Aberrant DNA methylation in pediatric patients with acute lymphocytic leukemia




Aberrant methylation of promoter-associated cystosine-guanine (CpG) islands is an epigenetic modification of DNA frequently observed in adult patients with acute lymphocytic leukemia (ALL). This epigenetic modification has been associated with gene silencing, malignant transformation, and aging. It is not known whether there are epigenetic differences between pediatric patients and adult patients with ALL.


To investigate the methylation characteristics of pediatric patients with ALL and to determine whether DNA methylation can explain prognostic or biologic differences between pediatric and adult patients, the authors analyzed the methylation status of 7 promoter-associated CpG islands in 16 pediatric patients with ALL and compared them with the methylation characteristics of a cohort of adult patients with ALL. The genes analyzed included the estrogen receptor gene (ER), multidrug resistance gene 1 (MDR1), p15, C-ABL, CD10, p16, and p73.


The mean methylation densities of ER, MDR1, CD10, p15, and C-ABL were 25.4%, 16.4%, 5.23%, 4.24%, and 4%, respectively. P16 was methylated in 11.7% of patients, and p73 was methylated in 17.6% of patients. One patient (6.2%) had methylation of 0 genes, 15 patients (93.7%) had methylation of ≥ 1 gene, and 4 patients (25%) had methylation of 3–4 genes. Methylation of all these genes was < 2% (or methylation specific polymerase chain reaction negative) in nonneoplastic tissues. A significant inverse correlation was observed between methylation of CD10 and CD10 expression. No differences were observed between the methylation characteristics of pediatric patients and adult patients.


The results indicate that DNA methylation is common in pediatric patients with ALL and that methylation of the genes studied does not account for prognostic differences between pediatric patients and adult patients with ALL. Cancer 2003;97:695–702. © 2003 American Cancer Society.

DOI 10.1002/cncr.11090

Acute lymphocytic leukemia (ALL) encompasses a heterogeneous group of lymphoid malignancies.1 The prognosis of patients with ALL depends on the age of the patient, cytogenetics, and immunophenotype. The prognosis of pediatric patients with ALL is much better compared with the prognosis of adult patients, with long-term survival rates of > 80%1 compared with adult patients, for whom the survival rate drops to 35–45%.2 There are genotypic differences in ALL between different age groups. The presence of the Philadelphia (Ph) chromosome, a feature that indicates a poor prognosis, is more frequent in adult and adolescent patients with ALL compared with in pediatric patients.3 Children age < 1 year who also have a poor prognosis frequently have abnormalities of the MLL gene.4 Hyperdiploidy and the TEL-AML1 abnormality both associated with a better outcome) are observed more frequently in children between the ages of 1 year and 9 years.5 In contrast, it is not known whether there are epigenetic differences between the two age groups.

DNA methylation is the addition of a methyl group to the cytosine in a cytosine-guanine (CpG) pair. Methylation of CpG islands in gene promoter regions has been associated with gene silencing and is a physiologic phenomenon observed in X-linked and imprinted genes.6 In contrast, aberrant methylation of tumor suppressor genes is observed frequently in human malignancies,7 including acute leukemias.8–12 Methylation of certain subsets of genes increases with age in neoplastic and nonneoplastic tissues, suggesting that DNA methylation may play a role in aging and age-related disorders.13 To study the methylation characteristics of pediatric patients with ALL, we analyzed the methylation status of 7 promoter-associated CpG islands in 16 patients.



Sixteen pediatric patients with ALL who were treated at the University of Texas M. D. Anderson Cancer Center (UTMDACC) were included in this study. Patients were selected based on sample availability. All patients were diagnosed in 1994 and 1995 and were treated on the Children's Cancer Group (CCG) 1922 or CCG 1882 trials for patients with standard and high-risk ALL, respectively. Samples consisted of bone marrow aspiration clots that were obtained at the time of initial diagnosis prior to chemotherapy administration. The percentage of bone marrow blasts was > 90% in all patients. No patients were positive for the Ph chromosome.

