Systemic treatment for advanced hepatocellular carcinoma

INTRODUCTION — Hepatocellular carcinoma (HCC) is an aggressive tumor that usually occurs in the setting of chronic liver disease and cirrhosis. It is typically diagnosed late in the course of the chronic liver disease [1]. The patient's hepatic reserve, as indicated by the Child-Pugh classification, often dictates therapeutic options (table 1).

This topic will discuss systemic treatment approaches for patients with advanced unresectable HCC for whom liver transplantation, and liver-directed therapy are not appropriate. An overview of treatment for HCC, surgical treatment, liver-directed nonsurgical therapies, and the clinical manifestations and diagnosis of HCC are reviewed separately.

●(See "Overview of treatment approaches for hepatocellular carcinoma".)

●(See "Surgical resection of hepatocellular carcinoma".)

●(See "Liver transplantation for hepatocellular carcinoma".)

●(See "Localized hepatocellular carcinoma: Liver-directed therapies for nonsurgical candidates who are eligible for local ablation".)

●(See "Localized hepatocellular carcinoma: Liver-directed therapies for nonsurgical candidates not eligible for local thermal ablation".)

●(See "Clinical features and diagnosis of hepatocellular carcinoma".)

EVOLUTION OF SYSTEMIC THERAPY

●Historical context – Previously, palliative systemic therapy with conventional cytotoxic chemotherapy agents had not been used routinely for patients with advanced HCC for a number of reasons:

•HCC is considered a relatively chemotherapy-refractory tumor, in part due to the high rate of expression of drug resistance genes.

•Systemic chemotherapy is usually not well tolerated by patients with significant underlying hepatic dysfunction.

•Chemotherapy may also be less effective overall in patients with significant cirrhosis. This was illustrated in an evaluation of predictive factors among 147 patients receiving chemotherapy for HCC [2]. There were no objective responses among patients with a poor performance status, ascites, portal vein tumor thrombus, or serum bilirubin >2 mg/dL [2].

•The role of cytotoxic chemotherapy, which has modest efficacy for HCC, at best, has diminished with the advent of newer immunotherapy and molecularly targeted therapy approaches, which demonstrated better efficacy and tolerability. Cytotoxic chemotherapy has been removed from guidelines for advanced HCC by both the NCCN and ESMO [3,4]. However, at some institutions, chemotherapy is still offered to select patients for third- or fourth-line therapy if not deemed optimal candidates for other approved therapies. (See 'Systemic chemotherapy' below.)

●Advances of targeted therapies and immunotherapy – Since 2008, there has been a resurgence of interest and enthusiasm for systemic therapy of HCC with the emergence of data showing that:

•The molecularly targeted agents sorafenib and regorafenib improve survival over best supportive care alone.

•Atezolizumab plus bevacizumab and tremelimumab plus durvalumab improve survival over first-line sorafenib. Given their more favorable toxicity profile [5], and greater efficacy, immunotherapy-based combinations have now become preferred first-line therapy options over sorafenib or other molecularly targeted agents. (See 'Choice of therapy' below.)

•Lenvatinib has been shown to be noninferior to first-line sorafenib.

•An overall survival (OS) benefit has been shown in the second-line setting for immune checkpoint inhibitors nivolumab plus ipilimumab, after progression on sorafenib.

●Uncertainties – These results and US Food and Drug Administration approvals for many of these treatments have radically changed the treatment landscape for advanced HCC. However, the rapid evolution of immunotherapy and molecularly targeted therapy for advanced HCC has also led to several areas of ongoing uncertainty [6]:

•All of these targeted agents are expensive, and in almost all cases, the demonstration of benefit has been limited to those with well-preserved liver function (ie, no worse than Child-Pugh class A cirrhosis (table 1)). There still may be a role for cytotoxic chemotherapy in selected medically appropriate patients who cannot obtain these drugs and who have retained adequate hepatic function and a reasonable remaining life expectancy. (See 'Systemic chemotherapy' below.)

•For patients who have access to all agents, and sufficiently preserved liver function, the best way to sequence the available systemic treatment options for advanced HCC has not been established.

•As discussed in more detail below, the best way to integrate these new systemic therapies into the treatment algorithm for HCC is in flux, particularly for patients with unresectable yet liver isolated HCC not amenable to transplantation or ablation.

PLACING SYSTEMIC THERAPIES IN THE CONTEXT OF HCC MANAGEMENT

Treatment algorithms — Treatment options for HCC are divided into surgical therapies (ie, resection, and liver transplantation) and nonsurgical therapies, which may be liver directed (ie, radiofrequency/microwave ablation, arterially directed therapies, external beam radiation therapy) or systemic (predominantly immunotherapy with immune checkpoint inhibitors and molecularly targeted therapies). Treatments are generally selected based upon disease extent and underlying liver function (algorithm 1). An alternative treatment algorithm is available from the Barcelona Clinic (figure 1).

Algorithmic approaches such as these can help to conceptualize the available treatment options for HCC, but they may not be applicable in all settings. Furthermore, attempts to generate algorithmic approaches to the treatment of HCC are difficult since new treatments (especially systemic therapies) and indications for various treatments are evolving rapidly, and few studies have addressed the sequencing of active regimens, or the integration of modern systemic therapy into local liver-directed therapies.

This is particularly relevant for patients with liver-limited disease who are not surgical candidates because of large tumor size, number of lesions, or technically inoperable disease, and do not meet extended transplant criteria. In this setting, locoregional liver-directed therapies have been the de facto mainstay of treatment because of high local response rates and favorable safety profile, especially in the context of minimally effective systemic therapies. However, some of these patients will never receive systemic therapy because of disease progression or treatment-related decline in hepatic function. (See "Localized hepatocellular carcinoma: Liver-directed therapies for nonsurgical candidates not eligible for local thermal ablation".)

Now that more effective systemic therapies have emerged that have been shown to prolong survival, how these therapies should be integrated into the treatment algorithm for such patients is in flux. A major question is should patients be deemed refractory to locoregional therapy before switching to systemic therapy? Would earlier use of systemic therapy or combination strategies (eg, immunotherapy or molecularly targeted therapy plus locoregional therapy) improve outcomes over either systemic therapy or locoregional therapy alone? Emerging data suggest that there are at least two patient populations, those with extensive vascular invasion, and a high intrahepatic tumor burden, who have a poor outcomes with locoregional therapies, and may be better candidates for upfront systemic therapy. Furthermore, data from the LAUNCH trial suggest the potential for better outcomes for transarterial chemoembolization plus lenvatinib versus lenvatinib alone in individuals with a high tumor burden and/or macrovascular invasion [7]. These data are discussed in more detail elsewhere. (See "Overview of treatment approaches for hepatocellular carcinoma", section on 'Large intrahepatic tumor burden' and "Overview of treatment approaches for hepatocellular carcinoma", section on 'Patients with portal vein tumor thrombus'.)

The limitations of treatment algorithms for HCC and a general approach to treatment of HCC are discussed in detail elsewhere. (See "Overview of treatment approaches for hepatocellular carcinoma", section on 'Treatment algorithms'.)

Indications for systemic therapy — Systemic therapy is an appropriate option for patients with advanced unresectable HCC who are not amenable to curative or locoregional therapy and have adequate performance status and underlying liver function. Clinical trials are preferred whenever available.

Candidates for systemic therapy include those with extrahepatic spread, or tumor confined to the liver who progressed after locoregional therapies, or who have extensive vascular invasion (figure 2) or a large intrahepatic tumor burden (diffuse and/or bilateral lobar involvement) unsuitable for locoregional approaches (algorithm 1). (See "Overview of treatment approaches for hepatocellular carcinoma", section on 'Large intrahepatic tumor burden' and "Overview of treatment approaches for hepatocellular carcinoma", section on 'Patients with portal vein tumor thrombus'.)

GENERAL CONSIDERATIONS

Estimate life expectancy and assess liver function — Among patients with HCC, survival is often determined not by tumor aggressiveness or the impact of a systemic treatment, but by the degree of hepatic dysfunction. Several staging or prognostic scoring systems have been developed to guide the prognosis and treatment of patients with HCC. Historically, the Child-Pugh classification (table 1) was the most commonly used in clinical studies to define a population with relatively well preserved liver function, although it was neither developed nor validated in patients with HCC [8]. Despite this, we still endorse the use of the Child-Pugh classification for assigning the degree of hepatic dysfunction for the purpose of treatment selection, as it is easy to understand and corresponds to the selection factors used in most clinical studies to date.

Other prognostic scoring systems include the American Joint Committee on Cancer; tumor, node, metastasis (TNM); Okuda; Cancer of the Liver Italian Program; Chinese University Prognostic Index; Groupe d'Etude et de Traitement du Carcinome Hepatocellulaire; BCLC; Hong Kong Liver Cancer systems; and most recently, the albumin-bilirubin (ALBI) score. These systems were developed based on both surgical and nonsurgical patients with HCC, and they generally aim to classify patients with HCC broadly into those with early-, intermediate-, and advanced-stage disease. One of the most promising is the ALBI score. (See "Staging and prognostic factors in hepatocellular carcinoma", section on 'Staging and prognostic scoring systems'.)

Patients with advanced HCC are a heterogeneous group. There are substantial differences in overall survival (OS) that are due to factors not captured within the subgroups identified by these staging systems, and it is not clear that any of these staging systems provides a more meaningful prognostic discrimination among those with advanced disease compared with others [9,10].However, there is some evidence that outcomes of nonsurgical treatment for HCC might be better predicted by one of these prognostic scoring systems [11,12]. As examples:

●A prognostic scoring system for patients with advanced HCC not amenable to locoregional therapy has been proposed (the Advanced Liver Cancer Prognostic System [ALCPS]) that allows the separation of three prognostic groups with significantly different three-month survival [11]. The stratification is based on the assignment of a weighted point value to 11 different patient and tumor characteristics, with the total number of points corresponding to an estimated likelihood of three-month survival (table 2).

●A new model has been developed based on serum albumin and bilirubin alone (the ALBI score) that provides a simple, objective, and discriminatory method of assessing liver function in patients with HCC; it was validated in geographically distinct cohorts of patients undergoing surgery for localized disease and sorafenib for advanced disease [13]. It seems to provide better prognostic discrimination for patients with HCC than the Child-Pugh score; among patients with Child-Pugh class A cirrhosis (who constituted the vast majority of the studied populations), two prognostically distinct subgroups emerged. The objective nature and simplicity of ALBI may diminish interobserver variation (as occurs with the grading of ascites and encephalopathy in the Child-Pugh scoring system) and help to refine prognostic estimates in patients treated for HCC, particularly among those with better liver function. Independent validation of the model is needed. A calculator is available for the ALBI score (calculator 1). (See "Staging and prognostic factors in hepatocellular carcinoma", section on 'Albumin-bilirubin (ALBI) score'.)

While prognostic scoring systems such as these can objectively help clinicians select appropriate candidates for treatment and participation in clinical trials of new agents, there are no prospective data addressing their ability to predict which patients might benefit from any form of therapy. The updated guidelines from BCLC suggest that, for the purpose of treatment selection underlying liver function be evaluated by more than just the conventional Child-Pugh classification [14], but beyond jaundice, refractory ascites, and encephalopathy (which reflect non-preserved liver function); they are vague as to how best to categorize a patient as having "preserved liver function."

