Need for Subgroup Analyses in Melanoma Trials

Transcript:Georgina Long, BSc, MBBS: Keith, I want to make a comment on all of this. Jason said before about the high LDH, it’s looking like nivolumab would be, maybe, very effective, and Jeff supported that. One thing that I think we’re crying out for is some consistency across the trials in terms of reporting subgroups and reporting endpoints in those subgroups.

So, for example, as you discussed before, Jason, with dabrafenib and trametinib, we’ve done an analysis looking at the long-term survivors—reporting on the 3-year survival, people with a normal LDH, people with a high LDH, people with a LDH two times the upper limit of normal, PFS in those groups, and OS in those groups. I would like to see those subgroups pulled across all treatments so that we can make a better assessment and have better discussions. Although I can tell you the 3-year survival in a normal LDH patient on dabrafenib and trametinib is 70%, I can’t tell you that for single-agent anti-PD-1. And I can’t tell you that for nivolumab/ipilimumab yet; that’s because it hasn’t accrued. I can’t even tell you the 1-year overall survival from the CheckMate-069 data, that’s the ipilimumab versus ipilimumab plus nivolumab. I think there needs to be some consistency across the board in looking at these subgroups and endpoints so that we can have really good discussions about first-line treatment.

Jeffrey S. Weber, MD: Well, it’s not going to be 70%. It’s 62% or 64% depending on the group at 2 years…

Georgina Long, BSc, MBBS: No, that’s the overall group.

Jeffrey S. Weber, MD: Yes.

Georgina Long, BSc, MBBS: But I want to know the normal LDH.

Jeffrey S. Weber, MD: Oh, breaking it out.

Georgina Long, BSc, MBBS: Yes. Every trial needs to be…

Jeffrey S. Weber, MD: I agree.

Keith T. Flaherty, MD: Only because, in the real world, you have a patient walking in your office who’s not the average patient.

Georgina Long, BSc, MBBS: Yes.

Keith T. Flaherty, MD: It’s a big digital patient who’s got features that you can glean, like gender and age.

Jeffrey S. Weber, MD: All of us agree that BMS (Bristol-Myers Squibb) needs to tell us with the breakouts, which, to be honest, were not preplanned. They were not stratified. So, these are post hoc analyses with all the dangers inherent.

Georgina Long, BSc, MBBS: That’s what we’ve got though.

Jeffrey S. Weber, MD: For the pooled CheckMate-067 and CheckMate-069 trials, you have 400 patients who got ipilimumab/nivolumab. Within those groups, you’re going to have a significant number. I’m sure they all had LDH, they obviously all had sites of disease and disease burden assessed. I would love to see the breakouts similar to what you presented and in your recent JCO article. I agree.

Georgina Long, BSc, MBBS: And Merck needs to hear the same message, as well. Anyone developing any compounds in this space, these are the endpoints we need so we can have proper conversations.

Keith T. Flaherty, MD: We’ll leave it and summarize that. Basically, patients with lower disease burden do better on all of our therapies, patients with high disease burden do less well on all of therapies.

Georgina Long, BSc, MBBS: Correct.

Jeffrey S. Weber, MD: Although, was that really true at that presentation we heard yesterday? You were there, I think.

Georgina Long, BSc, MBBS: Which one?

Jeffrey S. Weber, MD: There was a presentation from UCSF on prognostic factors. Unless I heard it wrong, it looked as if with PD-1 antibodies, it went the other way with LDH. High LDH, I thought, did a little better and females did worse, which was interesting.

Keith T. Flaherty, MD: Right. I saw that poster, an interesting point. We do have the benefit of trials now that accrued hundreds and hundreds, if not thousands, of patients in aggregate. So, post writing with 130 some odd patients, which is intriguing in a liver metastasis element of that, as well. But this is just, over time, the type of evidence that we need to actually be able to talk to patients about, comparative efficacy.

Georgina Long, BSc, MBBS: And to be fair, Merck did present something in 2014 from their phase I study. Again, it wasn’t the 655 patients we now have; that showed that LDH was strongly associated with a poor response, high LDH, poorer response, and poorer overall survival for pembrolizumab. But we need to do it now that we’ve got more data, more patients, more numbers, and more mature data to make these. Yes, they’re post hoc, but that’s all we have.

Keith T. Flaherty, MD: Jonathan, just one last question about the kinetics of response issue. As I was talking about with Jeff a moment ago, ipilimumab/nivolumab does seem to have a very high response rate and pretty rapid onset. So, going back to the presurgical context, has your group had any experience? It’s a tough regimen, admittedly, and maybe could complicate matters going into the operating room. But is that a regimen you could use in a presurgical context?

Jonathan S. Zager, MD: You could definitely use it. I think, right now, our algorithm for approaching the patient would be BRAF plus or minus MEK. If not, then we use pembrolizumab alone. But definitely you could use ipilimumab/nivolumab. I guess in the patients with very high disease burden or rapidly progressing, or younger patients, they would be ideal candidates to use the combination therapy in a presurgical setting.

Transcript Edited for Clarity