Salvage Ipilimumab plus Nivolumab after Anti-PD-1/PD-L1 Therapy in Advanced Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the seventh most common malignancy in the world but represents the second leading cause of cancer death (1, 2). Despite a recognizable at-risk population and effective HCC surveillance, up to half of patients diagnosed with HCC present with advanced, incurable disease (1, 2). Until recently, multi-kinase inhibitors were the only approved therapy for advanced, unresectable HCC, but recent clinical studies have established a role for immune checkpoint inhibitors, including anti-PD-(L)1 therapies, in this disease across multiple contexts (3–10).

On the basis of the phase III IMbrave150 trial, atezolizumab plus bevacizumab was the first immunotherapy combination regimen to show an improved overall survival (OS) benefit compared with sorafenib in the front-line setting for advanced HCC (3, 11). The efficacy of targeting CTLA-4 and PD-1 was shown by the CHECKMATE-040 study, demonstrating that ipilimumab plus nivolumab could induce durable responses in a subset of patients with HCC previously treated with sorafenib (4, 5, 12). More recently, the phase III HIMALAYA trial showed that tremelimumab plus durvalumab was associated with a significant improvement in OS compared with sorafenib in the first-line setting (10).

Thus far, all treatments for HCC after the first-line of therapy have been approved on the basis of trials enrolling patients intolerant or refractory to multi-kinase inhibitors, and, as a result, the optimal sequence of therapy for patients previously treated with immune checkpoint inhibitors is unknown. Prospective comparative effectiveness trials are unlikely to be conducted, so we will likely depend on retrospective series demonstrating efficacy and safety of different regimens in this setting.

While combined PD-(L)1/CTLA-4 blockade following anti-PD-(L)1 failure has demonstrated efficacy as a salvage regimen in other cancer types, it is unknown whether this treatment sequence is effective in HCC (13, 14). In particular, it is unclear whether CTLA-4 inhibition after progression on atezolizumab plus bevacizumab or other anti-VEGF plus anti-PD-(L)1 combinations have utility in HCC given that such regimens have been the preferred front-line treatments for eligible patients. Herein, we assess the clinical benefit and safety of ipilimumab plus nivolumab in patients previously treated with anti-PD-(L)1 therapy in HCC.

Patients were identified on the basis of retrospective review of the electronic health record at the Johns Hopkins Hospital (July 1, 2016–November 15, 2022), University of Texas Southwestern Medical Center (March 1, 2020–November 1, 2022), and Parkland Health and Hospital System (March 1, 2020–November 1, 2022). Patient sex and race were obtained from the electronic health record. Eligible patients included all adults (18 years of age or older) with a diagnosis of HCC based on histology or imaging characteristics meeting American Association for the Study of Liver Diseases (AASLD) guideline criteria who had received at least one dose of anti-PD-(L)1 therapy with discontinuation for progression or alternative treatment, specified further below, followed by at least one dose of ipilimumab plus nivolumab (15). All patients had subsequent imaging and/or laboratory monitoring to determine disease progression. Patients with prior anti-CTLA-4 treatment or patients with fibrolamellar HCC, sarcomatoid HCC, or mixed cholangiocarcinoma/HCC were excluded. This study was approved by the Institutional Review Board at each of the respective sites and conducted in accordance with the Declaration of Helsinki.

All patients received ipilimumab 3 mg/kg plus nivolumab 1 mg/kg, except for one patient who received ipilimumab 1 mg/kg plus nivolumab 3 mg/kg based on a concern for potential increased risk of immune-related adverse events (irAE) in the setting of underlying autoimmune disease. Patients received ipilimumab plus nivolumab every 3 weeks for up to four cycles followed by 480 mg nivolumab maintenance therapy every 4 weeks until disease progression, death, intolerance, or patient preference. Per routine practice at each site, patients had imaging every 6–12 weeks and laboratory monitoring prior to every cycle to determine disease progression. Radiographic response to treatment was assessed using cross-sectional imaging per RECIST v1.1. In 4 patients, clinical progression was determined by the treating physician based on signs or symptoms consistent with progressive disease (PD) and a concomitant rise in alpha-fetoprotein levels (16).

The primary outcome in our study was the objective response rate (ORR), defined as patients who achieved complete response (CR) or partial response (PR). The secondary outcome variables in our study were progression-free survival (PFS), calculated as the time from the first dose of ipilimumab plus nivolumab to disease progression or death from any cause, and OS, calculated as the time from the first dose of ipilimumab plus nivolumab to death from any cause. We also evaluated the duration of response (DOR), defined as the time from the first dose of ipilimumab plus nivolumab to disease progression in patients who achieved a CR or PR. irAEs were graded using the NCI Common Terminology Criteria for Adverse Events v4.0.

