Conflict of interest management

As outlined by IOM standards, all financial relationships of expert panel members that might result in actual, potential, or perceived conflicts of interest were individually reported. Disclosures were made prior to the onset of manuscript development and updated on an annual basis. In addition, panel members were asked to articulate any actual or potential conflicts at all key decision points during guideline development, so that participants would understand all possible influences, biases, and/or the diversity of perspectives on the panel. Although some degree of relationships with outside interests are to be expected among experts, panel candidates with significant financial connections that may compromise their ability to fairly weigh evidence (either actual or perceived) were not eligible to participate in guideline development.

Recognizing that guideline panel members are among the leading experts on the subject matter under consideration and guideline recommendations should have the benefit of their expertise, any identified potential conflicts of interests were managed as outlined in SITC’s disclosure and conflict of interest resolution policies. As noted in these policies, panel members disclosing a real or perceived potential conflict of interest may be permitted to participate in consideration and decision-making of a matter related to that conflict, but only if deemed appropriate after discussion and agreement by the expert panel.

The financial support for the development of this guideline was provided solely by SITC. No commercial funding was received.

Recommendation development

Panel recommendations are based on literature evidence, where possible, and clinical experience, where appropriate.19 Consensus for the recommendations here was generated by open communication and scientific debate in small-group and whole-group settings, surveying and responses to clinical questionnaires, as well as formal voting in consensus meetings.

For transparency, a draft of this CPG was made publicly available for comment during the development process and prior to publication. All comments were evaluated and considered for inclusion into the final manuscript according to the IOM standard.

Evidence rating

The evidence- and consensus-based recommendations of the panel were refined throughout the development process in order to obtain the highest possible agreement among the experts, however, the minimum threshold was defined as 75% approval among the voting members. Evidence supporting panel recommendations was graded according to the Oxford Centre for Evidence-Based Medicine (OCEBM) Levels of Evidence Working Group ‘The Oxford Levels of Evidence 2’ (2016 version). A summary of the OCEBM grading scale may be found in table 1. The level of evidence (LE) for a given recommendation is expressed in parentheses following the recommendation (eg, LE: 1). Recommendations without an associated LE were based on expert consensus.

Initial HCC diagnosis

The initial diagnostic workup of HCC typically comprises a histologic analysis of tumor samples obtained by biopsy or surgery, cross-sectional imaging, a detailed analysis of the liver’s condition with laboratory studies, and an assessment of the potential etiology of the HCC including investigations of HBV and HCV viral status. Guidelines for surveillance screening, initial diagnosis, and staging of HCC have been developed by multiple organizations including, but not limited to, the American Association for the Study of Liver Diseases (AASLD),7 the American College of Gastroenterology (AGC),20 the European Society for Medical Oncology (ESMO),21 the European Association for the Study of the Liver (EASL)22 and the Japan Society of Hepatology (JSH).23 These organizations and others have also put forth guidelines for non-immunotherapeutic approaches for the treatment of HCC.

HCC may be identified using computerized tomography (CT) and magnetic resonance imaging (MRI) with Liver Imaging Reporting And Data Systems (LI-RADS).24 The LI-RADS system provides a standardized approach for radiologists to communicate with the treating physicians and provides a certain level of confidence that a lesion in a cirrhotic liver or a liver at risk for cirrhosis presents as HCC on imaging.25 LI-RADS staging ranges from LR-1, for lesions that are definitely benign, to LR-5, which represents 100% probability of being HCC. The LI-RADS system acknowledges that limitations exist, and has included an LR-NC (for non-categorizable) category where diagnostic possibilities cannot be meaningfully narrowed. LI-RADS is endorsed by the AASLD,7 as well as by the Organ Procurement and Transplantation Network (OPTN)/United Network for Organ Sharing (UNOS). Contrast agents have also greatly enhanced the diagnostic accuracy of MRIs. Multiple meta-analyses have determined that gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI has superior sensitivity, specificity, and diagnostic odds ratio (OR) as compared with multidetector CT.26–28 However, most published guidelines do not recommend one imaging modality over the other.

Patients with HCC often present with underlying cirrhosis—two conditions with independent mortality risks. It is essential for a care team comprised of multiple specialties, including perspectives from both hepatology and oncology, to be established early so that a treatment plan that addresses all of the complex needs of a patient with HCC may be developed.29 A multidisciplinary tumor board review of liver lesions is recommended for HCC diagnosis and management plans, particularly for patients with tumors that may be eligible for transplant, surgery, or liver-directed treatments.

Historically, avoiding tumor biopsy has been acceptable practice in patients with cirrhosis and imaging characteristics consistent with HCC. One concern of performing biopsies in this disease has been the putative risk for tumor dissemination outside the liver via needle track seeding. The occurrence of needle track seeding appears to be uncommon in the published literature, however, with incidence rates estimated to be as low as 2.7% overall, or 0.9% per year.30 While biopsy may be less encouraged in certain clinical scenarios such as in patients where liver transplants are being considered, histologic diagnosis is increasingly encouraged for the diagnosis of HCC, particularly for more advanced tumors requiring systemic therapy. Other primary liver tumors such as cholangiocarcinoma or mixed cholangiohepatoma can occasionally present very similarly to HCC, and the treatment for these tumors can be distinct. Other entities such as metastatic neuroendocrine cancers can similarly demonstrate arterial enhancement on multiphase imaging.31 Additionally, in rare instances, tissue biopsy may uncover certain genetic alterations that render a patient eligible for a tissue-agnostic therapy or a clinical trial.

HCC staging

An ideal staging system in HCC serves two purposes: treatment indication and prognostic prediction. A variety of staging systems have been developed, and their performance and validation varies. While some staging systems focus on pathology, others incorporate radiological characteristics, serum biomarkers, liver function, and performance status. In most solid tumors, staging is performed at the time of surgery using resected specimens. The Tumor-Node-Metastases (TNM) classification, developed by the American Joint Committee on Cancer (AJCC), classifies the primary tumor (T) based on size, number, and vascular invasion.32 However, the TNM classification is not currently used to guide treatment for HCC. Also, importantly, the TNM classification should not be confused with the radiologic T-staging system used by LI-RADS and OPTN/UNOS, which is summarized in table 2.

Radiographic T-staging is of limited pretreatment prognostic and predictive value for patients being considered for systemic therapy, as the system does not take into account liver function, which is an important risk factor for patients with HCC. Several alternative staging or scoring systems have been developed, including the Barcelona-Clinic Liver Cancer (BCLC) system,33 Cancer of the Liver Italian Program (CLIP),34 Japan Integrated Staging (JIS),35 Chinese University Prognostic Index (CUPI),36 Groupe d’Etude et de Traitement du Carcinome Hépatocellulaire (GETCH)37 plus many others.

The BCLC system, summarized in table 3, has gained wide recognition and has been endorsed by multiple international hepatology associations including AASLD6 and EASL. Several large-scale cohort studies have validated the BCLC system, including in Korean patients with treatment‐naïve HCC,38 US patients,39 and Italian patients undergoing radical surgery.40 In addition, scoring by the BCLC system has been reported and studied in subgroup analyses for most of the phase III studies done in advanced HCC.

HCC is a highly heterogeneous disease with varied underlying etiologies depending on geography and demographics. Studies comparing the performance of staging systems for predicting prognosis have returned conflicting results depending on the patient population investigated and the treatments administered. The JIS score showed the best ability to predict OS by disease stage in an analysis of Japanese patients,41 whereas an analysis of 1,713 prospectively enrolled patients with HCC in Taiwan found that CLIP was the best prognostic model in patients undergoing both curative and non-curative treatments.42 In the advanced and metastatic disease setting, another comparison of the prognostic value of different systems determined CLIP and CUPI to be the most reliable staging systems for patients with HCV and HBV etiologies, respectively.43 In these analyses of patients with advanced disease in need for systemic therapy, BCLC and TNM lacked prognostic value.