Screen for viral hepatitis — Many patients with HCC have underlying liver disease, especially infection with hepatitis B virus (HBV) or hepatitis C virus (HCV), and they may be at risk for viral reactivation during active systemic therapy, including immune checkpoint inhibitor immunotherapy [15]. We agree with an updated year 2020 provisional clinical opinion from the American Society of Clinical Oncology, which endorses universal HBV screening for all patients beginning systemic anticancer therapy (cytotoxic chemotherapy, immunotherapy, molecularly targeted therapy) using three tests, hepatitis B surface antigen (HbSAg), hepatitis B core antibody (anti-HBc), total immunoglobulin (Ig) or IgG, and antibody to hepatitis B surface antigen (anti-HBs) [16]. The finding of chronic HBV (HbSAg-positive) or past HBV (HbSAg-negative and anti-HBc-positive) infection requires HBV reactivation risk assessment to determine the need for antiviral prophylaxis. Their recommended approach to HBV screening, monitoring, and antiviral prophylaxis is outlined in the algorithm (algorithm 2), and discussed in more detail elsewhere. (See "Hepatitis B virus reactivation associated with immunosuppressive therapy".)

In addition, clinicians should also consider testing for chronic HCV infection prior to initiating potentially immunosuppressive chemotherapy, although the data are less compelling than for HBV testing. Patients with chronic HCV infection who are receiving chemotherapy should undergo serial monitoring of liver function tests, and if there is no dramatic change, continued chemotherapy treatment without dose modification is appropriate. (See "Screening and diagnosis of chronic hepatitis C virus infection".)

Predictive markers of response — There are few clinically applicable markers to predict disease response to any form of systemic therapy for advanced HCC, with the exception of ramucirumab, which appears to be beneficial for patients with high initial alpha-fetoprotein (AFP) levels. (See 'Ramucirumab' below.)

However, data are evolving:

●As described in detail below, treatment benefit from first-line sorafenib appears to be higher in patients with hepatitis C viral (HCV) infection than in those with other underlying risk factors for HCC. However, whether patients with other etiologies of chronic liver disease should be offered a first-line treatment other than sorafenib (eg, lenvatinib or combined atezolizumab plus bevacizumab) is unclear; there are no prospective studies that have examined treatment response to any other form of molecularly targeted therapy based upon etiology of underlying liver disease. (See 'Benefit based on etiology of liver disease' below.)

●Emerging data suggest that immunotherapy approaches (eg, front-line atezolizumab plus bevacizumab, or second-line pembrolizumab or nivolumab/ipilimumab) might be less effective for HCCs that arise in the setting of nonalcoholic steatohepatitis (NASH) as compared with other etiologies [17]. A meta-analysis included 1656 patients treated with front-line atezolizumab plus bevacizumab in the IMBrave 150 trial, or front-line nivolumab in the Checkmate-459 trial, or second-line pembrolizumab in the KEYNOTE-240 trial. (See 'Atezolizumab plus bevacizumab' below and 'Nivolumab and pembrolizumab' below and 'Pembrolizumab' below.)

When the data from these three trials were stratified according to underlying etiology of chronic liver disease, the survival benefit of immunotherapy was limited to those with underlying viral liver disease (HR 0.64, 95% CI 0.48-0.84) but not nonviral liver disease (HR 0.92, 95% 0.77-1.11). In two validation cohorts of patients with advanced HCC treated with immunotherapy targeting the programmed cell death 1 receptor pathway, those with nonalcoholic fatty liver disease (NASH, which constituted a small minority of treated patients) had a median survival that was significantly worse than that of patients with other underlying etiologies (5.4 versus 11 months in one cohort, and 8.8 versus 17.7 months in the second cohort). The authors postulated, based on preclinical studies, that the lower responsiveness to immunotherapy in patients with NASH was possibly attributed to NASH-related aberrant T-cell activation causing tissue damage leading to impaired immune surveillance.

While intriguing, these results should be viewed as hypothesis-generating only. Validation in prospective trials stratifying outcomes of various immunotherapy treatments according to etiology of chronic liver disease is needed.

Response assessment — For patients receiving systemic therapy for advanced HCC, reimaging is typically carried out using cross-sectional imaging (computed tomography [CT] or magnetic resonance imaging [MRI]) after the initial six to eight weeks of therapy and then every two to three months thereafter. We use modified RECIST for HCC (table 3) rather than standard RECIST criteria for solid tumors to assess treatment effect. Clinical status and serial measurements of the tumor marker AFP (if initially elevated) should also be considered in addition to imaging when contemplating treatment changes. (See "Assessment of tumor response in patients receiving systemic and nonsurgical locoregional treatment of hepatocellular cancer", section on 'Criteria for response assessment' and "Assessment of tumor response in patients receiving systemic and nonsurgical locoregional treatment of hepatocellular cancer", section on 'AFP'.)

Conventionally, treatment response of HCC to systemic therapy has been assessed by radiologic imaging using conventional response criteria, such as the Response Evaluation Criteria in Solid Tumors (RECIST) (table 4). However, criteria such as these may not accurately reflect response to treatment or tumor viability, particularly in the era of targeted therapy with agents such as sorafenib, which are cytostatic rather than cytotoxic. Early assessment of response to molecularly targeted agents such as these can be challenging since tumor necrosis, extension, and radiologic appearance can be inhomogeneous. As an example, the predominant patterns of HCC response after sorafenib on dynamic contrast-enhanced MRI are T1 hyperintensity due to coagulation, T2 tumor hyperintensity due to edema, and diminished enhancement in the arterial phase due to necrosis. On CT, sorafenib decreases HCC enhancement without necessarily affecting tumor size. As with locoregional therapies, residual thick and nodular foci of arterial enhancement within HCC suggest residual viable tumor.

Furthermore, an increase in tumor size due to necrosis has been reported in patients treated with multikinase inhibitors, such as sorafenib, a phenomenon referred to as "pseudoprogression" [18].

An HCC expert consensus proposed the use of modified RECIST, which mainly measures tumor viability, in assessing response for patients with HCC treated with targeted agents, such as sorafenib (table 3) [19-21]. (See "Assessment of tumor response in patients receiving systemic and nonsurgical locoregional treatment of hepatocellular cancer", section on 'Measuring tumor dimensions versus tumor viability'.)

Immunotherapy — The patterns of response to treatment with immunotherapy (eg, immune checkpoint inhibitors) differ from those with molecularly targeted agents or cytotoxic chemotherapy. (See 'Checkpoint inhibitor immunotherapy' below.)

As an example, in early trials of immune-based therapeutics in melanoma, a unique response pattern was seen, termed "pseudoprogression," in which lesions appeared to be larger or new lesions were seen related to changes in imaging characteristics rather than tumor growth. This is particularly common in liver lesions. Even new small intrahepatic lesions of 1 to 2 cm in size could reflect pseudoprogression of prior subcentimeter lesions. Infiltration of lymphocytes and inflammatory changes related to immune activation that eventually leads to a tumor response can make radiographic assessment challenging early in the course of treatment. Transient rises in AFP can also be misleading in this setting [22]. In addition, responses can take appreciably longer to become apparent compared with cytotoxic therapy. As a result of all of these issues, we continue therapy as long as there are no additional signs of progressive disease, such as deterioration in performance status or identification of new distant metastases. (See "Systemic treatment of metastatic melanoma lacking a BRAF mutation", section on 'Assessing treatment response'.)

Immune-related response criteria have been proposed to properly recognize the nontraditional patterns of response occasionally seen with checkpoint inhibitors and some other immunotherapies. This subject is discussed in more detail elsewhere. (See "Principles of cancer immunotherapy", section on 'Immunotherapy response criteria'.)

FIT PATIENTS WITH PRESERVED LIVER FUNCTION AND FUNCTIONAL STATUS

First-line therapy — Systemic therapy is appropriate for patients with advanced unresectable HCC who are unsuitable for locoregional therapy and whose liver function is adequate to tolerate therapy (ie, Child-Pugh class A or B cirrhosis (table 1)).

Choice of therapy — The following represents our recommended approach to first-line therapy:

●Our preferred approach is participation in an ongoing clinical trial testing new therapeutic strategies.

●If a clinical trial is not available or trial participation is not feasible, for healthy patients with an Eastern Cooperative Oncology Group performance status of 0 or 1 (table 5), no worse than Child-Pugh class A cirrhosis, have not recurred following liver transplantation, are not receiving therapeutic anticoagulation, and following management of esophageal varices, if applicable, we suggest atezolizumab plus bevacizumab rather than other systemic agents.

For patients who are unable to receive bevacizumab, tremelimumab plus durvalumab is an alternative to atezolizumab plus bevacizumab. Immunotherapy approaches should be avoided in patients who recur following transplantation because of the high rate of allograft rejection.

For less fit patients who have a recurrence following liver transplantation or another contraindication to atezolizumab plus bevacizumab, we suggest monotherapy with sorafenib or lenvatinib rather than conventional cytotoxic chemotherapy. Both are US Food and Drug Administration (FDA)-approved for first-line therapy, and it is not clear that either drug is superior to the other. Where available, donafenib or durvalumab are alternatives to sorafenib or lenvatinib.

●If monotherapy with a molecularly targeted agent is chosen for initial therapy, American Society of Clinical Oncology (ASCO) guidelines suggest that the choice of treatment with lenvatinib or sorafenib should be made through a discussion with the clinician and patient, and should include factors such as medical history, severity of liver disease, viral etiology of liver disease, treatment-related toxicity and cost, goals of treatment, and patient preference [23]. However:

•Some clinicians, including the author, prefer lenvatinib rather than sorafenib for most patients because of the better tolerability (especially for hand-foot skin reaction, alopecia, fatigue, and anorexia), and the higher objective response rates (ORRs) and longer time to tumor progression (TTP) in the REFLECT trial. (See 'Lenvatinib' below.)

Other clinicians prefer sorafenib, given the longer duration of experience with this drug than with lenvatinib, the lack of a survival benefit for lenvatinib over sorafenib in the REFLECT trial, and the approval of two second-line regimens (regorafenib and nivolumab plus ipilimumab) after failure of first-line sorafenib. There are no data and no approved drugs for treatment of advanced HCC after failure of first-line lenvatinib. (See 'Sorafenib' below and 'Regorafenib' below and 'Checkpoint inhibitor immunotherapy' below.)

•Treatment with tyrosine kinase inhibitors (TKIs) may be less effective in patients with more advanced liver disease. Consensus-based National Comprehensive Cancer Network (NCCN) guidelines suggest that sorafenib use in those with Child-Pugh class B disease be limited to those whose score is no higher than 7 (table 1). NCCN guidelines suggest limiting use of lenvatinib to individuals with no worse than Child-Pugh class A cirrhosis.

•The survival benefit from sorafenib appears to be higher in patients with hepatitis C virus (HCV) infection than in those with other underlying risk factors, including hepatitis B virus (HBV), based upon multiple meta-analyses. In the REFLECT trial, there was a trend toward improvement for lenvatinib over sorafenib in the HBV subgroup, but the differences were not significant. (See 'Benefit based on etiology of liver disease' below and 'Lenvatinib' below.)

●If sorafenib is chosen, to improve early tolerability, we typically start at 200 mg twice a day and increase the daily dose in 200 mg increments approximately every five days until the target dose of 400 mg twice daily is reached.

Sorafenib is feasible in conjunction with immunosuppressive therapy in patients with a recurrence of HCC following orthotopic liver transplantation (OLT), but it may be more toxic in this setting. Treatment should be closely monitored due to the potential for severe side effects and the need for dose modification. There are no data on the safety and efficacy of lenvatinib in conjunction with immunosuppressive therapy in patients with an HCC recurrence after OLT. (See 'Efficacy following TACE or liver transplantation' below.)

The safety and benefit of combining sorafenib or lenvatinib with cytotoxic chemotherapy are not yet definitively established, and we suggest against these strategies outside of the context of a clinical trial. (See 'Is there a role for sorafenib plus chemotherapy?' below.)

