Diffuse Large B-cell Lymphoma Dataset Research Articles

Causal relationships of gut microbiota, plasma metabolites, and metabolite ratios with diffuse large B-cell lymphoma: a Mendelian randomization study.

Recent studies have revealed changes in microbiota constitution and metabolites associated with tumor progression, however, no causal relation between microbiota or metabolites and diffuse large B-cell lymphoma (DLBCL) has yet been reported. We download a microbiota dataset from the MiBioGen study, a metabolites dataset from the Canadian Longitudinal Study on Aging (CLSA) study, and a DLBCL dataset from Integrative Epidemiology Unit Open genome-wide association study (GWAS) project. Mendelian randomization (MR) analysis was conducted using the R packages, TwoSampleMR and MR-PRESSO. Five MR methods were used: MR-Egger, inverse variance weighting (IVW), weighted median, simple mode, and weighted mode. Reverse MR analyses were also conducted to explore the causal effects of DLBCL on the microbiome, metabolites, and metabolite ratios. Pleiotropy was evaluated by MR Egger regression and MR-PRESSO global analyses, heterogeneity was assessed by Cochran's Q-test, and stability analyzed using the leave-one-out method. 119 microorganisms, 1,091 plasma metabolite, and 309 metabolite ratios were analyzed. According to IVW analysis, five microorganisms were associated with risk of DLBCL. The genera Terrisporobacter (OR: 3.431, p = 0.049) andgenera Oscillibacter (OR: 2.406, p = 0.029) were associated with higher risk of DLBCL. Further, 27 plasma metabolites were identified as having a significant causal relationships with DLBCL, among which citrate levels had the most significant protective causal effect against DLBCL (p = 0.006), while glycosyl-N-tricosanoyl-sphingadienine levels was related to higher risk of DLBCL (p = 0.003). In addition, we identified 19 metabolite ratios with significant causal relationships to DLBCL, of which taurine/glutamate ratio had the most significant protective causal effect (p = 0.005), while the phosphoethanolamine/choline ratio was related to higher risk of DLBCL (p = 0.009). Reverse MR analysis did not reveal any significant causal influence of DLBCL on the above microbiota, metabolites, and metabolite ratios (p > 0.05). Sensitivity analyses revealed no significant heterogeneity or pleiotropy (p > 0.05). We present the first elucidation of the causal influence of microbiota and metabolites on DLBCL using MR methods, providing novel insights for potential targeting of specific microbiota or metabolites to prevent, assist in diagnosis, and treat DLBCL.

Identifying CD1c as a potential biomarker by the comprehensive exploration of tumor mutational burden and immune infiltration in diffuse large B cell lymphoma.

Tumor mutational burden (TMB) is a valuable prognostic biomarker. This study explored the predictive value of TMB and the potential association between TMB and immune infiltration in diffuse large B-cell lymphoma (DLBCL). We downloaded the gene expression profile, somatic mutation, and clinical data of DLBCL patients from The Cancer Genome Atlas (TCGA) database. We classified the samples into high-and low-TMB groups to identify differentially expressed genes (DEGs). Functional enrichment analyses were performed to determine the biological functions of the DEGs. We utilized the cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT) algorithm to estimate the abundance of 22 immune cells, and the significant difference was determined by the Wilcoxon rank-sum test between the high- and low-TMB group. Hub gene had been screened as the prognostic TMB-related immune biomarker by the combination of the Immunology Database and Analysis Portal (ImmPort) database and the univariate Cox analysis from the Gene Expression Omnibus (GEO) database including six DLBCL datasets. Various database applications such as Tumor Immune Estimation Resource (TIMER), CellMiner, konckTF, and Genotype-Tissue Expression (GTEx) verified the functions of the target gene. Wet assay confirmed the target gene expression at RNA and protein levels in DLBCL tissue and cell samples. Single nucleotide polymorphism (SNP) occurred more frequently than insertion and deletion, and C > T was the most common single nucleotide variant (SNV) in DLBCL. Survival analysis showed that the high-TMB group conferred poor survival outcomes. A total of 62 DEGs were obtained, and 13 TMB-related immune genes were identified. Univariate Cox analysis results illustrated that CD1c mutation was associated with lower TMB and manifested a satisfactory clinical prognosis by analysis of large samples from the GEO database. In addition, infiltration levels of immune cells in the high-TMB group were lower. Using the TIMER database, we systematically analyzed that the expression of CD1c was positively correlated with B cells, neutrophils, and dendritic cells and negatively correlated with CD8+ T cells, CD4+ T cells, and macrophages. Drug sensitivity showed a significant positive correlation between CD1c expression level and clinical drug sensitivity from the CellMiner database. CREB1, AHR, and TOX were used to comprehensively explore the regulation of CD1c-related transcription factors and signaling pathways by the KnockTF database. We searched the GETx database to compare the mRNA expression levels of CD1c between DLBCL and normal tissues, and the results suggested a significant difference between them. Moreover, wet experiments were conducted to verify the high expression of CD1c in DLBCL at the RNA and protein levels. Higher TMB correlated with poor survival outcomes and inhibited the immune infiltrates in DLBCL. Our results suggest that CD1c is a TMB-related prognostic biomarker.

CD5 Gene Signature Identifies Diffuse Large B-Cell Lymphomas Sensitive to Bruton's Tyrosine Kinase Inhibition.

