Noninvasive diagnostic value of indocyanine green retention test in patients with esophagogastric varices in liver cirrhosis

Objective 

This study aimed to evaluate the value of the indocyanine green retention test (ICG-R15) as a noninvasive predictor of gastroesophageal varices (GOV) in cirrhosis.

Methods 

A total of 629 patients with cirrhosis who were admitted to Mengchao Hepatobiliary Hospital of Fujian Medical University between June 2016 and September 2020 (450 in the trial group and 179 in the validation group) were selected. The trial group was divided into a GOV group and a non-GOV group based on gastroscopic findings. The two group’s ICG results, clinical hematological, biochemical, and ultrasound parameters were compared. A predictive model for the noninvasive diagnosis of GOV was developed by applying binary logistic regression analysis and compared with previous common noninvasive diagnostic methods, followed by model validation in a validation group.

Results 

ICG-R15 is positively correlated with the degree of varicose veins. Univariate and multifactorial logistic regression analyses showed that PLT, PT, ICG-R15, splenic length, splenic vein diameter, and the presence of ascites were independent predictors of GOV. Predictive equation 1 for GOV was constructed based on the above factors: Logistic (Y1) = 1.327 × PLT + 0.691 × PT + 1.110 × ICG-R15 + 0.581 × SVD + 0.815 × SD + 1.130 × presence of ascites − 2.133, which predicted GOV in cirrhosis with an AUC of 0.861, sensitivity 84.5%, and specificity 74.1%. As the internal diameter of the splenic vein and the long diameter of the spleen are influenced by subjective factors and data are not easily available, we removed these two factors and constructed the prediction equation logistic (Y2) = 1.516 × PLT + 0.856 × PT + 1.135 × R15 + 1.297 × presence of ascites − 1.513. AUC for predicting GOV was also high at 0.850, with a sensitivity of 75.3% and a specificity of 83.2%. Similar results were obtained in the validation group. However, the previous GOV prediction models ARPI, AAR, FIB-4, Lok, FI, and King had AUCs of 0.731, 0.643, 0.781, 0.763, 0.735, and 0.730, respectively, in our trial group, with only low-to-moderate diagnostic efficacy.

Conclusion 

ICG-R15 correlates with the severity of GOV in cirrhosis, and the models based on it can further improve the noninvasive diagnostic efficacy of GOV.

留言 (0)

沒有登入
gif