Approximately 20% of stage 4 high-risk neuroblastoma patients are alive and disease-free 5 years after disease onset while the remaining experience rapid and fatal progression. Numerous findings underline the prognostic role of methylation of defined target genes in neuroblastoma without taking into account the clinical and biological heterogeneity of this disease. In this report we have investigated the methylation of the PCDHB cluster, the most informative member of the “Methylator Phenotype” in neuroblastoma, hypothesizing that if this epigenetic mark can predict overall and progression free survival in high-risk stage 4 neuroblastoma, it could be utilized to improve the risk stratification of the patients, alone or in conjunction with the previously identified methylation of the SFN gene (14.3.3sigma) that can accurately predict outcome in these patients. We have utilized univariate and multivariate models to compare the prognostic power of PCDHB methylation in terms of overall and progression free survival, quantitatively determined by pyrosequencing, with that of other markers utilized for the patients' stratification utilizing methylation thresholds calculated on neuroblastoma at stage 1–4 and only on stage 4, high-risk patients. Our results indicate that PCDHB accurately distinguishes between high- and intermediate/low risk stage 4 neuroblastoma in agreement with the established risk stratification criteria. However PCDHB cannot predict outcome in the subgroup of stage 4 patients at high-risk whereas methylation levels of SFN are suggestive of a “methylation gradient” associated with tumor aggressiveness as suggested by the finding of a higher threshold that defines a subset of patients with an extremely severe disease (OS <24 months). Because of the heterogeneity of neuroblastoma we believe that clinically relevant methylation markers should be selected and tested on homogeneous groups of patients rather than on patients at all stages.
Citation: Banelli B, Merlo DF, Allemanni G, Forlani A, Romani M (2013) Clinical Potentials of Methylator Phenotype in Stage 4 High-Risk Neuroblastoma: An Open Challenge. PLoS ONE 8(5): e63253. https://doi.org/10.1371/journal.pone.0063253
Editor: Hiromu Suzuki, Sapporo Medical University, Japan
Received: November 30, 2012; Accepted: March 31, 2013; Published: May 22, 2013
Copyright: © 2013 Banelli et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the Italian Ministry of Health, core grant “Ricerca Corrente” to the IRCCS AOU San Martino-IST. B.B. is the recipient of the “Young Investigators” grant GR-2008-1143408 from the Italian Ministry of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Neuroblastoma (NB), a neoplasia derived from ganglionic precursors of the sympathetic nervous system, is the most common extra cranial solid tumor of infancy. This tumor is highly heterogeneous and its clinical behavior ranges from spontaneous regression to rapid and aggressive progression, metastatic spreading and poor outcome . About half of the children with malignant NB have metastatic disease at diagnosis and, according to the International Neuroblastoma Staging System (INSS), are classified at stage 4 . The guidelines set for by the International Neuroblastoma Risk Group (INRG) subdivide these patients into three categories (high, intermediate and low risk) depending on clinical and biological criteria . This risk stratification is part of the clinical decisional tree and helps to choose the most suitable treatment.
In stage 4, high-risk NB approximately 80% of the patients experience rapid and fatal disease while the remaining subgroup performs well with most of the patients alive and free of disease 5 years after diagnosis . This dramatic difference suggests that the prediction of outcome for these patients is still inaccurate. In perspective, the precise classification of NB patients into risk classes according to clinical and molecular parameters is of the utmost importance for the selection of the most effective treatment. In this respect, NB is one of the first tumors where a molecular marker, the amplification of the MYCN oncogene, has been utilized to choose the optimal therapeutic protocol , . Nevertheless the identification of predictive biomarkers in NB is made difficult because of its clinical and biological heterogeneity.
The aberrant DNA methylation is considered a promising biomarker of outcome or response to treatment, and the potential clinical application of methylation analysis in cancer is actively investigated , , . The aberrant and concordant methylation at multiple loci, known as CpG Island Methylator Phenotype (CIMP), was initially described in colorectal tumors , and is considered a potential predictive biomarker in cancer. In neuroblastoma, CIMP was originally associated to clinically distinct subgroups of patients by quantitative methylation analysis conducted on a series of patients at stages 1–4, assigned at different INRG risk groups and with different clinical and biological features , .
In univariate analyses CIMP had a strong predictive power on outcome and its prognostic power was entirely recapitulated by the methylation of 17 genes of the Protocadherin B cluster (PCDHB) that is the most informative member of the Methylator Phenotype in this tumor . However, when other biomarkers commonly used in the clinical practice were included in a multivariate model, CIMP lost its prognostic power likely because the overall number of patients examined was insufficient or because of the known heterogeneity of NB .