All analyses were performed using R (V4.2.1), with the survival and survminer packages used to calculate PFS, OS, and DOR. Continuous variables were described with median, range, number, and percentage of the cohort. Differences among categorical variables were assessed using Fisher exact test. Median PFS, OS, and DOR were estimated using Kaplan–Meier analyses with log-rank test used to assess differences in PFS and OS across subgroups of interest, including response (responders vs. nonresponders), body mass index (BMI; overweight/obese vs. normal/underweight), and HCC etiology (viral vs. nonviral). A P value < 0.05 was considered statistically significant for all analyses.

The data generated in this study are not publicly available to preserve patient confidentiality and due to the Health Insurance Portability and Accountability Act, however, deidentified data are available upon reasonable request from the corresponding author.

The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the article; and decision to submit the article for publication.

We identified 32 patients with advanced HCC and prior anti-PD-(L)1 therapy who were subsequently treated with ipilimumab plus nivolumab (Table 1). The median age was 67 years, 59% (n = 19) of patients had viral-associated cirrhosis, with 47% (n = 15) due to hepatitis C and 13% (n = 4) due to hepatitis B; 66% (n = 21) of patients had Child-Pugh (CP) A disease prior to initiating ipilimumab plus nivolumab. With respect to Barcelona Clinic Liver Cancer (BCLC) stage, 13% (n = 4) of patients were stage B and 88% (n = 28) were stage C at the time of treatment.

Prior to treatment with ipilimumab plus nivolumab, 31%, 47%, and 22% of patients received one, two, and three or more lines of systemic treatments, respectively (Table 1). Prior lines of anti-PD-(L)1 therapy included immunotherapy doublet regimens comprised of atezolizumab plus bevacizumab (50%, n = 16), other anti-VEGF plus anti-PD-(L)1 combination therapies (31%, n = 10), and anti-PD-(L)1 monotherapy (19%, n = 6).

There were 28 patients (88%) who discontinued anti-PD-(L)1 therapy due to disease progression. Three patients (9%) stopped therapy for planned surgical resection after 2 months of anti-PD-(L)1 therapy. One patient (3%) stopped therapy due to lack of response to anti-PD-(L)1 therapy plus grade 2 esophagitis, managed with oral steroids, with transition to a multi-kinase inhibitor. The best overall response (BOR) to prior PD-(L)1 blockade was PD in 56% of patients (n = 18), stable disease (SD) in 38% of patients (n = 12), and PR in 6% of patients (n = 2). Of note, the 4 patients who discontinued therapy for surgical resection and lack of response to therapy had a BOR of SD.

The ORR with ipilimumab plus nivolumab after prior anti-PD-(L)1 therapy was 22%, including 1 patient (3%) with a CR and 6 patients (19%) with a PR. In addition, 8 patients demonstrated SD (25%), for a disease control rate of 47% (Fig. 1). One patient's response was not evaluable, as they developed a terminal irAE prior to response assessment, and an additional 16 patients (50%) had PD.

Among patients with an objective response to salvage ipilimumab plus nivolumab, none had an objective response to prior anti-PD-(L)1. Objective response was seen in 3 of the 12 patients who previously demonstrated SD and 4 of the 18 patients who previously demonstrated PD. Among the 4 patients who discontinued therapy for surgical resection and lack of response to therapy, 2 had an objective response to ipilimumab plus nivolumab, with both patients discontinuing first-line anti-PD-(L)1 therapy for surgical resection. There was no difference in BOR based on BMI (overweight/obese vs. normal/underweight, P = 0.22) or HCC etiology (viral vs. nonviral, P = 0.09).

The median PFS was 2.9 months [95% confidence interval (CI): 2.1–not reached], and the median OS was 9.2 months (95% CI: 5.9–not reached; Fig. 2A and B). The DOR in patients who responded to ipilimumab plus nivolumab was 6.0 months (95% CI: 4.1–not reached). Objective response to ipilimumab plus nivolumab was associated with improved PFS (2.4 months vs. not reached, P = 0.004) and OS (5.9 months vs. not reached, P = 0.02; Supplementary Fig. S1A and S1B). There were no statistically significant differences in PFS and OS after salvage ipilimumab plus nivolumab when stratifying patients by BMI or HCC etiology (Supplementary Table S1).

