Immunohistochemical Classification of De Novo, Transformed, and Relapsed Diffuse Large B-Cell Lymphoma Into Germinal Center B-Cell and Nongerminal Center B-Cell Subtypes Correlates With Gene Expression Profile and Patient Survival

Abstract

Abstract Context.—Diffuse large B-cell lymphoma (DLBCL) can be assigned to prognostic subgroups, including germinal center B-cell (GCB) and activated B-cell subgroups, by using gene expression profiling and, reportedly, immunohistochemistry for CD10, Bcl-6, and multiple myeloma-1/interferon regulatory factor-4 (MUM1/IRF4). Objective.—To compare 2 commercial MUM1/IRF4 antibody formulations for accuracy in subtyping DLBCL against gene expression profiling, compare subtyping to patient survival, and evaluate the usefulness of GCB and non-GCB subtyping in relapsed and transformed DLBCL. Design.—Evaluation of 2 commercial MUM1/IRF4 antibodies, ICSTAT/M17 and Mum-1p, by using 40 cases of de novo, relapsed, and transformed DLBCL; and comparison of the results obtained with gene expression profiling and survival. Results.—Immunohistochemistry predicted the gene expression profiling subtype 71.8% and 69.2% of the time overall with use of the Mum-1p and ICSTAT/M17 antibodies, respectively, and 100% and 91.7% of the time when MUM1/IRF4 expression determined subtype. Gene expression profiling and immunohistochemistry revealed nearly identical 5-year overall survival rates for the GCB vs non-GCB subtypes (68.0% for GCB vs 24.7% for non-GCB with use of gene expression profiling [P = .03] and 70.2% vs 18.4%, respectively, with use of immunohistochemistry [P < .001]). When de novo, transformed, and relapsed cases were analyzed separately, 5-year overall survival rates were also significantly different. Conclusions.—Immunohistochemistry can be used to subclassify DLBCL, including a very small series of transformed and relapsed cases, into GCB and non-GCB subtypes and predict survival rates similar to those predicted by use of gene expression profiling. The 2 MUM1/IRF4 antibodies performed similarly.