Abstract
Background & Aims
We performed a meta-analysis of data from the KEYNOTE-240 and KEYNOTE-394 studies to obtain a more precise estimate of the pembrolizumab treatment effect in participants with previously treated advanced hepatocellular carcinoma (HCC).
Methods
Participants with confirmed HCC and disease progression after treatment with or intolerance of sorafenib or oxaliplatin-based chemotherapy (KEYNOTE-394 only), Barcelona Clinic Liver Cancer stage C or B disease not amenable to or refractory to locoregional therapy, and one or more measurable lesion per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 were randomly assigned 2:1 to receive pembrolizumab or placebo for ≤35 cycles. Data from the KEYNOTE-240 and KEYNOTE-394 intention-to-treat populations were pooled, and the treatment effect was evaluated for pembrolizumab and placebo separately.
Results
In total, 578 and 288 participants who received pembrolizumab and placebo, respectively, were included in this analysis. Compared with placebo, pembrolizumab improved overall survival (hazard ratio 0.79, 95% CI 0.67–0.93), progression-free survival (per RECIST v1.1 by blinded independent central review [BICR]; hazard ratio 0.76, 95% CI 0.64–0.89), and objective response rate (per RECIST v1.1 by BICR; 15.4% vs. 2.8%, for an estimated treatment difference of 12.5%; 95% CI 8.8–16.2). Subgroup analyses showed that the treatment effect of pembrolizumab was generally similar across baseline participant characteristics, including viral status, Barcelona Clinic Liver Cancer stage, and geographic region.
Conclusions
Meta-analysis of KEYNOTE-240 and KEYNOTE-394 showed that pembrolizumab provides clinically meaningful improvement in overall survival, progression-free survival, and objective response rate. This analysis expands on findings from each study individually and provides further evidence of the global benefit of pembrolizumab as second-line therapy for advanced HCC after prior sorafenib- or oxaliplatin-based therapy.
Impact and implications
To obtain a more precise estimate of the pembrolizumab treatment effect in participants with previously treated advanced hepatocellular carcinoma, we performed a meta-analysis of efficacy using pooled participant data from the phase III KEYNOTE-240 and KEYNOTE-394 studies. Subgroup analyses showed that the treatment effect of pembrolizumab was generally similar across baseline characteristics, including viral status, Barcelona Clinic Liver Cancer stage, and geographic region. This meta-analysis provides further evidence of the global benefit of pembrolizumab as second-line therapy for advanced hepatocellular carcinoma.
Clinical Trials Registration
Registered at ClinicalTrials.gov NCT02702401 (KEYNOTE-240) and NCT03062358 (KEYNOTE-394).
Keywords: Programmed cell death ligand 1, Randomized, Pooled, Post hoc, Global
Graphical abstract
Highlights
-
•
Pembrolizumab has shown antitumor activity in HCC in KEYNOTE-240 and KEYNOTE-394.
-
•
Compared with placebo, pembrolizumab improved OS, PFS, and ORR in both studies.
-
•
The treatment effect of pembrolizumab was similar across patient subgroups.
-
•
Pembrolizumab continues to provide benefit as second-line therapy for advanced HCC.
Introduction
Primary liver cancer is a leading cause of cancer-related mortality, with an estimated 830,000 deaths globally in 2020.1 The global annual incidence of primary liver cancer is rising and is predicted to exceed 1 million cases by 2025.2 Hepatocellular carcinoma (HCC) accounts for approximately 90% of all cases of primary liver cancer, and most patients are diagnosed with advanced-stage disease not amenable to curative treatment approaches.[1], [2], [3] First-line treatment options for advanced HCC have expanded from only the multikinase inhibitors sorafenib and lenvatinib, now to the combination of the programmed cell death ligand 1 inhibitor atezolizumab and vascular endothelial growth factor (VEGF) inhibitor bevacizumab as the preferred first-line therapy for most patients with advanced HCC.2,4,5 In addition, durvalumab and tremelimumab are now approved in the first-line setting.6 In the second-line setting, the multikinase inhibitors regorafenib and cabozantinib and the VEGF receptor antibody ramucirumab (for patients with α-fetoprotein levels >400 ng/dl) are approved based on results of phase III studies.2,5 These agents have been shown to improve survival by slowing progression but have demonstrated single-digit objective response rates (ORRs).2 Despite these recent advances, treatment availability differs across the globe, and a high unmet need exists for second-line treatment options that prolong survival, induce durable responses, and are tolerable.
