Model predicts when TB blood test misses infection

Interferon-Gamma Release Assay (IGRA) is a commonly used blood test to help diagnose Pulmonary Tuberculosis (PTB), but it can sometimes give false-negative results, leaving infections undetected. To explore why this happens, a research team led by Wei Zhang examined clinical and laboratory data from 143 people diagnosed with PTB. The researchers separated patients into two groups based on test results: an observation group of 63 patients who were IGRA negative but pathogen positive, and a control group of 80 patients who were both IGRA positive and pathogen positive. To make fair comparisons and reduce bias from differing patient characteristics, they used Propensity Score Matching (PSM) to balance the groups; after matching, each group included 55 patients. With balanced groups, the team compared routine blood tests and other indicators to search for patterns linked to false-negative IGRA readings. The goal was practical: identify measurable factors that increase the chance of a missed IGRA, and build a tool clinicians can use to estimate that risk so patients who might be missed by IGRA can receive timely attention and treatment.

Using Propensity Score Matching (PSM) to control for confounding factors, the study compared clinical characteristics and laboratory indicators between matched groups of PTB patients. After matching, significant differences emerged in several routine blood measures: white blood cell count (WBC), neutrophil count (NEUT), lymphocyte count (LYM), red blood cell count (RBC), hemoglobin (HGB), and albumin (ALB) (P < 0.05). The team then applied logistic regression analysis to determine which of these were independent predictors of false-negative IGRA results. Logistic regression identified RBC and ALB as influencing factors for IGRA false negatives. Based on these findings, the researchers constructed a nomogram model to predict the likelihood of a false-negative IGRA. The nomogram showed a good overall fit (χ2 = 6.444, P = 0.598) and moderate discrimination with an area under the receiver operating characteristic curve (AUC) of 0.703 (95% CI: 0.605-0.800). Reported performance metrics included accuracy 0.682 (95% CI: 0.586-0.767), sensitivity 0.691 (95% CI: 0.569-0.813), specificity 0.673 (95% CI: 0.549-0.797), PPV 0.679 (95% CI: 0.556-0.801), and NPV 0.685 (95% CI: 0.561-0.809). Decision curve analysis indicated net benefit from using the nomogram when the threshold probability ranged from 0.15 to 0.75.

The study’s findings suggest that lower red blood cell counts (RBC) and lower albumin levels (ALB) are associated with a higher risk of a false-negative Interferon-Gamma Release Assay (IGRA) result in patients with Pulmonary Tuberculosis (PTB). By translating those measurable laboratory values into a nomogram, clinicians have a practical tool to estimate the probability that an IGRA result is falsely negative for an individual patient. The model’s AUC of 0.703 indicates it performs better than chance and could serve as a useful adjunct to existing diagnostic pathways, though it is not perfect. Importantly, decision curve analysis showed the nomogram provides clinical net benefit over a wide range of probability thresholds (0.15–0.75), meaning it could help guide decisions about when to pursue additional diagnostic testing or to treat empirically despite a negative IGRA. Wei Zhang and colleagues highlight that incorporating simple blood tests like RBC and ALB into prediction tools can improve early diagnosis and management strategies for PTB patients who might otherwise be overlooked by IGRA alone.