A non-invasive model based on the virtual portal pressure gradient to predict the first variceal hemorrhage in cirrhotic patients.

BACKGROUND AND AIMS: This study aimed to establish a non-invasive model based on the virtual portal pressure gradient (vPPG) to predict the first variceal hemorrhage (VH) in patients with cirrhosis.

METHODS

This single-center study prospectively enrolled cirrhotic patients as the training and validation cohorts during different time periods. The PPG-detection software (PPGS 1.0) was used to perform vPPG calculation, which involves 2 steps including three-dimensional (3D) reconstruction of portal vein tree and subsequent application of computational fluid dynamics. All patients were given standard primary prophylaxis against VH and followed up for 2 years. Data from the training cohort were assessed using univariate and multivariate Cox regression and Kaplan-Meier analyses, by which a nomogram with its dynamic form was developed to estimate the probability of VH.

RESULTS

In the training cohort (n = 128), 37 (28.9%) experienced VH during 2-year follow-up. Four variables including vPPG ≥ 10.5 mmHg (p < 0.001), PLT < 56 × 109/L (p = 0.048), albumin < 32 g/L (p < 0.001) and INR ≥ 1.2 (p = 0.022) were identified as independent risk factors of VH, among which vPPG showed the best diagnostic performance (AUC 0.875). Subsequently, these predictors were incorporated into the nomogram, of which C-indexes were 0.891 and 0.926 for the training and validation cohorts, respectively. Calibration curves demonstrated a great calibration ability of the model. At the threshold probabilities of 0.1-0.6 (1 year) and 0.1-1.0 (2 years), this nomogram could offer more net benefits in decision curve analysis.

CONCLUSIONS

The vPPG-based nomogram could be used for risk stratification of the first VH in patients with cirrhosis.