Recent advances in systemic therapy for hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is one of the most common and lethal malignant tumors in the world. Therapeutic options for advanced HCC are limited. Systemic treatment, especially with conventional cytotoxic drugs, is usually ineffective. For more than a decade, sorafenib has been the only systemic drug that has been proven to be clinically effective for treating advanced HCC. However, over the past three years, the rapid progress of molecular targeted therapies has dramatically changed the treatment landscape for advanced HCC. Immune checkpoint therapies are now being incorporated into HCC therapies, and their combination with molecular targeted therapy is emerging as a tool to enhance the immune response. In this review, we summarize the development and progress of molecular targeted agents and immunotherapies in HCC.


Background
Worldwide, liver cancers are predicted to rank sixth for incident cases and be the third leading cause of cancerrelated death in 2020, with approximately 905,677 new cases and 830,180 deaths each year ( Fig. 1A-C) [1]. Based on annual estimates, the World Health Organization predicts that more than 1.3 million people will die from liver cancer by 2040 (Fig. 1D) [2]. HCC is the most frequently occurring tumor among all primary liver cancers, accounting for 75%-85% of cases [3]. HCC is usually diagnosed at an advanced stage, for which there remain limited effective treatment options [4]. When applying palliative treatment at an advanced stage, the median post-diagnostic survival ranged from 6 to 12 months [5]. Systemic treatments may benefit patients with advanced-stage HCC [6].
Until 2007, there were no effective treatment options for patients diagnosed with advanced-stage disease or patients who transitioned into advanced-stage disease after other treatments failed [7]. Sorafenib is the first and only systemic drug approved by the U.S. Food and Drug Administration (FDA) as standard treatment for advanced HCC between 2007 and 2016. However, rapid advances during the last 3 years have led to the approval of other molecular targeted drugs and several immune checkpoint inhibitors (ICIs), which have been added as therapeutic tools for advanced HCC (Tables 1, 2 and 3). In addition, the recent successful combination of atezolizumab and bevacizumab marks an important change in the first-line treatment of HCC (GO30140, NCT NCT02715531) [8]. To date, systemic therapy for advanced HCC includes molecular targeted therapy, immune checkpoint inhibitors, or a combination of both (Fig. 2).
In this review, we present a summary of systemic drugs already approved for use in advanced HCC and those that have shown positive results in preclinical or clinical trials, providing a perspective on the future of advanced HCC therapies.

Molecular targeted therapies
Over the last few decades, advances in molecular cell biology have greatly contributed to our knowledge of the molecular mechanisms of tumorigenesis and its progression, which in turn provides opportunities for the development of novel molecular targeted agents that inhibit molecular abnormalities as promising cancer treatments [9]. At present, molecular targeted therapy mainly includes tyrosine kinase inhibitors (TKIs) and/or monoclonal antibodies.

Sorafenib
Sorafenib is an oral TKI that targets the angiogenesis and proliferation pathways of tumors by blocking vascular endothelial growth factor receptor (VEGFR) 1-3, platelet-derived growth factor receptor (PDGFR) βpathways, and Raf-MEK-ERK signaling, all of which participate in the pathogenesis of HCC (Fig. 4) [11,12].
At ASCO 2007, the results of the SHARP (Sorafenib Hepatocellular Carcinoma Assessment Randomized Protocol; NCT00105443) were reported, which demonstrated the effectiveness of sorafenib against HCC. The primary endpoints were overall survival (OS) and time to symptomatic progression. Compared to the placebo group, the sorafenib treatment group had significantly prolonged OS (median OS 10.7 vs. 7.9 months, hazard ratio [HR]=0.69; 95% confidence interval [CI], 0.55-0.87; p<0.0001), while the time to symptomatic progression was not significantly different between the two groups (median time 4.1 vs. 4.9 months, HR=1.08, p=0.77). Statistically significant differences were found for secondary endpoints: the median time to radiologic progression improved by 3 months (5.5 vs. 2.8 months, P<0.001), and the sorafenib group had a higher disease-control rate than the placebo group (43% vs. 32%, P = 0.002). Common adverse effects observed in the sorafenib group included diarrhea, weight loss, hand-foot skin reactions, alopecia, anorexia, and voice changes, which were more frequent than those in the placebo group (P<0.001). Notably, 97% of the patients included in this trial were assessed as Child-Pugh liver function class A [13]. Based on these data, sorafenib was approved by the U.S. FDA in November 2007 for advanced HCC as first-line standard treatment.
Although sorafenib has a notable survival benefit in patients with advanced HCC, it has yet to be maintained. Owing to adverse events (AEs), many patients receiving sorafenib suffer from disease progression after dosage reductions or discontinuing treatment [14,15]. In addition, 27% of patients in the SHARP trial had no initial response to sorafenib [13]. This reveals that resistance to sorafenib (both primary and acquired) greatly limits its survival benefit. In a study predicting survival in advanced HCC patients who permanently discontinued sorafenib, the median post-sorafenib survival was only 4.1 (95% CI 3.3-4.9) months. Liver decompensation, performance status (PS), tumor progression, and/or extrahepatic tumor spread were significantly associated with post-sorafenib survival [16].

