Immune checkpoint inhibitors (ICI) cause durable tumour responses in some patients, but other treated patients do not respond to ICI. Profiling immune changes in responders and non-responders to ICI provides mechanistic insight that will guide development of predictive biomarkers, and strategies to improve ICI outcomes. Our study profiled dynamic changes in T cell repertoires from responding and non-responding animals with bulk T cell receptor beta (TCRβ) and single cell 5’VDJ sequencing.
We utilised an established bilateral tumour model that controlled variation in host genetics and tumour mutation burden. Bilateral tumours either grew or regressed symmetrically following treatment with ICI. Bilateral tumours within an animal also had highly similar TCRβ repertoires, allowing us to sample one tumour in time whilst leaving the other as a readout for ICI response. A source of variation came from the naturally diverse TCRβ repertoire between animals, allowing us to assess TCRβ repertoire’s contribution to ICI outcome. Tumours were sampled prior to, and multiple days post ICI.
ICI responders had reduced tumour TCRβ diversity earlier in time compared to non-responders, and differences were even observed prior to ICI. Reduced TCRβ diversity corresponded with expansion of limited number of TCRβ clonotypes, of which some were specific for tumour associated self-antigen. We clustered similar TCRβ sequences by machine learning and found multiple TCRβ clusters that increased earlier in time in responders, suggesting clonotype expansion against a wider range of antigens. Single cell RNAseq revealed that clonally expanded CD8+ TILs in non-responders post ICI had increased T cell exhaustion and dysfunction gene expression compared to their counterparts in responders.
ICI responding tumours get a head start in terms of T cell clonal expansion, and non-responding tumours are characterised by late expansion of exhausted TIL clones. Unique dynamic changes occurring early during ICI could inform potential biomarkers of response.