Leukemia (2003) 17, 1845–1850 & 2003 Nature Publishing Group All rights reserved 0887-6924/03 $25.00 www.nature.com/leu Concurrent methylation of multiple in childhood ALL: Correlation with phenotype and molecular subgroup MI Gutierrez1, AK Siraj1,, M Bhargava2,8, U Ozbek3, S Banavali4, MA Chaudhary5, H El Solh6 and K Bhatia1,7

1King Fahad National Centre for Children’s and Research, Riyadh, Saudi Arabia; 2Department of Hematology, All India Institute of Medical Sciences, New Delhi, India; 3Department of , Institute for Experimental Medicine, Istanbul University, Istanbul, Turkey; 4Department of Medical Oncology, Tata Memorial Hospital, Mumbai, India; 5Department of Biostatistics, Epidemiology and Scientific Computing, Riyadh, Saudi Arabia; 6Department of Pediatric Hematology-Oncology, King Faisal Specialist Hospital and Research Centre, Riyadh, Saudi Arabia; and 7International Network for Cancer Treatment and Research, Brussels, Belgium

Multiple genes have been shown to be independently hyper- functionally relevant genes via aberrant DNA methylation has methylated in lymphoid malignancies. We report here on the been linked to diverse human pathologies.7,8 extent of concurrent methylation of E-cadherin, Dap-kinase, O6MGMT, , , p15 and p14 in 129 pediatric ALL cases. Cancer has been recognized as a genetic disease for long While most of these genes demonstrated methylation in a time, but more recent data indicate that it can also be proportion of cases, O6MGMT, p16 and p14 were infrequently considered as an epigenetic disease. The normal epigenetic methylated (11, 7 and 3%, respectively). Methylation of at least equilibrium is dramatically altered in tumor cells. Global one was found in the vast majority (83%) of cases. To hypomethylation of the genome occurs simultaneously with determine the extent and concordance of methylation we hypermethylation of CpG islands in critical genes such as tumor- calculated a methylation index (MI ¼ number of methylated 9,10 genes/number of studied genes) for each sample. The average suppressor genes. Genes silenced by this mechanism include MI was 0.28, corresponding to 2/7 methylated genes. MI was genes that participate in every cellular process: (Rb, correlated with standard prognostic factors, including immu- p16, p15), DNA repair (BRCA1, MLH1, MGMT), nophenotype, age, sex, WBC and presence of specific trans- (DAP-kinase), drug-detoxification (GSTP1), etc. These changes locations (TEL-AML1, BCR-ABL, E2A-PBX1 or MLL-AF4). We are usually tumor-specific and therefore, not all the genes are determined that children X10 years old and children presenting methylated in all tumor types. Furthermore, some types of with high WBC (X50 Â 109/l) both associated with a higher MI cancer demonstrate a characteristic ‘methylator phenotype’ in (Po0.01 and o0.05, respectively). T-ALLs demonstrated a lower MI (median ¼ 0.17) than precursor B ALLs (median ¼ 0.28). which multiple genes (mostly MINTs, methylated in tumors) are 11–13 Among the different molecular subgroups, MLL-ALLs had the concurrently methylated. It is not clear yet as to what highest MI (mean ¼ 0.35), while ALLs carrying the t(1;19) had causes these different patterns of methylation, but it is likely that the lowest MI (mean ¼ 0.07). The most common epigenetic some selective growth advantages are gained during transform- lesion in childhood ALL was methylation of E-cadherin (72%) ation. It has been well documented that silencing of tumor- independent of the molecular subtype or other clinicopatho- suppressor genes by hypermethylation contributes to logical factors. 14,15 Leukemia (2003) 17, 1845–1850. doi:10.1038/sj.leu.2403060 the transformation process. Keywords: ; chromosomal translocations; Several genes have been shown to be independently immunophenotype; tumor-suppressor genes; hematological hypermethylated in hematological malignancies.16–20 Recent neoplasia studies have focused upon methylation analysis of multiple genes. Melki et al 21 first reported concurrent methylation of calcitonin, estrogen (ER), E-cadherin, p15, p16 and Hic1, while Rb was unmethylated in AML. Toyota et al 22 Introduction studied other loci in AML and showed concurrent methylation of MyoD, Pitx2, Gpr37 and Scd4 from 12 loci analyzed in The major epigenetic modification of human genomic DNA is correlation with estrogen-receptor hypermethylation. Recent the methylation of cytosine residues located within the reports demonstrate significant methylation in adult ALL and stability in relapsed tumors.23,24 However, no specific methyl- dinucleotide CpG. Hypermethylation occurs in a nonrandom, 11–13 tissue-specific manner. It has been estimated that 3–4% of all ator phenotype as described in solid tumors has been 1 reported in hematological malignancies. A correlation with cytosines are methylated in normal human DNA. The distri- 25 bution of CpG dinucleotides is not uniform throughout the prognosis has been suggested for individual genes such as and O6MGMT.26 Moreover, hypermethylation of p16 have been genome; some regions are CpG-rich and thus denominated CpG 27 islands.2 These CpG islands are more common in the 50 region associated with disease progression inT-ALL, while epigenetic 0 modification of the ER has been associated with better outcome of genes (5 untranslated, promoter and 1). CpG islands are 28 usually unmethylated in normal tissues, except in the case of in AML. Few reports that have studied childhood ALL mostly include imprinted genes, X in women, germline- or tissue- 29–31 3–6 single candidate genes and only limited data are available specific genes. Promoter hypermethylation causes a tran- 32 scriptional block in the genes involved. Indeed, silencing of on the extent of concurrent hypermethylation. It is likely that the biology of childhood ALL is different from adult ALL Correspondence: Dr K Bhatia, PO Box 3354, MBC # 98-16, Riyadh and that different leukemogenic events control these 11211, Saudi Arabia, Fax: +966 1 246 5422 disease entities. At least one study suggests that in spite The first two authors contributed equally. of different and characteristic genetic lesions in adult and 8Presently at Sir Ganga Ram Hospital, New Delhi, India. childhood ALL, epigenetic lesions may not be different in these Received 27 January 2003; accepted 12 May 2003 diseases.32 Methylation of genes in childhood ALL MI Gutierrez et al 1846 To determine clearly the methylation pattern in childhood Materials and methods ALL and to define if different subtypes show differences, we have studied a large series of samples at multiple loci. We Clinical samples determined the extent of concurrent hypermethylation of seven genes in 129 primary childhood ALLs and correlated these data Bone marrow aspirates (BM) or peripheral blood lymphocytes with clinicopathological features. (PB) were obtained from 129 patients presenting with ALL at the time of diagnosis and prior to chemotherapy, following IRB approved projects and Institutional guidelines. The patients Table 1 Clinicopathological characteristics of the patients at were o1–16 years of age, with a median age of 6 years (Table 1). presentation The inclusion of leukemic PB sample was restricted to those that contained more than 25% peripheral blasts. Mononuclear cells Total Precursor B T lineage were isolated using Ficoll-Hypaque density gradient centrifuga- N 129 99 (77%) 30 (23%) tion and genomic DNA was extracted with the Purogene Sex (Gentra, Minneapolis, MN, USA). Male 88 62 26 Female 41 37 4 Ratio M:F 2.1:1 1.7:1 6.5:1 Sodium bisulfite modification of DNA

