COSMIC-312: lessons learned from another combination immune checkpoint and multi-kinase inhibitor trial
Study background
The therapeutic landscape of advanced hepatocellular carcinoma (HCC) has radically evolved over recent years. Combination strategies with immune checkpoint inhibitors (ICIs) and vascular endothelial growth factor receptor (VEGFR) targeting tyrosine kinase inhibitors (TKIs) have garnered great interest, especially because of previous data demonstrating an augmentation in antitumour immune response. First line treatment remains that of atezolizumab and bevacizumab, a combination which improved progression-free survival (PFS) and overall survival (OS) by unprecedented magnitude as compared to monotherapy multi-kinase inhibitors (MKIs) (1), as well as other combination therapy options such as durvalumab/tremelimumab (2) and ipilimumab/nivolumab (3). It is postulated that combining immune and vascular endothelial growth factor (VEGF) pathway inhibitors may promote an immune-permissive environment and thereby enhance the response from ICIs. COSMIC-312 was therefore designed to compare patients who received MKI alone vs. the synergistic combination of cabozantinib and atezolizumab so as to better assess the clinical efficacy and safety as well as treatment outcomes using such strategies. Cabozantinib specifically targets VEGFR-2, similar to other TKI, but stands out by additionally targeting AXL and c-MET kinases involved in sorafenib resistance.
Trial design and research findings
The final results of COSMIC-312, a multi-centre international open-label randomised phase 3 trial was recently reported in Lancet Gastroenterology and Hepatology (4). The trial recruited patients with advanced HCC naïve to prior systemic oncological therapy and with good liver function (Child-Pugh class A with no documented hepatic encephalopathy or clinically meaningful ascites within 6 months), and randomised them to receive to combination cabozantinib 40 mg daily plus intravenous atezolizumab 1,200 mg every 3 weeks, oral cabozantinib 60 mg daily, or oral sorafenib 400 mg twice daily. Randomisation was stratified by disease aetiology, geographic region, and presence of extrahepatic spread or macrovascular invasion. The study was designed to assess dual prior endpoints of PFS (assessed by a blinded independent radiology committee) and OS between the combination treatment arm and sorafenib arm.
Thirty percent of the enrolled patients had an aetiology of hepatitis B virus (HBV), 31% had hepatitis C virus (HCV) and 39% had non-viral aetiologies. Extrahepatic disease or macrovascular invasion was present in most patients (69%) and most were from regions outside of Asia (71%).
Selected key efficacy and toxicity endpoints of the study are summarized in Table 1. Although median progression free survival was improved (6.9 months in the combination group vs. 4.3 months for sorafenib group, hazard ratio (HR) 0.74 [99% confidence interval (CI): 0.56–0.97], there was no improvement in median overall survival (mOS) between the two arms [16.5 months in combination group vs. 15.5 months in sorafenib group, HR 0.98 (96% CI: 0.78–1.24)]. In pre-planned subgroup analyses however, patients with HBV etiology had improved PFS and OS with the combination arm vs. the sorafenib arm.
