Classifying Germinal Center Derived Lymphomas—Navigate a Complex Transcriptional Landscape

Abstract

Classification of lymphoid neoplasms is based mainly on histologic, immunologic, and (rarer) genetic features. It has been supplemented by gene expression profiling (GEP) in the last decade. Despite the considerable success, particularly in associating lymphoma subtypes with specific transcriptional programs and classifier signatures of up- or downregulated genes, competing molecular classifiers were often proposed in the literature by different groups for the same classification tasks to distinguish, e.g., BL versus DLBCL or different DLBCL subtypes. Moreover, rarer sub-entities such as MYC and BCL2 “double hit lymphomas” (DHL), IRF4-rearranged large cell lymphoma (IRF4-LCL), and Burkitt-like lymphomas with 11q aberration pattern (mnBLL-11q) attracted interest while their relatedness regarding the major classes is still unclear in many respects. We explored the transcriptional landscape of 873 lymphomas referring to a wide spectrum of subtypes by applying self-organizing maps (SOM) machine learning. The landscape reveals a continuum of transcriptional states activated in the different subtypes without clear-cut borderlines between them and preventing their unambiguous classification. These states show striking parallels with single cell gene expression of the active germinal center (GC), which is characterized by the cyclic progression of B-cells. The expression patterns along the GC trajectory are discriminative for distinguishing different lymphoma subtypes. We show that the rare subtypes take intermediate positions between BL, DLBCL, and FL as considered by the 5th edition of the WHO classification of haemato-lymphoid tumors in 2022. Classifier gene signatures extracted from these states as modules of coregulated genes are competitive with literature classifiers. They provide functional-defined classifiers with the option of consenting redundant classifiers from the literature. We discuss alternative classification schemes of different granularity and functional impact as possible avenues toward personalization and improved diagnostics of GC-derived lymphomas.

Transcriptomic Maps of Colorectal Liver Metastasis: Machine Learning of Gene Activation Patterns and Epigenetic Trajectories in Support of Precision Medicine

“…Components of the PRC2 such as EZH2 and SUZ12 (located in spot C) can facilitate these transitions and promote CRC stem-like cells (as indicated by the stemness marker LGR5 in spot C, see below) via repressive interactions with their targets (located in the plasticity plateau) in concert with changes to histone and DNA methylation in CIMP genes. Notably, the bimodality between active and repressed chromatin states and their association with proliferative and immunogenic states was described for other tumors such as lymphomas [92] and gliomas [11]. In a more general sense, the interplay between high-expression-genetic and low-expression-epigenetic states, as identified in the CRLM expression landscape, is a key condition for the reorganization of transcriptional programs that facilitate tumor development via the emergence of new attractor states better adapted to the changing TME [8].…”

“…These related to molecular hallmarks of cancer, namely avoiding immune destruction (spot A), tumor-promoting inflammation and angiogenesis (spot E), sustaining proliferation (spot C), deregulating cellular energetics, metabolic activity and DNA repair (spot B), and CIN-related genomic instability (spot D) (Figure 9) [91]. The spot-genes provide robust signatures for classifying the LMS, meeting a trade-off criterion of balancing the interpretability and specificity of molecular markers [92].…”

“…Hence, these personalized portraits can be evaluated by experts in a diagnostic pipeline to assess the specific transcriptional state of a CRLM sample. Such "manual" inspection of molecular portraits could be automated, for example, through neuronal network machine learning, in future stages of technical development (see [92] for a proof of principle application to another cancer type). Moreover, transcriptomic maps can be transformed into prognostic landscapes visualizing the association between gene expression, cellular programs, and their corresponding HR.…”

Subtyping or not subtyping—Quo vadis for precision medicine of colorectal cancer

“…When considering sparse data such as somatic mutations, biological enrichment of the data becomes necessary for better interpretability of the results. In this context also, a trade-off balancing accuracy and stability becomes essential ( 20 ): small numbers of marker genes, e.g., the four ECM-markers extracted by Chai et al ( 15 ), can lead to noisy scorings in practical applications and could be substituted by larger sets of marker genes (also called metagenes) making the scores more robust without significant loss of accuracy. Large stability gains can be reached at the small cost of classification accuracy utilizing metagene approaches.…”

Machine learning extracts marks of thiamine’s role in cold acclimation in the transcriptome of Vitis vinifera

“…SOM portrayal considers the multidimensional nature of gene regulation and pursues a modular view on co-expression, reduces dimensionality and, most importantly, supports visual perception in terms of individual, case-specific expression portraits. The pipeline has been applied to a series of data types and issues, e.g., in the context of molecular oncology ( Binder et al., 2022 ; Loeffler-Wirth et al., 2022 ; Ashekyan et al., 2023 ) and health-related population studies ( Nikoghosyan et al., 2019 ; Schmidt et al., 2020a ), which all have proven the analytic strength of the methods in complex, multi-dimensional omics data. In the context of vine genomics, oposSOM has been applied so-far as “SOMmelier” to microarray SNP data to discover the dissemination history of Vitis vinifera as seen by vine genomes ( Nikoghosyan et al., 2020 ).…”