A Model-Informed Early Prediction of Methotrexate-Induced Acute Kidney Injury in Pediatric Patients with Osteosarcoma Using Real-World Data: A Multi-center Pharmacokinetic Study

Abstract

Background and objective: High-dose methotrexate is central to pediatric osteosarcoma treatment, but delayed elimination increases the risk of acute kidney injury, compromising treatment intensity. Current therapeutic drug monitoring often identifies toxicity too late for intervention. This study aimed to develop a population pharmacokinetic model using real-world data to identify early predictors of methotrexate-associated acute kidney injury.Methods: We conducted a multicenter study including 248 pediatric, adolescent, and young adult patients with osteosarcoma (1809 high-dose methotrexate courses) from referral centers in Argentina, Chile, Spain, and the USA. A population pharmacokinetic model was developed using Monolix. Early predictors of acute kidney injury were assessed using receiver operating characteristic curve analysis, and optimal sampling times earlier than those currently used in the clinics for rescue measures were explored through simulation.Results: A two-compartment population pharmacokinetic model adequately described the data up to 48 h after the start of infusion. Time-varying serum creatinine was included as a key covariate for methotrexate clearance. The alpha disposition half-life was the most accurate early predictor of moderate-to-severe acute kidney injury (area under the curve-receiver operating characteristic: 0.956 (95% confidence interval (CI) 0.914-0.998); threshold: 3.4 h (95% confidence interval 2.7-3.7), sensitivity: 0.94 [95% CI 0.81-1.0], and specificity: 0.96 [95% CI 0.73-0.99]). A two-point early sampling strategy (C4 and C8-11) precisely estimated alpha disposition half-life.Conclusions: Early identification of impaired methotrexate clearance is feasible using model-based metrics available within the first 11 h after the end of infusion by means of the alpha disposition half-life. This approach may guide supportive care offering a practical tool for model-informed precision dosing in pediatric patients with osteosarcoma.