Table 1
| Variable | Combination arm | Sorafenib arm | Cabozantinib arm |
|---|---|---|---|
| Efficacy endpoints—overall population | |||
| mPFS | 6.9 m (99% CI: 5.7–8.2) | 4.3 m (99% CI: 2.9–6.1); HR 0.74 (99% CI: 0.56–0.97) | 5.8 m (99% CI: 5.4–8.2) |
| mOS | 16.5 m (96% CI: 14.5–18.7) | 15.5 m (96% CI: 12.2–20.0); HR 0.98 (96% CI: 0.78–1.24) | 14.5 m (96% CI: 12.7–20.0) |
| ORR | 13.0% (95% CI: 9.9–16.5%) | 4.6% (95% CI: 2.2–8.3%) | 7.4% (95% CI: 4.1–12.2%) |
| DCR (%) | 78 | 65 | 84 |
| Efficacy endpoints—HBV subgroup | |||
| mPFS (HBV patients only) | 6.8 m (95% CI: 5.6–7.2) | 2.7 m (95% CI: 1.6–4.1); HR 0.52 (95% CI: 0.37–0.75) | – |
| mOS (HBV patients only) | 20.4 m (95% CI: 16.9–24.2) | 14.9 m (95% CI: 7.0–19.1); HR 0.63 (95% CI: 0.42–0.95) | – |
| ORR (HBV patients only) | 17.3% (95% CI: 11.2–25.0%) | 0.0% (95% CI: 0.0–5.6%) | – |
| Toxicity endpoints | |||
| Dose reductions for AEs (%) | 62 | 43 | 66 |
| G3–4 TEAEs (%) | 66 | 48 | 57 |
| Others | |||
| Subsequent systemic anti-cancer therapy (%) | 26 | 42 | 35 |
| Subsequent MKI use (%) | 19 | 26 | 23 |
| Subsequent ICI use (%) | 6 | 22 | 16 |
| Median daily dose of MKI (mg) | 23.2 (IQR, 17.2–32.7) | 614.0 (IQR, 444.8–790.4) | 38.7 (IQR, 24.8–50.4) |
AEs, adverse events; CI, confidence interval; DCR, disease control rate; G, grade; HBV, hepatitis B virus; HR, hazard ratio; ICI, immune checkpoint inhibitor; IQR, interquartile range; MKI, multi-kinase inhibitors; m, months; mOS, median overall survival; mPFS, median progression-free survival; ORR, overall response rate; TEAEs, treatment-emergent adverse events.
What have we learned from this trial
Two other phase 3 combination ICI and MKI trials have been conducted and reported, the CARES-310 study (5), comparing combination camrelizumab and rivoceranib to sorafenib, and LEAP-002 (6), comparing combination pembrolizumab and lenvatinib to lenvatinib. Together with these new results from COSMIC-312, the three trials teach us several lessons and provides food-for-thought for the community moving forward, as summarised in Table 2.
Table 2
| Trial | Patient population | Key efficacy endpoints | TRAEs |
|---|---|---|---|
| CARES-310 (NCT03764293): camrelizumab + rivoceranib vs. sorafenib | • HBV: 75% | • mPFS 5.6 vs. 3.7 m (HR 0.52, 95% CI: 0.41–0.65) | • TRAEs: 97% vs. 93% |
| • BCLC-C: 74% | • mOS 22.1 vs. 15.2 m (HR 0.62, 95% CI: 0.49–0.80)* | • G3-4: 81% vs. 52% | |
| • AFP >400 ng/mL: 36% | • Dose reduction: 47% vs. 32% | ||
| LEAP-002 (NCT03713593): pembrolizumab + lenvatinib vs. lenvatinib | • HBV: 49% | • mOS 21.2 vs. 19.0 m (HR 0.84, 95% CI: 0.71–1.00) | • TRAEs: 96% vs. 96% |
| • BCLC-C: 78% | • mPFS 8.2 vs. 8.0 m (HR 0.87, 95% CI: 0.73–1.02) | • G3–4: 62% vs. 57% | |
| • AFP >400 ng/mL: 30% | • Dose reduction: 22% vs. 26% | ||
| COSMIC-312 (NCT03755791): atezolizumab + cabozantinib vs. sorafenib | • HBV: 30% | • mPFS 6.9 vs. 4.3 m (HR 0.74, 95% CI: 0.56–0.97) | • TRAEs: 99% vs. 99% |
| • BCLC-C: 69% | • mOS 16.5 vs. 15.5 m (HR 0.98, 95% CI: 0.78–1.24) | • G3–4: 66% vs. 48% | |
| • AFP >400 ng/mL: 35% | • Dose reduction: 66% vs. 48% |
Italicized are statistically significant results. *, interim analysis. AFP, alpha-fetoprotein; BCLC-C, Barcelona Clinic Liver Cancer Stage C; CI, confidence interval; G, grade; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazard ratio; ICI, immune checkpoint inhibitor; MKI, multi-kinase inhibitors; mOS, median overall survival; mPFS, median progression-free survival; TRAEs, treatment-related adverse events.
