AT SABCS 2015
SAN ANTONIO (FRONTLINE MEDICAL NEWS) – The immune checkpoint inhibitor pembrolizumab appears safe and modestly active in women with estrogen receptor (ER)–positive, HER2-negative advanced breast cancer that expresses programmed death–ligand 1 (PD-L1), according to preliminary results of a phase Ib trial presented at the San Antonio Breast Cancer Symposium.
Nineteen percent of women screened for the trial, KEYNOTE-028 , had archival or fresh tumors from nonirradiated sites that tested positive for this ligand by immunohistochemistry, reported lead investigator Dr. Hope S. Rugo, clinical professor in the department of medicine (hematology/oncology) and director of the breast oncology clinical trials program at the University of California, San Francisco.
A total of 25 of these women received at least one dose of pembrolizumab (Keytruda), an antibody to the PD-1 (programmed death–1) cell surface receptor that blocks interaction of the receptor with its ligands, thereby preventing the inactivation of T cells. (The antibody is currently approved by the Food and Drug Administration for the treatment of melanoma and non–small cell lung cancer.)
With a median duration of follow-up of 7.3 months, the overall response rate as assessed by investigators using RECIST criteria was 12%, and the clinical benefit rate was 20%, Dr. Rugo reported. Responses were durable, lasting anywhere from about 9 weeks to more than 44 weeks.
The rate of grade 3 or 4 adverse events was 16%. Only a single patient had to have interruption of pembrolizumab because of the development of an immune adverse event.
“Based on these data, we believe that further investigation of immune therapies in ER-positive, HER2-negative breast cancer, particularly using combination therapies that can expand the T-cell compartment, are warranted,” Dr. Rugo maintained.
She noted that the trial’s findings differ somewhat from those of a similar trial testing avelumab, an investigational antibody against the PD-L1 ligand itself, that was presented at the symposium ( abstract S1-04 ). That trial found a much higher PD-L1 positivity rate in screened women, 55%, and a much lower overall response rate, 3%.
These differences may have been due to use of different immunohistochemical assays for PD-L1, lack of data for some patients in the other trial, and/or differences in the target of the agent used (PD-1 vs PD-L1), she speculated. “It’s clear that as we move forward in this field of immunotherapy, which we are all very excited about, that standardization of assays assessing PD-L1 positivity are going to be critical,” she concluded.
“After how many studies will we be convinced [given] the discordance between different antibodies [and] scoring mechanisms, and the very, very weak enrichment in responders, that PD-L1 staining is not an adequate biomarker for these agents?” asked session attendee Justin Balko, Pharm.D., Ph.D., of Vanderbilt University, Nashville, Tenn.
“I think that we are going to need a consortium as well as development of guidelines for the best methods to test PD-L1 expression, whether that’s an immunohistochemical test, looking at mRNA, or looking at pathway activation,” Dr. Rugo replied. “Whether we need to include for example TILs [tumor-infiltrating lymphocytes] in evaluation or quantification of lymphocytes, which may or may not be practical, remains to be seen,” she said, adding that this research will require a collaborative effort.
Another attendee speculated that the disparate findings for the two trials may have been due to reliance on RECIST criteria in the KEYNOTE-028 trial. “I wonder if you had the opportunity to explore any of the immune-related response criteria that have been evaluated in other settings like melanoma,” she said.
“As a breast cancer field in immunotherapy, we are kind of behind everybody else, and we are just now starting to use immune RECIST as part of our assessments for the studies going on now prospectively in breast cancer,” Dr. Rugo replied. “ So no, we didn’t use immune RECIST. But it’s an interesting area to go back and look at, to see whether or not we can reclassify any of the responses.”
Giving some background to the research, she noted that PD-L1 expression has been inversely correlated with ER expression. A previous trial found an overall response rate of about 19% in triple-negative breast cancer (SABCS 2014, abstract S1-09 ).
For KEYNOTE-028, patients were screened for PD-L1 positivity, even though pembrolizumab targets PD-1, because there is much greater variability in ligand levels than in receptor levels when it comes to determining T cell inactivation, Dr. Rugo explained.
Fully 80% of the 25 women treated had received three or more prior lines of therapy for their advanced disease. They were give pembrolizumab every 2 weeks with assessments every 8 weeks initially.
The grade 3 or 4 adverse events observed were cases of autoimmune hepatitis, increased gamma-glutamyl transferase levels, muscle weakness, nausea, and septic shock. However, there were no treatment-related deaths.
In patients who developed any-grade immune-related adverse events of interest, just one – a patient with autoimmune hepatitis – had to have treatment interruption.
All of the responses observed were partial responses. The median time to response was 8 weeks, and the median duration of response was not reached, with a range from about 9 to 44 weeks. The three patients with a response had been in the study for at least 26 weeks as of the data cutoff for analysis.
“Cost is an issue with these drugs,” said attendee Dr. Mark Graham II, an oncologist with Waverly Hematology Oncology, Cary, N.C. “I’m particularly interested in your nonresponders and the fact that we can see pseudoprogression with these drugs. So for the eventual nonresponders, …what is the likely number of cycles that will be necessary to conclude that a patient is not an initial responder?”
That number is difficult to accurately pin down, as patients in the trial were taken off the agent as early as 8 weeks along if they had any evidence of progression, Dr. Rugo said. “I really don’t think we know the answer yet, but if you went out to 16 weeks, all the patients that we saw who were going to respond had responded by 16 weeks.”
An accurate answer will likely require future studies, she added. “I think it’s most complicated in triple-negative disease, where a small phase Ib trial showed a very long median time to response. So this is a good question and it remains to be answered.”
If sufficient tissue is available, the investigators plan to look for other biomarkers of pembrolizumab benefit, such as proliferation markers and intrinsic subtype, according to Dr. Rugo. “Probably those more in-depth investigations are going to occur with future studies, just because of the need to acquire enough tissue to do them. This trial was exploratory,” she said.
Dr. Rugo disclosed that her institution receives research funding from Merck & Co. – the trial sponsor – and Genentech.
