Abstract
Interferon-γ release assay (IGRA) is a widely used blood test for detecting TB infection. However, a positive result of IGRA cannot differentiate active tuberculosis (ATB) infection from inactive tuberculosis (IATB). In this study, we established a nomogram model for predictive risk of ATB, differentiated from IATB, based on the concentration of interferon-γ (IFN-γ) of QuantiFERON-TB Gold In-Tube Test (QFT-GIT) and clinical characteristics. Participants with a positive QFT-GIT result were recruited and divided into a training and validation cohort according to hospitalisation date. The nomogram model for the differential diagnosis of ATB from IATB was established according to gender, age, pleural effusion (PE), and the concentration of IFN-γ in the Nil, TB antigen, and mitogen tube of QFT-GIT in the training cohort by logistic regression and validated in the validation cohort, and then combined with adenosine deaminase (ADA) to evaluated the performance value in ATB cases with PE. The area under receiver operating characteristic curve (AUC) of the diagnostic nomogram model, which we called the NSMC-ATB model for ATB diagnosis was 0.819 (95% CI 0.797-0.841), with sensitivity 73.16% and specificity 75.95% in training cohort, and AUC was 0.785 (95% CI 0.744-0.827), with sensitivity 67.44% and specificity 75.14% in validation cohort. A combination of the NSMC-ATB model and ADA performed better than the NSMC-ATB model and ADA alone in predicting ATB cases with PE, as AUC was 0.903 (95% CI 0.856-0.950) with sensitivity 78.63% and specificity 87.50%. We established an effective diagnostic nomogram model, called the NSMC-ATB model to differentiate ATB from IATB. Meanwhile, the combination of the NSMC-ATB model and ADA improved the performance value of ATB with PE.