Lenvatinib
As first-line treatment, most drugs subsequently tested in phase III trials failed to improve the prognosis compared to sorafenib; they also did not improve survival compared with placebo as second-line therapy. These drugs include erlotinib [17], brivanib [18], sunitinib [19], linifanib [20], everolimus [21], and tivantinib [22] (Fig. 4, Table 1). Insufficient antitumor activity, toxicity in a cirrhosis state, and inadequate patient selection were cited as reasons for these failures. Sorafenib has remained the sole effective first-line treatment option for over a decade until lenvatinib was finally proven to be non-inferior to sorafenib in terms of OS [23].

Regorafenib
Regorafenib is an oral inhibitor that inhibits many angiogenic and tumorigenic kinases, such as VEGFR1-3, tyrosine kinase with immunoglobulin and epidermal growth factor homology domain 2 (TIE2), PDGFRβ, FGFR1, B-RAF, RET, and KIT ( Fig. 4) [27]. Regorafenib may exert a more far-reaching antiangiogenic activity owing to a combined blockade of the VEGFR2 and TIE2 pathways, which is supported by the results from a preclinical study by Tsai and Lee [28] showing a cooperative effect when combining anti-TIE2 and sorafenib, leading to an observed increase in overall survival in a melanoma model. On April 27, 2017, the U.S. FDA expanded the indication for regorafenib as a second-line treatment for advanced HCC patients previously treated with sorafenib [29]. Approval was based on a randomized, placebocontrolled international phase III trial designed to assess the safety and efficacy of regorafenib in patients with HCC progressing during sorafenib treatment (RESORCE; NCT01774344) [30]. Regorafenib had a longer OS than placebo (mOS, 10.6 vs. 7.8 months, HR= 0.63, p<0.0001) and prolonged progression-free survival and time to progression by mRECIST (mPFS 3.1 vs. 1.5 months, HR=0.46, p<0.0001; mTTP was 3.2 vs. 1.5 months, HR=0.44, p<0.0001)). The regorafenib group had an ORR of 11% versus 4% compared to the placebo group (p=0.0047), and the disease control rate (DCR) was 65% versus 35% (p<0.0001). Of note, the RESORCE trial was performed in patients who had progressed on prior sorafenib treatment; thus, the efficacy of regorafenib among sorafenib-intolerant patients has not yet been determined.
Additional analyses from the RESORCE trial suggested that regorafenib might offer clinical benefits independent of the latest sorafenib dose or the time to progression after previous sorafenib treatment and that the occurrence of AEs was independent of the last sorafenib dose [31]. Multiple clinical trials are currently evaluating the efficacy of regorafenib together with ICIs (Table 3).

Cabozantinib
Cabozantinib is a small-molecule tyrosine kinase inhibitor of MET and VEGFR 1-3, RET, KIT, AXL, and FLT3, all of which are associated with tumor pathogenesis (Fig.  4) [32,33]. Cabozantinib has remarkable antitumor activity in HCC through dual inhibition of MET and VEGFR2 [33]. In an international, phase III double-blinded trial, 773 patients who progressed on no more than two prior systemic treatments including sorafenib were randomized to cabozantinib or placebo (CELESTIAL; NCT01908426).
The primary endpoint of the CELESTIAL trial was that OS was significantly improved (mOS 10.2 vs. 8.0 months, HR=0.76, p=0.005). The secondary efficacy endpoints of the trial showed that cabozantinib significantly improved  [35]. Several trials evaluating the combinations of cabozantinib and ICIs in patients with advanced HCC are ongoing, including in combination with nivolumab and atezolizumab (Table 3). Of note, in the CELESTIAL trial, 192 (27%) patients had received two previous systemic anticancer drugs, meaning that cabozantinib was used as third-line therapy. Within this subgroup, the OS was not significantly different (mOS 8.6 months vs. 8.6 months). However, cabozantinib improved the PFS (mPFS 3.7 vs. 1.9 months, HR=0.58, 95% CI, 0.41-0.83), suggesting that advanced HCC patients could use cabozantinib as a possible third-line option [34].

Ramucirumab
Ramucirumab, a humanized recombinant IgG1 monoclonal antibody, selectively binds to VEGFR-2, preventing activation of the VEGF pathway ( Fig. 4) [36], and it showed anti-tumor activity in advanced HCC in a phase II trial [37]. A prospective study focusing on ramucirumab applied to treat the progression of advanced HCC while on sorafenib (REACH; NCT01140347) showed that ramucirumab did not lead to a significant improvement in OS compared to placebo (mOS 9.2 vs. 7.6 months, HR =0.87, p=0.14). However, in a subgroup with high baseline α-fetoprotein (AFP) concentrations (≥ 400 ng/mL), ramucirumab significantly prolonged the OS with a median OS of 7.8 months vs. 4.2 months (HR=0.674, p=0.006) [38].
To further investigate this result, the biomarker-driven REACH-2 trial (NCT02435433) was designed, evaluating ramucirumab versus placebo in advanced HCC patients with elevated AFP levels (≥400 ng/mL) who received prior sorafenib treatment. The primary endpoint of the REACH-2 trial showed that compared to the placebo group, OS was longer in the ramucirumab group (mOS 8.5 vs. 7.3 months, HR=0.71, p=0.0199). Ramucirumab also increased the PFS (mPFS 2.8 vs. 1.6 months, HR= 0.452, p<0.0001), time to radiographic progression (mTTP 3.0 vs. 1.6 months, HR=0.427, p<0.0001), and disease control rate (59.9% vs. 38.9%, p=0.0006). REACH-2 was the first phase III trial to successfully screen for therapeutic candidates with biomarkers among patients with advanced HCC [39].
Based on these results, on May 10, 2019, the U.S. FDA authorized ramucirumab as a second-line treatment for HCC patients whose AFP was no less than 400 ng/mL [40].

