|Home | About | Journals | Submit | Contact Us | Français|
The mainstay of acute myeloid leukemia chemotherapy is the nucleoside analog cytarabine (ara-C). Numerous studies suggest that the intracellular concentrations of the ara-C active metabolite, ara-CTP, vary widely among patients and, in turn, are associated with variability in clinical response to acute myeloid leukemia treatment. Thus, genetic variation in key genes in the ara-C metabolic pathway – specifically, deoxycytidine kinase (a rate-limiting activating enzyme), 5´ nucleotidase, cytidine deaminase and deoxycytidylate deaminase (all three are inactivating enzymes), human equilibrative nucleoside transporter (ara-C uptake transporter) and ribonucleotide reductase (RRM1 and RRM2 – enzymes regulating intracellular deoxycytidine triphosphate pools) – form the molecular basis of the interpatient variability observed in intracellular ara-CTP concentrations and response to ara-C. Understanding genetic variants in the key candidate genes involved in the metabolic activation of ara-C, as well as the pharmacodynamic targets of ara-C, will provide an opportunity to identify patients at an increased risk of adverse reactions or decreased likelihood of response, based upon their genetic profile, which in future could help in dose optimization to reduce drug toxicity without compromising efficacy. The pharmacogenetic studies on ara-C would also be equally applicable to other nucleoside analogs, such as gemcitabine, decitabine, clofarabine and so on, which are metabolized by the same pathway.
Cytarabine (1-β-arabinofuranosylcytosine [ara-C]) a deoxycytidine nucleoside analog, has been the most effective chemotherapeutic agent used in the treatment of acute myeloid leukemia (AML) since the 1960s [1–3]. In addition to AML, ara-C is also used in the treatment of other hematological malignancies, such as acute lymphoblastic leukemia, chronic myelocytic leukemia, erythroleukemia and mantle-cell lymphoma. Despite the fact that front-line chemo-therapy regimens containing ara-C induce complete response in 65–80% of newly diagnosed AML patients, clinical outcome is suboptimal as most of the patients relapse with resistant disease and poor response to subsequent therapies. Thus, development of cellular resistance and inefficient response to first-line chemotherapy remains the greatest impediment to successful AML therapy.
The intracellular penetration of ara-C depends on its plasma concentration [4–6]. Standard low-dose (LD) ara-C (100–200 mg/m2) regimens achieve a steady state plasma concentration of 0.5–1 µM, but high-dose (HD) ara-C (2–3 g/m2) regimens achieve plasma concentrations greater than 10 µM. Although ara-C is a component of virtually all standard induction regimens for AML, the optimal dose remains to be determined. During recent years, increasing use of HD ara-C has significantly improved survival, particularly in the translocation (8:21) and inversion (16) AML subtypes. Unfortunately, 15–30% of patients still show no response to treatment and 30–80% of patients experience relapse. Although the adult AML patients who receive HD ara-C have a higher probability of 4 years relapse-free survival than those who receive LD ara-C, HD ara-C use has been associated with serious side effects, including fatal toxicities .
At ara-C plasma concentrations of 0.5–1 µM (achieved with standard LD ara-C regimens), expression levels of the uptake transporter human equilibrative nucleoside transporter (hENT1/SLC29A1) are crucial for ara-C uptake into the cell, but at higher plasma concentrations the simple diffusion rate of ara-C exceeds that of pump-mediated transport . Inside the cell, ara-C phosphorylation by deoxycytidine kinase (DCK) is the rate-limiting step in its activation (Figure 1). The resulting ara-C monophosphate (ara-CMP) is then further phosphorylated by pyrimidine kinases to the active 5´-triphosphate derivative, ara-cytidine-5´-triphosphate (ara-CTP). Conversely, the enzyme 5´-nucleotidase (NT5C2) can dephosphorylate ara-CMP back to ara-C. Ara-C and ara-CMP can both be converted into the inactive forms, ara-U and ara-UMP, by the action of the enzymes cytidine deaminase (CDA) and deoxycytidylate deaminase (DCTD), respectively. DNA incorporation of ara-CTP in place of deoxycytidine triphosphate (dCTP) results in chain termination, blocking DNA and RNA synthesis and causing leukemic cell death, which, in turn, is associated with therapeutic response of ara-C [9–11]. Thus, the intracellular concentration of ara-CTP is related to the clinical efficacy of ara-C chemotherapy. Biochemical modulation of ara-CTP metabolism by other nucleoside analogs, such as fludarabine and clofarabine, has been shown to be feasible in clinical studies of adult AML  and pediatric acute leukemias . The addition of a nucleoside analog as a biochemical modulator has been shown to increase ara-CTP accumulation in the circulating blast cells in both adult [14,15] and pediatric AML .
