Association between Tumor Volume Change on MRI with Surgical Margin Status, Pathological Response, and Local Control Following Pre-Operative Radiation Therapy for Soft Tissue Sarcoma

Abstract

The clinical significance of radiographic progression during pre-operative radiation therapy (RT) for soft tissue sarcoma (STS) remains unclear. We sought to evaluate associations between radiographic change on T1 post-contrast (T1c) and T2 weighted magnetic resonance imaging (MRI) with percent pathological response (PR%), positive surgical margins (+SM), and local control (LC). We retrospectively identified patients with STS undergoing neoadjuvant RT who had both pre- and post-RT MRI prior to surgical resection. Gross tumor volumes were contoured on pre- and post-RT T1c and T2 MRI sequences and relative change in volume from baseline was calculated. Radiographic classification was defined as response (>30% reduction), progression (>30% increase), or stable (≤30% reduction or ≤30% increase). Chi squared, Fishers Exact, and Kruskal Wallis (KW) tests were used to assess differences between groups. Linear and binary logistic regression models used to assess associations between MRI response and PR% and +SM, respectively. LC was modeled with Kaplan Meier methods and log rank tests. A total of 68 STS patients were identified, with a median follow up of 49 months (range 7-229). With a median age of 60.5 years (25-88) and tumor size of 10.8cm (2.7-25.7), the most common histologies were undifferentiated pleomorphic sarcoma (UPS; 32.4%) and myxoid liposarcoma (ML; 16.2%), and were primarily grade 2-3 disease (89.7%). With a median RT dose of 50 Gy in 25 fractions (44-60Gy), the median radiographic volume change was 2% (-86.4 to 953.6%) and -2.1% (-89.6 to 962.5%) for T1c and T2, respectively. Radiographic classification of response/stable/progression was 25.4%/49.2%/25.4% and 27.9%/52.5%/19.7% for T1c and T2, respectively. Histology (ML vs. UPS) and grade (1 vs. 3) were predictors for radiographic response on both T1c (72.7% vs 18.8%, p = 0.03 and 71.4% vs. 10.4%, p = 0.03) and T2 (71.5% vs. 18.2%, p = 0.02 and 71.4% vs. 14.6%, p = 0.002), respectively. With 6 +SM (8.8%), the rate of +SM for response/stable/progression was 20%/10.3%/0% in T1c (p = 0.2) and 5.9%/12.5%/0% in T2 (p = 0.5). As a continuous variable, neither relative change on T1c (p = 0.2) or T2 (p = 0.4) were associated with +SM. With an overall median PR% of 64% (0-100%), the median PR% was significantly different for response/stable/progression for both T1c (95%/42%/73%, p = 0.02) and T2 (95%/50%/87.5%, p = 0.04). Radiographic change on neither T1c (p = 0.4) or T2 (p = 0.5) were associated with PR% on a continuous basis. With a total of 4 local recurrences, there was no significant difference in LC by radiographic classification on either T1c (p = 0.65) or T2 (p = 0.85). While radiographic response may be correlated with pathological response, radiographic progression on either T1c or T2 following neoadjuvant RT was not associated with a detriment in surgical margins or local control. These findings suggest that STS radiographic "pseudoprogression" is not associated with worse outcome.