Real-World Application of a Machine Learning-Based Early Recurrence Model for Guiding Adjuvant Chemotherapy Use in
Won-Gun Yun1,2, Youngmin Han1, Yoon Soo Chae1
1Department of Surgery and Cancer Research Institute, Seoul National University College of Medicine, Seoul, Korea.
Background/Aims:
Although recent randomized controlled trials have reported the efficacy of adjuvant chemotherapy for resected biliary tract cancer, discrepancies remain between recommendations and real-world practice. Therefore, we aimed to assess the efficacy of adjuvant chemotherapy in patients with gallbladder cancer and used machine learning-based risk stratification to predict recurrence to avoid unnecessary chemotherapy.
Methods:
Patients who underwent surgery between 2005 and 2022 and were histologically diagnosed with stage 2 or above gallbladder cancer were included. The patients were stratified by risk of early recurrence according to the machine learning-based algorithm suggested by Catalano and colleagues.
Results:
Among 395 patients, 204 (51.6%) and 191 (48.4%) were determined to have a low and high risk of early recurrence, respectively. Although the 5-year overall survival rates were not significantly different between the adjuvant chemotherapy and surveillance groups (87.2% vs 83.3%; p=0.233) in the low-risk patients, the adjuvant chemotherapy was associated with a significantly higher 5-year overall survival rate in the high-risk patients (52.1% vs 37.8%; p=0.003).
Conclusions:
Machine learning-based prediction of early recurrence is helpful in selecting patients who may benefit from adjuvant chemotherapy. These findings may help reduce medical expenses by avoiding unnecessary adjuvant chemotherapy in patients with low risk of early recurrence.
