Evolution of systemic treatment for advanced hepatocellular carcinoma

Abstract

Advanced hepatocellular carcinoma (HCC) was considered an inherently refractory tumor in the chemotherapy era (1950–2000). However, systemic therapy has evolved to molecular targeted therapy and immunotherapy, and nine treatment regimens have been approved worldwide during the past 20 years. The approved regimens target tumor angiogenesis or tumor immunity, the two cancer hallmarks. Recently, the combination of atezolizumab (antiprogrammed cell death ligand 1) and bevacizumab (anti‐vascular endothelial growth factor) has improved the efficacy of systemic therapy in treating advanced HCC without excessive toxicities or deterioration of quality of life. This review summarizes the major advances in systemic therapy and provides future perspectives on the next‐generation systemic therapy for advanced HCC.

1. INTRODUCTION

Hepatocellular carcinoma (HCC) was considered until recently a tumor type for which systemic therapy had a limited effect. The combination of an immune checkpoint inhibitor (ICI) and an anti‐vascular endothelial growth factor (anti‐VEGF) monoclonal antibody (atezolizumab and bevacizumab), approved in 2020, has increased the tumor response rate to approximately 30% and overall survival (OS) to approximately 20 months. Now, advanced HCC is considered highly treatable. This review describes the astonishing evolutional process (Figure 1).

FIGURE 1.

Evolution of systemic treatment for advanced hepatocellular carcinoma approved by US Food and Drug Administration. CTLA‐4, cytotoxic T‐lymphocyte antigen‐4; ICI, immune checkpoint inhibitor; PD‐1, programmed cell death protein 1; PD‐L1, programmed cell death ligand 1; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor. Created with BioRender.com

2. CYTOTOXIC CHEMOTHERAPY

Cytotoxic chemotherapy has a limited therapeutic effect on HCC. Multiple regimens of cytotoxic chemotherapy, including monotherapy and multiagent combination therapy, have been investigated in patients with advanced HCC. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ However, no cytotoxic chemotherapy regimen has resulted in survival benefits in Phase III randomized controlled studies. ⁶ , ⁷ , ⁸ Even in the era of molecular targeted therapy, a combination of cytotoxic chemotherapy and targeted agents failed to provide additional benefits. ⁹

The failure of cytotoxic chemotherapy may be attributable to several factors. First, patients with advanced HCC usually have various degrees of hepatic dysfunction secondary to the underlying liver disease. The preexisting hepatic dysfunction renders chemotherapy too toxic, particularly to those with liver cirrhosis or hypersplenism. Second, HCC cells are often intrinsically resistant to cytotoxics. ¹⁰ Increased dihydropyrimidine dehydrogenase activity, ¹¹ high P‐glycoprotein expression, ¹² and complex xenobiotic‐metabolizing enzyme expression ¹³ are examples of reported mechanisms of intrinsic resistance to cytotoxic chemotherapy in HCC.

3. MOLECULAR TARGETED THERAPY

Since the year 2000, molecular targeted drugs have become key treatment agents in several cancer types. The strategy involves identifying the driver mutation of a specific cancer type and then targeting the driver mutation‐related pathways with drugs. However, no definite driver mutations have been identified for HCC, and common genetic alterations in HCC—such as TP53, CTNNB1, ARID1A, CDKN2A, SMARCA4, and SF3B1—are not druggable. ¹⁴ , ¹⁵ Attempts at using inhibitors of common oncogenic pathways, such as epidermal growth factor receptor inhibitor erlotinib and mTOR inhibitor everolimus, have failed to demonstrate survival benefits. ¹⁶ , ¹⁷ , ¹⁸ The first biomarker‐driven phase III “METIV‐HCC” study, which investigated MET inhibitor tivantinib in patients with high‐MET‐expression HCC, ended with negative results. ¹⁹

The only effective molecular targeted therapy for HCC is the use of antiangiogenics (Table 1). Hypervascularity is a well‐known feature of HCC. ²⁰ , ²¹ The VEGF/vascular endothelial growth factor receptor (VEGFR) pathway is the major signaling pathway of tumor angiogenesis. ²² The VEGF/VEGFR pathway can be inhibited using two strategies: multitarget kinase inhibitors and monoclonal antibodies (Figure 2).

TABLE 1.