●These recommendations assume that patients have access to all available therapies. Molecularly targeted and immunotherapy agents are expensive. For patients who are unable to obtain these agents, systemic chemotherapy remains an option. (See 'Systemic chemotherapy' below.)

●In general, first-line treatment is continued until disease progression or intolerance. Issues related to response assessment are discussed in detail elsewhere. (See "Assessment of tumor response in patients receiving systemic and nonsurgical locoregional treatment of hepatocellular cancer", section on 'Measuring tumor dimensions versus tumor viability'.)

Atezolizumab plus bevacizumab — Combined therapy with atezolizumab plus bevacizumab has been shown to improve survival as compared with front-line sorafenib monotherapy. For healthy patients with no worse than Child-Pugh class A cirrhosis (table 1), an excellent performance status, who have no contraindications to bevacizumab, and have not recurred following liver transplantation, we suggest atezolizumab plus bevacizumab rather than sorafenib monotherapy. This recommendation is consistent with a year 2020 guideline from ASCO [23], year 2021 guidelines from the Society for Immunotherapy of Cancer [24], the NCCN, the Barcelona Clinic Liver Cancer, a position paper from the European Association for Study of the Liver [25], and the American Gastroenterological Association [25]. However, the survival benefit compared with other first-line treatment options must be balanced by the higher cost for dual antibody therapy and the potential for adverse events.

Immunotherapy approaches should be avoided in patients who recur following liver transplantation because of the high rate of allograft rejection. (See 'Checkpoint inhibitor immunotherapy' below.)

Atezolizumab is, a humanized monoclonal antibody immune checkpoint inhibitor that binds to programmed cell death ligand 1 (PD-L1), and bevacizumab is a monoclonal antibody targeting vascular endothelial growth factor (VEGF); there are increasing data to support this combination for first-line treatment of advanced HCC [26-28]. Combination therapy was directly compared with sorafenib monotherapy in the phase III IMBrave 150 trial, which randomly assigned 501 previously untreated patients with advanced unresectable HCC and no worse than Child-Pugh class A cirrhosis 2:1 to atezolizumab (1200 mg intravenous [IV] every three weeks) plus bevacizumab (15 mg/kg IV every three weeks, after atezolizumab) or to sorafenib (400 mg twice daily) [27]. Due to the risk of bleeding, patients enrolled in this trial were required to have undergone esophagogastroduodenoscopy within six months of treatment initiation, and to have treatment of esophageal varices, when necessary. The trial excluded patients who had a myocardial infarction or stroke with the previous three months, were on therapeutic anticoagulation, or had coinfection with HCV or HBV.

In the latest analysis, at a median follow-up of 15.6 months, median OS with combined therapy was significantly better (19.2 versus 13.4 months, estimated HR for death 0.66, 95% CI 0.52-0.85) [29]. ORRs were nearly threefold higher with combined therapy (30 versus 11 percent). Treatment-related grade 3 or 4 adverse events occurred in a similar percentage of patients in each group (43 versus 46 percent), but hypertension, transaminase elevation, and proteinuria were more frequent with combined therapy; diarrhea was more common with sorafenib.

In a separate analysis of patient-reported outcomes from this trial, patients who received atezolizumab plus bevacizumab reported significantly longer delays in median time to deterioration of quality of life (not estimable versus 4.2 months), physical functioning (13.1 versus 5.6 months), and role functioning (9.1 versus 4.3 months) [28]. Combination therapy was also associated with a reduced risk of deterioration of several disease-related symptoms (eg, anorexia, diarrhea, fatigue, pain) compared with sorafenib alone.

Largely based upon these data, the combination of atezolizumab plus bevacizumab was approved by the US FDA on May 29, 2020, for treatment of patients with unresectable or metastatic HCC who have not received prior systemic therapy. Further support for first-line atezolizumab plus bevacizumab was provided in a year 2020 network meta-analysis of eight trials of molecularly targeted therapy and immunotherapy in the first-line setting (totaling 6290 patients), which concluded that OS was superior for combination of atezolizumab plus bevacizumab compared with sorafenib (HR 0.58, 95% CI 0.42-0.80), lenvatinib (HR 0.63, 95% CI 0.44-0.89), and nivolumab (HR 0.68, 95% CI 0.48-0.98) [30].

Few data are available on the safety of immunotherapy following liver transplantation; however, allograft rejection rates may be as high as 39 percent because of stimulation of the host immune response by these agents [31]. Based on these data, immunotherapy approaches should be avoided in patients who recur following transplantation.

Tremelimumab plus durvalumab — Combined therapy with durvalumab plus a priming dose of tremelimumab has been shown to improve survival over first-line sorafenib monotherapy. For healthy patients with no worse than Child-Pugh class A cirrhosis (table 1), an excellent performance status, have not recurred following liver transplantation, and are unable to receive bevacizumab, tremelimumab plus durvalumab is an alternative to atezolizumab plus bevacizumab.

Combinations of durvalumab, an anti-PD-L1 monoclonal antibody, plus a single priming dose of the anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) monoclonal antibody tremelimumab appear to be tolerable and clinically active in patients who progressed on, were intolerant to, or refused sorafenib [32]. The comparative efficacy of this regimen for first-line therapy was addressed in the phase III HIMALAYA trial, in which 1171 patients with no worse than Child-Pugh class A cirrhosis and ineligible for locoregional therapy were randomly assigned to a single dose of tremelimumab (300 mg) plus durvalumab (1500 mg followed by durvalumab alone [1500 mg every four weeks]), or durvalumab alone (1500 mg every four weeks), or sorafenib (400 mg twice daily); enrollment to a fourth arm, tremelimumab 75 mg every four weeks plus durvalumab 1500 mg every four weeks was discontinued when interim analysis suggested similar results to durvalumab alone [33]. Combined therapy with durvalumab plus a single priming dose of tremelimumab provided significantly longer median OS compared with sorafenib (16.4 versus 13.8 months, HR for death 0.78, 95% CI 0.65-0.93). Combined therapy was also associated with a higher objective response rate (20 versus 5 percent) and three-year OS (31 versus 20 percent), although similar median progression-free survival (PFS; 3.8 versus 4.1 months). There was also a trend towards better median OS with durvalumab monotherapy compared with sorafenib alone. These results are described below. (See 'Durvalumab' below.)

Largely based on these results, the US FDA has approved the combination of tremelimumab plus durvalumab for patients with unresectable HCC [34].

Lenvatinib

●Lenvatinib alone – In the REFLECT trial, lenvatinib was shown to be noninferior to first-line sorafenib, which improves survival over supportive care alone. For most patients who are ineligible for or lack access to atezolizumab plus bevacizumab or durvalumab plus tremelimumab, we prefer lenvatinib over sorafenib because in the REFLECT trial, compared with sorafenib, lenvatinib was better tolerated, achieved a higher objective response rate, and delayed the time to tumor progression.

Lenvatinib is an inhibitor of VEGFR-1, VEGFR-2, and VEGFR-3, as well as fibroblast growth factor receptors (FGFR) 1 to 4, platelet-derived growth factor receptor (PDGFR) alpha, RET, and KIT. Activity in advanced HCC is supported by the following studies:

•A phase II trial of first-line lenvatinib (12 mg once daily) in 46 patients with advanced HCC who did not qualify for surgical resection or local therapies reported a partial response in 17 patients (37 percent) and stable disease in an additional 19 patients (41 percent) [35]. Median TTP was 7.8 months, and median OS was 18.7 months. Toxicity was prominent, with hypertension, hand-foot skin reaction, decreased appetite, and proteinuria in 76, 65, 61, and 61 percent, respectively. Dose reductions were needed more often in patients with a low body weight.

•A subsequent randomized noninferiority trial (the REFLECT study) compared lenvatinib (12 mg once daily for body weight ≥60 kg, 8 mg daily for <60 kg) versus sorafenib (400 mg twice daily for all patients) in 954 patients with unresectable HCC and no prior systemic therapy (99 percent Child-Pugh class A) [36]. Patients with involvement of >50 percent of the liver or invasion of the main portal vein or biliary tree were excluded. The predefined noninferiority margin (primary endpoint OS) was 1.08. Lenvatinib was noninferior to sorafenib for median OS, the main endpoint (13.6 versus 12.3 months, HR 0.92, 95% CI 0.79-1.06), the objective response rate was higher (24 versus 9 percent), and median TTP was longer (7.4 versus 3.7 months, HR 0.66, 95% CI 0.57-0.77). From a toxicity standpoint, the rate of grade 3 or 4 hypertension was higher with lenvatinib (23 versus 14 percent), while hand-foot skin reaction was more frequent with sorafenib (52 versus 37 percent any grade, 11 versus 3 percent grade 3 or worse), as was alopecia of any grade (25 versus 3 percent).

Largely based on the REFLECT trial, lenvatinib was approved in Japan in March 2018 for treatment of unresectable HCC, and it received approval in the United States for first-line treatment of unresectable HCC in August 2018. Consensus-based guidelines from the NCCN suggest limiting use of lenvatinib to individuals with no worse than Child-Pugh class A cirrhosis.

●Other lenvatinib-based regimens

•Lenvatinib plus pembrolizumab – We do not offer the combination of lenvatinib plus pembrolizumab as initial therapy for advanced hepatocellular carcinoma. Although initial studies suggested clinical benefit for this combination [37], the addition of pembrolizumab to lenvatinib did not meet the prespecified boundaries for PFS and OS in a randomized phase III trial (LEAP-002) [38].

•Lenvatinib plus TACE – Additional experience with lenvatinib plus TACE is needed in Western populations, in which the cause of HCC is more often alcoholic cirrhosis than HBV, before it can be concluded that this combination is a preferred approach over systemic therapy alone, especially upfront immunotherapy.

Better outcomes with combined lenvatinib plus TACE versus lenvatinib alone were suggested in the Chinese phase III LAUNCH trial in which 338 patients with advanced HCC (predominantly HBV-related) and no prior local or systemic treatment were randomly assigned to lenvatinib or lenvatinib plus "on-demand" TACE [7]. Although median OS was significantly longer with combined therapy (17.8 versus 11.5 months, HR 0.45, 95% CI 0.34-0.55), grade 3 or 4 adverse events were also more common. This trial is described in more detail elsewhere. (See "Overview of treatment approaches for hepatocellular carcinoma", section on 'Lenvatinib alone or with TACE'.)

Sorafenib — Sorafenib has been shown to improve survival over best supportive care alone. Sorafenib is an option for first-line therapy in individuals who are not eligible or lack access to front-line atezolizumab plus bevacizumab, and for those with no worse than Child-Pugh class B7 cirrhosis (table 1), who are not eligible for lenvatinib. For most patients, we prefer lenvatinib, given its better tolerability, higher response rate, and longer TTP in the REFLECT trial. The safety of this drug in patients with less well-preserved liver function is discussed below. (See 'Less fit patients and those with Child-Pugh B cirrhosis' below.)

Sorafenib (Nexavar) is a multitargeted, orally active small molecule TKI that inhibits Raf kinase and the vascular endothelial growth factor receptor (VEGFR) intracellular kinase pathway [39]. It was the first systemic agent to show an OS benefit for advanced HCC in a placebo-controlled trial (the SHARP trial).

The following data are available:

●The multicenter European SHARP trial randomly assigned 602 patients with inoperable HCC and Child-Pugh class A cirrhosis to sorafenib (400 mg twice daily) or placebo [40]. OS, the primary endpoint, was significantly longer in the sorafenib-treated patients (10.7 versus 7.9 months), as was time to radiologic progression (5.5 versus 2.8 months). Similar to the phase II trial described above, ORRs were low according to RECIST (seven partial responses [2 percent]). (See 'Response assessment' above.)

