Genomic and Spatial Proteomic Characterization of the Microenvironment of Diffuse Large B-Cell Lymphoma in African American Patients

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common aggressive non-Hodgkin lymphoma in the U.S, with well-established differences in clinical presentation and survival among African American (AA) patients compared to non-Hispanic white patients (e.g., Shenoy et al., Cancer, 2011). However, lack of molecular characterization of AA DLBCL patients represents a disparity; very little is known about the tumor-specific factors that might drive inferior survival for these patients. Here, we used a multi-omics approach to characterize the molecular and spatial features of the lymphoma microenvironment (LME) in AA patients with DLBCL.

Methods

Whole-exome and bulk RNA sequencing (RNA-seq) were performed on samples from 28 patients (10 females, 18 males; age 36–88 years; 27 baseline and 1 relapse) with self-reported AA race. Global genetic ancestry was estimated with PLINK population stratification software. PLINK presented membership in 26 populations belonging to 5 superpopulations, with the sum totaling 1. The superpopulation with the highest weight was chosen as the marker for a specific patient. Somatic mutations were calculated using a supervised machine learning (ML) approach. A custom GRIDSS-based pipeline (Cameron et al., Genome Biol, 2021) was used to detect gene translocations. LME types were predicted as described by Kotlov et al. (Cancer Discov, 2021). Deconvolution by Kassandra was used to define composition of LME cell subsets and biological processes from bulk RNA sequencing data (Zaitsev et al., Cancer Cell, 2022). DLBCL genomic subtypes were determined using LymphGen (Wright et al., Cancer Cell, 2020). For cases with available tissue, multiplex immunofluorescence (mIF) imaging was performed using a 25-marker panel to characterize the LME. A section of a normal lymph node was used as the control. In samples analyzed with mIF imaging, we categorized ~0.5 million cells into 38 subpopulations of B cells, T cells, stromal cells, and macrophages.

Results

PLINK assigned and 25 to African ancestry. Ten patients (36%) presented with the unfavorable immune-inflamed LME, compared with 15–20% observed in prior DLBCL datasets with predominantly white patients (Kotlov et al., Cancer Discov, 2021). Also, 4 (14.2%) had immune-depleted LME, 10 (36%) had mesenchymal LME, and 19 (68%) had the GCB subtype. Genes most frequently mutated were SOCS1 (27%), TNFRSF14 (27%), KMT2D (26%), and TNFRAIP3 (19%). LymphGen did not detect a predominant subtype. Ten patients had translocations of BCL2 (n=7), BCL6 (n=2), or MYC (n=1) with the IGH/IGK/IGL loci as partners with no double hit LBCL identified. Deconvolution based on RNA-seq data indicates B-cell dominance, with 40.1% among all samples (median = 41%).

Among 21 patients analyzed with mIF imaging (n = 38 tissue sections), we identified 11 spatially organized cellular neighborhoods, representative of different B cell subsets existing in varying proportions within the studied cohort. Independent analysis of the imaging-based retrieval of cell compositions revealed Most infiltrating CD4⁺ and CD8⁺ T cells were PD-1^- ICOS^- (median of 76.8% and 82% of respective T cell populations). Macrophages (CD68⁺ cells) constitute a mean of 11.5% (median of 9.8%) of the total cell population. Approximately 5% of macrophages were positive for scavenger receptor CD206 and showed pro-tumor activity. We also detected HLA-DR⁺ CD206⁺ macrophages, which may promote tumorigenesis (Heng et al., J Transl Med, 2023).

Conclusion

In the first-ever report of LME in AA patients with DLBCL, we found an equal dominance of immune-inflamed and mesenchymal LME and a lack of B cell markers, associated with different prognoses (e.g., Kotlov et al., Cancer Discov, 2021). Future studies should utilize larger, multiethnic cohorts to confirm these findings and explore that could specifically improve survival outcomes for AA patients with DLBCL.