Validation of noninvasive methods to predict the presence of gastroesophageal varices in a cohort of patients with compensated advanced chronic liver disease

Background and Aim: The aim was to validate noninvasive methods to predict the presence of gastroesophageal varices (GEV) in patients with suspected compensated advanced chronic liver disease. Methods: We retrospectively reviewed clinical and radiological data collected prospectively between September 2013 and September 2015. We reviewed 442 consecutive patients with suspected compensated advanced chronic liver disease measured by transient elastography (TE) and a gastroscopy. We evaluated platelets, spleen diameter, TE, liver stiffness × spleen size/platelets (LSPS), variceal risk index (VRI), Baveno VI strategy, and Augustin algorithm. Results: One hundred sixty-one out of 442 patients were included. Patients with GEV were compared with patients without GEV and showed statistically significant differences in platelet count (117 SD 51 vs 149 SD 62; P = 0.02), spleen diameter (13.0 SD 1.9 vs 11.5 SD 2; P = 0.003), and TE (28 SD 15 vs 19 SD 10; P = 0.001). Single methods (platelet count and TE) diagnosed correctly 51% and 71.4% of patients. Combined methods (LSPS, VRI, Baveno VI, and Augustin algorithm) diagnosed correctly 78%, 83.6%, 45.3%, and 57.1% of patients. Patients with GEV misdiagnosed: platelets 5/161 (3.1%), TE 6/161 (3.7%), LSPS 16/159 (10%), VRI 18/159 (11.3%), Baveno VI 3/161 (1.8%), and Augustin algorithm 6/161 (3.7%). Rate of unnecessary gastroscopies: platelets 46%, TE 25%, LSPS 13%, VRI 6%, Baveno VI 53%, and Augustin algorithm 39.1%. Conclusions: A significant number of patients were classified correctly using TE, LSPS, and VRI; however, LSPS and VRI had unacceptable rates of misdiagnoses. TE is the best noninvasive single method and the Baveno VI strategy the best combined method.