Hepatocellular Carcinoma: Practical Implications of Emerging Data - Episode 17
Richard S. Finn, MD: Last year, we had exciting data with atezolizumab-bevacizumab. Preceding that was the disappointment of the CheckMate 459 study. That was frontline nivolumab versus sorafenib, and to be honest, even a few months before that, we had the KEYNOTE-240 study, which was pembrolizumab versus placebo. When we saw that pembrolizumab versus placebo didn’t meet its primary end point, a lot of us started getting concerned: How will nivolumab beat sorafenib? Tony, what are your thoughts on CheckMate 459? Were you shocked by those results? Any surprises that came out of that?
Tanios S. Bekaii-Saab, MD, FACP: To say I’m shocked, the answer is no. I have it on the record, so it’s not posthumous. I have it on the record predicting difficulty for nivolumab as single agent to beat sorafenib, and not because sorafenib is a fantastic drug, by any means. It’s a problem of how we measure effect, as all of you have mentioned. One is this: Every one of those agents as singlet benefits few; it doesn’t benefit most. The problem is, when you start measuring around the medians, you’re going to be in trouble, because the median is really where most of those patients who will not respond to one will not respond to the other. They’re going to collapse that median, and they’re going to come close together.
With the nivolumab versus sorafenib, I must say, a little bit of a disappointment in the fact that the survival benefit was not, even at the tail end, much more impressive. We must be missing the mark with having the proper biomarker for singlet versus doublet versus now we’re looking at triplets. Unfortunately, the answer is no, but also glad that this was a short-term disappointment because then the atezolizumab-bevacizumab came to actually bring back up the immune therapy landscape back into HCC. Pembrolizumab versus placebo versus a nonactive control, whereby we know every TKI [tyrosine kinase inhibitor] actually won the war versus a placebo, but then we have pembrolizumab not doing the same, including a hazard ratio around the PFS of 0.71, which, if we look at regorafenib or cabozantinib or even ramucirumab in 400 ng/mL, the hazard ratio was 0.49, which tells us that there is a very small subgroup of patients that benefits just from the 1.
I think that discussion is over at this point. There may be a few patients, as Anthony mentioned—maybe a Child-Pugh score of 7 and 8, where you feel compelled that the TKIs are going to be problematic, and a doublet doesn’t make as much sense. But perhaps these are the patients you give a hail Mary nivolumab-pembrolizumab perhaps. Other than that, I think single-agent PD-1 or PD-L1 inhibitor is certainly not an option for 99% of our patients at this point.
Richard S. Finn, MD: One thing we saw from both CheckMate 459 and KEYNOTE-240, the single-agent I/O studies in front line and second line is a lot of patients: 20% to 30% of patients in the control arm went on to get a single-agent PD-1 inhibitor. Certainly that could contaminate an OS end point. But 2 things: You mentioned PFS was not dramatically different, which you would think is where the second line wouldn’t affect as much, the second line choice or the choice beyond progression. Atezolizumab-bevacizumab did improve OS, despite all these other things, which suggests that maybe our best chance for improving OS is up front. But sequencing will still be an issue, right? From the beginning of our discussion, you’d say 80% plus or minus might be the ones who get a doublet up front, but still there will be a role for single-agent TKIs for various reasons. Anthony, is single-agent I/O done in liver cancer?
Anthony El-Khoueiry, MD: I largely agree with Tony that it’s not necessarily done, but there will be a very small subgroup of patients where you may find it useful. For the patients with compromised liver function, you may feel like that’s their best option, or for patients who did not qualify for atezolizumab-bevacizumab up front, they’re going to get single-agent PD-1 in later lines of therapy. It’s a much smaller population. Tony just really pointed to the key issue, which is that benefiting a small subgroup of patients very well was not sufficient when we look at medians to get these agents’ approvals. Unless we have a biomarker to select for single agent, it’s going to be difficult to have significant usage of these drugs as single agents.
Richard S. Finn, MD: To the other extreme, Anthony, we talked a lot about doublets and single agent. Are triplets on the horizon?
Anthony El-Khoueiry, MD: Well, yes. We’re oncologists, so we like to pile things on top of one another. We saw some data from the GI Cancer Symposium in 2020, in January—this seems so long ago. We saw data on ipilimumab-nivolumab plus cabozantinib. This is another cohort of CheckMate040. In this cohort, there were 2 arms, nivolumab plus cabozantinib versus nivolumab-ipilimumab plus cabozantinib. What’s intriguing is, if you look at the triplet, the response rate was again in the 30% range or high 20%. The response rate did not go beyond what we see as doublets. However, with the nivolumab-cabozantinib doublet, there was a median OS of 21 months. With the triplet, the median OS had not been reached. Similarly with the triplet, if you look at disease-control rate, it was really around 90%. That comes at a cost: 70% grade 3 and 4 treatment-related adverse events, right? Especially elevated liver enzymes, so liver toxicity was more significant in the triplet arm. Will there be room for these triplets? It’s going to depend on the risk-benefit ratio—how much toxicity are we paying for how much benefit. That remains to be seen at this point. It continues to be very experimental, in my opinion.
Transcript Edited for Clarity