Critical Appraisal of Guideline Recommendations on Systemic Therapies for Advanced Hepatocellular Carcinoma A Review

Abstract

IMPORTANCE

The combination of immune checkpoint inhibitors with antiangiogenic agents has revolutionized the treatment landscape of advanced hepatocellular carcinoma (HCC). However, due to rapid publication of new studies that attained their predefined primary end points, a lack of robust cross-trial comparison of first-line therapies, and diverging clinical guidelines, no clear-cut treatment flowchart and sequence of therapies are available. This critical analysis of the recommendations for the management of advanced HCC from the main scientific societies in the US and Europe adopted an integrated approach to provide information on the clinical benefit (overall survival and progression-free survival) and safety profile of these therapies using the European Society for Medical Oncology (ESMO)–Magnitude of Clinical Benefit Scale (MCBS) score and an ad hoc network meta-analysis.

OBSERVATIONS

There is a major consensus among guidelines that atezolizumab plus bevacizumab has a primacy as the recommended first-line treatment of choice in advanced HCC. On progression after immunotherapy-containing regimens and for patients with contraindications for immunotherapies, most guidelines maintain the established treatment hierarchy, recommending lenvatinib or sorafenib as the preferred options, followed by either regorafenib, cabozantinib, or ramucirumab. Thus far, the first-line immune-based regimen of tremelimumab plus durvulumab has been integrated only in the American Association for the Study of Liver Diseases guidance document and the latest National Comprehensive Cancer Network guidelines and is recommended for patients with a high risk of gastrointestinal bleeding. Overall, in the first-line setting, both atezolizumab plus bevacizumab and sintilimab plus IBI305 (a bevacizumab biosimilar) received the highest ESMO-MCBS score of 5, indicating a substantial magnitude of clinical benefit. In a network meta-analysis, no significant differences in overall survival were found among the various combination regimens. However, the newly reported combination of camrelizumab plus rivoceranib was associated with a significantly higher risk of treatment-related adverse events compared with atezolizumab plus bevacizumab (relative risk, 1.59; 95% CI, 1.25-2.03; P < .001).

CONCLUSIONS AND RELEVANCE

This narrative review found that atezolizumab plus bevacizumab is regarded as the primary standard of care for advanced HCC in the first-line setting. These findings from integrating the recommendations from scientific societies’ guidelines for managing advanced HCC along with new data from cross-trial comparisons may aid clinicians in decision-making and guide them through a rapidly evolving and complex treatment landscape.

Hepatocellular carcinoma (HCC) is a global health problem with increasing incidence and mortality.¹ Approximately 50% to 60% of patients with HCC are treated with systemic therapies.² First-line therapies for advanced HCC include 4 regimens, and second-line therapies and further lines of treatment include up to 5 distinct alternatives in Western countries.^1-3 Three additional molecular treatments have been reported as superior to sorafenib in China.^2,3 The increasing number of effective regimens has been associated with improvement in the median overall survival (OS) of patients with advanced HCC from 8 months for the placebo arm⁴ to 19.2 months for atezolizumab plus bevacizumab, the current standard of care in first-line treatment.⁵ Both the US (American Association for the Study of Liver Diseases [AASLD],⁶ American Society of Clinical Oncology [ASCO],⁷ National Comprehensive Cancer Network [NCCN],⁸ and American Gastroenterological Association [AGA]⁹) and European (European Society for Medical Oncology [ESMO]¹⁰ and European Association for the Study of the Liver [EASL]¹¹) guidelines or guidance for management of HCC have recommended these therapies. Nonetheless, due to the fast pace of publication of new studies that attained their predefined primary end points and the lack of high-level evidence to support recommendations in specific clinical settings (ie, for second-line treatment following atezolizumab plus bevacizumab; for patients with Child-Pugh class B liver disease), no clear-cut flowchart and sequence of therapies exist.

In this narrative review, we conducted a critical analysis of the existing evidence on HCC and the strength of recommendations by the main scientific societies that produced the guidelines for the management of HCC.^6-11 In addition, due to the lack of cross-trial comparison of the magnitude of clinical benefit among drugs that have been approved for first-line treatment and the lack of direct comparison of the safety profiles of these approved drugs, we provided information on the clinical benefit and safety profile of these therapies using the ESMO–Magnitude of Clinical Benefit Scale (MCBS) score and an ad hoc network meta-analysis. Overall, we aimed to guide clinical decision-making by integrating the recommendations from published guidelines and supplementing them with new data (Figure 1).

Figure 1. Proposed Treatment Algorithm for Advanced Hepatocellular Carcinoma (HCC).

Scientific societies shown are those that endorse the treatment regimen. AASLD indicates American Association for the Study of Liver Diseases; AGA, American Gastroenterological Association; ASCO, American Society of Clinical Oncology; BCLC, Barcelona Clinic Liver Cancer; EASL, European Association for the Study of the Liver; ECOG, Eastern Cooperative Oncology Group; ESMO, European Society for Medical Oncology; NCCN, National Comprehensive Cancer Network; PD, progressive disease.

^a For patients with serum α-fetoprotein >400 ng/mL before treatment.