During patients’ first line of anti-PD-(L)1 therapy, 7 (22%) experienced irAEs, including grade 2 hypothyroidism (n = 3, 9%), grade 2 hyperthyroidism (n = 1, 3%), grade 2 esophagitis (n = 1, 3%), grade 1 dermatitis (n = 1, 3%), and grade 1 transaminitis (n = 1, 3%). Among patients who experienced irAEs with the first line of anti-PD-(L)1, 3 (9%) experienced irAEs when treated with ipilimumab plus nivolumab. These 3 patients included: a patient who experienced grade 2 esophagitis, who then discontinued ipilimumab plus nivolumab due to grade 4 colitis; a patient who experienced grade 2 hyperthyroidism, who experienced grade 3 pneumonitis, grade 1 esophagitis, and grade 2 inflammatory arthritis with ipilimumab plus nivolumab, all managed with steroids, allowing the patient to continue nivolumab maintenance; and 1 of the 3 patients who experienced grade 2 hypothyroidism, who developed grade 5 hepatitis, as detailed below.

During treatment with ipilimumab plus nivolumab, irAEs were more frequent as compared with the first line of anti-PD-(L)1 therapy, as expected. irAEs occurred in 13 patients (41%) during salvage ipilimumab plus nivolumab (Table 2). The most common irAE was pneumonitis (n = 4, 13%), followed by colitis (n = 3, 9%), rash (n = 3, 9%), and hepatitis (n = 3, 9%). A grade 5 treatment-related adverse event of autoimmune hepatitis was observed in 1 (3%) patient. This fatal irAE occurred after two doses of therapy in a patient with CP A5 cirrhosis. Grade 3–4 irAEs were observed in 6 (19%) patients. Among patients who developed irAEs with ipilimumab plus nivolumab, 5 discontinued treatment due to toxicity, including 1 patient with Crohn disease who received flipped dose ipilimumab plus nivolumab. Two of the 5 patients were rechallenged with ipilimumab plus nivolumab: 1 patient developed recurrent autoimmune hepatitis, requiring permanent discontinuation of therapy, while the other had no further complications. Neither of these 2 patients experienced irAE with prior anti-PD-(L)1 therapy. Notably, this regimen was found to be safe in patients with CP B cirrhosis, with no increased risk of irAE in this population.

This study demonstrates that salvage ipilimumab plus nivolumab in HCC is associated with clinically meaningful objective response and disease control. In addition, objective responses with ipilimumab plus nivolumab were achieved in patients with primary resistance to anti-PD-(L)1 therapy, adding to growing preclinical and clinical evidence that combined PD-(L)1/CTLA-4 blockade can overcome PD-(L)1 resistance in HCC and other tumor types (10, 13). The response rate observed in this study (22%) was somewhat lower than that reported for ipilimumab plus nivolumab in the prospective Checkmate 040 study (32%; refs. 4, 5, 10). However, this may reflect the immune checkpoint refractory population in this study, as the Checkmate 040 study enrolled patients with no prior exposure to anti-PD-(L)1 therapy. There were no discernable differences in safety or efficacy of salvage ipilimumab plus nivolumab across patient subsets such as nonviral or viral etiologies, but this requires further confirmation, as our study is likely underpowered to detect weak associations.

This retrospective study was conducted in the setting of a rapidly evolving treatment landscape for advanced HCC. At the time of this report, bevacizumab plus atezolizumab is the preferred first-line therapy for patients with advanced HCC, and there is little available data to guide treatment decisions in subsequent therapy lines (15, 17, 18). On the basis of expert consensus opinion, clinical guidelines from AASLD, American Society of Clinical Oncology, National Comprehensive Cancer Network (NCCN), and others recommend the use of multi-target tyrosine kinase inhibitors in patients progressing on bevacizumab plus atezolizumab, but some patients may not be candidates for such therapies, including patients with bleeding events on bevacizumab plus atezolizumab (15, 17, 18). Our retrospective data suggest that ipilimumab plus nivolumab is a treatment option for patients with prior exposure to anti-PD-(L)1 therapy and can be effective even in patients who do not respond to bevacizumab plus atezolizumab or other anti-PD-(L)1–based regimens.