A genetic classifier termed LymphGen accurately identifies diffuse large B-cell lymphoma (DLBCL) subtypes vulnerable to Bruton's tyrosine kinase inhibitors (BTKis), but is challenging to implement in the clinic and fails to capture all DLBCLs that benefit from BTKi-based therapy. Here, we developed a novel CD5 gene expression signature as a biomarker of response to BTKi-based therapy in DLBCL. CD5 immunohistochemistry (IHC) was performed on 404 DLBCLs to identify CD5 IHC+ and CD5 IHC- cases, which were subsequently characterized at the molecular level through mutational and transcriptional analyses. A 60-gene CD5 gene expression signature (CD5sig) was constructed using genes differentially expressed between CD5 IHC+ and CD5 IHC- non-germinal center B-cell-like (non-GCB DLBCL) DLBCLs. This CD5sig was applied to external DLBCL data sets, including pretreatment biopsies from patients enrolled in the PHOENIX study (n = 584) to define the extent to which the CD5sig could identify non-GCB DLBCLs that benefited from the addition of ibrutinib to rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP). CD5 expression was observed in 12% of non-GCB DLBCLs. CD5+ DLBCLs displayed transcriptional features of B-cell receptor (BCR) activation and were enriched for BCR-activating mutations known to correlate with BTKi sensitivity. However, most CD5+ DLBCLs lacked canonical BCR-activating mutations or were LymphGen-unclassifiable (LymphGen-Other). The CD5sig recapitulated these findings in multiple independent data sets, indicating its utility in identifying DLBCLs with genetic and nongenetic bases for BCR dependence. Supporting this notion, CD5sig+ DLBCLs derived a selective survival advantage from the addition of ibrutinib to R-CHOP in the PHOENIX study, independent of LymphGen classification. CD5sig is a useful biomarker to identify DLBCLs vulnerable to BTKi-based therapies and complements current biomarker approaches by identifying DLBCLs with genetic and nongenetic bases for BTKi sensitivity.

Developing Deep Learning Pipeline of Whole-Slide Images for Enhanced Diffuse Large B Cell Lymphoma (DLBCL) Subtyping and Outcome Prediction: Leveraging Self-Attention Transformer for Training and Inference

Background: Deep learning algorithms can help to analyze whole-slide images (WSI) in lymphoma pathology, identifying deeper features and patterns that may not be easily discernible to human observers. This pilot project is focused on diffuse large B-cell lymphoma (DLBCL), a heterogeneous disease with diverse genetic alterations. By leveraging self-attention trained clusters and transformers, the project aims to identify patterns and associations between mutational status and overall survival, potentially enhancing personalized treatment strategies and improving data reliability. Methods: This study employed a computer vision pipeline learning system to classify DLBCL subtypes using self-discovery of discriminatory features from scanned WSI. The workflow is shown in Figure 1a. First, we segmented tiles or patches from the gigapixel-sized WSI of 223 lymphoma cancer biopsy slides sourced from The Cancer Genome Atlas (TCGA), DLBCL, and Stanford DLBCL-Morph datasets. to optimize extraction of relevant features. For feature extraction, we utilized self-supervised pretraining with a Vision Transformer (ViT) network on a dataset of 1,515,000 patches. These patches were grouped into morphologically similar clusters using K-means, representing various lymphoma proliferation patterns. Dimensionality reduction with UMAP allowed for computational efficiency and feature visualization. Extracted features were used to predict overall survival using Bag-of-Words (BoWs). We further enhanced the model by incorporating geometric information, including nuclear characteristics from HoverNet. We assessed the model's performance using key metrics such as area under the curve (AUC) and accuracy. Additionally, genomic mutations with a frequency of 10% or higher from the TCGA were incorporated for enhanced predictive capability, including PIM1, SGK1, CARD11, KMT2D, SOCS1, BTG1, MUC16 and FAT4. Correlations and hierarchical clustering were performed on mutations, patches, and outcomes (Figure 1b). Results: Using the self-trained ViT encoder as the backbone and random forest, we demonstrate an accuracy of 0.88 on a weakly supervised task with all samples. The performance is significantly improved when compared to ResNet-50 trained on ImageNet. In addition, saliency maps from multi-attention heads provide excellent interpretations of morphological characteristics, including tumor stroma, cell location and necrosis. With the feature embeddings extracted by ViT, 10 morphologically distinct clusters were identified. The algorithms identified morphologically similar and dissimilar clusters among tiles that represent variation in the distribution of lymphoma cells as seen in clustering results. There was a strong correlation between clusters one and four, the mutated genes, and an increased probability of poor outcomes. When this was visualized by the pathologist, cluster 4 showed a high level of necrosis in the WSI. Finally, using the segmented WSI cortex with clustering distribution, our method achieves an AUC of 0.77 for predicting the vital status of alive/dead as an outcome. Conclusions: Our pipeline effectively leverages WSI to employ machine learning and deep learning tools for disease classification and outcome prediction in DLBCL. By computationally analyzing the entire tumor landscape, it captures tumor heterogeneity and disease risk, establishing correlations between patch-level characteristics, genomic mutations and overall outcomes. The Vision Transformer (ViT) plays a pivotal role in this process, successfully identifying specific features in histopathology tissue by leveraging self-attention mechanisms, which enables precise feature extraction and accurate analysis for DLBCL classification. Notably, the ViT model achieves high performance across various tasks without the need for external labeled data, owing to its self-supervised learning capability that enables independent learning and extraction of meaningful information. Based on this discovery cohort from this dataset, we will update the analysis using a larger, independent external cohort from institutional archives, and the results will be presented at the annual meeting.

CD79 Expression Is Associated with Cell-of-Origin and Outcome in Diffuse Large B-Cell Lymphoma