In vitro studies have demonstrated that the intracellular concentrations of ara-CTP are higher in ara-C sensitive cells than in resistant cells . Furthermore, leukemic cells from patients with chronic myelogenous leukemia (which is not responsive to ara-C) have only half the ara-CTP levels as compared with leukemic cells from patients with AML (which is responsive to ara-C) . Intracellular ara-CTP levels are also positively correlated with clinical response after HD-ara-C treatment . The primary mechanism underlying ara-C resistance, thus, appears to be the insufficient intracellular levels of the active triphosphate metabolite ara-CTP, which may be owing to:
The intracellular dCTP pools, in turn, are regulated by the enzyme ribonucleotide reductase (RR) (consisting of RR motif subunits RRM1 and RRM2).
Thus, the observed interpatient variation in clinical response and sensitivity to ara-C could be owing, in part, to the interpatient variation in expression and/or activity of proteins that affect ara-C uptake, activation and degradation (Figure 1). SNPs and small insertions or deletions in these genes could result in amino acid changes that could in turn influence the protein and/or activity of the respective genes. Furthermore, the mRNA expression of these candidate genes could be influenced by SNPs in their regulatory regions. Thus, genetic polymorphisms in genes within the metabolic pathway of ara-C could influence their expression and/or activity, which could in turn, influence the intracellular ara-CTP levels and its clinical efficacy. In addition, variation in the expression or activity of pharmacodynamic targets of ara-C could also contribute to the observed variation in response.
Studies addressing these candidate genes are discussed below and are also highlighted in Table 1.
hENT1 (SLC29A1), an uptake transporter of the nucleoside analogs ara-C, gemcitabine, fludarabine and cladribine, is responsible for up to 80% of ara-C influx into human leukemic cells [20,21]. Thus, when given at the standard dose, ara-C intracellular concentration is critically dependent on uptake by hENT1. Considerable interpatient variation in the hENT1 mRNA expression levels has been observed in AML patients; further in vitro cytotoxicity assays using diagnostic samples from AML patients indicated a significant inverse correlation of hENT1 expression with the lethal dose 50 (LC50) values for ara-C and other nucleoside analogs [22,23]. When given ara-C in standard induction therapy, AML patients, whose blasts were negative for hENT1 expression, had a shorter disease-free survival (7.7 months vs 13 months) and an approximately eight-fold higher risk of early relapse than hENT1-positive patients . Greater cellular sensitivity to ara-C of infant acute lymphoblastic leukemia (ALL), which has high incidence of mixed lineage leukemia as compared with ALL cells from older children, was attributed to high hENT1 mRNA expression in leukemic cells from patients with mixed lineage leukemia gene rearranged ALL . Further hENT1 content measured by flow cytometry was also shown to be associated with sensitivity to ara-C in AML and ALL cells [26,27]. In pancreatic cancer patients treated with gemcitabine, high hENT1 expression was associated with longer overall survival, disease-free survival and time to disease progression , as well as with time to relapse in patients with gemcitabine-sensitive tumors . Recently, in a large prospective randomized trial, in a cohort of pancreatic adeno-carcinoma patients, hENT1 protein levels were significant determinants of patients receiving gemcitabine .
hENT1 is localized on chromosome 6q21.1–21.2 and its coding region was sequenced in Caucasians, African–Americans, Asians and other ethnic groups . hENT1 demonstrated exceptionally low mutability as compared with the other transporters. Within the coding region, four synonymous and two nonsynonymous changes (Ile216Thr, Glu391Lys) were identified . Figure 2 depicts a multispecies alignment of the hENT1 protein sequence for the protein region harboring these nonsynonymous changes. As depicted in Figure 3, the amino acids in hENT1 are not as conserved as other proteins are. In vitro, functional assays demonstrated that these polymorphisms had no effect on hENT1-mediated uptake of nucleosides and nucleoside analogs . This indicated that the coding variants might not contribute to the interindividual variation of nucleoside analogs.
