PD-L1 Testing Standard of Practice

Joyce O’Shaughnessy, MD: I want to hear from everybody about what you are doing about testing for PD-L1 now in your practice. It almost seems like, the CPS [combined positive score] of 10 or more was seen in 38% of the population, and the SP142 was 40%, 41% in the IMpassion130. The first question is, although they're different antibodies and though the SP142 was just immune cells, the CPS is both tumor cells and immune cells, still it's interesting that they're coming in around 40% of the patients.

First of all, are they the same patients? And secondly, from a practical standpoint, is it going to be where if I get any test that is PD-L1 positive by the definition for that antibody, that I'm going to say I can use whatever checkpoint inhibitor I want?

Hope, is there a real overlap between the 40% in the 2 trials?

Hope S. Rugo, MD, FASCO: It's interesting, and they'll have to go back on the available tumor samples and use the other testing methods. This is something we did in IMpassion130 and data we presented at ESMO [the European Society for Medical Oncology annual meeting] last year, when we had in-person meetings. What we did was we looked at the biomarker-evaluable population, where about 46% of patients were SP142 positive. This is an antibody where 1% or more of the immune cells need to test positive for PD-L1. We compared it to 22C3 and another antibody, and 22C3 uses the CPS scoring system where you make a ratio that includes both tumor cells and immune cells. In that, 80% of patients were positive.

If you use a score of 1 for CPS, you're measuring most tumors, and that is almost identical to what was seen in the neoadjuvant KEYNOTE-522 trial. Seventy-five percent of patients had a CPS score of 1 or greater. When we looked at the overlap, almost all except for 1% of patients who were positive by SP142, were positive by the CPS score. It does appear that it encompasses the same group of patients. When you do modeling, it's a little bit different, but in this patient group, more patients were positive with CPS.

If you used SP142, you basically got the population that was also CPS positive, so it makes me think by leading in…. And then when they looked at the benefits by progression-free survival and overall survival in IMpassion130, the double positives got the benefit. It was the SP142 that stood out, so you could have a hypothesis that this 39% or so in KEYNOTE-355 are almost a similar group of patients who are positive by SP142, even though we don't know for sure. It does suggest that, it is very unfortunate, but about 40% of our patients can benefit; the rest we need to do more for.

Joyce O’Shaughnessy, MD: Thanks, that helps us in practice a little bit. It kind of simplifies things. Tiffany, what are you guys doing at Memorial Sloan Kettering Cancer Center in terms of testing? And how do you think you’ll interpret, now that we have 2 different antibodies, and soon we will have pembrolizumab available to us in practice. What antibody are you guys going to use, do you know yet?

Tiffany A. Traina, MD: Great question. We have been using the SP142 antibody, given the atezolizumab indication and the data that we have from IMpassion130 already. I think Hope raises some interesting points about the degree of overlap, and maybe we are looking at the same population when you use the CPS of greater than 10 and the Ventana SP142. It'll be interesting to see what ultimately the indications are and where positivity is set for CPS score.

Honestly, of the hazard ratios and the potential benefit seem so close, even for CPS greater than 1, although that didn't meet statistical significance. I think for now we are still using SP142 as our antibody of choice. Another point that hasn't really been raised is what tissue is used for testing and some of the limitations there. With using SP142 as the antibody of the choice, we know we are unable to do that assay on cytology. We're unable to do that on bone. We tend to prefer using this assay on, say, the breast primary when trying to distinguish what the PD-L1 status is, and I'm curious if perhaps the CPS score may allow for some different testing. I don't know if anyone's experienced that in using the CPS assay.

Joyce O’Shaughnessy, MD: Yes, I want to clarify what you just said there, Tiffany. You said for the SP142, it’s not advisable to use it for bone. Was there another tissue that wasn't advisable?

Tiffany A. Traina, MD: Yes. You cannot do it on cytology on FNA you need core tissue for that.

Hope S. Rugo, MD, FASCO: That's interesting. I think that what has happened, not just in IMpassion130, we don't have the data from KEYNOTE-355 yet, but also from other studies that Sherene Loi, MD, and others have done from the earlier trials with single-agent checkpoint inhibitors, is that the PD-L1 positivity varies by site of disease. People thought that LDH might be a marker of resistance when your LDH is high, and liver metastases were. It's fascinating that that was the lowest percentage of positivity we saw in IMpassion130.

It was a small number of patients, but still, fit into all the other spectrum of data where breast, much more likely to be positive, lung, and soft tissue. In fact, the 1 long-term survivor I have on pembrolizumab from the KEYNOTE trial in the second-line cohort is somebody who had lung metastases. It's fascinating that maybe it's more immune sensitive.

Joyce O’Shaughnessy, MD: Interesting. Denise, what about you guys? What are you doing right now with testing around the atezolizumab selection of patients? Do you think that the new KEYNOTE data will make any difference in your testing?

Denise Yardley, MD: I think so. Traditionally, based on the FDA approval, we've been using the SP142, but we do a lot of other molecular testing and profiling where we get signals of positivity. That doesn't correspond to the tests that the FDA requires for approval. Now with the exciting data with pembrolizumab, we're having to rethink the strategy of how we're going to incorporate selecting which tests for which patient population. And then moving forward with health care providers, how they are going to look at these 2 immunotherapies as they hit the market for triple negatives.

I think that will be giving us some direction as well, but for the most part we've gotten comfortable with the SP142, and now we're going to have to rethink the algorithms as we have all these diagnostic companion tests that are required by the FDA to be able to utilize those agents based on the exciting data.

Joyce O’Shaughnessy, MD: Yes, and it is reassuring to me that if…it happens to be that not all the laboratories carry all of the antibodies, etc, if we get an SP142, 1 or more, it's likely that that's going to be a CPS of 10 or more. It's not 100% overlap perhaps, but it's likely, and vice versa. Personally, I'll feel comfortable, because not all the labs are going to have all of the assays. I take a little bit of comfort as well that the CPS of 1 or more was very close to statistical significance. Even if you were off a hair, give or take, giving the patient the benefit of the doubt of the checkpoint inhibitor doesn't seem to be out of the ballpark.

Transcript Edited for Clarity

Related Videos
Video 1 - "HR+/HER2- Early-Stage Breast Cancer: Background and Risk Stratification "
Don S. Dizon, MD
Vijayakrishna Gadi, MD, PhD, and Megan Kruse, MD
Patrick I. Borgen, MD
Henry Kuerer, MD, PhD, FACS, CMQ