Monotherapy with immune checkpoint inhibitors
The success of anti-cytotoxic T lymphocyte associated antigen 4 (CTLA-4) antibodies in blocking immune checkpoints in advanced melanoma patients offers hope for immunotherapy of tumors. HCC carcinogenesis is typically accompanied by chronic liver inflammation, and HCC could be immunogenic [41]. In general, immune tolerance occurs in HCC mainly due to myeloidderived suppressor cells (MDSCs), alterations of immune checkpoint molecules (such as CTLA-4 and programmed cell death protein-1 [PD-1]), and enrichment of T-regulatory cells (Tregs) [42]. The first evidence showing that ICIs could make a significant difference in HCC treatment came from a phase II trial in which the safety and antitumor activity of tremelimumab (CTLA-4 blockade) supported further studies in patients with advanced HCC who developed HCV-induced cirrhosis [43] ( Table 2). In addition, other clinical trials of ICI monotherapy for HCC have been performed ( Table 2).

Nivolumab
Nivolumab, a monoclonal antibody, blocks the PD-1 signaling pathway (Figs. 3 and 4) and restores the anti-tumor immune activity. It has been authorized by the U.S. FDA for application in several tumors (e.g., melanoma, nonsmall cell lung cancer [NSCLC], and kidney cancer) [44].
CheckMate 040 (NCT01658878), a noncomparative phase I/II trial, was designed to investigate the safety and efficacy of nivolumab for advanced HCC patients by escalating and expanding the dose of the medication. Prior sorafenib was allowed. Safety and tolerability represented the primary endpoints for the dose-escalation phase (n=48), while ORR was the primary endpoint for the doseexpansion phase (n=214). During the dose-escalation phase, the safety profile of nivolumab was evaluated, and the patients' tolerability of the medication was acceptable: 25% of the patients (12 of 48) had grade 3 or 4 treatmentrelated adverse events (TRAEs), and the incidence of TRAEs appeared to be dose-independent. In the doseexpansion phase, the ORR was 20% (42 of 214 patients, 95% CI, [15][16][17][18][19][20][21][22][23][24][25][26], with thirty-nine partial responses and three complete responses. The median duration of response was 9.9 months, and the mPFS was 4.0 months (95% CI, 2.9-5.4). Seventy-nine of 138 patients (57%) had their disease under control, with most disease stabilizations lasting for at least 6 months. Among uninfected untreated patients, the ORR was 23%, and the nine-month overall survival rate was 82%, which supported that Presentation of tumor-associated antigen by the major histocompatibility complex (MHC) expressed by APCs results in the release of an activation signal in combination with a costimulatory signal via the B7-CD28 pathway, leading to activation of T cells in the lymph node; B7 also binds to CTLA-4 with a higher affinity than that of CD28, in which case T cells cannot be activated. PD-1 on T cells inhibits antigen-specific T cell activation by interacting with its ligands PD-L1 and PD-L2. Immune escape is induced through the PD-1/PD-L1 axis, as well as the B7/CTLA-4 axis. This figure was adapted from Kudo, et al. [44]. (B) VEGF modulates the immunosuppressive TME, and TKIs restore this suppressive effect. Red arrows represent promotion effects. APCs, antigen presenting cells; CTL, cytotoxic T lymphocyte; CTLA-4, cytotoxic T-lymphocyte antigen 4; iDC, immature dendritic cell; matDC, mature dendritic cell; MDSCs, myeloid-derived stem cells; PD-1, programmed cell death protein 1; PD-L1, programmed cell death-ligand 1; TAMs, tumor-associated macrophages; TME, tumor microenvironment; Tregs, regulatory T cells. Note: This is an open access article distributed under the Creative Commons Attribution License that permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0) nivolumab deserved further investigation as first-line therapy for advanced HCC patients [45]. Based on the ORR and the durable objective responses, on September 22, 2017, the U.S. FDA accelerated the approval of nivolumab to treat HCC patients who were first treated with sorafenib [46].
Another research team studied first-line treatment (nivolumab vs. sorafenib) for advanced HCC patients through a randomized and multicenter phase III study (CheckMate 459; NCT02576509) and reported its data at the 2019 Congress of European Society for Medical Oncology (ESMO). The primary endpoint, OS, was not statistically significant, as the mOS was 16.4 and 14.7 months in the nivolumab group and sorafenib group, respectively (HR=0.85, p=0.0752). Interestingly, patients receiving nivolumab had a higher complete response rate (ORR, 15% vs. 7%, respectively). Eighty-one patients (22%) taking nivolumab and 179 patients (46%) taking sorafenib had grade 3/4 TRAEs [47]. Although the above findings may not influence the current standard treatment, they give us insight that immunotherapy may also have a position in the first-line treatment for advanced HCC.
Following KEYNOTE-224, a double-blind, randomized, phase III study (KEYNOTE-240; NCT02702401) was designed to further assess the safety and efficacy of   Table 2). Pembrolizumab combined with molecular targeted drugs are now being evaluated through several phase I/II trials (Table 3).

