Background/Purpose: Anti-PD-1/PD-L1 checkpoint inhibitor therapies have produced remarkable results in stimulating immune responses against tumors. Checkpoint inhibitor therapy can also trigger a variety of autoimmune responses; however, the mechanisms underlying these immune-related adverse events (IrAEs) remain unclear. Checkpoint inhibitor-related arthritis (CIrA) occurs in ~5% of treated patients, and synovial fluid from these patients provides a unique opportunity to study the phenotypes of immune cells from the inflamed site. The extent to which the inflammatory response in CIrA resembles RA or PsA is unknown.

Methods: We conducted detailed immunophenotyping of cryopreserved synovial fluid mononuclear cells (SFMC) from CIrA following anti-PD-1/PD-L1 treatment (n=9), seropositive RA (n=5), and PsA (n=5) patients using a 39-marker mass cytometry (CyTOF) panel. CIrA patients had varied cancers, and 5/9 had mono/oligoarticular arthritis. We identified and custom-conjugated an anti-PD-1 CyTOF antibody using a clone that is not blocked by pembrolizumab or nivolumab, allowing detection of PD-1 in CIrA samples. We used FlowSOM to cluster either CD4 or CD8 T cells into 15 ‘metaclusters’ (cell populations) based on their multi-dimensional phenotypes. We used Kruskal-Wallis or Mann-Whitney tests to identify significantly altered populations (p< 0.05), which we confirmed by biaxial gating.

Results: T cells were the most abundant population in CIrA (50% of SFMC. 53% CD4, 40% CD8), followed by monocytes (24%) and NK cells (8%); they were comparable to those in RA and PsA. Cells in CIrA samples showed expression of PD-1 comparable to that in RA and PsA. However, expression of PD-L1 and PD-L2, both frequently expressed on monocytes, was increased in CIrA compared to RA and PsA (Fig. 1, PD-L1: 2.2-fold increased over RA/PsA, p=0.0021; PD-L2: 2.4-fold increased over RA/PsA, p< 0.0001). FlowSOM analysis of CD4 T cells highlighted distinct cell populations in CIrA and RA. RA samples contained one significantly expanded population (metacluster5, 30% of CD4s in RA, p=0.006), which contained PD-1hiICOS+ T peripheral helper (Tph) cells. Tph cells were not expanded in CIrA. In contrast, CIrA samples showed a specific expansion of a distinct population of PD-1+CD38hi CD127-cells (metacluster2, 10% of CD4s in CIrA, 2.4-fold increased over RA/PsA, p=0.0047), which we confirmed by biaxial gating (Fig. 2). Interestingly, analysis of CD8 T cells by FlowSOM revealed expansion in CIrA of a CD8 cell population with a similar phenotype to the expanded CD4 population: PD-1+CD38hiCD127-(CD8 metacluster2, 30% in CIrA, 3.4-fold increased over RA/PsA, p=0.0002). Over 40% of PD-1+CD38hi CD127-CD8+T cells in CIrA expressed Ki67, suggesting active proliferation (Fig. 3).

Conclusion: CyTOF analysis of SFMC revealed uniquely expanded CD4 and CD8 T cell populations in CIrA that are not shared with RA or PsA. Both the expanded CD4 and CD8 populations show a PD-1 ⁺ CD38 ^hi CD127 ^– phenotype. These cell phenotypes may be directly enhanced by PD-1/PD-L1 blockade and may contribute to the amplified immune response. Functional analysis of these cells may reveal key mechanisms driving CI-associated IrAEs.

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/mass-cytometry-immunophenotyping-of-synovial-fluid-from-checkpoint-inhibitor-related-arthritis/