Suresh S. Ramalingam, MD, FASCO: We’ve seen some clinical reports in the past few weeks with TMB [tumor mutational burden] in large phase III trials. For instance, the chemotherapy plus PEMBRO [pembrolizumab] studies, which were the KEYNOTE-189 and the KEYNOTE-021 studies, went back and looked at tumor tissue—they sequenced them. They wanted to see if the tumors had a threshold of 175 mutations higher than that or lower, to see if there is a differential effect of the chemo versus PEMBRO regimen. Both of these studies showed no difference in outcomes for the chemo plus PEMBRO as a function of TMB. In that experience, TMB was not a useful biomarker. But at this meeting, we heard very different data when you are talking about PEMBRO alone as monotherapy. What can you tell us about these results, Johan?
Johan F. Vansteenkiste, MD, PhD: Well, it seems like when you use chemotherapy plus immunotherapy, you fade out quite a lot of the biomarker data. That is the case for PD-L1 [programmed death-ligand 1] as well. The chemo plus immunotherapy is superior to chemotherapy alone in all PD-L1 subsets with some variation according to expression. It’s the same. We have seen the same for chemotherapy plus immunotherapy with tumor mutational burden. Pembrolizumab alone can be administered either in patients above 50% tumor proportion score, which is what we do in Europe, and it can be also be administered in patients above 1%, which is currently common practice in the United States.
In that setting, tumor mutational burden seems to add extra value to the prediction of a better outcome of the patients. If you give pembrolizumab alone … I think it’s unfortunate we didn’t see these data. If TMB is adding extra value for what was reported, it slightly adds extra value if you’re above 50%. It also adds extra value if you’re above 1%. What I want to know is what does it add in those who have a PD-L1 expression between 1% and 49%? We didn’t see that yet. But certainly there seems to be a complementary role there.
Suresh S. Ramalingam, MD, FASCO: The PEMBRO monotherapy data with TMB being a predictive marker aligns with the NIVO [nivolumab] data that were presented a couple of years ago in The New England Journal of Medicine, where high TMB was correlated with NIVO therapy efficacy compared with chemotherapy. As we talk about PD-L1 TMB, it’s hard not to discuss the IMpower110 study that was presented at the ESMO [European Society for Medical Oncology] Congress, ATEZO [atezolizumab] versus platinum-based doublet chemotherapy. This was presented by David R. Spigel. We know that the PEMBRO versus chemo experience resulted in favorable outcomes for PEMBRO. NIVO versus chemo did not. In the MYSTIC trial, DURVA [durvalumab] versus chemo did not result in improved survival. Talk about what you took away from the IMpower110 report, Johan.
Johan F. Vansteenkiste, MD, PhD: IMpower110 was the atezolizumab study comparing single-agent atezolizumab versus chemotherapy. What was seen there is that the effect was in terms of outcome, and it was an early interim analysis. It was not a final overall survival analysis. It was an interim overall survival analysis. What came out is that the tumors that had tumor or immune cell score, immunohistochemistry score of 3, that there was a significant overall survival benefit when atezolizumab was added to chemotherapy. For those who had an immunohistochemistry of 2 and 3, it was borderline. Further follow-up is needed. In those who had an immunohistochemistry of 1, 2, or 3, that has not been analyzed yet.
Again, in that data set, what we would like to see is how they behave as category 3, category 2, and category 1 separately. This is because they always fuse 3, then 2 and 3, and then 1, 2, and 3, which means in the lower categories, you always have a carryover of the best effect. That makes it difficult to interpret the data. Gradually, I think they will come, but we don’t have them yet. The message at the present time is atezolizumab as a single agent was a valid option in patients who have TC3 or IC3 scoring on their biomarker.
Suresh S. Ramalingam, MD, FASCO: How common is it for people in the United States, Pasi, to test for TC3, IC3 for a diagnostic work-up?
Pasi A. Jänne, MD, PhD: I think it’s the exception rather than the rule. Most reports either from our own pathologists or from commercial vendors are just the PD-L1 tumor proportion score. I don’t think there’s a separate testing for immune cells versus tumor cells. Maybe with the data that we have today in the frontline setting with atezolizumab, that will change, assuming there is a regulatory approval in that setting. We’ll see if that practice changes.
Suresh S. Ramalingam, MD, FASCO: Cho?
Byoung Chul Cho, MD, PhD: We rarely test immune cell in tumor specimen. We usually look at the PD-L1 expression in the tumor cell by doing SP263 and 22C3 assay. It’s really hard to see immune cells, staining of PD-L1, in the tumor specimen.
Johan F. Vansteenkiste, MD, PhD: Pathologists don’t like it, huh? It’s a very challenging thing.
Pasi A. Jänne, MD, PhD: Also, the other thing it would be nice to know is if the benefit is the same in the TC3 and the IC3, or are they separate populations? Are they the same population? I think that’s a degree of nuance we haven’t quite yet seen either.
Johan F. Vansteenkiste, MD, PhD: Fortunately, this data will be reanalyzed using the 22C3 and the SP263, so that will give part of an answer to that question.
Pasi A. Jänne, MD, PhD: Yeah, so that will give you some harmonization as well if most pathology practices already use those other antibodies.
Transcript Edited for Clarity