As no functional coding SNPs are known to modulate the function of hENT1 and because expression of hENT1 is reported to affect sensitivity to ara-C, transcriptional regulation of hENT1 is the most likely contributor to the observed variability in hENT1 expression. Although the hENT1 promoter has so far not been completely characterized, it is reported to have hypoxia inducible factor 1 (Hif-1) binding sites. Hif-1-mediated repression of hENT1 expression has been observed in hypoxia [32,33]. Thus, any genetic variation in the hENT1 locus that disrupts or creates binding sites for Hif-1 and other transcription factors that are important for hENT1 expression could alter hENT1 expression.
Analysis of the 1.6 kb of proximal hENT1 promoter upstream of the initiation codon, identified SNPs at −1345C>G, −1050G>A and −706G>C. TRANSFAC® analysis (Biobase International, Wolfenbuttel, Germany) predicted −1345C>G to result in the loss of the FTF/LHR-1 binding site, −706G>C to result in the loss of a barbiturate-inducible element and the gain of binding sites for MyT1 zinc finger transcription factor, AhR nuclear translocator and X-box binding protein. Further analysis identified haplotypes associated with hENT1 expression . Analysis of the hENT1 gene in a Japanese population identified 30 novel polymorphisms, including two nonsynonymous Asp59Glu and Ala430Thr changes . The functional significance of these nonsynonymous changes remains to be determined. Figure 3 summarizes the potentially functional SNPs identified in hENT1. The prognostic significance of hENT1 indicates that hENT1 expression or genetic variants that affect hENT1 expression, may provide a new tool for patient–risk stratification.
Deoxycytidine kinase (EC 22.214.171.124) is a pyrimidine salvage enzyme involved in the phosphorylation of deoxycytidine, deoxyadenosine and deoxyguanosine. It is a rate-limiting enzyme that catalyzes the first step in the activation of nucleoside analogs such as ara-C, gemcitabine, cladribine, fludarabine and clofarabine. Reduced or loss of DCK activity in cell lines resistant to ara-C has been demonstrated in multiple studies [36–38]. Transfection of the DCK gene in DCK-resistant cell lines had been demonstrated to restore in vitro sensitivity to ara-C [39,40]. The activity and expression of DCK varies widely in normal and malignant cells and tissues. There is a 50-fold variation in DCK mRNA expression in patient leukemic cells , a 35-fold variation in DCK mRNA in primary AML cells, a 36-fold variation in liver tissue and a 150-fold variation in human liver metastases of colorectal cancer origin . Although a strong correlation between DCK protein levels and activity has been reported, no correlation between mRNA and enzymatic activities measured using multiple substrates (ara-C and cladribine) was observed in blood cells from chronic lymphocytic leukemia patients . DCK mRNA expression is related positively to AML treatment outcome. When treated with ara-C, patients with higher DCK expression demonstrate longer event-free survival than patients with low DCK expression . Furthermore, the pretreatment levels of DCK have also been related to gemcitabine sensitivity in solid tumors of different origins . These studies provide evidence for the involvement of DCK in the activation of ara-C and other nucleosides, thereby making it a key candidate for pharmacogenetic studies.
