Retroperitoneal sarcoma often requires comprehensive resection, leading to severe postoperative morbidity. The lack of disease-procedure specific tools for morbidity risk and the questionable accuracy of existing tools (ACS-NSQIP and P-POSSUM) in RPS surgery drove this study to assess these calculators' accuracy. Retrospective analysis of primary RPS cases undergoing surgery at two sarcoma-referral centers was conducted. Predicted morbidity/mortality rates at 90days postsurgery, classified by Clavien-Dindo (CD) and Comprehensive Complication Index (CCI), were compared with observed data. Accuracy was assessed by Brier Score and area under the curve (AUC). Inflammatory Biomarkers Prognostic Index (IBPI) also was tested. A total of 567 patients (median age 62years; 53.6% male) with a median of four resected organs were included. 59% experienced surgical complications by 90 days postoperation, graded CD ≥ 3 in 30.5%, median CCI 20.9, with a mortality rate of 1.6% (8/567). Reoperation was required in 68 of 567 patients (12%). Thirty-day mortality was 1.1%. Severe complications occurred after 30th postoperative day in 3.5% cases. ACS-NSQIP predicted below-average complication for 65.1%, average for 16.9%, and above-average for 18% of patients. P-POSSUM predicted a 66% rate of morbidity and 4% mortality. None of the prediction tools were accurate, with Brier scores ranging 0.155-0.231 and no AUC ≥ 0.7. IBPI accuracy for predicting severe infective complication was low (AUC 0.58, Brier 0.161). The significant morbidity burden after MVR necessitates reliable evaluation, especially in frail patients. Given the limitations of ACS-NSQIP and P-POSSUM, a dedicated prediction tool for perioperative events in RPS candidates for MVR needs urgent development.
Read full abstract