Published phase III trials of molecular targeted therapy for advanced HCC

Trial	N	Enrollment time	Experimental arm	Control arm	OS (months)	TTP/PFS (months)	RR% DCR% RECIST	RR% DCR% mRECIST
First‐line
SHARP 23 NCT00105443	602	2005–2006	Sorafenib	Placebo	10.7 vs. 7.9 HR: 0.69 p < 0.001*	a 5.5 vs. 2.8 HR: 0.58 p < 0.001	2 vs. 1 43 vs. 32
Asia‐Pacific 24 NCT00492752	226	2005–2007	Sorafenib	Placebo	6.5 vs. 4.2 HR: 0.68 p = 0.014	a 2.8 vs. 1.4 HR: 0.57 p = 0.0005	3.3 vs. 1.3 35 vs. 17
SUN1170 25 NCT00699374	1074	2008–2010	Sunitinib	Sorafenib	7.9 vs. 10.2 HR: 1.30 p = 0.0014	b 3.6 vs. 3.0 HR: 1.13 p = 0.2286	6.6 vs. 6.1 51 vs. 52
BRISK‐FL 26 NCT00858871	1150	2009–2011	Brivanib	Sorafenib	9.9 vs. 9.5 HR: 1.06 p = 0.3730	a 4.1 vs. 4.2 HR: 1.01 p = 0.8532		9 vs. 12 65 vs. 66
LIGHT 27 NCT01009593	1035	2010–2011	Linifanib	Sorafenib	9.1 vs. 9.8 HR: 1.046	a 5.4 vs. 4.1 HR: 0.759 p = 0.001	13.0 vs. 6.9
SEARCH 16 NCT0901901	720	2009–2011	Sorafenib + erlotinib	Sorafenib + placebo	9.5 vs. 8.5 HR: 0.929 p = 0.408	a 3.2 vs. 4.0 HR: 1.135 p = 0.18	6.6 vs. 3.9 44 vs. 53
REFLECT 29 NCT01761266	954	2013–2015	Lenvatinib	Sorafenib	13.6 vs. 12.3 HR: 0.92c (95% CI 0.79–1.06)	b 7.4 vs. 3.7 HR: 0.66 p < 0.0001	19 vs. 7 73 vs. 59	24 vs. 9 76 vs. 61
Second‐line or later
BRISK‐PS 28 NCT00825955	395	2009–2011	Brivanib	Placebo	9.4 vs. 8.2 HR: 0.89 p = 0.3307	a 4.2 vs. 2.7 HR: 0.56 p < 0.001	12 vs. 2 71 vs. 49
EVOLVE‐1 17 NCT01035229	546	2010–2012	Everolimus	Placebo	7.6 vs. 7.3 HR: 1.05 p = 0.68	a 3.0 vs. 2.6 HR: 0.93	2.2 vs. 1.6 56 vs. 45
METIV‐HCC 19 NCT01755767	340	2012–2015	Tivantinib	Placebo	8.4 vs. 9.1 HR: 0.97 p = 0.81	b 2.1 vs. 2.0 HR: 0.96 p = 0.72	0 vs. 0 50 vs. 50
REACH 35 NCT01140347	565	2010–2013	Ramucirumab	Placebo	9.2 vs. 7.6 HR: 0.87 p = 0.14	b 2.8 vs. 2.1 HR: 0.63 p < 0.0001	7 vs. <1 56 vs. 46
RESORCE 30 NCT01774344	573	2013–2015	Regorafenib	Placebo	10.6 vs. 7.8 HR: 0.63 p < 0.0001*	b 3.1 vs. 1.5 HR: 0.46 p < 0.0001		11 vs. 4 65 vs. 36
REACH‐2 36 NCT02435433	292	2015–2017	Ramucirumab	Placebo	8.5 vs. 7.3 HR: 0.71 p = 0.0199*	b 2.8 vs. 1.6 HR: 0.45 p < 0.0001	5 vs. 1 60 vs. 39
CELESTIAL 31 NCT01908426	707	2013–2017	Cabozantinib	Placebo	10.2 vs. 8.0 HR: 0.76 p = 0.005*	b 5.2 vs. 1.9 HR: 0.44 p < 0.001	4 vs. 1 64 vs. 33

Abbreviations: DCR, disease control rate; HCC, hepatocellular carcinoma; HR, hazard ratio; mRECIST, modified RECIST for hepatocellular carcinoma; N, patient number; OS, overall survival; PFS, progression‐free survival; RECIST, Response Evaluation Criteria in Solid Tumors; RR, response rate; TTP, time to progression.