Firstly, ICI and MKI combination in the COSMIC study demonstrated increased toxicity vs. sorafenib, though it appears that the toxicities were likely driven by cabozantinib rather than the use of a combination strategy, and also because of generally lower toxicity levels with sorafenib. The adverse events and safety data were comparable to that of the CELESTIAL trial of single agent cabozantinib vs. placebo, and the most common grade 3–4 treatment-related adverse events such as hypertension, transaminitis, and palmar-plantar erythrodysesthesia was comparable between the combination and the single agent cabozantinib arms in this study. In both CARES-310 and COSMIC-312, patients treated in the combination arm were more likely to require dose reductions in treatment compared to patients receiving monotherapy MKI. In COSMIC-312, this was despite the starting dose of cabozantinib already being lower (40 mg in the combination vs. 60 mg in the control arm). Patients required significant dose reduction during treatment with an average daily dose of cabozantinib at 24.2 mg, and dose modifications were required in both the combination and monotherapy arms.
It is difficult to conclude if the relative dose intensity (RDI) of cabozantinib determines its therapeutic efficacy. Previous reports have not shown a clear relationship between RDI and clinical efficacy for cabozantinib in patients with advanced HCC patients (7,8), unlike that for lenvatinib (9,10), and many of these studies used either very low or very high cut-offs to dichotomise these two groups of patients. For instance, two previous studies used rather disparate cut-offs—a Japanese study (8) used a low cut-off for RDI over 4 weeks of ≥40% (defined as high exposure to the drug), whereas an Italian study (7) used a high cut-off for RDI over 8 weeks of ≥90%, making it difficult to compare the outcomes, therapeutic thresholds, and toxicities across these arbitrarily assigned RDI cut-offs. Additionally, MKI-related toxicities are presently more adequately recognised and effectively mitigated by practising clinicians these days.
In addition, overall response rate and median duration of response appears lower in COSMIC-312. In comparison, in the treatment of metastatic renal cell carcinoma, MKI-ICI combinations still demonstrate an OS and PFS benefit despite the occurrence of grade 3 or more adverse events in more than half of the patient population (11). However, MKI-ICI combinations like cabozantinib and atezolizumab failed to demonstrate any OS benefit, and that of pembrolizumab and lenvatinib (as per the phase III LEAP-002 study) (6) unfortunately also missed the threshold for significance. Phase II or larger phase I dose expansion cohorts are likely still crucial in ICI-TKI combination studies for patients with HCC, as the tolerated dose in this population group may be different from that in other tumour types receiving the same combination. Further investigation into the impact of dose-related parameters on efficacy endpoints may need to be explored.
Secondly, the three trials again highlight the perennial question of which endpoints are relevant, and the correlation between PFS and OS. In COSMIC-312 the lack of OS benefit despite PFS benefit, may be at least partially attributable to post-trial exposure to other systemic therapy options (20% of patients in the combination arm and 37% in the sorafenib arm proceeded on to receive second line systemic treatment), with more patients in the sorafenib arm getting other MKIs, as well as a significant proportion receiving subsequent ICI therapy. This brings into question whether sequential treatment with exposure to different systemic agents is sufficient for some patients, instead of upfront combination treatment. This in turn also highlights the need for more granularity on the impact of upfront combination treatment on liver function preservation and quality of life (12), which may in turn affect the ability of a patient to be exposed to as many active agents as possible during one’s cancer journey; and if so, also paves the way for subsequent trials to consider exploring de-escalation strategies.
