Response: Outcomes with R-CHOP in DLBCL leave ample room for improvement

In commenting on our paper¹  and the accompanying Inside Blood,²  Dunleavy and colleagues correctly point out that the outcome for patients with Bcl-6–positive diffuse large B-cell lymphoma (DLBCL) treated with cyclophosphamide, doxorubicin, vincristine, and prednisone with rituximab (R-CHOP) still leaves much room for improvement. We agree that the data in Table 5 were misinterpreted, leading Moskowitz to the mistaken conclusion that R-CHOP plus maintenance rituximab (MR) provided a failure-free survival (FFS) of 82%, a result that “will be difficult to improve upon.”^2(p4198)  Rather than a randomized comparison of 4 treatments, this study was a comparison of CHOP and R-CHOP induction, and—for responders only—a comparison of maintenance rituximab and observation (OBS). Table 5 includes only responding patients (ie, a favorable subset) and shows 2-year FFS from the time of second randomization. As reflected in the confidence intervals, there is wide variation, and the numbers are small in the Bcl-6–negative group. While there is no difference in FFS among the 3 rituximab-containing groups, the difference in outcome for the CHOP plus OBS–responding patients according to Bcl-6 expression is nonetheless striking, but will require confirmation in large datasets.

The outcome for patients treated with R-CHOP induction is best seen in Figure 1¹ , based on a weighted analysis removing the effect of MR. As such, these curves should not be labeled with patient numbers and an erratum will be published to remove them. Nonetheless, FFS for the Bcl-6–positive cases leaves ample room for improvement, as suggested by Dunleavy and colleagues. Novel approaches that are based on the biology of the individual disease subsets are required.

Our data underscore the heterogeneity of DLBCL and the need to develop strategies for the different biologic entities that constitute this broad category of non-Hodgkin lymphoma (NHL). We agree that the Bcl-6–positive and Bcl-6–negative groups are molecularly heterogeneous as evidenced by the fact that Bcl-6–positive and –negative cases are represented in both the germinal center B cell–like (GCB) and activated B cell–like (ABC) subsets as shown in Dunleavy et al's communication. The additional complexity of DLBCL is shown by the results of gene expression studies from the Shipp laboratory demonstrating 3 reproducible subsets of DLBCLs, including a cluster termed B-cell receptor/proliferation that includes both cell-of-origin GCB and ABC subsets.³  We agree that these findings argue strongly for the inclusion of gene expression profiling in the context of DLBCL clinical trials. The current Cancer and Leukemia Group B (CALGB) study includes gene expression profiling on all cases to be performed in Dr Staudt's laboratory in the National Cancer Institute (NCI) intramural program. This study will establish the feasibility of doing correlative studies on fresh lymphoma specimens within the cooperative groups. Immunohistochemical studies performed on paraffin-embedded tissue, such as those done in our trial, have their own set of issues, including reproducibility of staining and scoring, but may provide valuable information about biology and prognosis without the requirement for fresh tissue.