Tanios S. Bekaii-Saab, MD, FACP: Amit, I want to go to you. Are there any biomarkers for I/O [immunotherapy] At least candidate biomarkers—I’m not talking about clinical because we know there are no clinical biomarkers right now.

Amit Singal, MD: We’ve tried to look for them, and 1 of the ones that was of interest was PD-L1 status. Unfortunately, the data that have come out don’t show any value in HCC [hepatocellular carcinoma] in terms of being a biomarker for selecting I/O or any other therapies. The only biomarker we have is AFP for sorafenib, as you know. It’s the Holy Grail. I forget who mentioned it in terms of moving toward precision medicine, but unfortunately, in HCC, we’re not quite there. Hopefully folks who are continuing to do trials will get biopsies and store blood, and we’ll find something in the future but nothing for tomorrow.

Tanios S. Bekaii-Saab, MD, FACP: Arndt, please.

Arndt Vogel, MD: I just want to ask, do we have 1 clinical biomarker? Most of you have seen the Nature paper…recently. What do you think about this idea of NASH [nonalcoholic steatohepatitis] and obesity and I/O? Do you think it’s specific for I/O? Do you think it has any role for I/O combinations? What’s your take on that?

Mark Yarchoan, MD: I have a lot to say on this. If I could for just a minute, the study has been overinterpreted and misapplied. We’ve done an analysis of the TCGA [The Cancer Genome Atlas], the immune microenvironment across the 3 major subsets of HCC: viral hepatitis B, hepatitis C, and nonviral are all very similar. The response rate of PD-1 or PD-L1 combinations across the 3 subsets are virtually identical. There’s this study largely in mouse data showing that NASH may be less sensitive to checkpoints. They do an analysis in which they look at global studies and the nonviral population that maybe there’s a signal for less benefit. The problem is that this is driven by geography, and most of the patients with hepatitis B are in Asia, where sorafenib consistently underperforms because of the lack of availability of second-line options when some of these studies were done. This absolutely should not be used to guide clinical practice. It’s important to discourage this mouse study from driving clinical decisions at this point.

Amit Singal, MD: I agree with Mark. The preclinical data in that paper were elegant. I don’t want to take anything away from preclinical data, but the clinical data have been overstated and given way too much emphasis. I’d strongly caution against withholding—that’s what I would say it is. It’s withholding a proved therapy from somebody who has nonviral liver disease based on this 1 study. As Mark pointed out, etiology was not as stratification variable for any of those trials. By definition, when you just do the subgroup analysis, those imbalances in those 2 groups are that it’s no longer randomized. There’s notable confounders. If this was a retrospective cohort study that had found the same thing, we wouldn’t be talking about it. We wouldn’t be blowing this out of proportion. Because it was a secondary analysis of randomized data, people have given it level 1 value, but they’re not level 1 data. It’s an interesting observation. I hope that there are continued studies to go down this route. But in terms of clinical practice, this shouldn’t enter into anyone’s mind when making clinical decisions.

Tanios S. Bekaii-Saab, MD, FACP: Arndt, I’ll give you the final word on this.

Arndt Vogel, MD: We should be required. I was caught by the end because we did not need to say that it’s not working in a patient with NASH. The idea was to say that the activity of I/O might be less pronounced in NASH. Sometimes when you try to sell a paper, it might be overestimated. There are lots of good preclinical data. In respect to the clinical use of this paper, we still treat our NASH patients with I/O. I agree with that.

Tanios S. Bekaii-Saab, MD, FACP: In many ways, I wish the first plots would end up going into supplements without being main features in any. They mostly elicit an interesting question, but they should not change much.

Arndt Vogel, MD: This is a general question. If you look at natural medicine at the moment, you get a paper published only when you have clinical data. You don’t have a mouse study alone—you don’t get published anymore in Natural Medicine Journal.

Tanios S. Bekaii-Saab, MD, FACP: Understood. I agree that it’s a question worthy of continuing to discuss clinically, perhaps we don’t have that level of evidence.

Mark Yarchoan, MD: Nivolumab itself is beautiful. I just think it’s been misinterpreted.

Amit Singal, MD: Agreed. Quick clinical data are elegant. The fact that the group hold it up with clinical data should be applauded. They’re a comprehensive paper, but to Mark’s point, we can’t say that we should change clinical.

Tanios S. Bekaii-Saab, MD, FACP: I don’t think Arndt was suggesting that.

Amit Singal, MD: I know he wasn’t. To anyone who’s listening, it’s the type of thing where we don’t want to change clinical practice right now. This paper should inspire more data and more study.

Tanios S. Bekaii-Saab, MD, FACP: Absolutely.

Transcript Edited for Clarity