A MULTICENTRE retrospective study of 685 patients with unresectable hepatocellular carcinoma (uHCC) found that progressive disease (PD) was the leading reason for discontinuation of atezolizumab plus bevacizumab, while older age and poorer liver function identified patients at greater risk of toxicity-related discontinuation.
Atezolizumab plus bevacizumab is a standard first-line treatment for uHCC. Because treatment can be discontinued for tumour progression, adverse events or worsening liver function, understanding which pathway is most likely to occur could help clinicians tailor monitoring during treatment.
The study assessed 685 consecutive patients treated with the regimen. Discontinuation was attributed to PD adverse events (AEs), hepatic dysfunction, response or strategy change, or other causes.
PD accounted for the largest cumulative incidence of discontinuation at one year, at 39.4%, followed by AEs at 15.3% and hepatic dysfunction at 5.6%. Response or strategy change accounted for 4.2%, while other causes accounted for 4.1%.
Baseline alpha-fetoprotein (AFP) predicted PD-driven discontinuation, with the hazard increasing by 18% for each log10 increase in AFP plus one. By contrast, toxicity-related pathways were more closely linked to patient and liver characteristics.
AE-driven discontinuation was independently associated with older age, an ECOG performance status of at least 1 and poorer albumin-bilirubin (ALBI) scores, which reflected liver function. Hepatic dysfunction-driven discontinuation was also strongly associated with age and ALBI.
The researchers then combined AFP with modified ALBI (mALBI) to define four clinical risk groups. This separated patients whose treatment was more likely to stop because of tumour progression from those at greater risk of toxicity-related discontinuation.
Patients with low AFP and poor mALBI had the highest one-year toxicity-related discontinuation, with a cumulative incidence of 33.9%. Conversely, patients with high AFP and good mALBI had predominantly PD-driven discontinuation, with a one-year cumulative incidence of 52.0%.
The findings suggested that baseline tumour burden and hepatic reserve contributed to different treatment-stopping pathways, rather than all discontinuations representing the same clinical event.
The AFP×mALBI framework could help clinicians anticipate whether progression or treatment-related toxicity was more likely to drive discontinuation, potentially informing monitoring, toxicity preparedness and supportive care.
However, the study was retrospective and the risk framework had not undergone external validation. It should therefore be viewed as hypothesis-generating rather than as a validated tool for treatment selection. Further validation will be needed to establish whether AFP and mALBI can reliably guide clinical monitoring in routine uHCC care.
Reference
Suda G. Competing risks of treatment discontinuation of atezolizumab plus bevacizumab in unresectable hepatocellular carcinoma. JHEP Rep. 2026;DOI:10.1016/j.jhepr.2026.101988.
Featured image: Sebastian Kaulitzki on Adobe Stock
Each article is made available under the terms of the Creative Commons Attribution-Non Commercial 4.0 License.