Liver function assessment is a critical component of HCC treatment that is required for every patient. Some of the staging systems embed within them the Child-Pugh classification, recognizing the need for assessing the extent of liver functionality as part of the staging of the disease. The Child-Pugh score evolved over time from the original system built in 1973 to help assess for survival of patients with bleeding esophageal varices.44 Of note, in the original Pugh effort, none of the patients had HCC. The system evolved into a five-parameter staging system which consists of three laboratory values (serum albumin, bilirubin, and prothrombin levels) and two clinically assessed variables (presence and degree of ascites and hepatic encephalopathy). A final score ranging from 5 to 15 is calculated based on the range of laboratory values and severity of clinical symptoms, and then classified into one of three classes: A (5–6), B (7–9), and C (10–15).44 45 Median survival of untreated HCC has been shown to be approximately 2.5 times lower in patients with Child-Pugh B disease compared with those with Child‐Pugh A.46 Recently, however, the limitations and subjectivity involved in the grading of clinical variables have called into question Child-Pugh scores in assessing liver function in HCC.47

The albumin-bilirubin (ALBI) grade, a simpler model to assess liver function based only on serum albumin and bilirubin, has been validated in study cohorts from multiple geographic regions and multiple clinical scenarios, including patients undergoing resection and sorafenib treatment. The score is calculated as (log₁₀ bilirubin [µmol/L]×0.66) + (albumin [g/L] × −0.0852), leading to three possible grades: ALBI Score ≤ −2.60 (ALBI grade 1), ALBI Score > −2.60 to ≤ −1.39 (ALBI grade 2), and ALBI Score > −1.39 (ALBI grade 3).48 The ALBI grade demonstrated superior prognostic value to the Child-Pugh score in a study of patients with HCC treated with radioembolization, particularly within patients with Child-Pugh A disease.49 ALBI grade also predicts OS after surgical resection (p<0.001), transarterial chemoembolization (TACE) (p<0.001) and sorafenib treatment (p<0.001), with independent prognostic value across BCLC stages, geographic regions (p<0.001),50 and for cancers being treated with immunotherapy.50

Diagnostic biomarkers

Several biomarkers have been put forward to predict prognosis in HCC, yet none are currently routinely used to guide treatment decisions for patients being considered for immunotherapy. Serum alpha-fetoprotein (AFP) has been the most widely used marker to increase the suspicion for a diagnosis of HCC, and has been included in international guidelines.21 51 However, the value of AFP as a surveillance marker remains controversial,52 and establishing a threshold value to diagnose HCC remains a challenge.52 53 Tumor-derived AFP has also been implicated in impaired dendritic cell function.54 Glypican 3 (GPC3), an antigen that is highly expressed on tumor cells and minimally present on healthy tissues,55 has been proposed as a serum biomarker for HCC and is being pursued as a target for chimeric antigen receptor (CAR) T cell therapies.56 However, neither AFP nor GPC3 have demonstrated predictive power for patients being treated with ICIs, although this is an active area of research.

The GALAD score, which determines risk of HCC based on patient sex, age, and serum levels of AFP, AFP isoform L3, and des-gamma-carboxy prothrombin has been validated for detection of HCC in patients with non-alcoholic steatohepatitis (NASH) with and without cirrhosis.57 A combination of GALAD and ultrasound (GALADUS) score has been shown to further improve performance, with an area under the curve of 0.98 (95% CI 0.96 to 0.99; cut-off −0.18; sensitivity 95%; specificity 91%) in a single-center cohort of 111 patients with HCC and 180 controls with cirrhosis or chronic HBV.58 In March of 2020, the FDA granted breakthrough device designation to the Elecsys GALAD score to aid in early diagnosis of HCC (for further discussion of immunotherapy-specific biomarkers, including PD-L1 status, see the Patient selection and management section).

Panel recommendations

• A multidisciplinary tumor board review of liver lesions is recommended for HCC diagnosis and the development of a management plan.

• Notwithstanding that LI-RADS-5 is nearly 100% specific for HCC (LE: 1), histologic confirmation is recommended for patients with unresectable disease particularly prior to the initiation of systemic therapy. Histologic diagnosis is mandatory for non-cirrhotic patients.

• Despite the controversy regarding the scoring and staging systems that could be used, before initiation of systemic therapy, an evaluation of liver function, including aspartate transaminase (AST)/alanine aminotransferase (ALT), bilirubin, prothrombin time (PT)/international normalized ratio (INR), albumin, plus platelets, is critical (LE: 2).

• For patients being considered for immunotherapy, an HCC-specific staging system incorporating liver function assessment is suggested (LE: 2).

• To evaluate patients prior to receiving immunotherapy, Child-Pugh classification would be the most appropriate to date (LE: 1) to measure liver function.

Available agents and indications

For more than 10 years, sorafenib was the only systemic therapy approved by the FDA for the treatment of HCC. In recent years, ICI-based regimens have become standard of care in the first-line setting as well as for the treatment of disease that has progressed on prior sorafenib treatment. As of 2022, two ICI-based combinations, atezolizumab with bevacizumab and tremelimumab with durvalumab, had full FDA approval for first-line treatment of HCC, and pembrolizumab monotherapy as well as nivolumab in combination with ipilimumab had accelerated approvals as second line options. Results of the landmark trials leading to these approvals are described in table 4. Further details for each indication are discussed in chronological order of their FDA approvals.

Prior sorafenib therapy

In 2017, nivolumab received accelerated approval as monotherapy for the treatment of patients with HCC with progression following or intolerance to sorafenib. Approval was based on data from a cohort of patients from the CheckMate 040 Trial, a phase I/II, open-label, multicenter study. Among the 154 patients treated with nivolumab, 22 (14.3%; 95% CI 9.2% to 20.8%) had an objective radiologic response based on Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 criteria. Three patients (1.9%) had complete responses (CRs) and 19 (12.3%) had partial responses (PRs). DORs ranged from 3.2 months to 38.2+ months with 91% lasting 6 months or longer and 55% lasting 12 months or longer. The overall response rate (ORR), based on modified RECIST (mRECIST), was 18.2% (28 patients; 95% CI 12.4% to 25.2%) and the CR rate was 3.2% (5 patients) with a PR rate of 14.9% (23 patients). No differences in response rates were observed across PD-L1 expression levels.13 Postregistration studies support the safety of single-agent nivolumab in patients with Child-Pugh B disease59 where treatment is associated with shorter OS compared with Child-Pugh A disease (7.3 months vs 16.3 months; p<0.001).60 Data from cohort 5 of CheckMate 040, which included 25 sorafenib-naïve and 24 sorafenib-treated patients with Child-Pugh B7-B8 advanced HCC, also showed safety and efficacy for single-agent nivolumab in a setting of mild to moderate liver impairment61 (for further details on immunotherapy in special patient populations, see the Patient selection and management section). Continued accelerated approval for nivolumab monotherapy was contingent on the confirmatory trial CheckMate 459 (described below).

On March 10, 2020, nivolumab in combination with ipilimumab received accelerated approval by the FDA to treat patients with HCC who were previously treated with sorafenib. Approval was based on the results of an additional cohort (cohort 415) from CheckMate 040. In the CheckMate 040 cohort 4, 148 patients were randomized 1:1:1 to three different treatment arms to evaluate different dosing regimens of the combination: high-dose ipilimumab, low-dose ipilimumab and continuous nivolumab/ipilimumab for arms A, B, and C, respectively. For approval, efficacy was evaluated in the 49 patients who received nivolumab at 1 mg/kg and ipilimumab at 3 mg/kg every 3 weeks for 4 doses, followed by single-agent nivolumab every 2 weeks until disease progression or unacceptable toxicity. Data from all arms support anti-HCC activity of the regimen, however. The ORR reported for accelerated approval was 33% (n=16; 95% CI 20% to 48%), with 4 CRs and 12 PRs. DORs ranged from 4.6 months to 30.5+ months, with 31% of responses lasting 24 months or longer.62 An updated analysis at a minimum follow-up of 44 months found ORRs by blinded independent central review of 32%, 31%, and 31% for arms A, B, and C, respectively. Median DORs were 17.5 months, 22.2 months, and 16.6 months for arms A, B, and C, and median OS for each arm was 22.2 months, 12.5 months, and 12.7 months, respectively.63

