Tiffany A. Traina, MD: For women with metastatic triple-negative breast cancer, I think it is really essential to understand the PD-L1 [programmed death-ligand 1] status of a patient's tumor. In IMpassion130, testing for PD-L1 could have been performed on either the primary tumor or a metastatic focus. Remembering that this was a first-line metastatic disease trial, even when the primary tumor was tested, often the time between treatment and that tissue collection was only about 60 days. These tumor specimens were largely representing untreated breast cancer.
The primary tumor appeared to have a greater likelihood of testing PD-L1 positive, and that is something to keep in mind as you're considering which tissue to send for PD-L1 testing. I'll remind you the assay used in IMpassion130 was the VENTANA PD-L1 SP142 assay, and the definition of PD-L1 positivity was 1% or greater staining in the immune cells. There were some nice subsequent reports on biomarker analysis after the initial presentation of these data.
It is really important to recognize that in PD-L1—negative patients, there was no advantage with the addition of atezolizumab to nab-paclitaxel. In all the different exploratory biomarkers examined, PD-L1 positivity by SP142 was the predictive biomarker of importance. There have been several other PD-L1 assays available in other solid tumors, and these include the Dako 22C3 antibody and the other VENTANA antibody, PD-L1 SP263.
There was a wonderful presentation at ESMO [the European Society for Clinical Oncology Congress] this past year by Hope Rugo, MD, and colleagues exploring, in a post hoc analysis, the performance of these other PD-L1 assays in the tumor specimens from IMpassion130. I think the conclusion there is that these other PD-L1 assays were able to detect or find more patients classified as PD-L1 positive.
In this subset, there were about 600 patients who had tumor samples available for additional PD-L1 testing. If you recall, the SP142 antibody was positive in about 41% of the IMpassion130 population. The 22C3 antibody classified almost 80% of tumors as PD-L1 positive. The SP263 antibody classified about 75% of tumors as PD-L1 positive. The real question is whether that positive result is predictive of benefit to atezolizumab. Unfortunately, there was suboptimal concordance between those assays and SP142, whereas all tumors that tested PD-L1 positive by SP142 were likely also to be found as PD-L1 positive by SP263 or 22C3. Positivity by those other assays in the absence of SP142 positivity did not seem to predict benefit to atezolizumab. In short, SP142 PD-L1 positivity is the predictive biomarker for benefit from atezolizumab in addition to nab-paclitaxel.
There has been work looking at other biomarkers beyond PD-L1 status to help predict benefit from checkpoint inhibitors, such as atezolizumab. In some of the exploratory analyses related to IMpassion130, there have been analyses for BRCA status, PD-L1 status by other assays beyond SP142, and even stromal TILs [tumor-infiltrating lymphocytes]. We do know that stromal TILs are prognostic in triple-negative breast cancer and that patients who have a high proportion of stromal TILs tend to respond better. That also applies to responding well to use of a checkpoint inhibitor.
Transcript Edited for Clarity