1AP-HP Sorbonne Université, Hôpital Universitaire Pitié-Salpêtrière, Service d'Hépato-gastroentérologie, Paris, France.
2AP-HP Sorbonne Paris Nord, Hôpitaux Universitaire Paris Seine Saint-Denis, Service d'Hépatologie, Bobigny, France.
3AP-HP Sorbonne Université, Hôpital Universitaire Pitié-Salpêtrière, Service de radiologie diagnostique, Paris, France.
4AP-HP Sorbonne Université, Hôpital Universitaire Pitié-Salpêtrière, Service de radiologie interventionnelle, Paris, France.
5Sorbonne Université, INSERM, Centre de recherche Saint-Antoine (CRSA), Institute of Cardiometabolism and Nutrition (ICAN), Paris, France.
6INSERM UMR 1138, Centre de recherche des Cordeliers, Paris, France.
Background: Atezolizumab-bevacizumab is the new standard for advanced hepatocellular carcinoma (HCC) but its impact on portal hypertension (PHT) is unknown. We aimed to identify predictive factors of acute variceal bleeding (AVB) and to monitor PHT parameters under treatment.
Methods: We conducted a prospective study including all cirrhotic patients treated with atezolizumab-bevacizumab since 2020. We performed monitoring of PHT using upper endoscopy at inclusion and at 6 months and hepatic venous pressure gradient (HVPG) at inclusion, 3 and 6 months after the beginning of treatment. We also included a retrospective series of patients treated with sorafenib. Time-to-events data were estimated by Kaplan-Meier with the log-rank test, along with Cox models.
Results: Forty-three patients treated with atezolizumab-bevacizumab were included (male 79.1%, Child-Pugh A 86%). At baseline, 48.8% were treated with curative anticoagulation, 16.3% already experienced AVB and 25.6% had large oesophageal varices (EV). Sorafenib group characteristics were similar. Vascular invasion was present in 60.5% and median was HVPG 8.5 mm Hg. No significant modification in HVPG and EV size was observed at 6 months in the whole cohort but also when considering vascular invasion and radiological response. 14% presented AVB within a median time of occurrence of 3 months, without bleeding-related death. In multivariate analysis, history of AVB (HR = 10.58, p = .03) was associated with AVB. AVB incidence was higher in atezolizumab-bevacizumab compared to sorafenib group (21% vs. 5% at 1 year, p = .02).
Conclusions: Atezolizumab-bevacizumab treatment was associated with a higher risk of AVB compared to sorafenib. A history of AVB was associated with AVB during follow-up, which questions the use of bevacizumab in this setting.