Introduction: CD79B is a target of polatuzumab vedotin, an antibody-drug conjugate, which significantly improved the prognosis of both previously untreated and relapsed/refractory patients with diffuse large B-cell lymphoma (DLBCL). However, the biological and clinical significance of CD79B protein and gene expression have not been fully explored in DLBCL, thus we aimed at examining these relationships. Methods: We retrospectively analyzed de novo DLBCL patients, who were diagnosed and received rituximab-based immunochemotherapy from 2008 through 2018 in the Okayama Hematology Study Group from Japan. Immunohistochemistry (IHC) staining was performed using a CD79B antibody (AT107-2), and protein expression was assessed based on H-score as described in a previous study (Sehn LH et al. JCO 2020), integrating with publicly available representative bulk RNA sequencing DLBCL datasets (BCC cohort from Ennishi D et al. JCO 2019 and NCI cohort from Schmitz R et al. NEJM 2018). We also performed CD8 and MHC class-I IHC to evaluate the tumor microenvironment. Gene expression profile-based cell-of-origin (COO) classification was performed including double-hit signature (DHITsig), recently renamed the dark zone signature (DZsig), using the NanoString DLBCL90 assay. In addition, simultaneous epitope and transcriptome measurement in single cells from lymphoid tissues was conducted. Results: CD79B IHC was evaluable in 576 cases. DLBCL90 assay classified the entire cohort into 293 ABC (50.9 %), 189 GCB (32.8 %), 31 DZsig-positive (5.4 %) and 63 unclassified (10.9 %). Furthermore, we dichotomized the cohort into 288 CD79B high cases and 288 CD79B low cases according to the median CD79B H-score. A dynamic range of CD79B protein expression was observed across COO, where ABC-DLBCL showed the lowest values followed by GCB-DLBCL and DZsig-positive-DLBCL, in ascending order (Kruskal-Wallis test, P < .00001; Figure A). Indeed, CD79B low cases were significantly enriched in ABC-DLBCL (58 %) compared to GCB-DLBCL (26 %) and DZsig-positive-DLBCL (2 %), respectively (Chi-squared test, P < .001). Consistently, we revealed that CD79B expression was the lowest in ABC-DLBCL compared to GCB-DLBCL and DZsig-positive-DLBCL at the transcriptomic level (Kruskal-Wallis test, P = .011 for BCC cohort and P = .022 for NCI cohort). In addition, we identified different CD79B staining patterns, composed of 433 with cytoplasmic pattern (75 %), 86 with membranous pattern (14.9 %), and 52 cases being IHC negative. These patterns significantly varied across COO (Fisher's exact test, P = .015) and CD79B H-score was the highest in the membranous pattern followed by the cytoplasmic pattern (Kruskal-Wallis test, P < .0001). Of note, the composition of CD8 positive T-cells in CD79B low tumors was significantly higher than that of CD79B high tumors (Wilcoxon rank sum test, P < .0001). However, MHC class-I expression was decreased in CD79B low cases compared to CD79B high (Wilcoxon rank sum test, P = .0002), suggesting an immune escape mechanism with downregulation of MHC class-I in the presence of cytotoxic T-cells, which is often seen in solid cancers. Significant association of CD79B expression with COO further prompted us to evaluate CD79B expression in normal germinal center B cells. Notably, the single-cell simultaneous epitope and transcriptome analysis (CITEseq, n = 2) and single-cell RNAseq analysis (n = 6) of reactive lymph nodes revealed that both CD79B gene and protein expression were the lowest in cells exhibiting plasmablastic signatures, followed by light zone and dark zone B cells (Kruskal-Wallis test, P < .0001; Figure B), supporting the relation of CD79B expression to COO subtype in DLBCL. Regarding prognostic impact, CD79B low group had significantly poorer prognosis in the entire DLBCL cohort (Log-rank test, P = .0005 for overall survival (OS) and P = .008 for progression-free survival (PFS)) and in ABC-DLBCL (Log-rank test, P = .003 for OS and P = .031 for PFS). Moreover, CD79B protein expression was significantly associated with OS after adjusting for International Prognostic Index in the entire DLBCL cohort (Cox regression model; P = .035). Conclusion: Our study identifies distinct CD79B expression patterns across COO subtypes, with CD79B low cases enriched in ABC-DLBCL and demonstrating poorer prognosis, suggesting its potential as a prognostic marker and for targeted therapies.

Optimizing Circulating Tumor DNA Limits of Detection for DLBCL during First Line Therapy

Background: Circulating tumor DNA (ctDNA) assessment is effective in diffuse large B-cell lymphoma (DLBCL) monitoring and risk stratification, with prognostic utility throughout first-line (1L) therapy. DLBCL ctDNA assays vary in analytical sensitivity, or limit of detection (LOD), which range from parts per thousand to below 1 part per million (1 in 10 6). With differing assays and time-points assessed in DLBCL studies, the relationship between analytical and clinical sensitivity for outcome prediction remains unclear. We assessed the prognostic ability of ctDNA at various LODs and used modeling strategies to project the efficacy of assays for minimal residual disease (MRD) detection. Methods: We previously reported a DLBCL dataset assessing plasma ctDNA before, during and after curative-intent 1L anthracycline-based treatment from multiple prospective studies, including RCHOP or EP[O]CH with acalabrutinib, lenalidomide, obinutuzumab, and polatuzumab (Roschewski et al. ASH 2022). In this study, ctDNA was evaluated by Phased Variant Enrichment Detection and Sequencing (PhasED-Seq, Foresight Diagnostics), a ctDNA MRD assay with LOD below 1 in 10 6. Samples from this study and a prospective DLBCL cohort treated with standard 1L therapy at Samsung Medical Center were considered in this analysis. We assessed the prognostic ability of ctDNA detection, considering LODs from 1 in 100 (10 2)to 1 in 10 6 to predict progression free survival (PFS) before, during and after 1L treatment. We assessed the desired sensitivity for ctDNA MRD during treatment by generating patient-specific log-linear models assuming exponential decay. We projected the distribution of variant allele frequencies (VAFs) during therapy from cycles 2 to 5 for patients who progressed to determine the minimal acceptable analytical sensitivity. Results: We included 230 patients consisting of 588 ctDNA samples, with 201 before therapy, 71 at C2D1, 101 at C3D1, 70 at C4D1 and 145 at end of therapy (EOT). Median follow-up was 17.5 months and 62 patients (27%) progressed. To evaluate the impact of LOD at treatment milestones, we applied a threshold for ctDNA positivity ranging from 1 in 10 2 to 1 in 10 6. Increased LOD had no effect on the performance of ctDNA before therapy (Figure A & B). At C2D1, if the LOD was at least 1 in 10 4, there was no difference in MRD prognostic performance. Starting at C3D1, improving the LOD for ctDNA positivity down to 1 in 10 6 showed superior predictive power for PFS at 24 months. The power of ctDNA to predict PFS at 24 months improved later in therapy, with area under the receiver operator curves (AUROCs) for PFS24 of 0.68, 0.73, 0.77, 0.88, and 0.86, at pretreatment, C2D1, C3D1, C4D1, and EOT time-points respectively (Figure A). To further understand the desired LOD for ctDNA MRD, we developed personalized models of ctDNA VAFs. Exponential models fit the data well in 43/44 cases with ≥ 3 MRD-detectable samples through C4D1, with a median correlation of 0.91, confirming their utility for modeling ctDNA. We generated log-linear models for 106 patients with ≥ 2 MRD-detectable samples, and used log-fold change in VAF per cycle (i.e. the slope) to compare patients by progression. Median log-fold change in VAF per cycle was worse for patients who progressed at -1.1 (IQR -1.4, -0.7) compared to those who remained disease-free with median -1.5 (IQR -2.0, -1.1) (p<0.0001). Patients with primary refractory DLBCL on EOT imaging had less robust log-fold change per cycle with median of -0.9 (IQR -1.3, -0.5) compared to those who relapsed after CR with median -1.3 (IQR -1.7, -0.9) (p=0.0004). We used the distribution of slopes to project VAFs for cycles 2 to 5 for patients who progressed to determine the LODs that provide acceptable clinical sensitivity. We found the analytical sensitivity for detecting DLBCL extended lower for each cycle, demonstrating the need for a LOD at least 1 in 10 6 for robust MRD detection at late time-points during 1L treatment. Conclusion: When using an ultrasensitive assay, MRD assessment at later timepoints better predicts PFS than at early timepoints. While the technical LOD does not affect disease burden assessment before therapy, using more sensitive assays during and after therapy improves disease detection and outcome prediction. Utilizing the most sensitive ctDNA MRD assays in 1L DLBCL therapy will maximize the efficacy of MRD-driven therapeutic strategies and MRD as a surrogate endpoint in future trials.

