Abstract 3392: Comprehensive molecular profiling of melanoma brain metastases (MBMs) and patient (pt)-matched extracranial metastases (ECMs)

Abstract

Abstract BACKGROUND: MBMs are a common and lethal complication of advanced melanoma. An improved understanding of the molecular features of MBMs could facilitate the development of more effective treatments for patients (pts). Further, identification of differences between MBMs and ECMs would indicate the need for organ-specific treatment strategies. METHODS: Transcriptomic capture and Illumina RNA-sequencing (RNA-seq) were performed on (i) surgically resected formalin-fixed, paraffin-embedded (FFPE) MBMs (98 tumors from 84 pts) and (ii) surgically-resected ECMs from the same pts (54 from 38 pts). The EdgeR/limma/voom pipeline was used to perform differential gene expression (DGE) analyses. Pathway analyses were performed via Ensemble Gene Set Enrichment Analysis (EGSEA). The ESTIMATE and MCPCounter R packages were used to assess immune infiltrates from voom-transformed counts. Illumina whole exome sequencing (WES) was performed on 27 matched pairs of MBMs and ECMs from 17 pts. MuTect algorithm was utilized to assess non-synonymous, stop-gain, and stop-loss mutations. DNA methylation profiling was performed on 16 pt-matched pairs of MBMs and ECMs by the Illumina MethylationEPIC BeadChip platform; data was analyzed via the ChAMP pipeline. Quantitative analysis of CD3 and CD8 immune markers was performed by immunohistochemistry (IHC) on 31 MBMs and 49 pt-matched ECMs. RESULTS: Overall patterns of mRNA expression, mutational burden, and gene methylation were largely similar between pt-matched pairs of MBMs and ECMs, as unsupervised hierarchical clustering was driven primarily by patient identification. However, EGSEA of RNA-seq data identified significant (FDR q-value < 0.10) differences in immune networks (decreased in MBMs) and neuronal differentiation factors (increased in MBMs) in the pt-matched tumors. MCPCounter analysis revealed significant (p < 0.05) depletion of all classes of immune cells except neutrophils in MBMs compared to ECMs. IHC confirmed decreased TCD3 (p = 0.005) and TCD8 (p = 0.013) infiltrating cells in MBMs. ESTIMATE ImmuneScores among MBMs inversely correlated with expression of VSIG4, a relative of the B7 family believed to inhibit T-cell proliferation, and were significantly (p = 0.0088) increased in MBMs previously treated with radiation (XRT). Increased ImmuneScores also associated with significantly (p = 0.00002) improved overall survival (OS) from surgery for MBM. Significant correlations in mean promoter region methylation changes and RNA-seq log fold-changes were identified in 5/10 (50%) of the neuronal factors overexpressed in MBMs, suggesting an epigenetic mechanism for their differential expression. CONCLUSIONS: Significant differences in immune and neuronal gene networks were detected in MBMs compared to patient-matched ECMs, and expression of immune genes in MBMs positively correlated with previous XRT and OS. Citation Format: Grant M. Fischer, Ali Jalali, Aron Joon, Michael T. Tetzlaff, Alexander J. Lazar, Fernando Carapeto, Mariana P. Macedo, Courtney W. Hudgens, Jennifer L. McQuade, Khalida Wani, Brandy Conner, Bhavana Singh, Michael A. Davies. Comprehensive molecular profiling of melanoma brain metastases (MBMs) and patient (pt)-matched extracranial metastases (ECMs) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3392.