**3. The Advent of Immune Checkpoint Inhibitors**

Despite the benefit in OS with sequential TKI treatment, the prognosis of patients with advanced HCC remains poor [9,23]. The reasons for this include, besides the sub-optimal tumoricidal activity of these drugs, the progression of the underlying liver disease, the advanced median age of this cohort of patients (approximately 70 years), and the presence of substantial comorbidities, which are very frequent in these subjects and, overall, make them a particularly vulnerable cohort [24].

In this context, ICIs have increasingly been investigated in the last years, with extremely encouraging results both in the first- and second-line setting, further boosting a rising number of clinical trials using ICIs alone or combined with other anti-tumoral drugs or with locoregional treatment. The rationale for the use of ICIs in HCC relies on the fact that HCC arises in a context of chronic inflammation and an altered tumor microenvironment, with the presence of tumor-infiltrating lymphocytes expressing PD1, which is a recognized key enabling factor beyond tumor cell-intrinsic molecular aberrations [25–27]. Moreover, the presence of PD1-expressing lymphocytes in HCC samples has been correlated with this outcome [28,29]. In this regard, Sia et al. have recently proposed a novel HCC classification based upon the tumor immune status: according to this classification, about 30% of HCCs could be categorized into an 'immune class', with high levels of immune cell infiltration, expression of PD-1 and/or PD-L1, activation of interferon-γ signaling, and markers of cytolytic activity [30]. Within this class, two distinct subclasses have been identified: the 'active immune' and the 'exhausted immune' classes, characterized by markers of an adaptive T-cell response or of an exhausted immune response, respectively [30]. The latter subclass is the ideal target of immunotherapy. The in-depth description of the molecular mechanisms involved in the tumor microenvironment of HCC is beyond the aim of this article, but it is worth pointing out that interactions between cancer cell antigens and the antigen-presenting cells lead to a priming of T-cells and their eventual migration into the tumor microenvironment. Physiologically, the T-lymphocytes' recognition of neoplastic antigens is followed by a T-cell-mediated killing of cancer cells [31]. This process is finely modulated at a local and general level by several mechanisms including immune checkpoints, which play a pivotal role in such modulation, as they suppress T-cell activity to inhibit eventual over-activation of the immune system and maintain self-tolerance. Thus, immune checkpoints physiologically prevent hyperimmune responses leading to tissue damage. Malignancies exploit these molecular mechanisms (immune checkpoints) to escape from the immune system recognition. In other words, ICIs act as anti-neoplastic agents by inhibiting negative feedback pathways of the immune system that mediate immune escape.

The most largely studied immune checkpoints are PD-1 and cytotoxic T-lymphocyteassociated protein 4 (CTLA-4). The pathological activation of PD-1 by its ligands, in particular PD-L1, expressed by cancer cells, can result in the immune escape of the tumor [32,33]. CTLA-4, which is mainly expressed on T-cells, regulates T-cell activity in physiological conditions, preventing an excess in T-cell responses and a hyperactivation of the immune response. Inversely, in pathological (neoplastic) conditions, CTLA-4 activation inhibits in the activation, proliferation, and production of tumor antigen-activated T-cells in the tumor microenvironment [32,33]. In the HCC tumor microenvironment, T-regulators (T-regs) express both CTLA-4 and PD-1 [28,32].

#### **4. Immune Checkpoint Inhibitors in HCC**

Tremelimumab, a CTLA-4-blocking monoclonal antibody, was the first ICI showing benefits in the treatment of HCC. This agent was tested in 2013 by Sangro et al. in a Phase II open-label trial that enrolled 21 patients with advanced HCC who were either sorafenibnaïve (76.2%) or -experienced, and a significant proportion of them were classified as Child–Pugh class B (43%) [10]. The positive results in terms of both safety and antitumor activity (partial response rate (PRR) 17.6%; disease control rate (DCR) 76.4%; TTP 6.48 months (95%CI, 3.95–9.14)), were instrumental in stimulating the research in immune checkpoint blockade in both first- and second-line treatment of HCC. In the last years, the effects of ICIs in HCC have been tested alone or in combination with other ICIs or combined with agents targeting the VEGFR. Currently available immunotherapy-based regimens and those under Phase III clinical investigation are summarized in Figure 1.