This study shows that a lack of response to prior anti-PD-(L)1 therapy does not preclude a subsequent response to combined PD-(L)1/CTLA-4 blockade, suggesting that CTLA-4 blockade enhances antitumor immunity and/or activates immune cell programs unaffected by anti-PD-(L)1 monotherapy (10, 13). While targeting either immune checkpoints may enhance the immune system's antitumor response, they do so via different mechanisms. CTLA-4 expressed on T cells blunts the immune system's response to tumor antigens by blocking binding of B7 on antigen-presenting cells (APC) to CD28 on T cells, resulting in suppression of APC function and impaired activation of CD8⁺ T cells (19). The interaction between PD-1 on cytotoxic lymphocytes and PD-L1 on APCs and tumor cells transmits an inhibitory signal that induces CD8⁺ T-cell apoptosis and prevents T cell–mediated apoptosis (19–21). This mechanism is consistent with analyses of melanoma tissue after treatment with either anti-PD-(L)1 monotherapy, anti-CTLA-4 monotherapy, or in combination, which show distinct gene expression profiles associated with each type of treatment (22).

Salvage ipilimumab plus nivolumab was associated with irAEs in a subset of patients, with no new safety signals observed. irAE rates in this cohort were lower than those seen in prospective trials evaluating combined PD-(L)1/CTLA-4 blockade with ipilimumab plus nivolumab in HCC (4, 5, 10). This finding is potentially attributable to selection bias, as treating physicians likely only considered patients for salvage ipilimumab plus nivolumab in the setting of prior anti-PD-(L)1 therapy if they had tolerated prior immune checkpoint inhibition. Despite this finding, the single case of fatal hepatitis in our retrospective cohort underscores the potential toxicity and need for careful monitoring with ipilimumab plus nivolumab in any line of therapy.

Despite being the largest study to date that has evaluated outcomes of combined PD-L(1)/CTLA-4 blockade in HCC after failing anti-PD-(L)1 therapy, our cohort is small, which limits the generalizability of our findings and our ability to compare patient outcomes based on important clinical and demographic factors. In addition, while all patients received at least one form of anti-PD-(L)1 therapy prior to ipilimumab plus nivolumab, they received different types and numbers of systemic treatments, making our patient population heterogeneous. Future studies should include larger populations with more stringent inclusion criteria to better identify factors that influence response to treatment.

In our cohort, ipilimumab plus nivolumab resulted in ORRs comparable with those seen in other studies of anti-PD-(L)1/CTLA-4 combination therapy. There was an acceptable safety profile with no new safety signals; however, we did observe one grade 5 irAE of autoimmune hepatitis. Overall, these findings suggest that salvage ipilimumab plus nivolumab has meaningful clinical activity in select patients with advanced HCC who failed prior anti-PD-(L)1 therapy, including regimens incorporating anti-VEGF agents. Larger prospective studies are warranted to better define the optimal sequence of therapy in advanced HCC.

A.G. Singal reports personal fees from Genentech, AstraZeneca, Bayer, Eisai, Exelixis, and Boston Scientific outside the submitted work. W.J. Ho reports grants from Sanofi and NeoTX outside the submitted work. M. Yarchoan reports grants and personal fees from Genentech, Exelixis, AstraZeneca; grants from Bristol-Myers Squibb and Incyte; personal fees from Eisai, Replimune, Hepion; and other from Adventris Pharamceuticals outside the submitted work. No disclosures were reported by the other authors.

S.L. Alden: Conceptualization, data curation, formal analysis, investigation, methodology, writing-original draft, writing-review and editing. M. Lim: Conceptualization, data curation, writing-review and editing. C. Kao: Conceptualization, data curation, formal analysis, supervision, methodology, writing-review and editing. D. Shu: Resources, data curation. A.G. Singal: Resources, data curation, supervision, writing-review and editing. A. Noonan: Resources, data curation, writing-review and editing. P. Griffith: Resources, data curation. M. Baretti: Resources, data curation, writing-review and editing. W.J. Ho: Conceptualization, resources, data curation, writing-review and editing. I. Kamel: Conceptualization, resources, data curation, formal analysis, supervision, investigation, methodology. M. Yarchoan: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing-review and editing. D. Hsiehchen: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing-review and editing.

A.G. Singal's research is supported by NIH R01 MD012565 and R01 CA256977. M. Yarchoan's research is supported by NCI, grant number R01-CA265009, to M. Yarchoan; the NCI Specialized Program of Research Excellence in Gastrointestinal Cancers, grant number P50CA062924, to M. Yarchaon; NCI, grant number U01CA212007, to M. Yarchaon; NIH Center Core Grant, grant number P30CA006973, to M. Yarchaon. I. Kamel's research is supported by Siemens Healthineers. W.J. Ho's research is supported by Sanofi and NeoTX.