• atezolizumab bevacizumab
• clinical factor
• hepatocellular carcinoma
• radiological factor

Patients (pts) who received Atezolizumab plus bevacizumab (AtezoBev) for hepatocellular carcinoma (HCC) can be categorized as primary progressors (PP) if the disease showed progression at the first CT scan, or as disease-controlled (DC) if they achieved a complete or partial response or stable disease. Disease responders (DR) are pts with a complete or partial response.
The identification of PP is challenging. The main aim of the study was to identify clinical and radiological factors as potential predictors of AtezoBev resistance.
A prospective analysis included pts treated with AtezoBev in a real-life setting from the ARTE multicentric Italian database from October 2020 to November 2023. We considered the baseline clinical characteristics of pts to identify the association with radiological responses.
We collected data from a single-institution retrospective cohort (Pisa cohort) to investigate clinical and radiological factors linked to primary resistance to AtezoBev. We included 37 pts in the AtezoBev group and a control group, consisting of 34 pts treated with lenvatinib (Lenv), between September 2019 and February 2024. CT images were reviewed by independent radiologists of Azienda Ospedaliera Universitaria Pisana.
We included 212 pts from the ARTE database. After a median follow-up of 14.57 months (95% CI 14.19-17.28), the median overall survival (OS) was 19.70 months (95% CI 15.26-29.47). The median OS of DR pts was not reached, pts who achieved SD 27.80 months (95% CI 26.61 to 30.10), and PP 6.12 months (95% CI 4.24 to 7.80). PP (n =74) had a higher prevalence of AFP ≥400 ng/mL, BCLC C stage, pts under 65 years old, and a lower prevalence of MetALD etiology than DC (n = 138) (p < 0.05). In the univariate overall survival (OS) analysis, we found only statistical differences among patients with BCLC stage and AFP levels (p < 0.05). DR (n= 40) and PP differ significantly in terms of the BCLC stage, MetALD etiology prevalence, and presence of extrahepatic disease (p < 0.05). In the univariate OS analysis, only BCLC status was a prognostic factor (p = 0.008).
In the Pisa retrospective cohort, after a median follow-up of 24.51 months (95% CI 17.43-37.96), there were no statistical differences in median OS (AtezoBev 15.79 months and Lenva 18.75 months, p = 0.95).
In the AtezoBev group, we identify PP (n =14) and DC (n=23). PP had a higher prevalence in females, pts under 65 years old, AFP ≥400 ng/mL, use of opioids as concomitant medication, and pts with a recent diagnosis of HCC (< 12 months) compared to DC (p < 0.05). We didn't observe any differences in radiological features. Time from diagnosis of HCC to AtezoBev and AFP levels were statistically significant on both univariate and multivariate OS analysis (p = 0.003). PP and DR (n=3) differed in the time from diagnosis to AtezoBev (p = 0.01).
Lenv cohort was homogeneous despite a higher prevalence of patients with cirrhosis and non-viral etiology than the AtezoBev group (p < 0.05).
There was no significant difference between AtezoBev and Lenv in time from diagnosis to systemic treatment (p = 0.48). In AtezoBev, pts diagnosed with HCC < 12 months had a shorter OS than those diagnosed with 12 months equal or more (9.34 months versus 37.76 months p < 0.05). In the Lenv, there was no difference in OS between the two groups (14.80 months vs 18.75 months, p = 0.68).
In conclusion, time from diagnosis to systemic treatment may be a significant potential factor in predicting response to AtezoBev. Pts with more aggressive diseases might benefit less from AtezoBev. Both clinical and radiological factors could not help discriminate PP from DC. A large-scale study is required to confirm the utility of time from diagnosis to systemic treatment.