In CheckMate 459, a phase III trial evaluating the efficacy of nivolumab as a first-line monotherapy, the ORR was 15% in the nivolumab group and 7% in the group receiving sorafenib. Median OS was 16.4 months for nivolumab-treated patients and 14.7 months for sorafenib-treated patients (HR 0.85; 95% CI 0.72 to 1.02; p=0.0752). The difference in OS between the two groups did not meet prespecified thresholds for statistical significance (HR 0.84; p=0.0419).64 Nevertheless, nivolumab demonstrated a favorable safety profile, better response rate, improved tolerability, and better QOL outcomes when compared with sorafenib. A trend towards better survival and response rate was noted in patients with PD-L1 tumor proportion score (TPS) ≥1% (about 19% of the randomized subjects) measured by the Dako PD-L1 immunohistochemistry (IHC) assay, supporting the importance of predictive biomarker development. In a 4:5 vote, the FDA Oncologic Drugs Advisory Committee (ODAC) recommended rescinding the indication for nivolumab for the treatment of patients with HCC and prior sorafenib therapy. There was unanimous agreement from committee members that voting was difficult due to the many factors, including the earlier vote to maintain the indication for pembrolizumab monotherapy. Those in favor of continued accelerated approval for nivolumab in this patient population highlighted the unmet need for second-line options. Rationale against continuing the indication centered on the lack of OS benefit in CheckMate 459 and the inadequacy of the proposed alternative studies to generate satisfactory evidence for efficacy in the second-line setting. Discussion also narrowed in on whether data exist to recommend nivolumab monotherapy over an ipilimumab plus nivolumab combination regimen, including debate over whether the group of patients deemed unfit for the dual checkpoint inhibitor combination represent a new indication that was not formally defined nor evaluated in trials to date. In July 2021, the nivolumab monotherapy indication for HCC was voluntarily withdrawn.

Accelerated approval was granted to pembrolizumab in 2018 for patients with HCC who have previously received sorafenib based on results from the phase II KEYNOTE-224 Trial. The study enrolled 104 patients to receive single-agent pembrolizumab with advanced HCC and radiographic progression or intolerance to sorafenib. The ORR was 17% (95% CI 11% to 26%) and among the 18 patients who responded, there was 1 CR and 17 PRs. At data cut-off, 12 of the 18 responses were ongoing and the median DOR was not reached (range 3.1–14.6+ months). Of the responders, 89% had a DOR ≥6 months, and 56% had a DOR ≥12 months.14

The phase III KEYNOTE-240 confirmatory trial evaluating pembrolizumab versus placebo was negative based on the co-primary end point of median OS and PFS. Median OS was 13.9 months (95% CI 11.6 to 16.0) for pembrolizumab versus 10.6 months (95% CI 8.3 to 13.5) for placebo (HR 0.781; 95% CI 0.611 to 0.998; p=0.0238), and median PFS for pembrolizumab was 3.0 months (95% CI 2.8 to 4.1) versus 2.8 months (95% CI 1.6 to 3.0; HR 0.718; 95% CI 0.570 to 0.904; p=0.0022), but this did not meet statistical significance by the prespecified statistical plan.65 The study did confirm the single-agent response rate of pembrolizumab in this setting with an ORR of 18.3 (95% CI 14.0 to 23.4) and a DOR of 13.8 months (range 1.5–23.6+ months). Despite the confirmatory trial not meeting prespecified end points, when the FDA ODAC reviewed the accelerated approval in April 2021 the vote to maintain the indication for pembrolizumab was unanimous, citing unmet medical need for patients who cannot receive first-line atezolizumab with bevacizumab (described below) and who have disease progression with or become intolerant to TKIs.

First-line therapy

The first ICI regimen to receive full approval and the first to receive first-line approval for the treatment of HCC is atezolizumab in combination with bevacizumab for patients who have not received prior systemic therapy, which was approved in 2020. Approval was based on the global, open-label, phase III IMbrave150 trial, in which 501 patients with unresectable HCC were randomly assigned in a 2:1 ratio to receive either first-line atezolizumab plus bevacizumab or sorafenib monotherapy until unacceptable toxic effects or loss of clinical benefit occurred. At the primary analysis, the HR for death with atezolizumab plus bevacizumab compared with sorafenib was 0.58 (95% CI 0.42 to 0.79; p<0.001). The 12-month OS rate was 67.2% (95% CI 61.3% to 73.1%) with atezolizumab plus bevacizumab and 54.6% (95% CI 45.2% to 64.0%) with sorafenib. Median PFS was 6.8 months (95% CI 5.7 to 8.3) versus 4.3 months (95% CI 4.0 to 5.6) with atezolizumab plus bevacizumab versus sorafenib, respectively (HR for disease progression or death 0.59; 95% CI 0.47 to 0.76; p<0.001).16 In an updated post hoc survival analyses, median OS was 19.2 months with atezolizumab plus bevacizumab compared with 13.4 months with sorafenib (HR 0.66; 95% CI 0.52 to 0.85; p=0.0009). The OS rates at 18 months were 52% vs 40% with atezolizumab plus bevacizumab versus sorafenib, respectively.66

The combination therapy also delayed deterioration in QOL compared with sorafenib monotherapy. In terms of the tolerability profile, grade 3–4 adverse events (AEs) occurred in 57% of patients treated with atezolizumab with bevacizumab.16 Additionally, the development of anti-drug antibodies (ADAs) is a possibility in patients treated with atezolizumab.67 In IMbrave150, among 318 ADA-evaluable patients with HCC, 30% (n=94) tested positive for treatment-emergent ADAs at one or more post-dose time points. In exploratory adjusted analyses, patients who were ADA-positive at landmark week 6 had a similar OS with atezolizumab plus bevacizumab versus sorafenib, whereas those with ADA-negative status had an improved OS compared with sorafenib. However, similar PFS and ORR benefit was seen with the combination over sorafenib regardless of ADA status.68

The second ICI regimen to receive full approval for the treatment of HCC is tremelimumab in combination with durvalumab for patients with unresectable HCC, which was approved in 2022. Approval was based on the open-label, phase III HIMALAYA trial, in which 1,171 patients were randomly assigned (1:1:1) to receive tremelimumab in combination with durvalumab (n = 393), durvalumab (n = 389), or sorafenib (n = 389).69 The primary endpoint for this trial was OS. In HIMALAYA, tremelimumab was administered as a one-time single dose plus durvalumab every 4 weeks (single tremelimumab regular interval durvalumab, or STRIDE regimen). Median OS for the tremelimumab in combination with durvalumab arm was 16.43 months (95% CI 14.16 to 19.58), compared to 16.56 months (95% CI 14.06 to 19.12) in the durvalumab arm and 13.77 months (95% CI 12.25 to 16.13) in the sorafenib arm. Superiority was demonstrated for tremelimumab plus durvalumab compared to sorafenib, with an OS HR of 0.78 (96.02% CI 0.65 to 0.93; p = 0.0035). Delayed separation of Kaplan-Meier survival curves was observed, and in an analysis of the piece-wise constant treatment effects of tremelimumab in combination with durvalumab versus sorafenib, the HRs before and after the 9-month time points were 0.87 (95% CI 0.68 to 1.11) and 0.70 (95% CI 0.56 to 0.89), respectively. Secondary endpoints included investigator-assessed ORR (table 4) and PFS, which was not significantly different among the three groups.

The study was not powered to assess for efficacy of the durvalumab plus tremelimumab combination compared to durvalumab alone. OS with durvalumab monotherapy was found to be noninferior to sorafenib (HR 0.86 [95.67% CI 0.73 to 1.03]; noninferiority margin 1.08). As of 2023, anti-PD-(L)1 monotherapy in the first-line setting is being used in some cases for patients with advanced HCC considered to be at high risk for complications with dual-ICI therapy. Although evidence of survival benefit for anti-PD-(L)1 monotherapy over sorafenib or lenvatinib in phase III studies is lacking, this monotherapy approach could be an alternative treatment option for these patients.

In HIMALAYA, the median time to deterioration of patient-reported QOL was 7.5 months for tremelimumab in combination with durvalumab, 7.4 months for durvalumab monotherapy, and 5.7 months for sorafenib. Grade 3−4 AEs occurred in 50.5% of patients in the tremelimumab with durvalumab arm, 37.1% in the durvalumab monotherapy arm, and 52.4% in the sorafenib arm. IrAEs requiring high-dose glucocorticoid treatment occurred in 20.1%, 9.5%, and 1.9% across the three treatment arms, respectively. The frequency of hepatic/hemorrhage AEs was similar across all treatment arms. Anti-durvalumab ADAs (positive post-baseline only) were detected in 3.1% and 2.5% of patients receiving tremelimumab plus durvalumab or durvalumab, respectively, and anti-tremelimumab ADAs were detected in 11% of patients in the dual therapy arm.

