Clinical and pathological characteristics of primary duodenal adenocarcinoma and the construction of predictive models in China and SEER database.

e16472 Background: Primary duodenal adenocarcinoma (DA) is a rare tumor with a low incidence rate, high mortality rate, and lacks effective treatment methods and sensitive biomarkers for predicting prognosis. We aim to analyze the differences in clinical characteristics between Chinese and Western populations with DA, and to construct a clinical model to better predict the prognosis of patients with DA. Methods: 500 DA patients from SEER database and 232 DA patients from Renji Hospital, Shanghai Jiaotong University school were enrolled. A nomogram was constructed according to least absolute shrinkage and selection operator regression analysis. COX regression was analyzed by using R language. Results: We found that Chinese cohort presented a big difference with western population. The patients in Renji-cohort were of early stage when compared with SEER database. Multivariate analysis implied that age, gender and tumor TNM staging were significantly correlated with the prognosis of DA patients in the SEER database cohort. Tumor differentiation and N stage were significantly correlated with the prognosis in Renji-cohort. Nomogram concluded pathological and molecular markers was constructed and the AUC value for the training set was 0.776, and it was as high as 0.816 for the validation set, significantly higher than the model only conclude the TNM stage. We also studied the correlation between lymph node dissection and prognosis. The results showed that positive lymph nodes are related to prognosis, but the number of lymph nodes dissected is not related to prognosis. However, it seems that if the stage is early, the prognosis might be better without lymph node dissection, which still requires validation with a larger sample size. Conclusions: The established nomogram model provided visualization of the risk of DA prognostic factors.