^a The reported endpoint was time to progression.

^b The reported endpoint was progression‐free survival.

^c The primary endpoint was noninferiority design.

^* p value met statistical significance for the primary endpoint.

FIGURE 2.

Multifaceted unfavorable effects of VEGF/VEGFR signaling. Different mechanisms of action and examples of antiangiogenic therapies. MDSC, myeloid‐derived suppressor cells; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor. Created with BioRender.com

3.1. Multikinase VEGFR inhibitors

Sorafenib—a multikinase inhibitor of VEGFR‐1, VEGFR‐2, VEGFR‐3, platelet‐derived growth factor receptor (PDGFR)‐α, PDGFR‐β, RAF‐1, B‐RAF, c‐KIT, FMS‐like tyrosine kinase 3 (FLT3), and RET—is the first approved systemic therapy drug for advanced HCC. Although the objective response rate (ORR) for sorafenib has been modest (2%–3%), the survival benefit of sorafenib over placebo was consistent in two parallel studies, one conducted in Western countries (SHARP study) and the other in the Asia‐Pacific region (Asia‐Pacific study). ²³ , ²⁴

The success of sorafenib in 2007 spurred great enthusiasm to test different multikinase inhibitors with VEGFR inhibition in either the first‐line or second‐line setting ²⁵ , ²⁶ , ²⁷ , ²⁸ (Table 1). However, all phase III studies failed until 2017.

In 2018, lenvatinib became the second drug approved for first‐line treatment of advanced HCC. The phase III REFLECT trial tested noninferiority of OS with lenvatinib (a VEGFR‐1, VEGFR‐2, VEGFR‐3, fibroblast growth factor receptors (FGFR)‐1, FGFR‐2, FGFR‐3, FGFR‐4, PDGFR‐α, KIT, and RET inhibitor) to sorafenib. The median OS for lenvatinib was 13.6 months, which was noninferior to sorafenib (12.3 months; hazard ratio [HR]: 0.92, 95% confidence interval: 0.79–1.06). ²⁹

In the second‐line setting, regorafenib (a VEGFR‐1, VEGFR‐2, VEGFR‐3, PDGFR‐β, RAF‐1, B‐RAF, c‐KIT, FLT3, and RET inhibitor) showed significant survival benefit over placebo with median OS of 10.6 months versus 7.8 months and a HR of 0.63 in patients who tolerated and progressed on sorafenib (RESORCE trial). ³⁰ Subsequently, cabozantinib (a VEGFR‐1, VEGFR‐2, VEGFR‐3, MET, and AXL inhibitor) showed superiority to placebo (median OS: 10.2 vs. 8.0 months, HR: 0.76) in patients with disease progression after one or two systemic regimens including sorafenib (“CELESTIAL” trial). ³¹

3.2. Monoclonal VEGF/VEGFR antibodies

Bevacizumab, an anti‐VEGF antibody, is the first drug in the class. Several phase II studies have evaluated the efficacy of bevacizumab either as monotherapy or in combination with cytotoxic chemotherapy. ³² , ³³ , ³⁴ Although some studies have shown signs of efficacy, no confirmatory phase III study has been conducted.

Ramucirumab, an anti‐VEGFR‐2 antibody, failed to show survival benefit over placebo in the second‐line setting (REACH trial). ³⁵ In a subgroup analysis, ramucirumab showed potential survival benefits in patients with alpha‐fetoprotein (AFP) ≥400 ng/ml. Therefore, another phase III trial (REACH‐2) was conducted in patients with AFP ≥400 ng/ml. Ramucirumab led to a superior median OS over placebo (8.5 vs. 7.3 months, HR = 0.71). ³⁶

Both multikinase VEGFR inhibitors and anti‐VEGF/VEGFR antibody share common toxicities such as hypertension, bleeding, and proteinuria, which are associated with VEGF/VEGFR inhibition. ³⁷ Because of the broad range of non‐VEGFR off‐target effects, multikinase VEGFR inhibitors are often associated with various degrees of toxicity such as hand–foot skin reaction, diarrhea, and fatigue. ³⁸ These toxicities are generally dose dependent ³⁷ , ³⁹ and can be managed with dose modification or change in drug administration schema. ⁴⁰

