BACKGROUND: Although a number of studies have been conducted to investigate factors affecting colon cancer recurrence and patient overall survival after surgical treatment, no prognostic risk models have been proposed for predicting survival specifically after postsurgical recurrence. OBJECTIVE: We aimed to identify factors affecting the survival of the patients with recurrent colon cancer and to construct a nomogram for predicting their survival. DESIGN: This was a retrospective study. SETTINGS: This study used the Japanese Study Group for Postoperative Follow-Up of Colorectal Cancer database, which contains retrospectively collected data of all consecutive patients with stage I to III colorectal cancer who underwent surgical curative resection between 1997 and 2008 at 23 referral institutions. PATIENTS: A total of 2563 patients with stage I to III colon cancer who experienced recurrence after surgery were included in the present study. MAIN OUTCOME MEASURES: A nomogram predicting survival was constructed using a training cohort composed of patients from 15 hospitals (n = 1721) using a Cox regression hazard model analysis. The clinical applicability of this nomogram was validated in patients from the 8 remaining hospitals (the validation cohort; n = 842). RESULTS: Eight factors (age, location of the primary tumor, histopathological type, positive lymph node status, presence of peritoneal metastasis, number of organs involved in the first recurrence, treatment for recurrence, and the interval between initial surgery and recurrence) were identified as nomogram variables. Our nomogram showed good calibration, with concordance indexes of 0.744 in the training cohort and 0.730 in the validation cohort. The survival curves stratified by the risk score calculated by the nomogram were almost identical for the training and validation cohorts. LIMITATIONS: The study was conducted using the data until 2008, and more advanced chemotherapeutic agents and multidisciplinary therapies that might have improved the outcomes predicted by our nomogram were not available. In addition, treatment strategies for recurrence might differ between countries. CONCLUSIONS: Our nomogram, which is based on a nationwide multicenter study, is the first statistical model predicting survival after recurrence in patients with stage I to III colon cancer. It promises to be of use in postoperative colon cancer surveillance. See Video Abstract at http://links.lww.com/DCR/A687.
BACKGROUND: Although a number of studies have been conducted to investigate factors affecting colon cancer recurrence and patient overall survival after surgical treatment, no prognostic risk models have been proposed for predicting survival specifically after postsurgical recurrence. OBJECTIVE: We aimed to identify factors affecting the survival of the patients with recurrent colon cancer and to construct a nomogram for predicting their survival. DESIGN: This was a retrospective study. SETTINGS: This study used the Japanese Study Group for Postoperative Follow-Up of Colorectal Cancer database, which contains retrospectively collected data of all consecutive patients with stage I to III colorectal cancer who underwent surgical curative resection between 1997 and 2008 at 23 referral institutions. PATIENTS: A total of 2563 patients with stage I to III colon cancer who experienced recurrence after surgery were included in the present study. MAIN OUTCOME MEASURES: A nomogram predicting survival was constructed using a training cohort composed of patients from 15 hospitals (n = 1721) using a Cox regression hazard model analysis. The clinical applicability of this nomogram was validated in patients from the 8 remaining hospitals (the validation cohort; n = 842). RESULTS: Eight factors (age, location of the primary tumor, histopathological type, positive lymph node status, presence of peritoneal metastasis, number of organs involved in the first recurrence, treatment for recurrence, and the interval between initial surgery and recurrence) were identified as nomogram variables. Our nomogram showed good calibration, with concordance indexes of 0.744 in the training cohort and 0.730 in the validation cohort. The survival curves stratified by the risk score calculated by the nomogram were almost identical for the training and validation cohorts. LIMITATIONS: The study was conducted using the data until 2008, and more advanced chemotherapeutic agents and multidisciplinary therapies that might have improved the outcomes predicted by our nomogram were not available. In addition, treatment strategies for recurrence might differ between countries. CONCLUSIONS: Our nomogram, which is based on a nationwide multicenter study, is the first statistical model predicting survival after recurrence in patients with stage I to III colon cancer. It promises to be of use in postoperative colon cancer surveillance. See Video Abstract at http://links.lww.com/DCR/A687.