PRAME Expression Is Correlated with Treatment Outcome and Specific Features of the Tumor Microenvironment in Classical Hodgkin Lymphoma

Abstract

Background: The tumor-associated antigen PRAME is over-expressed in several types of cancer and is currently investigated as a therapeutic target for T-cell immunotherapy. Our previous integrative genomic study in diffuse large B-cell lymphoma (DLBCL) identified PRAME deletion to be correlated with patient outcome and an immunologically "cold" tumor microenvironment. However, it remains an open question whether PRAME expression significantly contributes to differential treatment outcomes and tumor microenvironment crosstalk across various B-cell lymphoma subtypes. Material and Methods: We performed an immunohistochemical (IHC) screen in a large cohort of B-cell lymphomas (de novo DLBCL; N=347, follicular lymphoma (FL); N= 166, mantle cell lymphoma (MCL); N= 180), and classical Hodgkin lymphoma (HL); N= 166) to assess PRAME expression as a prognostic biomarker. Moreover, to investigate PRAME-expression associated tumor microenvironment composition and function, we correlated PRAME IHC results with single cell RNA sequencing data of more than 127,000 cells from 22 HL tissue specimens. Results: PRAME IHC analysis revealed frequent PRAME over-expression in HL (115/166, 69%), followed by DLBCL (104/319, 33%), FL (13/166, 8%), and MCL (14/180, 8%). Interestingly, only HL showed a significant treatment outcome correlation, whereas other B-cell lymphoma subtypes did not. Specifically, using a previously published HL cohort (Steidl et al, NEJM 2010) PRAME-negative Hodgkin Reed Sternberg (HRS) cells indicated significantly shorter overall survival (P = 0.008) and disease-specific survival (P = 0.042 ). To characterize PRAME-specific microenvironment composition and function in HL, we analyzed T-, B-, NK-cell, and macrophage subsets in PRAME-positive (17 of 22 cases) vs -negative (5 of 22 cases) tumor samples using single cell RNA sequencing data. From 22 expression-based microenvironment cell clusters that were annotated and assigned to a cell type based on gene expression, all three CD4 helper T-cell clusters were de-enriched in PRAME-negative samples, and the CD4 non-Treg proportion was significantly lower in PRAME-negative samples (P = 0.049). Strikingly, when focusing on phenotypic features of cells within the CD4 non-Treg T-cell cluster, CXCL13 was identified as the most up-regulated gene in PRAME-negative samples. When interrogating published HRS cell transcriptome data (Steidl et al, Blood 2012), immune response pathways including chemokine receptors and chemokine ligands were up-regulated in PRAME-negative HRS cell samples. Of specific interest, CXCR5, the cognate receptor for CXCL13, was significantly upregulated as a member of the chemokine pathway (P = 0.0086) in PRAME-negative HRS cell samples. These results suggest that crosstalk between CXCL13 (produced in the microenvironment) and CXCR5 (expressed on HRS cells) contributes to tumor maintenance in PRAME-negative HL. Finally, to explore potential therapeutic approaches for PRAME-negative HL cells, we focused on 3 HL-derived cell lines (L540, L591, DEV) with low PRAME expression and exposed these lines to DNMT or HDAC inhibitors. DNMT inhibitor treatment showed clear restoration of PRAME expression in a dose dependent manner, but no restoration was found by HDAC inhibitor treatment. To investigate the effect of DNA methylation in transcriptional regulation of PRAME in HL cells, we performed bisulfite sequencing in the PRAME CpG promoter region in PRAME down-regulated (L540, L591, DEV) and up-regulated (HD-LM2, KMH-2, L1236) cell lines and found hypermethylation in PRAME low vs high cell lines. Moreover, the CpG promoter region was significantly demethylated by DNMT inhibitor treatment in cell lines with low PRAME expression. Conclusion: We discovered that PRAME protein expression was correlated with outcome in HL and identified specific T-cell subsets in PRAME-negative patients. PRAME restoration by DNMT inhibitors might represent a new therapeutic avenue in combination with modern immunotherapies, such as PRAME-specific T-cell therapy or PD1 inhibition. Disclosures Scott: Roche/Genentech: Research Funding; Janssen: Consultancy, Research Funding; NanoString: Patents & Royalties: Named inventor on a patent licensed to NanoSting [Institution], Research Funding; Celgene: Consultancy. Steidl:Nanostring: Patents & Royalties: Filed patent on behalf of BC Cancer; Bristol-Myers Squibb: Research Funding; Roche: Consultancy; Seattle Genetics: Consultancy; Bayer: Consultancy; Juno Therapeutics: Consultancy; Tioma: Research Funding.