Spleen and Liver Stiffness Is Positively Correlated with the Risk of Esophageal Variceal Bleeding

Abstract

Background/aims: Portal hypertension (PH) is a common complication of chronic liver disease and results in esophageal and gastric variceal bleeding, which is associated with a high mortality rate. Measurement of the hepatic venous pressure gradient (HVPG) is considered the gold standard for diagnosing PH and estimating the risk of varices and bleeding. In contrast, upper gastrointestinal (GI) endoscopy (UGE) can reliably demonstrate the presence of varices and bleeding. Both measures are invasive, and HVPG is mainly restricted to tertiary centers. Therefore, the development of noninvasive methods of assessing the severity of PH and the risk of variceal bleeding is warranted.

Methods: We retrospectively examined the correlation of spleen stiffness (SSM) and liver stiffness measurements (LSM) with the incidence of variceal bleeding among 143 patients who underwent combined liver and spleen elastography between 2013 and 2015.

Results: For 19 of 103 patients (16.8%), upper GI variceal bleeding was diagnosed and treated endoscopically. The median SSM of all patients was 35.3 kilopascals (kPa); the median LSM, 11.7 kPa. Patients with previous bleeding episodes had significantly higher SSM (75.0 kPa) and LSM (37.3 kPa) than those without a history of bleeding (SSM, 30.6 kPa; LSM, 8.2 kPa; p < 0.0001). Seventy-five patients (66.4%) underwent UGE in addition to SSM and LSM: 25 with no esophageal varices (EVs; SSM, 29.5 kPa; LSM, 11.4 kPa), 16 with EV grade 1 (SSM, 35.9 kPa; LSM, 33.4 kPa), 21 with EV grade 2 (SSM, 67.8 kPa; LSM, 27.0 kPa) and 13 with EV grade 3 (SSM, 75.0 kPa; LSM, 26.3 kPa). No statistically significant differences were found between respective grades of EV but were found between the presence and absence of varices. At a calculated cutoff level of 42.6 kPa (with application of 95% CI), SSM had sensitivity of 89% and specificity of 64% in determining the risk of bleeding, with a negative predictive value (NPV) of 0.97 (LSM sensitivity, 84%; LSM specificity, 80%; LSM NPV, 0.96 at LSM cutoff level of 20.8 kPa). When LSM (cutoff level, 20.8 kPa) and SSM (cutoff level, 42.6 kPa) were combined, the NPV was 1 (sensitivity, 100%; specificity, 55%).

Conclusion: SSM and LSM as determined by FibroScan (a noninvasive method of detecting PH) is positively correlated with upper GI variceal bleeding (optimal SSM cutoff level, 42.6 kPa; optimal LSM cutoff level, 20.8 kPa). No patients with both SSM and LSM below cutoff levels had a history of bleeding complications.