4. IMMUNOTHERAPY

Immune checkpoints are the molecular basis for physiological immune tolerance to prevent overreaction of the host immune system. However, cancer cells can hijack these regulatory processes and cause immune escape of tumors. ⁴¹ Cytotoxic T‐lymphocyte antigen‐4 (CTLA‐4) is expressed on the surface of T cells upon activation and then binds to CD80 or CD86 on antigen‐presenting cells, a process that leads to the attenuation of T‐cell response. ⁴² CTLA‐4 plays vital roles in the early antigen recognition process and T‐cell priming in lymphoid organs. Ipilimumab, an anti‐CTLA‐4 antibody, is the first approved anticancer drug of this category. ⁴³ Another pivotal pathway is the axis of programmed cell death protein 1 (PD‐1) and programmed cell death ligand 1 (PD‐L1). PD‐1 is expressed on T cells, and PD‐L1 is expressed on tumor or immune cells. The PD‐1 × PD‐L1 interaction sends negative regulatory signals to cytotoxic T cells in the tumor microenvironment. ⁴⁴ Multiple anti‐PD‐1 and anti‐PD‐L1 monoclonal antibodies have been approved for various cancer types and are currently the most widely adopted immunotherapeutic agents (Figure 3).

FIGURE 3.

Mechanism of immune checkpoint inhibitors (ICIs). At the antigen presentation site, while a T‐cell receptor (TCR)–peptide–MHC complex forms between T cells and antigen‐presenting cells, the activation or suppression of T cells depends on the second signal. CD80/86‐CD28 interaction sends a stimulatory signal inside T cells, whereas cytotoxic T‐lymphocyte‐associated protein 4 (CTLA‐4) competes with CD28, binds to CD80/86, and sends an inhibitory signal. At the tumor site, programmed cell death protein 1 (PD‐1) on effector T cells and programmed cell death ligand 1 (PD‐L1) on tumor cells form an inhibitory synapse. Anti‐CTLA‐4, anti‐PD‐1, and anti‐PD‐L1 block immune checkpoints and reinvigorate antitumor immunity. CTLA‐4, cytotoxic T‐lymphocyte antigen‐4; MHC, major histocompatibility complex; PD‐1, programmed cell death protein 1; PD‐L1, programmed cell death ligand 1; TCR, T‐cell receptor. Created with BioRender.com

4.1. Anti‐CTLA‐4 antibody

The earliest attempt at using an ICI in patients with HCC was in a phase II trial of tremelimumab, an anti‐CTLA‐4 antibody. Tremelimumab resulted in an ORR of 17.6%, a disease control rate (DCR) of 76.4%, and time to progression of 6.48 months in 17 evaluable HCC patients. ⁴⁵ Subsequent development of CTLA‐4 blockade in HCC has led to focus on combination with PD‐1 and PD‐L1 inhibitors.

4.2. Anti‐PD‐1 antibody

Nivolumab, an anti‐PD‐1 antibody, is the first approved ICI for HCC. In the pivotal phase I/II CheckMate 040 trial, nivolumab showed an ORR of 14%, a median duration of response (DOR) of 19.4 months, and a median OS of 15.1 months in sorafenib‐treated patients ⁴⁶ , ⁴⁷ (Table 2). A subsequent phase III CheckMate 459 trial of nivolumab was conducted in sorafenib‐naïve patients. Although the ORR of nivolumab was higher than that of sorafenib (15% vs. 7%), nivolumab failed to show significant superiority to sorafenib in terms of OS (16.4 vs. 14.7 months). ⁴⁸

TABLE 2.