Pembrolizumab received accelerated approval from the United States FDA in 2018 on the basis of the global, single-arm, open-label, phase II KEYNOTE-224 study in participants with advanced HCC and disease progression after treatment with sorafenib or intolerance of sorafenib treatment.7,8 This study demonstrated an ORR of 17% (18 of 104 participants), durable responses, and a manageable adverse event profile.8 A favorable benefit–risk profile for pembrolizumab compared with placebo was observed in the global, phase III KEYNOTE-240 study of participants with advanced HCC and disease progression after treatment with sorafenib or intolerance of sorafenib treatment; however, the study narrowly missed prespecified statistical significance criteria for overall survival (OS) and progression-free survival (PFS).9 Median OS at final analysis was 13.9 months with pembrolizumab vs. 10.6 months with placebo (hazard ratio [HR] 0.781, 95% CI 0.611–0.998, one-sided p = 0.0238 [prespecified p value boundary for OS superiority was 0.0174]), and median PFS at the first interim analysis, when PFS was prespecified to be tested, was 3.0 months with pembrolizumab vs. 2.8 months with placebo (HR 0.775, 95% CI 0.609–0.987, one-sided p = 0.0186 [prespecified p value boundary for PFS superiority was 0.002]).9 An ORR of 18.3% with pembrolizumab vs. 4.4% with placebo was observed at final analysis. The safety profile of pembrolizumab in KEYNOTE-240 was consistent with that observed in the KEYNOTE-224 study.8,9 Recently, pembrolizumab demonstrated statistically significant and clinically meaningful benefit in OS, PFS, and ORR compared with placebo in the KEYNOTE-394 study in participants from Asia with advanced HCC and disease progression during or after treatment with or intolerance of sorafenib- or oxaliplatin-based chemotherapy.10 At the final analysis, median OS was 14.6 months with pembrolizumab vs. 13.0 months with placebo (HR 0.79, 95% CI 0.63–0.99, one-sided p = 0.0180) and median PFS at the second interim analysis, when PFS was prespecified to be tested, was 2.6 months with pembrolizumab vs. 2.3 months with placebo (HR 0.74, 95% CI 0.60–0.92, one-sided p = 0.0032). Pembrolizumab was associated with a manageable adverse event profile, with adverse events consistent with the known safety profile of pembrolizumab in participants with previously treated advanced HCC.[8], [9], [10]
To obtain a more precise estimate of the pembrolizumab treatment effect in participants with previously treated advanced HCC, we performed a meta-analysis of efficacy using pooled participant data from the phase III KEYNOTE-240 and KEYNOTE-394 studies.
Participants and methods
KEYNOTE-240 and KEYNOTE-394 studies were conducted in accordance with principles of Good Clinical Practice and were approved by the appropriate institutional review boards and regulatory agencies. Written informed consent was provided by all participants before enrollment in each study.
Data sources and participant population
KEYNOTE-240 (ClinicalTrials.gov identifier: NCT02702401) was a global, randomized, double-blind, phase III study conducted in Asia, Australia, Europe, North America, and South America. KEYNOTE-394 (NCT03062358) was a randomized, double-blind, phase III study conducted in mainland China, Hong Kong, Republic of Korea, Malaysia, and Taiwan. The study design and full eligibility criteria for KEYNOTE-240 and KEYNOTE-394 have been published.9,10 Briefly, in both studies, eligible participants were aged ≥18 years and had confirmed advanced HCC, with radiographic disease progression during or after treatment with or intolerance of sorafenib or, in KEYNOTE-394 only, oxaliplatin-based chemotherapy and Barcelona Clinic Liver Cancer (BCLC) stage C or B11 disease not amenable to or refractory to locoregional therapy. Participants had Child-Pugh class A liver disease, an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, ≥1 measurable lesion confirmed by blinded independent central review (BICR; KEYNOTE-240) or investigator (KEYNOTE-394) per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, and otherwise adequate organ function. Participants who had main portal vein invasion or inferior vena cava or cardiac involvement of HCC based on imaging were excluded from both studies.
In both studies, participants were randomly assigned (2:1) to receive intravenous pembrolizumab 200 mg or placebo, both with best supportive care per the discretion of the investigator and in accordance with local practices. Randomization in KEYNOTE-240 was stratified by geographic region (Asia without Japan vs. non-Asia with Japan), macrovascular invasion (yes vs. no), and α-fetoprotein level (<200 ng/ml vs. ≥200 ng/ml). Randomization in KEYNOTE-394 was stratified by prior treatment (sorafenib vs. chemotherapy), macrovascular invasion (yes vs. no), and HCC etiology (HBV vs. other [HCV or uninfected]). Treatment was administered every 3 weeks for up to 35 cycles (∼2 years) or until unacceptable toxicity, disease progression, or withdrawal.
Outcomes and assessments
Outcomes of interest were OS, PFS, and ORR in the intention-to-treat (ITT) populations of KEYNOTE-240 and KEYNOTE-394; OS, PFS, and ORR across subgroups based on baseline demographic and disease characteristics; and OS adjusted for the use of poststudy anticancer treatment.
HBV infection was defined as HBV surface antigen-positive and/or detectable HBV DNA. HCV infection was defined as anti-HCV antibody-positive and detectable HCV RNA. OS was defined as the time from randomization to death attributable to any cause. PFS was defined as the time from randomization to the first documented disease progression per RECIST v1.1 assessed by BICR or death attributable to any cause, whichever occurred first. ORR was defined as the proportion of the participants in the analysis population who had a confirmed complete or partial response per RECIST v1.1 assessed by BICR. In each study, tumor imaging was performed every 6 weeks, and survival was assessed every 12 weeks.