The adult cohort consisted of a group of 61 patients with Ph negative ALL who we studied previously.11 Adult specimens consisted of bone marrow aspiration samples that were obtained at initial diagnosis and were stored in a tumor bank. All patients had been treated with the hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasome chemotherapy regimen at UTMDACC.14 Consent for sample collection was obtained following institutional guidelines.

Bisulfite Modification of DNA

Information describing methods for bisulfite modification of DNA and polymerase chain reaction (PCR) techniques used in this study can be found at http://www.mdanderson.org/leukemia/methylation. Bisulfite treatment of genomic DNA was performed as described previously.15 DNA was extracted using standard phenol-chloroform methods after samples had been deparaffinized using xylene and ethanol followed by digestion with proteinase K. After extraction, 2 μg of DNA were used for bisulfite treatment. DNA was denatured in 0.2 N NaOH at 37 °C for 10 minutes and was incubated with 3 M Na-bisulfite at 50 °C for 16 hours. DNA was then purified using the Wizard cleanup system (Promega, Madison, WI) and desulfonated with 0.3 N NaOH at 25 °C for 5 minutes. DNA was then precipitated with ammonium acetate and ethanol, washed with 70% ethanol, dried, and resuspended in H2O.

Primer Design

Primers sequences, coordinates, Genebank accession numbers, the number of expected restriction fragments, and PCR conditions are shown in Table 1. The genes that were analyzed had been studied previously in a cohort of adult patients.11 Different criteria were used for gene selection. It is known that ER,16p15,17, 18p16,19 and p7320, 21 are methylated in ALL. It also is known that MDR1 is methylated and silenced in leukemic hematopoietic cells.22C-ABL is methylated in Ph positive chronic myelogenous leukemia and in Ph positive ALL.11, 23 It has been shown that CD10 is methylated and silenced in prostate carcinoma cells24 and frequently is expressed in ALL.

Table 1. Genes Studied
Gene accession no.Primers (coordinates)Restriction enzyme (no. of sites)Annealing temperatures (no. of cycles)
  1. ER: estrogen receptor; MDR1: multidrug resistance gene 1.

ER (X03635)GGTTTTTGAGTTTTTTGTTTTG (300) and AACTTACTACTATCCAAATACACCTC (505)BstUI (1)60 (3), 57 (4), 54 (5), 51 (25)
MDR1 (AC002457)GTTATAGGAAGTTTGAGTTT (140829) and AAAAACTATCCCATAATAAC (141001)TaqI (1)54 (3), 51 (4), 48 (5), 45 (26)
CD10 (X79438)TTYGGTTTTAGTTTGGAGTT (3665) and CCCTTTAAACCTTTCTCCCT (3843)BstUI (1)58 (3), 56 (4), 54 (5), 52 (26)

For p73 and p16, we used methylation specific PCR (MSP), as described elsewhere.20, 25 For ER, MDR1, p15, C-ABL, and CD10, we used combined bisulfate restriction analysis (COBRA).26 To minimize overestimation of methylated alleles when using this method, the following points were used: 1) Primers were designed to contain a minimum number of CpG dinucleotides in their sequence to avoid biased amplification of methylated alleles. If primers contained CpG sites, then they were designed to amplify methylated and unmethylated equally (with a mixture of C or T used for the sense strand or G or A for antisense primers); 2) primers were designed to contain a maximum number of thymidines converted from cytosines to avoid amplification of the nonconverted genomic sequence; 3) amplification of genomic that was DNA not treated with bisulfite always was carried out to monitor lack of nonspecific amplification; 4) primers were designed to be within 300 base pairs of known transcription start sites; and 5) for each set of primers, we tested multiple restriction enzymes to confirm the methylation status and sequence of the fragment analyzed, and we performed mixing experiments (using methylated and unmethylated templates mixed at a known ratios) to exclude any potential amplification bias.

