Abstract
Abstract Background: Approximately 25% of patients with esophageal adenocarcinoma achieve pathological complete response (pathCR) after chemoradiotherapy with following surgery (trimodality therapy). The prediction for pathCR would provide an important guidance for decision making regarding surgical strategy. Previous studies have used the logistic regression model to predict pathCR, however, few studies utilized the machine learning models. In this study, we aimed to establish a predictive model for pathCR using the machine learning models using clinical data in patients with GEJ cancer treated by trimodality therapy. Methods: 512 patients with localized esophageal and GEJ adenocarcinoma who received trimodality therapy in MD Anderson Cancer Center between 2002 to 2020 were included. We first performed survival analysis to confirm the survival benefit of pathCR. Then, several prediction models commonly used in machine learning, including logistic regression, LASSO, Random Forest, BART, and xgboost were used to predict pathCR. Each method was trained and validated using 10-fold cross-validation. Results: A total of 125 patients (24 %) achieved a path CR. Patients who achieved path CR had significantly longer overall survival (OS) and relapse free survival (RFS) than <pathCR 387 patients (76%) (median OS, 132 months vs 58 months, respectively, p < 0.002; median RFS, 63 months vs 27 months, respectively, p < 0.001). The logistic regression model indicated non-signet ring cell subtype, T1/T2 stage, negative biopsy after chemoradiation were independent predictive factors for pathCR. Finally, we created machine learning models and LASSO showing the highest predictive ability with AUC value of 0.644 among those models including conventional logistic regression model. The percent partial AUC at lower false positive rate of 0.25 was higher than that of the conventional logistic regression model (% pAUC, 23.7 vs 21.8, respectively) Conclusion: We established the clinical predictive model for pathCR in patients with GEJ cancer treated by trimodality therapy using machine learning method with acceptable predictive ability. Addition of molecular features might further refine the models. Citation Format: Kohei Yamashita, Jacob M. Maronge, Melissa Pool Pizzi, Wayne L. Hofstetter, Aileen Chen, Matheus Sewastjanow-Silva, Ernesto Rosa Vicentini, Meita S. Hirschmann, Anh Ta, Namita Shanbhag, Ying Yuan, Jaffer A. Ajani. Clinical model for predicting pathological complete response in patients with esophageal and gastroesophageal junction adenocarcinoma after trimodality therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 951.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.