572 Single cell analyses reveal functional heterogeneity within an exhausted CD8⁺ TIL population that is correlated with response to checkpoint therapy in melanoma

<h3>Background</h3> Immunotherapy agents, including PD-1 pathway inhibitors, have revolutionized cancer treatment. However, most patients do not receive long term clinical benefit from PD-1-based immunotherapy, highlighting a major clinical need to better understand the underlying immunological drivers of response to immunotherapy. Preclinical studies have focused heavily on progenitor-like exhausted CD8 T cells (termed T_PEX), since prior work has demonstrated that the T_PEX population proliferates in response to PD-1 inhibitors more than other exhausted T cell (termed T_EX) subpopulations. However, these studies have also shown that checkpoint inhibitor (ICI) treatment drives T_PEX to differentiate into T_EX populations capable of mediating potent anti-tumor activity. These data have complicated our ability to identify which T cell populations are relevant to monitor in patients to understand mechanisms contributing to anti-tumor immune responses at the time of tumor sampling. We previously identified two major population of CD8 T cells in tumors: those with high PD-1 and CTLA-4 levels and those with low levels. Tumors with more PD-1 and CTLA-4 cells responded better to checkpoint inhibitors. In this follow-up study, we investigated whether this population correlated with progression-free and overall survival, as well as characterized heterogeneity in response. <h3>Methods</h3> Melanoma patients were recruited and treated either with nivolumab or a combination of nivolumab/ipilimumab. The drug responders and non-responders were classified based on radiological findings. Further, the combination of flow cytometry, bulk RNA seq, single cell RNA seq, and TCR sequencing was used to characterize the heterogeneity in the CD8 tumor-infiltrating lymphocytes (TILs) based on co-expression (CP^Hi) or lack (CP^Lo) of PD-1 and CTLA-4. <h3>Results</h3> In this study, we found higher levels of CP^Hi TILs correlated with objective response rate and survival in patients with melanoma. A substantial heterogeneity within this CP^Hi TIL population, including discrete cell features such as glycolytic activity and/or tissue residency, was observed. We also found that coinhibitory receptor expression was inversely linked with costimulatory receptor expression and correlated with distinct cellular features. Critically, transcriptional features of the T_PEX subset were predominantly present in the CP^Lo population and were not correlated with response, likely reflecting the fact that after ICI treatment, T_PEX differentiated into other subsets that mediate anti-tumor activity. <h3>Conclusions</h3> Our data show that the CP^Hi TIL population that correlates with response in melanoma is not one uniform population, but rather a heterogenous mix of different subpopulations with discrete cellular processes that may differentially contribute to anti-tumor immunity following checkpoint blockade. <h3>Acknowledgements</h3> We would like to acknowledge the Parnassus Flow Cytometry Core at the University of California, San Francisco, supported in part by National Institutes of Health (NIH) grant P30 DK063720 and NIH S10 Instrumentation Grant S10 1S10OD021822-01 for sorting human samples. K.E.P. was supported by The University of Texas Rising Stars Award, The Andrew Sabin Family Foundation Fellowship, and The University of Texas MD Anderson Cancer Center Melanoma SPORE Developmental Research Program grant. A.I.D. and M.D.R. were supported by a generous gift from the Cook, Amoroso, and Kelly Funds. A.I.D. was also supported by the Parker Institute for Cancer Immunotherapy. A.H.S. was supported by NIH grants P01AI039671, P01AI108545, and P01CA236749. J.M.S. was supported by grants from the NIH (7K08CA256044-3) and is a Cancer Prevention and Research Institute of Texas (CPRIT) Scholar in Cancer Research. <h3>Ethics Approval</h3> Studies using human melanoma specimens were approved by the UCSF Committee on Human Research (CC138510) and by the Institutional Review Board of UCSF (protocol 13-12246).