Abstract P1-07-05: Gene expression analysis of HOXB13-high and -low tumors reveals a dichotomous immune landscape

Abstract

Abstract Background: Despite the independent development and validation of the HOXB13/IL17RB gene expression ratio as a prognostic and predictive biomarker, the specific molecular underpinning of these two genes, as it relates to their ability to predict endocrine therapy benefit remains unknown. In this study, we performed gene-expression pathway analyses comparing ER+ HER2- HOXB13-high versus HOXB13-low breast tumors in The Cancer Genome Atlas (TCGA) RNAseq dataset. Methods: Gene expression, copy number, and clinical data from TCGA BRCA samples were obtained from a standardized Firehose run from the Broad Institute GDAC website (http://gdac.broadinstitute.org/runs/stddata__2015_11_01/data/BRCA/20151101). Patients with annotated receptor status ER+ and Her- were included, yielded a final set of 428 patients for downstream analysis. These patients were then stratified by HOXB13 expression. First, raw counts from the TCGA RSEM pipeline were normalized with DESeq2, and then we defined a cohort of 128 HOXB13-high patients (normalized HOXB13 counts >= 100), and 242 HOXB13-low patients (normalized HOXB13 counts <= 10). Genes differentially expressed between the HOXB13-high and HOXB13-low cohorts were identified using DESeq2. Genes were ranked by log2 fold change and used to find genesets enriched for upregulated or downregulated genes using GSEA in pre-ranked mode, interrogating either the Hallmark (H) or Curated (C2) sets of MSigDB genesets. Genesets with an FDR q-value of 0.25 or less were considered significant. We also examined literature-curated immunosuppressive and immunostimulatory genesets, as well as six published immune tumor subtypes from the TCGA project (https://www.cancer.gov/tcga). A metagene score for each sample was computed by averaging the log10(1+TPM) across the genes in the geneset. Differences in metagene scores, and in individual gene scores, were evaluated using the Wilcoxon signed-rank test. Five immune subtypes (https://www.cancer.gov/tcga) were represented among our samples, and differences between HOXB13-low and HOXB13-high samples were calculated using Fisher’s exact test. Results: Gene-expression analysis of ER+HER2- TCGA breast tumors revealed 610 genes differentially expressed (>2-fold change, padj< 0.01) between HOXB13-high (N=128) and HOXB13-low (N=242) tumors. Comparative GSEA of HOXB13-high versus low HOXB13-low tumors revealed 206 significant C2 genesets (p<0.005, FDRq <0.10) and 11 significant Hallmark genesets (p<0.01, FDRq <0.10). Frequently high-ranking significant C2 genesets were those associated with interferon response pathways (N=13) and estrogen signaling pathways (N=19), while frequently high-ranking significant Hallmark genesets were those associated with interferon response (N=2) and MYC targets (N=2). Comparative analysis of the immune landscape in ER+ HER- TCGA tumors demonstrated a significantly greater prevalence of HOXB13-low tumors within the previously described “Inflammatory”, “lymphocyte-deleted” and “wound-healing” immune subtypes. Analysis of immunomodulator (IM) gene expression revealed significantly higher levels of 52 of 56 IM genes in HOXB13-high versus HOXB13-low tumors. Both immunostimulatory genes (IFNG, CXCL10, CXCL11, GZMB, PRF1, ICOS, and IL15) and immunosuppressive genes (CTLA4, IDO1, FOXP3, CCL8, CD38, IL2RA, LAG3 and TIGIT) were differentially expressed in HOXB13-high tumors. Conclusions: HOXB13 expression in ER+ HER- breast cancer was associated with the interferon response pathway, which plays an important role in the tumor microenvironment. HOXB13-high tumors demonstrated a significantly altered immune expression profile compared with HOXB13-low tumors. Further investigation of the potential role of HOXB13 in modulating the tumor immune microenvironment is warranted. Citation Format: Emily Lachtara, Michael Lawrence, Dennis Sgroi. Gene expression analysis of HOXB13-high and -low tumors reveals a dichotomous immune landscape [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P1-07-05.