Frontline Cabozantinib/Atezolizumab Does Not Improve OS Over Sorafenib in Advanced HCC

The combination of cabozantinib and atezolizumab was not found to result in an improvement nor a detriment in overall survival vs sorafenib when used in previously untreated patients with advanced hepatocellular carcinoma, according to data from the phase 3 COSMIC-312 trial.

The combination of cabozantinib (Cabometyx) and atezolizumab (Tecentriq) was not found to result in an improvement nor a detriment in overall survival (OS) vs sorafenib (Nexavar) when used in previously untreated patients with advanced hepatocellular carcinoma (HCC), according to data from the phase 3 COSMIC-312 trial (NCT03755791).1

Based on these findings, Exelixis, Inc. shared that they do not intend to submit a supplemental new drug application to the FDA for the combination in this indication. However, they do plan to share full results at a future medical conference.

“Exelixis has a longstanding commitment to patients with liver cancer, exemplified by the 2019 approval of [cabozantinib] for previously treated advanced liver cancer, and we remain steadfast in our journey to further therapies for this and other difficult-to-treat cancers,” Vicki L. Goodman, MD, executive vice president of Product Development and Medical Affairs and chief medical officer of Exelixis, Inc., stated in a press release. “We are grateful to the investigators and patients who participated in the COSMIC-312 trial and contributed greatly to this research.”

COSMIC-312 set out to enroll approximately 840 patients with advanced HCC that was BCLC stage B or C and who were not amenable to curative treatment of locoregional therapy.2 Patients needed to have Child-Pugh A disease, an ECOG performance status of 0 or 1, and measurable disease per RECIST v1.1 criteria. Patients could not have previously received systemic therapy.

Study participants were randomized 2:1:1 to receive cabozantinib at 40 mg once daily plus atezolizumab at 1200 mg every 3 weeks, single-agent sorafenib at 400 mg twice daily, or cabozantinib monotherapy at 60 mg once daily. Treatment was continued until loss of clinical benefit or unacceptable toxicity.

Key stratification factors included disease etiology (hepatitis B virus, hepatitis C virus, or non-viral), region (Asian or other), and the presence of extrahepatic disease and/or macrovascular invasion (yes or no).

The co-primary end points of the trial were progression-free survival (PFS) with cabozantinib plus atezolizumab vs sorafenib in the primary intent-to-treat (ITT) population and OS in the ITT population. A key secondary end point was PFS with cabozantinib vs sorafenib in the ITT population.

Once 257 PFS events occurred in the PITT population, a final PFS analysis of the doublet vs sorafenib, as well as an interim OS analysis were conducted. At this time point, an interim PFS analysis of single-agent cabozantinib vs sorafenib was also slated to be performed. After 368 OS events occurred in the ITT population, a final OS analysis was conducted. A final PFS analysis was to be performed when a total of 283 PFS events occurred in the ITT population.

A total of 1283 patients were screened and 837 patients underwent randomization; of these patients, 432 comprised the cabozantinib/atezolizumab arm (n = 250, PITT population; n = 432, ITT population), 217 comprised the sorafenib monotherapy arm (n = 122, PITT population; n = 217, ITT population), and 188 comprised the cabozantinib monotherapy arm (n = 188, ITT population).

At a data cutoff of March 8, 2021, 429 patients received treatment with cabozantinib plus atezolizumab, 207 received treatment with single-agent sorafenib, and 188 received cabozantinib monotherapy. The median follow-up in the PITT population was 15.8 months (range, 12.8-27.0), and the median follow-up in the ITT population was 13.3 months (range, 6.4-27.0).

In the ITT population, the median age of patients across the arms was 64.0 years, and the majority were male, were enrolled in a region other than Asia, were not Asian, had an ECOG performance status of 0, and had BCLC stage C disease. All patients across the arms had Child-Pugh grade A disease.

In the PITT population, at a median follow-up of 15.8 months (range, 12.8-27.0), the median PFS achieved with cabozantinib plus atezolizumab (n = 250) was 6.8 months (99% CI, 5.6-8.3) vs 4.2 months (99% CI, 2.8-7.0) with sorafenib (n = 122; HR, 0.63; 99% CI, 0.44-0.91); P = .0012).

In the ITT population, at a median follow-up of 13.6 months (range, 6.4-27.0), the median OS with the doublet (n = 432) was 15.4 months (96% CI, 13.7-17.7) vs 15.5 months (96% CI, 12.1–not evaluable) with sorafenib (n = 217; HR, 0.90; 96% CI, 0.69-1.18; P = .438).

