Challenges of Sequencing and Biomarkers in Hepatocellular Carcinoma

Richard S. Finn, MD: There are a lot of other combinations being looked at. Amit, at ASCO GI [American Society of Clinical Oncology Gastrointestinal Cancers Symposium], there were a number of these TKI [tyrosine kinase inhibitor]–I/O [immuno-oncology] combinations. Many of these were with drugs that aren’t necessarily yet approved in liver cancer.

Amit G. Singal, MD, MS: Yes, there were a couple of studies that were presented, the ALTER-H003 study being 1. It was with agents that aren’t yet approved, but many of these show very similar findings to what we’ve seen in the larger phase 3 studies: high response rates, somewhere between 25% and 30%, and generally well tolerated in terms of the AE [adverse event] profile. There was nothing unexpected. 

These were generally earlier phases in terms of the trials, but they reinforce this idea that there’s a lot of smoke, a lot of promise, and overall make us quite excited to see continued phase 3 data to transform the field. All of it is coming together. It’s not as if there’s an aberrant signal; all these trials seem to be consistently showing very similar results across the board.

One of the things that’s going to be interesting—and we’ve alluded to this, as we see these responses somewhere between 25% and 30% is: Are we choosing the same 25% to 30%, or are we rescuing different subcohorts? As you transition from 1 line of therapy, you start to add more patients from whom you can induce responses and improve survival. 

We’ll start to figure this out as we have sequencing data. Hopefully, we’ll become smart about this. We’ll start to have biomarkers that can say up front whether a patient has a biomarker that allows them to respond to a single-agent PD-1 inhibitor therapy, so there is no need for the added toxicity of a doublet. Whereas we have another patient who doesn’t have that biomarker and would be responding to an I/O-TKI or an I/O-CTLA4 inhibitor, whatever that may be. The hope is that we’re not just blindly sequencing therapies and we will be able to choose who can respond to single-agent PD-1 and who needs 1 doublet vs another.

Richard S. Finn, MD: No one is going to argue with that. You raise a very provocative question. I look at the I/O field broadly across histologies, and biomarkers are hard to come by; they’re elusive. We may enrich MSI [microsatellite instability] high, TMB [tumor mutation burden] high, which aren’t necessarily frequent observations in liver cancer. But the immune system is so complex, especially with these TKI interactions, which we don’t have a good handle on.

Dan, what are your thoughts? Are we close to a biomarker? Where are we?

Daneng Li, MD: Unfortunately, we’re not. In a sense, it is very complex with the immune system. You can look at some of the ways the immune system is complex in terms of PD-1 expression or TMB, or MSI high, but a multitude of factors regarding immune markers and genomics are involved. Looking at the earlier atezolizumab-bevacizumab data that were presented by Andrew Zhu at AACR [American Association for Cancer Research Annual Meeting] looked at potentially angiogenesis markers combined with immune markers to create a scoring algorithm like kidney cancer.

Ultimately, as we get more research involved and become smarter, we’re going to start to find that it’s a combination of markers to create some type of scoring algorithm that’s going to be very complex but will lead us to start to predict response to the various different treatments we’re using.

Richard S. Finn, MD: Part of the phase 2 study of tremelimumab and durvalumab, Katie Kelley, MD, from UCSF [University of California San Francisco Helen Diller Family Comprehensive Cancer Center] presented some interesting biomarker work that was more pharmacodynamic, meaning patients who were on treatment and those who had a better response seemed to have an enrichment of activated T cells. But that doesn’t tell us whom to select upfront. That just tells us an MOA [mechanism of action].

Josep, your lab is active in this. What are your thoughts? Are we going to have something in the near future or not?

Josep M. Llovet, MD: There is a possibility, but it is very hard. The first thing is that, as you know, we have been collecting data from patients exposed to a single-agent TKI, and we end up with 11 gene signatures that we will be presenting in the near future. Still, these are controlled, retrospective studies, so we need to have the kind of data we see in the phase 3 trials and the biopsies obtained we see in the phase 3 trials, which is not always possible. We need to work with these data to validate whatever proposal.

I recall with the CheckMate040 trial also there was a paper in Journal of Hepatology with 4 genes predicting response, but these are just hypothesis-generating, which is nice, but we need to test that.

I would love to have a biomarker because this represents a profound impact in the outcome of the patient. The profile of adverse events is very good. But at this point, we don’t have any tools to recognize those, certainly not PD-1, as you know. Also, MSI high is present in 3% percent of patients we see, and high TMB is very low. So we are not there yet. 

Richard S. Finn, MD: Just a quick comment before we wrap up, Josep. Should we be checking our patients with advanced liver cancer for MSI? Do you do that?

Josep M. Llovet, MD: No. MSI is present in 3% or less of patients. Also, as you mentioned, it is not black-and-white. Not all MSI-high patients are responding to checkpoint inhibitors. In the frontline setting you are already exposing the patient to checkpoint inhibitors, so it is not the Holy Grail.

TRANSCRIPT EDITED FOR CLARITY