Association of baseline clinical factors with outcomes in patients with localized Ewing sarcoma treated on frontline trials with interval compressed chemotherapy (ICC): A report from the Children’s Oncology Group.

Abstract

10032 Background: Identifying clinical and biological factors associated with outcomes in localized Ewing sarcoma (ES) will enable risk-stratified clinical trials with the goal of improved outcomes for high-risk patients and decreased treatment toxicity for low-risk patients. The specific aims of this analysis in localized ES patients were 1) to classify extraosseous (EO) primary tumors with subdivisions into deep (viscera, glands, body cavities, muscle, nerves) and superficial (cutaneous and subcutaneous) sites, and 2) to understand the relationship between baseline clinical factors (age, sex, primary tumor site, size (maximum dimension and volume)) and event-free survival (EFS). Methods: The analytic cohort included ES patients treated with ICC on AEWS0031 and AEWS1031. Primary tumor sites were defined as pelvic, non-pelvic, and EO (deep vs. superficial). Post-enrollment EFS was the primary endpoint. Univariate analyses used the Kaplan–Meier method (logrank test). Multivariable analyses used Cox proportional hazards models. To assess the impact of tumor volume as a continuous variable, a subgroup analysis was conducted using AEWS1031 only, as tumor volumes were collected prospectively on this study. Visual exploration of effects of continuous variables on EFS event hazard used restricted cubic splines with 3 knots. Tests were performed at the 5% level. Results: AEWS0031 (n = 628) and AEWS1031 Regimen B (n = 283) yielded 911 patients. In univariate analyses, difference in risk of EFS event was observed between tumor sites ( P = 0.03). EO tumors had the highest estimated EFS compared to pelvic and non-pelvic (5 year EFS 84.7% vs. 72.4% and 75.9%). Superficial EO appeared to be a very low-risk group, albeit interpretations are limited by the small group size and singular event (one second malignant neoplasm among 15 patients). In multivariable analysis of the combined cohort, sex, race, and ethnicity were not prognostic. EO tumors may be associated with decreased hazard compared with non-pelvic bone primaries (HR 0.58, P = 0.07), and tumors ≥200 mL with increased hazard (HR 1.56, P < 0.01). In the subgroup analysis, tumor volume and age were prognostic and in visualizations age was non-linearly related to the hazard of EFS event, increasing until approximately 15 years. Tumor volume was non-linearly related to the hazard of EFS event and increased until approximately 400 mL. Conclusions: Patients with ES and primary tumors ≥200 mL continue to be at higher risk of an EFS event when treated with ICC, and EO tumors may be lower risk compared with other sites. Risk of an event appears to remain constant in ES patients ≥15 years or with primary tumors ≥400 mL. These findings should be validated in prospective trials and tumor biology integrated with clinical factors to improve risk stratification for localized ES.