PCR and Restriction Digestion

PCR reactions were carried out in 50-μL reactions. In each reaction, 2 μL of bisulfite-treated DNA were used as well as 1.25 mM dNTP, 6.7 mM MgCl2, 5 μL PCR buffer, 1 nmol primers, and 1 U of Taq polymerase. All PCR reactions were performed using a hot start at 95 °C for 5 minutes. After amplification, except for p73 and p16, PCR products were digested with specific restriction enzymes that digest alleles that were methylated prior to bisulfite treatment; then, products were separated on nondenaturing polyacrylamide gels. Gels were stained with ethidium bromide. The proportion of methylated product versus unmethylated product (digested vs. undigested) was quantitated by densitometric analysis, determining the density of methylation. Densitometric analysis was performed using a BioRad Geldoc 2000 digital analyzer equipped with the Quantity One software (version 4.0.3). For p73 and p16, MSP analysis was used as described previously.20, 25

Statistical Analysis

DNA methylation was measured using two different assays, depending on the gene studied. COBRA allows the quantitative determination of the density of methylation, which is the ratio of methylated (restricted) versus unmethylated (unrestricted) PCR product.26 MSP, in turn, allows a qualitative determination of methylation. The techniques were chosen based on the availability of well-standardized and reproducible laboratory protocols. When using COBRA, a sample was considered methylated if the methylation density was > 10%. This cut-off point was chosen based on the distribution of methylation previously found in other disorders10 and the assumption that methylation density levels < 10% would not be associated with gene silencing. To investigate possible clinical-biologic characteristics and methylation, correlation coefficients were computed using Spearman rank correlation coefficients. The variables analyzed included individual gene methylation values and the patient characteristics (age, gender, white blood cell [WBC] count, hemoglobin, platelet counts, cytogenetics, and immunophenotype). The definitions of complete remission (CR) and disease free survival (DFS) have been described elsewhere.14 Estimated 5-year DFS and overall survival (OS) were based on the Kaplan–Meier method, and differences were tested using the log-rank test. To compare the methylation characteristics of pediatric patients and adult patients, when COBRA was used, we used a t test for independent samples. For p73 and p16, we used the Fisher exact test. The Statistica software program (1999 version) was used to perform these computations.


Patient Population

We analyzed the methylation status of 7 promoter-associated CpG islands from paraffin embedded bone marrow aspiration clots from 16 pediatric patients with ALL. Patients were treated subsequently on the CCG 1922 or CCG 1882 protocols, based on their risk category. Methylation characteristics were compared with those of 61 adult patients.11 All patients in both age groups were Ph negative. The clinical-pathologic features of these patients are summarized in Table 2.

Table 2. Patient Characteristics
CharacteristicNo. of patients (%)
Pediatric cohort (n = 16 patients)Adult cohort (n = 61 patients)
  1. WBC: white blood cell; DFS: disease free survival; OS: overall survival; CALLA: common acute lymphocytic leukemia antigen.

Age (yrs)  
Male gender10 (62.5)40 (65.5)
WBC count > 30 × 109/L 7 (43.8)22 (36)
 Diploid 4 (25.0)19 (31.1)
 T(8;14), t(8;2);t(8;22) 1 (6.0) 4 (6.5)
 6q−, 14q+ 0 (0.0) 2 (3.2)
 Insufficient metaphases 5 (31.25)10 (16.3)
 Hyperdiploid 4 (25.0) 3 (4.9)
 Hypodiploid 1 (6.0) 2 (3.2)
 11q23 1 (6.0) 0.(0.0)
 Mature B 0 (0.0) 5 (8.2)
 T 2 (12.5) 7 (11.4)
 T-CALLA 1 (6.25) 3 (4.9)
 Precursor B 2 (12.5) 4 (6.5)
 CALLA11 (68.75)31 (50.8)
 Null 0 (0.0) 5 (8.2)
Complete remission rate (%)10093.4
Five-yr DFS (%)8838
Five-yr OS (%)9347

Methylation Status of Seven Promoter-Associated CpG Islands in Pediatric Patients with ALL