**Figure 1.** Possible HCC treatments with ICIs. Atezolizumab plus bevacizumab has been approved as a first-line treatment, whereas nivolumab with or without ipilimumab and pembrolizumab gained FDA approval as second-line treatments. Selected Phase III trials (orange squares) are testing ICIs alone or in combination or combined with other agents in the first and second-line setting, and in the adjuvant and neo-adjuvant setting as well.

#### *4.1. Immune Checkpoint Inhibitors in Monotherapy*

Following the encouraging results of the Phase II tremelimumab study, nivolumab, a monoclonal antibody targeting PD1, demonstrated a single-agent activity in the Phase Ib/II open-label, non-comparative, Checkmate 040 trial [11]. The initial trial included 262 sorafenib-naïve and -experienced patients assigned to a dose-escalation (48 subjects) or to a dose-expansion (214 subjects) phase. In the dose-expansion phase, the investigator-

assessed overall ORR was 20%, with 3 complete responses (CR) and 39 partial responses (PR). Particularly, ORR was 22.5% for sorafenib-naive and 18.7% for sorafenib-experienced patients. Median OS was 29 months for sorafenib-naïve group and 15 months for the sorafenib-experienced group. The most impressive was the duration of response of 9.9 months amongst patients who had an objective response, which led the US FDA to grant accelerated approval to nivolumab as second-line therapy for patients with advanced-stage HCC previously treated with sorafenib [11]. In this subgroup, the ORR confirmed by blinded independent central review was 14.3% by Response Evaluation Criteria In Solid Tumors (RECIST) 1.1 and 18.2% by modified RECIST (mRECIST) criteria. Of note, the median duration of response was the longest ever seen in a second-line setting: 16.6 months [34,35]. However, the expectations raised by the results of this study were disappointed in a subsequent Phase III randomized trial (CheckMate-459) testing nivolumab versus sorafenib, as the anti-PD1 agent failed to demonstrate superiority as compared to the TKI [14]. Still, the study results confirmed clinically meaningful improvements in OS (16.4 versus 14.7 months), ORR (15% for nivolumab versus 7% for sorafenib), and CR (14 versus 5 patients). Moreover, nivolumab demonstrated a favorable safety profile, consistent with previous reports and, of particular interest, the quality of life was better in the nivolumab treatment arm [14]. The long survival of the sorafenib arm (median OS of about 15 months) was an unexpected outcome that negatively impacted the study results and that probably reflects the improved tailored management of patients with advanced HCC in the last decades, as well as physicians' familiarity with the TKI.

Another ICI that has been tested with favorable outcomes in monotherapy for advanced HCC is pembrolizumab, a monoclonal antibody targeting PD-1. Promising results came from the Phase II trial KEYNOTE 224, which showed good responses (ORR 17%, DCR 61%) and a good safety profile of pembrolizumab in patients who were intolerant to, or progressed under, sorafenib [12]. These results prompted Finn et al. to conduct the KEYNOTE-240 trial enrolling 413 patients who failed sorafenib and who were randomized 2:1 to pembrolizumab or placebo [13]. The survival in the pembrolizumab arm was among the highest ever reached in the second-line setting, being approximately 14 months (95%CI, 11.6–16.0) for pembrolizumab versus 10 months (95%CI, 8.3–13.5) for placebo (HR, 0.781; 95%CI, 0.611–0.998; *p* = 0.0238). Nevertheless, even this study failed to reach statistical significance due to the long survival of the control arm, reflecting once more the advances in the clinical management of advanced HCC. The safety profile of the drug was good, confirming the positive results of the Phase II study and the previous experience with nivolumab.

