Omid Reza Zekavat, Mohammad Hadi Imanieh, Ali Behniapour, Reza Sadeghikhoo
Background Acute lymphoblastic leukemia (ALL) is the most common childhood cancer, accounting for 23% of pediatric malignancies worldwide. In Iran, the cumulative incidence is approximately 21 cases per 100,000 children under 15, with a five-year survival rate of 67%. Maintenance-phase chemotherapy includes 6-mercaptopurine, methotrexate, and vincristine. Materials and methods This prospective cohort study, conducted from October 2023 to October 2025 in southern Iran, comprised two separate cohorts: 117 children at the initiation of maintenance phase and 79 children in the post-treatment phase. Liver function tests (LFTs) were measured at baseline and after three months. Drug-induced liver injury (DILI) was diagnosed according to Hy's Law. Changes in laboratory parameters were statistically analyzed. Logistic regression was used for DILI prediction, evaluating AUC, sensitivity, and specificity. The synthetic minority oversampling technique (SMOTE) was applied for class balancing. Results During the maintenance phase, significant increases were observed in ΔALT (+20.21; p = 0.039) and ΔTotal Bilirubin (+0.10; p = 0.005). Post-treatment findings showed significant decreases in ΔALT (−28.46; p < 0.001) and ΔAST (−9.94; p = 0.001). The incidence of DILI was 6% during maintenance phase and 0% after treatment. Initial blast percentage showed modest discrimination for DILI (AUC = 0.68; sensitivity 85.7%, specificity 58.2%). Conclusion Chemotherapy-induced hepatotoxicity demonstrated short-term improvement after treatment cessation, with no cases of DILI meeting Hy's Law criteria at 3-month follow-up. An initial blast percentage >84% showed modest discrimination for DILI in the maintenance cohort and requires validation in larger populations before clinical screening.