We analyzed the methylation characteristics of ER, MDR1, p15, CD10, C-ABL, p73, and p16. The mean methylation density was 25.4% for ER, 16.4% for MDR1, 5.2% for CD10, 4.2% for p15, and 4% for C-ABL. Two samples (12.5%) had methylation of p16, and 3 patients (18.75%) had methylation of p73. If a methylation cut-off point of 10% was used to consider a sample as methylated, then 11 samples (68.7%) had methylation of ER, 8 samples (50%) had methylation of MDR1, 4 samples (25%) had methylation of p15, 2 samples (12.5%) had methylation of CD10, and 1 sample (6.2%) had methylation of C-ABL. For all of these genes, normal bone marrow showed no significant methylation (density > 2% or MSP negative; data not shown). Figure 1 shows representative examples of gene methylation. Table 3 summarizes the distribution of methylation densities. Methylation (> 2% or MSP positive) was lowest for C-ABL (1 sample) and highest for ER (12 samples).

Figure 1.

Examples of methylation analysis of multiple cytosine-guanine (CpG) islands in pediatric patients with acute lymphocytic leukemia. (A) Aberrant methylation was detected by bisulfite polymerase chain reaction (PCR) followed by restriction enzyme digestion (combined bisulfate restriction analysis [COBRA]).26 Unmethylated bands (top) and methylated bands (bottom; arrowheads) were quantitated by densitometry. (B) For p73 and p16 analysis, methylation specific PCR was used.20, 25 Two different PCR reactions were carried, one with primers specific for the methylated sequence (Meth) and one with primers for the unmethylated sequence (Unmeth). This technique does not allow quantitative analysis. A sample was considered methylated if a band was visualized in the methylated reaction. Appropriate controls (C) consisted of cell lines with known methylation of the gene studied. ER: estrogen receptor; MDR1: multidrug resistance gene 1.

Table 3. Distribution of Pediatric Methylation Densitiesa
GeneMethylation density: No. of patients (%)
< 2.0%2.0–9.9%10.0–49.9%> 50.0%
  • ER: estrogen receptor; MDR1: mutidrug resistance gene 1.

  • a

    The of number of patients are shown along with the range of methylation density as detected by combined bisulfate restriction analysis (COBRA). For p73 and p16, methylated specific polymerase chain reaction analysis was used only, allowing a quantitative estimation of methylation. Three patients (18.7%) were methylated at p73, and 2 patients (12.5%) were methylated at p16.

ER4 (25)1 (6.2)9 (56.2)2 (12.5)
MDR18 (50)0 (0.0)8 (50)0 (0.0)
P159 (56.2)3 (18.7)4 (25)0 (0.0)
CD1010 (62.5)4 (25)2 (12.5)0 (0.0)
C-ABL15 (93.7)0 (0.0)0 (0.0)1 (6.2)

Methylation of Multiple Promoter-Associated CpG Islands in Pediatric Patients with ALL

Using a methylation density cut-off point of 10% or considering MSP positive reactions as indicative of significant methylation, 1 sample (6.2%) had methylation of 0 genes, 5 samples had methylation of 1 gene (31%), 6 samples had methylation of 2 genes (38%), and 2 samples (13%) of had methylation of 3 or 4 genes. No samples had methylation of five or more genes. Figure 2 represents the methylation patterns of the pediatric and adult cohorts studied. If a methylation cut-off value of 15% was used instead of the 10% cut-off value, only 4 of 112 results (3.5%) were modified, which did not affect the conclusions.

Figure 2.

Methylation profile of the cohorts studied. The methylation status of 7 genes was analyzed in 16 pediatric patients with acute lymphoblastic leukemia (ALL (A) and in 61 adult patients with ALL (B). Each column represents a gene, and each row represents a patient. A sample was considered methylated for a particular gene if the density of methylation was ≥ 10% (for the estrogen receptor gene [ER], multidrug resistance gene 1 [MDR1], P15, C-ABL, and CD10) or if a band was present in the methylation specific polymerase chain reaction analysis for p73 and p16 (black boxes). White boxes indicate nonmethylated samples.

