[A dual-center study of enhanced CT-based nomograms predicts spontaneous portosystemic shunts on initial hemorrhage from rupture of esophageal gastric varices in cirrhosis].

Objective: To explore the value of an enhanced CT-based nomogram for predicting spontaneous portosystemic shunts (SPSS) on initial hemorrhage from rupture of esophageal gastric varices in cirrhosis. Methods: Clinical and imaging data from 212 cases with esophageal gastric varices in cirrhosis were retrospectively collected from two hospitals. The 166 cases collected from one hospital were used as the training group (87 patients in the hemorrhagic group and 79 patients in the non-hemorrhagic group), while 46 patients collected from the other hospital were used as the external validation group (18 patients in the hemorrhagic group and 28 patients in the non-hemorrhagic group). Logistic regression analysis was used to identify variables for constructing the optimal predictive model. The nomogram was used to illustrate the model. The training set was internally validated by the Bootstrap method. The area under the receiver operating characteristic curve (AUC) was used to evaluate the discrimination of the model. The accuracy and clinical benefit of the model were assessed using standard curves and clinical decision curves. Statistical analysis was performed using the independent samples t-test, Mann-Whitney U test, χ2 test, or Fisher's exact probability method according to the different data. Results: There were statistically significant differences in Child-Pugh grade, model for end-stage liver disease score, albumin, direct bilirubin, total bilirubin, prothrombin time, alanine aminotransferase, international normalized ratio, white blood cell count, hemoglobin, inferior vena cava diameter, portal vein diameter, splenic vein diameter, SPSS diameter, SPSS volume, and portal vein thrombosis between the two training groups (P<0.05). Univariate logistic regression analysis revealed statistically significant differences with statistical significance between the two groups for some indicators (P<0.05). Multivariate logistic regression analysis showed that SPSS volume, portal vein diameter, portal vein thrombosis, prothrombin time and hemoglobin had the optimal prediction model for establishment of the initial hemorrhage from rupture of esophagogastric variceal in cirrhosis. The AUCs for the training, internal validation, and external validation groups were 0.822, 0.806, and 0.802, respectively. The calibration curve demonstrated good consistency. The clinical decision curve indicated that the net benefit of the prediction model was significantly higher than all positive or negative predictions within a larger threshold range. Conclusion: An enhanced CT-based nomogram can accurately predict SPSS on the initial hemorrhage from rupture of esophagogastric varices in cirrhosis.