Combination therapy
To achieve a more satisfactory therapeutic response rate than ICI monotherapy, combination therapies based on ICIs are being developed to treat cancer. In 2018, Larkin et al. discovered through the CheckMate067 trial that the use of combined nivolumab and ipilimumab (CTLA-4 inhibitor) was more effective than monotherapy in patients with advanced melanoma [53]. Possible mechanisms for this combination therapy are as follows: 1) blocking the CTLA-4 pathway in lymph nodes increases the activation and proliferation of T cells, leading to CD8+ T cell infiltration; 2) blocking the PD-1/PD-L1 pathway not only maintains the killing capacity of cytotoxic T lymphocytes (CTLs) [44] but also improves the ability of antigen-presenting cells to present tumorassociated antigens; and 3) given the high expression of CTLA-4 and PD-1 on Tregs, the combination of these two inhibitors indirectly reduces the immunosuppressive tumor microenvironment (TME) (Fig. 3A) [54].
Another area of interest is the combination of TKIs with ICIs. It is worth noting that molecular targeted agents (sorafenib, lenvatinib, regorafenib, cabozantinib, and ramucirumab) with beneficial effects on survival in patients with HCC all share the same characteristics, that is, anti-angiogenesis, highlighting the importance of this hallmark in cancer treatment [55]. The restoration of tumor vascular function may help to enhance the ability of other drugs (e.g., ICIs) that can be used in combination with anti-angiogenic molecules to kill tumor cells [56,57]. TKIs with antiangiogenic activity exert immunomodulatory effects on the TME [58], which include promoting dendritic cell (DC) maturation [59], enhancing T cell trafficking and function [10], and reversing the hypoxia-induced immunosuppressive effect in tissue [60] (Fig. 3B).

Nivolumab + Ipilimumab
As discussed above, PD-1 plus CTLA-4 inhibitors promote the anti-tumor immune response through different complementary mechanisms that affect different signaling pathways. Nivolumab in combination with ipilimumab has been shown to be effective in many tumors (e.g., melanoma, NSCLC, renal cell carcinoma, mismatch repair-deficient/microsatellite instability-high metastatic colorectal cancer) [61][62][63][64]. In cohort 4 of CheckMate 040 (NCT01658878), 50 patients received 4 doses of nivolumab (1 mg/kg) with ipilimumab (3 mg/kg) every 3 weeks and then 240 mg of nivolumab every 2 weeks (arm A). In this randomized clinical trial, the dosage regimen of arm A showed a good safety profile, a promising objective response rate, and a durable response: the ORR was 32% per RECIST v1.1, with 4 complete responses and 12 partial responses; the median response duration was 17.5 months (4.6-30.5+ months), and 31% of the responses lasted over 24 months [45]. Based on these data, on March 10th, 2020, the dosage regimen of arm A received accelerated approval for advanced HCC patients who progressed on sorafenib treatment [65] ( Table 3).
Recently, the 44-month follow-up results from Check-Mate 040 were presented at the 2021 ASCO GI Conference. The mOS was 22.2 months in arm A. The 24month OS rate improved to 46% (95% CI, 32-59), and the 36-month OS rate was 42% (95% CI, 28-55). The ORR remained at 32%, and the median response duration remained at 17.5 months (5-47+ months) in arm A. The DCR was higher in arm A than in arms B and C (54%, 43%, 49%, respectively). Nivolumab + ipilimumab as a second-line treatment continued to show clinical responses and long-term survival benefits [66].
This combination became a promising treatment and is being compared with sorafenib via a phase III trial. In a global, open-label, phase III trial, 501 patients randomly received either atezolizumab + bevacizumab or sorafenib (IMbrave 150; NCT03434379). The efficacy phase showed that the atezolizumab + bevacizumab group had a significantly longer OS and PFS: the estimated 6-month and 12-month survival rates were 84 No serious adverse events with significant differences were observed between the two groups. The most common AEs, affecting more than 20% of patients in the atezolizumab + bevacizumab group, were hypertension (29.8%), fatigue (20.4%), and proteinuria (20.1%) [71]. On May 29, 2020, the U.S. FDA approved atezolizumab in combination with bevacizumab as a first-line treatment for advanced HCC patients based on the safety and efficacy revealed in IMbrave 150 [72] (Table 3).
At the 2021 ASCO Gastrointestinal Cancer Symposium, Finn et al. [73] updated the OS analysis for IMbrave 150. After an additional 12 months of follow-up, a sustained clinical efficacy benefit was achieved with atezolizumab + bevacizumab versus sorafenib: the mOS was 19.2 months in the combination group versus 13.4 months with sorafenib (HR= 0.66 [95% CI, 0.52-0.85], P=0.0009). The updated data showed an ORR of 29.8% (95% CI, 24.8-35) by RECIST v1.1 in the combination group compared to 11.3% (95% CI, 6.9-17.3) in the sorafenib group, and more patients in the combination group achieved a complete response than previously reported (CR= 7.7%). This combination regimen has the longest OS ever seen in first-line phase III studies, further confirming that this regimen can serve as a standard of care for patients with advanced HCC who have not received any systemic therapy before.

Targeted therapies currently in clinical trials
Currently, five targeted therapies have been approved for the treatment of advanced HCC. Of these five therapies, four are small-molecule kinase inhibitors, and one is a monoclonal antibody against VEGFR2 (Fig. 2). In addition to these approved targeted therapies, a variety of targeted therapies are in clinical development.

