Non-invasive predictors of the presence of large oesophageal varices in patients with cirrhosis

Dig Liver Dis. 2003 Jul;35(7):473-8. doi: 10.1016/s1590-8658(03)00219-6.


Background/aims: The usual clinical practice is to screen all patients with established cirrhosis at the time of diagnosis by upper endoscopy for the presence of varices. Patients with large varices should be treated with non-selective beta blockers to reduce the incidence of first variceal bleeding. However, fewer than 50% of cirrhotic patients have varices at screening endoscopy and most have small sized varices, with a low risk of bleeding. The aim of the present study was to determine whether clinical or laboratory non-endoscopic parameters could predict the presence of large oesophageal varices.

Patients/methods: Seventeen variables considered relevant to the prevalence of oesophageal varices were tested in 184 patients with cirrhosis, who underwent screening endoscopy. Small varices were regarded as those which flatten with insufflation or slightly protrude into the lumen, while large varices are those which protrude into the lumen or touch each other. None of the patients was on beta blockers or other vasoactive drugs or had a history of variceal bleeding.

Results: Oesophageal varices were present in 92 patients (50%), and large varices in 33 patients (17.9%). Variables associated with the presence of large oesophageal varices on univariate analysis were the presence of ascites and splenomegaly either by clinical examination or by ultrasound (p < 0.01), the presence of spiders (p = 0.02), platelet count (p < 0.0001), and bilirubin (p = 0.01). Factors independently associated with the presence of large oesophageal varices on multivariate analysis were platelet count, size of spleen and presence of ascites by ultrasound. Using mean values as cut-off points, it is noteworthy that only five out of 39 patients (12.8%) with platelets > or = 18(x 10(9)/l), spleen length < or = 135 mm and no ascites had varices. Moreover, all these patients had small sized varices. On the other hand, 15 out of 18 patients (83.3%) with a platelet count < 118 x 10(9)/l, spleen length > 135 mm and ascites had varices. Moreover, five out of those 18 patients had large varices (28.3%).

Conclusion: Thrombocytopenia, splenomegaly and ascites are independent predictors of large oesophageal varices in cirrhotic patients. We suggest that endoscopy could be avoided safely in cirrhotic patients with none of these predictive factors, as large varices are absent in this group of patients.

MeSH terms

  • Ascites / diagnosis*
  • Esophageal and Gastric Varices / blood
  • Esophageal and Gastric Varices / diagnosis*
  • Esophageal and Gastric Varices / etiology
  • Esophagoscopy
  • Female
  • Humans
  • Liver Cirrhosis / complications*
  • Male
  • Middle Aged
  • Multivariate Analysis
  • Platelet Count
  • Predictive Value of Tests
  • Splenomegaly / diagnosis*
  • Thrombocytopenia / diagnosis*