Treatment response of patients with tuberculosis and HIV co-infection: a retrospective analysis of secondary data from Shanghai, China, 2010-2020.

Abstract

Background: At present, there is a need for more substantial real-world evidence on the factors influencing the effectiveness of tuberculosis (TB) treatment in HIV/TB co-infected patients.

Objectives: This retrospective study aims to identify factors affecting TB treatment effectiveness in HIV/TB co-infected patients.

Design: Retrospective cross-sectional study.

Methods: We included 461 HIV/TB co-infected patients, deriving 742 samples based on each initial positive TB test period. A total of 7788 valid treatment records corresponding to 17 TB drug compositions and 150 clinical indicators (each > 100 records) were used for analysis. Data mining techniques were employed, including consensus clustering, Fisher's exact test, stratified analysis, multivariate logistic regression analysis, and three modeling approaches (logistic regression, support vector machine, and random forest).

Results: The TB treatment effectiveness of CD4⁺ T cell count ⩽ 42 is significantly lower than that of the sample group > 42 (aOR: 1.77, 95% CI: 1.15-2.74, p = 0.010). The TB treatment effectiveness of the "rifabutin and levofloxacin alone or in combination" group is significantly higher than that of the "other first- and second-line anti-TB drugs in combination" group (aOR: 0.10, 95% CI: 0.01-0.64, p = 0.022). Significant differences exist in factors between TB treatment effective and ineffective groups, including age (aOR: 2.12, 95% CI: 1.10-4.20, p = 0.027), pre-treatment high-density lipoprotein (HDL) cholesterol (aOR: 0.47, 95% CI: 0.25-0.89, p = 0.022), pre-treatment CD8⁺ T cell count (aOR: 0.55, 95% CI: 0.33-0.90, p = 0.019), pre-treatment neutrophil percentage (aOR: 0.68, 95% CI: 0.48-0.96, p = 0.030), rifabutin (aOR: 1.59, 95% CI: 1.09-2.32, p = 0.016), and cycloserine (aOR: 0.21, 95% CI: 0.03-0.77, p = 0.041). The best area under the receiver operating characteristic curve of the test set under three modeling methods is 0.560-0.763. Rate of lymphocyte percentage recovering to normal is significantly higher in the TB treatment-effective group than in the treatment-ineffective group (aOR: 1.83, 95% CI: 1.09-3.10, p = 0.022).

Conclusion: CD4⁺ T cell count of 42/μL assists TB treatment effectiveness evaluation. Rifabutin and levofloxacin show more therapeutic benefits. Lymphocyte percentage can serve as an effective TB therapeutic and diagnostic target. Age, pre-treatment factors (HDL cholesterol, CD8⁺ T cell count, and neutrophil percentage), rifabutin, and cycloserine are significantly associated with TB treatment effectiveness. Factors affecting TB treatment effectiveness for HIV/TB co-infected patients need more evidence.