Risk assessment of esophageal variceal bleeding in B-viral liver cirrhosis by a liver stiffness measurement-based model

Abstract

Objectives: Periodic endoscopic screening for esophageal varices (EVs) and prophylactic treatment for high-risk EVs (HEVs; (i) medium/large EVs and (ii) small EVs with red sign or decompensated cirrhosis) are recommended for cirrhotic patients. We assessed cumulative risks of future EV bleeding (EVB) using the liver stiffness measurement (LSM)-based model, LSM-spleen diameter to platelet ratio score (LSPS=LSM×spleen diameter/platelet count).

Methods: We prospectively enrolled 577 consecutive B-viral cirrhosis patients from 2005 to 2009, none of whom experienced EVB. All underwent laboratory workups, endoscopy, LSM, and ultrasonography. Those with HEVs took nonselective β-blockers as prophylaxis for EVB after diagnosis, if not contraindicated. The major end point was the first EVB event, examined using Kaplan-Meier and Cox-regression methods.

Results: Among whole population, 95.9% negative- /93.5% positive-predictive value by LSPS<3.5/LSPS≥5.5 were provided for predicting the presence of HEV at enrollment, respectively. Among patients with HEV (n=150), 25 experienced their first EVBs during follow-up (median, 29 months). To differentiate EVB risk, we divided them into subgroup 1 (LSPS<6.5) and 2 (LSPS≥6.5) according to LSPS 6.5, a point with maximum sum of sensitivity and specificity from time-dependent receiver-operating characteristic (ROC) curves (area under ROC curve=0.929). EVB risk was higher in subgroup 2 than subgroup 1 (P<0.001). Multivariate analysis found higher LSPS (P=0.003) a significant predictor, alongside large variceal sizes (P=0.004) and Child-Pugh classifications B/C (P=0.001). Notably, EVB risk of subgroup 1 was as low as that of low-risk EVs (P=0.507).

Conclusions: LSPS is a reliable predictor for EVB risk. According to risk stratification, different prophylactic treatments should be considered for subgroups with LSPS≥6.5.