Introduction: Diffuse large B-cell lymphoma (DLBL) is a biologically and clinically heterogeneous lymphoma. Gene expression profiling studies using cDNA microarray have led to the development of a new paradigm for the subclassification of DLBL into germinal center B-cell (GCB) and activated B-cell types with a favorable and an unfavorable prognosis, respectively. The expression pattern of CD10, BCL6 and MUM1 by immunohistochemistry (IHC) has been proposed as a surrogate to distinguish GCB from non-GCB types. This immunohistochemical model must be validated in other data sets.

Methods: We identified 57 cases of primary nodal DLBL for which the initial diagnostic biopsy, International Prognostic Index (IPI), and clinical follow-up were available. These patients (pts) received CHOP (n=51) or CHOP-like (n=6) chemotherapy, rituximab (n=12), radiation therapy (n=20), and high dose therapy with stem cell transplantation (SCT)(n=6) as planned primary therapy with curative intent. A tissue microarray (TMA) was constructed and cases were assessed for expression of CD10, BCL6, and MUM1. GCB and non-GCB immunophenotypes (IP) were defined as previously described (Hans et al. Blood 2004, 103:272–85). Clinical endpoints were disease progression and death. Analyses were performed using Cox proportional hazards testing.

Results: The pts consisted of 32 men and 25 women with median age of 60 (range 29–82) years. 26 pts progressed and 25 pts died. Median follow-up of surviving pts was 69.9 (range 8.3–148.1) months. 64%, 82%, and 58% of cases expressed CD10, BCL6, and MUM 1, respectively, and 69% had a GCB IP using the Hans criteria. Univariate analysis showed that high IPI and MUM1 expression predicted a higher risk of progression (P<.001 and P=.027) while IPI, MUM1, and GCB IP predicted a higher risk of death (P<.001, P=.031, and P=.023). BCL6 expression was associated with a trend toward a lower risk of progression (P=.10) but not death (P=.38). By multivariate analysis, high IPI and MUM1 expression remained independent predictors of both progression (HR 7.07, 95% CI 2.68–18.63, P<.001 and HR=3.88, 95% CI 1.51–9.95, P=.005, respectively) and death (HR 6.48, 95% CI 2.57–16.34, P<.001 and HR 3.21, 95% CI 1.25–825, P.015, respectively) while BCL6 expression predicted a lower risk of progression only (HR 0.29, 95% CI 0.11–0.74, P=0.01). Two re-analyses excluding pts that received rituximab or SCT as a planned part of primary therapy gave similar results.

Conclusions: We were unable to confirm the significance of GCB vs non-GCB types using CD10, BCL6, and MUM1 immuhistochemical expression patterns as independent predictors of either progression or death in this cohort of nodal DLBL. However, MUM1 expression (in addition to high IPI) was an independent predictor of a higher risk of both disease progression and death. BCL6 was an independent predictor of a lower risk of disease progression. BCL6 and MUM1 assessment using IHC in nodal DLBL provides useful prognostic information and simplifies pathological risk stratification of nodal DLBL from 3 to 2 markers. These data require confirmation in a larger, independent cohort of nodal DLBL pts treated with a rituximab-containing chemotherapy regimen.

Author notes

Corresponding author

Sign in via your Institution