Overview of First-Line Management of Advanced HCC

First-line management of advanced HCC has evolved substantially in the past few years, and updated global guidelines^6-11 have recognized immunology-oncology–based combinations as the preferred first-line option, retaining tyrosine kinase inhibitors (TKIs) as the alternative in case of contraindications to immunotherapies. After a decade of sorafenib being the only systemic agent associated with an increase in the median OS in patients with advanced HCC⁴ (Table 1^4,5,12-24), in 2017, lenvatinib became another option for patients after demonstrating noninferiority to sorafenib for its primary end point of OS but superiority for its secondary end points of progression-free survival (PFS) and objective response.¹² In terms of adverse event profile, lenvatinib is associated with higher-grade hypertension and proteinuria, while sorafenib is characterized by more severe hand-foot syndrome.

Table 1.

Overview of Phase 3 Clinical Trials for Systemic Therapies in Advanced HCC With Favorable Findings

	Trial characteristics		Clinical benefit				Quality metricsb
Source (trial name)	Arms	Patients, No.	OS, median (IQR), mo	PFS, median (IQR), mo	ORR, via RECIST 1.1, %	ESMO-MCBS score (magnitude of clinical benefit)a	Jadad score	Delphi score
First-line treatment
Llovet et al4 2008 (SHARP)c	Sorafenib	299	10.7 (9.4-13.3)	5.5 (4.1-6.9)	2.0	4 (Substantial)	10	9
	Placebo	303	7.9 (6.8-9.1)	2.8 (2.7-3.9)	1.0
Kudo et al12 2018 (REFLECT)	Lenvatinib	478	13.6 (12.1-14.9)	7.4 (6.9-8.8)	18.8	NE	8	6
	Sorafenib	476	12.3 (10.4-13.9)	3.7 (3.6-4.6)	6.5
Finn et al5 2020 and Cheng et al13 2022 (IMBrave150)c	Atezolizumab plus bevacizumab	336	19.2 (17.0-23.7)	6.9 (5.7-8.6)	27.3	5 (Substantial)	8	7
	Sorafenib	165	13.4 (11.4-16.9)	4.3 (4.0-5.6)	11.9
Ren et al14 2021 (ORIENT-32)	Sintilimab plus IBI305	380	Not reached	4.6 (4.1-5.7)	21.0	5 (Substantial)	8	6
	Sorafenib	191	10.4 (8.5 to not reached)	2.8 (2.7-3.2)	7.0
Abou-Alfa et al15 2022 (HIMALAYA)	STRIDE	393	16.4 (14.2-19.6)	5.4 (3.7-5.3)	20.1	3 (Intermediate)	8	6
	Durvalumab	389	16.6 (14.1-19.1)	3.7 (3.2-3.8)	17.0
	Sorafenib	389	13.8 (12.3-16.1)	4.1 (3.8-5.5)	5.1
Qin et al,16 2022c	Camrelizumab plus rivoceranib	272	22.1 (19.1-27.2)	5.6 (5.5-6.3)	25.4	4 (Substantial)	8	7
	Sorafenib	271	15.2 (13.0-18.5)	3.7 (3.8-3.7)	5.9
Qin et al17 2022 (RATIONALE-301)c	Tislelizumab	342	15.9 (13.2-19.7)	2.1 (2.1-3.5)	14.3	NE	8	7
	Sorafenib	332	14.1 (12.6-17.4)	3.4 (2.2-4.1)	5.4
Second-line treatment
Bruix et al18 2017 (RESORCE)	Regorafenib	379	10.6 (9.1-12.1)	3.1 (2.8-4.2)	11.0	4 (Substantial)	10	9
	Placebo	194	7.8 (6.3-8.8)	1.5 (1.4-1.6)	4.0
Abou-Alfa et al19 2018 (CELESTIAL)	Cabozantinib	470	10.2 (9.1-12.0)	5.2 (4.0-5.5)	4.0	3 (Intermediate)	10	9
	Placebo	237	8.0 (6.8-9.4)	1.9 (1.9-1.9)	1.0
Zhu et al20 2019 (REACH-2)	Ramucirumab	197	8.5 (7.0-10.6)	2.8 (2.8-4.1)	5.0	1 (Low)	10	9
	Placebo	95	7.3 (5.4-9.1)	1.6 (1.5-2.7)	1.0