High BECN1 Expression Negatively Correlates with BCL2 Expression and Predicts Better Prognosis in Diffuse Large B-Cell Lymphoma: Role of Autophagy.

Diffuse large B-cell lymphoma (DLBCL) is characterized by high molecular and clinical heterogeneity. Autophagy, a lysosome-driven catabolic process devoted to macromolecular turnover, is fundamental in maintaining normal hematopoietic stem cells and progenitors homeostasis, and its dysregulation plays a critical role in the initiation and progression of hematological malignancies. One main regulator of autophagy is BECLIN-1, which may interact alternatively with either BCL-2, thus allowing apoptosis, or PI3KC3, thus promoting autophagy. The altered expression of BCL2 and BECN1 correlates with lymphoma outcomes, but whether this is associated with dysregulated cross-talk between autophagy and apoptosis remains to be elucidated. Analysis of the TCGA database revealed that BCL2 and BECN1 mRNA expression were inversely correlated in DLBCL patients. In representative DLBCL cell lines exposed to doxorubicin, the cells highly expressing BCL-2 were resistant, while the ones highly expressing BECLIN-1 were sensitive, and this correlated with low and high autophagy flux, respectively. Venetoclax targeting of BCL-2 increased while the spautin-1-mediated inhibition of BECLIN-1-dependent autophagy reversed doxorubicin sensitivity in the former and in the latter, respectively. By interrogating the TCGA DLBCL dataset, we found that BCL2 and BECN1 acted as negative and positive prognostic markers for DLBCL, respectively. The differentially expressed gene analysis in the respective cohorts revealed that BCL2 positively correlated with oncogenic pathways (e.g., glucose transport, HIF1A signaling, JAK-STAT signaling, PI3K-AKT-mTOR pathway) and negatively correlated with autophagy-related transcripts, while BECN1 showed the opposite trend. Notably, patients with high BECN1 expression displayed longer survival. Our data reveal, for the first time, that the modulation of BECLIN-1-dependent autophagy influences the prognosis of DLBCL patients and provide a mechanistic explanation supporting the therapeutic use of drugs that, by stimulating autophagy, can sensitize lymphoma cells to chemotherapy.

SpatialSort: a Bayesian model for clustering and cell population annotation of spatial proteomics data.

Recent advances in spatial proteomics technologies have enabled the profiling of dozens of proteins in thousands of single cells in situ. This has created the opportunity to move beyond quantifying the composition of cell types in tissue, and instead probe the spatial relationships between cells. However, most current methods for clustering data from these assays only consider the expression values of cells and ignore the spatial context. Furthermore, existing approaches do not account for prior information about the expected cell populations in a sample. To address these shortcomings, we developed SpatialSort, a spatially aware Bayesian clustering approach that allows for the incorporation of prior biological knowledge. Our method is able to account for the affinities of cells of different types to neighbour in space, and by incorporating prior information about expected cell populations, it is able to simultaneously improve clustering accuracy and perform automated annotation of clusters. Using synthetic and real data, we show that by using spatial and prior information SpatialSort improves clustering accuracy. We also demonstrate how SpatialSort can perform label transfer between spatial and nonspatial modalities through the analysis of a real world diffuse large B-cell lymphoma dataset. Source code is available on Github at: https://github.com/Roth-Lab/SpatialSort.

17th International Conference on Malignant Lymphoma, Palazzo dei Congressi, Lugano, Switzerland, 13 - 17 June, 2023.

