Background: Many studies have been performed of various non-invasive liver fibrosis tests (NIFTs) which aim to predict degrees of liver fibrosis and the presence of esophageal varices in patients with cirrhosis. However, the use of NIFTs to predict variceal bleeding (VB) in the setting of upper gastrointestinal bleeding (UGIB) remains unexplored.
Objective: To evaluate the performance of NIFTs in predicting VB in patients with UGIB.
Materials and Methods: The authors prospectively enrolled consecutive patients who presented with UGIB and underwent
esophagogastroduodenoscopy between June 2018 and August 2019 at Rajavithi Hospital, Bangkok. Baseline clinical/lab
characteristics and NIFTs-scoring systems were evaluated including APRI, AAR, FIB-4, Fibrosis Index, Lok Index, GUCI, and King’s score.
Results: A total of 215 patients with UGIB were included. Their mean age was 56.4 years, their mean Glasgow-Blatchford score was 9.8, and 39.5% of them had VB. In overall analysis, the AUCs of NIFTs for predicting VB ranged between 0.686 and 0.867. GUCI and APRI (both at the cut-off of 0.5) showed the best performance in predicting VB with sensitivity of 95.3% and 90.6% and specificity of 73.1% and 75.4% respectively.
Conclusion: GUCI and APRI scores displayed good performance in predicting VB in patients presenting with UGIB regardless of
known cirrhosis status. They may therefore be helpful when selecting patients for prompt administration of vasoactive agents, antibiotics and urgent esophagogastroduodenoscopy.

Keywords: Biomarker, Cirrhosis, Endoscopy, Esophageal varices, Gastrointestinal hemorrhage, Liver fibrosis, Portal hypertension