Abbreviations: CELESTIAL, Study of Cabozantinib (XL184) vs Placebo in Subjects With Hepatocellular Carcinoma Who Have Received Prior Sorafenib; ESMO-MCBS, European Society for Medical Oncology–Magnitude of Clinical Benefit Scale; HCC, hepatocellular carcinoma; HIMALAYA, Study of Durvalumab and Tremelimumab as First-line Treatment in Patients With Advanced Hepatocellular Carcinoma; IMbrave150, A Study of Atezolizumab in Combination With Bevacizumab Compared With Sorafenib in Patients With Untreated Locally Advanced or Metastatic Hepatocellular Carcinoma; NE, not evaluable; ORIENT-32, A Randomized, Open-label, Multi-center Study to Evaluate the Efficacy and Safety of the Combination of Sintilimab and IBI305 Compared to Sorafenib in the First-Line Treatment of Patients With Advanced Hepatocellular Carcinoma; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; RATIONALE-301, A Randomized, Open-label, Multicenter Phase 3 Study to Compare the Efficacy and Safety of BGB-A317 Versus Sorafenib as First-Line Treatment in Patients With Unresectable Hepatocellular Carcinoma; REACH-2, A Study of Ramucirumab (LY3009806) Versus Placebo in Participants With Hepatocellular Carcinoma and Elevated Baseline Alpha-Fetoprotein; RECIST 1.1, Response Evaluation Criteria in Solid Tumors, version 1.1; REFLECT, A Multicenter, Open-Label, Phase 3 Trial to Compare the Efficacy and Safety of Lenvatinib (E7080) Versus Sorafenib in First-line Treatment of Participants With Unresectable Hepatocellular Carcinoma; RESORCE, Study of Regorafenib After Sorafenib in Patients With Hepatocellular Carcinoma; SHARP, A Phase III Study of Sorafenib in Patients With Advanced Hepatocellular Carcinoma; STRIDE, single tremelimumab regular interval durvalumab.

^aESMO-MCBS²¹ score of 4 or 5 indicates substantial clinical benefit; 3, intermediate clinical benefit; and 1 or 2, low clinical benefit. The ESMO-MCBS score can be calculated only for trials with a superiority design.

^bQuality assessment was performed using a modified version of the Jadad score and Delphi list.^22-24 Studies with a Jadad score of 6 or higher and a Delphi Score of 5 or higher were considered as high quality (eTable 5 in the Supplement).

^cThese studies included patients with main portal vein invasion.

IMBrave150 (A Study of Atezolizumab in Combination With Bevacizumab Compared With Sorafenib in Patients With Untreated Locally Advanced or Metastatic Hepatocellular Carcinoma) was the first trial to demonstrate substantial improvement in OS associated with the combination of atezolizumab, the programmed cell death 1 ligand 1 (PD-L1)–blocking antibody, and bevacizumab, the anti–vascular endothelial growth factor A antibody, compared with sorafenib.⁵ The reported median (IQR) OS was 19.2 (17.0-23.7) months with the atezolizumab-bevacizumab combination vs 13.4 (11.4-16.9) months with sorafenib¹³ (Table 1), and the objective response rate (ORR, measured with RECIST [Response Evaluation Criteria in Solid Tumors], version 1.1) was 30% for the atezolizumab-bevacizumab combination compared with 5% for sorafenib. The most frequent adverse events were associated with bevacizumab and included hypertension and proteinuria. All patients had to have an upper endoscopy within 6 months of starting the study drug. All-grade bleeding events (measured with the Common Terminology Criteria for Adverse Events, version 4.0 [National Cancer Institute]) were higher in the treatment vs control arm (25.2% vs 17.3%); however, grade 3 or 4 bleeding events were similar in both groups (6.4% vs 5.8%). Grade 5 gastrointestinal bleeding events were less common with the atezolizumab-bevacizumab combination (1.2%) and more common in patients with main portal vein invasion.²⁵ Similar results were recapitulated with the combination of sintilimab and IBI305, a bevacizumab biosimilar, in a population of Chinese patients.¹⁴

The combination of tremelimumab, a cytotoxic T-lymphocyte antigen-4 (CTLA-4) inhibitor, and durvalumab, a PD-L1 inhibitor—called the STRIDE (single tremelimumab regular interval durvalumab) regimen—has shown promise in the second-line treatment setting²⁶ and has now demonstrated a survival advantage vs sorafenib in the phase 3 HIMALAYA (Study of Durvalumab and Tremelimumab as First-line Treatment in Patients With Advanced Hepatocellular Carcinoma)¹⁵ trial conducted in first-line advanced HCC. This trial excluded patients with main portal vein invasion and did not require a baseline upper endoscopy given that there was no increased risk of bleeding considering the targets of the new regimen. The trial met its primary end point of improving survival with a median (IQR) OS of 16.4 (14.2-19.6) months with STRIDE vs 13.8 (12.3-16.1) months with sorafenib (Table 1). No differences were identified regarding PFS, and the ORR was 20.1% with STRIDE vs 5.1% with sorafenib. Overall, the rate of treatment-related adverse events (TRAEs) was generally lower with STRIDE, although, as expected, there were substantially more immune-related adverse events with STRIDE, and 20% of patients required corticosteroids for these indications.¹⁵

Although neither has regulatory approval, 2 phase 3 studies recently demonstrated noninferiority of immune checkpoint inhibitors (ICIs) vs sorafenib. The RATIONALE-301 (A Randomized, Open-label, Multicenter Phase 3 Study to Compare the Efficacy and Safety of BGB-A317 Versus Sorafenib as First-Line Treatment in Patients With Unresectable Hepatocellular Carcinoma) trial¹⁷ found tislelizumab, a programmed cell death 1 (PD-1) antibody, to be noninferior to sorafenib; similarly, the HIMALAYA trial¹⁵ had a durvalumab monotherapy arm that met this secondary end point. These regimens are both safe and well tolerated, with an ORR of approximately 15% for the ICI. Single-agent pembrolizumab currently has accelerated approval in the US based on a single-arm trial,²⁷ while confirmatory studies are being reviewed.^28,29 The combination of camrelizumab and rivoceranib has been reported as superior to sorafenib in a recent randomized clinical trial¹⁶ (Table 1).

