RESUMEN
Purpose: Radiation therapy (RT) is an integral part of Ewing sarcoma (EwS) therapy. The Ewing 2008 protocol recommended RT doses ranging from 45 to 54 Gy. However, some patients received other doses of RT. We analyzed the effect of different RT doses on event-free survival (EFS) and overall survival (OS) in patients with EwS. Methods and Materials: The Ewing 2008 database included 528 RT-admitted patients with nonmetastatic EwS. Recommended multimodal therapy consisted of multiagent chemotherapy and local treatment consisting of surgery (S&RT group) and/or RT (RT group). EFS and OS were analyzed with uni- and multivariable Cox regression models including known prognostic factors such as age, sex, tumor volume, surgical margins, and histologic response. Results: S&RT was performed in 332 patients (62.9%), and 145 patients (27.5%) received definitive RT. Standard dose ≤ 53 Gy (d1) was admitted in 57.8%, high dose of 54 to 58 Gy (d2) in 35.5%, and very high dose ≥ 59 Gy (d3) in 6.6% of patients. In the RT group, RT dose was d1 in 11.7%, d2 in 44.1%, and d3 in 44.1% of patients. Three-year EFS in the S&RT group was 76.6% for d1, 73.7% for d2, and 68.2% for d3 (P = .42) and in the RT group 52.9%, 62.5%, and 70.3% (P = .63), respectively. Multivariable Cox regression revealed age ≥ 15 years (hazard ratio [HR], 2.68; 95% confidence interval [CI], 1.63-4.38) and nonradical margins (HR, 1.76; 95% CI, 1.05-2.93) for the S&RT group (sex, P = .96; histologic response, P = .07; tumor volume, P = .50; dose, P = .10) and large tumor volume (HR, 2.20; 95% CI, 1.21-4.0) for the RT group as independent factors (dose, P = .15; age, P = .08; sex, P = .40). Conclusions: In the combined local therapy modality group, treatment with higher RT dose had an effect on EFS, whereas higher dose of radiation when treated with definitive RT was associated with an increased OS. Indications for selection biases for dosage were found. Upcoming trials will assess the value of different RT doses in a randomized manner to control for potential selection bias.