Multicenter Phase 2 Trial of Second-Line Regorafenib in Patients with Unresectable Hepatocellular Carcinoma after Progression on Atezolizumab plus Bevacizumab.

Introduction
Primary liver cancer is the sixth most frequently diagnosed cancer, with hepatocellular carcinoma (HCC) accounting for approximately 90% of cases [1]. For patients with early-stage HCC, curative approaches such as surgical resection, liver transplantation, and ablative techniques offer chances of long-term survival. However, most patients are diagnosed at an advanced stage, making systemic therapy the mainstay of treatment [2].
Atezolizumab (anti-PD-L1 [anti-programmed death-ligand 1]) plus bevacizumab (anti-VEGF monoclonal antibody) is the first immune checkpoint inhibitor (ICI)-based combination therapy that showed significant survival benefits over sorafenib in patients with unresectable HCC, rapidly positioning itself as the global first-line systemic therapy [3–5]. Dual ICI combinations using CTLA-4 (cytotoxic T-lymphocyte-associated protein 4) and PD-1 (programmed cell death protein 1)/PD-L1 inhibitors, such as STRIDE (tremelimumab plus durvalumab) and nivolumab plus ipilimumab, also proved superior in overall survival (OS) compared to sorafenib and lenvatinib/sorafenib monotherapy, respectively [6, 7]. Additionally, the combination of rivoceranib (anti-VEGFR2 tyrosine kinase inhibitor) and camrelizumab (anti-PD-1) demonstrated OS benefits over sorafenib [8]. Although the systemic treatment landscape for unresectable HCC has evolved rapidly over the past decade, atezolizumab-bevacizumab remains one of the most widely used regimens in the first-line setting.
Despite these advances, a significant proportion of patients do not respond to or progress on first-line atezolizumab-bevacizumab. Previous studies have shown that 50–60% of patients are medically fit to receive subsequent therapy after progression on first-line atezolizumab-bevacizumab [9, 10]. However, optimal post-atezolizumab-bevacizumab treatment has not been firmly established due to a lack of well-designed prospective studies. In daily clinical practice, multi-kinase inhibitors (MKIs) such as sorafenib, lenvatinib, regorafenib, and cabozantinib, approved for the management of advanced HCC before the era of ICIs, are commonly used. Until recently, the clinical outcomes of these MKIs as subsequent treatments after progression on ICIs have only been evaluated in retrospective studies [9–12]. Prospective clinical trials are needed to assess the efficacy and safety of previously approved MKIs following progression on new first-line standard therapy.
Regorafenib is an orally available MKI that targets angiogenesis (VEGFR1-3, TIE2), oncogenesis (KIT, RET, RAF1, and BRAF), and the tumor microenvironment (PDGFR and FGFR) [13, 14]. In the phase 3 RESORCE trial, regorafenib significantly improved median OS (10.6 months vs. 7.8 months) compared to placebo in patients with unresectable HCC who progressed on and tolerated sorafenib [15]. Since then, regorafenib has been established as one of the standard subsequent regimens following progression on sorafenib [16]. However, its clinical outcomes have not been prospectively evaluated in patients who progressed on first-line atezolizumab-bevacizumab. This study aimed to assess the efficacy and safety of regorafenib in patients with unresectable HCC who progressed after first-line atezolizumab-bevacizumab.

Patients and Methods

Study Design and Endpoint
This trial (NCT05134532) was an open-label, multicenter, investigator-initiated, single-arm phase 2 trial for patients with unresectable HCC, recruited from two sites in South Korea (Asan Medical Center, Seoul, and CHA Bundang Medical Center, Seongnam). The study adhered to the Declaration of Helsinki and Good Clinical Practice guidelines. Protocols were approved by the Institutional Review Board at each participating center (Asan Medical Center, 2021-1836; CHA Bundang Medical Center, 2021-05-048-076). Written informed consent was obtained from all patients prior to enrollment. The study received partial support from Bayer, providing regorafenib and research funding.

Eligibility
Key inclusion criteria included a diagnosis of HCC according to American Association for the Study of Liver Disease guidelines, unresectable or recurrent HCC, prior first-line atezolizumab-bevacizumab treatment (at least 2 consecutive cycles), age of at least 19 years, an ECOG (Eastern Cooperative Oncology Group) performance status of 0 or 1, a Child-Pugh score of 5 or 6, and adequate organ and bone marrow function. Key exclusion criteria included fibrolamellar carcinoma or mixed hepatocellular cholangiocarcinoma, a history of active cancer within 2 years before registration, and clinically significant bleeding within 28 days of registration.