The long-awaited trials LEAP-002 (A Phase 3 Multicenter, Randomized, Double-blinded, Active-controlled, Clinical Study to Evaluate the Safety and Efficacy of Lenvatinib in Combination With Pembrolizumab Versus Lenvatinib in First-line Therapy of Participants With Advanced Hepatocellular Carcinoma)³⁰ and COSMIC-312 (Study of Cabozantinib in Combination With Atezolizumab Versus Sorafenib in Subjects With Advanced HCC Who Have Not Received Previous Systemic Anticancer Therapy)³¹ did not reach their prespecified end points (eTable 1 in the Supplement). COSMIC-312³¹ was a first-line phase 3 trial comparing atezolizumab plus cabozantinib vs sorafenib. While the combination demonstrated longer PFS compared with sorafenib, it did not improve the median (IQR) OS (15.4 [13.7-17.7] months vs 15.5 [12.1 to not estimable] months; hazard ratio [HR], 0.90 [95% CI, 0.69-1.18]). The trial did not demonstrate any synergy between the drugs, which is reflected by a low ORR for the combination compared with anti–PD-L1 monotherapy (11% vs 14%-18%).^{15,17,28,32,33} While there was no clear explanation for this finding, in experimental studies, cabozantinib substantially increased neutrophil infiltration while reducing intratumoral CD8+PD-1+ T-cell proportions.³⁴ In addition, cabozantinib was used at a lower dose (40 mg) compared with previous studies. Similarly, LEAP-002 was a double-blind, phase 3 randomized clinical trial³⁰ of lenvatinib plus pembrolizumab vs lenvatinib alone that reported non-significant differences in OS (21.2 months vs 19.0 months). While the reasons for not reachingsuperiority were complex, including the over-performance of lenvatinib, which was likely associated with exposure of 9 months in the LEAP-002 trial compared with exposure of just 3 months in the REFLECT (A Multicenter, Open-Label, Phase 3 Trial to Compare the Efficacy and Safety of Lenvatinib (E7080) Versus Sorafenib in First-line Treatment of Participants With Unresectable Hepatocellular Carcinoma) trial¹² and a high percentage of patients receiving effective second-line therapies (52%), the unprecedented OS for the lenvatinib arm places this TKI in a position of potential primacy among the approved TKIs.

Scientific Societies’ Recommendations for First-Line Therapies

As the guidelines show (Figure 1, Table 2), the main recommendations for first-line therapies are as follows. First, for patients without contraindications for immunotherapy, the first option is atezolizumab plus bevacizumab.^6-11 The AASLD guidance and latest NCCN guidelines published after the HIMALAYA trial¹⁵ proposed the use of tremelimumab plus durvalumab as an alternative to atezolizumab plus bevacizumab in cases with a high risk of gastrointestinal bleeding that might contraindicate the use of bevacizumab.^6,8 Second, for patients with contraindications for immunotherapy, the first-line option is lenvatinib or sorafenib,^6-11 with no established preference.

Table 2.

Summary of Scientific Society Guidelines or Guidance Documents for Treatment of Advanced HCC

