To investigate the clinical characteristics of anti-tuberculosis drug-induced liver injury (AT-DILI) and identify the influencing factors of nonrecovery outcomes, so as to provide evidence for risk identification and follow-up the management during anti-tuberculosis therapy.

Clinical data of 136 patients with AT-DILI admitted to the Fourth People’s Hospital of Qinghai Province from January 2020 to May 2024 were collected, retrospectively, including demographic characteristics, underlying liver disease, time to liver injury onset, clinical pattern of AT-DILI, severity grade, laboratory indicators, treatment and outcomes. Patients were divided into improved group (112 cases) and unresolved group (24 cases) according to the clinical outcomes, and clinical characteristics were compared between the two groups by a unified set of variables. The influencing factors for unrecovery outcomes in patients with AT-DILI were analyzed by Univariate and multivariate Logistic regression analyses. Receiver operating characteristic (ROC) curves were established based on the predicted probabilities from the final Logistic regression model, which contained hepatitis B and gamma-glutamyl transferase (GGT) level, and to evaluate the discriminative performance of this model for nonrecovery outcomes in patients with AT-DILI.

Among the 136 patients, 86 (63.2%) were male and 50 (36.8%) were female; 69 (50.7%) patients were complicated with hepatitis B, and 18 (13.2%) patients received artificial liver support therapy. Overall, 112 patients (82.4%) recovered or improved, whereas 24 (17.6%) patients had nonrecovery outcomes. AT-DILI mainly occurred during 60 days after initiation of anti-tuberculosis therapy (118 cases, 86.8%). Cholestatic pattern was the most common (96 cases, 70.6%), and grade 4 liver injury was the most common severity grade (49 cases, 36.0%). Age distribution, whether complicated with hepatitis B, severity of liver injury, drug resistance status, GGT, serum sodium (Na), procalcitonin (PCT) and hemoglobin (Hb) between the improved group and unresolved group of patients with AT-DILI were significantly different (all P<0.05). Multivariate Logistic regression analysis showed that hepatitis B (OR=7.838, 95%CI: 2.039-30.130, P=0.003) and GGT level (OR=1.012, 95%CI: 1.001-1.024, P=0.046) were both influencing factors of unresolved outcomes for patients with AT-DILI. The combined model based on hepatitis B and GGT level showed certain discriminative performance for nonrecovery (AUC=0.886, 95%CI: 0.819-0.953, P<0.001).

AT-DILI often occurs in early stage of anti-tuberculosis treatment, so liver function monitoring within the first two months after medication administration should be emphasized. Concomitant hepatitis B and elevated GGT level are risk factors for unresolved outcomes in patients with AT-DILI, which can serve as reference indicators for early identification of high-risk patients and the follow-up management.