VEGFR
There are many growth factors involved in tumor angiogenesis, the most important of which is the vascular endothelial growth factor (VEGF) family [74]. In addition to the mentioned inhibitors targeting VEGFR that have been approved by the FDA (sorafenib, lenvatinib, regorafenib, cabozantinib, and ramucirumab), there are several anti-VEGFR agents currently in clinical trials. Tivozanib is a potent VEGFR 1-3 TKI with relatively lesser effects on KIT and PDGFRβ and it has dose-dependent activity against HCC in vivo [75]. According to the existing NCT01835223 data (Table 1) [76], tivozanib showed an ORR (per RECIST v1.1) of 21%, mOS of 9 months (90% CI, 5.4-27.8), mPFS of 24 weeks, and a 1-year OS rate of 40%. Unfortunately, the study did not proceed to stage 2 due to not meeting the prespecified statistical plan. However, tivozanib had better anti-tumor efficacy than sorafenib and lenvatinib as first-line therapy, with ORRs of 21% vs. 6.5% (sorafenib) and 18.8% (lenvatinib), respectively. Since small molecular inhibitors can at least partially reverse immune dysfunction, the combination of tivozanib and ICIs is a promising option. As previously mentioned, a trial of tivozanib in combination with durvalumab is currently ongoing (Table 3).
In addition, at the 2020 ASCO Virtual Scientific Program, two other TKIs designed by Chinese companies reported exciting phase III results. In a first-line treatment setting, donafenib (ZGDH3, NCT02645981, Table 1) had a significantly longer overall survival than sorafenib: the mOS was 12.1 vs. 10.3 months (HR=0.831, p=0.0363) ( Table 1). Additionally, donafenib exhibited favorable safety and tolerability compared to sorafenib: drug-related AEs of grade ≥ 3 occurred in 37.5% of patients in the donafenib group versus 49.7% in the sorafenib group (p=0.0018), and the incidence of treatment interruptions due to drug-related AEs was 30.3% in the donafenib group compared to 42.5% in the sorafenib group (P=0.0013) [77].
Apatinib has been investigated as second-line therapy in Chinese advanced HCC patients through a phase III study AHELP (NCT02329860, Table 1), which showed significantly longer OS and PFS compared to placebo and exhibited a tolerable safety profile. The mOS was 8.7 vs. 6.8 months (HR=0.785, p=0.0476), the mPFS was 4.5 vs. 1.9 months (HR=0.471, p<0.0001), the ORR was 10.7% vs. 1.5% per RECIST v1.1, and no new safety signals were noted [78].

c-MET
c-Met is a receptor tyrosine kinase for hepatocyte growth factor (HGF) [79], and abnormalities in c-Met are found in approximately 50% of HCC patients [80]. Aberrant c-MET activity is involved in HCC progression [81] and may lead to treatment resistance toward sorafenib [82]. c-MET inhibitors can be categorized into selective and nonselective. Nonselective c-MET inhibitors (e.g., tivantinib and cabozantinib) may exert anti-tumor effects, mainly owing to the inhibition of non-c-MET targets, and an "off-target" effect is associated with increased toxicity [83]. Conversely, selective c-MET inhibitors (e.g., tepotinib and capmatinib) are being tested in HCC and are thought to have reduced toxicity. Both selective and nonselective agents are being investigated in different trials for potential use in HCC patients as first-or second-line therapy (NCT01988493, NCT02115373; NCT01737827, NCT01964235, Table 1). Decaens et al. reported data of NCT02115373, showing that tepotinib was well tolerated and that the recommended phase 2 dose (500 mg) reached its primary endpoint in the phase 2 study: the 12-week PFS was 63.3% (one-sided binomial exact test: P < 0.0001) [84]. Selective c-MET inhibitors are more promising drug candidates. There are also ongoing studies evaluating the safety and tolerability of c-MET inhibitors combined with ICIs (NCT02795429, NCT02082210, Table 3).

TGFβ
The TGFβ pathway has dual anti-and pro-tumoral activities in cancer cells: in the early stage, it is beneficial for promoting cell cycle arrest and apoptosis [85], while in the advanced stage, it promotes tumor progression and metastasis by enhancing cell motility, epithelial-tomesenchymal transition (EMT), invasiveness and stemness [86]. In addition, TGFβ signaling is a major modulator of immune cell proliferation, differentiation, development, and survival [87], which suppresses CD8+ T cells, NK cells, and DCs and increases CD4+CD25+ Tregs by promoting the differentiation of M2-type macrophages [88], leading to immunosuppression in HCC. Microenvironmental remodeling by TGFβ creates a beneficial microenvironment for tumor development and metastasis [86]. These findings provide a rationale for blocking TGFβ signaling in HCC treatment, especially in immunotherapy. Galunisertib (LY2157299) is a TGFβ receptor 1 inhibitor [89] that has been studied as monotherapy and combined with sorafenib in extensive phase I/II trials for HCC [90].

Endoglin
Endoglin (also known as CD105), a type 1 intact transmembrane glycoprotein, is a co-receptor for TGF-β ligands and plays an important role in fibrogenesis and angiogenesis [91]. Endoglin is mainly found in immature tumor vascular endothelial cells and contributes to tumor angiogenesis [92,93]. Endoglin microvascular density has an essential and independent prognostic role in recurrent or metastatic HCC patients [94], and circulating serum endoglin could be detected in HCC patients [93,95,96], suggesting that endoglin may be a prognostic marker for HCC patients; however, it still needs further study. TRC105 (carotuximab) is an anti-endoglin antibody, yet it did not show adequate monotherapy efficacy in the phase II trial (NCT01375569) of TRC105 for post-sorafenib HCC patients [97]. The combination of TRC105 and sorafenib (NCT01306058) showed an ORR of 21% per RECIST at all four dose levels, with a mOS of 15.5 months and a mTTP of 3.8 months [98]. At the 2019 ASCO GI, Raghav et al. reported that TRC105 + sorafenib had encouraging signs of activity, with durable partial responses in 1 of 5 patients thus far in phase II (NCT01806064) [99].