Source, (society)	Type of recommendation	Child-Pugh class A
		First-line treatment	Second-line treatment
Singal et al,6 2023 (AASLD)	Scientific society guidance based on expert opinion	Preferred treatment is atezolizumab plus bevacizumab or tremelimumab plus durvalumab.	After first-line atezolizumab plus bevacizumab, preferred treatment is sorafenib or lenvatinib. Cabozantinib, regorafenib, or ipilimumab plus nivolumab may be used.
			After first-line tremelimumab plus durvalumab, preferred treatment is sorafenib or lenvatinib.
		If contraindications to atezolizumab plus bevacizumab, lenvatinib or sorafenib may be used.	After first-line sorafenib or lenvatinib, preferred treatment is cabozantinib, regorafenib, or ramucirumab (AFP ≥400 ng/mL). Ipilimumab plus nivolumab or pembrolizumab may be used.
		First-line treatment, Child-Pugh class B preferred treatment is sorafenib or lenvatinib single-agent ICI (durvalumab, pembrolizumab, or nivolumab).
Vogel and Martinelli,10 2021 (ESMO)	Scientific society guidelines based on ESMO-MCBS scorea	Preferred treatment is atezolizumab plus bevacizumab.	After first-line atezolizumab plus bevacizumab, sorafenib, lenvatinib, cabozantinib, regorafenib, or ramucirumab (AFP ≥400 ng/mL) may be used.
		If contraindications to atezolizumab plus bevacizumab, lenvatinib or sorafenib may be used.	After first-line sorafenib or lenvatinib, cabozantinib, regorafenib, or ramucirumab (AFP ≥400 ng/mL) may be used.
Bruix et al,11 2021 (EASL)	Scientific society guidance based on expert opinion	Preferred treatment is atezolizumab plus bevacizumab.	After first-line atezolizumab plus bevacizumab, preferred treatment is sorafenib or lenvatinib.
		If contraindications to atezolizumab plus bevacizumab, lenvatinib or sorafenib may be used.	After first-line sorafenib or lenvatinib, cabozantinib, regorafenib, or ramucirumab (AFP ≥400 ng/mL) may be used.
Gordan et al,7 2020 (ASCO)	Scientific society guidelines based on GRADEa	Preferred treatment is atezolizumab plus bevacizumab.	After first-line atezolizumab plus bevacizumab, preferred treatment is sorafenib or lenvatinib. Cabozantinib or regorafenib may be used.
		If contraindications to atezolizumab plus bevacizumab, lenvatinib or sorafenib may be used.	After first-line sorafenib or lenvatinib, preferred treatment is cabozantinib, regorafenib, or ramucirumab (AFP ≥400 ng/mL) or atezolizumab plus bevacizumab. Pembrolizumab or nivolumab may be used.
Su et al,9 2022 (AGA)	Scientific society guidelines based on GRADEa	Preferred treatment is atezolizumab plus bevacizumab.	After first-line treatment, cabozantinib, regorafenib, pembrolizumab, or ramucirumab (AFP ≥400 ng/mL) may be used.
		If contraindications to atezolizumab plus bevacizumab, lenvatinib or sorafenib may be used.
NCCN,8 2023	Scientific society guidelines	Preferred treatment is atezolizumab plus bevacizumab or tremelimumab plus durvalumab. Lenvatinib, sorafenib, durvalumab, or pembrolizumab may be used.	After first-line atezolizumab plus bevacizumab, preferred treatment (category 1a) is cabozantinib,a regorafenib,a ramucirumab (for AFP >400 ng/mL),a lenvatinib, or sorafenib. Ipilimumab plus nivolumab or pembrolizumab may be used.
		First-line treatment, Child-Pugh class B preferred treatment is sorafenib (Child-Pugh class B7) or nivolumab. Atezoilzumab plus bevacizumab may be used in Child Pugh class B in certain circumstances

Abbreviations: AASLD, American Association for the Study of Liver Diseases; AFP, α-fetoprotein; AGA, American Gastroenterological Association; ASCO, American Society of Clinical Oncology; EASL, European Association for the Study of the Liver; ESMO, European Society for Medical Oncology; ESMO-MCBS, European Society for Medical Oncology–Magnitude of Clinical Benefit Scale; GRADE, Grading of Recommendations, Assessment, Development, and Evaluation; HCC, hepatocellular carcinoma; ICI, immune checkpoint inhibitor; NCCN, National Comprehensive Cancer Network.

^aFor details on quality of evidence used in guideline development, see eTable 2 in the Supplement.

Overview of Second-Line Treatment of Advanced HCC

Three drugs have demonstrated their superiority to placebo in phase 3 clinical trials in the second-line setting after progression on sorafenib: cabozantinib, regorafenib, and ramucirumab. The phase 3 trial of cabozantinib (CELESTIAL [Study of Cabozantinib (XL184) vs Placebo in Subjects With Hepatocellular Carcinoma Who Have Received Prior Sorafenib]¹⁹) enrolled patients who received up to 2 previous systemic regimens. In the overall population, median (IQR) OS was prolonged by 2.2 months (10.2 [9.1-12.0] months with cabozantinib vs 8.0 [6.8-9.4] months with placebo; HR, 0.76 [95% CI, 0.63-0.92]) (Table 1). Given that sorafenib and regorafenib have similar molecular targets and adverse event profiles, the phase 3 trial of regorafenib (RESORCE [Study of Regorafenib After Sorafenib in Patients With Hepatocellular Carcinoma]¹⁸) included a narrower population of patients receiving second-line treatment who had tolerated but progressed with first-line sorafenib. Median (IQR) OS was improved by almost 3.0 months (10.6 [9.1-12.1] months with regorafenib vs 7.8 [6.3-8.8] months with placebo; HR, 0.63 [95% CI, 0.50-0.79]) (Table 1). Ramucirumab was initially evaluated in a phase 3 trial that did not attain its predefined end points (REACH [A Study of Ramucirumab Drug Product and Best Supportive Care Versus Placebo and BSC as 2nd-Line Treatment in Participants With Hepatocellular Carcinoma After 1st-Line Therapy With Sorafenib]³⁵), although a subsequent phase 3 trial in patients with an α-fetoprotein (AFP) of 400 ng/mL or greater (REACH-2²⁰) did attain its predefined end points. Median (IQR) OS was prolonged by 1.2 months (8.5 [7.0-10.6] months with ramucirumab vs 7.3 [5.4-9.1] months with placebo; HR, 0.71 [95% CI, 0.53-0.95]) (Table 1). To date, the approval of ramucirumab in patients receiving second-line treatment with an AFP greater than 400 ng/mL represents the only biomarker-indicated approval in HCC.

