Abstract: Objective To explore the correlation between the expression profiles of inflammatory cytokines and anti-tuberculosis treatment failure in type 2 diabetes mellitus (T2DM) patients complicated with active tuberculosis (ATB) (hereinafter referred to as ATB-T2DM patients). Method A total of 294 ATB-T2DM patients who were treated in People's Hospital of Xinjiang Uygur Autonomous Region from December 2020 to September 2023 were selected as the research subjects. All patients received 6 month standard anti-tuberculosis treatment. According to the treatment outcomes within 12 months after starting anti-tuberculosis treatment, they were divided into the treatment failure group(73cases) and the treatment success group(221cases). The basic clinical data and the expression profiles of serum inflammatory cytokines [interleukins (ILs), CXC chemokine ligands (CXCLs)] of the two groups were compared. Binary logistic regression analysis was used to evaluate the risk factors associated with treatment failure. The receiver operating characteristic curve (ROC) and the area under the curve (AUC curve) were employed to assess the efficacy of each risk factor in independently and jointly predicting treatment failure. The calibration curve and the decision curve were utilized to evaluate the accuracy of the joint prediction of each risk factor. Result Among the enrolled patients, 73 cases had treatment failure, with a treatment failure rate of 24.83%. There was a significant difference in the proportion of patients with pulmonary cavities between the treatment success group and the treatment failure group (P<0.05). The levels of IL-1β, IL-6, IL-10, CXCL2, CXCL5, CXCL8, and CXCL10 in the treatment failure group were significantly higher than those in the treatment success group (P<0.05). Binary logistic regression analysis showed that high levels of serum IL-1β, IL-6, IL-10, CXCL5, CXCL8, and CXCL10 were all risk factors for treatment failure (P<0.05). ROC curve analysis indicated that the AUCs of IL-1β, IL-6, IL-10, CXCL5, CXCL8, CXCL10 in independently and jointly predicting anti-tuberculosis treatment failure in ATB-T2DM patients were 0.649, 0.665, 0.617, 0.643, 0.626, 0.650, and 0.819 respectively. The joint prediction efficacy was significantly higher than that of each factor independently (P<0.05). The calibration curve suggested that the joint prediction model based on ILs and CXCLs had a good fit (Hosmer-Lemeshow χ²=10.068, P=0.260), and the decision curve indicated that this model had a high net benefit of use within the risk threshold range of 0-0.9. Conclusion The anti-tuberculosis treatment outcomes of ATB-T2DM patients are closely related to the levels of IL-1β, IL-6, IL-10, CXCL5, CXCL8, and CXCL10. The risk model for treatment failure constructed based on the above-mentioned indicators has a high predictive value.

Key words: Active tuberculosis, Diabetes mellitus, Interleukins, Chemokines, Treatment failure