Panel recommendations

• For first-line treatment of patients with advanced Child-Pugh A HCC, atezolizumab plus bevacizumab (LE:2) or tremelimumab plus durvalumab (LE:2) are recommended, unless any medications are contraindicated.

• General contraindications to bevacizumab include high risk of cardiac disease, stroke, hemorrhage, hemoptysis, gastrointestinal perforation, or non-healing wounds (LE: 1). (For contraindications to immunotherapy, see the Patient selection and management section). Consideration should be given to timing of prior events. Additional contraindications specifically relevant to HCC include untreated or incompletely treated gastroesophageal varices at risk for bleeding (LE: 2).

• For patients with contraindications to an ICI-containing combination therapy (ie, atezolizumab plus bevacizumab or tremelimumab plus durvalumab), lenvatinib or sorafenib should be considered as standard first-line therapy (LE:2).

• Nivolumab monotherapy has demonstrated activity in Child-Pugh B7-B8 HCC for both first-line treatment of sorafenib-naïve patients and for second-line treatment of patients who were intolerant to or progressed on sorafenib (LE: 3).

• For patients with good performance status who have progressed on first-line therapy and have not received prior immunotherapy, other non FDA-approved or conditionally approved anti-PD-1 checkpoint inhibitors may be considered as immunotherapeutic options (LE: 3).

Checkpoint inhibitors and novel combinations

Tremelimumab monotherapy has been evaluated in a pilot trial of patients with HCC with chronic HCV infection. Among the 17 patients who were assessable for tumor response, the PR rate was 17.6% and disease control rate (DCR) was 76.4% with a median time to progression of 6.48 months (95% CI 3.95 to 9.14). Significant drops in viral load were observed in the 20 patients who were evaluable for toxicity and viral responses, and no patients needed steroids because of severe irAEs.70 Single-agent tislelizumab (anti-PD-1),71 camrelizumab (anti-PD-1),72 and durvalumab (anti-PD-L1)73 are all also being studied in phase III trials.

In the CheckMate 040 trial cohort 6,74 the efficacy and safety profile of the triplet combination of cabozantinib, nivolumab, and ipilimumab were analyzed and compared with the cabozantinib plus nivolumab doublet. A total of 71 sorafenib-naïve or sorafenib-experienced patients with advanced HCC were randomized to either receive nivolumab 240 mg every 2 weeks with cabozantinib 40 mg daily (n=36) or nivolumab 3 mg/kg every 2 weeks with ipilimumab 1 mg/kg every 6 weeks and cabozantinib 40 mg daily (n=35). Although the study was not powered to directly compare efficacy of the triplet versus doublet regimens, numerically higher response rates (29% vs 19%), better PFS (median 6.8 vs 5.4 months) and improved median OS (not reached vs 21.5 months; 15-month OS rates: 70% vs 64%) were observed with the three-drug combination. Nevertheless, a higher rate of treatment-emergent AEs was also observed in the triplet arm, without the emergence of new safety signals in either treatment arms.

HCC is one of the most vascularized solid tumors and anti-angiogenic agents may complement immunotherapies. Multiple anti-angiogenic multikinase inhibitors are being evaluated in combination with checkpoint inhibitors for HCC. The combination with the most available data at the time of manuscript preparation is pembrolizumab plus lenvatinib. In Study 116, an ongoing phase Ib multicenter open-label study of lenvatinib plus pembrolizumab in 104 patients with unresectable HCC, the confirmed ORRs at data cut-off were 46.0% (95% CI 36.0% to 56.3%) by mRECIST and 36.0% (95% CI 26.6% to 46.2%) by RECIST v1.1 with median DORs of 8.6 months (95% CI 6.9 to not estimable [NE]) and 12.6 months (95% CI 6.9 to NE), respectively. Median OS was 22 months and treatment-related AEs of grade ≥3 occurred in 67% of patients.75 The ongoing phase III LEAP‐002 trial is also studying the combination and enrolling patients for treatment with pembrolizumab plus lenvatinib for first-line treatment of advanced HCC.76 Other ICI/TKI combination studies include avelumab with axitinib, which led to tumor shrinkage in 15 (68.2%) and 16 (72.7%) patients and an ORR of 13.6% (95% CI 2.9% to 34.9%) and 31.8% (95% CI 13.9% to 54.9%) by RECIST and mRECIST, respectively in one study.77 Cabozantinib is being combined with atezolizumab for patients who have not received prior systemic therapy for HCC in the phase III study COSMIC‐312.78

Integration with local and regional therapies

Locoregional therapies such as TACE and drug-eluting bead TACE (DEB-TACE) may induce immunogenic cell death, thus promoting CD8⁺ T cell infiltration into the tumor microenvironment, potentially synergizing with anti-PD-(L)1 therapy.79 Doxorubicin, which has been shown to cause immunogenic cell death,80 is the most commonly administered drug during TACE and DEB-TACE, and patients undergoing chemoembolization have been shown to develop AFP-specific CD4⁺ T cell responses81 as well as GPC3-specific cytotoxic T cell responses.82

A few studies have reported tolerable safety and initial efficacy outcomes with the combination of ICIs and locoregional therapies such as TACE and radiofrequency ablation (RFA). One trial enrolled 32 patients with HCC for tremelimumab therapy at two dose levels (3.5 mg/kg and 10 mg/kg intravenous [IV]) every 4 weeks for 6 doses, followed by infusions every 3 months until off-treatment criteria were met. On day 36, patients underwent subtotal RFA or chemoablation. Of the 19 evaluable patients, 5 (26.3%; 95% CI 9.1% to 51.2%) achieved PR. The median time to tumor progression was 7.4 months (95% CI 4.7 to 19.4) and median OS was 12.3 months (95% CI 9.3 to 15.4).83

Integration with transplant

Checkpoint inhibitors are considered contraindicated in patients undergoing transplantation due to fears of graft rejection.84 Reports have emerged of immunotherapy being used as salvage therapy in liver transplant recipients with malignancies other than HCC, but rejection was frequent. A review of 14 cases of liver transplant recipients who were treated with ICIs identified four cases of liver graft rejection and three cases with lethal outcomes.85 Another retrospective study including 39 patients with solid organ transplants reported permanent discontinuation of ICIs in 31% because of allograft rejection. Graft loss occurred in 81%, leading to death in 46%.86

Vaccines

Some vaccines have demonstrated manageable safety and preliminary efficacy in early phase trials in HCC. Although no antitumor effects or immune responses were detected among 40 patients with advanced HCC who were treated with low-dose cyclophosphamide in combination with a telomerase peptide vaccine (GV1001),87 other strategies have posted more promising results.

Several groups have attempted to develop peptide vaccines based on GPC3. One GPC3 peptide vaccine was well tolerated in a phase I trial that included 33 patients with advanced HCC. Vaccination induced a GPC3-specific cytotoxic T lymphocyte response in 90% of patients—there was 1 PR and 19 cases of stable disease at 2 months.88 That same vaccine was shown to lead to numerically lower rates of recurrence compared with surgery alone at 1 year (28.6% vs 54.3%) and 2 years (39.4% vs 54.5%) in the adjuvant setting in a phase II trial of 35 patients with HCC who had undergone resection.89

AFP-based vaccines have been shown to elicit T cell responses in early trials. Four immunodominant, human leukocyte antigen (HLA)-A*0201-restricted epitopes of AFP that are recognized by the human T cell repertoire have been identified.90 In a pilot phase I clinical trial that enrolled six HLA-A*0201 patients with AFP-positive HCC for intradermal vaccinations with the four peptides emulsified in incomplete Freund’s adjuvant, T cell responses were observed against most or all of the epitopes.91 Subsequently, a phase I/II trial that included 10 HLA-A*0201 patients with AFP-positive HCC who were immunized with intradermal vaccinations of the four AFP peptides pulsed onto autologous dendritic cells found statistically significant levels of AFP-specific T cells to at least one peptide by major histocompatibility complex (MHC) tetramer in 60% of participants.92