In parallel, 2 phase 3 trials of pembrolizumab were conducted: KEYNOTE 240 (A Phase III Study of Pembrolizumab vs Best Supportive Care as Second-Line Therapy in Subjects With Previously Systemically Treated Advanced Hepatocellular Carcinoma),²⁸ a global trial that did not reach statistical significance per the prespecified statistical plan, and KEYNOTE-394 (A Phase III Randomized Double-blind Study of Pembrolizumab Plus Best Supportive Care Versus Placebo Plus Best Supportive Care as Second-Line Therapy in Asian Subjects With Previously Systemically Treated Advanced Hepatocellular Carcinoma),²⁹ an Asian trial that achieved its primary OS end point. The dual ICI combination of ipilimumab, an anti–CTLA-4 therapy, and nivolumab (anti–PD-1) received approval in the US on the basis of its response rate of 32% in an arm of the CheckMate 040 (An Immuno-therapy Study to Evaluate the Effectiveness, Safety and Tolerability of Nivolumab or Nivolumab in Combination With Other Agents in Patients With Advanced Liver Cancer) trial,²⁶ which was numerically higher than in studies of anti–PD-1 alone, and the responses were durable. More than half of patients experienced serious TRAEs that required the use of corticosteroids in 50% of cases. Regular approval is pending for the confirmatory CheckMate 9DW trial³⁶ (A Study of Nivolumab in Combination With Ipilimumab in Participants With Advanced Hepatocellular Carcinoma).

Sequential Treatment Strategies

Rapid changes in the first-line management of advanced HCC have revealed knowledge gaps on how to sequence drugs after the establishment of immuno-oncology–based combinations. In reality, all approved drugs in the second-line setting (cabozantinib,¹⁹ regorafenib,¹⁸ ramucirumab,³⁵ and in the US, pembrolizumab²⁷ and ipilimumab plus nivolumab²⁶) received approval when only sorafenib was available. Their true activity after lenvatinib is not known, especially after newly approved ICI-doublet regimens. However, all of these drugs have proven that despite similar mechanisms of action, their sequential use was associated with increased survival in the settings in which they were studied. Although high-level evidence is lacking to support the use of the available drugs in sequence after newly established regimens, scientific societies endorse their use at disease progression regardless of prior treatment. Borrowing from other malignant neoplasms, for which a number of active regimens are approved (ie, breast cancer³⁷), recent advances in first-line management do not disqualify the use of previously proven active drugs.

Based on substantial activity, all eligible patients should receive an anti–PD-L1–based strategy during their disease course, with the largest benefit being in the first-line setting (Figure 1). The question is how patients should be treated afterward considering that the mechanisms of resistance are likely different between those who get no clinical benefit (primary resistance) and those with acquired resistance. Small studies have suggested some responses with combinations of CTLA-4 with anti–PD-1 after prior ICIs.³⁸ Recent clinical data reported that regorafenib has similar activity regardless of the first-line regimen that was used, including after prior ICI.³⁹ A study⁴⁰ analyzing ramucirumab after prior non-sorafenib regimens found similar safety and clinical benefit as those found in the phase 3 REACH-2 (A Study of Ramucirumab [LY3009806] Versus Placebo in Participants With Hepatocellular Carcinoma and Elevated Baseline Alpha-Fetoprotein) trial.²⁰ Ultimately, adopting a biomarker-based strategy to guide the choice of treatment regimen will be of great value. In this sense, 3 gene signatures (inflammatory signature,⁴¹ IFNAP [interferon and antigen presentation] signature,⁴² and T-effector signature⁴³) have been reported to identify patients responding to single-agent ICI or the atezolizumab-bevacizumab combination.

Scientific Societies’ Recommendations for Second-Line Therapies

As the guidelines show (Figure 1, Table 2; eTable 2 in the Supplement), the main recommendations for second-line therapies are as follows. First, following first-line atezolizumab plus bevacizumab, use of a multitarget TKI is recommended based on expert consensus opinion. The AASLD,⁶ ASCO,⁷ and EASL¹¹ guidelines state that sorafenib or lenvatinib are the preferred option, while the ESMO¹⁰ and NCCN⁸ guidelines do not provide a distinct recommendation after atezolizumab plus bevacizumab among 5 choices (sorafenib, lenvatinib, cabozantinib, regorafenib, and ramucirumab). Second, following first-line tremelimumab plus durvalumab, there is no clear guidance as this regimen is too new to be incorporated into most of the guidelines. Given that these patients are naive to any form of anti–vascular endothelial growth factor–targeted therapy, the AASLD guidance⁶ recommends the use of sorafenib or lenvatinib in the second-line setting in this population. Third, following first-line sorafenib or lenvatinib, the guidelines do not distinguish second-line recommendations after prior sorafenib or lenvatinib and support use of cabozantinib, regorafenib, or ramucirumab (AFP >400 ng/mL) as the preferred options.^6-11 The AASLD,⁶ ASCO,⁷ NCCN,⁸ and AGA⁹ guidelines also list immunotherapy as an option in the second-line setting despite limited high-level evidence in this context. As most patients with HCC who are candidates for immunotherapy currently receive immunotherapy-containing regimens in the first-line setting, it is likely that patients who are being considered for second-line immunotherapy would not have had access to immunotherapy-containing regimens when they started first-line therapy. The ASCO⁷ and AGA⁹ guidelines consider atezolizumab plus bevacizumab as an option for patients who have not been exposed due to the lack of availability of this regimen in the first-line setting after progression with TKIs.