FGF19/FGFR4
Among the family of tyrosine kinase receptors, fibroblast growth factor receptor 4 (FGFR4) is mostly expressed in the liver [100]. FGF19, one of the three endogenous FGFs, binds with the highest affinity to FGFR4 [101], and the FGF19 expression level is frequently reported to be associated with HCC prognosis [102]. After activation of the FGFR4 signaling pathway, HCC cells undergo proliferation, EMT, anti-apoptosis, angiogenesis, invasion, and drug resistance through activating four downstream pathways: PI3K/AKT/mTOR, RAS/RAF/MAPK, PLCγ/DAG/PKC, and GSK3β/β-catenin [103].. Abnormal expression of FGF19/FGFR4 promotes the development of HCC [102], making it a promising therapeutic target. Despite the relatively low selectivity of pan-FGFR inhibitors against FGFR4 [104], some TKIs specifically targeting FGF19/FGFR4 are being tested in different phases of clinical trials (Table 1) [105], among which the most promising candidate is BLU-554.
In a phase I trial (NCT02508467) to escalate and expand doses, fisogatinib (BLU-554) produced a clinical response in FGF19-positive patients with advanced HCC: the ORR per RECIST v1.1 was 17% (11 of 66 patients) in FGF19-positive patients and 0% in the FGF19-negative group; 53% of TRAEs in this study were grade 1/2, which were tolerable. This trial validates the effectiveness of blocking the FGF19/FGFR4 axis and the biomarker potential of FGF19 to screen HCC patients [106]. Another FGFR4 inhibitor, FGF401, has been investigated as monotherapy or in combination with PDR001 (NCT02325739).

Immune checkpoint therapies in clinical trials
Currently, two immune checkpoint inhibitors have been approved by FDA for the HCC treatment, both of which are PD-1 monoclonal antibodies. In addition to these approved therapies, there are a variety of promising immune checkpoints therapies in clinical trials.

Other immune checkpoint inhibitors
In addition to the well-known PD-1/PD-L1 and CTLA-4, a series of inhibitory immune checkpoint molecules involved in the immune tolerance of HCC have been reported, including lymphocyte activation gene 3 (LAG-3) [114], T cell immunoglobulin mucin-3 (TIM-3)/galectin-9 (GLA-9) [115], T cell immunoglobulin and ITIM domain (TIGHT) [116], and adenosine A2a receptor. Zhou et al. [114] found that a PD-L1 inhibitor in combination with a TIM-3, LAG-3, or CTLA-4 inhibitor results in a revitalization of in vitro tumor-infiltrating lymphocyte (TIL) responses in most patients and further enhances its effect compared to PD-L1 monotherapy. The use of drugs targeting TIM-3/GLA-9 is well summarized in Ref [117]. Next-generation ICIs targeting LAG-3, TIM-3, and TIGHT are expected to provide survival benefits for HCC patients.

Agonists of co-stimulatory checkpoint pathways
Immune checkpoints include inhibitory pathways and costimulatory pathways. Agonists of co-stimulatory checkpoint pathways such as OX40, GITR, ICOS, CD27, and CD28 are under investigation for treating solid tumors, which have also been summarized in Ref [118]. PF-04518600 is an OX4 agonist antibody. In 2020, the results of a phase I trial (NCT02315066) investigating its safety and tolerability in patients with advanced HCC were published, showing that PF-04518600 was welltolerated overall and provided meaningful disease control [119]. Glucocorticoid-induced tumor necrosis factor receptor family-related protein (GITR) is believed to promote the function of effector T cells and inhibit the function of Tregs [120].
Beek et al. found that agonistic targeting GITR enhances the function of tumor-infiltrating T cells in HCC, suggesting that it could be a promising target for immunotherapy [121]. Several antibodies targeting GITR including INCAGN01876 (NCT03126110), BMS-986156 (NCT04021043), and TRX518 (NCT02628574, NCT03861403) are being studied for solid tumors involving HCC.

