Adding pembrolizumab to standard chemotherapy significantly improved progression-free survival for patients with metastatic triple-negative breast cancer, but only if their tumors were enriched with comparatively high levels of the target programmed death ligand-1 (PD-L1), results of the KEYNOTE 355 trial showed.
Among 843 patients with triple-negative breast cancer (TNBC) randomized to receive either investigator’s choice of chemotherapy plus pembrolizumab (Keytruda) or placebo, patients whose tumors had a PD-L1 combined positive score (CPS) of 10 or higher had a median progression-free survival (PFS) of 9.7 months when treated with pembrolizumab and chemotherapy, compared with 5.6 months among patients treated with chemotherapy and placebo, reported Javier Cortes, MD, PhD, from the Vall d´Hebron Institute of Oncology in Madrid and Barcelona.
However, among patients with CPS between 1 and 10, there was no significant difference in PFS between the treatment arms, he said in a presentation made as a part of the 2020 American Society of Clinical Oncology virtual scientific program.
“The inclusion of taxanes and a known taxane/platinum–based regimen permits assessment of the clinical benefit of pembro in combination with several routinely used chemo partners. A trend toward improved efficacy with PD-L1 enrichment was observed in patients treated with pembro plus chemo. The improvement in progression-free survival with chemotherapy and pembrolizumab was observed across patient subgroups,” said Dr. Cortes.
In the KEYNOTE-522 study, adding pembrolizumab to chemotherapy in the neoadjuvant setting increased the likelihood that women with stage III or early node-positive TNBC would have a pathologic complete response and sustained clinical benefit.
KEYNOTE-355 examined whether pembrolizumab in combination with chemotherapy could provide additional benefit over chemotherapy alone in patients with previously untreated locally recurrent inoperable or metastatic TNBC.
Patients with previously untreated metastatic triple-negative breast cancer who had at least 6 months between definite surgery or last dose of adjuvant chemotherapy (whichever came last) and first disease recurrence were stratified by study chemotherapy received, tumor PD-L1 expression at baseline, and prior treatment with the same class of chemotherapy in the neoadjuvant and/or adjuvant setting.
The patients were then randomized in a 2:1 ratio to pembrolizumab plus chemotherapy based on the investigator’s choice of nab-paclitaxel, paclitaxel, or carboplatin-gemcitabine (562 patients) or to chemotherapy alone (281).
Pembrolizumab and placebo were administered in a double-blind fashion for up to 35 doses. Chemotherapy was given at the investigator’s discretion according to local guidelines. This trial was not powered or designed to compare differential efficacy of the various chemotherapy regimens, Dr. Cortes noted.
The trial had dual primary endpoints of PFS in patients with PD-L1–positive tumors (CPS > 10 and > 1) and in the intention-to-treat population, and overall survival both in PD-L1-positive patients and the ITT population. Overall survival results will be reported at a later date.
As noted before, the primary endpoint was met in the population of patients with CPS higher than 10, with median PFS of 9.7 among those receiving pembrolizumab versus 5.6 months among those receiving placebo, and an estimated 1 year PFS of 39.1% versus 23% for controls, translating into a hazard ratio for progression on pembrolizumab of 0.65 (P = .0012).
In the patients with CPS higher than 1, however, the median PFS was 7.6 months with pembrolizumab compared with 5.6 months with placebo, translating into a hazard ratio of 0.74. However, the results did not meet the prespecified boundary for significance. Because of this, the statistical significance in the ITT population was not tested.
“In patients with PD-L1 CPS 10 or higher tumors, the benefit of pembro/chemo on progression-free survival was generally consistent across most predefined subgroups, including eight geographic regions, ECOG performance status, on-study chemo, and prior treatment with the same class of chemo,” Dr. Cortes said.
Treatment-related adverse events occurred in 96.3% of the patients on pembrolizumab and 95% of patients on placebo. Grade 3 or greater adverse events occurred in 68.1% versus 66.9%, respectively. Two patients in the pembrolizumab arm died from a treatment-related event. There were no treatment-related deaths in the placebo arm.
The most common events were those typically associated with chemotherapy, including anemia, neutropenia, nausea, alopecia, fatigue, decreased neutrophil counts, and elevated liver transaminases. Immune-mediated adverse events of any grade occurred in 25.6% of patients in the pembrolizumab arm versus 6% of controls; none of these events were fatal.
“What is clear in this study is that again we’re seeing efficacy of pembrolizumab in combination with chemotherapy increases with increases in CPS,” according to the invited discussant Catherine M. Kelly, MB, BCh, from University College Dublin and Mater Misericordiae University Hospital in Dublin.
“The results from today’s KEYNOTE-355 appear consistent in terms of progression-free survival. However, it is ‘watch this space’ until we get overall survival data before we can make any further comparisons,” she added.
Questions that still need to be answered include which is the best test for measuring PD-L1, whether patients with CPS of 1 or more but less than 10 benefit from the treatment, which of the available chemotherapy regimens is the best partner for pembrolizumab, how to treat patients who don’t respond to the combination, and what are the implications for PD-1/PD-L1 inhibitors in late-stage disease if they are approved in the neoadjuvant or adjuvant setting, Dr. Kelly said.
The study was funded by Merck. Dr. Cortes disclosed honoraria from, a consulting/advisory role for, and institutional research funding from Merck and others. Dr. Kelly disclosed honoraria from MSD Oncology and others, and travel expenses from Pfizer and Roche.
SOURCE: Cortes J et al. ASCO 2020, Abstract 1000.