Immunotherapy Drives First-line Systemic Treatment for Advanced HCC

Anthony B. El-Khoueiry, MDAnthony B. El-Khoueiry, MD

Evidence-based decisions have leveraged the use of immunotherapy (IO) combination atezolizumab (Tecentriq) plus bevacizumab (Avastin) for patients who receive a diagnosis of hepatocellular carcinoma (HCC). Approved in 2020, the combination has become a category 1 recommendation in the National Comprehensive Cancer Network guidelines and set the standard for first-line treatment.^1,2

However, decisions may vary for patients who present with unfavorable clinical profiles. This includes those at an elevated risk of bleeding, Child-Pugh status, or contraindications such as autoimmune disease.2 Other contenders in the armamentarium include alternate immunotherapy combinations and single-agent tyrosine kinase inhibitors (TKIs) may hold a role for select patients with advanced HCC, but clearer stratification factors and real-world data are needed to solidify their role in the treatment landscape.

“If a patient is a candidate for IO and does not have contraindications to VEGF therapy, they’re going to get atezolizumab/bevacizumab,” Anthony B. El-Khoueiry, MD, said in a recent OncLive Peer Exchange^®. “If they’re not a candidate for combination—and that’s in the eye of the beholder to some extent, maybe liver function is weighing in—you can use single-agent sorafenib, or single-agent anti–PD-1/PD-L1, and if they have IO contraindications, they’re going to get a TKI. With the new regimens that come on board we’ll see the data and how positive or negative they are.”

El-Khoueiry and a panel of liver cancer experts held a case-based discussion centered on updates in the systemic treatment of patients with advanced HCC and the clinical factors that may influence decisions that diverge from the standard IO/IO approach.

The Standard of Care in Advanced HCC

To guide the discussion of first-line systemic treatment options, the case of a woman aged 63 years with chronic active hepatitis C infection was referenced. After starting antiviral therapy, she presented with elevations on liver function tests, which prompted a CT scan of the abdomen. Multifocal infiltrative liver lesions were observed in the right lobe with evidence of vascular invasion by the tumor and the patient received a diagnosis of HCC.

“This is a patient who has ChildPugh A cirrhosis, has prototypical BCLC [Barcelona clinic liver cancer] staging with a tumor invading a large blood vessel,” Pierre Gholam, MD, said. “She does not appear to have any significant contraindications to the combination of atezolizumab and bevacizumab and based on the evidence that would be our go-to first-line model.”

Mark Yarchoan, MD, concurred adding, “This is the preferred first-line systemic regimen for individuals who are eligible for it.” Yarchoan cited data from the phase 3 IMbrave150 study (NCT03434379) in which the combination was evaluated against single-agent sorafenib (Nexavar).^1,3 “This was a positive study regarding every end point that we care most about: overall survival [OS], progression-free survival [PFS], objective response rate [ORR], and even quality of life. You really need a reason not to use it in first line.”

Updated data from IMbrave 150, published in 2022, highlighted the prolonged clinically meaningful benefit with the combination in the first line.³ Further, real-world data of the regimen showed that the combination may be administered beyond the criteria of the IMbrave 150 study and support its role as an option for patients with liver impairment including ChildPugh B status.⁴

Among 202 evaluable patients, 154 had ChildPugh A classification and 48 had Child-Pugh B classification. The median OS for all-treated patients was 14.9 months (95% CI, 13.6-16.3) with a 6-month OS rate of 77% and a 12-month OS rate of 60%. When stratified by Child-Pugh classification the median OS for those with A classification was 16.8 months (95% CI, 14.1-23.9) vs 6.7 months (95% CI, 4.3-15.6) for those with B classification. The median PFS of the overall population was 6.8 months (95% CI, 5.2-8.5) and the was 7.6 months (95% CI, 6.2-8.9) for patients with Child-Pugh A and 3.4 months (95% CI, 2.6-4.2) for patients with Child-Pugh B classifications.

Responses, evaluated using RECIST 1.1 criteria, were comparable to the rates seen in IMbrave 150. In the overall population evaluable for response (n = 174), the objective response rate was 25% and was 26% and 21% in the Child-Pugh A (n = 140) and Child-Pugh B (n = 34) populations, respectively.⁴

Addressing the increased risk of bleeding, the investigators of the real-world analysis noted, “the proportion of patients who experienced bleeding events in our study was comparable between patients with Child-Pugh A and patients with Child-Pugh B, not diverging from rates observed in IMbrave150.”⁴

Moderator Tanios S. Bekaii-Saab, MD, asked for considerations that may deter from the use of the combination.

“When combining these agents [there is concern] for risk of bleeding, primarily related to portal hypertension,” Gholam said. “I think one would be well-advised to assess portal hypertension risk in this patient with Child-Pugh A status. We can now do this to some extent noninvasively, although certainly the broader hepatology and medical oncology community has relied on endoscopy to assess for the presence of esophageal or gastric varices. The fact that cross-sectional imaging does not show splenomegaly, abdominal varices, or esophageal varices is reassuring [for this patient]. Depending on where the tumor is located, [liver elastography] may or may not be a good test because, remember, if you have a tumor, this could alter the stiffness of the liver and may give you erroneous information. But that is a possibility. There are other tests for assessment of fibrosis, but, in general, nothing points to limitations in our ability to initiate atezolizumab/bevacizumab. There’s no uncontrolled hypertension and there doesn’t appear to be any underlying autoimmune disease that would be adversely affected by that combination.”

