ALBERT

Description: Abstract 3650 Background: Mantle cell lymphoma (MCL) is an aggressive lymphoma with poor prognosis. Cyclin D1 (CCND1) gene amplification is a marker of MCL. Glutathione (GSH) S-transferase pi (GSTP1) is a gene on the same amplicon as CCND1. Alternative splicing of CCND1 is mediated by an A/G single nucleotide polymorphism (SNP) in the exon 4 splice donor region. Inhibition of GSTP1 activity increased the sensitivity of MCL cell lines to cisplatin, cytarabine and bortezomib, but not doxorubicin suggesting its importance for drug resistance (Paret et al, 2005, ASH Abst #2806). Other GSH-related proteins were hypothesized to potentially be predictive biomarkers. We retrospectively analyzed DNA samples from MCL patients in a clinical trial with uniform first line chemoimmunotherapy treatment in order to study potential prognostic and predictive molecular markers including: GST M1/T1 gene copy number as well as SNPs for CCND1, GSTP1 and other GSH metabolism family of genes (described in Moyer et al, 2010 CEBP). Methods: DNA was available from 89 MCL patients treated with R-HyperCVAD from 1999 to 2002 at MD Anderson Cancer Center (MDACC). The CCND1 exon 4 (codon 242) SNP was evaluated using pyrosequencing. GST copy numbers were determined as previously described (Moyer et al., 2007 CCR). GSH-related gene SNP testing was performed at the Mayo Clinic Advanced Genomic Technology Center using an Illumina Golden Gate 384 plex panel [290 GSH-related SNPs (30 genes) and 94 additional SNPs in DNA damage repair and other pathways (19 genes] for GSTs and related GSH metabolism genes including GSH synthetase and GSH transporters. Comparison to ten year clinical data with an 8 year median follow-up was used under an IRB approved protocol (Romaguera et al., 2010, Br J Haematol). For overall survival (OS) and failure free survival (FFS), each SNP was first examined by the method of Kaplan and Meier and the log-rank test. To account for multiple testing (of 300+ SNP), we used a beta-uniform mixture model to model the sets of p-values for OS and FFS separately. SNPs that had a false discovery rate (FDR) adjusted p-value of less than 0.2 were evaluated with Cox proportional hazard models to examine the association between SNP and time to event. When there was evidence of non-proportional hazards, we fitted Bayesian parametric accelerated failure regression model. All tests are two-sided. All analyses were done using SAS (v 9.1, Cary, NC) and the R statistical project (http://www.r-project.org/). Results: Fifteen SNPs were significantly associated with FFS or OS in univariable analysis using a FDR q-value cut off of 0.2 (Table). Ten of these SNPs remained significant with FFS or OS after adjustment for age. Prognostic SNPs included GST (GSTO1, GSTO2, GSTA5) and DNA repair (LIG1, ERCC2, RAD52) genes. GST M1/T1 gene copy numbers and SNPs for GSTP1 and CCND1 were not found to be significantly associated with FFS and OS. Conclusion: DNA repair and GSH-associated gene SNPs correlated with OS and FFS in MCL patients treated with R-HyperCVAD at ten years follow-up. This is the largest MCL molecular marker study of which we are aware in a uniformly treated patient group and includes over 300 SNPs including CCND1, GSTP1, and GSH-associated metabolism pathway genes. Additional clinical trials may look at these SNPs in correlative studies. Statistically Significant SNPs Disclosures: No relevant conflicts of interest to declare.