Suboptimal performances of Baveno VII and AASLD 2024 criteria for detecting clinically significant portal hypertension with varices in untreated patients with HDV cirrhosis

Backgroundandaims:Presenceofgastroesophagealvarices (EV) on esophagogastroduodenoscopy (EGD) results from clinically significantportalhypertension(CSPH)andareassociatedwithanincreased risk of liver decompensation. Non-invasive tests (NITs) are increasingly used in clinical practice for assessing the risk for CSPH with EV, but disease-specific cut-offs for chronic hepatitis Delta (CHD) are missing. We investigated the performances of NITs-based criteria to predict EV in CHD patients. Method: Untreated patients with HDV-related compensated cirrhosis andavailableliver stiffness measurement (LSM)andEGDfromthe multicenter European SAVE-D and D-SHIELD cohorts were included. EGDandLSMwereperformedpriortoBulevirtide(BLV)start.Baveno VII (LSM ≥ 25 kPa) and AASLD (LSM >20 kPa + platelets [PLT] count <150×109/L) criteria were applied to rule in and rule out (LSM ≤15 kPa plus PLT 150×109/L) CSPH. In a subgroup of patients, spleen stiffness measurement (SSM)wasavailableforcombinedalgorithms: “rule in” if at least 2 of LSM >20kPa+PLT<150×109/L+SSM>50kPa; “rule out” at least 2 of LSM ≤ 15 kPa + PLT 150×109+SSM <21kPa. Results: Of the 660 compensated patients with cirrhosis included in the SAVE-D and D-SHIELD studies, 347 patients were included in this analysis 154 (44%) showed EV on EGD. 53 (62%) of 85 patients fulfilling Baveno VII “rule in” criteria had EV (62%), but 101 (66%) patients with EV had LSM <25 kPa. This criterion demonstrated a Specificity (Sp) of 83.4% and a Positive Predictive Value (PPV) of only 62.4%. 70 (65%) of the 108 patients fulfilling AASLD “rule in” criteria had EV, but 84 (55%) of patients with EV did not fulfill these criteria (Sp 80.3%, PPV 64.8%). 41/46 (89%) patients fulfilling the “rule out” criteria did not have EV on EGD, corresponding to Sensitivity (Sn) of 21.2% and Negative Predictive value (NPV) of 49.5%. 5 (11%) patients hadmissedvaricesaccording to“rule out” criteria. 216 (62%) and 185 (53%) patients fell in the “grey zone” according to Baveno VII and AASLD criteria, respectively, with a prevalence of varices of 44% and 41%, respectively. The addition of SSM (available in 74 patients)>50 kPatothe“rulein” criteria improved Sn (72.7%vs.45.5%) butreduced Sp (73.2% vs. 80.3%) while PPV remained substantially unchanged (68.6% vs. 64.8%). The corresponding positive likelihood ratio was 2.72. Conversely, incorporating SSM <20 kPa into the “rule out” criteria improved both Sn (32.6% vs. 21.2%) andNPV (53.2%vs. 49.5%). Conclusion: Baveno VII and AASLD criteria based on NITs, including SSM, to identify CSPH with EV in CHD demonstrate suboptimal performances compared to other liver disease etiologies. Avoiding screening endoscopy based on NITs in CHD patients could lead to misclassify patients who have varices and are at higher risk of decompensating events