In terms of considering other factors that may point clinicians to an alternative, those with nonviral HCC may benefit from treatment with the TKI lenvatinib (Lenvima).⁵ Data from a prospective study of patients with advanced HCC or intermediate HCC who were not eligible for locoregional therapies showed that those who received atezolizumab plus bevacizumab (n = 190) had inferior outcomes compared with lenvatinib (n = 569). Among all-treated patients, the median OS was 17.8 months (95% CI, 15.8-43.8) with the TKI vs 12.1 months (95% CI, 11.1-16.8) for patients treated with the combination (HR, 0.71; 95% CI, 0.50-1.06; P = .1028). An analysis of patients with nonalcoholic fatty liver disease (ANFLD)/ nonalcoholic steatohepatitis (NASH) showed an OS advantage favoring single-agent TKI treatment. The median OS was 21.2 months (95% CI, 18.4-30.6) with lenvatinib (n = 254) vs 12.1 months (95% CI, 10.0-16.8) with the combination (n = 82; HR, 0.46; 95% CI, 0.25-0.88; P = .0181).⁵

“When sorafenib came out there was an impression that there was an enhanced response in patients who had viral hepatitis,” Gholam said. “In fact, there were studies, some of which I participated in, which showed that perhaps sorafenib has a weak antiviral activity, reduces hepatitis C viral load by perhaps a half log. This has continued as TKIs came to the forefront.”

Gholam pointed to a study from Heikenwälder et al in a paper published in Nature which analyzed outcomes with immunotherapy in patients with NASH.⁶ Investigators noted that unconventional activation of CD8-positive/PD1-positive T cells drive carcinogenesis in the setting of fatty liver disease and, in turn, diminish response to IO therapy. “I think this issue is at the same time important and also poorly understood,” Gholam said. “Does that mean that I would make different treatment decisions if I had a patient in front of me who had NAFLD in terms of deciding what treatment option to give them? Within the parameters that outlined in the discussion, I would still choose the [atezolizumab/bevacizumab] option. But this is an important question to ask because fatty liver disease has now become the leading cause of liver disease worldwide. Hepatitis C is on the decline because it is universally curable.”

Alternative First-Line Combinations are Under Evaluation

HIMALAYA

Alternative combination strategies have been evaluated in the first-line setting vs single-agent TKIs; however there has been no movement to dethrone the approved IMbrave 150 regimen. Specifically, data from the phase 3 HIMALAYA trial (NCT03298451) have set the IO combination of durvalumab (Imfinzi) and tremelimumab on course as a viable option for this patient population.⁷ However, without an approval, it holds a place as a benchwarmer for the time being.

Investigators assessed a regimen known as STRIDE, which includes a single dose of tremelimumab (300 mg) plus durvalumab (1500 mg every 4 weeks). “HIMALAYA was intended to examine the concept of IO/IO combinations looking at the combination of anti–PD-L1 durvalumab with anti-CTL4 tremelimumab,” El-Khoueiry said. “The way tremelimumab was given in this study is with 1 loading dose based on preclinical evidence that the dose could alter the immune microenvironment, and then [treatment could [continue with] durvalumab as a single agent. This loading dose of tremelimumab was a higher dose than normally used at 300 mg compared with the traditional dose of 75 mg.”

Patients were randomly assigned to STRIDE (n = 393), single-agent durvalumab (n = 389), or sorafenib (n = 389). Median OS was 16.43 months (95% CI, 14.16-19.58) with STRIDE vs 13.77 months (95% CI, 12.25-16.13) with sorafenib (HR, 0.78; 96.02% CI, 0.65-0.93; P = .0035). The HR for OS for durvalumab vs sorafenib alone was noninferior (HR, 0.86; 95.67% CI, 0.73-1.03). The OS rates at 36 months were 30.7%, 24.7%, and 20.2%, respectively.

“Some of the strengths of this trial are that it’s really the largest randomized phase 3 study in first-line HCC,” El-Khoueiry said. “It is very mature with long follow-up. For the first time we’re seeing 36-month OS rates. Certainly, at 36 months, the combination still was superior to sorafenib, so those curves diverge and stayed separate for a long time.”

Yarchoan noted that cross-study comparisons between the 2 positive data sets presents challenges. “These are different trials, but when we need to pick a regimen, we do what we have to do, which is try to parse apart signals from the data,” he said. “The strength of IO/IO, as we all know, is the tail of the curve and sure enough, the tail of the curve for durvalumab/tremelimumab is very nice. We don’t have that length of follow-up yet for atezolizumab/bevacizumab, but the response rate is clearly higher.”