Introduction: CD79B is a target of polatuzumab vedotin, an antibody–drug conjugate, which may improve the prognosis of both previously untreated and relapsed/refractory patients with diffuse large B-cell lymphoma (DLBCL). However, the biological and clinical significance of CD79B protein and gene expression in DLBCL is largely unknown. Methods: We retrospectively analyzed de novo DLBCL patients, who were diagnosed and received rituximab-based immunochemotherapy from 2008 through 2018 in the Okayama Hematology Study Group from Japan. Immunohistochemistry (IHC) staining was performed using a CD79B antibody (AT107-2), and protein expression was assessed based on H-score according to a previous study (Sehn L et al. JCO 2020). We also performed gene expression profile-based cell-of-origin (COO) classification, including double-hit signature (DHITsig) which has been renamed to dark zone signature (DZsig) (Waleed A et al. Blood 2022), using the NanoString DLBCL90 assay. Results: CD79B IHC expression was evaluable in 602 cases. We idefined two groups according to median H-score of CD79B expression: CD79Bhigh and CD79Blow. The COO subtypes were assigned as follows: 308 patients (51%) with activated B-cell-like (ABC)-DLBCL, 196 (33%) with germinal center B-cell-like (GCB)-DLBCL, 32 (5%) with DZsig-pos DLBCL and 66 (11%) with unclassified (UNC). H-score of CD79B was the lowest in patients with ABC-DLBCL followed by GCB-DLBCL and DZsig-pos DLBCL in ascending order (Kruskal–Wallis test P < .00001; Figure A). Indeed, CD79Blow tumors were significantly enriched in ABC-DLBCL (57%) compared to GCB-DLBCL (40%) and DZsig-pos DLBCL (22%), respectively (both, P < .001). Consistently, using publicly available DLBCL datasets (Schmitz et al. NEJM 2018 and Ennishi et al. JCO 2019), we revealed that CD79B gene expression was the lowest in ABC-DLBCL compared to GCB- and DHITsig-DLBCL (Kruskal–Wallis test P = .01; Figure B and C). The association of CD79B expression with COO prompted us to evaluate CD79B expression in normal germinal center B cells. Notably single-cell transcriptomic analyses of six reactive lymphoid tissues from publicly available datasets revealed that the lowest expression of CD79B was found in plasmablasts followed by light zone B cells and dark zone B cells in ascending order (Kruskal–Wallis test P < .0001), supporting the differential expression of CD79Baccording to COO subtype. CD79Blow group had significantly shorter overall survival (OS) in the total DLBCL cohort (log-rank, P < .001) and within ABC-DLBCL (P = .001, Figure D and E). Moreover, CD79B protein expression was significantly associated with OS after adjusting for International Prognostic Index in the total cohort (Cox regression model; P< .001). Keywords: Aggressive B-cell non-Hodgkin lymphoma, Diagnostic and Prognostic Biomarkers Conflicts of interests pertinent to the abstract. K. Sunami Honoraria: Celgene, Sanofi, BMS, Ono, Janssen Research funding: Takeda, AbbVie, GSK, Chugai, Otsuka, MSD, Novartis, Astellas Amgen, Pfizer, Parexel, Kyowa Kirin, Symbio, Agios Y. Hiramatsu Honoraria: Chugai Pharmaceutical Co, Nippon Shinyaku Co, Bristol Myers Squibb, Sanofi K.K. I. Yoshida Honoraria: Kyowa Kirin, Chugai, Eisai, Jannsen, Nippon-shinyaku, Otsuka, Symbio, Takeda, Sumitomo Pharma, Meiji Research funding: Kyowa Kirin, Chugai Y. Maeda Research funding: Chugai, Nippon-shinyaku Other remuneration: Chugai, Eisai, Otsuka, Kyowa Kirin, Takeda D. W. Scott Consultant or advisory role: Abbvie, AstraZeneca, Janssen, Incyte Honoraria: AstraZeneca Research funding: Janssen, Roche/Genentech D. Ennishi Honoraria: Chugai, Eisai, Kyowa Kirin Research funding: Nipponshinyaku, Chugai

SPATIALLY‐RESOLVED TRANSCRIPTOMICS DEFINE CLINICALLY RELEVANT SUBSETS OF MACROPHAGES IN DIFFUSE LARGE B‐CELL LYMPHOMA

Introduction: Tumor-associated macrophages (TAMs) are abundant immune cells in the microenvironment of diffuse large B-cell lymphoma (DLBCL) and are implicated in tumor progression and therapy resistance. Conventional immunohistochemistry-based studies have relied on the oversimplified M1/M2 classification of TAMs which does not fully capture the functional diversity of macrophage biology, thus potentially contributing to their inconsistent prognostic significance in DLBCL. Spatial whole-transcriptomic analysis allows in-depth investigation of specific cell types in different spatial regions within normal and malignant lymphoid tissues, enabling detailed macrophage phenotyping. Here, we employed spatial whole-transcriptomic analysis in the characterization of CD68+ cells of DLBCL and reactive lymphoid tissues (RLTs) to identify novel macrophage subsets with biological and clinical significance. Methods: Digital spatial profiling with whole-transcriptomic analysis of CD68+ cells was performed in 47 DLBCL and 17 RLTs, to define macrophage signatures (termed “MacroSigs”) of distinct lymphoid spatial niches and clinical scenarios. Eight independent DLBCL datasets (4594 patients) with complete transcriptomic and survival information were used for validation of these spatial-derived MacroSigs. Results: Digital spatial profiling revealed previously unrecognized transcriptomic differences between macrophages populating distinct spatial compartments in RLTs (light zone (LZ)/dark zone (DZ), germinal center (GC)/interfollicular (IF) regions), and in between disease states (RLTs and DLBCL with or without relapsed disease). This transcriptomic diversity of macrophages was categorized into eight MacroSigs. Spatial-derived MacroSigs associate with specific cell-of-origin (COO) subtypes of DLBCL, of particular interest being the IF-MacroSig enriched in the unclassified COO (p < 0.005, 6/8 datasets). MacroSigs of relapsed-DLBCL and DZ were prognostic for shorter overall survival in multiple datasets (p < 0.05 in 5/8 datasets; p < 0.05 in 8/8 datasets, respectively). Projection onto a macrophage single-cell RNA-sequencing atlas reveals the Non-relapse-DLBCL MacroSig to depict HES1/FOLR2-like macrophages, while relapse-DLBCL-MacroSig represents IL1B-like monocytes, with unique therapeutic vulnerabilities for each. The research was funded by: Work in ADJ’s laboratory is funded by a core grant from the Cancer Science Institute of Singapore, National University of Singapore through the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiative. CT was supported by the Italian Foundation for Cancer Research (AIRC) Investigator Grant IG ID.22145; 5 × 1000 Grant ID.22759. Keywords: aggressive B-cell non-Hodgkin lymphoma, diagnostic and prognostic biomarkers No conflicts of interests pertinent to the abstract.

