Atezolizumab and bevacizumab (atezo/bev) combination therapy is increasingly used for advanced hepatocellular carcinoma (HCC), substantially advancing the systemic chemotherapy for this type of tumor. High-quality evidence for post-atezo/bev salvage therapy is lacking; however, the number of nonresponse or intolerance cases with this regimen is increasing in real-world clinical practice. Therefore, an effective post-atezo/bev salvage therapy is imperative. Based on the results of the IMbrave150 trial, atezo/bev was confirmed as a highly potent treatment, substantially altering the treatment landscape of HCC [1]. Currently, in almost all guidelines for the treatment of HCC, atezo/bev is recommended as the first-line therapy if there are no contraindications.

To select second-line therapy for non-response or intolerance to atezo/bev, two tyrosine kinase inhibitors (TKIs), sorafenib and lenvatinib, which were used mainly as firstline therapy before the approval of atezo/bev, are often chosen over other second-line agents because of their high anti-tumor objective response rate (ORR). Chon et al. [2] compared the efficacy of sorafenib and lenvatinib as second-line therapy, which was initiated between August 2019 and December 2022 after the failure of atezo/bev in patients with HCC. This was accomplished through a multicenter retrospective observational study. This study is noteworthy for balancing the clinical characteristics of both groups through propensity score (PS) matching and providing new salvage evidence after atezo/bev failure. In the PS-matched cohort, the ORR was comparable between the lenvatinib and sorafenib groups; however, the disease control rate (DCR) was higher in the lenvatinib group. Additionally, progression-free survival (PFS) was also superior in the lenvatinib group (median: 3.5 months vs. 1.8 months), whereas OS did not differ substantially between the two groups (median, 10.3 months vs. 7.5 months).

A subset of patients who discontinued atezo/bev treatment in the IMbrave150 trial received TKIs. Yoo et al. [3] analyzed cases, including those in phase III and I clinical trials, or given off-trial in daily clinical settings where TKIs were administered after atezo/bev. They reported a substantially higher median PFS in the lenvatinib group compared with that in the sorafenib group (6.1 months vs. 2.5 months), while there was no significant difference in median overall survival (OS) between the two groups.

In a phase III trial (REFLECT), lenvatinib demonstrated non-inferiority to sorafenib with OS as the primary endpoint. However, unlike sorafenib, lenvatinib exhibits a considerably high anti-tumor effect with an ORR according to the modified Response Evaluation Criteria in Solid Tumors, which considers the tumor blood flow status. Furthermore, real-world data prior to the advent of immune checkpoint inhibitors (ICIs) indicate that lenvatinib as a second-line therapy achieves comparable PFS to first-line therapy, especially in patients with good liver function classified as modified albumin-bilirubin grade 1/2a [4]. Additionally, lenvatinib was administered as a second-line therapy for post-ICI treatment [5]. In this cohort, although most patients received ICIs other than atezo/bev, the PFS and OS were favorable, with a median PFS of 10 months and an OS of 15 months. Considering that the median PFS of first-line atezo/bev in phase III clinical trials was 6.8 months [1], lenvatinib administration showed very promising results as a post-ICI second-line therapy. As ICIs are antibody agents that maintain high serum concentrations and affect T-cell responses for months [6], durable effects of ICIs can be expected after treatment cessation. From this perspective, the subsequent administration of lenvatinib may yield a temporary combination effect with ICIs, which may be attributed to the favorable PFS and OS of lenvatinib in the second-line setting after ICIs.

One mechanism proposed for the acquisition of resistance to ICIs involves the activation of the Wnt/β-catenin pathway in cancer cells, which is thought to impair the induction of dendritic cells and cytotoxic T lymphocytes into the tumor through the induction of activating transcription factor 3 (ATF3) and inhibition of C-C chemokine ligand (CCL) 4 and 5 [7,8]. Harding et al. [9] reported on ICI resistance in HCC harboring activating mutations in β-catenin, and we also found an association between activation of the Wnt/β-catenin and shortened PFS on ICI therapy [10]. Conversely, another report suggested that activating mutations in β-catenin are associated with increased fibroblast growth factor receptor 4 (FGFR4) signaling in tumor [11]. Lenvatinib is a strong inhibitor of FGFR4, with high efficacy observed in cases with high expression of FGFR4 in HCC.12 Therefore, it is possible that lenvatinib may be effective for cases of resistance to ICI therapy attributed to activation of the Wnt/β-catenin pathway, which is associated with activation of FGFR4 signaling.