Comparing Treatment Regimens

With the advent of combination therapies with superior clinical benefit compared with sorafenib, the previously established hierarchy and treatment flowchart⁴⁴ has been substantially modified (Figure 1). Despite the numerous phase 3 randomized clinical trials available (Table 1; eTables 3 and 4 in the Supplement), some of the recommendations from the guidelines lack head-to-head comparisons of first-line regimens or second-line regimens. This review supplements the current recommendations with (1) information on the magnitude of clinical benefit for each drug in the first-line and second-line settings using accepted metrics, such as the ESMO-MCBS score,²¹ and (2) a direct comparison of the clinical benefit (OS and PFS) and safety (total and TRAEs) between new first-line therapies vs sorafenib for clinical benefit and vs atezolizumab plus bevacizumabfor safety using a network meta-analysis.

ESMO-MCBS Score for First-Line and Second-Line Therapies

First, we scored each treatment regimen using the ESMO-MCBS 1.1,²¹ a tool that captures the magnitude of clinical benefit, adverse events, and quality-of-life outcomes associated with anticancer therapies and that has undergone rigorous statistical evaluation.⁴⁵ It has been applied by ESMO to the registration studies of all new cancer treatments that have been approved by the European Medicines Agency since 2015 and by the US Food and Drug Administration since 2020.⁴⁶ The ESMO-MCBS scale ranges from 1 to 5, with scores of 4 or 5 indicating substantial, 3 indicating intermediate, and 1 or 2 indicating low magnitude of clinical benefit. Systemic therapies for advanced HCC that were approved prior to 2021 are included in the updated ESMO guidelines.¹⁰ Additionally, we calculated ESMO-MCBS scores for 3 phase 3 trials of systemic therapies for advanced HCC that have not yet been assessed by ESMO (Table 1).

Overall, in the first-line setting, atezolizumab plus bevacizumab and sorafenib were scored by the ESMO guidelines as providing substantial benefit (ESMO-MCBS scores of 5 and 4, respectively). We also calculated the ESMO-MCBS scores for those drug combinations that have not been incorporated into the guidelines. Thus, the ESMO-MCBS scores were 5 for sintilimab plus IBI305, 4 for camrelizumab plus rivoceranib, and 3 for tremelimumab plus durvalumab. In the second-line setting, the ESMO guidelines ranked regorafenib first with a score of 4, followed by cabozantinib with a score of 3 (intermediate) and ramucirumab with a score of 1 (low). An overview of the ESMO-MCBS score for each regimen is provided in Table 1.

Network Meta-Analysis

In this network meta-analysis, we included 4 trials representing a total of 2397 patients (eFigure A in the Supplement) comparing atezolizumab plus bevacizumab,¹³ sintilimab plus IBI305,¹⁴ tremelimumab plus durvalumab,¹⁵ and camrelizumab plus rivoceranib¹⁶ to sorafenib. Two of the trials^13,15 were global, and 2 were conducted primarily in Chinese patients, most of whom had hepatitis B virus–related HCC^14,16 (eTable 3 in the Supplement). The clinical benefit analysis was performed using HRs and 95% CIs for OS and PFS from the published data (eTable 4 in the Supplement). For the safety analysis, we calculated the relative risk (RR) of TRAEs using the proportion of total and severe (≥grade 3) TRAEs reported in each trial (eTable 4 and eMethods in the Supplement).

In the first-line setting using sorafenib as the common comparator, all 4 combination regimens had substantially better OS, with an HR ranging from 0.57 (95% CI, 0.43-0.75) to 0.78 (95% CI, 0.65-0.93). While sintilimab plus IBI305 was ranked first, there were no significant differences between the different combination regimens (Figure 2A; eFigure B in the Supplement). In addition, camrelizumab plus rivoceranib, atezolizumab plus bevacizumab, and sintilimab plus IBI305 were associated with a significantly reduced risk of progression (as indicated by the PFS analysis) compared with sorafenib, whereas this was not the case for tremelimumab plus durvalumab (Figure 2B; eFigure C in the Supplement).

Figure 2. Network Meta-Analysis of Clinical Benefit and Safety of Combined Regimens vs Sorafenib in First-Line Treatment of Advanced Hepatocellular Carcinoma.

HR indicates hazard ratio; OS, overall survival; PFS, progression-free survival; RR, relative risk; SUCRA, surface under the cumulative ranking; TRAE, treatment-related adverse event.

In comparing the safety profiles of these regimens, we used atezolizumab plus bevacizumab, which is considered the standard of care by the guidelines, as the active comparator. This combination had a significantly lower risk of TRAEs compared with camrelizumab plus rivoceranib, sintilimab plus IBI305, and sorafenib and showed a similar profile as tremelimumab plus durvalumab (Figure 2C; eFigure D in the Supplement). Furthermore, when considering the RR for serious TRAEs (≥grade 3), camrelizumab plus rivoceranib was associated with a significantly higher RR of serious TRAEs compared with atezolizumab plus bevacizumab (RR, 1.59; 95% CI, 1.25-2.03; P < .001) (Figure 2D; eFigure E in the Supplement). Overall, while there were no significant OS differences in the clinical benefit of all 4 combination regimens, camrelizumab plus rivoceranib was associated with a substantially higher risk of TRAEs compared with atezolizumab plus bevacizumab.