In terms of selecting patients for the HIMALAYA regimen vs the IMbrave 150 regimen, Yarchoan said, “If you need a response for someone with a lot of disease where you really need to get some disease control, I think from the available data [it’s atezolizumab plus bevacizumab.” However, he adds that for patients who are “VEGF ineligible—patients with very recent surgery, recent stroke, recent heart attacks, recent bleeds—those are the patients for whom there’s no question that durvalumab/tremelimumab is a compelling regimen.”

El-Khoueiry concurred, “I would follow the same approach where atezolizumab/bevacizumab is the go-to regimen with durvalumab/ tremelimumab being the backup option. We’ll see how that regimen performs in real-world as well. That remains to be seen.”

The HIMALAYA combination was granted a priority review in April 2022 and the FDA is expected to decide on the application in the fourth quarter of 2022.⁸

COSMIC-312

Unlike HIMALAYA, data from the phase 3 COSMIC-312 (NCT03755791) failed to make meaningful gains with the novel combination of cabozantinib (Cabometyx) plus atezolizumab in the first-line setting. Approved as a single agent for patients with HCC who have been previously treated with sorafenib, when added to the immune checkpoint inhibitor, cabozantinib elicited mixed results.^9,10

At a median follow-up of 15.8 months, data showed the combination produced a clinical benefit in PFS; however, that benefit did not extend to OS. Investigators enrolled 837 patients and randomly assigned them to the cabozantinib/atezolizumab arm (n = 432), sorafenib arm (n = 217), or cabozantinib alone (n = 188). The median PFS was 6.8 months (99% CI, 5.6-8.3) with the investigative combination vs 4.2 months (99% CI, 2.8-7.0) with sorafenib (HR, 0.63; 99% CI, 0.44-0.91; P = .0012). At the interim analysis for OS, the median OS was 15.4 months (96% CI, 13.7-17.7) compared with 15.5 months (96% CI, 12.1–not estimable) in the combination and sorafenib groups, respectively (HR, 0.90; 96% CI, 0.69-1.18; P = .44).¹⁰

“COSMIC-312 was a bit of a disappointment to many of us,” Yarchoan said. “The study was positive regarding PFS, which was one of the coprimary end points, but missed the OS endpoint. The OS curves were at least, thus far, overlapping with the sorafenib curves. I’m not quite sure what happened. Cabozantinib is a very active agent in HCC. It’s an agent that I think many of us use in the second line where it performs very well.”

In terms of response, cabozantinib/atezolizumab elicited an objective response of 11% (95% CI, 8.1%-14.2%) compared with 4% (95% CI, 1.6%7.1%) with sorafenib. The median time duration of response was 10.6 months vs 8.8 months, respectively, and median time to progression was 7.0 months vs 4.6 months.¹⁰

“The response rate of 11% is numerically lower than we’ve seen with PD-L1 monotherapy,” Yarchoan added. “Although the challenge here is that the sorafenib arm also had a very low response rate of approximately 4%, which is lower than we’ve seen in other recent studies, which were in the range of 15%. I don’t know if the radiology [was different], if there was stage migration, for example treating patients who had TACE [transarterial chemoembolization] lesions that may be less likely to show true response.”

Yarchoan noted that with other agents showing an OS benefit in the front line, the combination will not move forward. Other studies, including the phase 3 LEAP-002 trial (NCT03713593), evaluating pembrolizumab (Keytruda) in combination with lenvatinib also failed to demonstrate a significant improvement in OS vs lenvatinib alone.¹¹

“It seems unlikely that any of those are going to displace the current standard,” BekaiiSaab said. “As exciting as it is to see another phase 3 trial, I don’t know if it’s going to break that ceiling.”

TKIs Still Hold a Role

Despite the combination therapies outperforming single-agent sorafenib in phase 3 trials, there is still a place for TKIs for select patient populations. “We talk about IO/IO combinations, but we still have other elements in our armamentarium—the TKIs, such as lenvatinib or sorafenib. Who are the patients for whom you would pick one of these TKIs as your first-line option?” Bekaii-Saab asked the panel.

Overall, El-Khoueiry said it would depend on the characteristics and patient preference. “The easy answer is patients who have clear contraindications to immunotherapy, those with active autoimmune disease,” he said. “There’s data indicating that patients with a distant history of autoimmune disease may be able to tolerate IO. It’s those with active autoimmune disease or disease that has required recent immunosuppression, who would be the bulk of patients who would get a TKI. Otherwise, I think it would be just a decision that an individual doesn’t want to have intravenous therapy or doesn’t want to travel for treatment. A TKI would be more convenient.”

In choosing the proper TKI for patients, Vogel said he would choose lenvatinib based on the “compelling” real-world data. “For 10 years we have used sorafenib [and] there is a time to change and try something new,” he said adding that sorafenib may still have a role for patients with Child-Pugh B status. “For those with good liver function, it’s lenvatinib.”

The evidence to support Vogel’s choice of lenvatinib comes from the REFLECT study (NCT01761266) in which the agent was evaluated against sorafenib. Overall, the agent was noninferior in terms of OS (HR, 0.92; 95% CI, 0.79-1.06); however, in an analysis characterizing outcomes based on characteristics, lenvatinib demonstrated early and durable response. The objective response among 478 patients was 18.5%, which investigators noted were comparable to the data reported with single-agent IO therapies.¹²