Combination therapies in clinical trials
Regorafenib enhances the killing effect of T cells by modulating various immunoreactive molecules, helping to present antigens, and triggering T cell differentiation and accumulation [122]. Tumor-associated macrophages associated with angiogenesis in the TME are characterized by TIE2 expression, which is strongly associated with cancer cell intravasation into the circulatory system [123]. This evidence provided a rationale to combine regorafenib and pembrolizumab as first-line treatment for advanced HCC, which is being studied in a multicenter dose-escalation phase IIb trial (NCT03347292 , Table 3).
Meanwhile, lenvatinib combined with pembrolizumab has shown promising antitumor activity and manageable toxicity in unresectable HCC through a phase Ib clinical trial (KEYNOTE-524/NCT03006926, Table 3), which supported a Breakthrough Therapy designation granted by the FDA in July 2019 [124]. Efficacy data from KEYNOTE-524 showed an ORR of 46% (95% CI, 36-56.3%) per mRECIST, the mPFS was 9.3 months (95% CI, 5.6-9.7) and mOS was 22 months (95% CI, 20.4-NE) [125]. As a single-arm study, lenvatinib + pembrolizumab from KEYNOTE-524 did not show evidence of meaningful improvement over approved therapies; therefore, a phase III trial is currently assessing the efficacy and safety of this combination compared to lenvatinib + placebo as a first-line option (LEAP-002/NCT03713593, Table 3).
Cabozantinib not only modulates peripheral and intratumor immune landscapes but also induces changes in the phenotype of MC38-CEA tumor cells, thereby increasing their vulnerability to cytotoxic effects mediated by T cells [126]. Immunogenic and immune subset modulations of cabozantinib support its clinical investigation in combination with immunotherapy. Multiple active trials are evaluating the efficacy of cabozantinib in combination with nivolumab/atezolizumab, including the COSMIC-312 phase III trial (NCT03755791, Table 3).
Camrelizumab was tested in combination with apatinib (VEGFR-2, RET, c-kit inhibitor) for treating advanced HCC patients in NCT02942329 [128]. At the data cutoff (June 15, 2018), 8 of 16 evaluable HCC patients achieved a partial response, with an ORR of 50% (95% CI, 24.7-75.4%). This combination has been evaluated in a phase III trial as a first-line treatment (NCT03764293) versus sorafenib and a phase II trial as a second-line treatment (NCT03463876) ( Table 3).
A study investigating tivozanib (selective VEGFR 1-3 TKI) in combination with durvalumab for untreated advanced HCC updated its phase Ib data at the 2021 ASCO GI Conference. NCT03970616 (Table 3) enrolled 7 patients and found that untreated advanced HCC patients tolerated this combination well, with two of them achieving a partial response [129]. The phase II portion of the study is ongoing.

Antibody-drug conjugates (ADCs)
Antibody-drug conjugates (ADCs) consist of a monoclonal antibody against a tumor associated antigen (TAA) on the surface of cancer cells that is conjugated to a cytotoxic drug (known as the payload) through a chemical linker [131]. This directed delivery strategy combines the precision of antibody targeting to antigens with the effective cell-killing activity of the payload, resulting in reduced systemic exposure and thus reduced toxicity [132]. ADCs are one of the rapidly growing areas in cancer drug development, with five ADCs having been approved by the FDA for clinical use (Adcetris, Kadcyla, Besponsa, Mylotarg, and Polivy) and over 100 ADCs being evaluated in clinical trials [133]. The discovery of a series of HCC-specific TAAs provides the rationale for applying ADCs in HCC treatment: glypican-3 (GPC3) [134], AFP, epithelial cell adhesion molecule (EpCAM) [135], tumor endothelial marker 1 (TEM1, or endosialin) [136], MAGE-A, NY-ESO-1 [137], and claudin-6 (CLDN6) [138]. In the past, there were pre-clinical studies of the anti-tumor effects of Anti-CD147 ILs-DOX [139], G7mAb-DOX [140], and MetFab-DOX [141] in HCC, as well as phase I/II trial (NCT01631552) of IMMU-132 [142] in solid tumors including HCC. Recently, Fu et al. designed and synthesized two anti-GPC3 ADCs with duocarmycin SA and pyrrolobenzodiazepine, which were named hYP7-DC and hYP7-PC, and demonstrated that these two ADCs had antitumor activity against both GPC3-positive HCC cell lines and xenograft models [143]. The hYP7-DC was also found to have synergistic effects with gemcitabine, and hYP7-PC induced tumor remission in a mouse model. Kong et al. developed a novel ADC drug, CLDN6-DM1 (DM1 being referred to as mertansine), and the preclinical data of CLDN6-DM1 showed robust antitumor effects against HCC cell lines and a xenograft mouse model, both as monotherapy and combined with sorafenib [144] (Table 4). All these efforts provide a new strategy for the treatment of HCC. Clinical trials of ADCs for HCC treatment in the near future are being anticipated.

Chimeric Antigen Receptor T Cells (CAR-T)
CAR is a fusion protein consisting of an antigen-binding domain (usually referring to the extracellular portion of the antibody), the signaling domains of the T cell receptor (TCR) chain, and additional co-stimulatory domains (e.g., CD28, OX40, CD137); therefore, CAR recognition is independent of major histocompatibility complex (MHC) restriction. Loss of MHC-associated antigen presentation by tumor cells is a key mechanism of cancer immune escape [148], and one strategy to overcome this escape is to genetically engineer autologous T cells with CAR, expand CAR T cells in vitro and infuse them back into patients to attack cancer cells [149]. In 2017, tisagenlecleucel (KYMRIAH) [150,151] and axicabtagene ciloleucel (Yescarta) [152,153] received successive FDA approval as CAR-T therapy for the treatment of acute lymphoblastic leukemia and diffuse large B-cell lymphoma. Breyanzi (lisocabtagene maraleucel) became the third FDA-approved CAR-T therapy in 2021 for certain types of non-Hodgkin lymphoma, including diffuse large B-cell lymphoma [154,155]. Positive results of CAR-T therapy in hematologic malignancies raised hope for CAR-T therapy against solid tumors [156]. In preclinical studies, engineered GPC3-CAR-T cells could effectively eliminate GPC3-positive HCC cells in HCC xenograft mouse models, demonstrating their clinical potential for treatment [157][158][159]. There are currently 20 clinical trials in progress testing CAR-T therapy for HCC, 12 of which target glypican-3 (GPC-3) ( Table 5). In 2020, Shi et al. [160] reported two sequential phase I studies, NCT02395250 and NCT03146234, which showed that GPC3-CAR-T cells have an initial safety profile and anti-tumor activity in advanced HCC patients. There were two confirmed partial responses of 13 patients, and one patient with sustained stable disease exhibited long-term survival