Despite the apparently "negative" results of these studies, likely due to issues related to their design requesting an overwhelming superiority of the tested ICIs over sorafenib, several positive aspects capturing the attention of researchers and clinicians were the overall objective response to nivolumab and pembrolizumab in 15–20% of cases, the durable antitumor responses, and the long-term OS in responding patients. Based on these peculiar results, the FDA granted conditional approval for these ICIs in the second-line setting.

Currently, results from the ongoing Phase III non-inferiority trial testing tislelizumab, a monoclonal antibody targeting PD-1, versus sorafenib (RATIONALE-301 trial) and those of the Phase III HIMALAYA study, testing durvalumab—an anti-PD-L1 monoclonal antibody—alone or in combination with tremelimumab versus sorafenib, are eagerly awaited [36].

As far as the safety profile of ICIs is concerned, the results of the pilot study by Sangro et al. on tremelimumab and those of the CheckMate and Keynote trials showed reassuring safety profiles for these agents, coherent with previous reports testing the use of these drugs in other cancer types [10–12,37]. As compared with the standard of care (i.e., sorafenib and lenvatinib), ICIs are generally better tolerated and have comparable or even lower rates of toxicity. The pathophysiology of adverse events (AEs) occurring during immunotherapy is related to their mechanism of action as the inhibition of physiological immune checkpoints may trigger immune-related AEs (irAEs) targeting the skin, gut, thyroid, adrenal glands, lung, and the liver itself, which may be a particularly worrisome complication in

a population with an already impaired liver function [38,39]. Most frequent any grade AEs in patients treated with ICIs for other cancer types are skin AEs (rash and pruritus), colitis, hyper- or hypothyroidism, hepatitis, and pneumonitis. Skin AEs occur in about 13–35% of cases, being grade > 3 only in a minority of cases (<3%) [38,40]. Grade 1 and 2 skin AEs are usually easily managed with emollients, oral anti-histamines, and topical steroids, whereas grade ≥ 3 reactions require oral corticosteroids administration and the discontinuation of the immunotherapy until the skin AE has reverted to grade 1 [38]. Thyroid dysfunction has been reported in a variable proportion of cases (5–20%), but these events are rarely severe and rarely require treatment discontinuation or hormonal replacement treatment or corticosteroids administration [38]. The frequency of colitis ranges from 2% to 22% [38,40], being more frequent and severe in patients treated with anti-CTLA4 agents [38,40]. Again, the incidence of high-grade colitis is very low, being around 1–2% [40]. Patients with non-severe diarrhea should be treated with anti-diarrheal, fluid replacement, and electrolytes; conversely, patients with grade ≥3 diarrhea or persistent grade 2 diarrhea should discontinue ICIs and receive intravenous (i.v.) corticosteroids. In case of lack of response to corticosteroids, infliximab should be prescribed [38]. Pneumonitis occurs in 2–4% of patients, with grade ≥ 3 events representing only 1% to 2% of cases [38,40], and the frequency of fatal pneumonitis and that of treatment discontinuation (due to this AE) are extremely low (0.2% and 0.2–4%, respectively) [38]. In the case of documented or high suspicion of immune-related pneumonitis, immunosuppressive treatment should be started immediately. In grade 1 to 2 pneumonitis, treatment consists of oral steroids (prednisone 1 mg/kg daily), whilst patients with grade 3 to 4 pneumonitis should be hospitalized and treatment should consist of high-dose i.v. corticosteroids. In these severe cases, immunotherapy should be permanently discontinued. With regards to the occurrence of treatment-related hepatitis, which occurs in a proportion of 5% to 10% of patients (among which 1–2% are grade 3) [38,40], in the presence of grade ≤ 2 transaminases elevation, checkpoint inhibitor therapy should be withheld and transaminases and bilirubin should be measured twice weekly. Persistent grade 2 elevation lasting longer than 2 weeks, after having ruled out other causes, should be treated with corticosteroids at a dose of 1 mg/kg/day (methyl)prednisolone or equivalent. Upon improvement, re-challenge with ICIs may be attempted after corticosteroid tapering. In the absence of improvement despite the initiation of corticosteroids, the dose should be increased to 2 mg/kg/day of (methyl)prednisolone or equivalent and checkpoint inhibitor therapy should permanently be discontinued [38]. In the instance of grade 3 or 4 transaminase or total bilirubin elevation, checkpoint inhibitor therapy should be permanently discontinued, and corticosteroids started at 1–2 mg/kg/day (methyl)prednisolone or equivalent. If the absence of response to corticosteroids within 2–3 days, mycophenolate mofetil should be added at 1000 mg twice daily. If no improvement is seen, liver biopsy should be considered. However, ICI-related hepatitis usually resolves within 4–6 weeks with appropriate treatment; therefore, if no improvement is detected in this time frame, other contributory causes should be reconsidered and the initial diagnostic work-up should be repeated.