Thirdly, although COSMIC-312 did not show an improvement in OS in the overall population, subgroups such that those with HBV aetiology or alpha-fetoprotein (AFP) >400 ng/mL seemed to benefit more. In the same vein, the CARES-310 trial was positive for survival endpoints, and included a much larger proportion of patients with HBV aetiology. A large real-world study examined and emphasised the effect of etiology on therapeutic efficacy in advanced unresectable HCC patients treated with atezolizumab/bevacizumab or lenvatinib. It was purported that there was a probable advantage of lenvatinib in the population with non-alcoholic steatohepatitis (NASH) (13). A meta-analysis of IMBrave150, Checkmate-459 and KEYNOTE 240, as well as a subgroup analysis of the IMBrave150 cohort, suggested a lower efficacy of atezolizumab/bevacizumab vs. sorafenib in the non-viral population (14). Subsequent trials will have to further explore such clinical parameters as potential biomarkers, such that one day we may truly be able to rank or prioritise systemic treatment options more precisely.
Lastly, each ICI and MKI is different and while the above points consider the ICI and MKI combination as a strategy, each unique pairing is likely to have its additional benefits, toxicities and considerations. There is still much we do not know yet, such as the best sequencing of these drug as individual options, much less as combination treatments. Better treatment selection biomarkers also need to be developed, be it baseline or on-treatment parameters (15) to help us predict and prognosticate better. Currently, there are no established biomarkers to predict response to immunotherapy or MKI, but we are continuing to probe this space—local data recently published in collaboration with our Korean colleagues investigated exploratory single-cell RNA sequencing analyses of peripheral blood mononuclear cells in patients treated with regorafenib and nivolumab, also an ICI-MKI combination. This analysis demonstrated that patients who had durability of response exhibited T cell receptor repertoire diversification and enrichment of genes in MK167+ proliferating CD8+ T cells. Such understanding is especially relevant now that we also have other ICI and anti-VEGF monoclonal antibody combination or combination ICI options available and choosing the optimal first line and subsequent therapies can prove challenging.
Conclusions
Whilst COSMIC-312 will not change the current treatment algorithm for systemic therapy in patients with aHCC, the results reinforce the concept that HCC is itself a heterogeneous disease, with patients of varying liver function, aetiology and disease pace. “The worst form of inequality is to try to make unequal things equal.” said Aristotle. Moving forward as a community, given the increasing number of systemic treatment options as well, we have the option for personalisation of treatment. Our next step is perhaps to figure out how to better tease out how select the best option for an individual based on both disease and patient factors, to find equipoise for each individual between efficacy and toxicity. Further therapeutic advances and interrogation of combination strategies will come from analysis of tumour tissue to capture such heterogeneity, and hone a more robust understanding of epigenetic modifications, molecular classifications, and relevant biomarkers that predict immune response. This may be through clinical or translational predictive or prognostic biomarkers, and re-exploring sequential use of therapies, de-escalation strategies and combinations with locoregional options, to not just improve efficacy but also to reduce toxicity.
Acknowledgments
None.
Footnote
Provenance and Peer Review: This article was commissioned by the editorial office, AME Clinical Trials Review. The article has undergone external peer review.
Peer Review File: Available at https://actr.amegroups.com/article/view/10.21037/actr-24-62/prf
Funding: None.
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://actr.amegroups.com/article/view/10.21037/actr-24-62/coif). J.J.X.L. reports grants from the NCCS Research Fund, honoraria for presentations from Sirtex and Eisai, and has participated on advisory boards for Beigene, Eisai, Servier, Roche, and Taiho. S.Y.L. reports support for attending meetings from Roche. D.T. reports grants from NMRC (CS-IRG-MOH–001231-00), honoraria for presentations from AstraZeneca, Roche, Eisai, and has participated in advisory boards for AstraZeneca, Eisai, GSK. S.P.C. reports honoraria for presentations from Pierre-Fabre, AstraZeneca, Roche, Juniper Biologics, and Amgen, as well as support for attending meetings from Taiho and AstraZeneca. The authors have no other conflicts of interest to declare.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.
