Identification Of A Novel Immunosuppressive Myeloid Gene Expression Signature For Clinical Biomarker Development.

Abstract Cell-type abundance GE signatures are useful for informing drug mechanism of action and may be useful in-patient selection for cancer immunotherapies. Recently, we validated natural killer cell- and dendritic cell-type specific expression signatures using a tiered strategy that included both computational and ex vivo validation. Here, we built upon this strategy to validate expression signatures for immunosuppressive myeloid cells (IMC). IMC play a critical role in impairing anti-tumor immunity and increased levels of peripheral IMC have been associated with advanced tumor progression and poor prognosis in various cancers. However, their low abundance in normal tissues and heterogeneous surface expression render their profiling difficult. To address this, we focused on validating GE signatures for IMC abundance using mRNA expression methods which are more clinically tractable than established flow cytometry methods that require intact cells. We implemented a three-stage GE signature validation strategy. First, we generated GE data from human ex vivo differentiated IMC (LIN- CD11B+ CD33+ HLADR-). Then, we evaluated the concordance of 16 previously published myeloid signatures with cell type abundance in a series of spike-ins with varying but known quantities of ex vivo differentiated IMC in a background of undifferentiated peripheral blood mononuclear cells (PBMCs). Gene-set variation analysis (GSVA) was used to score the signatures and Spearman's rank (rS) correlation coefficient was calculated to assess the significance of correlations between GSVA scores and IMC abundance. With this method, we validated 3 of the 16 published myeloid signatures (1. MacB3/PMID:26873985 (rS = 0.87; p = 8.8e-07); 2. MacB2_3w/PMID:26873985 (rS = 0.83; p = 6.7e-06); 3. TAM/PMID:27424807 (rS = 0.80; p= 2.6e-05)). Next, we generated a ‘de novo' signature by performing differential GE analysis and identified up-regulated genes in IMC in comparison to PBMCs with log2 fold change > 2, p-adj < 0.05). We further refined this list by selecting genes with significant concordance to IMC abundance (rS > 0.80, p-adj < 0.05) and confirmed their specificity to IMC in published scRNAseq studies [PMID: 30979687; 31033233; 32302573]. This led to a 6-gene ‘de novo' signature (MRC1, APOE, C1QA, C1QB, MMP9, SPP1) associated with IMC function with highly significant concordance (rS = 0.96; p = 1.2e-11) to IMC abundance. Lastly, we showed that in whole blood baseline samples from HNSCC patients (n=53) treated with danvatirsen + durvalumab [NCT02499328], our ‘de novo' signature was significantly elevated (p=0.033) in patients with progressive disease compared to complete responders. Application to additional studies will be required to more fully determine its clinical utility. Citation Format: Srimathi S. Srinivasan, Gozde Kar, Deanna L. Russell, Peter Gathungu, Minoo Rafati, David Whitston, Grant Duclos, Ben Sidders, J Carl Barrett, Tim Slidel, Patricia McCoon. Identification of a novel immunosuppressive myeloid gene expression signature for clinical biomarker development [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 390.