Molecular characterization of Richter syndrome identifies de novo diffuse large B-cell lymphomas with poor prognosis

Richter syndrome (RS) is the transformation of chronic lymphocytic leukemia (CLL) into aggressive lymphoma, most commonly diffuse large B-cell lymphoma (DLBCL). We characterize 58 primary human RS samples by genome-wide DNA methylation and whole-transcriptome profiling. Our comprehensive approach determines RS DNA methylation profile and unravels a CLL epigenetic imprint, allowing CLL-RS clonal relationship assessment without the need of the initial CLL tumor DNA. DNA methylation- and transcriptomic-based classifiers were developed, and testing on landmark DLBCL datasets identifies a poor-prognosis, activated B-cell-like DLBCL subset in 111/1772 samples. The classification robustly identifies phenotypes very similar to RS with a specific genomic profile, accounting for 4.3-8.3% of de novo DLBCLs. In this work, RS multi-omics characterization determines oncogenic mechanisms, establishes a surrogate marker for CLL-RS clonal relationship, and provides a clinically relevant classifier for a subset of primary “RS-type DLBCL” with unfavorable prognosis.

Mutational profiling of circulating tumor DNA and clinical characteristics in lymphoma: Based on next generation sequencing.

Liquid biopsy has been experimented with to identify the mutation of lymphoma based on next-generation sequencing (NGS). We applied NGS analysis to circulating tumor DNA (ctDNA) in 20 lymphoma patients. Then, we compared treatment outcomes, and clinical characteristics among these patients, then investigated mutational profiling. Two independent cohorts of 241 patients with mature B cell lymphoma in Mature B-cell malignancies data set (MBN) data set and 50 diffuse large B-cell lymphoma (DLBCL) patients in DLBCL data set, were used to examine the association between gene mutations and prognosis. We found ctDNA positive group had significantly more relapsed/PD (7/12, 58.3%) and less CR/PR patients (1/12, 8.3%) compared to negative group (0, 0%) (5/8, 62.5%) (p < 0.001). Somatic alterations were identified in 12 of 20 patients and the total 11 mutations were: Ataxia telangiectasia mutated (ATM), TP53, BCL2, BTG2, CD28, EP300, IDH2, IRF8, JAK3, NOTCH1, and NRAS. ATM (S2168L) was found in SLL and TLBL for the first time. BTG2 (c.292_293del), CD28 (P119T), IRF8 (E74D) and NOTCH1 (c.4348 G > A) were newly detected in DLBCL, angioimmunoblastic T-cell lymphoma, primary central nervous system lymphoma, and BCL for the first time respectively. We also disclosed an unreported mutation EP300 (c.1058_1059insC) in DLBCL. Our cases implied ctDNA detection consistent with the FISH of tissue samples to some extent, speculating new molecular subtypes of DLBCL, finding some potential drug-resistant mutations, and suggesting disease recurrence. Moreover, in MBN and DLBCL datasets, patients with TP53 mutation had a significantly shorter OS (all p < 0.05) in both circulating free DNA and tumor tissue. The mutations (no SNP) of NOTCH1 (all p < 0.05) significantly contributed to worse OS in the two cohorts.

Pathway importance by graph convolutional network and Shapley additive explanations in gene expression phenotype of diffuse large B-cell lymphoma.

Deep learning techniques have recently been applied to analyze associations between gene expression data and disease phenotypes. However, there are concerns regarding the black box problem: it is difficult to interpret why the prediction results are obtained using deep learning models from model parameters. New methods have been proposed for interpreting deep learning model predictions but have not been applied to genetics. In this study, we demonstrated that applying SHapley Additive exPlanations (SHAP) to a deep learning model using graph convolutions of genetic pathways can provide pathway-level feature importance for classification prediction of diffuse large B-cell lymphoma (DLBCL) gene expression subtypes. Using Kyoto Encyclopedia of Genes and Genomes pathways, a graph convolutional network (GCN) model was implemented to construct graphs with nodes and edges. DLBCL datasets, including microarray gene expression data and clinical information on subtypes (germinal center B-cell-like type and activated B-cell-like type), were retrieved from the Gene Expression Omnibus to evaluate the model. The GCN model showed an accuracy of 0.914, precision of 0.948, recall of 0.868, and F1 score of 0.906 in analysis of the classification performance for the test datasets. The pathways with high feature importance by SHAP included highly enriched pathways in the gene set enrichment analysis. Moreover, a logistic regression model with explanatory variables of genes in pathways with high feature importance showed good performance in predicting DLBCL subtypes. In conclusion, our GCN model for classifying DLBCL subtypes is useful for interpreting important regulatory pathways that contribute to the prediction.

Real-world data on diffuse large B-cell lymphoma in 2010-2019: usability of large data sets of Finnish hospital data lakes.

Background: Real-world data on diffuse large B-cell lymphoma (DLBCL) has remained incomplete. In Finland, health record data originally recorded in different hospital data record systems are collectively available via data lake technology, enabling efficient extraction and analysis of large data sets. The usability of Finnish data lake data in the assessment of DLBCL was evaluated. Methods: Adult DLBCL patients diagnosed between 2010 and 2019, home municipality in the Hospital District of Southwest Finland and data available in respective data lake were included. Results: The algorithmic determination of treatment lines and respective survival was successful. Patient characterization was feasible, albeit partly incomplete because of limited data content/availability and coverage. Stage, International Prognostic Index and cell of origin were available for 63.0, 68.3 and 28.4% of patients, respectively. Genetic aberrations were not structurally available or feasible to extract without a manual chart review. Conclusion: Finnish data lakes represent an efficient way to analyze large DLBCL data sets. The current study provides a tool for developing recording practices in routine care.

SGK1 mutation status can further stratify patients with germinal center B-cell-like diffuse large B-cell lymphoma into different prognostic subgroups.