The only grade 3 or 4 adverse effects that occurred significantly more often in the treated group were diarrhea (8 versus 2 percent) and hand-foot skin reaction (8 versus <1 percent). There were no differences in liver dysfunction or bleeding. The overall incidence of sorafenib-related side effects (particularly dermatologic effects) was low compared with that reported by others [41]. (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'Hand-foot skin reaction'.)

These results established sorafenib monotherapy as the new reference standard systemic treatment for advanced HCC and formed the basis for approval of sorafenib for unresectable HCC in the United States.

●The efficacy of sorafenib in Asian patients was the subject of a second placebo-controlled phase III trial in which 226 patients with Child-Pugh class A cirrhosis and no prior systemic therapy for HCC received sorafenib 400 mg twice daily or placebo [42]. Patients receiving sorafenib had significantly better median OS (6.5 versus 4.2 months) and TTP (2.8 versus 1.4 months). Grade 3 or 4 side effects included hand-foot skin reaction (11 percent), diarrhea (6 percent), and fatigue (3 percent).

The magnitude of the absolute survival benefit was markedly less in this trial than in the SHARP trial. In fact, the treated group in the Asian trial had a shorter survival duration than the control group in the SHARP trial (6.5 versus 7.9 months) despite the fact that both trials used the same entry criteria. Nevertheless, patients accrued to the Asian study were more ill at the start of therapy than those in the SHARP trial, with generally worse performance statuses and more advanced stages of disease [43].

Benefit based on etiology of liver disease — Treatment benefit from sorafenib appears to be much higher in patients with HCV than in those with other underlying risk factors. However, whether patients with other etiologies of chronic liver disease should be preferentially offered first-line treatment other than sorafenib (eg, lenvatinib) is unclear; there are no prospective studies that have examined treatment response to any other form of molecularly targeted therapy based on HCC etiology, although preliminary reports suggest that immunotherapy benefit may be limited in patients with nonalcoholic steatohepatitis. (See 'Lenvatinib' above and 'Predictive markers of response' above.)

Although the Asia Pacific trial showed a benefit for sorafenib in HBV-infected patients [42], an increasing body of data suggests that patients with HCV infection as the etiology of their cirrhosis may have a better response to sorafenib compared with those with other underlying causes of cirrhosis [44-47]. As examples:

●In an exploratory analysis of the phase III SHARP trial, although a survival benefit was seen in all subgroups treated with sorafenib, the difference in median OS between sorafenib-treated and placebo-treated patients was highest in those with HCV related cirrhosis (6.6 months [14 versus 7.4 months]); it was 3.6 months (9.7 versus 6.1 months) in patients with HBV related cirrhosis and 2.3 months (10.3 versus 8 months) in those with underlying alcohol-related liver disease [44].

●An individual patient data meta-analysis of three phase III trials in which sorafenib was the control arm [48-50] came to the same conclusion [45]. Hazard ratios (HRs) showed that survival was better with sorafenib only among those patients who were both HBV negative and HCV positive (median unadjusted survival 12.6 versus 10.2 months for the "other" treatments). By contrast, sorafenib did not lead to better survival for patients who were either HBV positive or HCV negative.

These differences in outcome according to hepatitis virus type could potentially explain some of the survival differences between the SHARP (in which only 18 percent had HBV) and Asian trials (in which 73 percent had HBV-related HCC) of sorafenib. However, the available data are scant, and further study is needed to establish the influence of underlying liver disease on sorafenib treatment responsiveness. (See 'Sorafenib' above.)

Better predictive biomarkers of response to sorafenib are needed. Many clinical (ie, dermatologic adverse events [51], hypertension, diarrhea [52,53]), biochemical (eg, baseline levels of aspartate aminotransferase and other plasma biomarkers [54,55]), and molecular [56-58] determinants of outcome have been suggested to predict benefit from sorafenib treatment, but none has been validated and accepted in clinical practice.

Efficacy following TACE or liver transplantation — Although limited, the available data support the safety and efficacy of sorafenib after transarterial chemoembolization (TACE). Furthermore, sorafenib might represent a feasible treatment option in conjunction with immunosuppressive therapies in patients with a recurrence of HCC following OLT, but treatment should be closely monitored due to the potential for severe side effects. Dose adjustment may be required.

Few studies have evaluated the efficacy of sorafenib after TACE. TACE aims to eradicate the blood supply to the tumors and thus, at least in theory, could limit delivery of drug to the liver. However, at least some data support the safety and efficacy of sorafenib in this setting [59].

On the other hand, several randomized trials have not been able to demonstrate a survival benefit for the addition of sorafenib to TACE versus TACE alone. These studies are addressed in detail elsewhere. (See "Localized hepatocellular carcinoma: Liver-directed therapies for nonsurgical candidates not eligible for local thermal ablation", section on 'Does sorafenib or lenvatinib add benefit to TACE?'.)

There is less certainty about the risk to benefit ratio of sorafenib in patients with recurrent HCC after OLT; in particular, there are sparse data on the safety of combining sorafenib with immunosuppressive therapies, such as mechanistic (previously called mammalian) target of rapamycin (mTOR) or calcineurin inhibitors. The following data are available:

●The largest experience comes from an open-label cohort study of 26 patients with an HCC recurrence after OLT who were not suitable for surgical resection or locoregional treatment and who were receiving a combination of an mTOR inhibitor and sorafenib [60]. Immunosuppressive therapy was initially switched to an mTOR inhibitor (in view of accumulating data suggesting an antiproliferative effect against HCC as compared with calcineurin inhibitors), with sorafenib started approximately 1.1 to 1.4 months later. (See "Liver transplantation for hepatocellular carcinoma", section on 'Degree and type of immunosuppression'.)

Ten of 26 patients started sorafenib at a dose of 800 mg daily, while 16 started at 400 mg daily. The overall disease control rate was 54 percent, median TTP was 6.8 months, and median OS was 19.3 months. There were only two cases of grade 3 or 4 hyperglycemia and one case of grade 3 or 4 mucositis thought possibly related to the mTOR inhibitor. The most common severe adverse event probably related to sorafenib was diarrhea (13 percent).

●On the other hand, higher rates of treatment-related toxicity have been noted by others [61-63]. As an example, in a review of 20 patients who experienced a tumor recurrence after OLT for HCC, 13 received sorafenib; calcineurin inhibitors were continued in four patients and were switched to an mTOR inhibitor in nine [61]. Side effects (mainly acute hepatitis, diarrhea, hand-foot skin reaction, and myelosuppression) prevented full dosing of sorafenib. Grade 3 to 4 adverse events were observed in 92 percent of patients, necessitating discontinuation of sorafenib in 77 percent. However, among patients receiving sorafenib plus an mTOR inhibitor, one patient achieved a partial response and four achieved stable disease.

Side effects — Side effects of antiangiogenic TKIs, such as sorafenib, may include hypertension, renal toxicity, arterial thromboembolism, bleeding, cardiotoxicity, thyroid dysfunction, hand-foot skin reaction, rash, pruritus, alopecia, problems with wound healing, potentially fatal hepatotoxicity, toxic/metabolic encephalopathy, and muscle wasting. (See "Cardiovascular toxicities of molecularly targeted antiangiogenic agents" and "Non-cardiovascular toxicities of molecularly targeted antiangiogenic agents" and "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'VEGFR/PDGFR inhibitors' and "Neurologic complications of cancer treatment with molecularly targeted and biologic agents", section on 'Axitinib, cabozantinib, lenvatinib, pazopanib, selpercatinib, sorafenib, sunitinib, and tivozanib'.)

Sorafenib has been associated with potentially fatal liver toxicity that is characterized predominantly by a hepatocellular pattern of liver damage with significant increases in transaminases, although this is rare; uncommonly, elevations in the international normalized ratio or hyperbilirubinemia may occur. Liver function tests should be monitored regularly during treatment. If transaminases are significantly increased without alternative explanation, such as viral hepatitis or progressing underlying malignancy, sorafenib should be discontinued. (See "Chemotherapy hepatotoxicity and dose modification in patients with liver disease: Molecularly targeted agents", section on 'Sorafenib'.)

Is there a role for sorafenib plus chemotherapy? — There is no established role for combining sorafenib with cytotoxic chemotherapy as initial treatment for advanced HCC. We do not pursue this approach outside of a clinical trial. Although a phase II trial suggested clinical benefit for the addition of sorafenib to doxorubicin [64], an overall survival benefit could not be confirmed in a randomized phase III trial, Cancer and Leukemia Group B (CALGB) trial 80802 [65]. The study was stopped prematurely by the data monitoring safety board after a planned interim analysis suggested futility for the combination.

Another randomized phase II trial (PRODIGE 10) also failed to demonstrate a PFS benefit for the addition of gemcitabine and oxaliplatin (GEMOX) to sorafenib as initial therapy for advanced HCC [66].

Donafenib — Donafenib is a derivative of sorafenib with a more favorable pharmacokinetic profile, largely because of reduced susceptibility to hepatic drug metabolizing enzymes [67]. Modestly superior efficacy and tolerability with donafenib was shown in a Chinese trial directly comparing both drugs for first-line treatment of advanced HCC [68]. Donafenib is approved in China for treatment of patients with unresectable HCC who have not previously received systemic treatment [69]; however, it is not available elsewhere.

Immunotherapy alone

Durvalumab — We consider durvalumab monotherapy to represent a reasonable alternative to sorafenib if patients are ineligible for combined therapy with either atezolizumab plus bevacizumab or durvalumab plus tremelimumab.

Durvalumab is an anti-PD-L1 monoclonal antibody. The noninferiority of durvalumab monotherapy over sorafenib for first-line treatment of advanced HCC was shown in a preliminary report of the three-arm HIMALAYA trial, which compared sorafenib versus durvalumab alone or with tremelimumab. The results of combined therapy versus sorafenib alone are described above. (See 'Tremelimumab plus durvalumab' above.)

Briefly, this trial randomly assigned 1171 patients with no worse than Child-Pugh class A cirrhosis and ineligible for locoregional therapy to a single dose of tremelimumab (300 mg) plus durvalumab (1500 mg followed by durvalumab alone [1500 mg every four weeks]), durvalumab alone (1500 mg every four weeks), or sorafenib (400 mg twice daily) [33]. Compared with sorafenib alone, durvalumab monotherapy was judged noninferior (median OS 16.6 versus 13.8 months, HR 0.86, 95% CI 0.73-1.03). Other benefits for durvalumab included a higher objective response rate (17 versus 5 percent), lower rate of grade 3 or 4 treatment-related adverse effects (13 versus 37 percent), and fewer patients discontinuing therapy because of adverse effects (4 versus 11 percent). Compared with combined durvalumab plus tremelimumab, median OS was similar with durvalumab monotherapy (median 16.4 versus 16.6 months), but three-year survival rates still favored combined therapy over durvalumab alone (30.7 versus 24.7 percent).

Nivolumab and pembrolizumab — First-line nivolumab could be considered for patients with advanced HCC in whom TKIs and antiangiogenic drugs are contraindicated or unavailable, but nivolumab monotherapy is not approved either in the United States or Europe for front-line therapy. (See 'Pembrolizumab' below and 'Nivolumab plus ipilimumab' below.)

The immune checkpoint inhibitors pembrolizumab and nivolumab plus ipilimumab are approved for treatment of HCC in patients previously treated with sorafenib. (See 'Checkpoint inhibitor immunotherapy' below.)

