulmonary tuberculosis who were seen between January 2023 and May 2025. After applying exclusion criteria, 369 patients entered the final analysis. For each participant, the researchers compiled clinical data, laboratory parameters, and tNGS reports, creating a dataset rich enough to test whether sequencing read counts tracked with established markers of bacterial burden.

The technical logic behind tNGS is central to the study. Unlike metagenomic sequencing, which reads all genetic material in a sample indiscriminately, targeted sequencing uses primers designed to enrich specific pathogen genomes before sequencing. In respiratory specimens such as bronchoalveolar lavage fluid, this enrichment means that the number of MTB reads recovered should, in principle, reflect the quantity of tubercle bacilli present. The researchers hypothesized that patients whose sputum or cultures yield detectable acid-fast bacilli or growing colonies would carry a heavier bacterial load and therefore produce more sequence reads.

That hypothesis held up in the data. High tNGS MTB sequence reads were closely associated with enhanced C-reactive protein, low body mass index, weight loss, and positivity on acid-fast bacilli smear and rapid culture. Patients with positive rapid culture or acid-fast bacilli detection exhibited substantially higher read counts than their bacteriological-negative counterparts. Statistical modeling confirmed that tNGS MTB sequence reads were significantly and independently predictive of bacteriological-positive status, meaning the association persisted after accounting for other clinical variables.

To quantify diagnostic performance, the researchers deployed a battery of analytical tools. Receiver operating characteristic analysis, which plots sensitivity against specificity across all possible cutoffs, yielded an area under the curve of 0.825 for distinguishing bacteriological-positive from bacteriological-negative pulmonary tuberculosis. The optimal cutoff was 10.163, at which the read count achieved a sensitivity of 68.2 percent and a specificity of 85.7 percent. In practical terms, the test correctly identified roughly two-thirds of the highly infectious patients while misclassifying fewer than one in seven of the bacteriological-negative group as positive.

Beyond the ROC curve, the team used decision curve analysis and calibration curves to assess clinical utility. Decision curve analysis evaluates the net benefit of acting on a test result across a range of risk thresholds, and the curves confirmed excellent clinical utility for the sequencing read count. Calibration analysis, which checks whether predicted probabilities match observed frequencies, further supported the reliability of the metric. Concordance with conventional diagnostic tests was evaluated using Cohen’s kappa coefficient, a statistic that measures agreement beyond chance.

Perhaps the most striking finding was the pattern of agreement with existing assays. tNGS sequence reads showed the highest agreement with acid-fast bacilli and rapid culture-positive detection, with a kappa of 0.559, which the authors characterize as moderate agreement. Notably, the sequencing read count outperformed existing immunological assays—such as the purified protein derivative skin test and the T cell spot test for tuberculosis infection—in diagnostic performance. Those immunological tools detect the host’s immune response rather than the pathogen itself, so they cannot distinguish active bacterial replication from past exposure or latent infection.

The implications for infection control are significant. Bacteriological-positive patients are the primary drivers of tuberculosis transmission, and identifying them quickly allows hospitals and public health programs to prioritize isolation, contact tracing, and early treatment. A sequencing read count that correlates with culture positivity could give clinicians a rapid, quantitative signal of infectiousness at the time of diagnosis, supporting the clinical application of tNGS in tuberculosis infection control and management that the authors envision.

The study does carry the limitations inherent to its retrospective, single-center design, and the authors present the read count as a promising potential indicator rather than a validated replacement for microbiological confirmation. Even so, the combination of an area under the curve of 0.825, strong specificity, and confirmed decision-analytic utility positions tNGS MTB sequence reads as a candidate biomarker worth prospective validation. As targeted sequencing becomes faster and more affordable, the idea that a pathogen-enriched read count could double as a transmission-risk gauge may reshape how tuberculosis diagnostics are deployed in the clinics and laboratories where the battle against this ancient disease continues.

Subject of Research: Diagnostic performance of targeted next-generation sequencing Mycobacterium tuberculosis sequence reads in pulmonary tuberculosis

Article Title: Diagnostic performance of tNGS mycobacterium tuberculosis sequence reads in differentiating bacteriological-positive from bacteriological-negative pulmonary tuberculosis

Article References: Zhou, Z., Xia, Y., Ma, X., Kang, X., Li, X., Wang, Q., Zhu, J., & Liu, S. (2026). Diagnostic performance of tNGS mycobacterium tuberculosis sequence reads in differentiating bacteriological-positive from bacteriological-negative pulmonary tuberculosis. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14649-w

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14649-w

Keywords: tuberculosis, targeted next-generation sequencing, Mycobacterium tuberculosis, diagnostic accuracy, bacteriological status, ROC curve, biomarker, infection control, BMC Infectious Diseases, sequence reads, acid-fast bacilli, rapid culture