808?Exploring resistance mechanism to pembrolizumab and ang-2 inhibitor trebananib (NCT03239145) using high-dimensional single-cell mass cytometry (CyTOF)

摘要

Background Angiogenesis is mediated by both the vascular endothelial growth factor (VEGF) family and the angiopoietin (Ang1-2)/Tie-2 pathway. 1-3 We demonstrated that increased soluble VEGF and ang-2 is associated with decreased benefit to immune checkpoint inhibitors (ICIs). 4 We then initiated a clinical trial combining pembrolizumab and ang-1/2 inhibitor (trebananib) ( NCT03239145 ) with expansion cohorts in microsatellite stable (MSS) colorectal cancer (CRC), ovarian and renal cell cancer. 5 We present the correlative analysis using high-dimensional single-cell mass cytometry (CyTOF) to characterize the effects of the combination therapy and examine differences between patients according to clinical benefit. Methods We used two separate CyTOF panels to monitor 48 markers of innate and adaptive immune populations in 26 evaluable patients who received the PR2D of trebananib (30mg/kg). Mass cytometry assay was performed on peripheral blood mononuclear cells of 26 patients at baseline (C1D1), 16 patients at cycle 3 day 1 (C3D1), and 4 patients at cycle 9 day 1 (C9D1). We compared immune cell markers between patients with clinical benefit (CB) and patients with no clinical benefit (NCB). Results Of 26 patients (16 CRC, 8 ovarian, 2 RCC), 11 patients had confirmed PR (3) or SD (8) resulting in CB of 42.3% while 15 patients had NCB. Independent of CB, there were statistically significant decreases from C1D1 to C3D1 in na?ve CD8 T cells (p=0.03), CD4 T central memory cells (p=0.05), and PD-1 /CD4 and CD8 T cells (p0.0001). In NCB patients, there was a statistically significant decrease from C1D1 to C3D1 in CD3 T cells (p=0.009), CD4 /CXCR3 T cells (p=0.02), CD8 /CXCR3 T cells (p=0.02), and CD8 T effector memory cells (p=0.03) while no significant changes in these T cell populations were observed in CB patients (figure 1). In NCB patients, monocytic myeloid-derived suppressor cells (p=0.003) and classical monocytes increased from C1D1 to C3D1 (p=0.01) while there was no significant change in this population in CB patients (figure 2). Interestingly, CB patients had higher activated CD56 NKp30 at baseline (p= 0.03) with increased cytolytic CD56 dim CD16 population from C1D1 to C3D1 (p= 0.04) compared to NCB patients (figure 3). Abstract 808 Figure 1 T cell subset analysis by cycle and clinical benefitDetection of T cell subsets at C1D1, C3D1, and C9D1. (A) CD3 T cells decrease from C1D1 to C3D1 in patients with no clinical benefit (p=0.009, n=16). CD3 T cells are significantly higher at C3D1 in patients with clinical benefit (p=0.02, n=16). (B) CD4 T cells decrease in patients with no clinical benefit from C1D1 to C3D1 (p=0.01, n=16). (C) CD8 T cells decrease in patients with no clinical benefit from C1D1 to C3D1 (p=0.03, n=16). (D) CD8 T effector memory cells decrease significantly in patients with no clinical benefit between C1D1 and C3D1 (p=0.03, n=16). (E) CD4 /CXCR3 cells decrease significantly in patients with no clinical benefit from C1D1 to C3D1 (p=0.02, n=16). (F) CD8 /CXCR3 cells decrease significantly from C1D1 to C3D1 in patients with no clinical benefit (p=0.02, n=16). Abstract 808 Figure 2 Myeloid cell subset analysis by cycle and clinical benefitDetection of myeloid cell subsets at C1D1, C3D1, and C9D1. (A) Myeloid cells increase significantly from C1D1 to C3D1 in patients with no clinical benefit (p=0.01, n=16). (B) Monocytic myeloid-derived suppressor cells increase significantly in patients with no clinical benefit from C1D1 to C3D1 (p=0.003, n=16). (C) Dendritic cells are significantly higher at C1D1 (p=0.02, n=26) and C3D1 (p=0.02, n=16) in patients with no clinical benefit. (D) Total monocytes significantly increase in patients with no clinical benefit from C1D1 to C3D1 (p=0.02, n=16). (E) Classical monocytes increase in patients with no clinical benefit from C1D1 to C3D1 (p=0.01, n=16). (F) M2 macrophages trend higher in patients with no clinical benefit at C3D1 (p=0.07, n=16). Abstract 808 Figure 3 NK cell subset analysis by cycle and clinical benefitFigure 3: Detection of NK cell subsets at C1D1, C3D1, and C9D1. (A) There is a trend towards increased CD56dim/CD16- cells in NCB patients compared to CB patients at C3D1 (p = 0.08). (B) CD56dim/CD16 cells are significantly higher in CB patients compared to NCB patients at C3D1 (p=0.04). (C) There were no significant differences in CD56bright cells according to cycle or clinical benefit. (D) CD3-/CD19-/CD56 cells are significantly higher in CB patients compared to NCB patients at C3D1. There is a trend towards decrease of this cell subset from C1D1 to C3D1 in NCB patients (p=0.06). (E) NKp30 /CD56 cells are significantly higher in CB patients compared to NCB patients at C1D1, but this was not significant at C3D1 (p=0.23). Conclusions Our findings suggest that the activity of anti-PD-1 and ang-2 peptibody (trebananib) combination is hindered by an increase in immunosuppressive myeloid cells leading to decrease in memory and effector T cell p