To explore the rationality of clinical classification of gallbladder carcinoma and its relationship with clinical prognosis.

Clinical data of 486 patients with gallbladder carcinoma who underwent radical resection in the First Affiliated Hospital of Xi'an Jiaotong University from January 1, 2011 to December 31, 2020 were retrospectively analyzed. The informed consents of all patients were obtained and the local ethical committee approval was received. Among them, 164 patients were male and 322 female, aged from 30 to 88 years, with a median age of 63 years. Clinical classification of gallbladder carcinoma: 234 cases were classified as abdominal type, 138 as hepatic type, 49 as hilar type and 65 as mixed type. The relationship between clinical classification and clinicopathological parameters was analyzed by rank-sum test and Chi-square test. The relationship between clinical classification and prognosis was assessed by Kaplan-Meier method and Log-rank test. The Tree-Augmented Naive (TAN) Bayesian network model was established for the survival prediction of gallbladder carcinoma, and the prediction efficiency of this model was assessed by the area under ROC curve (AUC).

Clinical classification of gallbladder carcinoma patients was associated with patients' age, preoperative TB, CEA, CA19-9 and CA125 levels (H=20.45, 56.90, 24.10, 55.62, 31.19; P<0.05) and the pathological classification, histological differentiation, vascular invasion, nerve infiltration, T stage and N stage (χ²=11.50, 15.44, 119.69, 38.95, 255.57, 25.34; P<0.05). The median survival of patients with abdominal, hepatic, hilar and mixed type gallbladder carcinoma was 44.7, 18.5, 17.0 and 10.0 months, respectively. Clinical classification was significantly related with clinical prognosis (χ²=71.14, P<0.05). TAN network survival prediction model showed that clinical classification of gallbladder carcinoma was associated with T stage, invasion of surrounding tissues and organs, CA19-9 and TB levels. ROC curve analysis revealed that the AUC of this model for predicting the 1-, 2-, 3-, 5-year survival of gallbladder carcinoma patients was 0.83, 0.78, 0.70, 0.69, and the accuracy was 0.76, 0.66, 0.73, 0.84, respectively.

Clinical classification of gallbladder carcinoma is correlated with the clinicopathological features and clinical prognosis of patients. TAN network survival prediction model based on clinical classification can effectively predict the prognosis. The data from our center can validate the rationality of this clinical classification.