Procedures
Eligible patients received regorafenib 160 mg orally for 3 weeks of every 4-week cycle (i.e., 3 weeks on, 1 week off) until disease progression, unacceptable toxicity, or withdrawal of consent. Treatment interruption and dose reduction were protocol driven and based on the severity of adverse events (AEs). Generally, study treatment could continue in the case of grade 1 or tolerable grade 2 treatment-related AEs (TRAEs) at the discretion of investigators. For intolerable grade 2 or grade 3 TRAEs, regorafenib was interrupted and followed by dose reduction upon improvement of AEs to grade 1 or baseline. For grade 4 TRAEs, study treatment was immediately interrupted.
Tumor response was assessed with computed tomography of the thorax and abdomen every 8 weeks. Additional investigations (e.g., bone scintigraphy) were arranged if clinically indicated. Responses were determined according to RECIST (Response Evaluation Criteria in Solid Tumors) v1.1. Safety was assessed according to the Common Terminology Criteria for Adverse Events v5.0.

Statistical Analysis
The efficacy analysis set was based on the intent-to-treat population. The primary endpoint was progression-free survival (PFS) as evaluated by investigators according to RECIST v1.1. Secondary endpoints included OS, time to progression (TTP), objective response rate (ORR), disease control rate (DCR), and TRAEs. PFS was defined as the time from the initiation of regorafenib to the first documentation of objective tumor progression or to the time of death from any cause. Patients who were alive without evidence of clinical or radiological progression were censored. OS was assessed as the time from the start of regorafenib to the date of death from any cause. In the absence of confirmation of death, survival time was censored at the last date the patient was known to be alive. TTP was defined as the time from the start of regorafenib to the date of the first documentation of objective tumor progression. The ORR was calculated as the proportion of patients with complete response (CR) or partial response (PR) among the intent-to-treat population, and DCR was defined as the proportion of those with CR, PR, or stable disease. The safety analysis set included patients who received at least one dose of regorafenib and visited the clinic for the assessment of AEs. The rate and severity of toxicities, as well as demographic data, were summarized using descriptive statistics. The hazard ratios and corresponding 95% confidence intervals (CIs) were computed using a Cox proportional hazards model. SPSS Statistics for Windows, v21.0 (IBM Corp., Armonk, NY, USA), was used for the analyses.
According to second-line clinical trials on tyrosine kinase inhibitors for treating HCC, the median PFS of the placebo arm is 2 months [15, 17]. The sample size was calculated based on the hypothesis that regorafenib treatment would improve the median PFS from 2 months to 4 months [18]. Under the conditions of an accrual time of 24 months and a follow-up duration of 12 months, with a one-sided alpha of 0.05 and a power of 80%, the sample size was calculated to be 36. Assuming a 10% dropout rate, the number of cases in this study was determined to be 40. The sample size was calculated using PASS v11.0 (NCSS, LLC, Kaysville, UT, USA).

Results

Baseline Characteristics
Forty-two patients were screened, and 40 were enrolled between December 2021 and April 2023. Baseline characteristics are summarized in Table 1. The median age was 56 years (range, 36–81 years); 77.5% of patients were men, and 60.0% had an ECOG performance status of 0. One patient was still receiving regorafenib treatment at the data cutoff (November 30, 2023). The most common etiology of HCC was hepatitis B (77.5%). All patients had Child-Pugh A liver function, with a Child-Pugh score of 5 in 75.0% (n = 30). Most patients had Barcelona Clinic Liver Cancer (BCLC) C stage HCC (97.5%). Extrahepatic metastasis was present in 85.0% of patients, with the most common metastatic site being the lymph nodes (60.0%, n = 24), followed by the lungs (42.5%, n = 17), bone (15.0%, n = 6), and peritoneum (5.0%, n = 2). Macrovascular invasion was observed in 25.0% of patients (n = 10), and main portal vein invasion was present in 10.0% (n = 4). The percentage of patients with baseline serum alpha-fetoprotein (AFP) levels exceeding 400 ng/mL was 40.0%. The median TTP of prior atezolizumab-bevacizumab was 4.1 months (95% CI: 1.9–6.4; interquartile range, 2.3–7.0 months). There was a median of 6 (interquartile range, 3.3–10.0) cycles of prior atezolizumab-bevacizumab treatment for a median duration of 3.5 months (95% CI: 0.5–6.5). During first-line atezolizumab-bevacizumab treatment, 1 patient (2.5%) and 14 patients (35.0%) achieved CR and PR as their best tumor responses, respectively.