Statistical analysis
The analysis plan for the meta-analysis was completed before KEYNOTE-394 was unblinded and data from the individual studies were merged. Efficacy was assessed using pooled data from the KEYNOTE-240 and KEYNOTE-394 ITT populations, with the treatment effect evaluated using meta-analysis of pooled participant data for the pembrolizumab and placebo arms separately. Median OS and PFS with 95% CIs were estimated based on the product-limit (Kaplan–Meier) method for censored data. A stratified Cox proportional hazards model using the Efron method for handling ties, with treatment as a single covariate, was used to estimate the OS and PFS HRs and their 95% CIs. Sensitivity analyses to adjust for the effect of poststudy anticancer therapies on OS were performed using the inverse probability of censoring weighting (IPCW) model12 and the simplified two-stage survival model without recensoring.13 The stratified Miettinen and Nurminen method with weights proportional to the stratum size was used to compare ORR between the pembrolizumab and placebo arms. Stratified analyses for the ITT populations were stratified by the protocol and, within each protocol, by strata used in the analyses of the respective protocol, corresponding to a one-stage approach.14 Specifically, the following 10 strata were used in the stratified analyses: (1) protocol KEYNOTE-240 plus macrovascular invasion (yes); (2) protocol KEYNOTE-240 plus macrovascular invasion (no) plus geographic region (Asia without Japan) plus α-fetoprotein (˂200 ng/ml); (3) protocol KEYNOTE-240 plus macrovascular invasion (no) plus geographic region (non-Asia with Japan) plus α-fetoprotein (˂200 ng/ml); (4) protocol KEYNOTE-240 plus macrovascular invasion (no) plus geographic region (Asia without Japan) plus α-fetoprotein (≥200 ng/ml); (5) protocol KEYNOTE-240 plus macrovascular invasion (no) plus geographic region (non-Asia with Japan) plus α-fetoprotein (≥200 ng/ml); (6) protocol KEYNOTE-394 plus macrovascular invasion (yes); (7) protocol KEYNOTE-394 plus macrovascular invasion (no) plus region (China) plus α-fetoprotein (˂200 ng/ml); (8) protocol KEYNOTE-394 plus macrovascular invasion (no) plus region (ex-China) plus α-fetoprotein (˂200 ng/ml); (9) protocol KEYNOTE-394 plus macrovascular invasion (no) plus region (China) plus α-fetoprotein (≥200 ng/ml); and (10) protocol KEYNOTE-394 plus macrovascular invasion (no) plus region (ex-China) plus α-fetoprotein (≥200 ng/ml). For subgroup analysis by protocol (i.e. KEYNOTE-240 and KEYNOTE-394 as two subgroups in the overall pooled ITT population), the analysis was stratified by the stratification variables used in the primary analysis of the individual study protocol; for other subgroups (i.e. age, sex, ECOG performance status, α-fetoprotein, extrahepatic spread, macrovascular invasion, viral status, HBV status, BCLC stage, prior locoregional therapy, prior surgical treatment, and region), the analysis was stratified by protocol.
The data cut-offs for KEYNOTE-240 were March 26, 2018 (PFS; first interim analysis time point for the study), and January 2, 2019 (OS and ORR; final analysis time point for the study). The data cut-offs for KEYNOTE-394 were June 30, 2021 (OS; final analysis time point for the study), and June 30, 2020 (PFS and ORR; second interim analysis time point for the study).
Results
Participants
Between May 31, 2016 and November 23, 2017, 413 participants were randomly assigned to receive pembrolizumab (n = 278) or placebo (n = 135) plus best supportive care in the KEYNOTE-240 study. Between May 31, 2017 and December 11, 2019, 453 participants were randomly assigned to receive pembrolizumab (n = 300) or placebo (n = 153) plus best supportive care in the KEYNOTE-394 study. The pooled ITT population included 578 participants in the pembrolizumab arm and 288 participants in the placebo arm (Fig. S1). The median follow-up, defined as time from randomization to data cut-off for final analysis of OS, was 21.4 months (range, 13.3–30.4 months) for KEYNOTE-240 and 33.8 months (range, 18.7–49.0 months) for KEYNOTE-394. Demographic and baseline disease characteristics were generally well balanced between treatment arms in each study (Table 1). By extension, the demographic and baseline characteristics were also balanced in the pooled population.
Table 1.
Baseline demographic and clinical characteristics in KEYNOTE-240 and KEYNOTE-394.
| Characteristics | KEYNOTE-240 |
KEYNOTE-394 |
||
|---|---|---|---|---|
| Pembrolizumab (n = 278) | Placebo (n = 135) | Pembrolizumab (n = 300) | Placebo (n = 153) | |
| Age, median (range), years | 67 (18–91) | 65 (23–89) | 54 (22–82) | 54 (22–78) |
| ≥65 | 169 (60.8) | 71 (52.6) | 69 (23.0) | 29 (19.0) |
| Male | 226 (81.3) | 112 (83.0) | 257 (85.7) | 126 (82.4) |
| Region | ||||
| China∗ | NA | NA | 255 (85.0) | 132 (86.3) |
| Ex-China† | NA | NA | 45 (15.0) | 21 (13.7) |
| Asia without Japan | 67 (24.1) | 31 (23.0) | NA | NA |
| European Union | 96 (34.5) | 43 (31.9) | NA | NA |
| Japan | 40 (14.4) | 19 (14.1) | NA | NA |
| United States | 21 (7.6) | 16 (11.9) | NA | NA |
| Other‡ | 54 (19.4) | 26 (19.3) | NA | NA |
| ECOG PS 1 | 116 (41.7) | 64 (47.4) | 176 (58.7) | 93 (60.8) |
| Child-Pugh class A | 277 (99.6) | 133 (98.5) | 300 (100.0) | 153 (100.0) |
| α-Fetoprotein ≥200 ng/ml | 129 (46.4) | 58 (43.0) | 169 (56.3) | 78 (51.0) |
| Extrahepatic spread | 195 (70.1) | 93 (68.9) | 232 (77.3) | 120 (78.4) |
| Macrovascular invasion | 36 (12.9) | 16 (11.9) | 33 (11.0) | 17 (11.1) |
| BCLC stage C | 222 (79.9) | 106 (78.5) | 277 (92.3) | 146 (95.4) |
| Viral status | ||||
| HBV positive§ | 72 (25.9) | 29 (21.5) | 236 (78.7) | 124 (81.0) |
| HCV positive§ | 43 (15.5) | 21 (15.6) | 5 (1.7) | 1 (0.7) |
| Prior first-line treatment | ||||
| Sorafenib | 278 (100.0) | 135 (100.0) | 272 (90.7) | 139 (90.8) |
| Oxaliplatin-based chemotherapy | NA | NA | 28 (9.3) | 14 (9.2) |
Data are n (%) unless otherwise specified. Percentages may not add up to 100 because of rounding.
