Hope S. Rugo, MD: With that general idea of trying to pick out treatments based on what type of disease somebody has, so HER2-low disease, how are we personalizing medication for patients? Of course 1 area where we really want to focus on personalizing medication—not that we don't want to do it for everybody—is triple-negative breast cancer. We've been hit by the poor options available to patients in the metastatic setting and for patients who don't have a great response in the early stage setting.
We have such limited options, and we have a very clear idea of what the outcome is for the majority of patients. We were excited to see data from the IMpassion130 trial with a checkpoint inhibitor, atezolizumab, added to nab-paclitaxel, but this only benefits a relatively small group of patients in the metastatic triple-negative setting. Tiffany, do you want to tell us a little bit about the IMpassion130 trial and what patients might benefit from atezolizumab?
Tiffany A. Traina, MD: Sure, I'm happy to. I was thinking back that it looks like we're right around probably the 1-year anniversary of when atezolizumab was FDA approved for this very indication based on IMpassion130. If you remember, that’s a study of about 900 women with triple-negative breast cancer, or metastatic disease, or locally advanced unresectable, and it was a first-line trial for these patients. The backbone was chemotherapy of nab-paclitaxel with a randomization to atezolizumab or a placebo.
If you recall, that primary end point for the study was both PFS [progression-free survival] and OS [overall survival], and what we saw in results was interesting. For the intention-to-treat population there was an improvement in progression-free survival, but what was most compelling was seeing the subset of patients with PD-L1—positive tumors. We’ll come back to how that PD-L1 was tested, but in that subset of patients who were PD-L1 positive, the difference in progression-free survival was more substantial, about an absolute difference of say 2 to 3 months, and about a 40% or so improvement in PFS.
When you move onto looking at the overall survival data, this is where things became a little bit more exciting. In the intention-to-treat population there wasn't a statistically significant difference in OS. However, just comparing in the subset of patients with PD-L1—positive disease, the group that received atezolizumab had about a 40% improvement in overall survival, with an absolute difference on the most recent analysis of almost 7 months. While that didn't have the power for a formal statistical analysis, it looked compelling and of interest for these patients.
Talking a bit about the PD-L1—positive definition is important. In IMpassion130, the assay used to establish PD-L1 positivity was the Ventana SP142 assay, and the definition of positivity was based on the percentage of immune cells positive. That is in contrast to PD-L1 testing in some of the other solid tumors where different assays were used, and different scores were used, such as combined positive scores [CPS] with assays such as the SP263 or 22C3. Hope, I know at ESMO [the European Society for Medical Oncology annual meeting] you performed an ad hoc analysis from the IMpassion130 data looking at other PD-L1–positive assays. In your data, from what was presented, the proportion of patients on IMpassion130 who had PD-L1 positivity using an SP142 assay was about 40% of those 900 patients.
In the analysis that you and your colleagues reported using SP263, another antibody and assay, there was about 75% PD-L1 positivity by that definition. I think with the 22C3 assay this was about 60% or so, maybe even more. Even higher than that, maybe 80% or so, were considered positive. There was concordance between SP142 and these other 2 antibodies, but only about 40% of overlap. When we try to define which patients are more likely to respond to the addition of atezolizumab, my preference based on the data I've seen is to stick with the SP142 assay as the predictive biomarker, if you will. I’m curious to hear what others think, and they are your data, Hope, so please elaborate on that if you think that there is other important biomarker nuance that we should be aware of.
Hope S. Rugo, MD: You did a great job of discussing that, and the information was in the biomarker evaluable subpopulation. We've all been very interested in the press release in February of this year from KEYNOTE-355 that suggested in an important population, patients who relapsed 6 months or greater after adjuvant or neoadjuvant therapy versus 12 months, that those patients benefited from the addition of pembrolizumab to gem [gemcitabine], carbo [carboplatin], paclitaxel, nab-paclitaxel as the options, but only if they had a CPS score of 10 or greater.
As you were talking about, when we were looking at this in the ESMO analysis, CPS scoring is supposed to be positive at 1 or greater, and that was this huge population compared to SP142. It is interesting. It does seem like you have to have more positivity, and CPS looks at tumor and ICs [immune cells] and looks at a ratio. Perhaps in the metastatic setting you really need more of the PD-L1 expression on immune cells, and you're seeing that with this higher CPS score. It's hard to know. We haven't seen the data presented yet from KEYNOTE-355 but will this year, so that'll be helpful.
It is a shame in some in some ways that it's only 41% in IMpassion130. People have found different percentages looking at their populations of patients, but it's clearly not everybody, and we still want to improve outcome for those patients.
Transcript Edited for Clarity