Predicting Benefit of Neoadjuvant Chemotherapy and Elective Nodal Irradiation in Pancreatic Adenocarcinoma: A
Garrett K Harada1, Jino Park1, Nicholas Peterson1
1Department of Radiation Oncology, Chao Family Comprehensive Cancer Center, University of California, Irvine, California, USA.
Background:
The relative benefit of neoadjuvant therapies remains controversial for patients with (borderline) resectable pancreatic ductal adenocarcinoma (PDAC). The purpose of this study was to create a model to predict response to multiagent neoadjuvant chemotherapy (NAC) followed by radiotherapy with elective nodal irradiation (ENI).
Methods:
Using the National Cancer Database (NCDB), we identified patients with cT1-4N0M0 PDAC diagnosed between 2006 and 2020 treated with multiagent NAC, radiation with ENI, followed by curative resection with nodal dissection. A LASSO logistic regression model was used to predict ypN0 status, with generation of a nomogram and assessment of outcomes in training and testing cohorts. Secondary endpoints of negative-margin resection and overall survival were also examined. Out-of-sample predictions were then made on a separate ENI-naïve cohort, with similar assessments of selected outcomes. The threshold for statistical significance was set to p < 0.05.
Results:
A total of 1053 patients were identified with a median age of 64.0 years (IQR = 57-70 years). The final model included pancreatic body tumor location, clinical T stage, time from diagnosis to radiation therapy and surgery, ENI dose, and duration of NAC, among others. Patients predicted for treatment response were more likely to be ypN0 (71.5% vs. 29.7%, p < 0.001), had more R0 resections (87.3% vs. 62.6%, p < 0.001), and improved OS after accounting for competing risks of perioperative death (SHR = 0.64, 95% CI = 0.46-0.89, p = 0.008). A similar significant trend was noted in the ENI-naïve cohort (N = 1258). Model AUC was 0.718 and 0.725 in training and testing cohorts, respectively.
Conclusions:
Using a machine learning approach, we define a nomogram capable of predicting treatment response to multiagent NAC followed by radiotherapy with or without ENI. Patients selected by this model had higher rates of ypN0, higher R0 resection rates, and improved OS.
More Related Videos
07:13Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
07:15Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020