Tumor lysate-based vaccines have also been evaluated in HCC. One study found that autologous tumor vaccination significantly delayed time to recurrence in 60 patients with HCC who had undergone curative resection. The 1-year, 2-year and 3-year recurrence rates in the 30 patients in the vaccine group were 16.7%, 29.2%, and 33.3%, respectively, compared with 30.8%, 53.8%, and 61.5%, respectively, in the control group.93 Another phase II trial of autologous dendritic cells pulsed with tumor lysate observed a radiologically determined DCR of 28% in 35 patients with advanced HCC.94 Hepcortespenlisimut-L, a tableted oral formulation derived from a heat-inactivated pooled blood of patients with HCC and viral hepatitis, has entered phase III trials in patients with HCC and demonstrated clear improvements in ALT, AST, alkaline phosphatase, and bilirubin levels compared with placebo.95

The dramatically high efficacy rates seen with RNA-based vaccines during the COVID-19 pandemic has re-invigorated the study of RNA-vaccinology—a concept with roots in the immunotherapy discipline.96 97 RNA has been used as both a vaccine platform and an adjuvant to boost immunogenicity for HCC-specific epitopes, such as HLA-A*02-restricted tumor-associated peptides.98 99

Adoptive cell therapies

To date, the most advanced clinical studies for cellular therapies in HCC are with cytokine-induced killer cells (CIKs), which are characterized by coexpression of CD3 and CD56 and can be generated by expanding human peripheral blood mononuclear cells in the presence of interferon-gamma (IFNγ).100 101 One randomized phase III trial of CIKs as adjuvant therapy for patients with HCC undergoing resection demonstrated a median recurrence-free survival (RFS) of 44 months in the cell therapy group and 30 months in the control group (HR 0.63; 95% CI 0.43 to 0.94; p=0.010 by one-sided log-rank test).102 A meta-analysis of 13 phase II and phase III trials involving CIKs for HCC that included a total of 1,212 patients found that cellular therapy was associated with a significantly improved 1-year survival (OR 0.25; 95% CI 0.12 to 0.52; p<0.001) and 2-year survival (OR 0.17; 95% CI 0.07 to 0.43; p<0.001), as well as a favorable DCR (OR 0.09; 95% CI 0.04 to 0.25; p<0.001) and ORR (OR 0.21; 95% CI 0.13 to 0.35; p<0.001).103

Allogenic natural killer (NK) cell-based adoptive therapies have also been evaluated in HCC. One study that included 40 patients with stage IV HCC found that NK cell therapy synergized with irreversible electroporation (IRE), leading to decreased AFP expression and higher median OS compared with IRE alone (10.1 months vs 8.9 months; p=0.0078).104 Allogenic NK cell therapy also showed synergy with cryoablation in a study that included 61 patients with advanced HCC. After a median follow-up of 8.7 months (range 3.9–15.1 months), median PFS and DCR were higher among the 35 patients who received cryoablation plus NK cells compared with the 26 patients treated with cryoablation alone (PFS 9.1 months vs 7.6 months; p=0.0107; DCR 85.7% vs 69.2%; p<0.01).105

Panel recommendations

• Clinicians should encourage patients' participation in clinical trials.

• Future biomarker development might help to select a subgroup of patients benefitting from single-agent nivolumab treatment. Designing a biomarker strategy based on pretreatment and on-treatment tissue and blood samples to assess immune cell changes and other correlatives is critical to elucidate mechanisms of response or resistance to immunotherapy in combination with local therapy in early-stage HCC.

• Studies evaluating combinations of other immunotherapies with ICIs should be based on solid scientific rationale.

• Future randomized studies to compare local therapy alone to local therapy combined with immunotherapy are essential to assess the expected synergy and favorable treatment outcome of the combination strategy.

Patient selection

In selecting the appropriate patient for consideration of treatment with a standard immunotherapy-based approach (as opposed to within the context of a clinical trial), there are both general oncologic considerations as well as HCC-specific or liver-specific considerations. It is critical to account for the singular nature of HCC, as it generally arises in a damaged and potentially dysfunctional liver. As many as 43% of patients with HCC will die as a direct result of cirrhosis as opposed to cancer progression.106 Therefore, clinical trials needed for patients with more advanced liver function decompensation than Child-Pugh B7 are encouraged, especially when the main factor behind liver function deterioration is HCC progression rather than the underlying liver disease. Additional considerations include the patient’s performance status and history of comorbidities, in particular the presence of any known autoimmune disorders. A patient’s eligibility for treatment with anti-VEGF therapy—either with TKIs (eg, sorafenib) or monoclonal antibodies (eg, bevacizumab)—will also inform a treatment plan. Liver-specific factors that need to be considered include the stage of the HCC and the indication for treatment, the underlying synthetic liver function, and disease etiology and its bearing—if any—on outcome. There are also certain specific situations such as recurrence in the setting of liver transplant that need further study, as well as the role of biomarkers in predicting efficacy or toxicity. For many of these considerations, the data are varied in terms of the weight of evidence, which should be taken into account in regard to the degree to which they should influence the physician’s decision.

Biomarkers for ICI efficacy and safety in HCC

ICIs provide benefit for only a subset of patients. The ability to identify intrinsic resistance to ICIs would allow patients to attempt other therapies, which could, most importantly, lead to better outcomes, while also saving healthcare resources. Unfortunately, validated blood or tissue biomarkers for ICI resistance are currently lacking in the clinical setting. Early studies have also returned conflicting results. High serum AFP levels are associated with increased sensitivity to the anti-VEGF receptor (VEGFR) monoclonal antibody ramucirumab.134 Post hoc subgroup analysis of randomized trials have shown that the HR for OS was slightly lower among patients with high AFP in KEYNOTE-240 (pembrolizumab vs placebo)65 and CheckMate 459 (nivolumab vs sorafenib),135 while the contrary was observed in IMbrave150 (atezolizumab with bevacizumab vs placebo).16 70 Furthermore, objective remissions occur irrespective of AFP levels after nivolumab or pembrolizumab monotherapies, or the combination of ipilimumab and nivolumab.

A number of features of the tumor microenvironment have been associated with HCC prognosis, including overall lymphocyte infiltration, density of Tregs, and tumor-associated macrophages (TAMs), especially if M2-polarized. In melanoma, the presence of conventional type 1 dendritic cells seems critical to promote a T and NK cell infiltrate and for the action of ICIs.136 In HCC animal models, β-catenin-mutations in HCC (which are present in around 25% of human HCCs) result in a paucity of intratumoral conventional type 1 dendritic cells,137 and it has been proposed that β-catenin defects may be used to identify patients with disease that will fail to respond to PD-1 blockade.138 This feature awaits investigation in clinical trials. Soluble factors also modulate the immune response against HCC. For example, transforming growth factor beta (TGF-β) downregulates antitumor responses through a variety of different mechanisms, and high levels of the cytokine shape the response to pembrolizumab.139

Pretreatment tumor infiltration by T cells and their activity status are key to determine response to ICIs in various cancers. In advanced HCC, CD4⁺ and CD8⁺ T cell infiltration showed weak correlations with survival after second-line treatment with PD-1 inhibitors in CheckMate 040.140 In the trial, deep antitumor responses were observed regardless of PD-L1 expression after nivolumab treatment, although the response rate was higher among patients with at least 1% of tumor cells expressing PD-L1.140

On the other hand, PD-L1 expression in tumor or stromal immune cells was higher among responders to pembrolizumab, but remissions also occurred in the absence of expression in both cell types.65 In CheckMate 459, median OS after nivolumab and sorafenib was 16.1 months versus 8.6 months among patients that had tumor PD-L1 expression ≥1% (HR 0.80), and 16.7 months versus 15.2 months among those that had tumor PD-L1 expression <1% (HR 0.84).135 Interestingly, macrophage infiltration, including M2-polarized TAMs, was not associated with clinical outcomes after nivolumab treatment. A meta-analysis including 894 patients across nine trials of ICIs in advanced HCC found a positive association between PD-L1 expression and response to therapy—especially for single-agent anti-PD-1. Strikingly, in the analysis, PD-L1 expression status had minimal association with response to therapy for patients being treated with anti-CTLA-4-containing combinations.141 Analytical heterogeneity in PD-L1 expression is substantial, however, and may contribute to the performance of this test as a predictive biomarker.142

Several inflammatory gene signatures are correlated with higher response rate and improved OS after nivolumab treatment.140 Interestingly, the most complex transcriptomic classifications of inflammatory HCC including a large number of genes143 were not identified as predictive of response in this analysis, suggesting that short gene signatures may be more relevant for clinical development. Regarding ICI combinations, objective remissions occurred with ipilimumab plus nivolumab irrespective of PD-L1 expression in tumor cells.144 An early burst of Ki67⁺CD8⁺ cells in the peripheral blood was also seen in one of the randomized expansion cohorts for Study 22, which evaluated combinations of durvalumab and tremilimumab at different dosing regimens,145 hinting that cytotoxic T cell proliferation after therapy may predict response. Altogether, though, it seems unlikely that a single biomarker could be used to inform clinical decisions in a timely fashion. However, it is probable that composed and integrative multifactorial indexes might help identify patient subsets who are likely to benefit, further underscoring the importance of obtaining pretreatment tumor biopsies for future translational studies.

