Age but Not Adiposity Predicts Asparaginase-Induced Hepatotoxicity during Induction Therapy for Adolescent and Young Adults with Acute Lymphoblastic Leukemia

Obesity is associated with increased treatment-related toxicity during therapy for acute lymphoblastic leukemia (ALL). During induction, hepatotoxicity due to L-asparaginase results in significant morbidity and chemotherapy delays/modifications. While hepatotoxicity is more common in older adolescent and young adult (AYA) patients, it cannot be predicted with high accuracy. Obesity defined by the surrogate marker of body mass index (BMI) has been shown to be a risk factor for hepatotoxicity; however, BMI does not accurately reflect body adiposity during ALL therapy due to concurrent muscle loss and adiposity gain. Because excess adiposity is generally believed to be responsible for most of the comorbidities of obesity, we hypothesized that body fat measured by direct assessment would be a better predictor of asparaginase-induced hepatotoxicity, and that non-obese patients with increased adiposity might be identified as an at-risk population which would not otherwise be recognized. We conducted an interim analysis of a prospective trial for an obesity intervention in pre-adolescent and AYA patients ≥10 years old newly diagnosed with B-ALL (NCT02708108). Whole-body dual-energy x-ray absorptiometry (DXA) performed at diagnosis and end of induction (EOI) directly quantified adiposity as total body fat percentage (TBF%) and fat mass (FM). Overweight or obesity (OW/OB) was defined using CDC-norms as a BMI percentile (BMI%) ≥85%. All patients were treated with Children's Oncology Group ALL regimens using a four-drug induction inclusive of glucocorticoids, daunorubicin, vincristine, and pegylated L-asparaginase. Obese and/or patients ≥15 years were considered at-risk for hepatotoxicity and thus eligible to receive exogenous supplementation with levocarnitine for hepatoprotection per provider discretion. The primary endpoint for the analysis was CTCAE Grade 3+ hepatotoxicity. Stepwise multivariable logistic regression models were constructed for the primary endpoint inclusive of TBF% and FM at diagnosis, change in FM during induction, and demographic candidate predictors. All tests were two-sided and significance was set at p