Reproducibility of gene expression signatures in diffuse large B-cell lymphoma
Diffuse large B-cell lymphoma (DLBCL) is a clinically, morphologically and biologically heterogeneous disease. Multiple gene expression profiles (GEP) identifying distinct biology associated subgroups have been identified, but with exception of the cell of origin (COO) classifier, none of these signatures have been validated in independent studies.
Here we set out to reproduce the MYC (Carey et al.), consensus clustering GEP (Monti et al.) and immune ratio (Keane et al.) stratifications on 175 RNA samples isolated from formalin-fixed and paraffin-embedded tissue samples of patients treated with rituximab, cyclophosphamide, doxorubicin, vincristine and prednisolone (R-CHOP) in the HOVON-84 trial. Expression levels were digitally quantified for 117 genes and classification algorithms were validated in the original datasets. Results of the HOVON-84 cases were linked to Lymph2Cx classified COO, MYC expression as determined by immunostaining and presence of MYC, BCL2 and BCL6 rearrangements as determined by a combination of fluorescence in situ hybridization and targeted sequencing.
The HOVON-84 patients were divided in 94 (54%) germinal center B-cell (GCB) type, 58 (33%) activated B-cell (ABC) type and 23 (13%) unclassified cases. A high MYC activity score was observed in 77 cases (44%) and was more frequent in ABC as compared to GCB DLBCL (68% vs 32%, P < 0.00001). In GCB DLBCL a high MYC activity was strongly associated with presence of MYC rearrangements (90%, P = 0.0002). The host response (HR) signature was present in 55 (31%) patients, while the B-cell receptor signaling, and oxidative phosphorylation clusters could not be reproduced. Combined the COO, consensus cluster and MYC activity score differentiate 6 gene expression clusters: GCB/MYC-high (12%), GCB/HR (16%), GCB/non-HR (27%), COO-Unclassified (13%), ABC/MYC-high (25%) and ABC/MYC-low (7%).
In conclusion, COO, MYC gene activity and the HR cluster of the consensus clustering signatures could be validated. These three signatures identify distinct subgroups based on different aspects of DLBCL biology, emphasizing that each classifier captures distinct molecular profiles.