Efficacy
Efficacy outcomes are summarized in Table 2. With a median follow-up duration of 10.1 months (95% CI: 8.3–12.0), the median PFS, the primary endpoint, was 3.5 months (95% CI: 3.0–3.9; Fig. 1a). The median OS was 10.5 months (95% CI: 7.1–13.8), and the 6-month OS rate was 65.0% (Fig. 1b). The median TTP was 3.5 months (95% CI: 3.0–4.0). Four patients achieved PR, resulting in an ORR of 10.0%. Stable disease and progressive disease were observed in 29 (72.5%) and 7 patients (17.5%), respectively, leading to a DCR of 82.5% (online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000543666).

Safety
The TRAEs occurring in ≥5% of patients are summarized in Table 3, and all AEs, regardless of treatment, are listed in online supplementary Table 1. The most common TRAEs of any grade were palmar-plantar erythrodysesthesia (PPES, n = 28, 70.0%), anorexia (n = 10, 25.0%), fatigue (n = 9, 22.5%), hyperbilirubinemia (n = 9, 22.5%), and aspartate aminotransferase increase (n = 8, 20.0%). The most common grade 3–4 AE was thrombocytopenia (n = 2, 5.0%). No new safety signals for regorafenib were observed.
Mean daily dose of regorafenib was 113.5 mg (standard deviation: 32.5) and the mean dose intensity of regorafenib was 84.4%. The mean relative dose intensity for the first 3 weeks (3W-RDI) was 78.7% (standard deviation: 19.6). A total of 35 patients (87.5%) experienced dose interruptions or reductions due to AEs. The most common causes of dose modifications were PPES (n = 11, 27.5%), general weakness (n = 7, 17.5%), and hepatic dysfunction (n = 3, 7.5%).

Subgroup Analysis
When stratified by the TTP on prior atezolizumab-bevacizumab (first quartile [<2.3 months] vs. second to fourth quartiles [≥2.3 months]), patients with longer TTP on prior atezolizumab-bevacizumab showed a tendency for better median PFS with regorafenib (3.8 vs. 2.5 months, p = 0.054). This was significantly associated with OS and ORR with regorafenib; the median OS was 15.0 months versus 3.6 months (p < 0.001; Fig. 2a, b), and the ORR was 13.3% versus 0%, respectively (p = 0.009). Neither median TTP nor tumor response with prior atezolizumab-bevacizumab was associated with PFS, OS, or ORR with regorafenib (online suppl. Table 2).
The presence of macrovascular invasion (n = 10, 25%) was associated with significantly poorer PFS (presence vs. absence; 2.5 vs. 3.5 months, p = 0.048) and numerically shorter OS (8.4 vs. 15.0 months, p = 0.096, Fig. 3a, b). Additionally, patients with baseline AFP ≥400 ng/mL were significantly associated with shorter PFS (2.6 vs. 5.3 months, p = 0.044) and showed a trend toward shorter OS (5.7 vs. 12.3 months, p = 0.257) compared to those with AFP <400 ng/mL (Fig. 3c, d).
In terms of baseline liver function’s potential impact on the efficacy of regorafenib, Child-Pugh score (5 [n = 30] vs. 6 [n = 10]) did not show statistically significant differences in PFS (median 3.9 vs. 2.5 months, p = 0.618), OS (12.3 vs. 6.6 months, p = 0.636), and ORR (13.3% vs. 0%, p = 0.556), although these outcomes tended to be numerically better with better liver function. These results were consistent in the subgroup analysis using ALBI (albumin-bilirubin) grade (1 [n = 17] vs. 2 [n = 23]): median PFS, 4.7 vs. 3.4 months, respectively, p = 0.603; median OS, 15.0 vs. 8.4 months, respectively, p = 0.096; ORR, 17.6% vs. 4.3%, respectively, p = 0.294 (online suppl. Table 3).
We also assessed the pattern of progression during prior atezolizumab-bevacizumab treatment [19]. Progression was related to the development of new extrahepatic lesions in 11 patients (27.5%), new intrahepatic lesions in 8 patients (20.0%), and tumor growth of existing lesions (intrahepatic or extrahepatic) in 31 patients (77.5%). The pattern of progression on prior atezolizumab-bevacizumab was not associated with PFS or OS on regorafenib (online suppl. Table 4).
Based on the findings from the post hoc exploratory analysis of phase 3 RESORCE trial [20], among the safety profiles, PPES of any grades (n = 25, 62.5%) was significantly associated with better PFS (yes vs. no; median 3.8 vs. 2.6 months, p = 0.015, Fig. 3e) and patients with PPES of any grades showed numerically longer OS (12.3 vs. 6.6 months, p = 0.625, Fig. 3f). There was no relationship between 3W-RDI and survival outcomes: 3W-RDI > mean (n = 21) vs. 3W-RDI ≤ mean (n = 19); PFS, 3.5 vs. 3.5 months (p = 0.838); and OS, 10.5 vs. 9.7 months (p = 0.907).