Overall, the available evidence suggests that, although common, irAEs can be easily managed in most cases by delaying the subsequent scheduled administrations, and with the administration of corticosteroids in severe cases [39]. In HCC studies, approximately 90–98% of patients experienced any AE during treatment, with up to 50% of them being grade 3 or higher [10–14]. However, similar rates of AEs have been recorded in randomized controlled trials in the respective placebo arms as well [13]. With regards to treatment-related AEs, grade ≥ 3 AEs have been reported in approximately 20% of cases for nivolumab and pembrolizumab monotherapy [13,14]; among them, the most frequent AE in the Keynote-240 and CheckMate-040 studies was aminotransferase increase (about 4–5% and 6–10%, respectively) [11,13]. This event is of particular concern in patients with cirrhosis due to the potential deterioration of liver function and to the peculiar risk of corticosteroid-related AEs in these subjects. However, current data show that ICIs are safe in well-selected cohorts of patients with cirrhosis and preserved liver

function (Child–Pugh class A), with no safety alerts as compared with patients without cirrhosis treated with ICIs for other cancer types [9,41]. The available evidence thus suggests that cirrhotic patients with HCC should not be at increased risk of liver irAEs, but close monitoring of liver function tests should be performed in cirrhotic patients treated with ICIs. Treatment-related serious AEs such as pneumonitis and colitis occurred in a minority of patients (<1%), as reported in the literature for immunotherapy in other cancer types [11,12]. Definite data on the safety and tolerability of ICIs in Child–Pugh class B patients, which represent a significant proportion of advanced HCC patients, are lacking. However, those from the CheckMate-040 trial are reassuring, since only 4 out of 49 patients with Child–Pugh class B reported treatment-related hepatic events, and only 2 of them needed treatment discontinuation [38]. Moreover, similar results regarding the safety of nivolumab and pembrolizumab in patients with Child–Pugh class B have been observed by Scheiner et al. in a real-life cohort of HCC patients [41]. Taken together, the available evidence suggests the safety profile of ICIs in the HCC population is good in selected cases with well-preserved liver function and that ICIs may be safely administered in Child–Pugh class B patients as well.

