Prognostic model for resected borderline and locally advanced pancreatic cancer after neoadjuvant chemotherapy.

Abstract

The current AJCC 8th has been reported to have a poor ability to predict the prognosis in patients with resected borderline resectable pancreatic cancer and locally advanced pancreatic cancer following neoadjuvant chemotherapy. This study was aimed to develop an improved prognostic model by incorporating pathology and parameters of biologic response (BR). A retrospective cohort study was conducted including patients who underwent curative-intent surgery following chemotherapy. We developed a modified ypT staging system and incorporated the BR, involving normalization of carbohydrate antigen 19-9 and reduction in the maximum standardized uptake value simultaneously after chemotherapy. The prognostic performance of the current pathologic system, modified pathologic system, and newly developed system incorporating pathology and BR were compared. In this study, 171 patients underwent surgery following chemotherapy. The modified T stage, which unified ypT2 and ypT3, demonstrated improved prognostic performance than the current staging system (area under the curve: 0.706 vs. 0.661). Biologic unresponsiveness was an independent prognostic factor for worse survival (hazard ratio 2.31, 95% confidence interval 1.50-3.55, P<0.001). The modified pathology with BR system demonstrated the highest discriminative ability in predicting 5-year overall survival than the current pathologic system (area under the curve: 0.785 vs. 0.661, P=0.010) and modified pathologic staging system (area under the curve: 0.785 vs. 0.706, P=0.002). The prognostic model, incorporating modified ypT staging and elevated carbohydrate antigen 19-9 levels and maximum standardized uptake value simultaneously, demonstrated improved results in predicting oncologic outcomes for patients who underwent surgery following neoadjuvant chemotherapy.