Recurrence Prediction Model for Early Gastric Cancer

Joo Hoon Kim, Hyo Song Kim, Kiyeol Kim, MD, 5 Ji Fu Lai, MD, Hyun Cheol Chung, MD, 2,3,4 Sung Hoon Noh, MD 1,2,3 Sun Young Rha, MD, 2,3,4

1Cancer Metastasis Research Center, 2National Biochip Research Center, 3Brain Korea 21 Project for Medical Science, 4Yonsei Cancer Center, Yonsei Cancer Research Institute, 5Department of Internal Medicine, 1 Department of Surgery, Yonsei University College of Medicine

5 Oral cancer research institute,

Purpose: Recurrence in early gastric cancer (EGC) after curative resection is rare, and in which type of EGC would recur is not well studied. We tried to make a predicting system for recurrence in EGC after curative resection.

Methods: From January 1987 to April 2005, 2923 patients with EGC underwent R0 curative resection in at Severance Hospital, Yonsei University were enrolled. Logistic regression was performed to identify independent risk factors for recurrence in EGC. Nomogram was developed on the basis of the Cox regression. The concordance index (c-index) was used for accuracy measure, with bootstrapping to correct for optimistic bias. Calibration plots were constructed.

Results: Out of enrolled 2923 patients, 79 patients were recurred (2.7%). The median time to recurrence was 20.5 months, and among them early recurrence developed within 2 years after surgery were 60.7%. In logistic results, lymph node metastasis and elevated gross type were independent risk factors for overall recurrence; whereas overall recurrence rate was higher in patients with both identified risk factors (17.5%). Meanwhile, male gender, elevated gross type and lymph node metastasis were significantly associated with early recurrence, and in patients with all identified risk factors above, the early recurrence rate was 12.2%. Nomogram for predicting the disease-free survival after curative surgery was constructed. Its c-index was 0.79 and the calibration appeared to be accurate as internal validation.

Conclusion: Recurrence in EGC after R0 resection is predictable by using common clinical characteristics. Analyzing the combination of independent risk factors, patients in high risk of early and overall recurrence could be identified. An internally validated nomogram is established to predict disease-free survival accurately and individually.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 892.