Subsequent Therapy
At the time of analysis, regorafenib was discontinued in 39 patients (97.5%), primarily because of disease progression (n = 35, 87.5%), AEs (n = 1, 2.5%), patient withdrawal (n = 1, 2.5%), and clinical deterioration (n = 1, 2.5%). After progression on regorafenib, 26 patients received subsequent treatment: sorafenib (n = 17), lenvatinib (n = 8), or an investigational immunotherapeutic agent (n = 1).

Discussion
In the current study, which prospectively investigated the clinical outcomes of second-line regorafenib in patients with unresectable HCC who had progressed on first-line atezolizumab-bevacizumab, the median PFS and OS were 3.5 months and 10.5 months, respectively, with an ORR of 10%. Although the primary endpoint (median PFS of 4.0 months) was not met, these outcomes are comparable to those of the pivotal phase 3 RESORCE trial for sorafenib-tolerant and -progressed patients, which reported a median PFS of 3.4 months according to RECIST v1.1 and a median OS of 10.6 months [15]. The AEs of regorafenib were manageable, and no new safety signals were observed. Our findings suggest that regorafenib may be considered as a subsequent therapy after progression on first-line atezolizumab-bevacizumab.
While there has been no global standard therapy after progression on first-line atezolizumab-bevacizumab, previously approved MKIs have been widely used as subsequent treatments, and multiple retrospective analyses have assessed their clinical outcomes. Mostly, sorafenib and lenvatinib, previous first-line therapies, were investigated, showing median OS ranging between 5.6 and 16.6 months and median PFS ranging between 1.8 and 6.1 months [10, 11, 21]. Meanwhile, regorafenib and cabozantinib, approved as subsequent therapies after progression on first-line sorafenib, have not been investigated in previous retrospective studies. Recently, a multinational, multicenter phase 2 trial of second- or third-line cabozantinib after progression on prior ICIs demonstrated a median PFS and OS of 4.1 and 9.9 months, respectively [22]. In the subgroup receiving cabozantinib as second-line treatment after first-line atezolizumab-bevacizumab (n = 13), the median PFS and OS were 4.3 and 9.9 months, respectively. These findings including our results suggest that regorafenib and cabozantinib may show comparable efficacy outcomes as subsequent therapies after progression on atezolizumab-bevacizumab, consistent with the results of pivotal phase 3 trials for sorafenib-pretreated patients [15, 17]. In a recent multinational phase 2 study of regorafenib plus pembrolizumab as second-line treatment after ICI-containing regimens, patients who progressed on prior atezolizumab-bevacizumab (n = 68) showed ORR of 5.9%, median PFS of 2.8 months (95% CI: 2.4–3.9 months) [23]. This trial used the reduced starting dose of regorafenib (90 mg once daily 3 weeks on 1 week off) and allowed dose escalation to 120 mg after the first cycle. Although the efficacy outcomes in this study seem to be numerically less than ours, it might be affected by discrepancies of patient baseline characteristics such as major vascular invasion (32% vs. 25% in ours) or reduced dose of regorafenib (90 mg vs. 160 mg).
In the subgroup analyses, patients with the shortest TTP (<2.3 months) on prior atezolizumab-bevacizumab were associated with significantly poorer OS (3.6 vs. 15.0 months, p < 0.001) and ORR (0% vs. 13.3%, p = 0.009) and numerically shorter PFS (2.5 vs. 3.8 months, p = 0.054) compared to those with longer TTP (≥2.3 months) on prior atezolizumab-bevacizumab. This indicates that poor tumor response to prior atezolizumab-bevacizumab may be a prognostic factor for subsequent regorafenib treatment. In line with these findings, it may be necessary to interpret the efficacy outcomes of prospective or retrospective studies for post-atezolizumab-bevacizumab second-line therapy with consideration of the proportion of patients with primary resistance to atezolizumab-bevacizumab in the overall study population.
There were no new safety signals for regorafenib identified in this study [15, 24]. Although the overall incidence of PPES (70%) was higher than that reported in the RESORCE trial (53%), the severity was predominantly mild (grade 1 or 2). It is important to note that the RESORCE trial included patients who had already tolerated sorafenib [15]. Consistent with previous post hoc analysis of the RESORCE trial, PPES of any grades were associated with better PFS in this study [20]. Most TRAEs were grade 1 or 2 in severity and were clinically manageable with dose reductions or interruptions. Our findings suggest that regorafenib may be used safely in patients with HCC without prior exposure to sorafenib [25].
Our study had a few limitations, including the small sample size and lack of a control group. Additionally, this study was conducted solely in Korea, which may limit its generalizability worldwide.
In conclusion, the efficacy and safety of regorafenib in this multicenter phase 2 study were consistent with those observed in the RESORCE trial, which included sorafenib-tolerant or -progressed patients. Our findings suggest that regorafenib has the potential to serve as a second-line therapy after the failure of first-line atezolizumab-bevacizumab. However, further studies in larger patient populations are needed to confirm these results.