BCLC, Barcelona Clinic Liver Cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; NA, not applicable.
Includes mainland China, Hong Kong, and Taiwan.
Includes Republic of Korea and Malaysia.
Includes Argentina, Australia, Canada, Chile, Colombia, Israel, Mexico, Norway, Russian Federation, and Turkey.
Based on local or central laboratory assessment. HBV positive was defined as hepatitis B surface antigen-positive and/or detectable HBV DNA. HCV positive was defined as anti-HCV-positive and detectable HCV RNA.
Efficacy
A total of 405 participants (70.1%) in the pembrolizumab arm and 229 participants (79.5%) in the placebo arm died. The HR for OS with pembrolizumab vs. placebo was 0.79 (95% CI 0.67–0.93) (Fig. 1A). Median OS was 14.2 months (95% CI 12.8–16.2 months) in the pembrolizumab arm vs. 12.5 months (95% CI 10.2–13.6 months) in the placebo arm. The estimated percentage of participants who were alive at 12, 24, and 36 months was 56.0% (95% CI 51.9–59.9%), 31.2% (95% CI 27.1–35.3%), and 21.4% (95% CI 16.8–26.3%), respectively, in the pembrolizumab arm and 50.7% (95% CI 44.8–56.3%), 22.8% (95% CI 17.8–28.2%), and 9.8% (95% CI 5.5–15.7%), respectively, in the placebo arm. Subgroup analyses showed a similar treatment effect of pembrolizumab on OS across subgroups, including viral status, BCLC stage, age, and geographic region (Fig. 1B). In particular, HRs for OS analyzed by subgroups based on viral status (infected [HBV positive or HCV positive] vs. uninfected) were consistent (0.76 [95% CI 0.62–0.93] vs. 0.85 [95% CI 0.65–1.11], respectively), although the 95% CI was broad for participants with an uninfected viral status.
Fig. 1.
Overall survival in the pooled intention-to-treat population.
(A) Kaplan–Meier estimates of OS in all participants. (B) Subgroup analysis of OS. For the overall pooled intention-to-treat population, the HR for OS was based on the Cox regression model with treatment as a covariate and stratified by protocol and, within each protocol, by strata used in the analysis of the respective protocol. For the subgroup analysis by protocol, the Cox regression model stratified by strata in the analysis of the respective protocol was used. For other subgroup analyses, the Cox regression model stratified by the protocol, but no other stratification variable, was used. BCLC, Barcelona Clinic Liver Cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; OS, overall survival.
Sensitivity analyses (pembrolizumab, n = 578; placebo, n = 288) adjusting for the use of poststudy anticancer therapies in both treatment arms by using the IPCW model showed an HR for OS of 0.64 (95% CI 0.51–0.81) with pembrolizumab vs. placebo. The two-stage model showed an HR for OS of 0.65 (95% CI 0.54–0.79) with pembrolizumab vs. placebo (Table S1).
A total of 440 participants (76.1%) in the pembrolizumab arm and 239 participants (83.0%) in the placebo arm experienced disease progression or died. The HR for PFS with pembrolizumab vs. placebo was 0.76 (95% CI 0.64–0.89) (Fig. 2A). Median PFS was 2.8 months (95% CI 2.7–2.9 months) in the pembrolizumab arm vs. 2.7 months (95% CI 1.6–2.8 months) in the placebo arm. The estimated percentage of participants who were alive without disease progression at 12 and 24 months was 16.8% (95% CI 13.4–20.7%) and 10.2% (95% CI 6.9–14.3%), respectively, in the pembrolizumab arm and 3.8% (95% CI 1.4–7.9%) and not applicable (i.e. no participant at risk), respectively, in the placebo arm (Fig. 2A). Subgroup analyses showed a similar treatment effect of pembrolizumab on PFS across subgroups (Fig. 2B). HRs for PFS analyzed by subgroups based on viral status (infected [HBV positive or HCV positive] vs. uninfected) were consistent (0.76 [95% CI 0.62–0.93] vs. 0.72 [95% CI 0.55–0.93], respectively).
Fig. 2.
Progression-free survival in the pooled intention-to-treat population.
(A) Kaplan–Meier estimates of PFS in all participants. (B) Subgroup analysis of PFS. For the overall pooled intention-to-treat population, the HR for PFS was based on the Cox regression model with treatment as a covariate and stratified by protocol and, within each protocol, by strata used in the analysis of the respective protocol. For the subgroup analysis by protocol, the Cox regression model stratified by strata in the analysis of the respective protocol was used. For other subgroup analyses, the Cox regression model stratified by the protocol, but no other stratification variable, was used. BCLC, Barcelona Clinic Liver Cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; NA, not applicable; PFS, progression-free survival.
