Abstract 1339: Transcriptomic profiles improve characterization natural killer cells and predict response to cytokine therapy in clinical trial for dogs with pulmonary metastases

Abstract

Abstract Introduction: Natural killer (NK) cells are key effectors of the innate immune system, but major differences between human and murine NK cells have been a barrier to translation. Outbred dogs are an important link for NK-based cancer immunotherapy studies, so better characterization of canine NK cells is needed. We used RNAseq to compare gene expression profiles of ex vivo dog NK cells to in vivo NK signatures from dogs with pulmonary metastases receiving inhaled recombinant human (rh) IL-15 in a phase I clinical trial. Methods: Eight dogs with spontaneous pulmonary metastases were enrolled on an IACUC and clinical trials review board-approved Phase I clinical trial of inhaled rhIL-15 using a 3+3 cohort design with escalating doses of inhaled rhIL-15. Blood was collected from study subjects immediately pre-treatment and on days 7, 14 and 21 after treatment initiation for isolation of NK cells (CD5dim) and RNAseq. We performed differential gene expression (DGE) comparing trial patients to healthy beagle CD5dim NK populations (resting) and ex vivo activated beagle NK cells using IL-15 and feeder line co-culture. We assessed global transcriptional profile, specific activating and inhibitory receptors of NK function, and principal component analysis (PCA) for variation between treatment groups (FDR <0.05). Results: Of 8 dogs, 2 demonstrated > 100-day survival with 1 stable disease and 1 partial response based on RECIST criteria. DGE revealed distinct transcriptional profiles between the ex vivo resting, IL-15 and co-cultured canine NK cells. Among treated patients, hierarchical clustering revealed that in vivo NK cell transcriptional signatures grouped by individual dog, and not amount of time exposed to treatment. PCA showed in vivo profiles of the long term survivors were distinctly separate from the non-responding patients (PC1 38%, PC2 12%). This suggests response to therapy could be determined by baseline NK cell characteristics rather than changes over time following treatment. Patient in vivo NK cell transcription profiles most closely resembled those of ex vivo resting NK cells and not IL-15 or co-culture activated (PC1 43%, PC2 19%), likely reflecting key differences in activation ex vivo vs in vivo. Key genes induced in vivo (>20X) post inhalation of rhIL-15 include DLA-DRA, B2M, and thymosin beta 4, while key genes induced ex vivo post rhIL-15 exposure include CD96, KLRB1, and SPP1/OPN. Conclusion: In this first transcriptomic sequencing of dog NK cells, we demonstrate distinct gene profiles of ex vivo activated NK cells from healthy donors compared to circulating NK cells from dogs receiving inhaled rhIL-15 on clinical trial. Baseline NK cell profiles appear to predict response more than changes over time. These data highlight the strength of the outbred dog model in speeding novel immunotherapy and biomarker studies. Citation Format: Alicia A. Gingrich, Taylor Reiter, Sean Judge, Daniel York, Mio Yanagisawa, Ian Sturgill, Rachel Brady, Kevin Stoffel, Arta Monjazeb, C. Titus Brown, Robert Rebhun, Robert J. Canter. Transcriptomic profiles improve characterization natural killer cells and predict response to cytokine therapy in clinical trial for dogs with pulmonary metastases [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1339.