recently presented at the San Antonio Breast Cancer Symposium.
Patients enrolled in KEYNOTE-355 – which is a phase 3, placebo-controlled trial of 847 patients – were stratified by CPS scores of at least 1 and at least 10, with the latter group in which adding pembrolizumab to chemotherapy was shown to significantly improve both overall survival and progression-free survival.
As it was unclear whether taking a more fine-grained approach would reveal specific CPS scores at which pembrolizumab would be beneficial, Javier Cortes, MD, PhD, International Breast Cancer Center, Barcelona, and colleagues divided the patients into four CPS levels: less than 1, 1-9, 10-19, and at least 20.
Patients with a CPS 10-19 and at least 20 given pembrolizumab alongside chemotherapy had an overall survival benefit of 29% and 28%, respectively, while the PFS improvement was 30% and 38%. In the CPS of less than 1 and 1-9 groups, there were no discernible benefits from adding the checkpoint inhibitor.
“Given the similar outcomes in the CPS 10-19 and the CPS ≥20 subgroups, a CPS of 10 or more is a reasonable cutoff to define the population of patients with metastatic TNBC that might have benefit from the addition of pembrolizumab to chemotherapy,” Dr. Cortes said. “In my opinion, these results provide further support for pembrolizumab in combination with chemotherapy as a good option, maybe a standard of care for some patients ... with local recurrent unresectable or metastatic TNBC whose tumors express PD-1 CPS ≥10.”
Invited discussant Hope S. Rugo, MD, said the study demonstrates that PD-L1 CPS of at least 10 is “clearly the optimal cutoff for differentiating benefit from pembrolizumab” and confirms the combination with chemotherapy as a “standard of care in this population”.
However, there are a number of outstanding questions in the metastatic setting, she said, including the test used to determine PD-L1 expression.
“Clearly the test that you order should be matched to the planned checkpoint inhibitor, and we look forward to additional data” on the relative overlap of the assays used in both the current study and in KEYNOTE-522.
However, IMpassion130 showed there is “incomplete overlap in terms of the two antibodies and tests that have been used to define PD-L1 positivity in breast cancer,” said Dr. Rugo, professor of medicine in hematology and oncology at the University of California, San Francisco.
“For excellent responders, can chemotherapy and eventually immunotherapy be discontinued, and when is it optimal? How long should we be continuing the combination and how long should we continue the checkpoint inhibitor alone?” she asked.
“Certainly in my own clinical practice,” Dr. Rugo explained, “in those excellent responders, it’s difficult to know when to stop the checkpoint inhibitor, but sometimes toxicity tells us the answer to that question. At some point, we need to stop therapy and understand what happens to those patients.”
She said that only 38% of patients in the current study benefited from pembrolizumab. “How can we amplify the immune response in those patients who do not have PD-L1–positive disease to further extend this benefit, and can we extend the efficacy to other subtypes? There are ongoing studies evaluating this question,” Dr. Rugo said.
Dr. Cortes said that KEYNOTE-355 showed the addition of pembrolizumab to chemotherapy led to clinically meaningful improvements in both PFS and overall survival versus chemotherapy alone in the first-line treatment of mTNBC.
However, that benefit was seen only in patients with a PD-L1 CPS of at least 10, while there was no statistically significant improvement in either PFS or overall survival in those with a CPS of at least 1.
He explained that 847 patients with previously untreated locally recurrent or metastatic TNBC, or those who had been treated at least 6 months prior to disease recurrence, were randomized 2:1 to pembrolizumab or placebo plus chemotherapy.
For the current analysis, they substratified patients by PD-L1 CPS into less than 1, which accounted for 24.9% of patients; 1-9, seen in 36.2%-38.4%; 10-19, accounting for 13.9%-14.1%; and at least 20, seen in 22.8%-24.7% of patients.
Dr. Cortes said the overall survival rate among patients with CPS of at least 10 was 70.5% for patients treated with pembrolizumab plus chemotherapy versus 81.6% for those assigned to placebo, at a significant hazard ratio of 0.73 (P = .0093).
Among patients with CPS of at least 1, the overall survival rate was 79.1% with pembrolizumab plus chemotherapy and 83.9% in those given placebo, at a nonsignificant hazard ratio of 0.86. This translated into an HR of 0.89 in the intention-to-treat analysis.
Turning to the novel subgroups, Dr. Cortes showed that the HR for overall survival for pembrolizumab versus placebo was nonsignificant in patients with CPS of at least 1, at 0.97, and in those with CPS 1-9, at 1.09.
However, the HRs were markedly improved in patients with CPD 10-19, at 0.71, and in those with CPS of at least 20, at 0.72, showing that the “relative benefit of adding pembrolizumab to chemotherapy was pretty much the same ... suggesting that CPS ≥10 could be a reasonable cutoff.”
In both of these groups, there was a sustained separation in the overall survival curves starting at around 10 months.
Turning to the PFS results, Dr Cortes said the event-free rate was 65.5% with the addition of pembrolizumab to chemotherapy in patients with PD-L1 CPS of at least 10, while those given placebo had a rate of 78.6%, at an HR of 0.66.
In patients with PD-L1 CPS of at least 1, the HR was 0.75, or 0.82 in the intention-to-treat analysis.
“As with overall survival,” he said, there was a “trend toward improved efficacy with PD-L1 enrichment with the addition of pembrolizumab to chemotherapy, although the PFS benefit in the pembro arm was slightly greater in the CPS ≥20 subgroup, compared to the CPS 10-19 subgroup.”
However, they highlighted that the difference was “small and the confidence intervals clearly overlapped.”
Why does PD-L1 expression play a role in response to pembrolizumab in mTNBC, but not in the early disease setting as seen in KEYNOTE-522?
“This is a question we have raised many, many times and have had many debates on,” Dr. Cortes said. “They are two completely different populations with the early breast cancer setting completely different to that in metastatic disease. Maybe the microenvironment plays a different role there, maybe we have to explore more in detail other biomarkers. I also think that different drugs were used in the neoadjuvant setting. We still have many unanswered questions.”
Dr. Rugo suggested that previous studies have given some clues to these questions with reductions in PD-L1 expression and tumor-infiltrating leukocytes observed between primary and metastatic disease.
The immune differences between primary and metastatic disease lead to immune escape, she said, adding: “This is clearly complicated by mutational complexity under the pressure of treatment.”
The study was funded by Merck Sharp and Dohme. Dr. Cortes and Dr. Rugo reported relationships with numerous pharmaceutical companies.