Hepatic Recompensation Before Systemic Therapy for Hepatocellular Carcinoma Yields Comparable Survival to Compensated Cirrhosis

BACKGROUND AND AIMS: The survival outcomes associated with hepatic recompensation in patients with advanced hepatocellular carcinoma (HCC) treated with first-line systemic therapies remain unclear. We compared survival from the initiation of first-line systemic treatments for advanced HCC among patients with compensated, decompensated, and recompensated cirrhosis. METHODS: A Latin American multicenter, prospective cohort study was conducted from 2018 to 2024, involving patients with HCC and Child-Pugh class A or B who received systemic therapy. At the time of first-line therapy, patients with cirrhosis were categorised as compensated (never decompensated), decompensated, or recompensated. Cox proportional hazards models were estimated. RESULTS: Among 306 patients receiving first-line systemic therapy (sorafenib: 60.5%, atezolizumab + bevacizumab: 29.7%, lenvatinib: 9.1%), 240 had cirrhosis, with 30.4% having a history of hepatic decompensation. Of these, 57.5% (95% CI 45.4%-69.0%) achieved hepatic recompensation over a median period of 12 months. At the time of first-line therapy, 69.6% were compensated, 17.5% recompensated, and 12.9% decompensated. Metabolic-associated steatotic liver disease (MASLD) was the most common underlying aetiology in the recompensated group. Median survival was significantly shorter in the decompensated group (8.6 months) compared to the compensated group (17.2 months) [aHR 1.91 (95% CI 1.04-3.5); p = 0.03], without a significant difference between the recompensated and compensated groups [aHR 1.28 (95% CI 0.79-2.1); p = 0.31]. Tumour progression was the primary reason for treatment discontinuation, and similar access to second-line therapies was observed between the compensated and recompensated groups. CONCLUSION: Patients with cirrhosis and advanced HCC who achieved hepatic recompensation might benefit from systemic therapies after a cautious observation period.