ATIM-40. CIRCULATING MYELOID-DERIVED SUPPRESSOR CELLS PREDICT FAVORABLE RESPONSE TO IMMUNE CHECKPOINT THERAPY IN A RANDOMIZED TRIAL OF NIVOLUMAB AND BEVACIZUMAB IN RECURRENT GBM Free

Glioblastoma (GBM) creates an immunosuppressive environment that presents a challenge to efficacy of immunotherapeutic approaches. Results from the CheckMate-143 trial demonstrated responses in 8% of patients with nivolumab, underscoring the need for further insight into the mechanisms and markers of immune suppression and response. Given a limited set of biomarkers predictive of immunotherapy response in GBM, we explored the changes in immune cell populations in nivolumab and bevacizumab-treated GBM patients pre and post-treatment in order to help predict response. In these studies, we utilized traditional and newly developed approaches, including mass cytometry time-of-flight (CyTOF), single-cell RNA sequencing, and 10X Genomics simultaneous cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq). We analyzed patients’ samples in a randomized, phase 2 study of nivolumab and bevacizumab at GBM first recurrence (NCT03452579). Nine patients were identified as responders or non-responders at 8 weeks after therapy initiation. Utilizing peripheral blood samples, we observed a 6.4-fold decrease in immunosuppressive myeloid-derived suppressor cells (MDSCs) between baseline and first imaging follow-up in responders compared to non-responders, with a 4.9-fold decrease in the granulocytic MDSC (G-MDSC) subtype in responders over non-responders. While no significant changes in overall T-cell numbers were noted, expression of PD-1 on CD4+ T cells was significantly elevated at baseline and follow-up in responders as compared to non-responders – signatures which were confirmed by CyTOF. Given these immunophenotypic changes, preliminary results of a detailed investigation of this cohort by CITE-seq indicate that responders had increased IL7R-positive T cells post-treatment, which was not observed in non-responders. These results are currently being validated in an additional 40 patients that have been enrolled. Altogether, differences in immunophenotypes that were specific to responders and non-responders were observed, and characterization of these immune populations may be helpful in identifying GBM patients likely to benefit from immunotherapy.