Age (years) The procedure was previously described. Briefly, 5 mg of DNA Range o1–16 o1–16 o1–15 1 Median 6 5 8 was first denatured with 0.4 M NaOH for 30 min at 42 C and o1211then incubated in 3 M sodium bisulfite, 10 mM hydroquinone 1–9 96 78 18 (Sigma, St Louis, MO, USA) at 551C for 16 h. Modified DNA was X10 31 20 11 finally purified with Gene Clean III kit (Bio101, Vista, CA, USA), desulfonated with 0.4 M NaOH for 15 min at 371C and finally WBC (x109/l) precipitated in ethanol and resuspended in dH2O. o10 50 48 2 10–49 33 25 8 50–99 13 9 4 Methylation-specific PCR (MSP) X100 33 17 16

Molecular subgroup A measure of 200 ng of sodium bisulfite modified DNA was used TEL-AML1 21 as template in PCR reactions containing 25 pmoles of each E2A-PBX1 4 primer, 200 mM each dNTP, 1 U Hotstart Taq polymerase mBCR-ABL 4 and 1 Â Q buffer (Qiagen, Valencia, CA, USA) in a final volume 11q23 7 of 25 ml. Table 2 shows all the primers used, specific for SIL-SCL 7 the methylated (M) and the unmethylated (U) forms and Undetermined 63 23 the PCR conditions. These conditions were derived from

Table 2 PCR conditions and primers specific for the methylated (M) and unmethylated (U) status of the genes in this study