Pembrolizumab is FDA-approved for two tissue-agnostic indications based on tumor-intrinsic characteristics. Approval for pembrolizumab for the treatment of microsatellite-high (MSI-H) or mismatch repair deficient (dMMR) tumors was based on a pooled ORR of 39.6% (95% CI 31.7% to 47.9%), with a 7% CR rate among 149 patients with 15 different tumor types in five single-arm multi-cohort multicenter trials: KEYNOTE-016,146 KEYNOTE-164,147 KEYNOTE-012148 KEYNOTE-028,149 and KEYNOTE-158.150 Approval for pembrolizumab for non-MSI-H/dMMR tumors with high mutation burden (TMB-H)—defined as ≥10 mutations per megabase (mut/Mb) as assayed by the FoundationOne CDx companion diagnostic—was based on KEYNOTE-158.151 No patients with HCC were included in the cohorts upon which the tissue-agnostic indications for pembrolizumab were approved, however.

TMB correlates with the number of neoantigens and response to ICIs in tumors with >20 somatic mut/Mb, such as melanoma.152 153 However, HCC is infrequently MSI-H/dMMR or TMB-H. One study that performed comprehensive genomic profiling of 755 patients with advanced HCC found a median TMB of 4 mut/Mb and that only six tumors (0.8%) were TMB-H. Furthermore, out of 542 cases assessed, only one (0.2%) was MSI-H.154 Another analysis found a rate for MSI-H as low as 6%.155

Markers of systemic inflammation like neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR) have shown a strong prognostic impact in HCC across tumor stages. Lower NLR has been associated with better outcomes after sorafenib,156 157 and similar trends are emerging from trials of ICIs. In CheckMate 040, patients progressing on nivolumab had a higher NLR and PLR than patients who had disease control as the best overall response.140 Consistent with this observation, a retrospective analysis of 103 patients who received nivolumab found that patients with Child-Pugh A disease who achieved PR or CR had significantly lower post-treatment NLR and PLR (p<0.001 for both) compared with patients who had stable or progressive disease.158

The composition of the gut microbiota, which has been linked to the promotion of HCC development and progression through secreted metabolites,159 160 may also predict response to treatment, although current analyses in the liver cancer setting are small and preliminary. Gut microbial diversity has been linked to ICI efficacy in epithelial tumors,161 and retrospective analysis has shown that antibiotic use is associated with worse outcomes with immunotherapy in lung and renal cancer,162 a finding that has also been replicated in a prospective trial including several additional tumor types.163 One pilot study of eight patients with HCC treated with anti-PD-1 therapy after progression on sorafenib found that patients with responsive disease displayed higher taxa richness and more gene counts in their microbiota compared with non-responders, with enrichment for 20 distinct species of bacteria, including Akkermansia muciniphila and the Ruminococcaceae family.164 The potential for the gut microbiota to shape responses to immunotherapy is an ongoing area of research, but, at present, the state of the data is not sufficient to alter management in this regard and clinical judgment outweighs other considerations.

An additional ongoing area of research is the identification of biomarkers for the prediction of which patients will experience irAEs with ICI therapy. Several studies have reported a link between various clinical and blood-based or serological factors and the onset of immune-related toxicity, although none have been prospectively validated for HCC. Patients with sarcopenia165 166 and of female sex167 168 have both been shown to have higher incidences of irAEs. Additionally, the composition of the gut microbiota may play a role in predicting which patients will develop ICI-associated colitis.169 Additional factors under active investigation for prediction of toxicity include elevated cytokine levels at baseline, such as interleukin (IL)-6168 and IL-17,170 as well as the presence of autoantibodies.171 172 Currently there are no clinically validated biomarkers to predict the risk of irAEs.

Recognition and management of irAEs

The same mechanisms by which immunotherapy drugs exert their therapeutic effects also underlie their unique toxicities—suppression of the inhibitory mechanisms that protect tissues from uncontrolled immune responses. Unlike AEs with chemotherapy or other treatment modalities, irAEs may be delayed in onset and have prolonged duration, sometimes months or years after initial exposure to therapy. The overall incidence and severity of irAEs reported in phase III trials of anti-PD-(L)1 agents varies depending on disease state and comorbidities. Most irAEs are of mild-to-moderate severity, but life-threatening events have been reported. A meta-analysis of fatal ICI-associated toxicities encompassing more than 16,000,000 adverse drug reactions from the medical records from the VigiBase-VigiLyze database found a total of 613 deaths related to ICIs. The fatalities related to anti-CTLA-4 therapy were most often from colitis (n=135, 70%), while fatalities associated with anti-PD-(L)1 were most often from pneumonitis (n=333, 35%), hepatitis (n=115, 22%), and neurotoxic effects (n=50, 15%).173 A systematic review including 48 clinical trials involving 7,936 patients treated with nivolumab monotherapy or combination nivolumab and ipilimumab found that the double regimen was associated with more all-grade and grade ≥3 irAEs categorized by system, organ, or class (p<0.05). Additionally, the ORR of nivolumab combined with ipilimumab was positively correlated with the incidence rate of skin (r=0.54; p=0.04) and gastrointestinal irAEs (r=0.60; p=0.02), but not endocrine, hepatic, pulmonary, or renal irAEs.174 Similarly a recent observational study including 331 patients with HCC receiving anti-PD-(L)1 monotherapy or combinations found that the emergence of treatment-related AEs of grade ≥2 while on ICI therapy predicted for improved OS (median 19.7 vs 11.0 months; HR 0.32; 95% CI 0.16 to 0.65; p=0.001) and increased ORR (30% vs 16%; χ² 5.9; p=0.01).175

Typically, the management of irAEs includes interruption of ICIs, corticosteroids, and occasionally the administration of immunomodulatory agents including tumor necrosis factor (TNF) inhibitors. Detailed recommendations on the recognition and management of ICI-associated AEs have been published elsewhere176 and the general principles contained therein may guide treatment decisions for irAEs, which are not specific to patients with HCC.

IrAEs specific to the treatment of HCC

Outside of immune-mediated hepatotoxicity, the commonly reported AEs in published trials leading to ICI approvals for HCC have been generally comparable to those seen in other disease settings. Pembrolizumab monotherapy showed a tolerable safety profile in KEYNOTE-224, with the most common irAEs of any grade being hypothyroidism (n=8, 8%) and adrenal insufficiency (n=3, 3%).14 In the cohort of patients receiving nivolumab monotherapy in CheckMate 040, the most common AEs were pruritus (n=9, 11%) and rash (n=11, 23%).13 The addition of ipilimumab to nivolumab, as evaluated in cohort 4 of CheckMate 040, was associated with a wider variety of toxicities with the most common AEs of any grade being rash (n=14, 29%), pruritus (n=22, 45%), diarrhea (n=12, 24%), decreased appetite (n=6, 12%), fatigue (n=9, 18%), adrenal insufficiency (n=7, 14%), and hypothyroidism (n=10, 20%).177 For the combination of atezolizumab with bevacizumab in IMbrave150, the most common adverse reactions were hypertension (n=98, 29.8%), fatigue (n=67, 20.4%), and proteinuria (n=66, 20.1%), and no serious AEs with a difference in incidence of >2% were noted between the atezolizumab with bevacizumab and sorafenib treatment groups. For the combination of tremelimumab plus durvalumab, the most common any grade treatment-related AEs were diarrhea (26.5%), pruritis (22.9%), and rash (22.4%).69

Drug-induced hepatotoxicity

HCC usually develops in a background of chronic liver disease, which itself may give rise to systemic manifestations. Cirrhosis is characterized by diffuse fibrosis of the liver, altered hepatic blood flow and portal hypertension, and progressive failure of liver functions. In parallel, other organs frequently develop secondary dysfunction. Many extrahepatic disorders associated with cirrhosis cause symptoms that may mimic irAEs and therefore lead to overdiagnosis or underdiagnosis of toxicities with immunotherapy. Late recognition of irAEs may delay treatment and worsen the prognosis. Overdiagnosis may result in inappropriate interruption of ICIs, complications caused by immunosuppressive therapy, unnecessary diagnostic procedures, and increased cost. Such disorders can also synergize in causing deteriorating organ function when irAEs occur. The most important cirrhosis-related disorders that may compromise the management of irAEs are summarized in table 5.