Special Populations

The phase 3 trials assessing systemic therapies in HCC have been restricted to patients with Child-Pugh class A liver disease. Patients with HCC and Child-Pugh class B liver disease have competing causes of mortality from liver failure and the cancer, and few prospective data are available to aid in the decision-making. The critical factor for survival in patients with Child-Pugh class B liver disease is the occurrence of liver decompensation, which is characterized by the presence of ascites in 75% of cases.⁴⁷ Therefore, patients with Child-Pugh class B liver disease without ascites, generally with a Child-Pugh score of B7, have a similar natural history related to liver dysfunction as patients with Child-Pugh class A liver disease. Clinical data and expert opinion suggest that Child-Pugh class B7 liver disease associated with cancer-related hypoalbuminemia and/or hyperbilirubinemia may sometimes benefit from systemic therapies.⁴⁸ For these patients, the safety of sorafenib is supported by prospective registry data,⁴⁹ and the safety of nivolumab is supported by cohort 5 of the CheckMate 040 phase 1 and 2 clinical trial,⁵⁰ although clinical benefit cannot be inferred from those studies. A retrospective analysis from the phase 3 CELESTIAL randomized trial^19,51 demonstrated a similar benefit with cabozantinib over placebo in a subgroup of patients whose liver function declined to Child-Pugh class B at week 8 of therapy as in the overall study population, supporting the continuation of TKI therapy in these patients. Only the AASLD,⁶ NCCN⁸ and AGA⁹ guidelines provide systemic therapy recommendations for patients with Child-Pugh class B HCC (Table 2), while the remaining scientific societies provide no systemic therapy recommendation, instead endorsing a cautious approach and emphasizing shared decision-making with a multidisciplinary team. In this sense, the AASLD⁶ guidelines recommend sorafenib, lenvatinib, or single-agent ICI for well-selected candidates with Child-Pugh class B liver disease.

In the first-line setting, treatment of patients with main portal vein invasion can be derived from a limited number of phase 3 trials accepting this inclusion criteria.^4,5,13,16,17 As the REFLECT trial¹² excluded patients with main portal vein invasion, most guidelines (ESMO,¹⁰ EASL,¹¹ AASLD,⁶ ASCO,⁷ and AGA⁹) recommend sorafenib over lenvatinib in this subpopulation. In the IMBrave150 trial,²⁵ grade 5 gastrointestinal bleeding was infrequent but more common in patients with main portal vein invasion who were treated with atezolizumab plus bevacizumab. Further data on the main portal vein invasion subgroup from 2 trials^16,17 are forthcoming.

Discussion

Alternative effective combinations in the first-line setting that were explored in Asia do not add a distinct clinical benefit or mechanism of action and, in some cases (ie, camrelizumab plus rivoceranib), are associated with a worse safety profile. Guideline recommendations for first-line strategies in patients with contraindications for immunotherapies or for second-line therapies after immunotherapy-based combinations are dominated by lenvatinib or sorafenib as the preferred options. Recent clinical benefit data of lenvatinib in a phase 3 trial³⁰ provide the rationale to consider this regimen with some primacy among TKIs. Few guidelines also advocate for the use of other TKIs, such as cabozantinib or regorafenib. In the second-line setting after immunotherapies, there is agreement that tumors progressing after the use of sorafenib or lenvatinib should be treated with either regorafenib, cabozantinib, or ramucirumab.

Additional research is needed to establish a standard of care for patients with Child-Pugh class B liver disease, particularly those without liver decompensation. Future trials are expected to challenge the new benchmarks reported by using first-line immunotherapy-based combinations as the control arm since sorafenib can no longer be considered the standard control arm. In this sense, a robust trial design should be developed with conservative estimations of effect size and restrictive use of primary end points. Similarly, regulatory agencies may provide a path for rescuing potentially effective drugs based on post hoc analysis of phase 3 trials.⁵² The recent data establishing the clinical benefit of adjuvant atezolizumab plus bevacizumab after resection or ablation vs best supportive care⁵³ for recurrence-free survival represent a major breakthrough and define the onset of a new era in the management of early HCC. Such information will influence subsequent therapies in case of tumor recurrence.

Limitations

The limitations of this review are related to the nature of the network meta-analysis.⁵⁴ Most of the evidence was derived from indirect comparisons, and the analysis was on a trial level rather than an individual patient level. Nonetheless, network meta-analysis is regarded as a useful tool to compare the clinical benefit and safety of novel systemic drug regimens in a rapidly evolving treatment landscape, wherein head-to-head comparisons are unlikely to become available.^55,56

Conclusions

There is a major consensus among scientific societies that atezolizumab plus bevacizumab has a primacy as the recommended treatment of choice in the first-line setting, whereas lenvatinib or sorafenib are considered alternative options in cases of contraindications for immunotherapy. The role of tremelimumab plus durvalumab and the specific niche of its target population (ie, patients at risk of gastrointestinal bleeding) need to be further defined. Findings from this narrative review may aid clinicians in decision-making and guide them through a rapidly evolving and complex treatment landscape.