Gene Primers Annealing Temp. (1C) MgCl2 (mM) No. of cycles E-cadherin M taattagcggtacggggggc 59 4.5 35 cgaaaacaaacgccgaatacg E-cadherin U ttagttaattagtggtatggggggtgg 59 4.5 35 accaaacaaaaacaaacaccaaataca DAP-kinase M ggatagtcggatcgagttaacgtc 59 4.5 35 ccctcccaaacgccga DAP-kinase U ggaggatagttggattgagttaatgtt 59 4.5 35 caaatccctcccaaacaccaa p73 M ggacgtagcgaaatcggggttc 62 4.5 35 accccgaacatcgacgtccg p73 U aggggatgtagtgaaattggggttt 62 4.5 35 atcacaaccccaaacatcaacatcca O6MGMT M tttcgacgttcgtaggttttcgc 56 3.5 35 gcactcttccgaaaacgaaacg O6MGMT U tttgtgttttgatgtttgtaggtttttgt 56 4.5 35 aactccacactcttccaaaaacaaaaca p14 M gtgttaaagggcggcgtagc 54 4.5 35 aaaaccctcactcgcgacga p14 U tttttggtgttaaagggtggtgtagt 56 4.5 35 cacaaaaaccctcactcacaacaa p15 M gcgttcgtattttgcggtt 57 3.5 35 cgtacaataaccgaacgaccga p15 U tgtgatgtgtttgtattttgtggtt 59 4.5 35 ccatacaataaccaaacaaccaa p16 M ttattagagggtggggcggatcgc 68 1.5 33 gaccccgaaccgcgaccgtaa p16 U ttattagagggtggggtggattgt 58 4.5 33 caaccccaaaccacaaccataa

Leukemia Methylation of genes in childhood ALL MI Gutierrez et al 1847 available by a real-time multiplex RT-PCR assay as previously described.33 Other 11q23 rearrangements were determined by Southern blot analysis with an MLL probe.

Results

Each MSP was first standardized using bisulfite treated IVM as a positive control for methylation and bisulfite-treated normal DNA as a positive control for the absence of methylation. M primers consistently yielded a band when IVM was used as template. Similarly, U primers efficiently amplified the fragment of the correct length from normal DNA (Figure 1a). None of the primers yielded any product in the absence of template or when non-bisulfite-treated DNA was used as template. The specificity of these reactions was further reconfirmed using cell lines with completely methylated or completely unmethylated loci that yielded the specific amplicon either with primers M or U but never with both. None of the cell lines used as controls for any given yielded cross-contamination or nonspecific PCR products (Figure 1a). The results were concordant with available published data. In order to analyze if any background methylation in normal lymphocytes was likely to interfere with the assessment of leukemic samples under these conditions, we analyzed the methylation status of all the genes using DNA from healthy blood donors (12 adults and five children). All these normal samples demonstrated clear absence of methylation of DAP- Figure 1 MSP analyses of controls (top) and ALL samples (bottom). kinase, p73, O6MGMT, p16, p15 and p14 (Figure 1a, lanes PBL M ¼ methylated reaction and U ¼ unmethylated reaction. In vitro 1–7). However, some background methylation of E-cadherin in methylated DNA (IVM) and normal DNA (N) were included in the normal controls was observed as previously reported.34,35 It is reactions with M primers and U primers, respectively. All other likely that these faint bands represent methylation in a small are shown in both reactions. DNA from the indicated cell lines demonstrate only methylated or only unmethylated products for each fraction of cells. Negative controls did not yield such bands. The gene. PBL ¼ normal lymphocytes from seven individuals. Three T-ALLs intensity of these bands was visibly different than those observed (14, 15, 16) and seven B-ALLs (14, 15, 16, 17, 18, 56, 57) are shown. in leukemic samples (in Figure 1 compare a with b). Densito- Leukemia sample numbers correspond to cases in Figure 2. metric measurements indicated that the ratio between methyl- ated and unmethylated bands was at an average 30-fold higher in patients than in normals. We then analyzed 129 primary ALL samples obtained before those used in the literature. In the analysis of each candidate treatment, including 99 of B-lineage and 30 of T-lineage. The gene, we included four types of controls to ensure specificity: (1) clinical characteristics are shown in Table 1. For most of the SssI in vitro methylated DNA (IVM) (Figure 1a, first lane), (2) samples (105/129, 81%) DNA was sufficient for determination DNA from a healthy individual (normal control) (Figure 1a, of the methylation status of all seven genes. However, for 16 second lane), (3) appropriate monoclonal cell lines (Figure 1a, cases (12%) we could analyzed six genes and for a minor second panel) and (4) non-template (blank). PCR products were number of samples (8/129, 6%) DNA was available for only five analyzed following electrophoresis on 4% agarose gels contain- or four genes. The loci were chosen because they were ing ethidium bromide. Specificity of each reaction was previously described as aberrantly methylated in lymphoid demonstrated when for each gene the M primers consistently malignancies. We used a methylation index (MI) to estimate and yielded a band from IVM, the U primers produced a band from compare the extent of methylation in ALL (MI: ratio between the normal DNA and no bands were obtained from DNA not treated number of methylated genes and the number of analyzed with bisulfite. Information on control monoclonal cell lines with genes). We compared bone marrow with peripheral blood completely methylated or completely unmethylated target genes (425% blasts) samples and no notable differences were was obtained from the literature and screened from the elucidated (median MI-BM ¼ 0.26 and median MI-PB ¼ 0.28) repository available in the Laboratory. based on the origin of the samples. Cloning various PCR products and sequencing at least three E-cadherin, DAP-Kinase, p73 and p15 showed methylation in clones from each further reconfirmed the M status of the p16 a significant proportion of cases (72, 30, 26 and 23%). By gene. contrast, O6MGMT, p16 and p14 were rarely altered (11, 7 and 3%, respectively). To detect methylation of p16 we used one of the strategies widely reported previously (MSP primers spanning Real-time multiplex RT-PCR for determination of 150 bp) and we confirmed our data by cloning and sequencing translocations PCR products. This confirmed that all products corresponded to methylated p16; 75% of the clones carried all 11 CpG sites The presence of the four most frequent translocations, that is, methylated and the remaining 25% of clones demonstrated 12;21, 1;19, 9;22 and 4;11, was determined whenever RNA was lesser densities but at least four methylated sites.