Patients with HCC and underlying liver disease are at high risk for decompensation with additional insult to the organ. Some studies have found that underlying liver disease as opposed to cancer progression is the ultimate cause of death in almost half of patients with HCC.106 Elevated liver enzymes without clinical impairment in hepatic function were commonly reported in all of the trials that led to approvals of ICIs for HCC.13–15 Grade 3 or 4 elevations in liver enzymes were reported in 16% of patients in the dose‐escalation arm of CheckMate 040 and in 12% of the patients in KEYNOTE‐224.14 In KEYNOTE-240, immune-mediated hepatitis events were seen in 10 patients (3.6%) in the pembrolizumab group, approximately 90% of which resolved.65

No prospective trials have defined the best treatment approach for drug-induced hepatotoxicity in patients with HCC receiving immunotherapy. The package inserts for pembrolizumab, nivolumab, and atezolizumab all recommend monitoring for changes in liver function and administering corticosteroids for hepatitis followed by a taper. ICIs should also be withheld or discontinued if liver enzymes or bilirubin become elevated, with the thresholds varying depending on baseline values and the drug regimen being given.178 Exclusion of other causes of acute liver damage—including toxicities from concomitant medications, use of herbal supplements, viral hepatitis, and particularly tumor progression—is key to adequate management.

Response evaluation

Measurement of response rate in HCC has been controversial. The WHO (WHO) criteria179 and the RECIST guidelines180 define standard measurement methods for converting radiology image observations into quantitative and statistically tractable frameworks for measuring changes in tumor size associated with therapy. However, assessments based solely on tumor size are misleading when applied to molecular targeted therapies and immunotherapies. For HCC in particular, poor correlation has been shown between the clinical benefits provided by sorafenib or locoregional interventional therapies and RECIST-based responses.181 Subsequently, the concept of ‘viable tumor’ was endorsed by the guidelines for the design of HCC clinical trials developed by AASLD182 and eventually incorporated into a formal proposal to amend standard RECIST criteria to address the unique complexity of HCC response assessment. The amended criteria were named mRECIST for HCC.183 In published trials of immunotherapy for HCC, RECIST v1.1 was used. In the immunotherapy setting, no significant differences exist between RECIST and mRECIST.

Several clinical investigations have shown that objective response measured by mRECIST predicts survival in patients treated by locoregional therapies. A meta-analysis including seven trials and 1,357 patients reported an OS HR (responders vs non-responders) of 0.39 (95% CI 0.26 to 0.61; p<0.0001).184 Another study found that EASL and mRECIST both outperformed the WHO criteria and RECIST for patients undergoing DEB-TACE.185 Recently, data from randomized trials confirmed that objective response by mRECIST predicts survival in patients with advanced-stage HCC receiving systemic therapies with TKIs, and suggested that objective response by mRECIST can be considered as a candidate surrogate end point of OS, although further research is needed to support this finding.186 187

Although late response after apparent disease progression on imaging has been reported in the context of immunotherapy for HCC,188 the overall incidence of pseudoprogression with ICI treatment is rare. Estimated rates of psuedoprogression across published studies range from 2%–10%.189 190 Also rare, though possible, is a rapid acceleration in tumor growth after anti-PD-(L)1 therapy, a phenomenon called hyperprogression.190 191 Although published evidence is limited, hyperprogression has been reported in small case series of patients with HCC treated with ICIs,192 and retrospective analyses.193 Importantly, the evidence to date has only reported hyperprogression in the setting of anti-PD-(L)1 monotherapy—it is unclear whether the addition of VEGF-directed antibodies to ICI therapy affects the likelihood of hyperprogressive disease after treatment.

Panel recommendations

• For patients with advanced-stage HCC and for patients with earlier-stage disease where liver-directed therapies are not considered appropriate or who have progressed after liver-directed therapy, the data at present supports first-line and subsequent-line ICI therapy use (LE: 2). Further studies are needed to confirm the efficacy of immunotherapy in the curative setting (neoadjuvant/adjuvant/perioperative) or in conjunction with intra-arterial therapies.

• In patients with HCC with cirrhosis, the data supports the use of immunotherapy in patients with underlying synthetic liver function consistent with well-compensated cirrhosis, specifically Child-Pugh A (LE: 2). The panel recognizes, however, that some carefully selected patients with Child-Pugh B may derive benefit (LE: 3).

• Patients who have contraindications for the use of TKIs or anti-VEGF therapies (eg, cardiovascular comorbidities) may be suitable for anti-PD-1 monotherapy (LE: 1).

• The panel recommends against the use of immunotherapy in the post-transplant setting (LE: 4) due to the high risk of graft failure, given known mechanisms of ICIs.

• Additional studies are needed to assess the potential risks and benefits of immunotherapy in the pretransplant setting.

• The panel agrees that patients can be considered for immunotherapy treatment irrespective of hepatitis viral etiology (LE: 3), though it is strongly recommended that patients with HBV be on concomitant antiviral medication and adherent.

• While patients living with HIV have not been included in clinical trials to date, the panel believes that this is not an absolute contraindication to treatment with immunotherapy as long as the appropriate HIV therapy is instituted as per expert guidance (LE: 2), while further dedicated studies to assess such therapies in patients living with HIV remain critical.

• Historical disparities in access to clinical trial participation for underrepresented groups should be considered, with efforts made to support diversity, equity, and inclusion.

• The panel recommends against the use of routine testing of biomarkers for predicting immunotherapy efficacy, which, at this point, remains exploratory.

• The panel recommends against the use of routine testing of biomarkers for predicting irAEs, which, at this point, remains exploratory.

• Response assessment can be performed according to mRECIST criteria in patients receiving locoregional interventional therapies (LE: 3).

• Limited data are available concerning the value of mRECIST and immune-related RECIST (irRECIST) criteria in the setting of HCC response assessment, especially in the context of ICI therapy. Further studies are needed to compare outcomes between patients with response to treatment by mRECIST versus irRECIST.

• Pseudoprogression, while a real phenomenon, occurs rarely (LE: 4). A comprehensive assessment is encouraged. In published trials, treatment beyond progression has been allowed.

• Hyperprogression may occur (LE: 4). It is uncommon, cannot be anticipated, and remains poorly understood.

• Caution should be exercised in translating response assessment models developed for clinical trials into clinical practice.

• For management of irAEs in patients with HCC, refer to general principles in published guidelines.

Patient and caregiver education

Prior to diagnosis, the majority of patients and their caregivers will likely be unfamiliar with HCC, and they may harbor misconceptions about the etiology of the disease, potentially leading to stigma and shame over and above the emotional distress associated with a cancer diagnosis.197 Among different types of cancer, HCC has been found to rank third highest in terms of levels of emotional distress experienced by patients.198 Rehabilitation, palliative care, and psycho-oncology have been insufficiently studied in liver cancer.

Perceived stigma surrounding liver disease may cause patients to delay care or avoid seeking social support, which negatively impacts QOL.199 The majority of HCC cases worldwide are secondary to HBV or HCV infection,4 with NASH increasingly becoming the primary cause in the US and Europe.5 However, a survey of HCC caregivers in the US found that 72% were under the mistaken impression that heavy alcohol use was the most common risk factor for liver cancer.200 Stigma surrounding HBV may be more pronounced in certain populations, such as people of Asian descent,201 202 so it is important for healthcare providers to be sensitive and culturally informed in their communications with patients.

HCC is a disease within a disease, and patients as well as their caregivers need to understand that their treatment journey will involve both the cancer itself as well as underlying liver damage. Patients with HCC often receive care from a multidisciplinary team that may include oncologists, hepatologists, surgeons, gastroenterologists, endocrinologists, and other specialists. In addition to the care team responsible for administering therapy targeting the tumor and the liver, patients will need ‘whole-person’ support for psychosocial and spiritual concerns, especially during end-of-life care.203 204 Depending on the stage of their disease (for more details on staging systems for HCC, see the Diagnostics and staging for patients with HCC section), a patient may be receiving information from a large number of different providers, especially in cases of intermediate-stage HCC.204 Additionally, practitioners from other specialties may have limited knowledge about the unique mechanisms of action of immunotherapies, and the accompanying potential for toxicities, making ongoing communication between a patient and their treating oncologist paramount.

