David Reardon, MD
Pembrolizumab had a 6-month progression-free survival (PFS) rate of 44% and a manageable safety profile for patients with recurrent PD-L1-positive glioblastoma multiforme (GBM), according to findings from the phase Ib KEYNOTE-028 trial presented at the 2016 Society for Neuro-Oncology Annual Meeting.
Median PFS was 3 months (95% CI, 2-9) and the 12-month PFS rate was 16%. The median overall survival (OS) was 14 months (95% CI, 10 to not reached), the 6-month OS rate was 85%, and the 12-month OS rate was 74%. Lead investigator, David A. Reardon, MD, noted that the Kaplan-Meier curves showed that a small tail of patients have durable benefit.
“The durability of benefit was the most striking indicator of the efficacy of pembrolizumab for these patients. Some of these patients are up over 80 weeks and maintaining progression-free survival, which is remarkable,” said Reardon, from the Dana-Farber Cancer Institute. “The efficacy was encouraging, but needs to be taken into the context of a very small single-arm study.”
The study enrolled 26 patients who were positive for PD-L1 on ≥1% of cells in tumor nests or positive bands in stroma. To determine PD-L1 status, tumor samples, which could be archival or newly obtained from nonirradiated lesion, were assessed through immunohistochemistry using a prototype assay and the 22C3 antibody clone.
The median age of patients was 55.5 years (range, 33-76) and the ECOG performance status was 0 for 11 patients (42%), 1 for 14 (54%), and missing for 1 (4%). Prior to enrollment, all patients had received standard of care therapies, and 31% had experienced 2 or more recurrences. The median time from original diagnosis to study enrollment was 20.0 months (range, 7.6-81.6).
After patients progressed on the study, 6 had surgery (23%), 1 had radiation (4%), and 10 had oncologic drugs and biologics (39%). Reardon explained that immunotherapies, such as pembrolizumab, may help subsequent treatments because of the effects on the immune system. So, this study is tracking subsequent treatments after patients progress.
One patient experienced a partial response (4%). Stable disease was achieved in 12 patients (48%). The disease control rate, which was 28%, was defined as the rate of complete response plus partial response plus stable disease for ≥6 months. Progressive disease occurred in 10 patients (40%).
Reardon shared images of target lesions in a patient with a confirmed partial response as assessed by RECIST v1.1 by the investigator. The lesion was visible at baseline and then no longer apparent at 40 weeks. A waterfall plot of best change from baseline in tumor size by RECIST v1.1 investigator review illustrated that 46% of the patients had experienced a decrease in target lesions, with 4 patients having a decrease of 30% or more.
The median time to response was 36.3 weeks (range, 7.7-37.6 weeks). The median response duration was 12.1 weeks (range, 8.3-12.1 weeks). The median duration of stable disease was 39.4 weeks (range, 7.1+ to 85.9+ weeks).
Overall, 15% of patients experienced a grade 3/4 adverse events (AE), a low enough rate for Reardon to label it as “Fairly reassuring.” Less than 10% of the patients experienced immune-related AEs, which included colitis in 2 patients (8%), hypothyroidism in 2 (8%), hyperthyroidism in 2 (8%), and drug eruption in 1 (4%). The median follow-up was 60.9 weeks (range, 11.7-94.0). No grade 4 cerebral edema occurred. No treatment-related deaths or discontinuations occurred.
“This agent was quite well tolerated for the glioblastoma population. There were really no unusual or unexpected side effects that occurred, which is always reassuring,” said Reardon.
Reardon stated that further investigation of anti-PD-1 therapy in this patient population is warranted. Immunocorrelative analyses are ongoing, including work to correlate the level of PD-L1 expression in a tumor with outcomes.
Reardon DA, Kim T-M, Frenel J-S, et al. Results of the phase Ib KEYNOTE-28 multi-cohort trial of pembrolizumab monotherapy in patients with recurrent PD-L1-positive glioblastoma multiforme (GBM). Neuro-Oncology. 2016; 18:abstract ATIM-35.<<< View more from the 2016 SNO Annual Meeting