Response: Multiple role of PRDM1β involvement in diffuse large B-cell lymphoma

We would first like to underline that our study did not focus on the assessment of PRDM1β expression as a prognostic marker in diffuse large B-cell lymphomas (DLBCL), but on its significance on the therapeutic response. Tam et al, using quantitative real time–polymerase chain reaction (RT-PCR), found a highly increased PRDM1 expression in U266, which is consistent with our results revealed by a semiquantitative method using the forward and reverse primer specifically located on exon 1β and exon 4 of the PRDM1 gene.¹  Meanwhile, we found that PRDM1 varies according to the type of B-lymphoma cells, some of them expressing PRDM1 at a lower level. This is the important reason why we further performed laser microdissection-based techniques to select only the lymphoma cells for the molecular studies.

At the protein level, using the monoclonal anti-PRDM1 antibody (clone ROS), Tam et al detected a negative or weak PRDM1 expression in DLBCLs. However, Garcia et al²  had previously published their immunohistochemical results using the same antibody, showing that PRDM1 is expressed on the DLBCL lymphoma cells. Therefore, we carried out a cooperative project with Dr Garcia's group, and observed a strong expression of PRDM1 in DLBCL patient samples. A comparison of the results from Dr Garcia group with our results would imply that the same antibody and the same method should be used in the 2 series. If the discrepancy persists, a further study, sequencing the protein involved, would represent an interesting investigation.

By Western blot, as we mentioned in our article, PRDM1β protein was identified as a fragment of approximately 70 kDa in B-lymphoma cell lines and in the DLBCL patient samples we studied. The 80-kDa PRDM1β protein described by Tam et al was identified in the myeloma cell line U266. We believe that DLBCL is a malignant hematologic disease distinct from myeloma, thus protein translation disturbance and/or modification might be involved in DLBCL. We fully agree with Tam et al that “differences in identification and interpretation of the PRDM1β signal in Western blots” could happen, but we would not consider it as a “nonspecific” band, because (1) it is not detected in normal human tonsil, which is known to have no expression of PRDM1β; (2) it can be down-regulated in lymphoma cells through rituximab alone or rituximab combined with doxorubicin; and (3) it varies according to the DLBCL samples.

Moreover, Tam et al also demonstrated in their previous study that inactivating mutations of PRDM1 occurred in 8 of 35 DLBCL patients.³  In our series of 82 patients, despite repeated tests of sequence analysis by our experienced group,^4,5  we did not find these mutations. To interpret this discrepancy, we first have to determine if the methods of analysis were identical. If the methods were identical, we could propose an exchange of the biologic material, to repeat the sequence analyses in the 2 series. When this double technical control is achieved, if the difference persists, we should take into account the different biologic characteristics between Asian and Western populations, possibly due to genetic/environmental background. Indeed, it has recently been reported that the occurrence of the non-GCB subtype of DLBCL was significantly higher in Asian than in Western countries, as defined by immunostainings on paraffin sections using antibodies against CD10, BCL-6, and IRF4.^6,7  Therefore, instead of a single exclusive factor, we propose that inactivating mutations of PRDM1 indicating a tumor-suppressor role and abnormal expression of functionally impaired PRDM1β isoform could both be involved, as well as other members of the PRDM gene family.

Indeed, this scientific exchange is interesting and we are open for active cooperation to investigate PRDM1 expression in DLBCL.