A phase II single-arm study of pembrolizumab with enzalutamide in men with metastatic castration-resistant prostate cancer progressing on enzalutamide alone

Abstract

Background: Checkpoint inhibitors can induce profound anticancer responses, but programmed cell death protein-1 (PD-1) inhibition monotherapy has shown minimal activity in prostate cancer. A published report showed that men with prostate cancer who were resistant to the second-generation androgen receptor inhibitor enzalutamide had increased programmed death-ligand 1 (PD-L1) expression on circulating antigen-presenting cells. We hypothesized that the addition of PD-1 inhibition in these patients could induce a meaningful cancer response.

Methods: We evaluated enzalutamide plus the PD-1 inhibitor pembrolizumab in a single-arm phase II study of 28 men with metastatic castration-resistant prostate cancer (mprogressing on enzalutamide alone. Pembrolizumab 200 mg intravenous was given every 3 weeks for four doses with enzalutamide. The primary endpoint was prostate-specific antigen (PSA) decline of ≥50%. Secondary endpoints were objective response, PSA progression-free survival (PFS), time to subsequent treatment, and time to death. Baseline tumor biopsies were obtained when feasible, and samples were sequenced and evaluated for the expression of PD-L1, microsatellite instability (MSI), mutational and neoepitope burdens.

Results: Five (18%) of 28 patients had a PSA decline of ≥50%. Three (25%) of 12 patients with measurable disease at baseline achieved an objective response. Of the five responders, two continue with PSA and radiographic response after 39.3 and 37.8 months. For the entire cohort, median follow-up was 37 months, and median PSA PFS time was 3.8 months (95% CI: 2.8 to 9.9 months). Time to subsequent treatment was 7.21 months (95% CI: 5.1 to 11.1 months). Median overall survival for all patients was 21.9 months (95% CI: 14.7 to 28 .4 months), versus 41.7 months (95% CI: 22.16 to not reached (NR)) in the responders. Of the three responders with baseline biopsies, one had MSI high disease with mutations consistent with DNA-repair defects. None had detectable PD-L1 expression.

Conclusions: Pembrolizumab has activity in mCRPC when added to enzalutamide. Responses were deep and durable and did not require tumor PD-L1 expression or DNA-repair defects.

Trial registration number: clinicaltrials.gov (NCT02312557).

Keywords: biomarkers; combination; drug therapy; immunotherapy; lymphocytes; prostatic neoplasms; tumor; tumor-infiltrating.

Figure 1

Radiographic and PSA changes. (A) Radiographs of the responders who had measurable disease. (B) Spider plot for changes in radiographic response for participants with measurable disease. (C) Changes in PSA for all participants. PSA, prostate-specific antigen; RECIST PD, RECIST progressive disease; RECIST SD, RECIST stable disease.

Figure 2

Distribution of mutational burden in responder’s versus non-responders. Boxplot showing the distributions of mutational burden (left) and neoepitope count (right) for responders versus non-responders. The center line in each box represents the median, with bottom and top boundaries of the box representing the first and third quartiles, respectively. Lines extending down and up from the box represent the minimum and maximum values, respectively. Non-responders are shown in light blue, and responders are shown in dark blue. There was no statistically significant difference between responders and non-responders in terms of mutational burden (median 6.5 vs 4.9 variants per megabase pair covered, p=0.42) or predicted neoepitope burden (679 vs 405.5 neoepitopes, p=0.42).

Figure 3

Pembrolizumab plus enzalutamide treatment induced systemic immune responses. (A) Serum CXCL10 levels pretreatment (C1D1; Cycle 1 Day 1) and post-treatment (C2D1, Cycle 2 Day1; C3D1, Cycle 3 Day 1) are shown (n=22). (B) The percent expression of CD38, HLA-DR, or Ki-67 in EM CD4 or CD8 T cells is shown (n=18). EM, effector memory; NS, not significant.

Figure 4

The functional status of EM CD8 T cells as a potential immune correlate with clinical response and outcome. (A) The percent expression of GzmB or perforin in EM CD8 T cells is compared between non-responders (NR; n=14) and responders (R; n=4). (B) Kaplan-Meier curves demonstrating that patients with higher median percentage of perforin, but not GzmB, EM CD8 T cells at baseline had prolonged PFS. (C) The frequency of MDSCs at baseline did not correlate with clinical response or outcome. EM, effector memory; GzmB, granzyme B; MDSCs, myeloid-derived suppressor cells; PFS, progression-free survival.