There are over a 100 driver gene mutations in patients with diffuse large B‐cell lymphoma (DLBCL), but their clinical significance remains unclear. Here, we first analyzed the DLBCL dataset from the UK‐based Haematological Malignancy Research Network. Patients were divided into high‐ and low‐risk groups based on whether lymphoma progressed within 24 months. Genes showing significantly different frequencies between groups were selected. Survival data for patients with the selected mutant genes were analyzed. The results were validated using two other large databases to evaluate the relationship between the selected mutant genes and prognosis. The mutation frequencies of 11 genes (MYD88[L265P], SGK1, MPEG1, TP53, SPEN, NOTCH1, ETV6, TNFRSF14, MGA, CIITA, and PIM1) significantly differed between the high‐ and low‐risk groups. The relationships between these mutant genes and patient survival were analyzed. Patients who harbored SGK1 (serum and glucocorticoid‐inducible kinase 1) mutations exhibited the best prognosis. Most patients with SGK1 mutation are germinal center B‐cell (GCB) subtype. Among patients with GCB DLBCL, those harboring SGK1 mutations exhibited better prognosis than those without SGK1 mutations. Most SGK1 mutations were single‐base substitutions, primarily scattered throughout the catalytic domain‐encoding region. Multiple SGK1 mutations were identified in a single patient. Thus, SGK1 mutations are a marker of good prognosis for DLBCL and occur predominantly in the GCB subtype of DLBCL. SGK1 mutation status can further stratify patients with GCB DLBCL into different prognostic subgroups.

Integrative Genomic Analysis Uncovers Unique Diffuse Large B Cell Lymphoma (DLBCL) Immune Environments and Identifies Associations with Specific Oncogenic Alterations

Introduction: Most patients diagnosed with diffuse large B cell lymphoma (DLBCL) are cured with combination chemoimmunotherapy, but 40% will develop relapsed or refractory (r/r) disease, which is often associated with a poor clinical outcome. PD-1 blockade therapy has been investigated in r/r DLBCL; however, response rates in unselected DLBCL patients are disappointing, highlighting the need for deeper understanding of DLBCL immune landscapes, as well as mechanisms that regulate the immune response to checkpoint blockade therapy (CBT) in this disease.In solid cancers, tumor-cell intrinsic oncogenic signaling strongly influences the immune environment and impacts clinical response to CBT. Despite the recent publication of large-scale genomic datasets in DLBCL, the impact of oncogenic signaling on the immune environment remains to be fully elucidated. In this study, we aimed to characterize immune landscapes associated with DLBCL, as well as the role of lymphoma-intrinsic alterations on shaping the immune environment in this disease.Methods: Using gene set variation analysis (GSVA) in a large cohort of primary DLBCLs (n = ~900), a sample-wise enrichment score was generated for gene sets associated with tumor infiltrating lymphocytes. Gene sets were manually curated to include signatures relating to IFNγ response, T helper cell subsets, CD8 + T cell exhaustion, macrophages, and dendritic cells. A DLBCL cell-of-origin (COO) signature was also included in the GSVA to control for the transcriptional and genomic effects of COO. Samples were hierarchically clustered into related groups. Multispectral immunofluorescence (mIF) for canonical T cell markers was used to confirm GSVA clustering. To mechanistically validate our findings, CRISPR/Cas9 gene editing was used to modulate candidate oncogenes and tumor suppressors genes (TSGs) in the syngeneic A20 murine lymphoma model.Results: GSVA performed on transcriptomes from a large genomic DLBCL dataset revealed four distinct DLBCL immune clusters, termed “ABC hot”, “ABC cold”, “GCB hot” and “GCB cold”, defined by differential expression scores of immune related gene sets (Fig 1A). Concordant with our previous work, DLBCLs with PD-L1 gene amplifications, which are associated with a “T-cell inflamed” tumor microenvironment, were enriched in the “ABC hot” cluster (Fig 1B). Conversely, double hit signature DLBCLs, known to be associated with decreased immune cell infiltration and a GCB COO, were enriched in “GCB cold” DLBCLs (Fig 1C).In an internal cohort of diagnostic DLBCL samples (n = 90) for whom RNA sequencing (RNAseq) and FFPE tissue were available, mIF analysis showed that both “ABC hot” and “GCB hot” DLBCLs had significantly higher ratios of CD8 + T cells to lymphoma cells compared to cold DLBCLs. “ABC hot” DLBCLs also had a significantly higher CD4 + T cell to lymphoma cell ratio (Fig 1D).Importantly, several mutations that correlated with particular DLBCL immune clusters were identified. The “ABC cold” cluster was significantly enriched for loss-of-function (LOF) mutations in TMEM30A and MYD88, whereas LOF mutations in ATM and FOXO1 were commonly observed in “GCB cold” DLBCLs. Finally, LOF mutations in SOCS1 and B2M were significantly enriched in “GCB hot” DLBCLs (Fig 1E, 1F).As LOF SOCS1 mutations were strongly associated with “GCB hot” DLBCLs and are also prevalent in other CBT-sensitive lymphomas, we hypothesized that SOCS1 LOF mutations would enhance lymphoma cell vulnerability to CBT due to increased IFNγ sensitivity resulting from unopposed JAK/STAT activation. To test this hypothesis, we generated Socs1 deficient A20 lymphoma cells. Compared to A20 WT, A20 Socs1-/- cells were characterized by increased pStat1 levels upon IFNγ stimulation (Fig 1G). Interestingly, A20 Socs1-/- tumors showed increased sensitivity to α-PD1 therapy compared to A20 WT in syngeneic hosts. Together, these data suggest that tumor-cell intrinsic JAK/STAT activation via SOCS1 -/- increases lymphoma cell sensitivity to IFNγ and α-PD1 therapy (Fig 1H).Conclusion: We have developed a novel immunogenomic platform to define the role of tumor-cell intrinsic alterations on the immune landscape of DLBCL. Confirmatory studies using in vitro and in vivo models validated the effect of key oncogenes and TSGs on the tumor microenvironment, and suggest these candidate genes may impact response to CBT in DLBCL. [Display omitted] DisclosuresSmith: Alexion, AstraZeneca Rare Disease: Other: Study investigator; Celgene, Genetech, AbbVie: Consultancy. Steidl: Trillium Therapeutics: Research Funding; Curis Inc.: Consultancy; Epizyme: Research Funding; Seattle Genetics: Consultancy; Bayer: Consultancy; AbbVie: Consultancy; Bristol-Myers Squibb: Research Funding. Kline: Seagen: Membership on an entity's Board of Directors or advisory committees; Morphosys: Consultancy, Membership on an entity's Board of Directors or advisory committees; Kite/Gilead: Speakers Bureau; Karyopharm: Membership on an entity's Board of Directors or advisory committees; Merck: Consultancy, Research Funding; Verastem: Research Funding; SecuraBio: Membership on an entity's Board of Directors or advisory committees; Regeneron: Membership on an entity's Board of Directors or advisory committees.

