10035 Background: Chemotherapy only cures 60–70% of DLCL patients (pts). While genes in both the GSH and ABC-transporter families have been implicated in drug resistance in other malignancies, their role in DLCL remains controversial. Methods: This study includes 94 DLCL pts with tissue expression data obtained prior to anthracycline-based chemotherapy from 2 oligonucleotide microarray datasets (Affymetrix) with annotated clinical information. Patient-level expression was derived for 10 key genes utilizing probe set matching and levels were adjusted for a common probe set signal. Freedom-from-progression (FFP) was analyzed using an accelerated failure-time regression model, stratified by the international prognostic index (IPI). CART software was used to develop a classification tree for the 2-year disease-free rate. Results: Based on IPI, 48% of pts were low risk, 23% were low-intermediate (int) risk, 24% high-int risk, and 12% were high risk. After therapy, 30 pts (32%) had disease relapse or progression within 2 yrs and 58 pts (62%) were disease-free; 6 (6%) had shorter follow-up. Expression of multi-drug resistance 1 (MDR1; ABCB1) was favorably associated with FFP (β=1.89, p=0.004), with a 100-point change in the expression signal resulting in a near-doubling of the time to failure (time ratio (TR) = 2.2, 95%CI: 1.3–3.6). Expression of GSH S-transferase μ, classes 1/2/4 (GSTM124) was also favorably associated with FFP (β=0.14, p=0.001); a 1000-point signal change resulted in a near-doubling of the time to failure (TR=1.8, 1.3–2.4). Based on our tree, 2 risk groups were identified using IPI and 3 genes: ABCB1, GSTM124, and GSH cysteine ligase modifier (GCLM). Thus, 89% (78%-96%) pts in the low-risk group (N=56) were disease-free at 2 yrs vs. 25% (11%-43%) pts in the high-risk group (p<0.001). Conclusions: We observed a significant favorable association between expression of genes in the GSH and ABC-transporter families, and FFP in pts treated for DLCL. Moreover, we generated 2 risk groups based on the IPI and the expression of 3 key genes, which we plan to validate in an independent dataset. Our findings suggest an additional role for these genes that are classically associated with drug resistance. No significant financial relationships to disclose.