Abstract:

Objective: To investigate the predictive value of baseline chest CT imaging features for predicting treatment outcomes in patients with non-drug-resistant pulmonary tuberculosis during the treatment phase. Methods: A retrospective case-control study design was employed. Patients with non-drug-resistant pulmonary tuberculosis admitted at Tianjin University Haihe Hospital between July 2017 and April 2024 were enrolled. Patients were categorized into a poor prognosis group (n=118) based on treatment outcomes and matched with a good prognosis group (n=183). Differences in baseline chest CT features and clinical indicators between the two groups were systematically compared using univariate analysis and logistic regression analysis. An efficacy prediction model was subsequently constructed. The predictive performance of the model was evaluated using the receiver operating characteristic (ROC) curve. Results: Multivariate logistic regression analysis identified the following as independent risk factors for poor treatment prognosis: retreated tuberculosis (OR=6.257, 95%CI: 2.977-13.153, P<0.001), baseline CT showing cavity (OR=1.994, 95%CI: 1.116-3.564, P=0.020), bronchiectasis (OR=3.610, 95%CI: 1.395-9.341, P=0.008), fibrous stripe (OR=1.889, 95%CI: 1.073-3.328, P=0.028), involvement of more than three lung lobes (OR=2.293, 95%CI: 1.241-4.237, P=0.008), and decreased serum creatinine (OR=3.074, 95%CI: 1.539-6.137, P=0.001). The final prediction model formula was established as: logit(p)=-2.327+1.834×(Retreated tuberculosis)+0.690×(Cavity)+1.284×(Bronchiectasis)+0.636×(Fibrous stripe)+0.830×(Involved lung lobes>3)+1.123×(Decreased serum creatinine). The model achieved an area under the curve (AUC) of 0.808 (95%CI: 0.759-0.857). Conclusion: The prediction model constructed based on baseline chest CT imaging features combined with clinical indicators showed good predictive value for the treatment outcomes of patients with non-drug-resistant pulmonary tuberculosis.

Key words: Tuberculosis,pulmonary, Prognosis, Factor analysis, statistical, Risk factors, Tomography, X-ray computed