The benefit of first-line immunotherapy for previously untreated patients with advanced HCC remains uncertain:

●Benefit for first-line pembrolizumab was suggested in an early report of a cohort of 51 patients enrolled in the uncontrolled phase II KEYNOTE-224 trial who had not received any prior systemic therapy for advanced HCC [70]. The objective response rate was 16 percent (all partial responses), and the median duration of response was 16 months (range 3 to 24+ months). The median PFS and OS durations were 4 and 17 months, respectively. There were no new toxicity signals, and the most common treatment-related adverse events were diarrhea, fatigue, hypothyroidism, and myalgias. One patient died with immune-mediated myocarditis and hepatitis.

●First-line nivolumab was directly compared with sorafenib in the phase III CheckMate 459 trial, which enrolled 743 patients with advanced, previously untreated HCC [71]. At a minimum follow-up of 22.8 months, nivolumab was associated with a twofold higher objective response rate (15 versus 7 percent) and more complete responses (4 versus 1 percent), but this did not translate into a significant benefit in either PFS (median 3.7 versus 3.8 months) or OS (median 16.4 versus 14.7 months; HR 0.85, 95% CI 0.72-1.02). Grade 3 or 4 treatment-related serious adverse events were reported in a similar number of nivolumab- and sorafenib-treatment patients (12 versus 11 percent), and there was no significant difference in the number of patients who discontinued therapy because of side effects.

Other regimens — Some regimens are effective initial therapies for advanced HCC but are only available in certain countries, while others are not used due to lack of efficacy.

●Camrelizumab plus rivoceranib (apatinib) – For fit patients with advanced or metastatic HCC and no worse than Child-Pugh class A cirrhosis, initial therapy with camrelizumab plus rivoceranib (ie, apatinib) improved PFS and OS relative to sorafenib in an international phase III trial. However, this combination is not available outside of China.

In an international, open-label phase III trial (CARES-310), 543 patients with systemic-therapy naïve, unresectable or metastatic HCC and Child-Pugh Class A liver function were randomly assigned to either camrelizumab (a programmed cell death protein 1 [PD-1] inhibitor) plus rivoceranib (ie, apatinib, an antiangiogenic agent that targets VEGFR2) or sorafenib [72]. Patients were included from China, Hong Kong, Taiwan, and South Korea (83 percent) as well as Europe and the United States (17 percent). Compared with sorafenib alone, camrelizumab plus rivoceranib improved PFS at median follow-up of eight months (5.6 versus 3.7 months, HR 0.52, 95% CI 0.41-0.65), OS at median follow-up of 15 months (22 versus 15 months, HR 0.62, 95% CI 0.49-0.80), and objective response rates (25 versus 6 percent). Grade ≥3 toxicity rates were higher for the combination (81 versus 52 percent).

Camrelizumab and rivoceranib is approved as a first-line treatment for patients with liver cancer by the Chinese National Medical Products Administration (NMPA) [73].

●Sintilimab plus bevacizumab — The combination of sintilimab, a PD-1 inhibitor, with a bevacizumab biosimilar improved OS and PFS compared with sorafenib alone in a randomized trial (ORIENT-32) conducted in China [74]. Sintilimab is available in China but does not have regulatory approval in the United States.

●Cabozantinib plus atezolizumab — We do not use the combination of cabozantinib plus atezolizumab as initial systemic therapy for advanced HCC. In a randomized phase III trial (COSMIC-312) of 837 patients with systemic therapy-naïve advanced HCC, cabozantinib plus atezolizumab failed to improve OS and increased treatment-related toxicity relative to sorafenib [75].

Second-line therapy — Second-line therapy is an option for patients whose tumors progress while on first-line therapy and whose performance status and liver function are sufficient to tolerate it. The best regimen is not established, and it depends, in part, on what was administered first-line. Most notably, all of the data examining the benefit of any second-line immunotherapy treatment were derived in patients initially treated with sorafenib, and there are no data on efficacy in those initially treated with atezolizumab plus bevacizumab. There are also no data to inform optimal sequencing of later lines of treatment after any first-line regimen [76]. The side effect profile of each individual regimen must be carefully considered in patients who have advanced liver disease and/or a short life expectancy.

The availability of effective second-line regimens has prompted a shift away from symptomatic progression as an endpoint to stop first-line therapy (as was used in the SHARP trial (see 'Sorafenib' above)) to radiologic progression as assessed by cross-sectional imaging that assesses both tumor viability and size. However, given the difficulties with radiographic imaging, particularly with molecularly targeted agents, we favor use of clinical status and serial assay of alpha-fetoprotein (AFP), if initially elevated, in conjunction with radiographic imaging if a change in treatment is being considered. (See 'Response assessment' above.)

Selection of therapy — The following represents our general approach to second-line therapy in suitable patients, which is consistent with ASCO guidelines [23]:

●We prefer that eligible patients are enrolled in clinical trials testing new treatment strategies, if possible.

●For patients who are not eligible for clinical trials or if they are not available, options for patients initially treated with atezolizumab plus bevacizumab or durvalumab plus tremelimumab include sorafenib, lenvatinib, regorafenib, cabozantinib, or apatinib, where available.

●For most patients who have progressed on or are unable to tolerate sorafenib or lenvatinib, and who have not undergone liver transplantation, we suggest nivolumab plus ipilimumab rather than other immunotherapy options or a second tyrosine kinase inhibitor. However, this regimen should not be administered to patients with Child-Pugh class B or C. Pembrolizumab is an appropriate alternative for those who are unable to tolerate the potential toxicities of nivolumab plus ipilimumab. (See 'Checkpoint inhibitor immunotherapy' below.)

While regorafenib and cabozantinib have also been approved for this patient group, immune checkpoint inhibitors have the potential for a higher objective response rate (including some complete responders) and the more favorable side effect profile. The higher response rate may translate into a real life benefit for patients with large or symptomatic tumors and may represent a preferred option for these patients, although this has yet to be confirmed in actual practice. (See 'Regorafenib' below and 'Checkpoint inhibitor immunotherapy' below.)

Lenvatinib is another option for treatment of patients who have no worse than Child-Pugh class A cirrhosis after failure of sorafenib. However, whether lenvatinib will work after failure of sorafenib is not known; there are no trials to inform this issue. (See 'Lenvatinib' above.)

Cabozantinib is another option for treatment of patients who have no worse than Child-Pugh class A cirrhosis, particularly for those who are intolerant of sorafenib or regorafenib. (See 'Cabozantinib' below.)

Ramucirumab is another option for the subset of patients with a high AFP level (>400 ng/mL). (See 'Ramucirumab' below.)

For patients initially treated with sorafenib, the choice of any of these TKIs or immunotherapy is empiric, as there are no comparator trials in this setting. A network meta-analysis of five trials of second-line therapy (pembrolizumab, cabozantinib, regorafenib, ramucirumab, brivanib) after failure of sorafenib concluded that the OS advantage relative to placebo was limited to second-line regorafenib (HR 0.62, 95% CI 0.51-0.75) and cabozantinib (HR 0.76, 95% CI 0.63-0.92) [30]. Indirect comparisons between the individual drugs were largely uninformative.

●For patients who have progressed on first-line lenvatinib and who have no worse than Child-Pugh class A cirrhosis, we also prefer nivolumab with ipilimumab, or pembrolizumab rather than sorafenib or another TKI. (See 'Nivolumab monotherapy' below and 'Pembrolizumab' below.)

Sorafenib is an option after failure of first-line lenvatinib, but whether it will work in this setting is not known.

Cabozantinib is another option for treatment of patients who have no worse than Child-Pugh class A cirrhosis, particularly for those who are intolerant of sorafenib or regorafenib. (See 'Cabozantinib' below.)

Ramucirumab is another option for the subset of patients with a high AFP level (>400 ng/mL). (See 'Ramucirumab' below.)

●For patients initially treated with a TKI, atezolizumab plus bevacizumab is another option for appropriately selected patients. Although there are no data on the efficacy of this combination after failure of a TKI, it is likely that most patients being considered for atezolizumab plus bevacizumab in this setting did not have access to this combination when they started treatment.

●These recommendations assume that patients have access to all available therapies. Molecularly targeted agents and immune checkpoint inhibitors are expensive. For patients who are unable to obtain these agents, systemic chemotherapy remains an option. The best regimen is not established. Options for fit patients include GEMOX, gemcitabine plus pegylated liposomal doxorubicin (PLD), or capecitabine monotherapy, especially for jaundiced individuals. (See 'Systemic chemotherapy' below.)

●In general, second-line therapy is continued until radiologic disease progression or intolerance. Issues related to response assessment during therapy are addressed elsewhere. (See "Assessment of tumor response in patients receiving systemic and nonsurgical locoregional treatment of hepatocellular cancer", section on 'Measuring tumor dimensions versus tumor viability'.)

Checkpoint inhibitor immunotherapy — Immunotherapy is our preferred option for second-line therapy in patients previously treated with sorafenib or lenvatinib who have not undergone liver transplantation. Options include nivolumab plus ipilimumab or pembrolizumab monotherapy. The role of immunotherapy as a first-line option is discussed above. (See 'Nivolumab and pembrolizumab' above.)

Immunotherapy approaches should be avoided in patients who recur following orthotopic liver transplantation because of the high rate of allograft rejection [31].

Nivolumab plus ipilimumab — An OS benefit has been shown in the second-line setting for combined nivolumab plus ipilimumab compared with nivolumab alone, after progression on sorafenib, and this approach could be considered if immunotherapy was not administered for front-line therapy.

Nivolumab is a fully human monoclonal antibody that targets the PD-1, restoring T cell immune activity directed against the tumor cell. Ipilimumab is an inhibitor of a different immune checkpoint (CTLA-4), and combined therapy with nivolumab effectively targets two different immune checkpoints that restrain the adaptive immune response. (See "Principles of cancer immunotherapy".)

The efficacy of combined therapy was addressed in the phase I/II CheckMate 040 trial, in which 148 sorafenib-treated patients were randomly assigned to one of three groups: nivolumab 1 mg/kg plus ipilimumab 3 mg/kg every three weeks for four doses, followed by nivolumab 240 mg every two weeks; nivolumab 3 mg/kg plus ipilimumab 1 mg/kg every three weeks for four doses, followed by nivolumab 240 mg every two weeks; or nivolumab 3 mg/kg every two weeks plus ipilimumab 1 mg/kg every six weeks [77]. All patients had no worse than Child-Pugh class A cirrhosis. When the three groups were combined, the objective response rate was 31 percent (more than twice as high as what was seen with nivolumab alone [14 percent]), and seven had a complete response. Almost one-third of the responses in each group were still ongoing at 24 months. In the latest analysis, when the three groups were compared, the best outcomes were seen with nivolumab 1 mg/kg plus ipilimumab 3 mg/kg every three weeks, followed by maintenance nivolumab. Of the 50 patients who received this combination, there were 16 objective responses (32 percent), four of which were complete, and the median OS was 22.2 months (versus 12.5 and 12.7 months in the other two groups) [78]. The median duration of response in the three groups was 17.5, 22.2, and 16.6 months, respectively, and the disease control rate was similar in the three groups (54, 43, and 49 percent, respectively).

Immune-mediated adverse events were also more common in this cohort (10 of 49, 20 percent), median time to onset was 1.3 months [79]. High-dose glucocorticoids were administered to 70 percent (median 14 days), and complete resolution occurred in 70 percent. Of the four patients who reinitiated treatment after symptom improvement, none had recurrence of hepatitis. Other immune-mediated side effects included rash in 35 percent (17 of 49; median time to onset 15 days), adrenal insufficiency in 18 percent (9 of 49; median time to onset 2.8 months), hypothyroidism or thyroiditis in 22 percent (5 of 49; median time to onset 1.4 months), colitis in 10 percent (5 of 49; median time to onset 2 months), pneumonitis in 10 percent (5 of 49; median time to onset 8.3 months), and infusion-related reactions in 8 percent (4 of 49).