Leukemia Methylation of genes in childhood ALL MI Gutierrez et al 1848 MI ranged from 0 to 1 with a median of 0.28, corresponding to compare the frequency with which another locus was to two genes/sample. The majority of samples (83%) showed methylated when that gene was methylated or unmethylated. methylation of at least one gene and one-fifth demonstrated We only identified a statistical association between methylation three or more methylated genes, indicated by an MI X0.43. in p16 and in DAP-kinase (Po0.005). Further coordination of In order to analyze the coexistence of methylation in different methylation at the seven loci was analyzed by the Wilcoxon loci in childhood ALL (Figure 2), we performed statistical rank sum test comparing the status of each gene (M or U) with analyses. For each gene, two-sided Fisher’s exact tests were used a MI calculated with the remaining genes. Again, DAP-kinase and p16 were the only genes that appeared more likely methylated in samples with greater extent of methylation (Po0.005 and o0.002, respectively). In contrast, methylation of p73, p15 and O6MGMT was equally distributed in all ranges of MI. We then correlated methylation status with clinicopatho- logical features at presentation, including age, sex, WBC and immunophenotype. Two associations were statistically signifi- cant. Leukemic samples from children X10 years old as well as leukemias from patients with WBC X50 Â 109/l had higher MI (Po0.025 and o0.05, respectively). Moreover, a lesser extent of methylation was observed in T-ALLs (median MI ¼ 0.17) than in B-ALLs (median MI ¼ 0.28). Similarly, while 14% of B-ALLs did not demonstrate methylation of any gene, 27% of T-ALLs fell in this category. We also analyzed each locus with respect to these clinical parameters and observed that leukemias in children X10 years are more likely to inactivate p73 (Po0.005). Although the frequency of each gene methylation was generally similar in both immunophenotypic groups, for DAP-kinase it was clearly higher in B-ALLs (35%) than in T-ALLs (13%) (Po0.025). To assess if the profile of methylation differed between molecular subgroups, we determined by real-time multiplex RT- PCR and Southern blot the presence of five common chromo- somal rearrangements. t(12;21) was detected in 21/99 B-ALLs, t(1;19) in 4/99 cases, t(9;22) in 4/99 cases, 11q rearrangements in 7/99 patients and SIL-SCL rearrangement in 7/30 T-ALLs. Although the numbers of each subgroup are small, some interesting observations were noted. E2A-PBX1 leukemias showed the lowest MI (mean 0.07) and MLL leukemias had the highest MI (mean 0.35). T-ALLs with the SIL-SCL rearrange- ment demonstrated a lower MI (mean 0.10) than the T-ALLs without this rearrangement (mean 0.27). We then analyzed the frequency of methylation of each of the seven genes in these different molecular subtypes. A statistically significant difference was detected between the presence of methylated DAP-kinase in MLL-rearranged leukemias compared to the TEL-AML1 subtype (71 and 14%, respectively, Po0.05).

