|
|
Back to 2016 Annual Meeting
Validated Asian Prognosis Tool Accurately Predicts Gastric Cancer Outcome of Ethnically Diverse Patient Population at an American National Comprehensive Cancer Center
Bryan S. Goldner*1, Kijun Song2, Taeil Son2, Jong Won Kim2, Laleh Melstrom1, Gagandeep Singh1, Sung Hoon Noh2, Yuman Fong1, Woo Jin Hyung2, Yanghee Woo1
1Department of Surgery, City of Hope National Medical Center, Duarte, CA; 2Yonsei University College of Medicine, Seoul, Korea (the Republic of)
Introduction: A novel prediction model was developed by an international collaborative group (G6+) for accurate determination of 5-year overall survival of gastric cancer patients. This prediction model was created using a single institution database of 12,399 patients and included clinically relevant factors not accounted for in the TNM staging system. This model has been validated using external data sets from Asia. However, the ethnic makeup of the United States is very diverse with cultures and ethnicities from throughout the world, bringing into question whether this nomogram will provide accurate prediction. In this study, we examined whether this prognostic tool can be used in a multiethnic cohort such that seen at an American comprehensive cancer center.
Methods: Using the gastric cancer database at a comprehensive cancer center, we examined the following characteristics for analysis: age, sex, gender, ethnicity, depth of tumor invasion, number of positive lymph nodes, total lymph nodes retrieved, presence of distant metastasis, extent of resection, histology, and survival. These data were processed through the recently published prognostic nomogram to obtain concordance index (C-statistic) using the bootstrap method and calibration was assessed. This was then compared to the current prognostic index, the TNM staging system.
Results: A total of 120 patients were identified during years 2004-2014. Ethnic breakdown included: 42.5% non-Hispanic white, 39.2% Asian, 11.7% Hispanic white, 2.5% Black, 0.8% Native American, 3.3% unknown. The majority of patients presented with poorly differentiated adenocarcinoma (45%), followed by moderately differentiated (25%), and signet ring type (23.3%). Distant metastatic disease was present in 14.2% of patients. Subtotal gastrectomy (75%) was the most common procedure performed. The novel prognostication model was superior to AJCC staging: a C-statistic of 0.861 (95% CI 0.797-0.925) compared to the 7th edition. TNM staging model C-statistic 0.647 (95% CI 0.567-0.727) (p-value <0.001).
Conclusions: Our study demonstrates superior accuracy using the novel prediction model for gastric cancer in the American patient population using a comprehensive cancer center validated dataset. This model takes into account factors not accounted for in the TNM staging system including: extent of lymphadenectomy, histology, and extent of resection, making this a more accurate tool, while maintaining global relevance due to the widespread and easy availability of these different factors. Unlike previous datasets used for validation, this is the first single institution dataset with a heterogenous ethnic makeup. Our data further supports its applicability to a very ethnically heterogeneous American gastric cancer population as well as its’ global application.
Back to 2016 Annual Meeting
|
