Endoscopic screening for esophageal varices is recommended in patients with advanced chronic liver disease, but many patients have no or only low-risk varices. Simple non-invasive markers could help to better select candidates for endoscopy. We aimed to assess the prevalence of esophageal varices and to evaluate the performance of liver stiffness measurement (LSM), platelet count, FIB-4 and APRI for predicting esophageal varices and high-risk varices in chronic hepatitis B (CHB).

We conducted a retrospective study including adult CHB patients who underwent both FibroScan and upper gastrointestinal endoscopy at F.Hached University Hospital. LSM (kPa), platelet count and routine laboratory parameters were collected. FIB-4 and APRI were calculated. Esophageal varices were graded (1–3). High-risk varices were defined as grade ≥2 and/or the presence of red wale signs. Advanced fibrosis was defined as F3–F4 by LSM. Receiver operating characteristic (ROC) curves were used to evaluate the ability of LSM, platelets, FIB-4 and APRI to predict any varices and high-risk varices.

Eighty CHB patients were analyzed. Esophageal varices of any grade were found in 18/80 (22.5%) patients, and high-risk varices in 5/80 (6.3%), all in patients with advanced fibrosis. Compared with patients without varices, those with varices had lower platelet counts (mean 103 vs 180×10⁹/L, p<0.001), higher FIB-4 scores (5.38 vs 1.93, p<0.001) and higher APRI values (1.52 vs 0.53, p<0.001). LSM tended to be higher in patients with varices (16.5 vs 9.9 kPa, p=0.10).

For the prediction of any varices, the area under the ROC curve (AUC) was 0.73 for LSM, 0.87 for FIB-4, 0.83 for APRI and 0.84 for platelet count. Optimal cut-offs for detecting any varices were approximately 4.9 for FIB-4 (sensitivity 68.8%, specificity 96.4%), 0.65 for APRI (sensitivity 87.5%, specificity 74.5%) and 7.3 kPa for LSM (sensitivity 87.5%, specificity 67.3%).

For high-risk varices, discrimination was very good, with AUCs of 0.96 for FIB-4, 1.00 for APRI and 0.75 for LSM, although based on a small number of events (n=5). In this setting, thresholds around 7.0 for FIB-4, 2.7 for APRI and 12 kPa for LSM all achieved 100% sensitivity and high specificity.

In this cohort of CHB patients, esophageal varices were present in about one quarter and high-risk varices in a small but clinically relevant proportion. Non-invasive markers, particularly FIB-4, APRI and platelet count, showed good to excellent performance for predicting esophageal varices and high-risk varices, and may help refine the selection of patients requiring screening endoscopy.