References
- Finn RS, Qin S, Ikeda M, et al. Atezolizumab plus Bevacizumab in Unresectable Hepatocellular Carcinoma. N Engl J Med 2020;382:1894-905. [Crossref] [PubMed]
- Abou-Alfa GK, Lau G, Kudo M, et al. Tremelimumab plus Durvalumab in Unresectable Hepatocellular Carcinoma. NEJM Evid 2022;1:EVIDoa2100070.
- Galle P R, Decaens T, Kudo M, et al. Nivolumab (NIVO) plus ipilimumab (IPI) vs lenvatinib (LEN) or sorafenib (SOR) as first-line treatment for unresectable hepatocellular carcinoma (uHCC): First results from CheckMate 9DW. J Clin Oncol 2024;48:LBA4008. [Crossref]
- Yau T, Kaseb A, Cheng AL, et al. Cabozantinib plus atezolizumab versus sorafenib for advanced hepatocellular carcinoma (COSMIC-312): final results of a randomised phase 3 study. Lancet Gastroenterol Hepatol 2024;9:310-22. [Crossref] [PubMed]
- Qin S, Chan SL, Gu S, et al. Camrelizumab plus rivoceranib versus sorafenib as first-line therapy for unresectable hepatocellular carcinoma (CARES-310): a randomised, open-label, international phase 3 study. Lancet 2023;402:1133-46. [Crossref] [PubMed]
- Llovet JM, 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;24:1399-410. [Crossref] [PubMed]
- Tovoli F, Dadduzio V, De Lorenzo S, et al. Clinical impact of relative dose intensity of cabozantinib during the first 8 weeks and of subsequent dose reductions in patients with unresectable hepatocellular carcinoma. Digestive and Liver Disease 2023;55:S224-5. [Crossref]
- Nobusawa T, Tsuchiya K, Yasui Y, et al. The impact of relative dose intensity on overall and progression-free survival during cabozantinib therapy for unresectable hepatocellular carcinoma. JCO 2023;41:569. [Crossref]
- Kirino S, Tsuchiya K, Kurosaki M, et al. Relative dose intensity over the first four weeks of lenvatinib therapy is a factor of favorable response and overall survival in patients with unresectable hepatocellular carcinoma. PLoS One 2020;15:e0231828. [Crossref] [PubMed]
- Takahashi A, Moriguchi M, Seko Y, et al. Impact of Relative Dose Intensity of Early-phase Lenvatinib Treatment on Therapeutic Response in Hepatocellular Carcinoma. Anticancer Res 2019;39:5149-56. [Crossref] [PubMed]
- Rini BI, Plimack ER, Stus V, et al. Pembrolizumab plus Axitinib versus Sunitinib for Advanced Renal-Cell Carcinoma. N Engl J Med 2019;380:1116-27. [Crossref] [PubMed]
- Starzer AM, Wolff L, Popov P, et al. The more the merrier? Evidence and efficacy of immune checkpoint- and tyrosine kinase inhibitor combinations in advanced solid cancers. Cancer Treat Rev 2024;125:102718. [Crossref] [PubMed]
- Casadei-Gardini A, Rimini M, Tada T, et al. Atezolizumab plus bevacizumab versus lenvatinib for unresectable hepatocellular carcinoma: a large real-life worldwide population. Eur J Cancer 2023;180:9-20. [Crossref] [PubMed]
- Pfister D, Núñez NG, Pinyol R, et al. NASH limits anti-tumour surveillance in immunotherapy-treated HCC. Nature 2021;592:450-6. [Crossref] [PubMed]
- Kim HD, Jung S, Lim HY, et al. Regorafenib plus nivolumab in unresectable hepatocellular carcinoma: the phase 2 RENOBATE trial. Nat Med 2024;30:699-707. [Crossref] [PubMed]
Cite this article as: Lee JJX, Lee SY, Tai D, Choo SP. COSMIC-312: lessons learned from another combination immune checkpoint and multi-kinase inhibitor trial. AME Clin Trials Rev 2025;3:30.
