BCL-6 and rituximab in diffuse large B-cell lymphoma: where are we?

We read with interest the paper by Winter et al¹  in which they report the impact of BCL-6 protein expression on outcome in diffuse large B-cell lymphoma (DLBCL) patients treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) with rituximab (R-CHOP) or without. These findings provide potentially important insights into DLBCL treatment and rituximab action. Perhaps most revealing is the apparent limitation of rituximab benefit to BCL-6⁻ DLBCL. This finding can mostly account for the 18% to 20% improvement in failure-free survival (FFS) observed in randomized studies of CHOP with or without rituximab when one considers that approximately a quarter of DLBCLs are BCL-6⁻ and rituximab improved FFS by 67% at 2 years in the present study.^2-4 

Our interpretation of these results differs from the accompanying Inside Blood commentary that concludes that the benefit of rituximab in BCL-6⁻ cases was based on “only” 8 patients and that the 82% FFS of R-CHOP at 2 years in BCL-6⁺ DLBCL will be difficult to improve upon. These conclusions are based only on responding patients. When one considers all patients and excludes the effect of maintenance, which confounds FFS, the effect of rituximab in BCL-6⁻ cases is based on 21 patients. Furthermore, the FFS of R-CHOP in BCL-6⁺ cases is approximately 40% at 3 years, a more mature time point given the 3.4-year median follow-up.

These results raise provocative biologic questions. We found that, when examined by microarray, BCL-6 mRNA varies considerably among previously defined DLBCL subgroups (Figure 1).^5,6  If one arbitrarily divides DLBCL in 2 groups based on the median BCL-6 mRNA expression, most (77%) germinal center B cell–like (GCB) DLBCLs are BCL-6⁺. However, BCL-6⁺ cases are also found among activated B cell–like (ABC; 28%), primary mediastinal B-cell lymphoma (PMBL; 38%), and unclassified (47%) DLBCLs. Conversely, BCL-6⁻ cases are most frequent in ABC but also present in other DLBCL subgroups. These considerations suggest that the BCL-6⁺ and BCL-6⁻ groups in the Winter et al study¹  were molecularly heterogeneous and argue for the inclusion of gene expression profiling in the context of DLBCL clinical trials.

The mechanism(s) by which rituximab exerts its effect in DLBCL is unclear. As Winter et al¹  indicate, ABC DLBCL depends on the constitutive activity of the IκB kinase and the NF-κB pathway for survival and may be modulated by rituximab.⁷  In vitro, however, rituximab treatment does not alter the IκB kinase activity of the OCI-Ly3 ABC cell line (L.M.G., unpublished observation, November 2006), unlike reports in Burkitt lines.⁸  Nevertheless, it remains possible that rituximab modulates the NF-κB pathway in vivo and directly induces apoptosis and/or sensitizes such tumors to CHOP. In this regard, we recently reported that rituximab significantly improves the outcome of PMBL, which also displays constitutive NF-κB activity.⁶ 

The present study suggests that BCL-6⁺ DLBCL requires alternative treatment strategies.^9,10  We previously reported that the dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (EPOCH) regimen may be superior to CHOP for DLBCL, suggesting it may overcome additional drug resistance.^11,12  To assess its effect in BCL-6⁺ DLBCL, we analyzed etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (DA-EPOCH-R) in 44 BCL-6⁺ untreated advanced-stage patients and found an 88% FFS at 43-month median follow-up, suggesting high activity (W.H.W., manuscript in preparation). The Cancer and Leukemia Group B (CALGB) is currently conducting a phase 3 randomized trial of R-CHOP versus DA-EPOCH-R in untreated DLBCLs with microarray to determine if DA-EPOCH-R represents a treatment advance and to investigate tumor biology on outcome.