Finally, in all three groups combined, virologic breakthrough, defined as a one-log increase in HBV DNA or HCV RNA from baseline, was observed in 7 of 82 HBV-infected patients (9 percent) and 4 of 39 HCV-infected patients (10 percent) overall. (See 'Screen for viral hepatitis' above.)

Largely based on this study, combination nivolumab plus ipilimumab was approved in March 2020 for treatment of HCC in patients previously treated with sorafenib [80]. The recommended doses are nivolumab 1 mg/kg followed by ipilimumab 3 mg/kg on the same day, every three weeks for four doses, then nivolumab alone (240 mg every two weeks or 480 mg every four weeks). Given the higher response rates and durability of responses, for patients with no worse than Child-Pugh class A cirrhosis who are being considered for immunotherapy and who are able to tolerate it, we prefer combined therapy with nivolumab plus ipilimumab over pembrolizumab alone.

Notably, there are no data on the efficacy of nivolumab alone or in combination with ipilimumab in patients previously treated with an anti-PD-L1-targeted therapy such as atezolizumab.

Pembrolizumab — In patients with advanced HCC who have progressed on sorafenib, pembrolizumab (a PD-1 inhibitor) is a reasonable alternative to nivolumab plus ipilimumab in patients who are unable to tolerate the potential toxicities of such combination immunotherapy.

The efficacy of pembrolizumab in advanced HCC refractory to sorafenib was initially demonstrated in a phase II trial (KEYNOTE-224) [81,82]. In two subsequent placebo-controlled phase III trials, pembrolizumab improved ORRs; it also improved OS and PFS in one of trials conducted in patients from Asia (KEYNOTE-394) [83,84]. Data are as follows:

●In an international, double-blind phase III trial (KEYNOTE-240), 413 patients with advanced HCC and Child-Pugh class A cirrhosis who had radiographic progression/intolerance of sorafenib were randomly assigned on a 2:1 basis to either pembrolizumab or placebo, in addition to best supportive care (BSC) [83]. At median follow-up of approximately 14 months, for pembrolizumab versus placebo, the differences in OS (median 14 versus 11 months, HR 0.78, 95% CI 0.61-0.998) and PFS (median 3 versus 2.8 months, HR 0.72, 95% CI 0.57-0.90) were not statistically significant because prespecified efficacy boundaries were not reached. However, pembrolizumab improved the ORR over placebo (18 versus 4 percent) and had more complete responders (six versus none). Treatment responses to pembrolizumab were also durable (median duration of response 14 months, range 2 to 24+ months).

●In another double-blind phase III trial (KEYNOTE-394), 413 patients with advanced, treatment-refractory HCC from Asia (mainland China, Hong Kong, the Republic of Korea, Malaysia, and Taiwan) were randomly assigned to either pembrolizumab or placebo, in addition to BSC [84]. Patients had Child-Pugh Class A cirrhosis and had either progressed on or were intolerant to sorafenib or oxaliplatin-based chemotherapy. At median follow-up of 34 months, compared with placebo, pembrolizumab improved PFS (median 2.6 versus 2.3 months, HR 0.74, 95% CI 0.60-0.92) and OS (median 15 versus 13 months, HR 0.79, 95% CI 0.63-0.99). Pembrolizumab also improved ORR over placebo (13 versus 1 percent), including six complete responders (2 percent). The grade ≥3 toxicity rate for pembrolizumab was 14 percent.

Based on data from KEYNOTE-394, pembrolizumab (table 6) was approved by the US Food and Drug Administration (FDA) for the treatment of patients with HCC secondary to hepatitis B who have received prior systemic therapy other than a PD-1/PD-L1 containing regimen [85].

Nivolumab monotherapy — The efficacy of nivolumab monotherapy was addressed in a phase I/II study (CheckMate 040) that included patients with advanced HCC and Child-Pugh class A or B (total score of 7 or less (table 1)) cirrhosis whose disease had progressed on sorafenib or who refused or were intolerant of the drug [86]. The combined report included 48 patients treated in the dose-escalation part of the study and 214 in the expansion cohort. In the dose-escalation part, nivolumab was administered IV every two weeks at doses from 0.1 to 10 mg/kg for up to two years; although a maximally tolerated dose was not reached, the expansion cohort was treated at a dose of 3 mg/kg (as long as not HBV infected). Overall, 68 percent of the expansion cohort had received sorafenib previously. Forty-nine of the 255 patients assessable for response had an objective antitumor response to nivolumab (15 percent of the escalated-dose cohort and 20 percent of the expansion cohort), six of which were complete. An additional 50 percent had stable disease. In the dose-escalation group, the median duration of response was 17 months (95% CI 6-24), and the median OS was 15 months. In the expansion cohort, the median duration of response was 9.9 months, and the data were insufficiently mature for calculation of median survival; however, 74 percent of patients remained alive at nine months.

Most adverse events were mild and transient; the most common grade 3 or 4 toxicities in the 48 patients in the dose-escalation phase were increased aspartate aminotransferase (10 percent), increased alanine aminotransferase (3 percent), increased lipase (3 percent), and increased amylase (2 percent). Immune-mediated hepatitis requiring systemic glucocorticoids occurred in 5 percent of treated patients. Toxicity was similar in the expansion cohort.

In a later analysis of the entire cohort, the durable benefits of nivolumab were observed both in sorafenib-naïve (objective response rate 23 percent, with 44 percent of responses ongoing) and sorafenib-experienced patients (objective response rate 16 to 19 percent) [87].

Although the FDA initially granted accelerated approval to nivolumab for treatment of HCC in patients previously treated with sorafenib in 2017, this approval was withdrawn in 2021 because of the negative findings of the phase III trial directly comparing nivolumab versus sorafenib as first-line therapy (see 'Nivolumab and pembrolizumab' above). The European Medicines Agency has not approved nivolumab monotherapy in this setting [88].

Avelumab — Avelumab is an anti-PD-L1 monoclonal antibody that is approved for advanced urothelial, renal cell, and Merkel cell cancer. Efficacy in 30 patients with advanced HCC and no worse than Child-Pugh class A cirrhosis who were previously treated with sorafenib was shown in a phase II trial, in which there were three partial responses (10 percent) and 19 patients (63 percent) with prolonged stable disease [89]. Tumoral overexpression of PD-L1 did not affect the antitumor response. The median time to tumor progression and OS durations were 4.4 and 14.2 months, respectively. Treatment was well tolerated with few grade 3 adverse events.

Toxicity considerations and response assessment — Notably, although patients with HCC who are treated with immune checkpoint inhibitors have a substantial increase in transaminases as compared with patients receiving these drugs for other indications (eg, lung cancer, melanoma), this has not translated into premature treatment discontinuation or treatment-related mortality [77,90]. Nevertheless, liver function tests should be monitored during therapy with these agents. (See "Hepatic, pancreatic, and rare gastrointestinal complications of immune checkpoint inhibitor therapy", section on 'Hepatotoxicity'.)

The patterns of response to treatment with immunotherapy differ from those with molecularly targeted agents or cytotoxic chemotherapy. Patients may appear to have a transient worsening of disease, manifested by either progression of known lesions or the appearance of new lesions, before disease stabilizes or regresses. In addition, responses can take appreciably longer to become apparent compared with cytotoxic therapy. (See 'Immunotherapy' above.)

Immune-related response criteria have been proposed to properly recognize the nontraditional patterns of response occasionally seen with checkpoint inhibitors and some other immunotherapies (table 7) [91]. This subject is discussed in more detail elsewhere. (See "Principles of cancer immunotherapy", section on 'Immunotherapy response criteria'.)

Tyrosine kinase inhibitors

Regorafenib — For patients progressing on sorafenib, regorafenib has been shown to improve OS over best supportive care alone. We consider regorafenib to be an alternative to nivolumab/ipilimumab or pembrolizumab monotherapy to second-line treatment after progression on sorafenib for patients who seemed to benefit from sorafenib, who maintain a good performance status and adequate liver function, and who are willing to trade treatment-related morbidity for the possibility of a small gain in median OS. However, for most patients we prefer immunotherapy. We would not consider regorafenib an appropriate option for patients who were intolerant of sorafenib. Regorafenib is also an option for patients progressing after first-line immunotherapy.

Regorafenib is an orally active inhibitor of angiogenic (including VEGFR-1, VEGFR-2, and VEGFR-3), stromal, and oncogenic receptor tyrosine kinases. It is structurally similar to sorafenib and targets a variety of kinases implicated in angiogenic and tumor growth-promoting pathways. A benefit for regorafenib in patients progressing after first-line treatment with sorafenib was suggested in the RESORCE trial, in which 573 patients who received sorafenib for at least 20 days at a dose of at least 400 mg daily, who had radiologic progression, and who maintained a performance status of 0 to 1 (table 5) and Child-Pugh class A liver function (table 1) were randomly assigned to regorafenib (160 mg once daily for three weeks on and one week off) or placebo [92]. Regorafenib was associated with significant prolongation in median OS (10.6 versus 7.8 months, HR for death 0.63), as well as significantly higher rates of objective antitumor response (11 versus 4 percent, none complete) and disease control (objective response plus stable disease; 65 versus 36 percent). Response assessment utilized modified RECIST for HCC. (See 'Response assessment' above.)

Treatment was relatively well tolerated; among the grade 3 or 4 adverse events occurring more often with regorafenib were hypertension (15 versus 5 percent), hand-foot skin reaction (13 versus 1 percent), fatigue (9 versus 5 percent), and diarrhea (3 versus 0 percent). Sixty-eight percent of patients treated with regorafenib required dose modification for adverse events (compared with 31 percent of the placebo group).

Largely based on these data, in April 2017, the FDA expanded the indications for regorafenib to include patients with HCC who had been previously treated with sorafenib.

Similar to sorafenib, we start at lower-than-recommended doses of regorafenib (typically 80 mg once daily) and escalate the dose as tolerated. Consensus-based guidelines from the NCCN recommend considering regorafenib only for patients with no worse than Child-Pugh class A cirrhosis.

Cabozantinib — For patients previously treated with sorafenib, cabozantinib has been shown to improve OS compared with supportive care alone. We consider cabozantinib to be an alternative to nivolumab/ipilimumab or pembrolizumab monotherapy for second-line treatment after progression on sorafenib. However, for most patients we prefer immunotherapy. Cabozantinib is also an option for patients progressing after first-line immunotherapy.

Cabozantinib is another potent inhibitor of several receptor tyrosine kinases, including the hepatocyte growth factor/c-MET, VEGFR-1, VEGFR-2, and VEGFR-3 [93]. High levels of MET expression have been associated with previous sorafenib treatment in patients with HCC, and with sorafenib resistance in preclinical models [94-96].

Efficacy of cabozantinib in patients with previously treated advanced HCC was shown in the phase III CELESTIAL trial, in which 707 patients with advanced and progressing HCC and no worse than Child-Pugh class A cirrhosis were randomly assigned to cabozantinib or placebo [97]. In the population of patients receiving second- or third-line treatment after prior treatment with sorafenib, median OS was significantly better with cabozantinib (10.2 versus 8.0 months), and the difference was more pronounced when the analysis was limited to patients whose only prior therapy was sorafenib (median OS 11.3 versus 7.2 months). In a later analysis, cabozantinib improved outcomes versus placebo across a range of baseline AFP levels [98].

The most common grade 3 or 4 adverse events with cabozantinib were palmar-plantar erythrodysesthesia (17 versus 0 percent in the placebo group), hypertension (16 versus 2 percent), increased aspartate aminotransferase (12 versus 7 percent), fatigue (10 versus 4 percent), and diarrhea (10 versus 2 percent) [97].