#### *4.2. Dual Immune Chechpoint Blockade*

Based on the hypothesis that anti-PD-1 and anti-CTLA4 agents may have a synergistic effect by inhibiting two different steps of the immune checkpoint system, combinations of anti-PD1 and anti-CTLA4 are underway. A Phase III trial with dual treatment with nivolumab plus ipilimumab, a CTLA-4 monoclonal antibody, in the first-line setting (CheckMate 9DW, NCT04039607) is underway. This trial was supported by the positive results observed in the cohort 4 (nivolumab plus ipilimumab) of the Checkmate-040 trial in the second-line setting [42]. In this study, patients were randomized 1:1:1 to either nivolumab 1 mg/kg plus ipilimumab 3 mg/kg, administered every 3 weeks (4 doses), followed by nivolumab 240 mg every 2 weeks (arm A); nivolumab 3 mg/kg plus ipilimumab 1 mg/kg, administered every 3 weeks (4 doses), followed by nivolumab 240 mg every 2 weeks (arm B); or nivolumab 3 mg/kg every 2 weeks plus ipilimumab 1 mg/kg every 6 weeks (arm C). Treatment combination had manageable safety, promising ORR, and durable responses. The arm A regimen showed the greatest benefits in terms of ORR (32% versus 27% and 29% in arms B and C, respectively) and OS (22.8 months (95%CI, 9.4—not reached) in arm A versus 12.5 months (95%CI, 7.6–16.4) in arm B and 12.7 months in arm C (95%CI, 7.4–33.0) [43]. Any grade treatment-related AE occurred in 94% of cases in arm A, 71% in arm B, and 79% of cases in arm C. Among them, 53% of patients in arm A, 29% of patients in arm B, and 31% of patients in arm C had grade 3 or 4 treatment-related AEs. Arm A also had higher rates of irAEs and irAEs leading to treatment discontinuation (18%), as compared with arms B and C (6% and 4%, respectively). Consequently, in arm A, 16% of patients stopped treatment: 6% of them due to treatmentrelated hepatitis, 6% due to pneumonitis, and 4% due to diarrhea/colitis [43]. However, most cases of patients presenting AEs continued treatment and the AEs resolved with standard management, while only 1 treatment-related death due to pneumonitis was reported (0.6%) [43]. Importantly, among patients who were re-challenged with nivolumab or ipilimumab after experiencing an irAE in any category, no patients experienced an event recurrence after the re-challenge [43]. Considering the outstanding OS and ORR obtained in arm A, these results suggest that nivolumab plus ipilimumab may provide improved efficacy in terms of ORR, and, potentially, of survival with an acceptable safety profile. Based on this evidence, this dual treatment received accelerated approval in the US as second-line treatment for HCC.

In the first-line setting, a Phase III trial (HIMALAYA) is testing the PD-L1 inhibitor durvalumab alone and in combination with tremelimumab, compared with sorafenib. This study was designed on the basis of the findings from a Phase I/II, randomized, open-label study that included patients progressing under, intolerant to, or refusing sorafenib [44]. Patients were randomized 1:2 to different tremelimumab plus durvalumab combinations,

and safety was the primary endpoint. Patients assigned to the high-dose tremelimumab arm (i.e., tremelimumab 300 mg plus durvalumab 1500 mg 1 dose followed by durvalumab every 4 weeks) had the highest confirmed ORR (duration of response not reached) and longest OS (18.7 months (10.8—not reached)) [44]. Grade 3 or 4 treatment-related AEs rates were comparable to those occurring in the nivolumab plus ipilimumab trials, being 35% in the high-dose (300 mg) tremelimumab arm and 25% in the low-dose (75 mg) tremelimumab arm. Discontinuation of the study drug due to AEs was 10.8% and 6% in the high- and low-dose arm, respectively, but no deaths were attributed to treatment.

In summary, dual checkpoint blockade may improve OS in HCC patients, but consistent evidence is still scarce. As might have been expected, the trials testing ICIs in dual treatment reported higher rates of AEs in comparison with ICIs used in monotherapy, but in most cases, the safety profile was consistent in presentation and management with that of monotherapy. Taking into consideration the poor prognosis of patients with advanced HCC, the benefit/risk ratio may still favor the dual treatment strategy. Current trials with dual checkpoint blockade are reported in Table 1.

**Table 1.** Ongoing clinical trials with immune checkpoint inhibitors, alone or in combination with other agents, in HCC.



**Table 1.** *Cont.*

HCC, hepatocellular carcinoma; DLTs, dose-limiting toxicities; AEs, adverse events; OS, overall survival; PFS progression-free survival; ORR, overall response rate.
