PD-L1 Assay PFS and OS Concordance


Sara M. Tolaney, MD, MPH: In the IMpassion130 study, we did see that the benefit of adding atezolizumab to nab-paclitaxel was really restricted to patients with PD-L1 positivity. In this study, PD-L1 positivity was defined by using the SP142 antibody on the Ventana Benchmark platform. And positivity was defined as having any staining on immune cells that was greater than 1%. And so generally speaking, when we’re now selecting patients to be treated with atezolizumab, we have been using the SP142 antibody to select patients. However, this is challenging in practice because many of us don’t have availability of SP142 testing, so 1 big question has arisen is, could we just use a different antibody to test for PD-L1 and would it equally predict benefit of atezolizumab?

One very good analysis that was done retrospectively from the IMpassion130 study was looking at all tumors from the patients and staining them for at least 3 different PD-L1 antibodies. The investigators stained the tumors for PD-L1 using SP142, or SP263, or 22C3 antibodies, and they really looked to see if PD-L1 positivity with each of these antibodies correlated with survival. And they also looked to see what the concordance was between staining with these antibodies.

They found that the patients who are in this biomarker subgroup (about 40% of patients were tested) had tumors positive for PD-L1 by testing with the SP142 antibody. But, when you look at 22C3, 81% of patients had PD-L1—positive disease, and for the SP263 antibody, it was 75% of patients. And so, in essence, what you end up seeing is that of the patients who test positive by the 22C3 antibody, almost all of them test positive by the SP142. So it’s only about 1% of patients who test for 22C3 who test positive who are not going to test positive by the SP142 antibody, and the results are very similar with SP263.

And so you can see that the concordance between these antibodies is actually poor. The concordance between SP142 and 22C3 is about 64%, and the concordance between SP142 and SP263 is also similar, about 69%—so again, very poorly concordant. As you would expect, given the poor concordance, we also see that these antibodies, such as 22C3 and SP263, are not as good as SP142 in predicting overall survival outcomes. And they actually tried to do this analysis looking at various cutoffs for the different antibodies. So, 22C3 traditionally has different ways of scoring, so you can use the combined positive score (what we call a CPS assay), or you could even potentially analyze it using the IC scoring system, as was used with SP142. No matter which way they analyze the antibodies with the different scoring systems, SP142 in the IC scoring system still performed the best in terms of predicting overall survival.

I think the bottom line from all these concordance assays is that it does not seem that the different antibodies are concordant, and it does seem that SP142 with the IC scoring system seems to be the best at predicting who’s going to benefit from atezolizumab. And so I think given this in our practice, we have turned to getting SP142 testing on our patients in order to predict who should get atezolizumab. In our lab—we actually have to send out to a central lab to get SP142 tested because at our hospital, we don’t have that assay validated. I think this is a challenge at many institutions where in other cancers, people are using different antibodies, and so not all hospitals have SP142 validated. It has required a send out at many institutions.

I think given the poor concordance between the various antibodies and seeing that 22C3 and SP263 did not perform as well as SP142 in predicting atezolizumab benefit, I do think that we should use SP142 testing when selecting patients who might benefit from atezolizumab. I think it’s possible if you send a different assay—such as sending out 22C3—you might select a patient that even though they test PD-L1—positive by that assay, they may not actually benefit from atezolizumab therapy. You may be overtreating patients with atezolizumab and potentially exposing some patients to toxicity.

Transcript Edited for Clarity

Related Videos
Erika P. Hamilton, MD
Funda Meric-Bernstam, MD
Aditya Bardia, MD, MPH
Sara M. Tolaney, MD, MPH
Expert on breast cancer
Expert on breast cancer
Colin E. Champ, MD
Christie J. Hilton, DO
Yuan Yuan, MD, PhD
Related Content