Based on these data, cabozantinib was approved in January 2019 for treatment of patients with HCC who have been previously treated with sorafenib. Consensus-based guidelines from the NCCN recommend considering cabozantinib only for patients with no worse than Child-Pugh class A cirrhosis, although safety and efficacy has been suggested in patients with albumin-bilirubin grade 1 or 2 hepatic dysfunction (calculator 1) [99]. (See 'Estimate life expectancy and assess liver function' above and "Staging and prognostic factors in hepatocellular carcinoma", section on 'Albumin-bilirubin (ALBI) score'.)

Ramucirumab — Ramucirumab is an appropriate second-line option after failure of sorafenib or front-line immunotherapy, but benefit appears limited to those with initially high AFP levels.

Ramucirumab is a recombinant monoclonal antibody of the immunoglobulin G subclass 1 (IgG1) class that binds to VEGFR-2, blocking receptor activation. A modest degree of activity in patients with no prior systemic treatment was suggested in a phase II trial of 42 patients; the objective response rate was 10 percent, and median OS was 12 months [100]. However, in the absence of a trial directly comparing ramucirumab with sorafenib, first-line ramucirumab cannot be considered a standard approach.

On the other hand, ramucirumab is an appropriate second-line option after failure of sorafenib, but benefit appears limited to those with initially high AFP levels:

●A trial of ramucirumab versus placebo in patients with advanced HCC following first-line therapy with sorafenib (the REACH trial) failed to show a significant survival advantage relative to placebo (median OS 9.2 versus 7.6 months) [101]. An unplanned subset analysis suggested the potential for a survival benefit in patients with a high initial level of AFP (>400 ng/mL) at diagnosis (median survival 7.8 versus 4.2 months).

●Benefit in this subgroup was confirmed in a randomized phase III trial (the REACH-2 trial), which randomly assigned 292 patients with HCC, no worse than Child-Pugh class A cirrhosis, and a serum AFP ≥400 ng/mL who had disease progression on first-line sorafenib to ramucirumab versus placebo [102]. Ramucirumab was associated with significantly better OS (8.5 versus 7.3 months, HR 0.71, 95% CI 0.53-0.95), although the magnitude of benefit (1.2 months) was numerically less than that seen in the subset of patients with a high AFP enrolled in the REACH trial (3.6 months). Ramucirumab also was associated with a higher objective response rate (5 versus 1 percent) and overall disease control rate (60 versus 39 percent). One potential advantage of ramucirumab over other molecularly targeted treatments for second-line therapy after progression on sorafenib is the absence of hand-foot skin reaction.

Largely based on this trial, ramucirumab was approved in May 2019 for second-line treatment of HCC in patients who have an AFP level ≥400 ng/mL and have been previously treated with sorafenib [103].

Apatinib — Apatinib is an orally active VEGFR-2 inhibitor that is approved for second-line treatment of advanced gastric cancer in China; it is not available in the United States or Europe.

Efficacy for second-line treatment of advanced HCC after failure of sorafenib and oxaliplatin-based chemotherapy was shown in a phase III randomized placebo-controlled trial of 393 patients with no worse than Child-Pugh class A or B (≤7) cirrhosis [104]. Apatinib improved OS (median 8.7 versus 6.8 months HR 0.785, 95% CI 0.617-0.998) and PFS, and was well tolerated.

LESS FIT PATIENTS AND THOSE WITH CHILD-PUGH B CIRRHOSIS — For patients who can tolerate it, we suggest sorafenib monotherapy rather than systemic chemotherapy as long as the total Child-Pugh score is <7 (table 1). Another option for individuals with no worse than Child-Pugh B7 cirrhosis is nivolumab.

The best way to manage systemic therapy in patients with Child-Pugh class B cirrhosis is not established. The vast majority of studies of immunotherapy or molecularly targeted therapy have been carried out in populations with preserved liver function (ie, no worse than Child-Pugh class A cirrhosis). There are few data on the safety and efficacy of any of these agents in patients with less well preserved liver function [105-108]. The majority of the published experience in this group is with sorafenib.

Sorafenib — The patients enrolled in the original first-line sorafenib trials had predominantly (over 95 percent) Child-Pugh class A cirrhosis. Despite the specific population studied in the SHARP trial, the original US Food and Drug Administration approval of sorafenib for HCC in the United States did not specify the underlying cirrhosis state. (See 'Sorafenib' above.)

However, the majority of patients with HCC have underlying cirrhosis and, therefore, competing causes of death from progressive HCC versus worsening cirrhosis. In a systematic review of 118 studies of patients with cirrhosis and no HCC, the one- and two-year survival rates for patients with Child-Pugh class A, B, or C cirrhosis were 95 and 90, 80 and 70, and 45 and 38 percent, respectively [109]. Death from cirrhosis could potentially mask treatment-related antitumor effect.

Furthermore, there are only limited clinical experience and few published data regarding the safety and efficacy of sorafenib in patients with Child-Pugh class B or more severe cirrhosis [110-113]:

●The safety and efficacy of sorafenib in patients with liver dysfunction were addressed in an analysis of 1586 patients receiving sorafenib and registered in the international GIDEON study prospective database [110]. The safety profile of sorafenib was similar in patients with Child-Pugh class A and B cirrhosis, but a greater percentage of Child-Pugh class B patients had serious adverse events (60 versus 33 percent), a greater number discontinued therapy due to adverse effects (40 versus 25 percent), and they had a higher rate of deaths during treatment up to 30 days from the last sorafenib dose (37 versus 18 percent).

The analysis also highlighted interspecialty differences in treating HCC with sorafenib. In North America, where oncologists prescribe sorafenib, patients often have worse cirrhosis as compared with other areas in the world where hepatologists and gastroenterologists are the prescribers. Despite this variation in treatment practice, toxicity rates (hand-foot syndrome, hematologic toxicities) were similar among all three specialties.

●The safety and efficacy of first-line sorafenib in patients with Child-Pugh class B HCC were addressed in a meta-analysis of 30 studies comprising 8678 patients, 79 percent with Child-Pugh class A cirrhosis and 19 percent with Child-Pugh class B cirrhosis [113]. While the median overall survival (OS) on sorafenib was less for those with Child-Pugh class B cirrhosis (4.6 versus 8.8 months), in contrast to the GIDEON analysis, both groups had similar rates of grade 3 or 4 adverse events during therapy and similar rates of treatment discontinuation without progression.

Underlying cirrhosis also makes it more difficult to clear drugs for which there is significant hepatic metabolism, such as sorafenib. There is controversy as to whether the initial dose of sorafenib needs to be modified in patients with hepatic dysfunction, as evidenced by the following reports:

●The safety of sorafenib in patients with elevated transaminase levels was studied in a subgroup analysis of data from the SHARP trial [114]. Patients with mild or moderate liver dysfunction (≥1.8 times upper limit of normal [ULN]) did not experience increased hepatic or other toxicity. Although median OS was diminished in both placebo-treated and sorafenib-treated patients with moderate liver dysfunction, median TTP and OS favored the sorafenib group regardless of transaminase levels. The authors concluded that treatment was safe and effective, even in patients with mild or moderately elevated baseline transaminase levels, and that hepatic function remained stable over the course of sorafenib therapy.

●On the other hand, results from a phase I study suggest that dose reduction to 200 mg twice a day is required in patients with a bilirubin 1.5 to 3 times ULN and that the drug cannot be tolerated with more severe hyperbilirubinemia. (See "Chemotherapy hepatotoxicity and dose modification in patients with liver disease: Molecularly targeted agents", section on 'Sorafenib'.)

Dosing considerations — Although the manufacturer recommends no sorafenib dose adjustment for Child-Pugh class B impairment, we recommend reduced-dose therapy (200 mg twice daily, at least initially) in patients with a total bilirubin 1.5 to 3 times the ULN and that the drug not be administered to patients with more severe degrees of hyperbilirubinemia.

For other patients with Child-Pugh class B cirrhosis, standard dosing from the onset with dose modification as needed is appropriate. However, to improve early tolerability, many clinicians start all patients (regardless of baseline hepatic function) at 200 mg twice a day and increase the daily dose in 200 mg increments approximately every five days, if tolerability is maintained, until the target dose is reached. The available evidence suggests that this approach results in lower rates of adverse events and drug discontinuation, while preserving efficacy [115,116]. Consensus-based guidelines from the National Comprehensive Cancer Network suggest that patients with Child-Pugh class B disease and a total score >7 are not good candidates for sorafenib (table 1) [117].

Nivolumab — Nivolumab monotherapy is an alternative to sorafenib for those with no worse that Child-Pugh B7 cirrhosis who have not undergone liver transplantation. (See 'Nivolumab monotherapy' above.)

Systemic chemotherapy — For patients who are not candidates for, or intolerant of sorafenib, who are not candidates for nivolumab and who are ineligible for clinical trials or if they are not available, we suggest chemotherapy with GEMOX for fit patients who can tolerate combination chemotherapy. Oral capecitabine or leucovorin-modulated fluorouracil are options for sick or jaundiced individuals. We choose the regimen based on the level of fitness and hepatic reserve.

The efficacy of cytotoxic chemotherapy is modest in patients with HCC, and in general, the duration of benefit is limited. No single regimen has emerged as superior to any other, although few randomized trials have been conducted. Despite objective responses that are occasionally complete, median survival in all studies has been short (<12 months in all cases).

The role of cytotoxic chemotherapy has diminished with the advent of newer immunotherapy and molecularly targeted therapy approaches, which demonstrated better efficacy and tolerability. Cytotoxic chemotherapy has been removed from guidelines for advanced HCC by both the NCCN and ESMO [3,4]. However, at some institutions, chemotherapy is still offered to select patients for third- or fourth-line therapy if they are not optimal candidates for other approved therapies.

The side effect profile of any chemotherapy regimen should be considered carefully in patients with advanced liver disease and a short life expectancy.

Patients suitable for multiagent chemotherapy — Multiple combination cytotoxic regimens have been tested in patients with advanced HCC. Most are of historical interest only. In practice, we most often use GEMOX for patients who are sufficiently fit to tolerate combination chemotherapy.

Gemcitabine based

●GEMOX is an active regimen that lacks substantial renal and hepatic toxicity [118-120]. In a phase II study involving 32 cirrhotics with previously untreated advanced HCC (lung metastases in 10, multifocal liver disease in 18, the rest node positive), gemcitabine (1000 mg/m² by fixed dose rate infusion) on day 1 was followed by oxaliplatin (100 mg/m²) on day 2, with both drugs repeated every two weeks [118]. Grade 3 or 4 toxicities included thrombocytopenia (27 percent), neutropenia (24 percent, with two cases of febrile neutropenia), anemia (9 percent), and neuropathy (9 percent). The objective response rate was 18 percent, and an additional 58 percent had disease stabilization. Median survival was 11.5 months. For unclear reasons, treatment seemed to be more effective in patients with nonalcoholic rather than alcoholic cirrhosis.

●In another phase II study, the combination of gemcitabine (1250 mg/m² on days 1 and 8) and cisplatin (70 mg/m² on day 1 of every 21-day cycle) was associated with an overall response rate of 20 percent [121]. Grade 3 to 4 toxicities included anemia (37 percent), neutropenia (13 percent), thrombocytopenia (7 percent), transaminitis (3 percent), and mucositis (3 percent). A second trial using a slightly different dosing regimen (cisplatin 25 mg/m² on days 1 and 8; gemcitabine 1000 mg/m² on days 1 and 8) reported a more favorable toxicity profile but a lower response rate (one partial response among 15 patients) [122].