Selected trials of ICI monotherapy for advanced HCC

Trial phase	N	Enrollment time	Experimental arm/control arm	OS (months)	TTP/PFS (months)	DOR (months)	CR/RR % DCR % RECIST	RR % DCR % mRECIST
NCT01008358 Phase 2	20	2009–2010	Tremelimumab (single arm)	8.2	a 6.5		0/17.6 76.4
CheckMate 040 46 NCT01658878 Phase 1/2 (Sorafenib‐experienced)	182	2012–2016	Nivolumab (single arm)	15.1		19.4	3/14 55	18 54
KEYNOTE‐224 49 , 50 NCT02702414 Phase 2 Second line	104	2016–2017	Pembrolizumab (single arm)	13.2	b 4.9	21.0	3.8/18.3 61.5	15 51
SHR‐1210‐II/III‐HCC 59 NCT02989922 Phase 2 Second line	220	2016–2017	Camrelizumab 3 mg/kg Q2W/Q3W (reported together)	13.8	b 2.1	NR (3.7–14.0)	0/15 44
KEYNOTE‐240 51 NCT02702401 Phase 3 Second line	413	2016–2017	Pembrolizumab/placebo	13.9 vs. 10.2 HR: 0.781 p = 0.0238 (NS)	b 3.0 vs. 2.8 HR: 0.718 p = 0.0022 (NS)	13.8 vs. NR	2.2/18.3 vs. 0/4.4 62 vs. 53
CheckMate 459 48 NCT02576509 Phase 3 First line	743	2016–2017	Nivolumab/sorafenib	16.4 vs. 14.7 HR: 0.85 p = 0.0752 (NS)	b 3.7 vs. 3.8 HR:0.93	23.3 vs. 23.4	4/15 vs. 1/7 55 vs. 58

Abbreviations: CR, complete response; DCR, disease control rate; DOR, duration of response; HCC, hepatocellular carcinoma; HR, hazard ratio; ICI, immune checkpoint inhibitor; mRECIST, modified RECIST for hepatocellular carcinoma; N, patient number; NR, not reached; NS, not significant; OS, overall survival; PFS, progression‐free survival; RECIST, Response Evaluation Criteria in Solid Tumors; RR, response rate; TTP, time to progression.

^a The reported endpoint was time to progression.

^b The reported endpoint was progression‐free survival.

^* p value met statistical significance for the primary endpoint (for phase III studies).

Pembrolizumab, another anti‐PD‐1 antibody, showed efficacy comparable with that of nivolumab in a phase II KEYNOTE‐224 trial and was subsequently approved for second‐line treatment (Table 2). ⁴⁹ , ⁵⁰ However, in the confirmatory phase III KEYNOTE‐240 trial, which compared pembrolizumab with placebo in a second‐line setting, pembrolizumab failed to meet preset margins of positivity for the co‐primary endpoints OS and progression‐free survival (PFS). ⁵¹

In view of toxicity, as the action mechanism of ICIs is to unleash T‐cell brakes, the occurrence of a wide range of immune‐related adverse events (irAEs) is expected. IrAEs are often unpredictable and can affect any organ. Although most irAEs are Grade 1 or 2 in severity, fatal irAEs such as pneumonitis and myocarditis may occur. ⁵² Nevertheless, ICIs are generally more tolerable than multikinase inhibitors. As demonstrated in the phase III CheckMate 459 study, nivolumab, compared with sorafenib, was associated with a low rate of Grade 3/4 treatment‐related adverse events (TRAEs; 22% vs. 49%) and improved health‐related quality of life. ⁴⁸

5. ICI‐BASED COMBINATIONS

Combination therapy, which may enhance efficacy at the expense of safety, is generally not feasible in patients with HCC who are vulnerable. However, the favorable and nonoverlapping safety profile, as well as high tumor response rate, of ICIs makes them an ideal partner in combinations (Table 3).

TABLE 3.