In a subsequent independent study, in a multivariate analysis that included age at diagnosis, stage and MYCN amplification, CIMP was found to have a strong influence on disease-free survival, but not on overall survival , .
In view of a possible translational application of methylation of PCDHB cluster to improve risk stratification in stage 4 NB patients at high risk, we have evaluated the predictive power of PCDHB methylation by quantitative analysis in stage 4 NB at high risk, the most common mode of presentation of this disease which represents a clinically relevant problem in terms of accurate patients stratification and improvement of outcome.
For comparison within the metastatic stage 4, we have included a group of stage 4 patients at intermediate/low risk of progression. In retrospective studies, we identified the SFN gene (14.3.3sigma) as a methylation target in aggressive NB ,  and found that quantitative methylation differences in SFN discriminated high-risk stage 4 patients with poor survival from those, at the same stage and in the same risk group, with favorable outcome independently from MYCN amplification, treatment, clinical response, histology and age at diagnosis .
We have conducted the present study in the same series of patients previously utilized for the SFN analysis considering all biological and clinical features currently used for patients' stratification including MYCN amplification.
The rationale of our work was to determine if PCDHB, alone or in conjunction with SFN could improve the risk stratification in these patients as a preliminary step in order to select members of a classifier predictive of outcome in stage 4, high-risk NB.
The Ethics Committee of the Giannina Gaslini Children Hospital of Genoa approved the collection, the storage in the Neuroblastoma Tissue Bank and the utilization of this material. Written informed consent was obtained for all patients from their parents or legal representative.
Planning of the study
In the present study we have analyzed a total of 121 NB patients at stage 4 diagnosed between 1990 and 2004. The clinical endpoints examined were the overall survival (OS) at 60 months and the progression free survival (PFS) in relation with the level of methylation of the genes examined.
To stratify the patients into risk groups we have utilized the criteria of the International Neuroblastoma Risk Group (INRG) Classification System . Accordingly, 106 stage 4 patients were considered at “high risk”: 100 patients were older than 18 months, and 6 were younger than 18 months at diagnosis but their tumor presented MYCN amplification. Within the stage 4 high-risk group we considered the patients as “short survivors” (HR-SS) if they died for disease within 60 months (N = 83) from diagnosis and as “long survivors” (HR-LS) if they survived more than 60 months (N = 23). We conducted the methylation analysis on high-risk patients first on a training set of 41 patients and then on a validation set of 65 patients.
As control group we have included 15 stage 4 patients at intermediate and low risk (I/LR: below 18 months of age at diagnosis, MYCN-single copy), all alive and free of disease 60 months after diagnosis.
The clinical characteristics of the study groups are reported in Table 1. The patients examined for the present study are the same described in a previous report  with the exception of 17 patients whose tumor samples were no longer available.
We retrieved the tumor DNA from the Italian Neuroblastoma Tissue bank . DNA (1 μg) was modified by sodium bisulfite treatment and the level of methylation for SFN and PCDHB was determined by pyrosequencing, a sequence-by-synthesis technique that allows the quantitative determination of the level of methylation of each CpG doublets within a target sequence . The primers were designed with the Pyrosequencing Assay Design Software (Qiagen, Milano, Italy).
The pyrosequencing methylation assays for PCDHB cluster and SFN have been previously described in details , . We have conducted the sequencing analysis utilizing the Pyro Q-CpG software (version 1.0.9) and the results of the pyrosequencing passed the quality controls built in the instrument. Blank reactions (for PCR and pyrosequencing) have been included in each assay to exclude cross contaminations. The specificity of the primers was determined in PCR reactions conducted on unmodified DNA to ensure that only the modified DNA was amplified.
The mean methylation values of the CpG doublets included in the target sequences were measured for PCDHB cluster and these values were considered for the statistical analysis.
The correlation of the percentage of methylation between SFN and the PCDHB cluster was assessed by computing the Pearson's linear correlation coefficient.
The statistical differences in the methylation levels were determined by the Student t-Test.
We defined overall survival as the time elapsed from the date of diagnosis and the date of death. Only cancer-related deaths were considered. Patients that survived were censored at the last date they were reported to be alive. Progression free survival was calculated from the date of diagnosis to the date of relapse, as reported in the clinical records.