The ORR was 15.4% (89 of 578 participants [95% CI 12.6–18.6%]) in the pembrolizumab arm vs. 2.8% (8 of 288 participants [95% CI 1.2–5.4%]) in the placebo arm, with an estimated treatment difference of 12.5% (95% CI 8.8–16.2%). Subgroup analysis showed that the estimated treatment difference in ORR was generally similar across subgroups (Fig. 3). Of note, the estimated treatment differences in ORR analyzed by subgroups based on viral status (infected [HBV positive or HCV positive] vs. uninfected) were consistent (10.0% [95% CI 5.3–14.4%] vs. 16.5% [95% CI 10.3–22.6%], respectively).
Fig. 3.
Subgroup analysis of the estimated treatment difference in objective response rate in the pooled intention-to-treat population.
For the overall pooled intention-to-treat population, the 95% CI for ORR difference was based on the Miettinen and Nurminen method stratified by protocol and, within each protocol, by strata used in the analysis of the respective protocol. For the subgroup analysis by protocol, the Miettinen and Nurminen method stratified by strata the analysis of the respective protocol was used. For other subgroup analyses, the Miettinen and Nurminen method stratified by the protocol, but no other stratification variable, was used. BCLC, Barcelona Clinic Liver Cancer; ECOG PS, Eastern Cooperative Oncology Group performance status; ORR, objective response rate.
Discussion
To obtain a more precise estimate of the treatment effect of pembrolizumab on OS, PFS, and ORR, a meta-analysis of pooled participant data from the phase III KEYNOTE-240 and KEYNOTE-394 studies was conducted. This analysis demonstrates that single-agent pembrolizumab provides clinically meaningful improvement in OS, PFS, and ORR compared with placebo in participants with advanced HCC previously treated with sorafenib or chemotherapy. Additionally, the treatment effect on OS, PFS, and ORR was generally consistent across subgroups based on baseline demographic and disease characteristics, including viral status, BCLC stage, and geographic region. Pembrolizumab also demonstrated a favorable safety profile in the second-line setting. Notably, when both the KEYNOTE-240 and KEYNOTE-394 studies were designed and conducted, immune checkpoint inhibitors were not yet approved for first-line treatment of HCC. Thus, a limitation of the current study is the uncertainty regarding the role of pembrolizumab after prior immune checkpoint inhibitor-based first-line therapy.
While KEYNOTE-240 and KEYNOTE-394 studies were being conducted, several new treatment options, including multikinase inhibitors, antiangiogenics, and immune checkpoint inhibitors, and new treatment combinations became globally available.2 The availability and use of these new treatments as poststudy systemic anticancer therapy likely influenced survival following discontinuation of study treatment. Our findings from sensitivity analyses of OS adjusting for the use of poststudy anticancer therapies suggest that these therapies may have attenuated the treatment effect estimate for pembrolizumab in this setting.
In KEYNOTE-240 and KEYNOTE-394, generally consistent efficacy results were observed for OS (HR 0.781 [95% CI 0.611–0.998] and 0.79 [95% CI 0.63–0.99], respectively), PFS (HR 0.775 [95% CI 0.609–0.987] and 0.74 [95% CI 0.60–0.92], respectively), and ORR (estimated treatment difference, 13.8% [95% CI 7.7–19.5%] and 11.4% [95% CI 6.7%–16.0%], respectively) with pembrolizumab vs. placebo.9,10 The current meta-analysis of pooled participant-level data from the KEYNOTE-240 and KEYNOTE-394 studies showed results comparable to those of both studies individually, with HRs for OS (0.79 [95% CI 0.67–0.93]) and PFS (0.76 [95% CI 0.64–0.89]) and an ORR of 15.4% vs. 2.8%, with an estimated treatment difference of 12.5% (95% CI 8.8–16.2%) favoring pembrolizumab over placebo. Of note, the HRs for OS and PFS and the estimated treatment difference in ORR observed in this meta-analysis comprising participants from global locations, including Asia (KEYNOTE-240), as well as only participants from Asia (KEYNOTE-394), suggest that the efficacy of pembrolizumab is consistent regardless of geographic region. Additionally, the analysis of the pooled populations showed similar benefit regardless of viral hepatitis status.
A strength of this meta-analysis is the similarity in trial design and inclusion and exclusion criteria between KEYNOTE-240 and KEYNOTE-394. KEYNOTE-240, conducted globally, and KEYNOTE-394, conducted in Asia only, were similar in design, with 2:1 randomization to pembrolizumab vs. placebo, and both studies included participants with advanced HCC treated with sorafenib. KEYNOTE-394 included a subset of participants who received oxaliplatin-based chemotherapy as well. Although previous treatment with oxaliplatin-based chemotherapy was permitted in KEYNOTE-394, >90% of participants received prior treatment with sorafenib. Both studies also had similar eligibility criteria and end points, including OS, PFS, and ORR. The similarities of both studies complement one another and allow for a robust estimate of the treatment effect of second-line pembrolizumab in participants with advanced HCC. However, the interpretation of this meta-analysis may be affected by the difference in the design of these studies, as a more conservative statistical assumption was used in KEYNOTE-394 for PFS and OS compared with a more aggressive assumption for KEYNOTE-240.