When broken down by disease etiology, the median PFS with the doublet vs sorafenib in the subset of patients with HBV (n = 109) within the PITT population was 6.7 months vs 2.7 months, respectively (HR, 0.46; 95% CI, 0.29-0.73). In the subset of those with HCV (n = 105), the median PFS was 7.9 months with the doublet and 5.6 months with the monotherapy (HR, 0.64; 95% CI, 0.38-1.09). In the non-viral subset (n = 158), the median PFS was 5.8 months and 7.0 months, respectively (HR, 0.92; 95% CI, 0.60-1.41).

The median OS with cabozantinib/atezolizumab vs sorafenib in the subset of patients with HBV (n = 191) within the ITT population was 18.2 months vs 14.9 months, respectively (HR, 0.53; 95% CI, 0.33-0.87). In the subset of patients with HCV (n = 203), the median OS was 13.6 months in the investigative arm and 14.0 months in the control arm (HR, 1.10; 95% CI, 0.72-1.68). In the non-viral subgroup (n = 255), the median OS was 15.2 months vs not reached, respectively (HR, 1.18; 95% CI, 0.78-1.79).

Additionally, within the ITT population, at a median follow-up of 13.1 months (range, 6.4-27.0), the median PFS achieved with single-agent cabozantinib (n = 188) was 5.8 months (99% CI, 5.4-8.2) vs 4.3 months (99% CI, 2.9-6.1) with sorafenib monotherapy (n = 217; HR, 0.71; 99% CI, 0.51-1.01; P = .0107).

In the ITT population, cabozantinib plus atezolizumab (n = 432) produced an objective response rate (ORR) of 11% (95% CI, 8.1%-14.0%) vs 3.7% (95% CI, 1.6%-7.1%) with sorafenib monotherapy (n = 217), and 6.4% (95% CI, 2.2%-11.0%) with cabozantinib monotherapy (n = 188).

In the doublet arm, 0.2% experienced a complete response as their best overall response, 11% achieved a partial response, and 67% had stable disease. Fourteen percent of patients experienced disease progression, and 7.9% were not evaluable. The median time to objective response was 4.0 months (range, 1.3-10.0), and the median duration of response was 10.6 months (95% CI, 7.1-12.7). The disease control rate in this arm was 78%.

Regarding safety, treatment-related adverse effects (TRAEs) that were grade 3 in severity occurred in 51% of those who received the doublet (n = 429), 30% of those given single-agent sorafenib (n = 207), and 52% received cabozantinib monotherapy (n = 188); grade 4 TRAEs were experienced by 2.8%, 1.9%, and 3.2% of patients, respectively. TRAEs that were grade 5 occurred in 1.9%, 0.5%, and 0.5% of patients, respectively.

Moreover, 6.1% of patients in the doublet arm discontinued both cabozantinib and atezolizumab due to TRAEs. Fourteen percent of patients in the doublet arm and 8.5% of patients in the single-agent cabozantinib arm discontinued cabozantinib and/or atezolizumab because of TRAEs. Additionally, 7.7% of patients in the single-agent sorafenib arm discontinued treatment with sorafenib.

The most common grade 3 or higher TRAEs experienced in the doublet arm included Palmar-plantar erythrodysesthesia (7.9%), hypertension (7.0%), alanine aminotransferase increased (6.3%), aspartate aminotransferase increased (6.5%), diarrhea (3.5%), fatigue (2.6%), and decreased appetite (0.9%).

Twenty patients on the doublet arm received subsequent systemic anticancer therapy, which included TKIs (14%), VEGF(R)-targeted antibodies (2.1%), immune checkpoint inhibitors (3.9%), or other agents that were not specified (4.9%).


  1. Exelixis announces final overall survival results from phase 3 COSMIC-312 trial of cabozantinib in combination with an immune checkpoint inhibitor in patients with previously untreated advanced liver cancer. News release. Exelixis, Inc.; March 14, 2022. Accessed March 15, 2022.
  2. Kelley RK, Yau T, Cheng A-L, et al. Cabozantinib (C) plus atezolizumab (A) versus sorafenib (S) as first-line systemic treatment for advanced hepatocellular carcinoma (aHCC): results from the randomized phase III COSMIC-312 trial. Ann Oncol. 2021;33(1):114-116. doi:10.1016/j.annonc.2021.10.008