Prediction of the varices needing treatment with non-invasive tests in patients with compensated advanced chronic liver disease

Backgrounds & Aims The Baveno VI guidelines proposed criteria including liver stiffness (LS) and platelet count to avoid screening endoscopy in patients with compensated advanced chronic liver disease (cACLD). This study was performed to validate the Baveno IV criteria and to compare its diagnostic accuracy with other non-invasive models. Methods Patients with cACLD who underwent laboratory tests, upper gastrointestinal endoscopy and abdominal ultrasound within 6 months of transient elastography were included. Results A total of 1218 patients with cACLD were included. VNT occurred in 249 patients (20.4%). With the Baveno VI criteria, the VNT miss rate was 1.9% with a 25.7% saved endoscopy rate. Using two criteria of LS 110 x 10(9) cells/L or LS 120 x 10(9) cells/L, the saved endoscopy rate was 39.1% while maintaining the VNT miss rate <5%. The optimal LS and platelet count-based criteria for predicting VNT differed according to the underlying liver disease. The area under the receiver operating characteristic curve of LS-spleen diameter to platelet score (LSPS) was 0.780 (95% confidence interval: 0.774-0.820), which was significantly higher than other models. The optimal cut-off value of the LSPS for predicting VNT was 1.47. Conclusion Liver stiffness and platelet count-based criteria are useful for discriminating patients with very low risk of having VNT among patients with cACLD and are partly affected by the type of underlying liver disease. Conversely, the LSPS is a predictor of VNT in patients with cACLD regardless of the type of underlying liver disease.