Approximately 90% of HCC cases arise in the setting of cirrhosis resulting from both viral and nonviral etiologies, and these are known to vary by geographic region.1,2 Despite the pooling of the KEYNOTE-240 and KEYNOTE-394 studies, the sample size for participants with HCV remained small (KEYNOTE-240: pembrolizumab, n = 43; placebo, n = 21; KEYNOTE-394: pembrolizumab, n = 5; placebo, n = 1). The low number of participants with HCV in KEYNOTE-394 (n = 6) was not surprising, given that a primary risk factor in Asia is chronic HBV infection.15,16 The absence of a formal integrated meta-analysis of safety is a limitation of this study; however, the safety profile of pembrolizumab was generally consistent between the KEYNOTE-2409 and KEYNOTE-39410 studies individually and was consistent with the known safety profile of pembrolizumab monotherapy in other indications.[17], [18], [19]
This meta-analysis of KEYNOTE-240 and KEYNOTE-394 showed improvement in OS, PFS, and ORR with pembrolizumab vs. placebo for the treatment of advanced HCC in the second-line setting, regardless of HBV or HCV status and differences in geographic region. These pooled data expand on findings from the KEYNOTE-224,8 KEYNOTE-240,9 and KEYNOTE-39410 studies individually and provide additional evidence of the efficacy of second-line pembrolizumab for the treatment of advanced HCC in a global population, with increased precision because of the larger sample size. Along with recent findings of a numerical trend toward improvement in both OS and PFS with first-line lenvatinib plus pembrolizumab vs. lenvatinib monotherapy in participants with advanced HCC (LEAP-002; NCT03713593), although prespecified statistical significance for OS or PFS were not met,20 these results provide additional evidence for pembrolizumab in the treatment of advanced HCC. Furthermore, ongoing phase III studies evaluating pembrolizumab as monotherapy in the adjuvant treatment setting in participants with HCC (KEYNOTE-937; NCT03867084) and as combination therapy with lenvatinib plus transarterial chemoembolization in participants with intermediate-stage HCC not amenable to curative treatment (LEAP-012; NCT04246177) will clarify the overall body of evidence for pembrolizumab in the treatment of advanced HCC.
Conclusions
In summary, the data reported in this meta-analysis provide further evidence for the benefit of pembrolizumab as second-line therapy for advanced HCC in a global population of participants and may be a second-line option for those who do not receive immunotherapy in the frontline setting.
Abbreviations
BCLC, Barcelona Clinic Liver Cancer; BICR, blinded independent central review; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; IPCW, inverse probability of censoring weighting; ITT, intention-to-treat; ORR, objective response rate; OS, overall survival; PD-L1, programmed cell death ligand 1; PFS, progression-free survival; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version v1.1; VEGF, vascular endothelial growth factor.
Financial support
This work was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA. The funder contributed to the study design, data collection, data analysis, and data interpretation, in collaboration with the authors; all authors had full access to the data. Investigators and site personnel collected data, which was housed in MSD’s database. The funder provided financial support for editorial and writing assistance. The corresponding author had full access to all the data and had final responsibility for the decision to submit for publication.
Authors’ contributions
Full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis: RSF, PY. Conceptualization: RSF, A-LC, MK, JL, ABS, SQ. Formal analysis: PY. Writing – original draft: MB, HYL, ABS. Writing – reviewing and editing: all authors.
Data availability statement
Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA (MSD) is committed to providing qualified scientific researchers access to anonymized data and clinical study reports from the company’s clinical trials for the purpose of conducting legitimate scientific research. MSD is also obligated to protect the rights and privacy of trial participants and, as such, has a procedure in place for evaluating and fulfilling requests for sharing company clinical trial data with qualified external scientific researchers. The MSD data-sharing website (available at: https://externaldatasharing-msd.com/) outlines the process and requirements for submitting a data request. Applications will be promptly assessed for completeness and policy compliance. Feasible requests will be reviewed by a committee of MSD subject matter experts to assess the scientific validity of the request and the qualifications of the requestors. In line with data privacy legislation, submitters of approved requests must enter into a standard data-sharing agreement with MSD before data access is granted. Data will be made available for request after product approval in the United States and the European Union or after product development is discontinued. There are circumstances that may prevent MSD from sharing requested data, including country- or region-specific regulations. If the request is declined, it will be communicated to the investigator. Access to genetic or exploratory biomarker data requires a detailed, hypothesis-driven statistical analysis plan that is collaboratively developed by the requestor and MSD subject matter experts; after approval of the statistical analysis plan and execution of a data-sharing agreement, MSD will either perform the proposed analyses and share the results with the requestor or will construct biomarker covariates and add them to a file with clinical data that is uploaded to an analysis portal so that the requestor can perform the proposed analyses.
