Front-line induction chemotherapy regimens containing cytosine arabinoside (Ara-C) and anthracyclines result in 80% complete remission rate in childhood acute myeloid leukemia (AML) but their cure rate is about 35 – 50%, one of the lowest of all childhood cancers. Understanding the factors that contribute to emergence of chemoresistant leukemic cells are crucial to improving treatment outcomes. We are interested in studying the role of genetic variation in Ara-C transport and biotransformation pathway genes in the efficacy and toxicity of treatment of childhood AML. Human cytidine deaminase (CDA; EC; gene localized to 1p35– 36.2) is a salvage pathway enzyme that irreversibly catalyzes the hydrolytic deamination of Ara-C and Ara-CMP to inactive uracil nucleosides. Several studies have suggested an important role for increased levels of CDA in the development of resistance to Ara-C. Yue and colleagues identified three single nucleotide polymorphisms (SNP) in the CDA gene, one of which (G208A) leads to an alanine to threonine change at codon 70 (A70T) (

Yue et al,
). They showed that the 70T allele had 40% activity with cytidine as substrate and only 32% activity toward Ara-C as the 70A allele. These results suggest that variation in the CDA gene might contribute to Ara-C sensitivity. This variant, while common in Japanese, was not observed by us in a panel of primarily Caucasians of mixed-European ancestry. We sequenced 1.6 Kb of the 5′-upstream region of CDA in 24 laboratory control samples and identified six SNPs (A-92G, C-205G , C-451T, C-897A, A-1138G, G-1172A). TRANSFAC matrix (www.genomatix.de) predicted that three of these, (A-92G, C-451T, and C-897A) would alter consensus transcription factor binding site sequences. We cloned the five naturally occurring haplotypes (ACC, GTC, ATC, ACA and GCC) using the TOPO-TA cloning kit, subcloned into luciferase expression vectors and then transfected Cos-1 cells using the Lipofectamine 2000 protocol. Gene expression was assayed using the Promega Dual-Luciferase Reporter Assay System and read on a Molecular Devices HT Analyzer. Luciferase activity was measured 24 and 48 hours after transfection for six replicates of every condition during two separate transfections. To correct for differences in transfection efficiencies, experimental (Photinus pyralis) luciferase activities were normalized by co-transfection with control (Renilla reniformis) luciferase plasmid. We observed that haplotype combinations ACC, GTC, ACA and GCC drive luciferase expression at approximately 3x that of ATC haplotype (p <0.0001) at 24 hours (2.5x) and this difference persists at 48 hours (3.3x). When reanalyzed as single SNP genotypes, most of the differences in expression were significant, but the magnitude of difference was reduced, suggesting that no single SNP accounts for the expression differences observed at the haplotype level. This leads to the hypothesis that individuals carrying the ATC haplotype (13% of the population; derived from the testing of 94 alleles) would exhibit lower levels of CDA expression and are more sensitive to Ara-C exposure. Experiments are underway to quantify CDA transcription in people of known CDA haplotypes. We also plan to genotype a large cohort of children with de novo AML for these three SNPs and correlate clinical outcomes in individuals carrying the low- versus the high-expressing haplotypes.

Author notes

Corresponding author