Hiroshi Nagata1, Soichiro Ishihara2, Koji Oba3,4, Toshiaki Tanaka5, Keisuke Hata5, Kazushige Kawai5, Hiroaki Nozawa5. 1. Department of Surgical Oncology, Faculty of Medicine, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan. hinagata-tky@umin.ac.jp. 2. Surgery Department, Sanno Hospital, International University of Health and Welfare, Tokyo, Japan. 3. Department of Biostatistics, School of Public Health, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan. 4. Interfaculty Initiative in Information Studies, The University of Tokyo, Tokyo, Japan. 5. Department of Surgical Oncology, Faculty of Medicine, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan.
Abstract
BACKGROUND: Detection of peritoneal metastasis remains challenging due to the limited sensitivity of current examination methods. This study aimed to establish a prediction model for estimating the individual risk of postoperative peritoneal metastasis from colon cancer to facilitate early interventions for high-risk patients. METHODS: This study investigated 1720 patients with stages 1-3 colon cancer who underwent curative resection at the University of Tokyo Hospital between 1997 and 2015. The data for the patients were retrospectively retrieved from their medical records. The risk score was developed using the elastic net techniques in a derivation cohort (973 patients treated in 1997-2009) and validated in a validation cohort (747 patients treated in 2010-2015). RESULTS: The factors selected using the elastic net approaches included the T stage, N stage, number of examined lymph nodes, preoperative carcinoembryonic antigen level, large bowel obstruction, and anastomotic leakage. The model had good discrimination (c-index, 0.85) and was well-calibrated after application of the bootstrap resampling method. Discrimination and calibration were favorable in external validation (c-index, 0.83). The model presented a clear stratification of patients' risk for postoperative peritoneal recurrence, and decision curve analysis showed its net benefit across a wide range of threshold probabilities. CONCLUSIONS: This study established and validated a prediction model that can aid clinicians in optimizing postoperative surveillance and therapeutic strategies according to the individual patient risk of peritoneal recurrence.
BACKGROUND: Detection of peritoneal metastasis remains challenging due to the limited sensitivity of current examination methods. This study aimed to establish a prediction model for estimating the individual risk of postoperative peritoneal metastasis from colon cancer to facilitate early interventions for high-risk patients. METHODS: This study investigated 1720 patients with stages 1-3 colon cancer who underwent curative resection at the University of Tokyo Hospital between 1997 and 2015. The data for the patients were retrospectively retrieved from their medical records. The risk score was developed using the elastic net techniques in a derivation cohort (973 patients treated in 1997-2009) and validated in a validation cohort (747 patients treated in 2010-2015). RESULTS: The factors selected using the elastic net approaches included the T stage, N stage, number of examined lymph nodes, preoperative carcinoembryonic antigen level, large bowel obstruction, and anastomotic leakage. The model had good discrimination (c-index, 0.85) and was well-calibrated after application of the bootstrap resampling method. Discrimination and calibration were favorable in external validation (c-index, 0.83). The model presented a clear stratification of patients' risk for postoperative peritoneal recurrence, and decision curve analysis showed its net benefit across a wide range of threshold probabilities. CONCLUSIONS: This study established and validated a prediction model that can aid clinicians in optimizing postoperative surveillance and therapeutic strategies according to the individual patient risk of peritoneal recurrence.