Clinical-Biologic Associations of DNA Methylation in Pediatric Patients with ALL

To analyze possible correlations between methylation and clinical-biologic characteristics, we analyzed the correlations between the degree of methylation of a single gene or group of genes and the following patient characteristics: age, gender, WBC count, hemoglobin and platelets at presentation, cytogenetics, immunophenotype, CR, DFS, and OS. The only significant association found was an inverse correlation between the level of CD10 expression (as determined by flow cytometry) and the level of CD10 methylation (correlation coefficient = − 0.56; P = 0.000). The two samples that had significant CD10 methylation had very low levels of CD10 expression (0% and 1%). No correlation was found between the methylation of any gene or genes and CR, DFS, OS, cytogenetics, immunophenotype, risk category, or any of the other characteristics that were included in the analysis.

Comparison between Pediatric and Adult Methylation Characteristics

To determine whether the methylation characteristics of the pediatric group differed from the methylation patterns of adult patients with ALL, we compared the pediatric group with an adult cohort of patients.11 Because cytogenetic abnormalities may influence the methylation characteristics of the patients studied, we excluded Ph positive patients from the adult cohort. Adult patients were considered methylated using the same criteria that were used for the pediatric group. MDR1 was methylated in 34 samples (55%), ER was methylated in 32 samples (52%), p15 was methylated in 19 samples (31.1%), p73 was methylated in 15 samples (24.5%), CD10 was methylated in 9 samples (14.7%), p16 was methylated in 2 samples (3.2%), and C-ABL was methylated in 1 sample (1.6%). Nine samples (14.7%) had methylation of 0 genes, 20 samples (32.7%) had methylation of 1 gene, 14 samples (22.9%) had methylation of 2 genes, 9 samples (14.7%) had methylation of 3 genes, 8 samples (13.1%) had methylation of 4 genes, 0 samples had methylation of 5 genes, and 1 sample (1.6%) had methylation of 6 genes. No sample had methylation of more than 6 genes. No statistically significant differences were observed in terms of gene specific methylation or methylation patterns between both groups. Using a cut-off value of 15%, only 15 of 427 results (3.5%) were affected. A summary of pediatric and adult gene specific mean methylation results is shown in Table 4. No differences were observed between the two age groups.

Table 4. Methylation Characteristics of the Genes Studied
GeneMean (95%CI)Pediatric rangeAdult rangeP value
Pediatric cohortAdult cohort
  • 95% CI: 95% confidence interval; ER: estrogen receptor;; MDR1: multidrug resistance gene 1.

  • a

    For p73 and for the pediatric analysis of p16 methylation, methylation specific polymerase chain reaction analysis (a nonquantitative technique) was used. The Fisher exact test was used to compare p73 values. To compare p16 values, an adult sample was considered positive for methylation if the density was > 5%. Using this cut-off value, two samples were positive.

ER25.39 (13–37)18.2 (12–24)0.0–65.00.0–41.150.26
MDR116.43 (7–26)30.06 (22–38)0.0–46.70.0–100.00.1
P154.24 (1–7)11.88 (7–18)0.0–17.40.0–89.80.12
c-abl4.05 (−4–13)0.68 (−0.3–2.0)0.0–64.90.0–95.540.65
Cd105.23 (−0.7–11.2)5.92 (2–10)0.0–41.970.0–65.380.65
P16a2.0 (11.7%)1.46 (−0.4–3.0)0.0–52.130.13
P73a3.0 (17.6%)15.0 (24.5%)0.62


In this study, we analyzed the methylation characteristics of 16 pediatric patients with ALL and compared them with the methylation pattern observed in a cohort of adult patients. Our results indicate that the methylation characteristics of pediatric patients with ALL are not different from those of the older group; therefore, prognostic differences cannot be accounted for based on differences of DNA methylation.