For survival analysis, besides the already defined levels of methylation thresholds for SFN (85%)  and for the PCDHB cluster (40%) , we determined, according to the Receiver Operating Characteristic (ROC) curves, a new methylation threshold for PCDHB (58.3%) specific for high risk stage 4 patients that could distinguish between long and short survivors. Survival curves were computed according to the Kaplan-Meier method and were compared using the log-rank test.
The Cox proportional-hazards multiple regression model was used to study the relation between DNA methylation of SFN and PCDHB and the overall and progression free survival. We included in the regression model the following factors known as clinically relevant in the INGR classification of stage 4 high risk patients: age at diagnosis (<18 months or ≥18 months), MYCN amplification (single copy or amplified) and ferritin serum level (<92 or ≥92 ng/ml).
A p value below 0.05 was considered statistically significant. Statistical analyses were done with the IBM® SPSS® Statistics 20 software.
We have analyzed by quantitative pyrosequencing the methylation of the PCDHB cluster in 121 NB samples derived from patients at stage 4 divided into training and validation sets of high-risk patients and in a control set of intermediate/low risk patients. We have matched these data with a previous SFN methylation analysis conducted on the same patients . The clinical characteristics of the patients are summarized in Table 1 and the complete clinical and biological data are shown in Table S1.
Similarly to what we have observed for the SFN gene , the overall distribution of the PCDHB methylation levels observed in our series of high risk stage 4 patients did not follow the bimodal distribution described by Abe et al  for a population of patients with neuroblastoma representative of all stages of the disease ( Figure S1).
We have evaluated the correlation between methylation levels of the SFN gene and of the PCDHB cluster by computing the Pearson's linear correlation coefficient in high risk (HR) and intermediate/low risk (I/LR) patients and subdividing the high risk patients in high risk-long and -short survivors groups (HR-LS and HR-SS, respectively) as defined in methods section. The result of this analysis are presented in Figure S2 and show that methylation levels of PCDHB and SFN were poorly correlated.
The comparison between the high risk (training and validation sets) and intermediate/low risk stage 4 patients showed that the mean methylation levels of the PCDHB cluster were significantly different between the two risk groups (Figure 1; PCDHB mean values: HR-T = 56.8, HR-V = 60.4 I/LR = 41.93, p<0.001). This result indicated that the methylation of the PCDHB cluster, as also observed also for SFN gene , was associated with risk stratification in stage 4 patients.
The lower and upper boundaries of each box are the 25th and 75th percentile. Black bars represent median values; wiskers are the smallest and largest values that are not outliers (defined as larger than 1.5 and smaller than 3 box lengths from 25th and 75th percentiles). The black star is the upper outlier.
The relation between DNA methylation and MYCN amplification in NB is controversial and it is still an open question , , , . According to our previously published data, hypermethylation of SFN in stage 4 high risk NB patients was independent from MYCN amplification . In a large series of NB patients at stages 1–4, methylation of the PCDHB cluster was found to be related to MYCN amplification; this association remained true also in our series of stage 4 high-risk patients (Figure 2, p<0.001 and p = 0.019 in training and validation sets, respectively).
It has been previously shown that the methylation threshold of 40% for the PCDHB cluster, determined by MSqPCR assay, predicts overall survival and progression free survival in NB patients at stages 1–4 , . In a previous report we described a methylation pyrosequencing assay for PCDHB cluster and showed that it provided similar results to MSqPCR when tested on a set of patients selected to be at the extreme ends of the INRG classification system in terms of outcome and disease progression . Interestingly, the threshold determined by pyrosequencing was essentially identical to that found by MSqPCR on a distinct patient series  (39.15% versus 40% respectively).
The Kaplan-Meier plot for high-risk stage 4 patients categorized into two groups according to the 40% threshold of methylation for PCDHB, showed no significant association with progression free survival and overall survival (Figure S3).
We have previously shown that in NB, higher methylation levels were associated with higher aggressiveness . In agreement with this finding, we observed that in high-risk, stage 4 patients that survived less than 24 months, the methylation level of SFN was significantly higher than that of the patients that died between 25 and 60 months (Figure S4).
Accordingly, a second threshold of methylation for the SFN gene (90%) identified the patients with a median survival halved from 36 to 18 months compared to patients with methylation comprised between 85 and 90% (Figure 3 and Table S2). Thus, methylation levels above 90% appeared to characterize a subset of patients with extremely aggressive disease and were suggestive of a linear trend of SFN methylation associated with poor outcome in NB at high risk (Log-rank test for trend P<0.0001 for OS and PFS; the same results were obtained also analyzing the training and validation set separately even if the size of the two groups was relatively small, data not shown).
Hazard Ratio and p values are reported in Table S2.
