CheckMate 9DW in advanced hepatocellular carcinoma: A new standard or a niche strategy?

INTRODUCTION

As per GLOBOCAN 2022, hepatocellular carcinoma (HCC) ranks 6^th in incidence and 3^rd in mortality worldwide.^[1] The therapeutic landscape of HCC has undergone a profound transformation, evolving from tyrosine kinase inhibitors (TKIs) to combinations of immune checkpoint inhibitors (ICIs) with either TKIs or other ICIs. The recently published results from CheckMate 9DW exemplify this shift. In this study, nivolumab plus ipilimumab was compared with lenvatinib (85%) or sorafenib (15%) in patients with unresectable advanced HCC. The dual ICI regimen demonstrated an overall response rate (ORR) of 36%, a median progression-free survival (PFS) of 9.1 months, and a median overall survival (OS) of 23.7 months.^[2] The rationale for combining an anti-PD1 and anti-Cytotoxic T-lymphocyte associated protein 4 (CTLA-4) monoclonal antibody is well founded. It is postulated that there is high-level expression of CTLA-4 on T-regulatory cells (T-regs), which when inhibited by ipilimumab restores the activation of cytotoxic T-lymphocytes, resulting in a positive modulation of the immunosuppressive tumor microenvironment.^[3]

EFFICACY COMPARISON WITH HIMALAYA

Given the similar mechanism of action, it is reasonable to view these results alongside those of single tremelimumab with regular interval durvalumab (STRIDE), the other approved dual ICI regimen, which evaluated durvalumab in combination with a single priming dose of tremelimumab in patients with unresectable advanced HCC. However, caution is warranted when making cross-trial comparisons, as differences in inclusion criteria, baseline patient characteristics, and study design can significantly impact observed outcomes. Baseline patient characteristics were generally well aligned across trials, with both STRIDE and CheckMate 9DW excluding patients with main portal vein invasion. The STRIDE regimen demonstrated an ORR of 20% and a median OS of 16.4 months. While the survival curves for STRIDE and CheckMate 9DW initially overlapped for the first 2 years, they began to diverge thereafter. Notably, HIMALAYA provides the longest follow-up reported for any active regimen in advanced HCC, showing an impressive 5-year OS rate of 19.6% with STRIDE compared to 9.4% with sorafenib.^[4] Both STRIDE and CheckMate 9DW are notable for early progressors (40% versus 20%, respectively), likely reflecting the presence of immune non-responders, a phenomenon less commonly seen with ICI-VEGF combinations. Although median PFS was similar between the experimental and control arms in CheckMate 9DW (9.1 vs. 9.2 months), these values are higher than those reported for the STRIDE regimen (3.78 months).^[5] However, median PFS may be less informative than the duration of response, which has consistently been longer and more pronounced with pure ICI combinations, reflecting durable responses driven entirely by the immune system. With STRIDE, 21.1% of patients were excellent long-term responders (participants who survived >48 months beyond randomization), and among these patients, the median duration of response was not reached. In CheckMate 9DW, 47% of patients in the nivolumab/ipilimumab arm had an ongoing response at 3 years, with an impressive 30.4-month median duration of response. Even though the study design is open-labeled, this may not be regarded as a true limitation as there was good concordance for efficacy assessments such as ORR and PFS between investigator assessment and blinded independent central review. The main stratification factors in CheckMate 9DW included etiology, macrovascular invasion, extra-hepatic spread, and baseline alpha-fetoprotein (AFP) levels but conveniently avoided control arm heterogeneity. At present, many oncologists prefer lenvatinib in view of its better response and favorable time to tumor progression rates. Hence, the trial mirrors actual prescribing patterns and improves external validity. As lenvatinib has superior performance compared to sorafenib, a lenvatinib-heavy arm raises the bar and reduces the apparent benefit of nivolumab-ipilimumab. Hence, the control arm selection is more of a design nuance than a bias and further strengthens the OS benefit reported. Furthermore, one could argue that the comparator is not atezolizumab-bevacizumab which is regarded as the standard worldwide. However, it should be kept in mind that this is a contextual issue rather than a fatal flaw as TKIs were still accepted as standard comparators when the trial was initiated. IMbrave-150 and CheckMate 9DW differ in patient selection, geography, etiology, and subsequent treatments. Hence, indirect comparisons between the two regimens should be fraught with caution.