Discussion

There is growing evidence that DNA methylation plays a key role in silencing critical genes in cancer.9,10 We report in this study the pattern and degree of aberrant methylation of multiple genes in primary childhood ALL. We included samples from 99 patients with precursor B ALL and 30 patients with T-ALL. Seven candidate genes were chosen because they were epigenetically modified in lymphoid malignancies and associated with decreased expression in previous studies. These included E-cadherin, DAP-kinase, p73, p16, p15, p14 and O6MGMT. Figure 2 Concurrent methylation of seven genes in childhood Hypermethylation of their 50 regions had been assayed by ALL. Precursor B ALLs that carry a known translocation (TEL-AML1, methylation-specific PCR (Table 2, Figure 1). It is likely that BCR-ABL, E2A-PBX1 or involving band 11q23) are shown in the first additional candidate genes are also relevant, including p21, panel, precursor B ALLs that do not carry any of these rearrangements and a small proportion of cases that could not be analyzed are shown p57, HIC, ER, etc and other novel loci. Further studies are in the middle panel and T-ALLs are shown in the right panel. Green needed to elucidate the complex epigenetic lesions. depicts methylation, yellow depicts unmethylation and blue depicts The data presented here, nonetheless, demonstrate that of the gene. multiple epigenetic lesions are frequent events in childhood

