Molecular Classification of Small B-Cell Lymphomas Using Digital Multiplexed Gene Expression Applicable to Formalin-Fixed Paraffin-Embedded Tissues

Background: Small B-cell lymphomas encompass a group of lymphoid neoplasms with high heterogeneity. However, the lack of relatively specific diagnostic markers for most of these diseases make their diagnosis challenging.

Methods: Using NanoString platform, a random forest-based molecular classification model was trained on 203 malignant and 98 non-malignant formalin-fixed paraffin-embedded (FFPE) tissues. Candidate genes were selected from microarray gene expression data of 891 small B-cell lymphomas. All malignant biopsies were derived from individuals diagnosed as a well-defined small B-cell lymphoma entity, including 17 chronic lymphocytic leukemia/small lymphocytic B-cell lymphoma (CLL/SLL), 74 follicular lymphoma (FL), 34 classical mantle cell lymphoma (cMCL), 5 non-nodal mantle cell lymphoma (nnMCL), 56 mucosa-associated lymphoid tissue lymphoma (MALT), 3 splenic marginal zone lymphoma (SMZL), and 14 nodal marginal zone lymphoma (NMZL). A leave-one-out prediction (CLL/SLL, FL, MCL, MZL, and non-malignancy) was applied to test the accuracy of prediction of the model, and Gini Index of each gene was taken into consideration in simplifying the molecular assay applicable to routine test.

Results: A total of 154 candidate genes were initially selected for training the molecular classification model. Genes were initially selected for training the molecular classification model. Thirty-two genes were specifically highly expressed in a single subgroup with Gini Index more than 0.001. The most important genes for classification were CCND1, ZBTB32, MME, BHLHE41, SOX11, and RGS13. The most distinguishable entity was CLL/SLL (ZBTB32, CLNK, and CD200), followed by cMCL/nnMCL (CCND1 and SOX11), and FL (MME, RGS13, and ELL3). Using the molecular classification model trained on the most specific 33 genes, the consistency of diagnosis were 97.5%.

Conclusions: In conclusion, our molecular model robustly recognized different subtypes of small B-cell lymphomas in leukemic FFPE samples, which provides a novel approach to precise diagnosis and classification of small B-cell lymphomas.

Figure

View largeDownload slide

Figure

View largeDownload slide

Close modal