# Prognosis Prediction Model After Upfront Surgery for Resectable Pancreatic Ductal Adenocarcinoma: A Multicenter Study (OS-HBP-2)

## TL;DR

This study developed a model to predict survival after surgery for resectable pancreatic cancer, validated using data from 603 patients across multiple centers.

## Contribution

A novel internally validated prognosis prediction model for overall survival after surgery in resectable pancreatic ductal adenocarcinoma patients.

## Key findings

- The 1-, 3-, and 5-year overall survival rates were 83.7%, 48.2%, and 37.5%, respectively.
- Key predictors of survival included tumor size, vascular contact, CA19-9 levels, and the modified Glasgow Prognostic Score.
- The model showed an area under the curve of 0.68 and a concordance index of 0.63.

## Abstract

Although several prognostic nomograms have been developed to estimate the prognosis of patients with pancreatic ductal adenocarcinoma (PDAC), further investigation is re-quired to optimize the prognostic assessment tools for patients with resectable PDAC. This multicenter study (n = 603) aimed to investigate the prognostic factors for survival and develop a prognosis prediction model after upfront surgery in patients with resectable PDAC. Using the results of the multivariate analyses, a prognosis prediction model for overall survival was constructed. This study suggests that our model may be useful and can be internally validated for predicting overall survival following upfront surgery in patients with resectable PDAC.

Background/Objectives: Upfront surgery (UFS) remains the standard treatment for patients with resectable pancreatic ductal adenocarcinoma (PDAC). We aimed to investigate the prognostic factors for survival after UFS in patients with resectable PDAC and to develop a prognostic prediction model. Methods: This multicenter, retrospective study included 603 patients who underwent UFS for resectable PDAC between January 2013 and December 2017. Univariate and multivariate analyses were performed to identify prognostic factors for overall survival (OS). We constructed a prognostic prediction model for OS after UFS. An internal validation was performed to evaluate the discriminative performance of the model. Results: The 1-, 3-, and 5-year OS rates were 83.7%, 48.2%, and 37.5%, respectively. The Cox proportional hazards model showed that tumor size > 2 cm (hazard ratio [HR] 1.50, p = 0.001); tumor contact with the portal and superior mesenteric veins of ≤180° (HR 1.47, p = 0.003); carbohydrate antigen 19-9 levels of 40 to 500 U/mL (HR 1.59, p = 0.002) and ≥500 U/mL (HR 2.16, p < 0.001); and a modified Glasgow Prognostic Score of two (HR 1.56, p = 0.038) were predictors associated with OS. The prognostic prediction model for 5-year OS demonstrated an area under the curve of 0.68. The calibration plots indicate a concordance index of 0.63. Conclusions: We identified the preoperative prognostic factors for OS and developed a prognostic prediction model to estimate OS in patients undergoing UFS for resectable PDAC. Our model may be useful and internally validated for predicting OS.