To determine if methylation of PCDHB followed a similar trend we calculated by ROC analysis a second threshold of methylation for the PCDHB cluster predictive of outcome taking into consideration only the high-risk patients at stage 4. The resulting thresholds (58.3%) were the best compromise between specificity (80%) and sensitivity (58%). According to this new threshold, methylation of the PCDHB cluster in the training set showed a significant association with progression free and overall survival (p = 0.0269 HR = 2.09 and p = 0.0148 HR = 2.33 respectively). However, this association was not maintained in the validation set (OS: p = 0.5684 HR = 0.86; PFS: p = 0.6509 HR = 0.88) (Figure 4).
The Hazard Ratio (HR) and the corresponding p values (Cox Long-Rank test) are reported.
Finally, the possible prognostic power of PCDHB cluster methylation was compared to that of the SFN gene in a multivariate model to take into account the clinically relevant parameters considered for the risk stratification of stage 4 NB patients . In this stepwise analysis the methylation of PCDHB cluster was excluded by the model before the MYCN amplification and ferritin serum level while SFN remained a statistically significant parameters to predict survival (Table 2).
The initial characterization of the methylator phenotype in NB was carried out in a series of patients representative of the entire spectrum of this disease and thus at all INSS stages and assigned at different risk groups . The heterogeneity in this patients' series made difficult to reach a sufficiently large number of cases to be analyzed in a robust multivariate model that included the biomarkers utilized in the clinical practice. From this pioneering study, the methylation of the PCDHB cluster above the established threshold of 40% was correlated with the reduced patients' survival and was considered the most informative member of CIMP in NB , , . Interestingly, lower levels of PCDHB methylation are associated with a less aggressive behavior not only in NB but also in Wilm's and breast cancer ,  indicating that this biomarker could be relevant in many tumor types.
In a subsequent neuroblastoma study, CIMP was proven to be a predictor of progression-free survival superior to MYCN amplification, stage and age at diagnosis . In a independent report, utilizing a different technical approach, a similar level of PCDHB methylation distinguished patients at the opposite ends of the INRG classification system: the low-stage patients at favorable outcome and the stage 4 patients at unfavorable outcome .
Because of the known clinical and biological heterogeneity of neuroblastic tumors, we hypothesized that if stage-specific NB biomarkers are used, it could be possible to stratify more precisely the patients and identify subgroups of patients with distinct clinical characteristics. Indeed, we believe that the analysis of a homogeneous set of patients could give a clear answer about the potentiality of the methylation of PCDHB cluster to predict the outcome in the most aggressive group of neuroblastoma: the high risk at stage 4.
We demonstrated that the methylation threshold of 85% for SFN is a strong and independent predictor of outcome in high-risk, stage 4 patients . Along this line we have now determined the methylation level of PCDHB in the same clinical series in the attempt to define if the prognostic power of PCDHB and SFN methylation could be increased by combining the information derived from the two biomarkers and if they were detecting the same or distinct subgroups of patients.
The weak correlation between PCDHB and SFN methylation observed in our study immediately suggested that the two biomarkers identify partially overlapping but not identical patients subgroups. In stage 4, as already observed for SFN, the PCDHB cluster showed higher methylation in high-risk respect to the intermediate and low risk subgroups indicating that the level of methylation of these genes followed the risk stratification set by the INRG. Nevertheless, the methylation threshold of PCDHB (40%) previously defined as predictive of outcome in NB at stages 1–4 did not predict outcome when only stage 4 high-risk patients were considered.
CIMP in cancer is generally associated with a worse prognosis (reviewed in ) although examples of improved outcome were described , . In neuroblastoma, a higher number of methylated genes  or higher methylation levels in specific gene signatures ,  are predictive of poor prognosis.
In a study conducted on stage 4 high-risk patients, we demonstrated the strong association between the level of methylation of SFN and outcome . This association is now further strengthened by the observation that patients with methylation levels over 90% had a median survival time halved respect to patients with the methylation level between 85 and 90% (18 vs. 36 months).
In view of this result we hypothesized that the PCDHB methylation threshold predictive of outcome in high risk, stage 4 patients could be different from that calculated on the entire NB patients' population. Therefore, by ROC analysis we determined a second, higher, threshold of methylation only from stage 4 patients at high risk and showed that, in the training set, it could significantly distinguish two groups of patients according to their OS and PFS. However this threshold lost its predictive power when assayed on the validation set of patients.