The DCK gene is localized on chromosome 4q13.3-q21.1 and the coding region is composed of seven exons. The proximal (1.5 kb) DCK promoter and all the coding exons were sequenced in 24 normal and 24 leukemic Chinese patients, five regulatory SNPs (−125G>T, −201G>T, −289T>A, −360C>G and −740G>C) and one synonymous coding SNP were identified. Two regulatory SNPs (−360C>G and −201C>T) were in linkage disequilibrium and were associated with lower DCK mRNA expression, as well as reduced transcriptional activation in reporter assays . Furthermore, the −360G/−201T haplotype was associated with a good clinical response when compared with the −360C/−201C haplotype. The −360C>G change resulted in the creation of activator protein 2 (AP2) sites and the −201C>T change abolished binding sites for Sp1 and AP2. Joerger et al. reported six novel SNPs in DCK: −243G>T, −135G>C, 261G>A, 364C>T(Pro121Ser), 727A>C (Lys242Gln) and intron 6 T>A (predominantley found in Caucasians) . Recently, we performed deep sequencing of the DCK proximal promoter and all seven coding exons in the International HapMap Project panels  with European (Centre d’ Etude du Polymorphisme Humain [CEPH], Paris, France, panel, n = 90) and African (Yoruba people in Ibadan, Nigeria: [YRI] n = 90) ancestry. A total of 64 genetic polymorphisms, including three nonsynonymous coding changes (Ile24Val, Ala119Gly and Pro122Ser) were identified on the DCK locus . All three coding changes were conserved in different species (Figure 2), implying that these codons might have functional significance. Functional studies using recombinant DCK proteins indicated the association of these polymorphisms with reduced DCK activity (using cladribine as a substrate), with the 24Val, 119Gly and 122Ser isoforms demonstrating 85, 66 and 43% activity as compared with the wild-type (WT) isoform. Lymphoblast cell lines harboring the coding changes were also associated with lower DCK activity. Furthermore, DCK 119Gly and 122Ser variant isoforms had significantly lower Km (p < 0.01) and Vmax (p < 0.001) as compared with the DCK WT protein . Both Alal119Gly and Pro 122Ser changes occur in close proximity to ERS motif, which interacts with the substrate within the active site of DCK. Lymphoblast mRNA expression levels of DCK in these samples indicated ethnic differences, with significantly higher expression in African ancestry as compared with European ancestry. Using Haploview, it was observed that SNPs at the DCK locus were in strong linkage disequilibrium, resulting in four block/groups within the CEPH population and nine block/groups of SNPs within the YRI population. Interestingly, SNPs in block 1/group 1 demonstrated significant frequency differences in these ethnic panels (allele frequency of 0.05 and 0.77 in CEPH and YRI, respectively). These differences in allele frequencies could contribute to ethnic differences observed in phenotypes. SNPs from block 1/group 1 were also associated with DCK mRNA expression. For a block 1 representative SNP at 35708C>T (that occurs in LD with 13 intronic and one other 3´UTR SNP) in the 3´UTR of DCK, the 35708T allele was associated with significantly higher DCK mRNA expression in both European and African ancestry panels as compared with allele 35708C . In addition, SNPs at −245G>C, 36113C>T and 29377C>T within the African ancestry panel were associated with DCK mRNA expression levels (Figure 4). An exploratory analysis in AML patients indicated association of the 35708C allele (3´UTR SNP) with lower leukemic blast cell intracellular levels of ara-CTP when compared with patients homozygous for the 35708T allele . Resequencing of DCK in 240 DNA samples from African–American, Caucasian–American, Han Chinese–American and Mexican–American subjects identified 28 polymorphisms in DCK including the three nonsynonymous changes indicated earlier . Functional analysis revealed lower activities for these isoforms (Val24: 70%; Gly119: 42% and Ser122: 66%) when compared with WT. Furthermore, the isoforms with multiple variants namely, Val24/Gly119/Ser122, Gly119/Ser122 and Val24/Gly119, demonstrated 105, 82 and 44% activity as compared with WT . Substrate kinetics studies indicated no change in apparent Km and Vmax values except for Val24/Ser122 isoform. The differences in the results of these two studies might be due to the use of different model systems (recombinant bacterially expressed protein vs mammalian Cos-1 cells) or different substrates used (purine analog cladribine vs pyrimidine analog gemcitabine) to study substrate kinetics. More recently, analysis of the DCK gene in a Japanese population identified 29 polymorphisms including Pro122Ser coding change, and further analysis indicated considerable ethnic differences in DCK SNPs and haplotypes . Figure 4 highlights selected DCK SNPs with potential functional or clinical implications.
