DEVELOPMENT OF A MULTIVARIABLE MODEL TO PREDICT POSITIVE SURGICAL MARGIN IN PATIENTS WITH BORDERLINE RESECTABLE AND LOCALLY ADVANCED PANCREATIC DUCTAL ADENOCARCINOMA
Weizheng Ren*1,2, Dimitrios Xourafas1, Stanley W. Ashley1, Thomas Clancy1
1Surgery , Brigham and Women's Hospital, Boston , MA; 2Chinese PLA General Hospital, Beijing, China
Introduction: Despite the goal of achieving a negative resection margin of at least 1mm (R0), up to 90% of patients with borderline/locally advanced pancreatic ductal adenocarcinoma (BR/LA-PDAC) undergoing surgery will have positive resection margins (R1), irrespective of neoadjuvant therapy. It would be therefore advantageous to preoperatively predict the margin status (R) of these patients.
Methods: Preoperative assessment data as well as the R status of patients with BR/LA-PDAC who underwent a pancreatectomy between 2008 and 2018 at a single institute were retrospectively reviewed. Logistic regression analysis was used to evaluate the association between R1 margin status and relevant clinicopathological characteristics. Significant predictors of R1 resection on univariate analysis (p<0.1) were entered into a stepwise selection multivariable model using the Akaike information criterion. A nomogram predicting an R1 resection was also constructed.
Results: A total of 142 patients with BR/LA-PDAC were included in the analysis. An R0 resection was achieved in 82 patients (57.7%). The nomogram constructed identified that lymphadenopathy at diagnosis according to preoperative imaging (OR=2.1, p=0.05), pancreaticoduodenectomy (OR=3.8, p=0.04), a decrease of CA 19-9 serum levels (more than 50% decrease from diagnosis to restaging, OR=0.44, p=0.10), preoperative treatment with FOLFIRINOX (OR=0.46, p=0.10) and no vascular involvement (PV/SMV) at restaging (OR=0.14, p=0.01) were predictive of an R1 resection. The prediction model incorporating these 5 factors demonstrated good accuracy with an area under the receiver operating characteristic curve (ROC) of 0.74 and an internal validation with a bootstrapping generated mean ROC of 0.75 (95% confidence interval: 0.67-0.82, Figure1). Based on the proposed model, the prognostic nomogram yielded a probability of achieving an R1 resection ranging from <5% (0 factors included) to >70% (5 factors included) (Figure2).
Conclusions: Relevant preoperative clinicopathological characteristics accurately predict positive resection margins in patients with BR/LA PDAC before surgery. With further development, these results may be routinely used to preoperatively guide the surgical planning of patients with BR/LA PDAC.
Figure1. ROC curve of the prediction model
Figure2. Nomogram prediction of positive resection margins (R1): The ‘‘Total points’’ line is the sum of points of predictors (0-284 points), which corresponds to the ‘‘R1 probability’’ line indicating the probability of achieving an R1 resection.
Back to 2020 Abstracts