Leukemia Methylation of genes in childhood ALL MI Gutierrez et al 1849 ALL (Figure 2). In total, 83% of samples carried at least one extent of methylation can be used as an independent prognostic methylated locus and one-fifth carried three or more altered factor. Indeed, reports on single genes also suggested loci. A study with a smaller cohort (N ¼ 16) and only three genes a prognostic value of aberrant epigenetic lesions.25,26,28,31 in common (p15, p16, p73), observed 93% of cases with some The association with WBC may indicate that promoter level of methylation and 13% with X3 modified promoters.32 It hypermethylation accumulates with increasing tumor burden is noteworthy that the frequencies of methylation do not differ and most likely with tumor progression.27,30 Wong et al 31 had notably between the studies. Both studies also indicate that p15 previously shown that the frequencies of p15 and p16 methylation is more common than p16 inactivation. In our methylation in ALL were higher in adults than in children. larger series, this was even more notable in T-ALLs. There is still Our data suggest that the overall epigenetic modification is age- controversy on the role of these -dependent kinase dependent. Although Garcia-Manero et al 32 reported no inhibitors in lymphoid malignancies,17,36–38 suggesting that differences between adult and pediatric ALL because no their inactivation may be critical only in a still undetermined statistical significance was achieved, at least in p73, p15 and molecular subtype of leukemia. For example, it has been MDR1, there is a trend towards higher methylation in adults. In reported that p16 methylation is frequent in MLL-leukemias, but our larger cohort, methylation of p73 reached statistical we identified only one out of seven such cases. This is still significance. Lack of p73 expression has been found in a different than none of the other 29 cases carrying translocations. proportion of ALLs.19 Although no inactivating mutations were Alternatively, geographic variations in methylation patterns may found, hypermethylation of the promoter was associated with explain some differences.39 O6MGMT was rarely methylated, reduced disease-free survival.42 Indeed, different etiologies distinguishing childhood ALL from other lymphoid malignancies between childhood and adult ALL may explain differences such as diffuse large B-cell lymphomas.26 Although it has been observed in methylation of certain genes like E-cadherin. For reported that 20% of adult ALLs do not express p14,40 we did example, we found a higher frequency of methylated E-cadherin not find a significant level of methylation in pediatric cases. Two than previously reported for ALL – 53% – that included adults.18 possibilities exist; either p14 inactivation is specific for adult Although this difference may be due to geographic variations, it ALL, or other mechanisms of inactivation participate in silencing can also be due to different age populations. p14 in childhood ALL. These examples support the hypothesis A direct clinical application of studies of methylation in tumor that aberrant CpG island hypermethylation is not only tissue- cells is the use of aberrant DNA methylation at certain sites as specific but also disease-specific and suggests an important role a of malignant transformation.43–45 Indeed, some in pathogenetic mechanisms. promoter hypermethylation can occur early in tumor progres- CpG island hypermethylation of the studied genes has been sion. Furthermore, the positive PCR signal from such a lesion previously described as associated with gene silencing. We have will not be masked by the presence of normal cells. These also preliminary analyzed expression of DAP-kinase and advantages could facilitate both early detection and assessment E-cadherin by real-time quantitative RT-PCR and clearly of minimal residual disease. We found that 72% of childhood observed that the presence of methylated CpG sites leads to a ALL carry methylation of E-cadherin indicating that this is an reduced level of transcripts (data not shown). early epigenetic event during leukemogenesis. Since an During statistical analyses of concomitant methylation in our extremely low level of methylated E-cadherin promoter was cohort of patients, only p16 was significantly associated with detected in blood cells from healthy individuals, appropriate DAP-kinase (Po0.05), suggesting that these two pathways, that quantitative assays need to be first established. The presence of is, allowing cell cycle to proceed and preventing apoptosis, may methylated E-cadherin across subtypes of ALL strongly suggests collaborate. Such a positive association was previously ob- that this could be a convenient biomarker for early detection, for served in tumors from head and neck.41 Furthermore, these two example of CNS involvement, and in follow-up studies. loci were more likely to be methylated concurrently with other genes and therefore, they associated with a higher MI (Po0.005). Overall, methylation was not randomly distributed; however, discrete and unique methylator phenotypes as in solid References tumors could not be established. Clearly, larger series are necessary to dissect these patterns that emerge from the data 1 Ehrlich M, Gama-Sosa MA, Huang LH, Midgett RM, Kuo KC, presented. McCune RA et al. Amount and distribution of 5-methylcytosine in Comparison between precursor B-ALLs and T-ALLs demon- human DNA from different types of tissues of cells. Nucleic Acids strated the existence of some differences in the pattern of Res 1982; 1: 2709–2721. methylation between lineages. Indeed, epigenetic modification 2 Bird AP. CpG-rich islands and the function of DNA methylation. Nature 1986; 321: 209–213. of DAP-kinase was significantly lower in T-ALLs (13%) than in 3 Jones PA. The DNA methylation paradox. Trends Genet 1999; 15: B-ALLs (35%) (Po0.025). A similar observation was reported in 34–37. NHL from both lineages.16 The extent of methylation also 4 Tremblay KD, Saam JR, Ingram RS, Tilghman SM, Bartolomei MS. differed between lineages. T-ALLs were less methylated (median A paternal-specific methylation imprint marks the alleles of the MI ¼ 0.17) than B-ALLs (median MI ¼ 0.28) and a higher fraction mouse H19 gene. Nat Genet 1995; 9: 407–413. of T-ALLs had no methylation (27%) compared to B-ALLs (14%). 5 Riggs AD, Pfeifer GP. X-chromosome inactivation and cell memory. Trends Genet 1992; 8: 169–174. These differences could reflect different etiologies and/or risk 6 Busslinger M, Hurst J, Flavell RA. DNA methylation and the factors and may have relevant clinical implications because the regulation of globin . Cell 1983; 34: 197–206. extent of methylation will most likely determine the success of 7 Singal R, Ginder GD. DNA methylation. Blood 1999; 93: demethylating agents in future chemotherapeutic regimens. 4059–4070. Methylation of each locus and the MI for each sample were 8 Charache S, Dover G, Smith K, Talbot CC, Moyer M, Boyer S. correlated to clinical data, specifically to standard prognostic Treatment of sickle cell anemia with 5-azacytidine results in increased fetal hemoglobin production and is associated with factors from the patients at presentation. Some interesting nonrandom hypomethylation of DNA around the gamma-delta- 9 observations arose. Age (X10 years) and WBC (X50 Â 10 /l) beta-globin gene complex. Proc Natl Acad Sci USA 1983; 80: positively correlated with MI. It remains to be seen whether the 4842–4846.

