Immunohistochemical Prognostic Marker for Diffuse Large B Cell Lymphoma in Patients Treated with CHOP Like Chemotherapy: Validation of the Algorithm by Hans Et Al. and Analysis of Individual Markers

Background) Recently, many predictive models of the clinical outcome of diffuse large B cell lymphoma (DLBCL) have been proposed for better stratification of subgroups. However, studies of the protein expression have reported conflicting results, predominantly due to the lack of universally accepted cut-off value for positivity. Here we retrospectively analyzed the expression of CD10, BCL6, MUM1, BCL2, and CD5 in Korean DLBCL patients treated with CHOP like regimen using the widely used algorithm by Hans et al. (Blood 2004) and compared the results with different cut-off values. We also investigated the clinical significance of individual biomarkers.

Methods) Of the DLBCL cases retrieved from the file of Pathology, Asan medical center (1999–2004). Ninety four (94) cases with adequate tissue were included. Clinical characteristics, treatment and outcome of these patients were collected from the AMC data base. Immunohistochemical stains were performed on tissue microarray (TMA) sections of the archival material. According to Hans’ algorithm, cases were divided into germinal center B-cell like (GCB) (either CD10⁺ or BCL6⁺MUM1⁻)(19.1%) and activated B cell-like (ABC) (CD10⁻BCL6⁻ or BCL6⁺MUM1⁺), with a universal cut-off value of 30% or more of the positively-stained tumor cells. The percentage of the positivity was also recorded in each case. The four cutpoints (0%, 1–30%, 31–50%, 51–75%, and 76–100%) were investigated for each marker for best correlation with the clinical outcome.

Results) During the 5-year median follow up, median overall survival was not reached. Hans’ algorithm divided the cases into 18 GCB (19.1%) and 76 ABC tumors (80.9%), with no clinical significance (3-year PFS: GCB group 63%, ABC group 62%). Among individual biomarkers, CD5⁻ group (cut point: any positivity), bcl-6⁺ group (cut point: any positivity) and CD10⁺ group (cut point: 30%) had longer PFS, and bcl-6⁺ group (cut point: any positivity) had longer OS in univariate analysis. Multivariate Cox regression analysis showed CD10 (p=0.003) as independent prognostic marker in PFS; and bcl-6 (p=0.03) as independent prognostic marker in OS.

Conclusion) The present study shows the lack of predictability of Hans’ algorithm in DLBCL patients treated with CHOP like regimen among Koreans, while CD10, Bcl-6, and CD5 had the prognostic significance using different cut-off values. The result suggests that the proposed cut-off value may not be applied universally for best correlation, and that the optimal cut-off value may need to be optimized for individual laboratory.