Selected published or ongoing trials of ICI‐based combinations for advanced HCC

Trial phase	N	Enrollment time	Experimental arm/control arm	OS (months)	TTP/PFS (months)	DOR (months)	CR/RR % DCR % RECIST 1.1	RR % DCR % mRECIST
ICI‐antiangiogenic therapy combinations
GO30140 62 Group A NCT02715531 Phase 1b First line	104	2016–2018	Atezolizumab + bevacizumab	17.1	a 7.3	NR	12/36 71	15/39 71
GO30140 62 Group F NCT02715531 Phase 1b First line	119	2018–2019	Atezolizumab + bevacizumab/atezolizumab	NR vs. NR	a 5.6 vs. 3.4 HR: 0.55 p = 0.011	NR vs. NR	2/20 vs. 5/17 67 vs. 49	5/27 vs. 5/17 68 vs. 49
IMbrave150 53 , 54 NCT03434379 Phase 3 First line	336	2018–2019	Atezolizumab + bevacizumab/sorafenib	19.2 vs. 13.4 HR: 0.66 p = 0.0009*	a 6.9 vs. 4.3 HR: 0.65 p = 0.0001*	18.1 vs. 14.9	8/30 vs. <1/11 74 vs. 55	12/35 vs. 3/14 73 vs. 55
JVDJ 55 NCT02572687 Phase 1 HCC cohort	28	2015–2017	Durvalumab + ramucirumab	10.7	a 4.4	NR	0/11 61
Second‐line or later
KEYNOTE‐524 58 NCT03006926 Phase 1b First line	100	2017–2019	Pembrolizumab + lenvatinib	22.0	a 8.6	12.6	1/36 88	5/41 86
RESCUE 63 NCT03463876 Phase 2 First line Second line	First: 70 Second: 120	2018–2019	Camrelizumab + apatinib	First line: NR Second line: NR	First line: a 5.7 Second line: a 5.5	First line: 14.8 Second line: NR	First line: 1/34 77 Second line: 2/23 76	First line: 9/46 79 Second line: 2/25 76
CheckMate 040 64 NCT01658878 Phase 1/2 Cohort 6 first line or second line	71		Nivolumab 3 mg/kg Q2W + ipilimumab 1 mg/kg Q6W + cabozantinib/nivolumab 240 mg Q2W + cabozantinib	NR vs. 21.5	a 6.8 vs. 5.4		0/19 vs. 0/29 75 vs. 83
LEAP‐002 65 NCT03713593 Phase 3 First line	750	Active, recruitment completed	Lenvatinib + pembrolizumab/lenvatinib + placebo	Primary endpoint	Primary endpoint (BICR PFS, RECIST)
COSMIC‐312 66 NCT03755791 Phase 3 First line	740	Active, recruitment completed	Atezolizumab + cabozantinib/cabozantinib/sorafenib	Primary endpoint	Primary endpoint (BICR PFS, RECIST)
SHR‐1210‐III‐310 NCT03764293 Phase 3 First line	510	Recruiting	Camrelizumab + apatinib/sorafenib	Primary endpoint	Primary endpoint
ICI‐ICI combinations
CheckMate 040 59 , 67 NCT01658878 Phase 1/2 Cohort 4 3 arms Second line	50 49 49	2016–2016	Arm A N1I3 Arm B N3I1 Arm C N3I0.5 (refer to footnote for detailed regimens)	Arm A 22.2 Arm B 12.5 Arm C 12.7		Arm A 17.5 Arm B 22.2 Arm C 16.6	Arm A 8/32 54 Arm B 6/31 43 Arm C 2/31 49
Study 22 61 NCT02519348 Phase 2 Part 2 and 3 4 arms Second line	75 104 69 84	2017–2019	Arm 1 T 300 + D Arm 2 D Arm 3 T Arm 4 T75 + D (refer to footnote for detailed regimens)	Arm 1 18.7 Arm 2 13.6 Arm 3 15.1 Arm 4 11.3	Arm 1 a 2.2 Arm 2 2.1 Arm 3 2.7 Arm 4 1.9	Arm 1 NR Arm 2 11.2 Arm 3 24.0 Arm 4 13.2	Arm 1 1/24 45 Arm 2 0/11 38 Arm 3 0/7 49 Arm 4 2/10 37
HIMALAYA 68 NCT03298451 Phase 3 First line	1324	Active, recruitment completed	Durvalumab/(I)durvalumab + tremelimumab/(II)durvalumab + tremelimumab/sorafenib (refer to footnote for regimen (I)(II))	Primary endpoint
CheckMate 9DW NCT04039607 Phase 3 First line	650	Recruiting	Nivolumab 1 mg/kg + ipilimumab 3 mg/kg Q3W * 4 Followed by nivolumab 240 mg Q2W/sorafenib or lenvatinib	Primary endpoint

