We sought to determine the subcellular distribution of 30+ immune-related markers including PD-L1 in stage III melanoma tumor tissues (nuclear, cytoplasmic, surface) using novel imaging mass cytometry technology that enables simultaneous detection of over 30 immune and tumor markers. Metal-tagged antibodies were used to stain stage III melanoma tissues similarly to an IHC-based protocol, followed by laser ablation and mass cytometry using the Helios® and Hyperion® technology (Fluidigm). Markers for immune subsets included CD45, CD3, CD4, CD8, PD-1, Granzyme B, FoxP3, CD11b, CD11c, CD107, CD20, Vista, and CD68. Markers for tumor cell phenotyping included PD-L1 (intracellular and extracellular domain specific), E-cadherin, B-catenin, PD-L2, and keratin 8/18. Structural, inflammatory, survival, and signaling markers included Ki67, p53, p-tyrosine, Bcl-2, Cl-caspase 3, histone H3, collagen-1, actin, and arginase-1. We further validated subcellular distribution of PD-L1 in the same patient tumor cores using fluorochrome-labeled antibodies and 100X confocal microscopy to show difference in surface-only, cytoplasmic-only, and nuclear PD-L1. We then assessed the association between subcellular PD-L1 and immune populations in the tumor microenvironment, along with overall and recurrence-free survival. We wrote custom algorithms for data processing to identify novel immune subsets, ran SPADE and viSNE analyses for cross-validation, and employed Imaris Bitplane software to validate our own algorithms for immune population identification. Overall, we found that intracellular-only PD-L1 in tumor tissues has the highest correlation (Pearson r) to CD11b, FoxP3, PD-1, arginase, Ki67, PD-L2 and nuclear stain iridium, while extracellular (surface) PD-L1 in tumors has the highest correlation (Pearson r) to FoxP3, PD-1, arginase, Ki67, PD-L2 and Vista. We identified two novel immunophenotypes (histone-3+/p-tyrosinehi /Vistalo / PD-L1-ICDlo / PD-L1-ECDlo and CD11bhi / PD-L1-ICDhi / PD-L1-ECDhi) that are positively associated with overall patient survival. Of note, only phospho-Tyrhi was significantly associated with RFS. This work highlights the use of cutting-edge technology that can employ the highest multiplexing available for assessing novel interactions of 30+ tumor tissue markers for identification of novel immunophenotypes and biomarker in stage III melanoma patients. Our novel study showcases a novel methodologic approach to high-throughput analysis of retrospective profiling of the melanoma tumor immune microenvironment on a single IHC slide, followed by appropriate and rigorous cross-validation using Fluidigm’s novel CyTOF imaging technology. Our study provides a groundwork for this powerful, novel technology in translational and clinical research for biomarker identification and identification of novel combinatorial markers that may predict overall and recurrence-free survival in stage III melanoma.

Citation Format: Heather G. Hambright, Anand V.R. Kornepati, Dai Ogata, Roland R.L. Bassett, Suhendan Ekmekcioglu, Elizabeth A. Grimm, Tyler J. Curiel. High-dimensional (30-plex) imaging mass cytometry on tissue microarray identifies novel PD-L1-inclusive immunophenotypes associated with overall survival in stage III melanoma [abstract]. In: Proceedings of the AACR Special Conference on Melanoma: From Biology to Target; 2019 Jan 15-18; Houston, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(19 Suppl):Abstract nr B10.