Matched Cohort Analysis of Ultra-hypofractionated Versus Standard Fractionation Preoperative Radiation Therapy for Soft Tissue Sarcoma.

Abstract

Objective: This study compares toxicity and oncologic outcomes in a matched cohort of soft tissue sarcoma (STS) patients receiving ultra-hypofractionated preoperative radiation therapy (RT) or standard fractionated RT.

Methods: This IRB-approved study included patients with STS of the extremity, pelvis, or trunk treated with preoperative RT followed by surgical resection. Patients received either standard RT or ultra-hypofractionated RT (≥30 Gy over 5 fractions) between 2016 and 2023 with intensity-modulated RT at a single institution. Ultra-hypofractionated RT patients proceeded to surgical resection 0 to 7 days after RT and standard fractionated RT group 4 to 6 weeks after completion. The cohorts were matched based on tumor location and type of surgical closure. An inverse propensity weighting (IPW) method was used to balance group covariates.

Results: A total of 74 patients were included in this study. 37 patients treated with ultra-hypofractionated RT were matched with 37 patients treated with standard fractionation RT. Median follow-up time was 21.00 [IQR 11.00, 45.00] months for ultra-hypofractionated RT and 29.00 [IQR 13.00, 43.00] months for standard fractionated RT (P=0.58). Rates of major wound complications (MWC) were 44.4% ultra-hypofractionated RT versus 29.7% standard RT (P=0.289). On logistic regression, MWC (OR 1.9, 95% CI 0.97-3.76, P=0.06) and wound dehiscence (OR 3.91, 95% CI 1.81-8.73, P=0.0006) were more common in the ultra-hypofractionated RT group. Clinically significant late toxicity (grade ≥2 fibrosis, joint stiffness, or edema) did not differ significantly. There was no difference in local control (P=1.00) or distant metastases (P=0.465).

Conclusions: Ultra-hypofractionated RT for STS results in excellent disease control. To reduce the risk of MWC, we have adopted delayed surgical resection for ultra-hypofractionated RT patients of 4 to 6 weeks.