The dosage of nivolumab (1 mg/kg) and ipilimumab (3 mg/kg) is different from the usual doses employed in landmark dual ICI trials such as Checkmate 214^[6] and Checkmate 9LA^[7] in metastatic renal cell and non-small cell lung cancer, respectively. Dosing in CheckMate 9DW was informed by initial findings from the CheckMate 040 study, which recently reported its 5-year update.^[8] The data showed that reverse dosing (nivolumab 3 mg/kg and ipilimumab 1 mg/kg) was associated with relatively poor outcomes. Importantly, while higher doses of anti-CTLA-4 may enhance efficacy, they were also linked to increased toxicity in CheckMate 040.

TOXICITY?? A CAUSE FOR CONCERN

Compared to STRIDE, CheckMate 9DW demonstrated a higher incidence of grade 3/4 immune-related adverse events, with 28.9% of patients requiring high-dose steroids versus 20.1% in the HIMALAYA trial. Any grade adverse events occurred in 84% of patients in the experimental arm versus 91% in the lenvatinib or sorafenib group. Immune-related pruritus (27%) and rash (17%) accounted for the majority of grade ½ adverse events. Transaminitis was the most common grade ¾ adverse event. Despite this, reassuring quality of life outcomes were presented at the recent European society of Medical Oncology Gastro-intestinal (ESMO GI) congress, with patients reporting less bother due to side effects from treatment with nivolumab/ipilimumab compared with lenvatinib/sorafenib as per the validated Functional Assessment of Cancer Therapy–Hepatobiliary (FACT-Hep) GP5 scores. In addition, treatment-related deaths and discontinuations due to adverse events were more frequent in CheckMate 9DW.^[2] The reasons for death in the dual ICI arm included immune-related hepatitis (4 patients), liver failure (3 patients), hepatic insufficiency (1patient), immune-related colitis (1 patient), autoimmune hemolytic anemia (1 patient), and dysautonomia (1 patient). Real-world experience and toxicity patterns with 3mg/kg ipilimumab are needed before we can routinely implement this strategy in our clinics. Treatment-related adverse events (TRAE) with potential immune etiology generally occurred early in treatment, mostly within the 1^st month, and were managed using established algorithms. Treatment discontinuations due to TRAE (18% of patients) and use of high-dose steroids for management of immune-related adverse events did not appear to negatively impact the efficacy of nivolumab/ipilimumab. Likewise, exploratory data from HIMALAYA suggest that patients who developed any immune-related adverse events had better survival.^[9] To guide treatment decisions in clinical practice, survival outcomes from CheckMate 9DW will likely be compared with those from IMbrave150, which remains the preferred regimen among oncologists worldwide due to its favorable balance of efficacy and tolerability. The 30% ORR and 19.2 months median OS with atezolizumab-bevacizumab may seem inferior, but the different populations included in both trials have to be considered. CheckMate 9DW excluded patients with main portal vein invasion (VP4) and other high-risk features and had a greater proportion of patients with early-stage disease [Barcelona Clinic Liver Cancer (BCLC) A/B: 26%]. This contrasts with IMBRAVE-150, where 82% of trial participants had BCLC stage C disease, and 24% of patients had high-risk features including VP4, biliary infiltration, and high tumor burden (>50% liver involvement).^[10] In addition, the incidence of macrovascular invasion was more prevalent in IMbrave-150 (38%), compared to CheckMate 9DW (23%). An exploratory analysis of IMbrave150, excluding patients with VP4 and macrovascular invasion, demonstrated a median OS comparable to that observed in CheckMate 9DW in a similar subgroup of patients.^[11]

DOES ETIOLOGY OF HCC FACTOR INTO DECISION-MAKING?

Differential response to therapy by etiology is a matter of debate in HCC. This stemmed from initial observations in the REFELCT trial where sorafenib and lenvatinib had a trend toward improved survival in Hepatitis C and B, respectively.^[12] Furthermore, subgroup analysis from IMBRAVE-150 indicated a better performance for atezolizumab-bevacizumab in viral compared to non-viral etiologies.^[10] It was later deduced that the hazard ratio for atezolizumab-bevacizumab in non-viral subgroups may have exceeded unity due to the strong performance of sorafenib in these subsets. In CheckMate 9DW, like in HIMALAYA, the efficacy of nivolumab/ipilimumab was preserved across etiologies. Furthermore, a recent updated meta-analysis confirms the benefit of ICI-based therapies in non-viral subgroups.^[13] Hence, etiology should not be considered while making therapeutic decisions in advanced HCC.

BASELINE LIVER FUNCTION: AN IGNORED ENTITY!!