Note: Regimens of CheckMate 040 cohort 4: Arm A N1I3‐(nivolumab 1 mg/kg + ipilimumab 3 mg/kg) Q3W * 4, followed by nivolumab 240 mg Q2W. Arm B N3I1‐(nivolumab 3 mg/kg + ipilimumab 1 mg/kg) Q3W * 4, followed by nivolumab 240 mg Q2W. Arm C N3I0.5‐nivolumab 3 mg/kg Q2W + ipilimumab 1 mg/kg Q6W. Regimens of Study 22: Arm 1 T300 + d‐tremelimumab 300 mg * 1 + durvalumab 1500 mg Q4W. Arm 2 D‐durvalumab 1500 mg Q4W. Arm 3 T‐tremelimumab 750 mg Q4W * 7, then Q12W. Arm 4 T75 + D: tremelimumab 75 mg Q4W * 4 + durvalumab 1500 mg Q4W. Regimen (I) and (II) of HIMALAYA: Regimen (I)‐tremelimumab 75 mg Q4W * 4 + durvalumab 1500 mg Q4W. Regimen (II)‐tremelimumab 300 mg * 1+ durvalumab 1500 mg Q4W.

Abbreviations: BICR, blinded independent central review; CR, complete response; DCR, disease control rate; DOR, duration of response; mRECIST, modified RECIST for hepatocellular carcinoma; N, patient number; NR, not reached; NS, not significant; OS, overall survival; PFS, progression‐free survival; RECIST, Response Evaluation Criteria in Solid Tumors; RR, response rate; TTP, time to progression.

^a The reported endpoint was progression‐free survival.

^b The reported endpoint was time to progression.

^* p value met the statistical significance for the primary endpoint (for phase III studies).

5.1. ICI–antiangiogenic therapy combination

In the phase III randomized controlled IMbrave150 study, the combination of atezolizumab and bevacizumab was found to be superior to sorafenib in both PFS (6.9 vs. 4.3 months, HR: 0.65) and OS (19.2 vs. 13.4 months, HR: 0.66). ⁵³ , ⁵⁴ In addition, ORR (30% vs. 11%) and quality of life were better in the combination arm. The results led to universal approval and wide adoption in the major therapeutic guidelines (Figure 1).

Various ICI and antiangiogenic therapy combinations are under investigation (Table 3). However, despite the success of the ICI and anti‐VEGF combination, the combination of anti‐VEGFR antibody ramucirumab and anti‐PD‐L1 durvalumab did not obtain similar results. ⁵⁵ The possible reasons for the differential effects between the ICI–anti‐VEGF and ICI–anti‐VEGFR combinations are unclear.

Multikinase VEGFR inhibitors have targets beyond VEGFR. Consequently, they may modulate antitumor immunity through VEGF/VEGFR‐independent mechanisms. ⁵⁶ , ⁵⁷ KEYNOTE‐524, a phase Ib study, evaluated the pembrolizumab and lenvatinib combination in a first‐line setting, which resulted in an ORR of 36%, a median PFS of 8.6 months, and a median OS of 22.0 months. ⁵⁸ These promising results are comparable with those of the atezolizumab–bevacizumab combination. At least three phase III studies of ICI–multikinase inhibitor combinations in first‐line settings are ongoing (Table 3).

5.2. ICI–ICI combinations

Another strategy is to combine different ICIs, such as anti‐PD‐1 or anti‐PD‐L1 in combination with anti‐CTLA‐4, to potentiate T‐cell response. A cohort in the CheckMate 040 trial investigated three combinations of nivolumab (anti‐PD‐1) and ipilimumab (anti‐CTLA‐4) at various dosages in a second‐line setting. The most promising efficacy was shown in the arm with the highest dose intensity of ipilimumab (1 mg/kg nivolumab and 3 mg/kg ipilimumab every 3 weeks for four cycles, followed by 3 mg/kg nivolumab every 2 weeks). This combination yielded a complete response rate of 8%, an ORR of 32%, a median DOR of 17.5 months, and a median OS of 22.8 months and was granted accelerated approval by the US Food and Drug Administration. ⁶⁰ Similarly, the durvalumab–tremelimumab combination exhibited the best efficacy at the highest dosage of tremelimumab. ⁶¹ Two phase III studies of these regimens in the first‐line setting are ongoing (Table 3).