Bispecific T cell Engager (BiTE)
BiTEs are generated by two different single-chain variable fragments (scFvs) covalently linked by small peptides which are derived from the antigen-binding domains of anti-CD3 and anti-TAA antibodies [161]. BiTEs are believed to possess remarkable efficacy for tumor immunotherapy because of their mini size, high flexibility, and high-affinity binding between effector and target cells [162]. Blinatumomab, an anti-CD19/CD3-BiTE, has been successful in treating hematologic malignancies and was approved by the FDA for treating acute lymphocytic leukemia [163]. Zhang et al. reported that the anti-EpCAM BiTE 1H8/CD3 had effective antitumor activity against HCC cells both in vitro and in vivo [145]. Kebenko et al. [146] reported a phase I study of solitomab (MT110, targeted EpCAM and CD3, NCT00635596) for advanced solid tumors. The study was associated with dose-limiting toxicities, based on which it failed to escalate the dose to potentially therapeutic levels. ERY974 [147], an anti-GPC3/CD3 BiTE, had been evaluated for safety and efficacy in GPC3+ HCC patients in a phase I trial (NCT02748837), with no results yet (Table 4). To improve the efficiency of BiTEs and reduce on-target off-tumor toxicities, the identification of target molecules abundantly expressed by tumor cells compared to nonmalignant cells is necessary. In addition, BiTEs face the complex TME of solid tumors and drug resistance due to the loss of the target antigen expression. These are all key issues to be addressed in the future.

Suggested systemic treatment strategies
Recently, approved systemic agents do offer more options for the treatment of advanced HCC patients; however, the lack of head-to-head comparative studies between first-and second-line regimens for systemic treatment has led to confusion when making clinical decisions. Thus, systematic review and network metaanalysis are particularly important at this stage. A metaanalysis of 14 clinical trials conducted by Sonbol et al. showed that atezolizumab + bevacizumab was superior in the first-line setting compared to sorafenib (HR= 0.58), lenvatinib (HR=0.63), and nivolumab (HR=0.68). In the second-line setting, all drugs showed a PFS benefit compared with placebo, with only regorafenib and cabozantinib converting to OS benefit. In addition, cabozantinib and regorafenib showed a PFS advantage over ramucirumab and pembrolizumab, but only regorafenib had a statistically significant OS benefit compared to ramucirumab (HR=0.71). For patients with advanced HCC whose AFP levels were at or above 400 ng/mL, there was no significant difference in PFS or OS among regorafenib, cabozantinib, and ramucirumab [164]. Park et al. incorporated 13 first-line studies and 11 secondline studies to obtain similar results. In first-line therapies, atezolizumab + bevacizumab had the greatest OS benefit, while lenvatinib had the greatest ORR benefit. In second-line therapies, cabozantinib had the greatest PFS benefit and ORR benefit compared to placebo [165]. In 2020, ASCO assembled an expert panel to perform a

Conclusions
Undoubtedly, the systemic treatments of advanced HCC have ushered in prime time (Fig. 4). Angiogenesis and immune evasion are two common features of cancers [167]. As a pioneer in the era of immune combination therapy, the atezolizumab + bevacizumab regimen is the first combination proven to be more effective than sorafenib in the first-line treatment of advanced HCC, which is regarded as a milestone and an encouraging breakthrough in the treatment of advanced HCC. Notably, with atezolizumab + bevacizumab being promoted as a preferred first-line treatment [164][165][166], the following questions arise: are the existing second-line treatment options still valid? Can the existing first-line TKIs become the second-line treatment after immunotherapy? These points remain to be addressed by future studies. As mentioned earlier, several phase III trials in the firstline setting are underway (e.g., CheckMate-9DW, LEAP2, COSMIC-312, HIMALAYA, Table 3), and the potential first-line treatment options are likely to continue to increase, making clinical decisions more challenging.
Our review not only lists the new drugs approved by FDA for HCC treatment, but also summarizes novel drug candidates under investigation. Selective inhibitors of VEGFR, c-MET, TGFβ, endoglin, and FGFR4 have shown good tolerability and efficacy in phase I/II or phase III studies, making these compounds promising for future treatment. In the field of immunotherapy, in addition to targeting PD-1/PD-L1 and CTLA-4, an increasing number of ICIs, such as LAG-3, TIM-3, and GITR, are gradually demonstrating their efficacy. Meanwhile, as more and more HCC-specific TAAs are identified, other immunotherapeutic approaches, including ADCs, CAR-T, and BiTEs, are being further developed.
The development of new drugs for HCC treatment may not be as impressive as for some other tumors, and few predictive biomarkers have yet been identified (except AFP for ramucirumab and perhaps FGF19 for FGFR4 inhibitors). Based on the results of recent clinical trials, it seems that a single drug may not be sufficient for the treatment of HCC [168], therefore combination therapy represents a major research direction for the systemic treatment of advanced HCC. In addition, it is important to find biomarkers that can predict the treatment response to guide systemic therapy strategy. With the continued development of new therapeutic strategies, it is expected that the treatment outcomes for HCC will be greatly improved in the future.