Wenlei Yang1, Fangfang Liu, Ruiping Xu, Wei Yang, Yu He, Zhen Liu, Fuyou Zhou, Fanxiu Heng, Bolin Hou, Lixin Zhang, Lei Chen, Fan Zhang, Fen Cai, Huawen Xu, Miaoping Lin, Mengfei Liu, Yaqi Pan, Ying Liu, Zhe Hu, Huanyu Chen, Zhonghu He, Yang Ke. 1. Key laboratory of Carcinogenesis and Translational Research (Ministry of Education/Beijing), Laboratory of Genetics, Peking University Cancer Hospital & Institute, Beijing, People's Republic of China Anyang Cancer Hospital, Henan Province, People's Republic of China Cancer Hospital of Shantou University Medical College, Guangdong Province, People's Republic of China Department of Non-communicable Disease Epidemiology, London School of Hygiene & Tropical Medicine, London, UK Key laboratory of Carcinogenesis and Translational Research (Ministry of Education/Beijing), Department of Information Technology Service, Peking University Cancer Hospital & Institute, Beijing, People's Republic of China Linkdoc AI Research (LAIR), Beijing, People's Republic of China.
Abstract
OBJECTIVE: To construct a prediction model for more precise evaluation of prognosis which will allow personalized treatment recommendations for adjuvant therapy in patients following resection of esophageal squamous cell carcinoma (ESCC). BACKGROUND: Marked heterogeneity of patient prognosis and limited evidence regarding survival benefit of various adjuvant therapy regimens pose challenges in the clinical treatment of ESCC. METHODS: Based on comprehensive clinical data obtained from 4,129 consecutive patients with resected ESCC in a high-risk region in China, we identified predictors for overall survival (OS) through a two-phase selection based on Cox proportional hazard regression and minimization of Akaike information criterion. The model was internally validated using bootstrapping and externally validated in 1,815 patients from a non-high-risk region in China. RESULTS: The final model incorporates nine variables: age, sex, primary site, T stage, N stage, number of lymph nodes harvested, tumor size, adjuvant treatment, and hemoglobin level. A significant interaction was also observed between N stage and adjuvant treatment. N1+ stage patients were likely to benefit from addition of adjuvant therapy as opposed to surgery alone, but adjuvant therapy did not improve OS for N0 stage patients. The C-index of the model was 0.729 in the training cohort, 0.723 after bootstrapping, and 0.695 in the external validation cohort. This model outperformed the seventh edition American Joint Committee on Cancer staging system in prognostic prediction and risk stratification. CONCLUSION: The prediction model constructed in this study may facilitate precise prediction of survival and inform decision-making about adjuvant therapy according to N stage.
OBJECTIVE: To construct a prediction model for more precise evaluation of prognosis which will allow personalized treatment recommendations for adjuvant therapy in patients following resection of esophageal squamous cell carcinoma (ESCC). BACKGROUND: Marked heterogeneity of patient prognosis and limited evidence regarding survival benefit of various adjuvant therapy regimens pose challenges in the clinical treatment of ESCC. METHODS: Based on comprehensive clinical data obtained from 4,129 consecutive patients with resected ESCC in a high-risk region in China, we identified predictors for overall survival (OS) through a two-phase selection based on Cox proportional hazard regression and minimization of Akaike information criterion. The model was internally validated using bootstrapping and externally validated in 1,815 patients from a non-high-risk region in China. RESULTS: The final model incorporates nine variables: age, sex, primary site, T stage, N stage, number of lymph nodes harvested, tumor size, adjuvant treatment, and hemoglobin level. A significant interaction was also observed between N stage and adjuvant treatment. N1+ stage patients were likely to benefit from addition of adjuvant therapy as opposed to surgery alone, but adjuvant therapy did not improve OS for N0 stage patients. The C-index of the model was 0.729 in the training cohort, 0.723 after bootstrapping, and 0.695 in the external validation cohort. This model outperformed the seventh edition American Joint Committee on Cancer staging system in prognostic prediction and risk stratification. CONCLUSION: The prediction model constructed in this study may facilitate precise prediction of survival and inform decision-making about adjuvant therapy according to N stage.