Safety is a concern in combination therapy. The most tolerable combination regimen is anti‐PD‐1/PD‐L1 with monoclonal VEGF antibody. Compared with sorafenib, the atezolizumab–bevacizumab combination resulted in numerically comparable Grade 3/4 adverse events (AEs), severe AEs, and dose interruption rates but with higher quality of life. ⁵³ The quality of life finding was partly attributed to the different natures of AEs between sorafenib and the atezolizumab–bevacizumab combination. The most common AEs of sorafenib include diarrhea and palmar–plantar erythrodysesthesia syndrome, which are symptomatic and significantly affect daily life. By contrast, the most common AEs of the atezolizumab–bevacizumab combination are infusion reaction, hypertension, and proteinuria, which are either transient or even asymptomatic. Conversely, an ICI in combination with a multikinase inhibitor is expected to increase the severity of AEs and decrease tolerability. As shown in the phase Ib KEYNOTE‐524 study, the pembrolizumab–lenvatinib combination caused Grade 3–5 TRAEs in as many as 67% of patients and led to significantly more frequent treatment interruption or discontinuation. ⁵⁸ Similarly, for anti‐PD‐1 or a combination of anti‐PD‐L1 and anti‐CTLA‐4, a high dose of anti‐CTLA‐4 was associated with improved efficacy but also caused increased toxicity. In CheckMate 040, arm A (1 mg/kg nivolumab and 3 mg/kg ipilimumab every 3 weeks for four cycles, followed by 3 mg/kg nivolumab every 2 weeks) was associated with 53% of grade 3 and 4 TRAEs, nearly double the number of grade 3 and 4 TRAEs as the other two arms with a low dosage of anti‐CTLA‐4 (29% and 31%, respectively). Therefore, while pursuing efficacious combination regimens, strategies for mitigating toxicities must be explored.

6. FUTURE PERSPECTIVES

6.1. Novel agents and combinations

A combination of nivolumab, ipilimumab, and cabozantinib resulted in an ORR of 26%, a DCR of 83%, and a median PFS of 6.8 months. High rates of Grade 3 and 4 TRAEs and treatment discontinuation (71% and 20%, respectively) were noted. ⁶⁴ Dual immune checkpoint blockade by using nivolumab and relatlimab (anti‐LAG‐3) is currently under investigation (NCT04567615). Using the atezolizumab–bevacizumab combination as the backbone, trials tested the add‐on immunomodulation in the tumor microenvironment, such as for tiragolumab (anti‐TIGIT) and tocilizumab (anti‐IL6R; Morpheus Liver, NCT04524871). ⁶⁹ FGF19‐FGFR4 dependence and FGFR4 kinase domain mutation have been demonstrated to be potential mechanisms of resistance to sorafenib, ⁷⁰ , ⁷¹ and early‐phase clinical trials have been conducted on selective FGFR4 inhibitors or combinations with other agents such as ICIs. ⁷² , ⁷³

6.2. Integration of locoregional and systemic treatments

Traditionally, the treatment of HCC was stage‐guided, with locoregional treatment for early or intermediate stages and systemic treatment for the advanced stage. However, because of the increase in potent systemic treatments, locoregional and systemic therapies are no longer mutually exclusive at any disease stage. For patients receiving curative or locoregional therapy, systemic treatment may serve as adjuvant therapy to reduce the likelihood of relapse and prolong survival. This strategy is being tested in several phase III studies. ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ In patients with intermediate HCC who are expected to obtain borderline benefits from locoregional therapy, such as patients with a single large tumor or multiple diffuse tumors, systemic treatment may have greater efficacy than locoregional treatment. ⁸⁰ For patients with initial advanced disease that is later down‐staged through immunotherapy, the integration of locoregional therapy may lead to long‐term disease control.

7. CONCLUSIONS

After decades of intensive research on systemic therapy for HCC, the journey has culminated at the combination regimen of atezolizumab and bevacizumab, which has pushed the efficacy of systemic therapy for HCC to an unprecedented high. The success of this combination paves the road for treatments with even higher efficacy. HCC is thus becoming a highly treatable disease, and systemic therapy will certainly play a major role in all stages of HCC.

CONFLICT OF INTEREST

All authors declare no conflict of interest.

Wu T‐C, Shen Y‐C, Cheng A‐L. Evolution of systemic treatment for advanced hepatocellular carcinoma. Kaohsiung J Med Sci. 2021;37:643–653. 10.1002/kjm2.12401