Baveno VI criteria validation and development of a new non-invasive strategy for varices rule-out in patients with HCC (SAPHE-R: Stiffness, ALBI, Platelets, HEpatocellular carcinoma Rule)

Background and Aims: Accuracy of Baveno criteria for avoiding endoscopy in patients with hepatocellular carcinoma (HCC) is uncertain, and systematic upper gastrointestinal endoscopy (UGE) may not be necessary. We evaluated Baveno VI criteria performance in this population and developed an alternative non-invasive rule to exclude esophagogastric varices of any grade, rather than high-risk varices (HRV). Methods: Patients with BCLC 0-C HCC and compensated advanced chronic liver disease (cACLD) from Italian Liver Cancer (ITA.LI.CA) database, with available liver stiffness measurement (LSM), UGE, and blood tests within six months of diagnosis, were included. Independent predictors of varices identified by logistic regression were combined through grid-search to derive a rule-out strategy (SAPHE-R), validated by bootstrap-resampling and in an independent cohort (Vienna). Results: The derivation cohort comprised 598 patients (age, 69 years; male sex, 75.3%; viral etiology, 69.3%; BCLC 0/A, 77.9%; ALBI 1, 54.2%; varices, 33.4%; HRV, 11.2%). Baveno VI criteria missed 8.5% of varices and 6.0% of HRV, sparing 31.1% of UGE. Independent predictors of varices were incorporated into SAPHE-R (LSM≤16 kPa, platelet count ≥160×10⁹/L, ALBI grade 1), which missed 4.0% of varices (HRV 3.0%, spared UGE 17.6%). In the validation cohort (n=149), SAPHE-R missed 4.8% of varices (HRV 3.1%, spared UGE 14.8%). Results remained consistent in sensitivity analyses. In an exploratory analysis, no HVPG-defined CSPH was observed among SAPHE-R rule-out patients (0/22). Conclusions: Baveno VI criteria showed suboptimal performance in patients with HCC and cACLD, whereas SAPHE-R provided a reliable non-invasive strategy to rule out varices, facilitating variceal risk assessment in patients with HCC.