Leukemia Methylation of genes in childhood ALL MI Gutierrez et al 1850 9 Esteller M, Herman JG. Cancer as an epigenetic disease: DNA 28 Li Q, Kopecky KJ, Mohan A, Willman CL, Appelbaum FR, Weick methylation and chromatin alterations in human tumors. J Pathol JK et al. methylation is associated with improved 2002; 196: 1–7. survival in adult acute myeloid leukemia. Clin Cancer Res 1999; 5: 10 Costello JF, Fruhwald MC, Smiraglia DJ, Rush LJ, Robertson GP, 1077–1084. Gao X et al. Aberrant CpG-island methylation has non-random and 29 Nakamura M, Sugita K, Inukai T, Goi K, Iijima K, Tezuka T et al. tumor-type-specific patterns. Nat Genet 2000; 25: 132–138. p16/MTS1/INK4A gene is frequently inactivated by hypermethyl- 11 Toyota M, Ahuja N, Toyota MO, Herman JG, Baylin SB, Issa JP. ation in childhood acute lymphoblastic leukemia with 11q CpG island methylator phenotype in colorectal cancer. Proc Natl translocation. Leukemia 1999; 13: 884–890. Acad Sci USA 1999; 96: 8681–8686. 30 Maloney KW, McGavran L, Odom LF, Hunger SP. Acquisition of 12 Toyota M, Ahuja N, Suzuki H, Itoh F, Toyota MO, Imai K et al. p16(ink4a) and p15(ink4b) gene abnormalities between initial Aberrant methylation in gastric cancer associated with the CpG diagnosis and relapse in children with acute lymphoblastic island methylator phenotype. Cancer Res 1999; 59: 5438–5442. leukemia. Blood 1999; 93: 2380–2385. 13 Strathdee G, Appleton K, Illand M, Millan DWM, Sargent J, Paul J 31 Wong IHN, Ng MHL, Huang DP, Lee JCK. Aberrant p15 promoter et al. Primary ovarian carcinomas display multiple methylator methylation in adult and childhood acute leukemias of nearly all phenotypes involving known tumor suppressor genes. Am J Pathol morphologic subtypes: potential prognostic implications. Blood 2001; 158: 1121–1127. 2000; 95: 1942–1949. 14 Esteller M. CpG island hypermethylation and tumor suppressor 32 Garcia-Manero G, Jeha S, Daniel J, Williamson J, Albitar M, genes: a booming present, a brighter future. 2002; 21: Kantarjian HM et al. Aberrant DNA methylation in pediatric 5427–5440. patients with acute lymphocytic leukemia. Cancer 2003; 97: 15 Sakai T, Toguchid J, Ohtani N, Yandell DW, Rapaport JM, Dryja 695–702. TP. Allele-specific hypermethylation of the retinoblastoma tumor- 33 Siraj AK, Ozbek U, Sazawal S, Sirma S, Timson G, Al-Nasser A suppressor gene. Am J Hum Genet 1991; 48: 880–888. et al. Pre-clinical validation of a monochrome real-time multiplex 16 Katzenellenbogen RA, Baylin SB, Herman JG. Hypermethylation assay for translocations in childhood acute lymphoblastic leuke- of the DAP-Kinase CpG island is a common alteration in B-cell mia. Clin Cancer Res 2002; 8: 3832–3840. malignancies. Blood 1999; 93: 4347–4353. 34 Melki JR, Vincent PC, Brown RD, Clark SJ. Hypermethylation of 17 Herman JG, Civin CI, Issa JP, Collector MI, Sharkis SJ, Baylin SB. E-cadherin in leukemia. Blood 2000; 95: 3208–3213. Distinct patterns of inactivation of p15-ink4b and p16-ink4a 35 Bornman DM, Mathew S, Alsruhe J, Herman JG, Gabrielson E. characterize the major types of hematological malignancies. Methylation of the E-cadherin gene in bladder neoplasia and in Cancer Res 1997; 57: 837–841. normal urothelial from elderly individuals. Am J Pathol 2001; 159: 18 Corn PG, Smith BD, Ruckdeschel ES, Douglas D, Baylin 831–835. SB, Herman JG. E-cadherin expression is silenced by 5’CpG 36 Baur AS, Shaw P, Burri N, Delacretaz F, Bosman FT, Chaubert P. island methylation in acute leukemia. Clin Cancer Res 2000; 6: Frequent methylation silencing of p15-INK4b (MTS2) and p16- 4243–4248. INK4a (MTS1) in B-cell and T-cell lymphomas. Blood 