Considering these findings, lenvatinib should likely be considered as a second-line therapy after atezo/bev therapy in patients with good liver function. However, the effectiveness of lenvatinib is correlated with its relative dose intensity (RDI), and decreasing RDIs are observed as liver function declines [13]. According to a report from a phase I clinical trial, lenvatinib serum concentrations increased in patients with Child-Pugh B compared to those in Child-Pugh A cases. Furthermore, dose adjustments based on body weight are required for lenvatinib administration; however, it is not clear how dose adjustments based on body weight should be made, considering the increase in blood concentrations in Child-Pugh B cases. When lenvatinib is used as second-line therapy or beyond, cases with compromised liver function may increase, which may make the administration of lenvatinib more challenging.

Sorafenib is characterized by strong inhibition of c-Raf and b-Raf serine/threonine kinases compared with lenvatinib. As a first-line therapy, sorafenib showed a lower ORR than that with lenvatinib. However, according to sorafenib clinical trials using large HCC cohorts, there was no significant difference in efficacy between starting doses of 800 mg and 400 mg. Therefore, initiation with 400 mg of sorafenib daily can be considered, particularly in patients with underlying conditions or impaired liver function, considering the risk of treatment interruption. Clinical trials exploring the safety and efficacy of Child–Pugh B showed no significant difference in PFS between Child–Pugh A and Child–Pugh B cases [14,15]. Child–Pugh B cases are limited to those with Child–Pugh scores of 7 or 8, and compared with Child–Pugh A cases, Child–Pugh B cases exhibit inferior OS. Regarding safety, although the overall incidence of adverse events (AEs) was similar, a higher incidence of severe and liver-related AEs was observed in Child–Pugh B cases. Nevertheless, given that an increase in cases with deteriorated liver function may be observed after the failure of first-line therapy, the abundant evidence of sorafenib in Child–Pugh B cases shows this agent as an option for second-line therapy.

In addition to these two TKIs, ramucirumab, a monoclonal antibody targeting vascular endothelial growth factor receptor 2 (VEGFR2), can also be selected as a secondline therapy for patients with HCC with serum α-fetoprotein (AFP) levels of 400 ng/mL or higher. Kuzuya et al.16 reported 13 HCC cases for which ramucirumab was administered after atezo/bev, with an ORR of 5.4% and a DCR of 69.2%. They noted a substantial extension in survival among cases showing disease control and AFP reduction at six weeks after treatment initiation. Currently, a randomized comparative trial of lenvatinib and ramucirumab is being conducted for the treatment of advanced HCC with AFP values of 400 or higher after ICI treatment (SELECT-400). Additionally, the efficacies of regorafenib and cabozantinib have been investigated in patients with HCC who are unresponsive to ICIs, prompting expectations of comparative performance among molecular-targeted agents (Table 1) [17,18].

Another large-scale phase III clinical trial (IMbrave251) is underway to investigate the significance of adding atezolizumab to sorafenib or lenvatinib as a second-line treatment for patients who have experienced disease progression after atezo/bev (Table 1). This trial provides new evidence regarding post-atezo/bev therapeutic options. In addition to atezo/bev, the durvalumab and tremelimumab (durva/treme) combination is available as a first-line therapy, and exploration of the sequencing of these two combination immunotherapies is anticipated. In second-line therapy, no regimen is an option, and there are many challenges to be addressed regarding the efficacy and safety of regimens for second-line and subsequent treatments. There is ongoing discussion on whether to retry ICI-based regimens, such as durva/treme, or use TKIs for patients unresponsive to atezo/bev [19]. Additionally, new drugs with different mechanisms of action and novel immunotherapies are being developed (Table 1).

In contrast, Zhu et al. [20] identified key molecular correlations of atezo/bev combination therapy using cases enrolled in the GO30140 phase 1b or IMbrave150 phase 3 trial. Zeng et al. [21] developed artificial intelligence (AI) and found that AI applied to HCC digital slides predicted PFS in patients with HCC treated with atezo/bev. These efforts should play a critical role in achieving long survival with atezo/bev; they could also provide important information for the selection of ideal salvage therapy after atezo/bev.

Currently, several agents are available for systemic chemotherapy of HCC, making it difficult to conduct prospective comparative trials for all combinations when determining the sequence of drug therapy. Large-scale observational clinical studies are ongoing to address this issue (Table 1) and provide new evidence based on real-world data to construct an ideal salvage therapy after ICI-based therapies for HCC.