Conflicts of interest
RSF reported receiving payments for consultancy, travel, writing assistance, and article processing for the present work and grants paid to his institution for the present work, all from Merck, and grants to his institution from Adaptimmune, Bayer, Bristol Myers Squibb, Eisai, Eli Lilly, Genentech, Pfizer, and Roche; consulting fees paid to him from AstraZeneca, Bayer, Bristol Myers Squibb, CStone, Eisai, Eli Lilly, Exelixis, Genentech, Hengrui, Pfizer, and Roche; payment to him for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Genentech; and payments to him from AstraZeneca and Hengrui for participation on a data safety monitoring board or advisory board, outside of the present work. XC reported receiving consulting fees from AstraZeneca, Bayer, BeiGene, Bristol Myers Squibb, EXE, Genentech/Roche, Ipsen Innovation, and MSD; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Amgen Taiwan, Bayer Yakuhin, Chugai Pharmaceutical, Novartis, and Ono Pharmaceutical; support for attending meetings and/or travel from IQVIA; and participation on a data safety monitoring board or advisory board for Abbisko Therapeutics, outside of the present work. PM reported serving in a consulting or advisory role for AstraZeneca, Bayer, Eisai, Ipsen, MSD, and Roche and receiving funding from Ipsen, outside of the present work. K-HL reported receiving honoraria from AstraZeneca, Eli Lilly, Novartis, and Pfizer and payment for expert testimony from Novartis, outside of the present work. MB reported receiving consulting fees from AstraZeneca, Bayer, Bristol Myers Squibb, Eisai, MSD, Roche, and Sirtex Medical; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from AstraZeneca, Bayer Pharma, and Roche; and support for attending meetings and/or travel from AstraZeneca, Bayer, and Roche, outside of the present work. A-LC reported receiving consulting fees from AstraZeneca, Bayer, BeiGene, Bristol Myers Squibb, Eisai, Genentech/Roche, Ipsen Innovation, MSD, Omega Therapeutics, Ono Pharmaceutical, and Sanofi; payment or honoraria for lectures, presentations, or speakers’ bureaus from Amgen Taiwan, Ipsen Innovation, and Roche; support for attending meetings and/or travel from AstraZeneca and Roche; and participation on a data safety monitoring board or advisory board for Abbisko Therapeutics, outside of the present work. HYL reported receiving payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Bayer, Eisai, and Roche; participation on a data safety monitoring board or advisory board for AstraZeneca, Bayer, Eisai, Ipsen, and Roche; and role as the president of the Korean Society of Medical Oncology, outside of the present work. MK reported receiving support for the present work from MSD, and grants or contracts from AbbVie, EA Pharma, Eisai, GE Healthcare, Gilead Sciences, Otsuka, Sumitomo Dainippon Pharma, Taiho, and Takeda; and payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Bayer, Chugai, Eli Lilly, Eisai, MSD, and Takeda, outside of the present work. JE reported receiving grants or contracts from BeiGene, Boston Scientific, and Bristol Myers Squibb; consulting fees from AstraZeneca, Basilea, Bayer, BeiGene, Boston Scientific, Bristol Myers Squibb, Eisai, Incyte, Ipsen, Merck Serono, MSD, Roche, Servier, and Taiho; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Incyte, Ipsen, Roche, and Servier; and support for attending meetings and/or travel from Amgen, outside of the present work. PY is an employee of MSD China, Shanghai, China. CL and JL are employees of MSD China, Beijing, China. ABS is an employee of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA, and owns stock in Merck & Co., Inc., Rahway, NJ, USA. KG, PC, WW, LZ, YP, T-TC, WL, and SQ report no disclosures.
Please refer to the accompanying ICMJE disclosure forms for further details.
Acknowledgements
The authors thank the patients and their families and caregivers and the primary investigators and site personnel for participating in the KEYNOTE-240 and KEYNOTE-394 studies. The authors also thank Olga Kuznetsova, PhD, and Scot Ebbinghaus, MD (of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA), for their contributions to this analysis and development of the manuscript. Medical writing and/or editorial assistance was provided by Obinna T. Ezeokoli, PhD, Shane Walton, PhD, CMPP, and Lauren D’Angelo, PhD, of ApotheCom (Yardley, PA, USA). This assistance was funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway NJ, USA.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jhepr.2025.101350.
