Noninvasive Prediction of Clinically Significant Portal Hypertension in Hepatocellular Carcinoma: A Tumor-Adjusted Model.

Abstract

Noninvasive assessment of clinically significant portal hypertension (CSPH) using liver stiffness measurement (LSM) and platelet count (per Baveno VII criteria) may be influenced by tumor-related factors in patients with hepatocellular carcinoma (HCC). We evaluated the relationship between LSM and hepatic venous pressure gradient (HVPG) and developed a tumor-adjusted model for CSPH prediction in HCC. In this multicenter study, 452 patients with compensated cirrhosis and HCC who underwent LSM and HVPG within three months were retrospectively analyzed. The correlation between LSM and HVPG was assessed, including stratification by tumor characteristics. A tumor-adjusted predictive model (Baveno-HCC) was developed in a derivation cohort (n=316) and validated in a separate cohort (n=136). LSM correlated moderately with HVPG (r=0.593, p<0.001), with reduced accuracy in patients with large (>5 cm) right-lobe tumors. Baveno VII criteria maintained high sensitivity and negative predictive value (>90%) for ruling out CSPH but showed limited rule-in performance, particularly in patients with larger tumors. The Baveno-HCC model, incorporating LSM, platelet count, spleen size, and tumor burden, improved rule-in performance with higher positive predictive value (PPV) while preserving rule-out accuracy. Improvements were most pronounced in patients with tumor burden >5 cm, with PPV increasing from 46.2% to 84.6% in the derivation cohort and from 56.3% to 100.0% in the validation cohort. Tumor burden and location affect LSM-based CSPH assessment. The proposed Baveno-HCC model improves CSPH prediction and supports more accurate risk stratification and clinical decision-making in patients with HCC.