Surgically treated pelvic liposarcoma and leiomyosarcoma: the effect of tumor size on cancer-specific survival

Introduction In soft tissue pelvic liposarcoma and leiomyosarcoma, it is unknown whether a specific tumor size cut-off may help to better predict prognosis, defined as cancer-specific survival (CSS). We tested whether different tumor size cut-offs, could improve CSS prediction. Materials and methods Surgically treated non-metastatic soft tissue pelvic sarcoma patients were identified (Surveillance, Epidemiology, and End Results 2004−2019). Kaplan-Meier plots, univariable and multivariable Cox-regression models and receiver operating characteristic-derived area under the curve (AUC) estimates were used. Results Overall, 672 (65%) liposarcoma (median tumor size 11 cm, interquartile range [IQR] 7-16) and 367 (35%) leiomyosarcoma (median tumor size 8 cm, IQR 5-12) patients were identified. The p-value derived ideal tumor size cut-off was 17.1 cm, in liposarcoma and 7.0 cm, in leiomyosarcoma. In liposarcoma, according to p-value derived cut-off, five-year CSS rates were 92 vs 83% (≤17.1 vs >17.1 cm). This cut-off represented an independent predictor of CSS and improved prognostic ability from 83.8 to 86.8% (Δ = 3%). Similarly, among previously established cut-offs (5 vs 10 vs 15 cm), also 15 cm represented an independent predictor of CSS and improved prognostic ability from 83.8 to 87.0% (Δ = 3.2%). In leiomyosarcoma, according to p-value derived cut-off, five-year CSS rates were 86 vs 55% (≤7.0 vs >7.0 cm). This cut-off represented an independent predictor of CSS and improved prognostic ability from 68.6 to 76.5% (Δ = 7.9%). Conclusions In liposarcoma, the p-value derived tumor size cut-off was 17.1 cm vs 7.0 cm, in leiomyosarcoma. In both histologic subtypes, these cut-offs exhibited the optimal statistical characteristics (univariable, multivariable and AUC analyses). In liposarcoma, the 15 cm cut-off represented a valuable alternative.