To our knowledge, this is the first comprehensive analysis of methylation of multiple CpG islands in pediatric patients with ALL and the first of such studies to compare different age groups. There are several limitations to our study. The first limitation relates to the number of patients. By studying a larger number of patients, we may have found differences that were not observed in this analysis between specific subgroups of patients with distinct clinical-biologic characteristics and methylation. The second limitation relates to the number of genes studied. By studying a larger number of genes, we may have generated a more comprehensive methylation profile, and we may have found other correlations between methylation and other patient characteristics that were not found in the current study. Third, DNA methylation of promoter CpG islands has been associated with gene silencing. In this study, we were unable to establish a direct correlation of methylation with gene expression. This was due to the source of DNA (paraffin embedded tissues). Previous studies have shown that, for all of the genes that were studied for the current report, methylation is associated with gene silencing.16, 17, 20, 24 The strong inverse association between methylation of CD10 and its expression found in this study serves as a marker to indicate that methylation results in gene silencing, at least for the CD10 gene.

The genes studied for this report were selected based on the results observed in the adult cohort.11 It has been shown that ER is methylated frequently in patients with leukemia, including pediatric patients with ALL,16 and it has been associated with a better prognosis in adult patients with acute myelogenous leukemia (AML).27 The current results confirm that ER is methylated frequently in pediatric patients with ALL. MDR1 encodes for P-glycoprotein, a chemotherapy efflux pump involved in chemoresistance. It has been found that MDR1 is methylated in several human malignancies,22 including ALL both at the time of initial presentation11 and at the time patients develop recurrent disease.12 We have found that MDR1 methylation correlates with MDR1 protein expression in adult patients with ALL.12 It is interesting to note that, in this series, 50% of patients had no methylation of MDR1, suggesting that MDR1 may be expressed in a significant subset of patients. P73 is a member of the p53 family28 that frequently is lost due to hypermethylation in patients with leukemias and lymphomas, including ALL.20, 21 The current results indicate that this phenomenon is not age-related. The expression of CD10 is a very common immunophenotypic feature of ALL blast cells. It is known that the promoter-associated CpG island of CD10 is methylated in nonhematopoietic tissues that do not express this antigen.29 Like adult patients, we found an inverse correlation between CD10 expression and methylation,11 indicating that methylation of the promoter is involved in CD10 silencing. Methylation of cell cycle regulatory genes, particularly p15, is frequent in leukemias. It has been proposed that methylation of p15 and the lack of p16 characterized these disorders,25 although it has been reported that p16 methylation is a frequent event in childhood ALL associated with abnormalities of the MLL gene.19 The specimen analyzed in the current study that showed cytogenetic abnormalities involving 11q23 had no methylation of p16. Methylation of p15 has been associated with a poor prognosis and abnormal cytogenetics in children with acute leukemias,18 a finding that was not observed in our study of adult patients,11 perhaps due to the patient population or to the techniques used to assess p15 methylation. Methylation of C-ABL has been observed in patients with chronic myelogenous leukemia30, 31 and in patients with Ph positive leukemias of the p210BCR-ABL variant.11, 23 In the current series of patients with Ph negative disease, one patient with complex cytogenetics had methylation of C-ABL, a phenomenon that we have observed in older patients.11

Several epigenetic alterations have been implicated in the etiology of aging,32 including age-dependent hypermethylation of promoter-associated CpG islands. Genes for which this relation has been observed include ER,33 insulin-like growth factor 2 (IGF2),34 and MyoD.35 The causes of age-related methylation are unknown, and its relation with oncogenesis is not understood fully. it is important to note that all of these observations were made in epithelial tissues. This study found no methylation differences between younger patients and older patients with ALL. It has been shown that abnormal hypermethylation of multiple promoter CpG islands also is frequent in younger patients with AML.10 These findings indicate that aberrant methylation may not have a role in the aging of nonepithelial tissues, although, to our knowledge, studies of DNA methylation in normal hematopoietic tissues from patients in different age groups have not been performed.

In summary, this is the first study to analyze the methylation characteristics of multiple genes in pediatric patients with ALL. The results indicate that the methylation characteristics of the individual genes studied and the methylation patterns observed are not different compared with the patterns observed in adult patients, suggesting that methylation abnormalities do not account for biologic or clinical difference between pediatric and adult patients with ALL.


The authors are grateful to Lisa McDonald, Mark Brandt, and Sherry Pierce for clinical data management and to Laura Sasse for editorial assistance.