The 5´ nucleotidases (NT5C2/cN II; EC 126.96.36.199]) are a group of enzymes with different substrate specificities that catalyze the dephosphorylation of common ribo- and deoxyribo-nucleoside phosphates, and thus maintain balanced nucleotide pools in cells . To date, seven human 5´ nucleotidases have been isolated and characterized. NT5C2 or ‘cytosolic 5´ nucleotidase II’ is ubiquitously expressed in human tissues. NT5C2 is an inosine monophosphate selective nucleotidase and its higher expression in multiple in vivo and in vitro studies, has been associated with the development of resistance to clinically important nucleoside analogs, such as ara-C, cladribine and gemcitabine . NT5C2 activity opposes that of DCK by dephosphorylating ara-CMP, thereby preventing the production of the active triphosphate form, ara-CTP . Galmarini et al. reported a higher expression of NT5C2 to be associated with a lower disease-free and overall survival in adult AML patients undergoing treatment with ara-C [24,53]. In non-small-cell lung cancer (NSCLC) patients treated with gemcitabine, a higher NT5C2 expression was related to lower overall survival . Although NT5C2 is involved in the inactivation of monophosphate forms of nucleoside analogs, thereby contributing to the development of resistance, it may also reflect the proliferation status of the leukemic cell and may act as a marker of disease aggressiveness. Intracellular ara-CTP production as well as ara-C sensitivity has been shown to be significantly associated with the DCK:NT5C2 ratio in the HL-60 leukemic cell line and its ara-C-resistant variants .
There is no published report so far identifying and characterizing the functional significance of genetic polymorphisms in NT5C2. However, data from international HapMap project indicates the presence of approximately 100 polymorphisms, including two synonymous and two nonsynonymous coding polymorphisms (Ala3Thr and Arg136Gln). The functional significance of these coding polymorphisms requires evaluation. Multispecies alignment of NT5C2 SNPs across the coding SNPs is shown in Figure 2.
Cytidine deaminase (EC 188.8.131.52) is the predominant inactivating enzyme in the ara-C metabolic pathway and is involved in irreversibly deaminating ara-C to ara-U. Elevated CDA expression has been associated with resistance to ara-C and other nucleoside analogs [56,57]. In AML patients, higher CDA levels have been associated with disease recurrence and lower CDA levels with longer duration of remission [58,59]. Ara-C resistance has been demonstrated in mammalian cells constitutively overexpressing human CDA . CDA expression and/or activity has also been implicated in sensitivity to gemcitabine in multiple studies . More recently, in a case study, extensive CDA activity has been implicated in a severe toxicity reaction to capecitabine, which is activated to 5-FU by carboxyl easterase, CDA and thymidine phosphorylase .
Deep sequencing of CDA in the Japanese , Caucasians and African–Americans  identified three nonsynonymous coding polymorphisms (A79C/Lys27Gln, G208A/Ala70Thr, T435C). All these coding changes were conserved in different species (Figure 2) implying that these codons might have functional significance. Functional studies using recombinant CDA demonstrated no influence on immunoreactive proteins but reduced activity for gemcitabine for the Gln27 CDA isoform . The apparent Km of the Gln27 isoform was also significantly higher than that of the WT Lys27 isoform in the substrate kinetic study . The A79C/Lys27Gln isoform has been associated with a reduced deamination of ara-C in vitro ; however, no change in CDA activity was observed between yeast transformants expressing CDA 27Lys or 27Gln variants for deamination of cytidine or ara-C. Furthermore, ara-C sensitivity of yeast transformants expressing Gln27 was also not different than the Lys27 isoform . Of the children with AML treated on the Children’s Cancer Group (CCG) 2941 and 2961 protocols, patients homozygous for the CDAGln27 polymorphism (considered the low activity allele) had increased risk of treatment-related mortality with ara-C-based therapy . In contrast to results from the studies indicated above, Lys27Lys was associated with lower CDA gemcitabine deamination activity in samples from lung cancer patients , with lower toxicity, time to progression and overall survival in NSCLC . The inconsistency in the results for the functional significance of the Lys27Gln allele precludes its clinical evaluation.
The 208G>A polymorphism identified in the Japanese population demonstrated significantly lower CDA activity and increased sensitivity to ara-C , but was not present in Caucasians or African–Americans . Within Japanese pancreatic cancer patients receiving gemcitabine, a patient homozygous for 208 G>A SNPs demonstrated a fivefold higher gemcitabine levels and showed severe hematologic and nonhematologic toxicities . Recently, within a larger cohort of Japanese gemcitabine-naive patients, the CDA haplotype with Ala70Thr (208G>A) SNP was associated with pharmacokinetics of gemcitabine , as well as with toxicity to gemcitabine , as well as with toxicity to gemcitabine . However, it should be noted that the three studies above originated from the same group and G208A SNP has been identified within the Japanese ethnic group [63,69,70] with an allele frequency of approximately 0.012 . Screening of 1.6 kb of the proximal promoter of CDA (upstream of exon 1) in Caucasians and African–Americans identified a haplotype defined by three SNPs (−92 A>G, −451C>T and −897C>A), which were associated with high levels of promoter and enzymatic activity . Whether this haplotype is associated with ara-C chemosensitivity or clinical response to ara-C remains to be determined. Recently, analysis in AML patients receiving ara-C identified CDA SNP −451 as an independent prognostic parameter for survival . Figure 5 highlights potentially significant SNPs at the CDA locus along with clinically significant observations.
The enzyme DCTD (EC 184.108.40.206) is involved in the deamination of ara-CMP to ara-UMP. However, its role in the pathogenesis of an ara-C-resistant phenotype is poorly defined. Although the expression of DCTD has not been demonstrated to be associated with ara-C sensitivity or clinical response , some studies have suggested a substantial role for DCTD in the metabolism of ara-CMP in T-lymphoblastic leukemia [75–77]. Recently, the screening of coding regions and the proximal promoter of DCTD in Caucasian and African ethnic groups identified a nonsynonymous SNP (Asn58Asp) with a significant loss of activity in in vitro assays .
Cytidine-5´-triphosphate synthetase ([CTPS] EC 220.127.116.11) catalyzes the conversion of UTP to CTP and is a key enzyme in pyrimidine biosynthesis. Inhibition of CTPS with cyclopentenyl cytosine (CPEC) has been shown to deplete CTP/dCTP pools, as well as increase sensitivity to ara-C in neuroblastoma cells , human T-lymphoblastic cell lines and myeloid leukemia cell lines [79–81]. Low levels of cellular dCTP pools due to CTPS inhibition facilitates ara-C phosphorylation (by reducing feedback inhibition of DCK) as well as incorporation of ara-CTP into DNA. CPEC-mediated inhibition of CTPS increases activation of DCK and the cytotoxity of neuroblastoma cell lines to gemcitabine .
Clustered mutations within the coding region of CTPS have been identified in ara-C-resistant strains of Chinese hamster ovary cells ; however, screening of the CTPS gene in 36 patients with recurrent and resistant leukemia failed to identify similar mutations . Data from the HapMap project indicates at least 24 SNPs within the CTPS locus (two in the 5´UTR, one in 3´UTR and the rest intronic). In addition, the SNP database indicates two synonymous SNPs, Gln167Gln and Val500Val, occurring with the allele frequency of 0.28 and 0.25, respectively. Future research warrants systematic evaluation of CTPS SNPs.
Ribonucleotide reductase catalyzes the reduction of ribonucleotides to their corresponding deoxyribonucleotides, which are the building blocks for DNA replication. Hence RR is a target of several anticancer agents, including nucleoside analogs and hydoxyurea (reviewed by Shao et al. ). The RR holoenzyme consists of the dimerized large and small subunits, RRM1 and RRM2. RR regulates intracellular pools of dCTP, which in turn, have been implicated in the development of ara-C resistance. Several cell lines and AML blasts with high levels of dCTP have been found to be resistant to ara-C [77,85–87]. dCTP regulates ara-C metabolism at three levels:
Nucleoside analogs, such as ara-C, gemcitabine, fludarabine, cladribine and clofarabine, act as inhibitors of RR after intracellular conversion to their dNDP or dNTP forms. Therefore biochemical modulation of ara-C by nucleoside analogs such as fludarabine and cladribine, has been shown to be feasible in adult and pediatric leukemia patients as they stimulate ara-CTP accumulation by the inhibition of RR [12,90–93]. Recently, the use of Triapine® (Vion Pharmaceutical Inc., CT, USA) a new potent RR inhibitor, has been demonstrated to stimulate accumulation of active metabolites of both gemcitabine and ara-C, in NSCLC cell lines . The therapeutic efficacy of RR inhibitors would thus be greatly influenced by any interpatient variability in the expression and/or activity of both RRM1 and RRM2 genes. Patients with low expression and/or activity of RRM1 or RRM2 might not derive much benefit from combination therapies with RR inhibitors as compared with patients with high expression.
RRM1 overexpression has been correlated with gemcitabine resistance in human pancreatic cancer , in NSCLC cell lines [96,97] and in mice . Preliminary analysis of gene expression in leukemic blasts from AML patients treated with ara-C, demonstrated a significant inverse correlation between RRM1 and RRM2 gene expression levels at diagnosis and the 24 h post ara-C treatment blast ara-CTP levels [MITRA ET AL,. UNIVERSITY OF MINNESOTA. UNPUBLISHED DATA].
Chemosensitivity to gemcitabine in 62 human cancer cell lines of various origin have been associated with the 2464G>A SNP (a silent coding change Ala744Ala) in RRM1 . Recently, the RRM1 promoter allelotype for SNPs −37A>C and −524C>T have been demonstrated to influence response rates to gemcitabine based chemotherapy . Although RRM1 2464G>A does not result in an amino acid change, it has been shown to influence mRNA expression of RRM1 . An association of the RRM1 haplotype for 2455A>G (Thr741Thr) and 2464G>A SNPs was linked with a lower frequency of chemotherapy-induced toxicity . However, so far there has been no systematic study to explore the role of genetic variation in RRM1/2 in the therapeutic efficacy of ara-C.
ATP-binding cassette transporter proteins (ABC drug transporters) have been implicated in drug resistance. Recently, it has been demonstrated that ABCC10 (MRP7) confers resistance to cytarabine in HEK293 cells by reducing accumulation of cytarabine  and ABCC11 (MRP8) has been demonstrated to have a differential expression in AML blasts as well as serve as a predictive marker for treatment outcome . Within MRP8-transfected LLC-PK1 cells, reduced accumulation of ara-C as well as its metabolites has been observed . Hence, these two ABC transporters are potential candidates for future pharmacogenetic studies.
In addition, genetic variations in the pharmacodynamic targets of ara-C (such as DNA polymerase α) and in genes of the apoptotic pathway could also contribute to the interpatient variation in clinical response to ara-C.
Cellular sensitivity to gemcitabine and ara-C in lymphoblastoid cells has been analyzed for association with genome-wide gene expression. Genes with the strongest association with IC50 values were involved mainly in cell death, cancer, cell cycle and nucleic acid metabolism . Some important genes include FKBP5, NT5C3, ES2, GCAT, MYBBP1A, TLE4 and ZNF278. Additional functional validation demonstrated downregulation of FKBP5 and NT5C3 to be associated with altered sensitivity to both drugs .
Using lymphoblastoid cell lines that are part of the International HapMap project (European [CEU] or African [YRI] ancestry); Hartford et al. analyzed more than 2 million single SNPs (narrowed down further for SNPs that affected mRNA expression) for association with susceptibility to ara-C . Four SNPs could explain 51% of the variability in sensitivity to ara-C among the CEU and five SNPs could explain 58% of the variation among the YRI. The new candidate genes identified were:
Although both these studies used lymphoblastoid cell lines for screening ara-C (or gemcitabine and ara-C) cytotoxicity and used genome-wide gene-expression arrays to identify the candidate genes associated with ara-C /gemcitabine cytotoxicity, there was almost no overlap between the top genes identified in these two studies. Although there could be a multitude of reasons for this, ranging from different array platforms to different cell lines, in the study by Hartford et al., one of the causes was suggested to be a narrowing of focus by concentrating on SNPs within genes that affect gene expression .
A few studies in the literature have tried to identify candidate genes associated with cellular resistance to ara-C. Two of the published studies have used ALL and/or CML cell lines to study the candidate genes in ara-C metabolic pathways. Takagaki et al. compared gene expression changes between CCRF-CEM (an ara-C sensitive ALL cell line) and K562 (an ara-C resistant CML cell line) and reported a downregulation of chaperone genes in the CCRF-CEM versus upregulation of hemoglobin genes in K562 cells . In the ara-C-resistant K562, upregulation of IGF-1R was observed compared with the corresponding ara-C-sensitive parental cell line . The major caveat in both these studies is that a very restricted set of genes were tested using either an in-house cDNA array or human cancer chip array.
Two additional reports, both from the same group, have used murine leukemia cell lines. They identified neurofibromin (NF1) deficiency as being associated with the development of resistance to ara-C . NF1 is a negative regulator of RAS signaling. However, whether NF1 or other key genes in the mTOR and MAPK pathways are predictive of ara-C sensitivity in human AML is not known. A more recent study used a mouse gene-expression array to characterize expression differences between ara-C-resistant and parental murine leukemia cell lines . Out of 929 genes with significant expression differences, 24 overlapped with prognostically important genes in human studies. However, as this study was performed in murine cell lines, the results of this study need to be evaluated in humans. Another limitation of this and the other studies previously mentioned, is the use of cell lines that were made resistant by growing them in increasing concentrations of ara-C over a long period. Ara-C-resistant cell lines prepared by this method usually exhibit multiple new chromosomal changes following exposure to increased doses of ara-C and this could influence the observed differences in expression.
Research on the role of pharmacogenomics in cytarabine response/resistance is still in its infancy. The key candidate genes in the activation pathway of ara-C, namely DCK, NT5C2, CDA, DCTD, SLC29A1 and RRM1/2, are being explored by various investigators and have been shown to harbor polymorphisms of functional significance. Since sensitivity to cytarabine depends on intracellular ara-CTP levels, genetic variations in these candidate genes could predict its antileukemic effect and the clinical outcome. Association studies are required in larger patient populations to confirm the observed results and to identify other candidate genetic variants of relevance to ara-C chemotherapy. In addition, the heterogeneity of AML itself, multidrug treatment regimens, gene–gene and gene–environment interactions should be accounted for in order to completely decipher the observed inter-patient variation in clinical response.
Although studies to identify the pharmacodynamic targets of ara-C in sensitive versus resistant cell lines are ongoing, future studies should be targeted in patient samples in order to identify the candidate genes contributing to resistance. Genome-wide association studies have also recently identified a few previously unknown candidate genes. However, whether these genes contribute to ara-C sensitivity in patients requires further evaluation. Once confirmed, the impact of genetic variation in genes of pharmacodynamic significance could be tested for ara-C sensitivity.
Future studies should be targeted at validating and confirming the pharmacogenetic markers that will identify patients with an altered response to ara-C therapy. This will result in an identification of the panel of candidate genes and potentially significant SNPs and would eventually be used as a diagnostic tool to predict response to ara-C. Such a panel of SNPs would be equally, if not more, useful markers for response to other nucleoside analogs such as cladribine, fludarabine, clofarabine, CNDAC and gemcitabine, as they all share a common pathway of drug activation.
Amit Kumar Mitra’s support with the bibliography and comments on this review are highly appreciated.
Financial & competing interest disclosure
The author is supported by a NIH grant R01CA132946. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in production of this manuscript.
Papers of special note have been highlighted as:
of considerable interest