1999; 94: 19 Kawano S, Miller CW, Gombart AF, Bartram CR, Matsuo Y, Asou 1773–1781. H et al. Loss of p73 gene expression in leukemias/lymphomas due 37 Batova A, Diccianni MB, Yu JC, Nobori T, Link MP, Pullen J et al. to hypermethylation. Blood 1999; 94: 1113–1120. Frequent and selective methylation of p15 and deletion of both 20 Thomas X, Teillon MH, Belhabri A, Rimokh R, Fiere D, Magaud JP p15 and p16 in T-cell acute lymphoblastic leukemia. Cancer Res et al. Hypermethylation of calcitonin gene in adult acute leukemia 1997; 57: 832–836. at diagnosis and during complete remission. Hematol Cell Ther 38 Guo SX, Taki T, Ohnishi H, Piao HY, Tabuchi K, Bossho F et al. 1999; 41: 19–26. Hypermethylation of p16 and p15 genes and RB 21 Melki JR, Vincent PC, Clark SJ. Concurrent DNA hypermethylation expression in acute leukemia. Leuk Res 2000; 24: 39–46. of multiple genes in acute myeloid leukemia. Cancer Res 1999; 39 Shen L, Ahuja N, Shen Y, Habib NA, Toyota M, Rashid A et al. 59: 3730–3740. DNA methylation and environmental exposures in human 22 Toyota M, Kopecky KJ, Toyota MO, Jair KW, Willman CL, Issa JP. hepatocellular carcinoma. J Natl Cancer Inst 2002; 94: 755–761. Methylation profiling in acute myeloid leukemia. Blood 2001; 97: 40 Taniguchi T, Chikatsu N, Takahashi S, Fujita A, Uchimaru K, 2823–2829. Asano S et al. Expression of p16-INK4A and p14-ARF in 23 Garcia-Manero G, Daniel J, Smith TL, Kornblau SM, Lee M-S, hematological malignancies. Leukemia 1999; 13: 1760–1769. Kantarjian HM et al. DNA methylation of multiple promoter- 41 Hasegawa M, Nelson HH, Peters E, Ringstrom E, Posner M, Kelsey associated CpG islands in adult acute lymphoblastic leukemia. K. Patterns of gene promoter methylation in squamous cell cancer Clin Cancer Res 2002; 8: 2217–2224. of the head and neck. Oncogene 2002; 21: 4231–4236. 24 Garcia-Manero G, Bueso-Ramos C, Daniel J, Williamson J, 42 Liu M, Li R, Hayashi Y, Zhu G, Guo S. Study on abnormal Kantarjian HM, Issa JP. DNA methylation patterns at relapse in expression of the p73 gene in childhood acute lympho- adult acute lymphocytic leukemia. Clin Cancer Res 2002; 8: blastic leukemia. Zhonghua Xue Ye Xue Za Zhi 2002; 23: 1897–1903. 239–242. 25 Roman-Gomez J, Castillejo JA, Jimenez A, Gonzalez MG, Moreno 43 Evron E, Dooley WC, Umbricht CB, Rosenthal D, Sacchi N, F, Rodriguez MC et al. CpG island hypermethylation is associated Gabrielson E et al. Detection of breast cancer cells in ductal with transcriptional silencing of the p21(CIP1/WAF1/SDI1) gene lavage fluid by methylation-specific PCR. Lancet 2001; 357: and confers poor prognosis in acute lymphoblastic leukemia. 1335–1336. Blood 2002; 99: 2291–2296. 44 Wong TS, Chang HW, Tang KC, Wei WI, Kwong DLW, Sham JST 26 Esteller M, Gaidano G, Goodman SN, Zagonel V, Capello D, Botto et al. High frequency of promoter hypermethylation of the death- B et al. Hypermethylation of the DNA repair gene O6-methyl- associated protein kinase gene in nasopharyngeal carcinoma and guanine DNA methyltransferase and survival of patients with its detection in the peripheral blood of patients. Clin Cancer Res diffuse large B-cell lymphoma. J Natl Cancer Inst 2002; 94: 26–32. 2002; 8: 433–437. 27 Nosaka K, Maeda M, Tamiya S, Sakai T, Mitsuya H, Matsuoka M. 45 Chan MWY, Chan LW, Tang NLS, Tong JHM, Lo KW, Lee TL et al. Increasing methylation of the CDKN2A gene is associated with Hypermethylation of multiple genes in tumor tissues and voided the progression of adult T-cell leukemia. Cancer Res 2000; 60: urine in urinary bladder cancer patients. Clin Cancer Res 2002; 8: 1043–1048. 464–470.

Leukemia