Evaluation of gene expression assays to classify large B-cell lymphoma by molecular subtype using formalin-fixed paraffin-embedded tumor biopsies

Abstract

Gene expression profiling (GEP) can classify large B-cell lymphoma (LBCL) into activated B-cell (ABC), dark zone signature (DZsig), and germinal center B-cell (GCB) subtypes, which can inform prognosis and treatment strategies. We evaluated the concordance of Clinical Laboratory Improvement Amendments (CLIA) assays with the previously validated RNA-based DLBCL90 nanoString assay (DLBCL90) and an RNAseq-based approach (gneSeqCOO and gneSeqDZsig) for classification of DZsig status and cell-of-origin (COO) in 95 formalin-fixed paraffin-embedded tumor specimens. The correlation between DLBCL90 COO and gneSeqCOO linear predictor scores (LPS) was R 2 = 0.95 ( P < 0.0001), with an overall classification concordance of 87% (95% confidence interval [CI] 79–93), and 1.3% frank discordance (ABC to GCB or vice versa). The correlation between DLBCL90 DZsig and gneSeqDZsig LPS was R 2 = 0.92 ( P < 0.0001), with class concordance of 85% (95% CI 74–93) and 0% frank class discordance. These data demonstrate high concordance between DLBCL90 and an RNAseq classifier for COO and DZsig providing CLIA options for clinical gene expression classification of LBCLs.