Integrating bulk and single-cell transcriptome to identify novel gene markers for germinal center B cells (GCB) subtype of diffuse large B-cell lymphoma

The gold standard cell-of-origin (COO) classification based on the gene expression profiling (GEP) is robust in diffuse large B cell lymphoma (DLBCL). However, its clinical application remains limited. The qualitative transcriptional characteristic on account of the within-sample relative expression orderings (REO) of genes, which is highly stable against batch effects and variations of specimen quality, presents great potential for clinical use. In this study, we developed a qualitative signature to distinguish the germinal center B cells (GCB) subtype of DLBCL from non-GCB subtypes. The signature consisting of 19 gene pairs (19-GPS), was identified from gene pairs with reversal REO patterns between the GCB and non-GCB subtypes. It was validated across four independent datasets with accuracies of 89.37%, 84.34%, 88.29%, and 92.72%, respectively. Patients classified as the GCB subtype by the 19-GPS had significantly higher overall survival (OS) and progression-free survival (PFS) rates compared to the non-GCB patients. In the validation datasets, the 19-GPS based COO classification demonstrated enhanced performance in prognostic stratification compared to the gold standard COO classification. Three marker genes in 19-GPS, namely HMCES, ZFAND4, and NLRP4, were identified with prognostic value for DLBCL. The expression of NLRP4 in B cells may be related to the transformation of follicular lymphoma (FL), a lymphoma derived from GCB, into DLBCL. All three aforementioned genes may be associated with the responsiveness of malignant human B-cell Lines to R-CHOP therapeutic agents. These results demonstrated that 19-GPS signature was effective and might potentially assist in guiding the prognosis and treatment options for DLBCL.