Prevalence and Prognostic Value of BCL2 and MYC Protein Expression within ABC and GCB Subtypes in Patients with Previously Untreated, Diffuse Large B-Cell Lymphoma: Analysis from the Phase III MAIN Study

Introduction. Diffuse large B-cell lymphoma (DLBCL) is the most common form of non-Hodgkin lymphoma (NHL). Although more than half of patients are cured with standard immunochemotherapy (R-CHOP) in first-line therapy, the disease relapses or is refractory to R-CHOP in ~40% of cases, with limited options for second-line treatment. Molecular characteristics, including Cell-of-Origin (COO), BCL2 expression and concomitant BCL2/MYC expression, contribute to differences in outcome for DLBCL patients (Alizadeh, Nature 2000, Iqbal Clin Cancer Res. 2011, Johnson JCO. 2012). Understanding these molecular risk factors is potentially of value to guide treatment decisions and optimize therapy. Here, we used assays validated for formalin-fixed, paraffin-embedded (FFPE) specimens to retrospectively assess the prevalence and prognostic value of BCL2 and MYC in patients from MAIN, a Phase III trial that evaluated bevacizumab plus R-CHOP in frontline, CD20-positive DLBCL (NCT00486759). This analysis will help assess markers relevant for risk assessment and may guide use of new investigational agents in DLBCL.

Methods. Tissue microarrays (TMAs) from FFPE tumor samples were evaluated using immunohistochemistry (IHC) for BCL2 (clone 124 DAKO), and MYC (clone Y69 Epitomics). FFPE cell pellets from 26 NHL cell lines were also stained as above. BCL2 staining was scored on a 0-3 intensity scale; the BCL2 cutoff was determined by the expression level that conferred sensitivity (<1.0 µM EC50) to the BCL2 inhibitor venetoclax on the NHL cell lines. Samples were coded positive if ≥50% of tumor cells showed a cytoplasmic intensity score of 2+ or 3+. Samples were coded MYC-positive if ≥40% of tumor cells showed any level of MYC nuclear staining above background. Tumors positive for both BCL2 and MYC were classified as “Double-Positive” (DP). Fluorescence in situ hybridization for BCL2 rearrangements used the Vysis LSI BCL2/IGH probe. Gene expression was evaluated using the BioMark RT-qPCR platform (Fluidigm) on FFPE material. Classification of tumor samples into Activated B cell (ABC) or Germinal Center B-cell type (GCB) COO subtypes was determined by adapting the linear predictor score (Wright, PNAS 2003) to the Fluidigm qRT-PCR. Progression-free survival (PFS) was estimated using the Kaplan-Meier method, and hazard ratio (HR) was estimated using Cox models. Outcome between study arms did not differ, thus all patients were grouped for the assessment of prognostic biomarkers.

Results. Baseline tissue samples were available from 230/787 (29%) patients from MAIN, with baseline characteristics and survival similar to the overall ITT population. TMAs were available for IHC evaluation in 184/230 (80%) samples, 88/184 (48%) of which were BCL2 positive. BCL2-positivity rates in various sub-sets were: 58% of IPI high, poor prognostic group vs. 41% of IPI low, good prognostic group, and 64% of ABC vs. 40% of GCB samples (Figure 1).

BCL2 expression appeared to associate with adverse PFS in the GCB subtype (HR adjusted for IPI, 3.89; 95% CI: 1.30–11.68, Figure 2a) but not in the ABC subtype (HR adjusted for IPI: 0.81; 95% CI: 0.3–2.2, Figure 2b).

BCL2 mRNA expression or BCL2 gene rearrangements by FISH showed no association with PFS.

Thirty-two of 174 (18%) samples evaluated for MYC and BCL2 by IHC were DP (95% CI: 13%–25%). PFS was similar in non-DP and DP group (HR adjusted by IPI, 0.77; 95% CI: 0.40–1.48), but the DP group appeared to have inferior OS (HR adjusted by IPI, 0.45; 95% CI: 0.22–0.93). These data should be interpreted cautiously due to the small number of PFS/OS events. 

BCL2 IHC, BCL2 rearrangements and MYC IHC data were available for 40 ABC and 59 GCB samples. DP by IHC was observed in 10/40 (25%) ABC subtype and 10/59 (17%) GCB subtype samples. BCL2 translocations were observed in 6/40 (15%) of ABC, vs. 19/59 (32%) of GCB samples.

Conclusions. Using validated assays developed for FFPE tissue testing, we determined the prevalence and prognostic value of BCL2 in a frontline DLBCL population enrolled in a clinical trial. BCL2 expression appeared to be higher in the poor prognostic groups (high IPI and ABC), and appeared to be an independent prognostic factor in the GCB subtype. BCL2 expression, DP and COO may identify patients with DLBCL who could benefit from BCL2 inhibition; the assays employed here are currently in use to evaluate BCL2 as a predictive biomarker for the BCL2 inhibitor venetoclax in lymphoma trials.