Prognostic impact of BCL2 and MYC expression and translocation in untreated DLBCL: Results from the phase III GOYA study.

Abstract

Introduction: DLBCL pts with tumours co-expressing BCL2 and MYC (dual-expressor, DE) or with dual gene translocations (double-hit, DH) have poor outcomes but prognostic relationships between cell-of-origin (COO) subtype (ABC vs GCB) and BCL2/MYC are unclear. We report predefined exploratory analyses of the prognostic effects of BCL2 positivity (+), MYC+, DE and DH, in relation to COO, in the Phase III GOYA study (NCT01287741). Method(s): Pts with previously untreated DLBCL were randomised 1:1 to receive obinutuzumab or rituximab plus 6 or 8 cycles of CHOP. Using a Ventana investigational-use IHC assay (BCL2 antibody clone, 124; c-MYC, Y69), pretreatment tumour samples were analysed at a central laboratory. Samples stained within the limit of BCL2 (<=4 months, 30degreeC) and MYC (<=12 months, 30degreeC) antigen stability were included in primary analyses; sensitivity analyses included all samples. Scoring algorithm incorporated % of tumour cells stained and intensity: BCL2 IHC+, moderate/strong in >=50% of tumour cells; MYC IHC+, >=40% of tumour cells. Vysis LSI Dual Color Break Apart FISH Probes identified BCL2 and MYC translocations: FISH+, >=50%. COO classification of RNA extracts used a NanoString Lymphoma Subtyping gene expression assay. Univariate Cox regression analysis of investigator-assessed PFS was performed. Covariates for multivariate analysis were treatment arm (Tx), IPI score, no. of CHOP cycles and COO. (Figure Presented) Results: Baseline characteristics, including IPI score, were similar for biomarker evaluable and ITT populations. Prevalence of BCL2 IHC+, MYC IHC+, DE and DH was 49%, 83%, 42% and 3.6%, respectively. Prevalence by COO: BCL2 IHC+, 75% in ABC and 38% in GCB; MYC IHC+, 95% in ABC and 76% in GCB; DE, 72% in ABC and 29% in GCB; DH, 7% in GCB and 0% in ABC (19/20 DH pts were GCB; 1 unclassified). In univariate analysis, BCL2 IHC+, DE and DH were associated with poorer prognosis (Table). Multivariate analysis confirm the poor prognosis of BCL2 IHC+ pts, independent of Tx, IPI score, no. of CHOP cycles and COO. Context-dependent effects of MYC IHC+ suggest an association with poorer prognosis in BCL2 IHC- pts while BCL2 IHC+ pts drive a suggested prognostic effect in DE pts (Figure). Poor prognosis of DH pts was independent of Tx, IPI score and no. of CHOP cycles, but no. of pts was low. Conclusion(s): Robust identification and analysis of biomarker subgroups confirm the prognostic importance of DH and BCL2 IHC+, and demonstrate that the prognostic effect of BCL2 IHC+ is independent of COO in DLBCL pts in the GOYA study.