Abstract: Objective The aim of this study was to explore an non-invasive indicators for the prediction of gastroesophageal varices (GOV) in patients with primary biliary cholangitis (PBC). Methods A total of 67 patients with PBC were encountered in our hospital between January 2016 and December 2020, and all patients underwent esophagogastroduodenoscopy and ultrasonography. 32 patients received liver biopsies (LB) and 33 received liver stiffness measurement (LSM). The platelet count/spleen length ratio (PSR) were calculated. The prediction was carried out based on the area under the receiver operating characteristic curve (AUC). Results Out of the 67 patients with PBC in our series, 27 patients had GOV (mild in 7, moderate in 7 and severe in 13 cases) and 40 hadn’t; in 32 patients underwent LB, the liver fibrosis S_3-4 was found in 9 cases, S_1-2 was found in 23, and the GOV were found 3 cases in the former and in the latter (Fisher’s, P=0.203); in 33 patients with LSM results, the sensitivity (Se),specificity (Sp) and accuracy (Ac) were 100.0%, 54.2% and 66.7% in predicting GOV occurrence when the LSM＞20 kPa was set as the cut-off-value; the multivariate Logistic analysis showed the PSR was an independent risk factor for GOV happening; the AUC was 0.932, with the Se, Sp and Ac of 92.6%, 87.5% and 89.6% in predicting GOV occurrence when the PSR<0.734×10⁹/L/mm was set as the cut-off-value, and preventing 92.5% of patients from EGD examination. Conclusion The predictive efficacy of our established PSR is promising in predicting GOV in patients with PBC, which might be validated in clinical practice.

Key words: Primary biliary cholangitis, Gastroesophageal varices, Platelet counts/spleen length ratio, Non-invasive diagnosis