Immune Checkpoint blockade for the treatment of metastatic melanoma is associated with highly variable clinical outcomes between individuals in terms of both oncological benefit and immune-related adverse events. Early markers of response are urgently sought. Whereas numerous intra-tumoural determinants of sensitivity to immunotherapy are known, the identification of peripheral predictors of response is limited. We aimed to characterize the CD8+ T-cell transcriptomic and clonal changes to treatment across a large cohort of patients in an effort to gain further insight into the peripheral markers of response.
We performed paired-end 75bp read RNA-sequencing to assay the peripheral CD8+ response at baseline and across multiple cycles of treatment (n = 105 patients, 69 controls, 315 separate transcriptomes). We validated identified clonal subsets indicative of response using 10X Chromium single cell sequencing (16 samples, 8 patients), flow-cytometry and targeted PCR.
After adjusting for multiple testing, we identify >5,800 transcripts induced by treatment. These fall into discrete gene modules, with several markedly diverging between combination immunotherapy and anti-PD1 alone. Patients demonstrating a durable radiological response to checkpoint immunotherapy (absence of progression at 6 months) have significantly increased numbers of large peripheral CD8+ circulating clones by day 21 after treatment, compared to non-responders and controls. We replicate this observation in a separately recruited cohort. Large peripheral clones have a distinct gene expression profile, characterized by high expression of CCL5, BCL2L1and NKG7 amongst other genes.
We identify robust and reproducible predictors of 6 month clinical and radiological responses to immune checkpoint blockade in the transcriptomes of peripheral circulating CD8+ T-cells from metastic melanoma patients after 21 days of treatment. These observations can be used to further our understanding of determinants of patient response, and may provide a mechanism for early treatment stratification of patients with a non-favourable peripheral profile.
Oxford Radcliffe Biobank.
Wellcome Trust, Oxford NIHR Biomedical Research Centre, Cancer Research UK.
All authors have declared no conflicts of interest.