●PLD in combination with gemcitabine is modestly active. In a phase II trial, 41 patients received gemcitabine (1000 mg/m² on days 1 and 8) plus PLD (30 mg/m² on day 1) every 28 days [123]. There were three complete and seven partial responses (overall response rate 24 percent), median TTP was 5.8 months, and median OS was 22.5 months. Treatment was well tolerated, with grade 3 to 4 toxicity limited to neutropenia (17 percent) and thrombocytopenia (2 percent).

FOLFOX — Regimens containing oxaliplatin plus short-term infusional fluorouracil and leucovorin (FOLFOX) are most commonly used in the treatment of advanced colorectal cancer. (See "Systemic therapy for nonoperable metastatic colorectal cancer: Selecting the initial therapeutic approach", section on 'FOLFOX versus FOLFIRI'.)

Modified FOLFOX 4 (table 8) was directly compared with single-agent doxorubicin (50 mg/m² every three weeks) in an Asian trial of 371 patients with advanced or metastatic HCC, 90 percent of whom had HBV infection as the underlying etiology of cirrhosis [124]. Although median progression-free survival (PFS) was significantly better with FOLFOX, it was still quite short (2.93 versus 1.77 months), and there was only a trend toward better median survival (6.4 versus 4.97 months, p = 0.07). FOLFOX was also associated with a higher objective response rate (8 versus 3 percent) and disease control rate (partial response plus stable disease; 52 versus 32 percent). Although the rate of sensory neuropathy was higher in the FOLFOX group (15 versus 0.6 percent), most cases were mild, and there were no significant differences in the rate of grade 3 or 4 toxicities (either hematologic or nonhematologic) between the groups. (See 'Doxorubicin and mitoxantrone' below.)

Interpretation of these results is hampered by the use of a relatively low dose of doxorubicin in the control arm. Furthermore, whether these results can be extrapolated to American patients who have alcohol or HCV, rather than HBV, as the underlying cause of cirrhosis is unclear.

Cisplatin based — Cisplatin-based combination regimens appear to result in higher objective response rates (ORRS) than non-cisplatin-containing regimens, although it is not clear that any regimen confers a survival benefit. Response rates with several two- and three-drug regimens are approximately the same:

●Cisplatin plus doxorubicin – Response rate 18 and 49 percent in two studies, one conducted in children (with predominantly HCC rather than the more chemosensitive hepatoblastoma) [125,126].

●Cisplatin, mitoxantrone, and continuous-infusion fluorouracil – Objective response rate 24 and 27 percent in two different studies [127,128].

●Cisplatin, epirubicin, and infusional fluorouracil – Response rate 15 percent [129].

●Cisplatin, doxorubicin, and capecitabine – Response rate 24 percent [130].

●Sequential low-dose infusional cisplatin plus infusional fluorouracil – Response rate 43 percent [131].

●Cisplatin plus capecitabine – Response rate 6 and 20 percent in two studies [132,133].

In our view, these cisplatin-based combinations are largely of historic interest only.

Less fit or jaundiced patients — For less fit or jaundiced individuals who might not be able to tolerate combination chemotherapy, capecitabine monotherapy is a reasonable option.

Fluoropyrimidines — Fluorouracil has acceptably low toxicity and broad antitumor efficacy. Although there is extensive hepatic metabolism, adequate doses can usually be administered in the setting of hepatic dysfunction or jaundice.

Response rates with fluorouracil monotherapy have been low. However, when given in combination with leucovorin, response rates as high as 28 percent have been reported [134], although lower rates have been noted by others [135].

The benefit of single-agent treatment with the orally active fluoropyrimidine capecitabine has been addressed in the following reports:

●Capecitabine monotherapy (1000 mg/m² twice daily for 14 of every 21 days) was associated with an objective response in 9 of 37 patients (25 percent) in one study, including one complete response [136]. Of the 37 patients, 22 had received no prior therapy for advanced disease.

●Others, using a metronomic capecitabine schedule (500 mg twice daily every day), also showed better activity for capecitabine monotherapy in patients with previously untreated disease [137]. Among 59 previously untreated patients, there were two complete responses, one partial response, and 30 with stable disease; in a second cohort of 31 patents previously treated with sorafenib, there were no objective responses, and stable disease was the best response in 10 patients.

However, the superiority of sorafenib over capecitabine monotherapy as initial therapy was shown in a randomized phase II trial of 52 previously untreated patients with advanced HCC [138]. Both median OS (five versus seven months) and median PFS (four versus six months) were inferior in the capecitabine arm. (See 'Sorafenib' above.)

Nevertheless, for patients who are unable to obtain treatment with molecularly targeted agents, such as sorafenib, fluoropyrimidine monotherapy is an option.

Doxorubicin and mitoxantrone — Historically, single-agent doxorubicin has been the most studied chemotherapy agent for advanced HCC. Although an early trial in 1975 reported dramatic clinical activity and a 79 percent response rate, subsequent work suggests that the true objective response rate with doxorubicin monotherapy is 20 percent or less with doses of 75 mg/m² [139-145]. Lower doses (≤60 mg/m² per cycle) are associated with even lower ORRs [146,147].

Despite the modest objective response rate, at least two controlled trials, involving 106 and 445 patients, respectively, suggest that doxorubicin is associated with a small survival advantage compared with either best supportive care alone (median survival 10.6 versus 7.5 weeks) [143] or nolatrexed (median survival 32 versus 22 weeks) [147]. The reason for the disparate survival outcomes in these two trials is unclear, but patient selection may have played a role.

Both epirubicin and mitoxantrone have an approximately similar level of antitumor efficacy compared with doxorubicin (response rate 10 to 25 percent) [148-151]. By contrast, the single-agent activity of pegylated liposomal doxorubicin (PLD) is limited [152-154]. Combination regimens containing PLD are discussed above. (See 'Gemcitabine based' above.)

Anthracyclines require dose modification in patients with liver dysfunction. (See "Chemotherapy hepatotoxicity and dose modification in patients with liver disease: Conventional cytotoxic agents", section on 'Anthracyclines'.)

CHILD-PUGH CLASS C CIRRHOSIS OR EXTENSIVE COMORBIDITY — Supportive care alone is appropriate for patients with Child-Pugh class C cirrhosis (table 1) and for those with a poor functional status or extensive comorbidity.

The majority of patients with Child-Pugh class C cirrhosis have liver function that is insufficient to tolerate systemic therapy, and systemic therapy is unlikely to benefit them because of their short survival [110,155]. As an example, in a retrospective report of experience with first-line sorafenib in 59 patients with HCC (26 with Child-Pugh class A, 23 with Child-Pugh class B, and 10 with Child-Pugh class C cirrhosis), the median survival times in the three groups were 8.3, 4.3, and 1.5 months, respectively [111]. These data support the view that sorafenib is unlikely to benefit patients with Child-Pugh class C cirrhosis.

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Hepatocellular carcinoma".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topic (see "Patient education: Liver cancer (The Basics)")

SUMMARY AND RECOMMENDATIONS

●Indications for systemic therapy – Systemic therapy is appropriate for patients with advanced unresectable HCC who are unsuitable for locoregional therapy and have adequate performance status and underlying liver function. Clinical trials are preferred whenever available. (See 'Indications for systemic therapy' above.)

●Importance of assessing hepatic reserve – HCC is an aggressive tumor often occurring in the setting of chronic liver disease and cirrhosis. Treatment options are influenced by Child-Pugh class (table 1), a measure of hepatic reserve. (See 'General considerations' above.)

We limit atezolizumab plus bevacizumab, lenvatinib, regorafenib, cabozantinib, and pembrolizumab to those with no worse than Child-Pugh class A cirrhosis, and do not offer sorafenib to patients with Child-Pugh class B disease and a total score >7. Nivolumab can be administered to patients with Child-Pugh class A or B disease.

●No worse than class A cirrhosis and good functional status

•Initial therapy – For patients whose functional status and liver function are adequate (ie, no worse than Child-Pugh class A cirrhosis (table 1), have no contraindications to bevacizumab, and have not recurred following liver transplantation, we suggest atezolizumab plus bevacizumab rather than other systemic agents (Grade 2C). (See 'Atezolizumab plus bevacizumab' above.)

For patients unable to receive bevacizumab, tremelimumab plus durvalumab is an alternative to atezolizumab plus bevacizumab. (See 'Tremelimumab plus durvalumab' above.)

Immunotherapy should be avoided in patients who recur following liver transplantation because of the high rate of allograft rejection. (See 'Checkpoint inhibitor immunotherapy' above.)

For patients who are ineligible for or lack access to atezolizumab plus bevacizumab or durvalumab plus tremelimumab, we prefer lenvatinib over sorafenib. Durvalumab or donafenib (where available) monotherapy are reasonable alternatives to sorafenib in this setting. (See 'Lenvatinib' above and 'Sorafenib' above and 'Durvalumab' above.)

•Subsequent therapy – We offer second-line therapy to patients whose tumors progress on first-line therapy or who are unable to tolerate the first-line treatment, and who maintain adequate performance status and preserved liver function. The optimal way to sequence available treatments is not established. We prioritize clinical trials. (See 'Second-line therapy' above.)

-Treatment of patients who have progressed on atezolizumab plus bevacizumab or durvalumab plus tremelimumab must be individualized as data are limited. Options include sorafenib, lenvatinib, regorafenib, cabozantinib, or apatinib, where available.

-For most patients who have progressed on or are unable to tolerate sorafenib or lenvatinib, we suggest nivolumab plus ipilimumab (table 9) rather than other immunotherapy options or a second tyrosine kinase inhibitor (Grade 2C). However, this regimen should not be administered to patients with Child-Pugh class B or C. (See 'Checkpoint inhibitor immunotherapy' above and 'Nivolumab plus ipilimumab' above.)

Pembrolizumab (table 6) is an appropriate alternative for those who are unable to tolerate the potential toxicities of nivolumab plus ipilimumab. (See 'Pembrolizumab' above.)

Another option following treatment with lenvatinib or sorafenib is to switch to the other agent or to regorafenib. However, whether lenvatinib will work after failure of sorafenib is not known; there are no trials to inform this issue. (See 'Lenvatinib' above.)

-Cabozantinib is an option for patients who are intolerant of sorafenib, lenvatinib, or regorafenib. (See 'Cabozantinib' above.)

-Ramucirumab may be preferred for the subset of patients with a high AFP level (>400 ng/mL). (See 'Ramucirumab' above.)

-If access to targeted therapies is limited, systemic chemotherapy remains an option for later lines of therapy. Options for fit patients include gemcitabine plus oxaliplatin, gemcitabine plus pegylated liposomal doxorubicin, or capecitabine monotherapy. (See 'Systemic chemotherapy' above.)

●Less fit patients or class B cirrhosis

•For less fit patients who have no worse than Child-Pugh B7 cirrhosis (table 1), we suggest monotherapy with sorafenib or nivolumab rather than cytotoxic chemotherapy (Grade 2C).

•For relatively fit patients with Child-Pugh class B cirrhosis who can tolerate combination chemotherapy and are not candidates for, or intolerant of sorafenib or nivolumab, and for whom clinical trial participation is not feasible, we suggest chemotherapy with GEMOX rather than an alternative systemic therapy regimen (Grade 2C). (See 'Systemic chemotherapy' above.)

•Oral capecitabine or leucovorin-modulated fluorouracil are options for sick or jaundiced individuals. (See 'Less fit or jaundiced patients' above.)

●Class C cirrhosis or extensive comorbidity – Supportive care alone is appropriate for those with Child-Pugh class C cirrhosis, a poor functional status, or extensive comorbidity. (See 'Child-Pugh class C cirrhosis or extensive comorbidity' above.)