Currently, immunotherapy is only approved for patients with advanced disease. Patients may be unfamiliar with the stages of liver cancer and the difference between treatments with curative intent and palliative therapy. Further complicating matters, patients may have preconceived notions shaped by media portrayals of high-profile immunotherapy success stories, while being less knowledgeable about the realistic efficacy and potential toxicities with treatment.205 Early referral to palliative care has been shown to improve QOL in patients with non-small cell lung cancer,206 yet palliative care is underutilized in patients with end-stage liver disease.207 208 Patients with cirrhosis who are ineligible for transplant are also underserved with appropriate palliative care.209 It is important for patients to understand that immunotherapy for HCC, even if it may extend OS, is a palliative treatment used in advanced stages of the disease and not curative in intent, so that they may be referred to advanced care planning early on in their treatment.

Considerations for administration, dosing and monitoring

The tolerability of immunotherapy is, for the most part, better than conventional cancer treatments, although future combination strategies (eg, ICIs with TKIs) may be associated with less favorable toxicity profiles.210 The administration, dosing, and monitoring considerations for immunotherapy may be distinct from what a patient or caregiver is expecting based on experience with prior therapies or conversations with other healthcare providers who do not specialize in immunotherapy. Therefore, it is important to discuss the potential for irAEs and the signs and symptoms of expected toxicities with patients and caregivers. Additionally, patients must understand how liver comorbidities may affect the efficacy of immunotherapy for HCC (see the Patient selection and management section for considerations for healthcare providers). It is important for patients and caregivers to have clear and detailed instructions for when to contact their healthcare providers due to symptoms of irAEs, and examples of call parameters are provided in box 1.

The ICIs that are currently approved for HCC are typically given as IV infusions, whereas TKIs such as sorafenib are oral medications. Because immunotherapy is usually administered at an infusion center, access to care may be a challenge for some patients, especially those in rural areas.211 212 However, a potential benefit of the requirement for inperson infusions is the opportunity for contact with a treating physician if AEs do occur. Additionally, although patients receiving palliative chemotherapy have been found to prefer oral administration over IV, the majority are not willing to accept a decreased response rate or shorter DOR,213 which is likely also true when deciding between IV immunotherapy versus other oral medications.

Multiple liver-specific assessment instruments have been developed to monitor QOL in patients with HCC, including the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core-18 (EORTC QLQ-HCC18),214 the Functional Assessment of Cancer Therapy-Hepatobiliary (FACT-Hep),215 the FACT Hepatobiliary Symptom Index (FHSI),216 and the QOL-liver cancer (QOL-LC).217 However, multiple systematic reviews have found that the most used assessment tool is the EORTC Quality of Life Questionnaire Core-30 (EORTC QLQ-C30).218 219

Special considerations for patients with HCC

Patients with HCC have been found to have lower health-related QOL (HRQOL) than the general population, especially for measures of physical, psychological, and functional well-being, as well as hepatobiliary symptoms.218 Both physical and psychological factors may influence a patient’s QOL, and a person’s self-perception and coping mechanisms may modulate their status. In patient interviews, HCC has been found to be perceived as a long-term and chronic disease that cannot be cured but might be controlled, and coping strategies can include focusing as much as possible on managing HCC and its symptoms, emotional responses, and leading a normal life.220 Those mental constructs can affect feelings about physical symptoms, and it has been demonstrated that patients with negative illness perceptions who use more emotion-oriented coping had worse HRQOL.221 However, rigorous studies on interventions targeting disease perception or coping mechanisms are currently lacking.

Pain, particularly upper quadrant abdominal pain, is common in patients with HCC.222 Pain management may be difficult because approximately 80% of patients with HCC have cirrhosis,223 and liver damage can alter drug pharmacokinetics. Perhaps due to confusion about efficacy and safety for opioid and non-opioid analgesics, patients with HCC are undertreated for pain.224 However, some generally safe options for pain management for patients with impaired liver function exist, including opioids, non-steroidal anti-inflammatory drugs (NSAIDs) in some cases, topical lidocaine patches (which have low levels of systemic absorption) for localized analgesia, tricyclic antidepressants, and anticonvulsants such as gabapentin (which is not metabolized by the liver).225

Financial toxicity is a major concern for patients with cancer,226 and immunotherapies are among the most expensive agents on the pharmaceutical market.227 228 Patients with cancer shoulder the burden of an increasing number of out-of-pocket costs for their treatment, even if they have insurance coverage.229 Treatment may cause both material and psychological financial hardship, and the risk factors for each vary. Patients of younger age, female sex, non-white race, and who change employment because of cancer are more likely to experience material financial hardship, whereas psychological hardship is more likely among those who are uninsured or have lower family income.230 The degree to which cancer causes financial burden has been shown to be the single most important predictor for poor QOL,231 and healthcare costs for HCC are substantial. In both North America and Asia, costs are highest for patients with HCC in the terminal phase of care.232 Although a comprehensive analysis of the healthcare costs associated with immunotherapy in the HCC setting has not yet been performed, oncologists should communicate with patients about how treatment may affect their financial well-being, as health insurance may not cover the costs of immunotherapy drugs.

Importantly, however, immunotherapy has generally been associated with favorable QOL outcomes compared with previous standards of care. In the landmark trials leading to FDA approval of checkpoint inhibitors for HCC, no adverse effects on QOL were observed when outcomes were reported for the patients receiving immunotherapy. Nivolumab was associated with stable patient-reported outcomes, including indicators of health status and QOL regardless of prior sorafenib in CheckMate 040.13 It is noteworthy to mention that even in a subcohort of patients from the CheckMate 040 trial with impaired liver function (Child-Pugh B), the AE profile was comparable to what was seen in patients with Child-Pugh A disease.59 Additionally, IMbrave150 provided a large and rich data set on patient-reported QOL outcomes, which complemented the efficacy data, with a reporting rate of greater than 90%. The study found that fewer patients treated with the combination of atezolizumab with bevacizumab experienced QOL deterioration compared with those receiving sorafenib. Furthermore, for the patients who did experience QOL deterioration on immunotherapy, the onset was later.233 In HIMALAYA, durvalumab in combination with tremelimumab treatment was also associated with delayed worsening of disease-related symptoms, physical functioning, and global QOL compared to sorafenib.69 Pembrolizumab also was shown to preserve HRQOL in a prespecified exploratory analysis of patients enrolled in KEYNOTE-240. Among the 271 and 127 patients randomly assigned to pembrolizumab and placebo, respectively, who completed the EORTC QLQ-C30 and the HCC supplement EORTC QLQ-HCC18, changes in both scores were similar across arms and global health status/QOL scores were stable.234 It will be important to prospectively study QOL outcomes in future immunotherapy trials, especially as new combination regimens advance through clinical development.

Panel recommendations

• Patient and caregiver education for HCC should include an overview of the liver’s function in the body, an explanation of underlying liver diseases such as HBV, HCV, and NASH, and a discussion of how immunotherapy works to treat their cancer.

• Patients must know which provider is coordinating their treatment, and they need to have clear instructions to promptly report any signs or symptoms of potential immune-related toxicities.

• Patients need counseling on the goals of treatment in advanced HCC, which is not curative in most patients, despite significant advances. Management of HCC should include focus on supportive care for uncontrolled symptoms and inclusion of palliative care specialists.

• Patients should receive education on the expected toxicities associated with immunotherapies, including hepatitis, colitis, pneumonitis, and immune-related endocrinopathies. Detailed call parameters should be provided to promptly report signs and symptoms of irAEs.

• Assessment of patients’ physical function and symptoms should be performed before, during, and after therapy.

• Patients should be referred to a treatment team including a social worker and a financial manager to assist in navigating healthcare costs and identifying support systems.

• Conversations should be initiated with patients about how the costs of immunotherapy treatment will be covered, including contributions from private insurance, Medicare and Medicaid, clinical trials, patient assistance programs, or compassionate use as needed.

• Patients should be provided information about local advocacy and support groups specific to primary liver cancer.