Combination of non-endoscopic parameters as predictors of large esophageal varices.

BACKGROUND AND AIM: Bleeding from gastroesophageal varices is a life-threatening complication of cirrhosis. Prophylactic therapy for large varices considerably reduces such bleeding. Current guidelines recommend upper GI endoscopy as the gold standard for detecting high-grade varices. However, in peripheral centers, this is often unfeasible or inaccessible thus requiring non-endoscopic predictors of large oesophageal varices.

METHODS

This prospective study involved 100 newly diagnosed cirrhosis patients without any history of variceal bleeding. Demographic, clinical, bio-chemical and ultrasonographic parameters were recorded. Esophageal varices were classified endoscopically as small and large. Multivariate logistic regression was performed to identify predictors of variceal size and presence. Cut-offs of significant indicators were determined by ROC analysis.

RESULTS

69% patients had small and 31% had large varices. Univariate analysis revealed significant differences between the grade of spleen, blood parameters, platelet count, international normalized ratio, serum albumin, spleen size, portal vein diameter (PVD) and platelet count to spleen diameter ratio in the two groups. Multivariate analysis revealed that INR [OR: 2.432(95% CI: 1.192 - 4.958)], splenomegaly at USG [OR: 2.138 (95% CI: 0.662 - 6.911)] and PVD [OR: 1.318 (95% CI: 0.937 - 1.853)] were the most significant predictors for large varices.

CONCLUSIONS

Multivariate prediction of large varices based on a combination of non-endoscopic parameters can be utilized in place of single parameter based predictions.