Background. Extranodal MZL of MALT type (MALT lymphomas) represents 5-8% of all human B-cell lymphomas can occur at any anatomical site. We and others have previously highlighted the genetic, biological and clinical peculiarities of this tumor, differences due in part to the diverse primary sites of disease. Almost by definition, the cellular composition of MALT lymphomas can be very heterogenous, encompassing small B cells resembling MZ cells, monocytoid cells, and small lymphocytes, in addition to scattered immunoblasts, centroblast-like cells and plasma cells. This issue and the possible presence of anatomical site-specific nonmalignant cell biopsies make single cell (sc) RNA-Seq ideal for studying MALT lymphomas. scRNA-Seq has the potential to provide detailed information on the transcriptome of non-neoplastic cells and give insight on the heterogeneity of the lymphoma cell population.

Methods. Cells were analyzed with the 10× Chromium Chip and the Single Cell 3′ Reagent Kit v3. Libraries were run on an Illumina Nextseq 500. A minimum of 50'000 reads per cell were sequenced.

FASTQ files were mapped to human reference genome (HG38) using the cellranger pipeline. Data went through quality control and downstream analysis in R (v3.6.1) using Seurat (v3.0). Cell clusters were generated with the FindNeighbors/FindClusters functions and cells annotated with a mixed approach considering known biomarkers and unsupervised computational methods (celldex, SingleR). Non-linear dimensional reduction (UMAP) was used to visualize clustering results and Monocle 2.0 for pseudo-time analysis.

Results. ScRNA-seq was performed on pulmonary (n=2), gastric and parotid (n=1, each one) MZLs. Publicly available data for a second parotid case were included in the analyses. A total of ~19K cells passed QC with a median of 1,368 detected genes per cell. Each specimen contained mostly B cells (median 76%; range 49-85%), followed by T cells (median 21%; 12-47%), plasma cells (median 2%; 1.6-2.7%), epithelial and myeloid cells (median <=1%). Each sample contained both neoplastic B cells (median percentage 49%; range 8-79%), defined as those bearing a monoclonality for κ (n.=4) or λ (n.=1), and normal B cells (median 26%; 6-41%). The latter mainly comprised naïve and memory B cells.

When we merged the five cases together, neoplastic B cells had highly patient-specific profiles, whereas nonmalignant B cells and plasma cells shared similar gene expression profiles across different cases. CD4+T cells also showed patient specific-signatures, while CD8+ cells co-clustered, indicating a more homogenous phenotype across patients.

Transcripts expressed across MZL cells were enriched for IRF4 and BCL6 targets, genes higher in splenic MZL vs normal spleen, genes expressed in memory B cells, plasma cells, ABC-DLBCL with high STAT3, and in OxPhos DLBCL. Examples of genes expressed in MZL cells vs normal B cells were PAX5 and CXCR4 (pulmonary MZLs), TNFRSF1B and KRAS in parotid MZL, PIK3CD in gastric MZL. SPIBand SOX5, among other genes, showed high expression levels in malignant cells from all individuals compared to the other cells. MZL cells showed high inter- and intra-tumoral heterogeneous expression of potential therapeutic targets (for example, MS4A1/CD20, CD19, BCL2, MCL1).

Pseudo-temporal analysis identified a main trajectory that ordered naïve B cells and healthy memory B cells to plasma cells, while the differentiation toward malignant B cells was inferred as a distinct branch.

Analysis of immune-related ligand-receptor (L-R) pair interactions showed that most of the B cell subclusters provided costimulatory signals to CD4+ or CD8+ T cells via CD22-PTPRC interactions. Another suggestive costimulatory interaction occurred through CD70-CD27, involving malignant B cells, CD4+, CD8+ cells and also Tregs. Regarding coinhibitory pairs, the CD86-CTLA4 axis was frequently identified between CD4+/Tregs subclusters and malignant B cells but not between non-malignant B and T cells.

Conclusions. scRNA-Seq highlighted peculiar inter- and intra-tumoral behaviors of B and T cells in MALT lymphomas. The results indicate that although each anatomical site can have a different molecular portrait, common gene expression programs are found in malignant cell populations of MALT as well as in cells of the tumor microenvironment.

Acknowledgments: supported by Barletta foundation and EASI-Genomics.

Rossi:BeiGene: Consultancy, Honoraria, Other: Travel Support, Research Funding; AstraZeneca: Consultancy, Honoraria, Other: Travel Support, Research Funding; BMS: Consultancy, Honoraria, Other: Travel Support; Janssen: Consultancy, Honoraria, Other: Travel Support, Research Funding; AbbVie: Consultancy, Honoraria, Other: Travel Support, Research Funding; Gilead: Other: honoraria, advisory board fees , Research Funding; MSD: Other: advisory board fees . Zucca:Abbvie: Other: travel grant; Celltrion Healthcare: Other: advisory board fees ; Astra Zeneca: Other: advisory board fees ; Mei Pharma: Other: advisory board fees ; Janssen: Research Funding; Bristol-Myers Squibb: Other: expert statements; Gilead: Other: travel grant, expert statements; MSD: Other: expert statements; Incyte: Other: advisory board fee, Research Funding; Roche: Research Funding; Celgene: Other: advisory board fees, Research Funding; BeiGene: Other: advisory board fee, Research Funding; Miltenyi Biomedicine: Other: advisory board fee. Novak:Bristol Myers Squibb: Research Funding. Bertoni:Spexis AG: Research Funding; Helsinn: Research Funding; Menarini Ricerche: Research Funding; NewG Lab Pharma: Research Funding; Astra Zeneca: Other: travel grant.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution