Letters to the Editor 1929 Research Council, The Danish Cancer Society, The John and logic malignancies and inherited blood disorders. Mol Ther 2006; Birthe Meyer Foundation, and the Toyota Foundation. 13: 26–41. 2 Kolb HJ. Graft-versus-leukemia effects of transplantation and donor L Wenandy1, T Kollgaard1, A Letsch2, lymphocytes. Blood 2008; 112: 4371–4383. RS Andersen1, D Stather2 , T Seremet1, IM Svane1,3, 3 Wang W, Meadows LR, den Haan JM, Sherman NE, Chen Y, L Vindeløv4, MH Andersen1 and P thor Straten1 Blokland E et al. Human H-Y: a male-specific histocompatibility 1Department of Hematology, Center for Cancer Immune antigen derived from the SMCY protein. Science 1995; 269: Therapy (CCIT), Herlev University Hospital, Herlev, Denmark; 1588–1590. 2Department of Hematology, Oncology, Charite´, 4 Spierings E, Wieles B, Goulmy E. Minor histocompatibility Berlin, Germany; antigens–big in tumour therapy. Trends Immunol 2004; 25: 56–60. 3 5 Rongcun Y, Salazar-Onfray F, Charo J, Malmberg KJ, Evrin K, Maes Department of Oncology, Herlev University Hospital, H et al. Identification of new HER2/neu-derived peptide epitopes Herlev, Denmark and 4 that can elicit specific CTL against autologous and allogeneic The Hematopoietic Cell Transplantation Laboratory, carcinomas and melanomas. J Immunol 1999; 163: 1037–1044. Department of Hematology, Rigshospitalet, 6 Muller MR, Grunebach F, Kayser K, Vogel W, Nencioni A, Brugger Copenhagen, Denmark W et al. Expression of her-2/neu on acute lymphoblastic leukemias: E-mail: [email protected] implications for the development of immunotherapeutic approaches. Clin Cancer Res 2003; 9: 3448–3453. 7 Schuler MM, Donnes P, Nastke MD, Kohlbacher O, Rammensee HG, Stevanovic S. SNEP: SNP-derived epitope prediction program References for minor H antigens. Immunogenetics 2005; 57: 816–820. 8 Andersen RS, Wenandy L, Sorensen RB, thor Straten P, Andersen 1 Baron F, Storb R. Allogeneic hematopoietic cell transplantation MH. Mcl-1 and anticancer vaccination: identification of an HLA- following nonmyeloablative conditioning as treatment for hemato- A2-restricted epitope. Leukemia 2007; 22: 668–669.

SNP analyses in cytarabine metabolizing enzymes in AML patients and their impact on treatment response and patient survival: identification of CDA SNP C-451T as an independent prognostic parameter for survival

Leukemia (2009) 23, 1929–1932; doi:10.1038/leu.2009.113; published online 21 May 2009 Ara-C

Cytarabine (Ara-C) is the most important element of standard acute myeloid leukemia (AML) treatment regimens since more cytidine deaminase than 40 years. However, response to Ara-C treatment is marked Ara-U Ara-C by significant inter-individual differences and the development (G-88A, A-92G, C-451T, C-897A, A79C, G208A) deoxycytidine of Ara-C resistance is still a major drawback. Differences in kinase DNA sequences, such as single nucleotide polymorphisms (C364T, A727C) (SNPs), may affect the expression and/or function of specific deoxycytidylate deaminase drug protein targets and explain, at least in part, the inter- Ara-UMP Ara-CMP individual variations in the response to specific treatments. (A172G) Ara-C is a deoxycytidine analog, which blocks DNA synthesis.1 Key enzymes within the Ara-C metabolic pathway are the deoxycytidine kinase (DCK), which is important for Ara-C activation and two enzymes that regulate Ara-C degrada- dCTP DNA Ara-CTP tion: cytidine deaminase (CDA), which catalyses the conversion nucleus of Ara-C into inactive Ara-U, and deoxycytidylate deaminases (DCTD), which catalyses the conversion of Ara-CMP into Figure 1 Key enzymes within the Ara-C metabolic pathway inactive Ara-UMP.1 The phosphorylation of Ara-C into Ara- (analyzed SNPs are indicated). CMP by DCK2 is the rate limiting step for Ara-C activation and is followed by conversion into Ara-CTP, which then competes with dCTP for incorporation into DNA and subsequent block of DNA synthesis (Figure 1).1,2 years. AML FAB-M3 (acute promyelocytic leukemia (APL)) In this study, we identified and correlated SNPs of these patients were not included and treated in other European trials. three critical enzymes within the Ara-C metabolic pathway with The study was approved by the responsible national ethics drug-induced toxicity and 5-year overall survival (OS) in 360 committees and was in agreement with the declaration of caucasian AML patients. All the patients were treated within Helsinki and registered with the NCT number 00180115. the multicenter trial AML96 of the German Study Initiative Written informed consent was given by all the patients. The Leukemia (DSIL) between 1996 and 2003 (patient character- treatment schedule was published earlier.3 Briefly, in indivi- istics: Table 1). Median observation time was 6.2 (4.3–9.0) duals aged 60 years or younger, first induction therapy consisted

Leukemia Letters to the Editor 1930 Table 1 AML patient characteristics earlier published protocols.4 Analyses of FLT3-ITD5 and NPM15,6 mutations and chromosome analyses were carried P 7 n n % Median (range) out as published earlier. Cytogenetic risk groups are as follows: unfavorable: À5/del(5q), À7/del(7q), hypodiploid karyotypes Sex (Male/Female) 360 180/180 50/50 (besides 45,X,ÀY or –X), inv(3q), abnl12p, abnl11q, þ 11, þ 13, Age (years) 360 57 (18–79) þ 21, þ 22, t(6;9); t(9;22); t(9;11); t(3;3), multiple aberrations; Bone marrow blasts (%) 329 64 (20–99) WBC (Gpt/L) 358 26.0 (0.5–465.9) intermediate: patients without low-risk or high-risk constella- CD34 positivity (%) 343 27 (0–98) tion; favorable: t(8;21) and t(8;21) combined with other aber- rations. To identify SNPs in the three key enzymes of Ara-C ECOG status 293 metabolism (Figure 1), we carried out in silico analyses with 0/1 155 53 8 9 2/3 128 44 three SNP databases: dbSNP (NIH), HapMap and ABI (Applied 10 4/5 10 3 Biosystems, Foster City, CA, USA). Analyses were focused on coding, nonsynonymous SNPs, which can influence protein FAB classification 360 structure, activity, stability or localization because of change of M0 14 4 the amino acid sequence and on promoter SNPs, which can M1 76 21 lead to changes in expression levels. Known, validated SNPs M2 134 37 M4 66 18 from the literature were also included. M5 60 17 Validation of identified SNPs was carried out through M6 8 2 allele-specific SNP-Assays by TaqMan-PCR on an ABI M7 2 1 Prism 7700 Sequence Detection System (ABI, Darmstadt, Germany) in accordance with the instructions provided by the Disease status 360 manufacturer. De novo 303 84 Secondary 57 16 Our analyses confirmed a number of SNPs within the DCK (C364T ¼ SNP7) and CDA (A79C ¼ SNP1 (dbSNP: rs2072671), FLT3-ITD 355 G-88A ¼ SNP2, A-92G ¼ SNP4, C-451T ¼ SNP5 (dbSNP: Positive 104 29 rs532545)) genes, whereas other putative SNPs (DCK Negative 251 71 (A727C ¼ SNP8), CDA (C-897A ¼ SNP6 (dbSNP: rs10916823), G208A ¼ SNP3), DCTD (A172G ¼ SNP9)) were identified to be NPM1 mutation 358 Positive 118 33 either false positives or not present in caucasians, showing only Negative 240 67 wild-type allele (Figure 2a and Table 2). SNP3 G208A could not be confirmed in our study in concordance with Fitzgerald et al., Cytogenetics 347 who did not detect this SNP in caucasians either. Two additional Low risk 20 6 DCK promoter SNPs (C-360G and C-201T) could not be Intermediate risk 253 73 analyzed because of failure in manufacturing. High risk 74 21 Differences in clinical or laboratory parameters were com- Abbreviations: AML, acute myeloid leukemia; FAB, French-American- pared between SNPs using the Mann–Whitney test for contin- British. uous and the w2-test for categorical variables. OS was calculated from the date of diagnosis until death (failure) or last follow-up date (censored). Disease-free survival was measured from the date of documented CR to relapse (failure), to death in CR of mitoxantrone 10 mg/m2 (days 4–8), cytosine arabinoside (failure) or to alive in CR at last follow-up (censored). The (Ara-C) 100 mg/m2 (days 1–8) and VP16 100 mg/m2 (days 4–8) Kaplan–Meier method was used to estimate survival probabil- ( ¼ MAV). Second induction consisted of 2 Â 1000 mg/m2 Ara-C ities, which were compared between different groups using (days 1–5) and m-AMSA 100 mg/m2 (days 1–5) ( ¼ MAMAC). the log rank test. The Cox model was applied to identify Patient with intermediate cytogenetic risk were referred to allo- independent prognostic variables for survival. A stepwise geneic hematopoietic stem cell transplantation from an HLA- backward-selection procedure was carried out to determine identical sibling donor if possible. Patients with intermediate the final model. The significance level was 0.05. To provide cytogenetic risk without a sibling, and low-risk patients were quantitative information on the relevance of results, 95% CIs of randomized to receive intermediate (2 Â 1000 mg/m2 every odds ratios and hazard ratios were computed. Statistical 12 h days 1–6)(I-MAC) or high-dose (2 Â 3000 mg/m2 every analyses were carried out with the SPSS (SPSS Inc., Chicago, 12 h days 1–6)(H-MAC) Ara-C þ mitoxantrone (10 mg/m2 day 4 IL, USA) and the S-PLUS programs (Insightful AG, Reinach, to 6), which was followed by autologous PBSCT (intermediate Switzerland). The deviation from the Hardy–Weinberg Equili- cytogenetic risk) or MAMAC (low cytogenetic risk). Patients with brium for the three relevant SNPs (CDA SNP1, SNP4 and SNP 5) high-risk cytogenetics were referred to allogeneic hematopoietic with significant patient numbers in all three possible genotypes stem cell transplantation including the option of unrelated was tested using the w2-test. For pairwise linkage disequilibrium hematopoietic stem cell transplantation. Patients over 60 years between the genetic markers three estimators, D, D0 and r of age received two induction cycles containing daunorubicin were computed. These analyses were carried out using the 45 mg/m2 (days 3–5) and Ara-C 100 mg/m2 (days 1–7) ( ¼ DA). free statistical computing environment R (Version 2.3.1) and Patients in complete remission (CR) received MAMAC (see its library genetics for genetic analyses. above). All patient material (bone marrow/peripheral blood) was Genotype frequencies were analyzed in the all patients for obtained at the time of diagnosis. Genomic DNA was extracted CDA SNP1-6 and DCK SNP7-8 and DCTD SNP9 (Figure 2a, from mononuclear cells after density gradient centrifugation Table 2). CDA SNP1, SNP4 and SNP5 frequencies were in (BIOCOLL, Biochrom, Berlin Germany). For the discrimination accordance with the Hardy–Weinberg prediction (P ¼ 0.82, of CD34-positive cells, CD34 monoclonal antibody QBEnd10 P ¼ 0.65 and P ¼ 0.65, respectively). These three SNPs were in (Coulter, Hamburg, Germany) was used according to the linkage disequilibrium to each other (Po0.0001 for each

Leukemia Letters to the Editor 1931 100 Table 2 Cytidine deaminase (CDA), deoxycytidine kinase (DCK) and deoxycytidilate deaminase (DCTD) genotypes as assessed in 360 AML patients 80 Gene/SNP Genotype/locus Change n % 60 CDA SNP1 A79C/exon 1 Lys27Gln Homozygous wt 1_1 (A_A) Lys 147 41 40 Heterozygous 1_2 (A_C) Lys/Gln 164 45 percent [%] percent Homozygous v 2_2 (C_C) Gln 49 14 20 CDA SNP2 G-88A/promoter Homozygous wt 1_1 (G_G) Change of 354 98 0 TF-binding site? Heterozygous 1_2 (G_A) 5 1 Homozygous v 2_2 (A_A) 1 1 A79C G-88A A-92G C364T A727C G208A A172G C-451T C-897A CDA SNP3 G208A/exon 2 Ala70Thr DCK CDA DCTD Homozygous wt 1_1 (G_G) Ala 360 100 Heterozygous 1_2 (G_A) Ala/Thr 0 0 1.0 Homozygous v 2_2 (A_A) Thr 0 0 0.9 CDA SNP4 A-92G/promoter Homozygous wt 1_1 (A_A) Change of 151 42 0.8 TF-binding site? Heterozygous 1_2 (A_G) 161 45 0.7 Homozygous v 2_2 (G_G) 48 13 0.6 CDA SNP5 C-451T/promoter 0.5 Homozygous wt 1_1 (C_C) Change of 147 41 TF-binding site? 0.4 Heterozygous 1_2 (C_T) 163 45 Homozygous v 2_2 (T_T) 50 14 overall survival overall 0.3 n=147 n=163 CDA SNP6 C-897A/promoter 0.2 Homozygous wt 1_1 (C_C) Change of 360 100 0.1 n=50 TF-binding site? Heterozygous 1_2 (C_A) 0 0 0.0 Homozygous v 2_2 (A_A) 0 0 0246810 years DCK SNP7 C364T/Exon 3 Pro122Ser Homozygous wt 1_1 (C_C) Pro 353 98 1.0 Heterozygous 1_2 (C_T) Pro/Ser 7 2 Homozygous v 2_2 (T_T) Ser 0 0 0.9 DCK SNP8 A727C/Exon 6 Lys242Gln 0.8 Homozygous wt 1_1 (A_A) Lys 359 99 Heterozygous 1_2 (A_C) Lys/Gln 1 1 0.7 Homozygous v 2_2 (C_C) 0 0 0.6 DCTD SNP9 A172G/Exon 4 Asn58Asp 0.5 Homozygous wt 1_1 (A_A) Asn 360 0 n=77 Heterozygous 1_2 (A_G) Asn/Asp 0 0 0.4 Homozygous v 2_2 (G_G) Asp 0 0

0.3 n=82 Abbreviations: DCTD, deoxycytidylate deaminases; SNP, single disease free survival nucleotide polymorphism; TF, transcription factor; v, variant; wt, wild 0.2 n=26 type. Putative SNPs, which showed only wt allele in our analyses, were left 0.1 out of the table. 0.0 0246810 years

Figure 2 Allelic discrimination of single nucleotide polymorphisms possible pair), which is reflected by the distribution of allelic (SNPs) within genes encoding Ara-C-metabolizing enzymes from variant combinations (D0 0.92, 0.92 and 0.96 for the SNP1/4, acute myeloid leukemia (AML) patients and correlation with patient SNP1/5 and SNP4/5 combinations). The high linkage disequili- survival. (a) Results of the allelic discrimination of all analyzed SNPs in brium of SNP A-92G and C-451T occurring with high and Ara-C-metabolizing enzymes in the AML patients. Gray bar, homo- 11 zygous wild-type (wt) allele; striped bar, heterozygous and black bar, similar frequencies are in concordance with the earlier results. homozygous variant allele. (b) Kaplan–Meier estimates of overall The CDA SNP1, SNP4 and SNP5 genotypes had no significant survival (n ¼ 360) and (c) disease-free survival (n ¼ 185) in AML influence on WBC, RBC, platelet or bone marrow blast cell patients in relation to the cytidine deaminase SNP5 (C-451T) geno- counts or CD34 expression at diagnosis. However, patients type. Solid line, patients with the genotype 1_1; dashed line, patients with the CDA SNP5 2_2-genotype had significantly higher with the genotype 1_2; dotted line, patients with the genotype 2_2. lactate dehydrogenase levels when compared with the 1_2 or

Leukemia Letters to the Editor 1932 1_1-genotypes (P ¼ 0.01), whereas no such difference was Acknowledgements seen for the other two SNPs. Correlation with the molecular markers FLT3 and NPM1 revealed that the CDA SNP5 The contribution of all physicians and patients in the AML96 2_2-genotype was more frequently positive for FLT3-ITD treatment trial of the DSIL (Deutsche Studieninitiative Leuka¨mie) mutation (46%) when compared with the genotypes 1_2 is highly appreciated. This work was supported by grants from (26%) and 1_1 (26%) (P ¼ 0.03). No such differences were, the Deutsche Jose´ Carreras Leuka¨mie-Stiftung eV (DJCLS R 05/11), however, seen for the other genotypes. Also, no imbalances the Deutsche Forschungsgemeinschaft (MA 2057/2-4) and the were seen with respect to age, disease status, French-American- Dietmar Hopp Foundation (Heidelberg) to UM.

British (FAB) classification, Eastern Cooperative Oncology 1,2 1,2 2 3 Group (ECOG) status or cytogenetic risk between the different U Mahlknecht , C-L Dransfeld , N Bulut , M Kramer , C Thiede3, G Ehninger3 and M Schaich3 genotypes. 1 Saarland University Medical Center, Department of Internal The correlation of CDA SNP genotypes with Ara-C treatment- Medicine, Division of Immunotherapy and Gene Therapy, related toxicity (overall toxicities: 4/360 grade III/IV eye toxicity, Jose´ Carreras Research Center, Homburg/Saar, Germany; 8/360 skin toxicity, 116/360 gastrointestinal toxicity, 62/360 2University of Heidelberg Medical Center, Department of liver toxicity, 20/360 central nervous system toxicity) showed Hematology/Oncology, Heidelberg, Germany and that patients displaying a CDA SNP1 and SNP4 1_1-genotype 3University of Dresden, Department of Internal Medicine I, had lower incidences for grade III/IV liver toxicity when Dresden, Germany compared with the respective 1_2 or 1_1-genotypes (13 vs 21 E-mail: [email protected] vs 23%; P ¼ 0.03 in the Cochrane–Mantel–Haenszel Test), whereas no significant correlation was observed with the other References SNPs or SNP combinations and other types of toxicity. Another aim of this study was to reveal whether there is 1 Stam RW, den Boer ML, Meijerink JP, Ebus ME, Peters GJ, any influence of CDA genotypes on treatment outcome. A total Noordhuis P et al. Differential mRNA expression of Ara-C- of 185/360 (51%) patients reached CR after Ara-C double- metabolizing enzymes explains Ara-C sensitivity in MLL gene- rearranged infant acute lymphoblastic leukemia. Blood 2003; 101: induction therapy. Whereas the CDA SNP1, SNP4 and SNP5 1270–1276. genotypes and their combinations had no significant influence 2 Coleman CN, Stoller RG, Drake JC, Chabner BA. Deoxycytidine on response to treatment, 5-year OS was worse for the CDA kinase: properties of the enzyme from human leukemic granulo- SNP5 2_2-genotype vs the 1_1-genotype (OS: 13 vs 25%, cytes. Blood 1975; 46: 791–803. P ¼ 0.05). Differences in the 5-year disease-free survival were 3 Schaich M, Ritter M, Illmer T, Lisske P, Thiede C, Schakel U et al. not statistically significant (2_2-genotype: 23%, 1_1-genotype Mutations in ras proto-oncogenes are associated with lower mdr1 gene expression in adult acute myeloid leukaemia. Br J Haematol 39%, P ¼ 0.08). Patients with the 1_2-genotype had an inter- 2001; 112: 300–307. mediate overall (23%) and disease-free (29%) survival (Figure 2b 4 Gramatzki M, Ludwig WD, Burger R, Moos P, Rohwer P, and c). Grunert C et al. Antibodies TC-12 (‘unique’) and TH-111 (CD96) In the multivariate Cox analysis considering CDA SNP1 characterize T-cell acute lymphoblastic leukemia and a subgroup (SNP A79C), SNP4 (promoter SNP G-88A) and SNP5 genotype of acute myeloid leukemia. Exp Hematol 1998; 26: 1209–1214. (promoter SNP C-451T), cytogenetic risk, WBC, platelet 5 Thiede C, Steudel C, Mohr B, Schaich M, Schakel U, Platzbecker U et al. Analysis of FLT3-activating mutations in 979 patients with count, lactate dehydrogenase, CD34 expression, bone marrow acute myelogenous leukemia: association with FAB subtypes and blast count, NPM/FLT3 mutation status, age and disease status, identification of subgroups with poor prognosis. Blood 2002; 99: the CDA SNP5 genotype proved to be an independent 4326–4335. prognostic factor for survival: patients with the 2_2-genotype 6 Thiede C, Koch S, Creutzig E, Steudel C, Illmer T, Schaich M et al. had a 450% increased risk for death compared with patients Prevalence and prognostic impact of NPM1 mutations in 1485 with the 1_1-genotype (hazard ratio 1.56 (95%-CI 1.09-2.24); adult patients with acute myeloid leukemia (AML). Blood 2006; 107: 4011–4020. ¼ P 0.02). 7 Mohr B, Bornhauser M, Thiede C, Schakel U, Schaich M, With these results of our study, we could reveal the clinical Illmer T et al. Comparison of spectral karyotyping and 12 relevance of earlier findings of Fitzgerald et al., who showed conventional cytogenetics in 39 patients with acute myeloid that the CDA promoter SNP C-451T functionally leads to leukemia and myelodysplastic syndrome. Leukemia 2000; 14: reduction in the CDA expression level and the level of CDA 1031–1038. enzymatic activity, which was already assumed to increase the 8 Smigielski EM, Sirotkin K, Ward M, Sherry ST. dbSNP: a database of single nucleotide polymorphisms. Nucleic Acids Res 2000; 28: risk of toxicity and adverse events associated with the standard 352–355. dosage of Ara-C therapy, which is in accordance with our data. 9 The International HapMap Project. Nature 2003; 426: 789–796. Despite Ara-C being the most important drug in AML therapy, 10 De La Vega FM, Dailey D, Ziegle J, Williams J, Madden D, patients receive a multiagent therapy including anthracyclines, Gilbert DA. New generation pharmacogenomic tools: a SNP mitoxantrone, amsacrine and etoposide in combination with linkage disequilibrium Map, validated SNP assay resource, and Ara-C. Therefore, these other agents could have had some high-throughput instrumentation system for large-scale genetic studies. Biotechniques 2002, Suppl: 48–50, 52, 54. influence on liver toxicity and treatment response independent 11 Sugiyama E, Kaniwa N, Kim SR, Kikura-Hanajiri R, Hasegawa R, of the examined CDA SNPs. Maekawa K et al. Pharmacokinetics of gemcitabine in Japanese Nevertheless, our results show that the CDA SNP C-451T cancer patients: the impact of a cytidine deaminase poly- is a significant, independent prognostic factor for survival, morphism. J Clin Oncol 2007; 25: 32–42. which further prompts the need for individualized chemother- 12 Fitzgerald SM, Goyal RK, Osborne WR, Roy JD, Wilson JW, apy in patients who are being given Ara-C or other nucleoside Ferrell RE. Identification of functional single nucleotide poly- morphism haplotypes in the cytidine deaminase promoter. analogs. Hum Genet 2006; 119: 276–283.

Leukemia