The degree of liver function as determined by albumin-bilirubin (ALBI) grade is a major predictor of survival outcomes in HCC treated with immunotherapy and targeted therapy. An exploratory analysis from IMbrave-150 showed that the median OS with atezolizumab-bevacizumab was not estimable (NE) in ALBI grade 1, but there was only a numerical PFS benefit with no OS benefit compared to sorafenib in ALBI grade 2.^[14] Furthermore, real-world experiences indicate that there can be significant deteriorations in the ALBI grade within the first 3 months of initiation of atezolizumab-bevacizumab, which may be negatively associated with OS.^[15] On the contrary, similar analysis from HIMALAYA has shown that the benefit of STRIDE is consistent across all ALBI grades.^[16] Alongside this, there is data to show that durvalumab-tremelimumab treatment can maintain liver function with no major deteriorations in ALBI grade.^[17] Unpublished data from CheckMate 9DW shows that nivolumab/ipilimumab has an impressive OS of 35.4 months in ALBI grade 1, compared to 16.9 months in ALBI grade 2/3. Most importantly, the hazard ratios in both groups were similar (0.75). Outside of a trial setting, the toxicities need to be taken into serious consideration while contemplating this regimen in higher ALBI grades. As discussed above, the early crossing of survival curves and the excess deaths within the first 6 months observed with pure ICI combinations in HCC are concerning. To explore this further, trial investigators conducted a competing risk analysis, which indicated that many of these early deaths were due to non-treatment-related causes. Notably, no excess mortality was observed in the ALBI-2 subgroup, despite several events being liver-related.

HCC: A BIOMARKER-FREE ZONE

At present, there is no validated predictive biomarker available to guide patient selection or predict response to immunotherapy in HCC. Unlike in other solid tumors, expression of markers such as PD-L1, microsatellite instability, and tumor mutational burden is very rare and has not shown consistent predictive value in HCC and therefore cannot be reliably used in clinical practice.^[18] In the setting of atezolizumab-bevacizumab, emerging evidence indicates that certain clinical and biochemical markers may provide some prognostic information. For example, patients who exhibit an early decline in serum AFP levels at 5 weeks and those with a low CRAFITY score have been observed to achieve more favorable outcomes.^[18] However, beyond their prognostic significance, these markers will likely also hold predictive value for response to other ICI-based combinations, given the likely overlap in mechanisms of response across these therapies.

THE JOURNEY AHEAD

Future trials are exploring other potential checkpoint inhibitors in HCC. Efforts to target LAG3 have so far been unsuccessful with negative results from Relativity-106 which tested relatlimab in combination with nivolumab and bevacizumab in HCC.^[19] Similarly, the development of tebotelimab, a bispecific antibody against LAG3 and PDL1 was stopped by the sponsors due to strategic reasons.^[20] The encouraging results from MORPHEUS-Liver^[21] have paved the way for the confirmatory IMBRAVE152/SKYSCRAPER-14 trial testing the addition of TIGIT inhibitor tiragolumab to atezolizumab-bevacizumab backbone.^[22] Furthermore, AdvanTIG-206 is exploring the combination of anti-TIGIT monoclonal antibody ociperlimab plus tislelizumab and BAT 1706 (VEGF inhibitor) as first-line treatment for advanced HCC.^[23] AMBER PART 2F is analyzing anti-TIM-3 cobolimab in combination with dostarlimab in treatment-naïve patients.^[24] The PRODIGE group, on the other hand, is testing the addition of ipilimumab to atezolizumab-bevacizumab in the TRIPLET-HCC trial.^[25] Expression of fibroblast growth factor 19 (FGF) denotes poor prognosis in HCC, and therapies targeting the same are in clinical development.^[26] Encouraging results for irpagratinib (FGFR4 inhibitor) in combination with atezolizumab as first-line treatment for advanced HCC with FGF19 overexpression were presented at the latest ESMO GI congress.^[27] Cellular therapies have also forayed into HCC, with tumor-associated antigen glypican-3 (GPC3) identified as a promising target. Phase 1 studies are currently exploring the role of GPC3-targeted CAR-T cell therapy in advanced HCC.^[28]

CONCLUSION

In essence, there is no single right or wrong option for the treatment of advanced HCC. In contrast to other malignancies, management of HCC requires a bespoke and nuanced approach. With the data from CheckMate 9DW, it is clear that nivolumab/ipilimumab will not replace the existing regimens; it just adds to the ever-expanding therapeutic arsenal. It has the potential to be the treatment of choice for young patients with a good performance status, symptomatic individuals with a heavy tumor burden, needing a quick time to response, and for those with uncontrolled hypertension and other cardiac co-morbidities. Furthermore, in a highly selected population, clinicians should take note of the high rate of treatment-related deaths. The reported deaths are of particular concern in a highly selected population with no portal vein invasion and over a quarter of patients with BCLC A/B. Hence, as the treatment of unresectable HCC broadens, nivolumab-ipilimumab might represent a double-edged sword and the onus is on clinicians to choose the right patients for this therapy.