Risk factors and construction of prediction model of perineural invasion of gallbladder carcinoma based on enhanced CT- image features / 西安交通大学学报(医学版)

ABSTRACT

【Objective】 To construct the prediction model of perineural invasion (PNI) in gallbladder carcinoma (GBC) based on preoperative enhanced CT image features and evaluate its prediction efficiency. 【Methods】 The clinical, imaging and pathological data of 180 GBC patients undergoing radical operation were retrospectively analyzed. They were divided into positive and negative groups according to the presence or absence of PNI. Preoperative enhanced CT imaging features (including presence of gallstones, imaging hepatic invasion, vascular invasion, T-stage, and hilar or retroperitoneal lymph node metastases) were evaluated by two radiologists. Independent sample t-test, Mann Whitney U test, and χ2 test were used to compare the correlation between CT signs and PNI. Logistics regression analysis was used to screen independent risk factors and establish the prediction model formula. ROC curve was used to evaluate the prediction efficiency of the prediction model and the corresponding area under the curve (AUC) was calculated. Hosmer-Lemeshow goodness of fit test was used to verify the prediction model. 【Results】 Unifactorial analysis showed that CA199, CA125, imaging hepatic invasion, vascular invasion (hepatic artery or portal vein), T-stage, and hilar or retroperitoneal lymph node metastasis were correlated with nerve invasion (P<0.05). Logistics multi-factor analysis results showed that CA199, imaging vascular invasion (hepatic artery or portal vein), and imaging T stage were independent risk factors for PNI. Based on the above independent risk factors, a prediction model formula was established and ROC curve was drawn, with an AUC of 0.807 (95% CI: 0.734~0.879), sensitivity of 0.792, specificity of 0.697, and the chi-square value of Hosmer-Lemeshow goodness of fit test of 0.594 (P=0.997), indicating that the predicted value of the model was close to the actual value. 【Conclusion】 Combining CA199, imaging vascular invasion, T-stage, and other preoperative clinically-enhanced CT features to establish a prediction model can effectively predict postoperative PNI of GBC.