Richter transformation of chronic lymphocytic leukaemia: a British Society for Haematology Good Practice Paper.

Richter transformation of chronic lymphocytic leukaemia: a British Society for Haematology Good Practice Paper.

A graph convolutional neural network for gene expression data analysis with multiple gene networks.

Spectral graph convolutional neural networks (GCN) are proposed to incorporate important information contained in graphs such as gene networks. In a standard spectral GCN, there is only one gene network to describe the relationships among genes. However, for genomic applications, due to condition- or tissue-specific gene function and regulation, multiple gene networks may be available; it is unclear how to apply GCNs to disease classification with multiple networks. Besides, which gene networks may provide more effective prior information for a given learning task is unknown a priori and is not straightforward to discover in many cases. A deep multiple graph convolutional neural network is therefore developed here to meet the challenge. The new approach not only computes a feature of a gene as the weighted average of those of itself and its neighbors through spectral GCNs, but also extracts features from gene-specific expression (or other feature) profiles via a feed-forward neural networks (FNN). We also provide two measures, the importance of a given gene and the relative importance score of each gene network, for the genes' and gene networks' contributions, respectively, to the learning task. To evaluate the new method, we conduct real data analyses using several breast cancer and diffuse large B-cell lymphoma datasets and incorporating multiple gene networks obtained from "GIANT 2.0" Compared with the standard FNN, GCN, and random forest, the new method not only yields high classification accuracy but also prioritizes the most important genes confirmed to be highly associated with cancer, strongly suggesting the usefulness of the new method in incorporating multiple gene networks.

Whole Transcriptome Data Analysis Reveals Prognostic Signature Genes for Overall Survival Prediction in Diffuse Large B Cell Lymphoma.

BackgroundWith the improvement of clinical treatment outcomes in diffuse large B cell lymphoma (DLBCL), the high rate of relapse in DLBCL patients is still an established barrier, as the therapeutic strategy selection based on potential targets remains unsatisfactory. Therefore, there is an urgent need in further exploration of prognostic biomarkers so as to improve the prognosis of DLBCL.MethodsThe univariable and multivariable Cox regression models were employed to screen out gene signatures for DLBCL overall survival (OS) prediction. The differential expression analysis was used to identify representative genes in high-risk and low-risk groups, respectively, where student t test and fold change were implemented. The functional difference between the high-risk and low-risk groups was identified by the gene set enrichment analysis.ResultsWe conducted a systematic data analysis to screen the candidate genes significantly associated with OS of DLBCL in three NCBI Gene Expression Omnibus (GEO) datasets. To construct a prognostic model, five genes (CEBPA, CYP27A1, LST1, MREG, and TARP) were then screened and tested using the multivariable Cox model and the stepwise regression method. Kaplan–Meier curve confirmed the good predictive performance of this five-gene Cox model. Thereafter, the prognostic model and the expression levels of the five genes were validated by means of an independent dataset. High expression levels of these five genes were significantly associated with favorable prognosis in DLBCL, both in training and validation datasets. Additionally, further analysis revealed the independent value and superiority of this prognostic model in risk prediction. Functional enrichment analysis revealed some vital pathways responsible for unfavorable outcome and potential therapeutic targets in DLBCL.ConclusionWe developed a five-gene Cox model for the clinical outcome prediction of DLBCL patients. Meanwhile, potential drug selection using this model can help clinicians to improve the clinical practice for the benefit of patients.

Genomic signatures in pediatric advanced stage Burkitt lymphoma/leukemia in a Chinese population detected by next generation sequencing

IntroductionBurkitt lymphoma/leukemia (BL/BAL) is the most common lymphoma in children, and the sporadic subtype is dominant in Chinese populations. MYC gene translocations are essential for sporadic BL/BAL (sBL/BAL), but other gene mutations also play important roles in sBL/BAL.Material and methodsClinical data of ten Chinese children with sBL/BAL were collected, next generation sequencing of tumor tissues was carried out. Cases of BL and diffuse large B cell lymphoma (DLBCL) were collected from a database, bioinformatics analysis was conductedResultsNine boys and one girl were enrolled, the average age at diagnosis was 100.10±13.39m; MYC rearrangements were confirmed. The patients received combination treatment of chemotherapy and rituximab. All patients achieved complete remission and were alive without relapse. Germline causal gene mutations were detected in four (40%) patients by whole-exome sequencing; ID3, BRCA2, ARID1A and SMARCA4 mutations, in addition to MYC mutations, were the most common somatic mutations. Gene functions in etiology were different between the BL and DLBCL datasets. The identified mutated genes were enriched and connected by GO or KEGG pathways, it seemed that the PI3K-Akt signaling pathway played important roles in the etiology of sBL/BAL; the EGFR-TKI resistance pathway was also analyzed.ConclusionsThe different gene mutations might be index to identify BL or DLBCL in case of difficult pathologic samples. Molecular hallmark of sBL/BAL is MYC translocation, whereas additional gene mutations also occur and play roles in the progression of the disease. sBL/BAL might be controlled by curbing the PI3K-Akt signaling pathway; the EGFR-TKI treatment might not be helpful in sBL/BAL.