Combining gene expression profiling and machine learning to diagnose B-cell non-Hodgkin lymphoma

Abstract

Non-Hodgkin B-cell lymphomas (B-NHLs) are a highly heterogeneous group of mature B-cell malignancies. Their classification thus requires skillful evaluation by expert hematopathologists, but the risk of error remains higher in these tumors than in many other areas of pathology. To facilitate diagnosis, we have thus developed a gene expression assay able to discriminate the seven most frequent B-cell NHL categories. This assay relies on the combination of ligation-dependent RT-PCR and next-generation sequencing, and addresses the expression of more than 130 genetic markers. It was designed to retrieve the main gene expression signatures of B-NHL cells and their microenvironment. The classification is handled by a random forest algorithm which we trained and validated on a large cohort of more than 400 annotated cases of different histology. Its clinical relevance was verified through its capacity to prevent important misclassification in low grade lymphomas and to retrieve clinically important characteristics in high grade lymphomas including the cell-of-origin signatures and the MYC and BCL2 expression levels. This accurate pan-B-NHL predictor, which allows a systematic evaluation of numerous diagnostic and prognostic markers, could thus be proposed as a complement to conventional histology to guide the management of patients and facilitate their stratification into clinical trials.

Translational Applications of Artificial Intelligence and Machine Learning for Diagnostic Pathology in Lymphoid Neoplasms: A Comprehensive and Evolutive Analysis

“…Gene expression profiling (GEP) of 414 DLBCL patients treated with CHOP/R-CHOP were used as inputs for a SVM model which accurately stratified them in two biologically distinct subgroups [57]. Furthermore, a random forest algorithm was trained and validated to discriminate the most frequent B-cell NHL categories among 510 cases of NHL, based on ligation-dependent RT-PCR and next-generation sequencing (NGS) [16].…”

“…Regarding this, several works have reported the crescent use of AI and ML tools in the diagnosis of hematological diseases [14,15]. Among hematological malignancies, lymphoid neoplasms (LN) constitute one of the most active foci of research in this area, and different AI algorithms have been developed to improve accuracy in lymphoma subtyping [16,17], validation of prognostic models [18], and prediction of chemotherapy response [19,20]. However, a global analysis of the major trends, leading producers, and scientific mapping of AI and ML applications to diagnostic pathology in LN has not yet been undertaken.…”

Histopathologic and Machine Deep Learning Criteria to Predict Lymphoma Transformation in Bone Marrow Biopsies

“…A recent study demonstrates the utility of using machine learning to capture information from gene expression analysis of B-cell lymphomas to aid in the subclassification by hematopathology. 27 Such assays are likely to become diagnostic companion tools and aid in the identification of targetable pathways. Potential hurdles for validating CNN use in a field with abundant subtypes is that large training sets have been shown to work best.…”

A novel diagnostic approach for the classification of small B-cell lymphoid neoplasms based on the NanoString platform

“…Previous studies have demonstrated that the classification of lymphoid neoplasms based on gene expression signatures was feasible and accurate [8][9][10][11] . To date, technologies have been developed to reliably quantify low-throughput gene expression in RNA from either fresh/frozen or formalin-fixed paraffinembedded (FFPE) tissue, allowing the development of clinically relevant RNA assays [9][10][11][12][13][14] .…”