Supplementary data
The following are the supplementary data to this article:
References
- 1.Sung H., Ferlay J., Siegel R.L., et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 2.Llovet J.M., Kelley R.K., Villanueva A., et al. Hepatocellular carcinoma. Nat Rev Dis Primers. 2021;7:6. doi: 10.1038/s41572-020-00240-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Park J.-W., Chen M., Colombo M., et al. Global patterns of hepatocellular carcinoma management from diagnosis to death: the BRIDGE study. Liver Int. 2015;35:2155–2166. doi: 10.1111/liv.12818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kudo M., Finn R.S., Qin S., et al. Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet. 2018;391:1163–1173. doi: 10.1016/S0140-6736(18)30207-1. [DOI] [PubMed] [Google Scholar]
- 5.Su G.L., Altayar O., O'Shea R., et al. AGA clinical practice guideline on systemic therapy for hepatocellular carcinoma. Gastroenterology. 2022;162:920–934. doi: 10.1053/j.gastro.2021.12.276. [DOI] [PubMed] [Google Scholar]
- 6.Abou-Alfa G.K., Lau G., Kudo M., et al. Tremelimumab plus durvalumab in unresectable hepatocellular carcinoma. NEJM Evid. 2022;1 doi: 10.1056/EVIDoa2100070. [DOI] [PubMed] [Google Scholar]
- 7.KEYTRUDA® (pembrolizumab) injection, for intravenous use. 03/2025. Merck Sharp & Dohme LLC; Rahway, NJ, USA: 2025. https://www.merck.com/product/usa/pi_circulars/k/keytruda/keytruda_pi.pdf [Google Scholar]
- 8.Zhu A.X., Finn R.S., Edeline J., et al. Pembrolizumab in patients with advanced hepatocellular carcinoma previously treated with sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial. Lancet Oncol. 2018;19:940–952. doi: 10.1016/S1470-2045(18)30351-6. [DOI] [PubMed] [Google Scholar]
- 9.Finn R.S., Ryoo B.-Y., Merle P., et al. Pembrolizumab as second-line therapy in patients with advanced hepatocellular carcinoma in KEYNOTE-240: a randomized, double-blind, phase III trial. J Clin Oncol. 2020;38:193–202. doi: 10.1200/JCO.19.01307. [DOI] [PubMed] [Google Scholar]
- 10.Qin S., Chen Z., Fang W., et al. Pembrolizumab versus placebo as second-line therapy in patients from Asia with advanced hepatocellular carcinoma: a randomized, double-blind, phase III trial. J Clin Oncol. 2023;41:1434–1443. doi: 10.1200/JCO.22.00620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Vogel A., Cervantes A., Chau I., et al. Hepatocellular carcinoma: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2018;29(Suppl 4):iv238–iv255. doi: 10.1093/annonc/mdy308. [DOI] [PubMed] [Google Scholar]
- 12.Robins J.M., Finkelstein D.M. Correcting for noncompliance and dependent censoring in an AIDS clinical trial with inverse probability of censoring weighted (IPCW) log-rank tests. Biometrics. 2000;56:779–788. doi: 10.1111/j.0006-341x.2000.00779.x. [DOI] [PubMed] [Google Scholar]
- 13.Latimer N.R., Abrams K.R., Lambert P.C., et al. Adjusting survival time estimates to account for treatment switching in randomized controlled trials—an economic evaluation context: methods, limitations, and recommendations. Med Decis Making. 2014;34:387–402. doi: 10.1177/0272989X13520192. [DOI] [PubMed] [Google Scholar]
- 14.Tierney J.F., Vale C., Riley R., et al. Individual participant data (IPD) meta-analyses of randomised controlled trials: guidance on their use. PLoS Med. 2015;12 doi: 10.1371/journal.pmed.1001855. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jayaraman T., Lee Y.-Y., Chan W.K., et al. Epidemiological differences of common liver conditions between Asia and the West. JGH Open. 2019;4:332–339. doi: 10.1002/jgh3.12275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhu R.X., Seto W.-K., Lai C.L., et al. Epidemiology of hepatocellular carcinoma in the Asia-Pacific region. Gut Liver. 2016;10:332–339. doi: 10.5009/gnl15257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Fuchs C.S., Doi T., Jang R.W., et al. Safety and efficacy of pembrolizumab monotherapy in patients with previously treated advanced gastric and gastroesophageal junction cancer: phase 2 clinical KEYNOTE-059 trial. JAMA Oncol. 2018;4 doi: 10.1001/jamaoncol.2018.0013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Adams S., Loi S., Toppmeyer D., et al. Pembrolizumab monotherapy for previously untreated, PD-L1-positive, metastatic triple-negative breast cancer: cohort B of the phase II KEYNOTE-086 study. Ann Oncol. 2019;30:405–411. doi: 10.1093/annonc/mdy518. [DOI] [PubMed] [Google Scholar]
- 19.Chung H.C., Ros W., Delord J.-P., et al. Efficacy and safety of pembrolizumab in previously treated advanced cervical cancer: results from the phase II KEYNOTE-158 study. J Clin Oncol. 2019;37:1470–1478. doi: 10.1200/JCO.18.01265. [DOI] [PubMed] [Google Scholar]
- 20.Llovet J.M., Kudo M., Merle P., et al. Lenvatinib plus pembrolizumab versus lenvatinib plus placebo for advanced hepatocellular carcinoma (LEAP-002): a randomised, double-blind, phase 3 trial. Lancet Oncol. 2023;24:1399–1410. doi: 10.1016/S1470-2045(23)00469-2. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA (MSD) is committed to providing qualified scientific researchers access to anonymized data and clinical study reports from the company’s clinical trials for the purpose of conducting legitimate scientific research. MSD is also obligated to protect the rights and privacy of trial participants and, as such, has a procedure in place for evaluating and fulfilling requests for sharing company clinical trial data with qualified external scientific researchers. The MSD data-sharing website (available at: https://externaldatasharing-msd.com/) outlines the process and requirements for submitting a data request. Applications will be promptly assessed for completeness and policy compliance. Feasible requests will be reviewed by a committee of MSD subject matter experts to assess the scientific validity of the request and the qualifications of the requestors. In line with data privacy legislation, submitters of approved requests must enter into a standard data-sharing agreement with MSD before data access is granted. Data will be made available for request after product approval in the United States and the European Union or after product development is discontinued. There are circumstances that may prevent MSD from sharing requested data, including country- or region-specific regulations. If the request is declined, it will be communicated to the investigator. Access to genetic or exploratory biomarker data requires a detailed, hypothesis-driven statistical analysis plan that is collaboratively developed by the requestor and MSD subject matter experts; after approval of the statistical analysis plan and execution of a data-sharing agreement, MSD will either perform the proposed analyses and share the results with the requestor or will construct biomarker covariates and add them to a file with clinical data that is uploaded to an analysis portal so that the requestor can perform the proposed analyses.