Statement of Ethics
This study was conducted in compliance with the Declaration of Helsinki and Good Clinical Practice guidelines. All protocols were reviewed and approved by the Institutional Review Board at each participating center (Asan Medical Center, 2021-1836; CHA Bundang Medical Center, 2021-05-048-076). Academic investigators and sponsors were responsible for the study design. All patients who participated in this trial provided written informed consent before enrollment (ClinicalTrials.gov identifier: NCT05134532).

Conflict of Interest Statement
J. Cheon received honoraria from Eisai and MSD, as well as research grants from Bayer. H.J. Chon received honoraria from Eisai, Roche, ONO, MSD, Bristol Myers Squibb, BeiGene, Sanofi, Servier, AstraZeneca, Menarini, and Boryung and received research grants from Roche, BeiGene, Dong-A ST, and Boryung Pharmaceuticals. H.-D. Kim received honoraria from AstraZeneca, Bristol Myers Squibb, Ono Pharmaceuticals, Boryung Pharmaceuticals, and Boostimmune; received research grants from AstraZeneca; and served as a consultant for Mustbio. R.S. Finn reports grants or contracts to their institution from Adaptimmune, Bayer, Bristol Myers Squibb, Eisai, Eli Lilly, Pfizer, Roche, and Genentech; consulting fees to themselves from Merck, AstraZeneca, Bayer, Bristol Myers Squibb, Exelixis, Cstone, Hengrui, Eisai, Eli Lilly, MSD, Pfizer, Roche, and Genentech; payment or honoraria to themselves from Genentech; and participation on a data safety monitoring or advisory board for AstraZeneca and Hengrui. S.L. Chan serves as an advisory member for AstraZeneca, MSD, Eisai, and Autem; received research funds from Celleron, Genorbio, Ipsen, MSD, Novartis, and SIRTEX; and honoraria from AstraZeneca, Bayer, BMS, Eisai, Hutchmed, Ipsen, MSD, and Roche. C. Yoo received honoraria from Servier, Bayer, AstraZeneca, Merck Sharp & Dohme, Eisai, Celgene, Bristol Myers Squibb, Ipsen, Novartis, Boryung, Mundipharma, and Roche and received research grants from Servier, Bayer, AstraZeneca, Ono Pharmaceuticals, Ipsen, Boryung, and Lunit Inc.

Funding Sources
This study was supported in part by Bayer. Regorafenib and research funding were provided by Bayer.

Author Contributions
Study design and conceptualization: J. Cheon, S.L. Chan, and C. Yoo. Methodology and formal analysis: J. Cheon and C. Yoo. Data collection: J. Cheon, B.-Y. Ryoo, H.J. Chon, H.-D. Kim, M.-H. Ryu, K-P. Kim, B. Kang, and C. Yoo. Data analysis and interpretation, writing – original draft, and writing – review and editing: J. Cheon, B.-Y. Ryoo, H.J. Chon, H.-D. Kim, M.-H. Ryu, K.-P. Kim, B. Kang, R.S. Finn, S.L. Chan, and C. Yoo.