DIAGNOSTIC PERFORMANCE OF PLATELET COUNT, SPLEEN SIZE, PLATELET-TO-SPLEEN RATIO, PLATELET-ALBUMIN BILIRUBIN SCORE, CHILD-TURCOTTE-PUGH SCORE, AND MODEL FOR END-STAGE LIVER DISEASE WITH SODIUM FOR PREDICTION OF ESOPHAGEAL VARICES IN CHRONIC LIVER DISEA

Abstract

Background: Esophageal varices are a major complication of portal hypertension in patients with chronic liver disease. Upper gastrointestinal endoscopy remains the reference method for detection and grading of esophageal varices. Several non-invasive parameters, including platelet count, spleen size, platelet-to-spleen ratio, platelet-albumin-bilirubin score, Child-Turcotte-Pugh score, and Model for End-Stage Liver Disease with sodium, have been evaluated as potential predictors. Objective: To evaluate the diagnostic performance of platelet count, spleen size, platelet-to-spleen ratio, platelet-albumin-bilirubin score, Child-Turcotte-Pugh score, and Model for End-Stage Liver Disease with sodium for predicting esophageal varices and large esophageal varices in patients with chronic liver disease. Methods: This retrospective observational study included 104 patients with chronic liver disease who underwent upper gastrointestinal endoscopy and had available data for platelet count, spleen size, bilirubin, albumin, Child-Turcotte-Pugh score, and Model for End-Stage Liver Disease with sodium. Esophageal varices were classified from the documented endoscopic findings as absent, small, or large. Platelet-to-spleen ratio was calculated as platelet count divided by spleen diameter. Platelet-albumin-bilirubin score was calculated using the published formula. Receiver operating characteristic analysis was performed for prediction of any esophageal varices and large esophageal varices. Results: Among 104 patients, 83 (79.8%) had esophageal varices and 21 (20.2%) had no documented esophageal varices. Large esophageal varices were documented in 26 (25.0%) patients. For prediction of any esophageal varices, the areas under the receiver operating characteristic curve were 0.509 for platelet count, 0.435 for spleen size, 0.495 for platelet-to-spleen ratio, 0.521 for platelet-albumin-bilirubin score, 0.513 for Child-Turcotte-Pugh score, and 0.558 for Model for End-Stage Liver Disease with sodium. For large esophageal varices, the corresponding areas under the receiver operating characteristic curve were 0.614, 0.522, 0.616, 0.390, 0.458, and 0.354, respectively. None of the individual parameters demonstrated robust discrimination for the presence of esophageal varices. Platelet count and platelet-to-spleen ratio showed modest discrimination for large varices, but their confidence intervals were broad. Conclusion: In this retrospective cohort, commonly available laboratory, imaging, and liver disease severity parameters demonstrated limited ability to predict endoscopically detected esophageal varices. Platelet count and platelet-to-spleen ratio showed some discriminatory signal for large varices, but their performance was insufficient to support replacement of endoscopic assessment. Prospective validation of composite non-invasive models incorporating liver and spleen stiffness is warranted.