Suresh S. Ramalingam, MD, FASCO: In the last part of this session I want to talk a little more about biomarkers. Pasi, we talked a lot about PD-L1 [programmed death-ligand 1] expression, but maybe you can start us off by talking to our audience about what is TMB and why that is even being discussed. What’s the rationale for using TMB [tumor mutational burden] as a biomarker?
Pasi A. Jänne, MD, PhD: TMB, or tumor mutational burden, is a reflection of the total number of mutations in an individual tumor—tumor-specific mutations that would create new immunogenic epitopes, where the immune system would try to go after that. Especially with checkpoint inhibition, you have more chances of the immune system to go after these neoepitopes or new proteins that are made by these mutations. Certainly, this is a biomarker that has been looked at fairly extensively. It first demonstrated that it predicts outcome for single-agent immune checkpoint inhibition and, of course, it has been looked at extensively now in the combination immune checkpoint inhibition as well as more recently with chemotherapy in checkpoint inhibition.
I think TMB is a complicated biomarker because there isn’t a standard in terms of what we call positive in terms of how many mutations. They’re often measured in mutations by megabase, and they’re extrapolated from our NGS [next-generation sequencing] reports. Or you get different results if you do more extensive sequencing like whole exome sequencing. You can do it from the tumor, from the blood, and so forth. There are various ways of doing that, and in fact, there are efforts underway in the US to harmonize on some of these efforts. So, it’s a fairly complex biomarker. I think the other challenge for the immunotherapy field of TMB is that you have to sequence somebody to get TMB. It is a 10- to 14-day or longer process to get the TMB results, whereas PD-L1 comes as an immunohistochemistry biomarker in 24 to 48 hours. That creates another level of complexity using TMB as a biomarker.
Suresh S. Ramalingam, MD, FASCO: Johan?
Johan F. Vansteenkiste, MD, PhD: Perhaps in what I just said I was not extremely optimistic about biomarkers for small cell lung cancer. But now that you mentioned TMB, in the study on the relapsed small cell lung cancer in the CheckMate trial, the tumor mutational burden was looked at in the way of tertiles—low, intermediate, and high. In that experience, both for nivolumab as well as for nivolumab-ipilimumab in relapsed small cell lung cancer, there were better response rates in those with high, not with low or intermediate but with high tumor mutation burden. Also, the overall survival curve had a plateau at 20% in those with TMB in relapsed small cell lung cancer, while the other 2 at 2 years were almost close to zero. So maybe there is something, a role for TMB in small cell lung cancer.
Suresh S. Ramalingam, MD, FASCO: If we were to look at TMB, obviously, Pasi, you mentioned what is the threshold, right?
Pasi A. Jänne, MD, PhD: Right.
Suresh S. Ramalingam, MD, FASCO: What threshold indicates who would benefit and who would not benefit? Do we have that, you think, along all the studies that we’ve looked at?
Pasi A. Jänne, MD, PhD: Well, I think there are some thresholds that have been identified based on panel-based sequencing. You have to have a sequence in the 300-to-400-gene range in order to be able to comfortably assign a threshold. Most of the commercial NGS panels that do that can assign that, and they typically put those in the reports. If you do whole-exome sequencing, that’s a whole different thing. Also, in cases where you sequence the germline to know if they are germline variants or not, I think that gives you another level of clarity where the TMB is. Some of these have come from clinical trials. The points have been picked. The values have been picked—some from blood, some from tumor. We have some, but I don’t think we have a consistent number to say that it has to be this across all panels before there’s a benefit. So there is some variability there.
Suresh S. Ramalingam, MD, FASCO: Cho, for blood-based TMB versus tumor-based TMB, how good is blood compared with the tumor?
Byoung Chul Cho, MD, PhD: There is a very good correlation between blood-based TMB and tissue-based TMB, especially if you look at the data from our MYSTIC trial. There is a concordance based on data, but we need a large amount of tumor tissue to have tumor mutational burden. That’s why we really need plasma-based TMB analysis. In the MYSTIC trial, the blood-based TMB was predictive. If we have blood TMB at the cutoff of 13 mutations per megabase, it was predictive for a PFS [progression-free survival] as well as overall survival to the durvalumab and tremelimumab combination versus chemotherapy. As Dr Pasi Jänne already mentioned, there is a lot of controversy of blood over tumor-based TMB regarding cutoff value and a platform, across platform validation. I do believe for the future that we need to elaborate blood- or tumor-based TMB for I/O [immuno-oncology]—I/O combination therapy development.
Suresh S. Ramalingam, MD, FASCO: Johan?
Johan F. Vansteenkiste, MD, PhD: It’s clear we don’t have to expect cutoff in TMB to be there because it’s a continuous biological phenomenon. It’s the same for PD-L1. Even when you are PD-L1 high, for our data you do better if you have 95% than when you have 55%. There is never an absolute cutoff in there. We need to find a test that is clinically applicable with a certain cutoff, but it’s a continuous biologic phenomenon.
Suresh S. Ramalingam, MD, FASCO: I absolutely agree. What do you think we know about the correlation between PD-L1 expression and TMB in a given tumor specimen? Are they linearly correlated or not?
Johan F. Vansteenkiste, MD, PhD: No, they are not. There is a certain concordance in, on average, one-third of the patients, but there are patients who are tumor mutational burden high and PD-L1 low. The reverse is there as well. So no, they’re not concordant at all, and perhaps that’s not a real surprise because it’s a biomarker that is present in a very different position in the cancer immunity cycle. They’re not necessarily linked, so most likely they are additive. That’s what we have to find out over the coming years. In some situations, 1 should have more weight. In other situations, the others should have more weight. Perhaps they should be both taken into account. I feel that we’re not there yet. We don’t understand the full picture, but hopefully over the coming years more clarity will arise there.
Transcript Edited for Clarity