LAUSR

BACKGROUND Gastric cardia adenocarcinoma (GCA) is a highly fatal form of cancer in humans. The aim of this study was to extract clinicopathological data of postoperative patients with GCA from the Surveillance, Epidemiology, and End Results database, analyze prognostic risk factors, and build a nomogram. METHODS In this study, the clinical information of 1448 patients with GCA who underwent radical surgery and were diagnosed between 2010 and 2015 was extracted from the SEER database. The patients were then randomly divided into training (n = 1013) and internal validation (n = 435) cohorts at a 7:3 ratio. The study also included an external validation cohort (n = 218) from a Chinese hospital. The study used the Cox and LASSO models to pinpoint the independent risk factors linked to GCA. The prognostic model was constructed according to the results of the multivariate regression analysis. To assess the predictive accuracy of the nomogram, four methods were used: C-index, calibration curve, time-dependent ROC curve, and DCA curve. Kaplan-Meier survival curves were also generated to illustrate the differences in cancer-specific survival (CSS) between the groups. RESULTS The results of the multivariate Cox regression analysis showed that age, grade, race, marital status, T stage, and log odds of positive lymph nodes (LODDS) were independently associated with cancer-specific survival in the training cohort. Both the C-index and AUC values depicted in the nomogram were greater than 0.71. The calibration curve revealed that the nomogram's CSS prediction was consistent with the actual outcomes. The decision curve analysis suggested moderately positive net benefits. Based on the nomogram risk score, significant differences in survival between the high- and low-risk groups were observed. CONCLUSIONS Race, age, marital status, differentiation grade, T stage, and LODDS are independent predictors of CSS in patients with GCA after radical surgery. Our predictive nomogram constructed based on these variables demonstrated good predictive ability.