Furthermore, a multivariate model that included the most relevant factors that predict outcome in this group of patients, underlined the predictive role of SFN. Overall these results indicate that, even if PCDHB methylation can correctly subdivide stage 4 patients in distinct risk groups, it is not a prognostic indicator in stage 4 high-risk neuroblastoma patients.
The conflicting results on the prognostic power of PCDHB cluster and of the SFN gene methylation in stage 4 high-risk patients probably reflect the different methodological approaches to examine these markers. SFN was selected through a candidate gene approach and validated in a retrospective study specifically focused on high risk stage 4 patients , , while the PCDHB cluster was identified by MS-RDA, a subtractive genome-wide analysis conducted between MYCN amplified NB cell lines and primary tumors with good prognosis . It is thus possible that this latter approach has favored the selection of genes methylated in MYCN amplified tumors. This could explain the strong relation between CIMP and MYCN amplification observed in this and in previous studies . In this respect the methylation of the CYP26C1 gene, another member of CIMP selected in the same study , although highly predictive of outcome in univariate analysis, was highly correlated to MYCN amplification in stage 4 high risk patients and was selected out in a multivariate model that includes MYCN amplification as confounder , . Similarly, PCDHB methylation in a stepwise Cox-regression model was the first marker to be excluded before MYCN amplification. Overall these results indicate that, MYCN amplification and CIMP, as defined in previous studies , , are linked phenomena and that the predictive power of CIMP observed in patients at stages 1–4, might be partially absorbed by MYCN amplification in high-risk stage 4.
Similarly to SFN , the distribution of PCDHB methylation values in high-risk stage 4 patients, was shifted toward higher levels of methylation. This result suggests that high-risk patients at stage 4 could be enclosed in the right end of the bimodal distribution described in a neuroblastoma population where all stages were represented . According to this hypothesis it is understandable why biomarkers selected from the neuroblastoma general populations could not reliably separate patients with divergent outcome in a high-risk subgroup .
Approximately 50% of the NB patients have metastatic disease at diagnosis and hence are classified at stage 4; the majority of them are at high risk of progression although the prediction of their outcome is still imprecise. Thus, patients in this group are those likely to benefit more from a marker predictive of prognosis to improve their risk stratification. According to our data, SFN methylation, differently from PCDHB, can identify patients at lower or much higher risk within this group of patients opening the possibility to design tailored therapies.
Many different methylation markers were found to predict with variable accuracy the outcome of NB patients. We believe that the transfer of these markers “from the bench to the bedside”, considering the great heterogeneity in neuroblastoma, could be more easily obtained focusing on more homogeneous groups of patients and carefully taking into account, in a multivariate approach, all the relevant clinical and biological characteristics of that particular set of patients.
Distribution of PCDHB cluster methylation in tumor samples from High Risk stage 4 patients (Long Survivors in red and Short Survivors in blue). Histograms represent the number of cases according to the percentage of PCDHB cluster methylation.
Correlation analysis between the mean methylation values of PCDHB and SFN in stage 4 NB patients subdivided according to risk class and survival.
Kaplan-Meier estimates of OS and PFS of the High-Risk patients assigned to groups according to the 40% thresholds of methylation for PCDHB (Training set: ≤40% N = 7; >40 N = 34). The Hazard Ratio (HR) and the corresponding p values (Cox Long-Rank test) are reported.
Distribution of methylation values for SFN in HR-SS patients subdivided in patients that died within 24 months and between 25 and 60 months from diagnosis. Black stars and black triangles are the upper and lower outliers, respectively.
Clinical characteristics of the patients and methylation analysis. For each patient are indicated the most relevant parameters as reported in the clinical records and the mean methylation values of the PCDHB cluster and of the SFN gene.
The authors thank the patients and their families for the constant co-operation and the physicians of the Giannina Gaslini Institute of Genova, Gian Paolo Tonini and Katia Mazzocco for providing clinical records and tumor samples from the neuroblastoma tissue bank.
Conceived and designed the experiments: BB MR. Performed the experiments: BB GA AF. Analyzed the data: BB MR DFM. Wrote the paper: BB MR DFM.
- 1. Maris JM, Hogarty MD, Bagatell R, Cohn SL (2007) Neuroblastoma. Lancet 369: 2106–2120.
- 2. Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV, et al. (1999) The International Neuroblastoma Pathology Classification (the Shimada system). Cancer 86: 364–372.
- 3. Cohn SL, Pearson AD, London WB, Monclair T, Ambros PF, et al. (2009) The International Neuroblastoma Risk Group (INRG) classification system: an INRG Task Force report. J Clin Oncol 27: 289–297.
- 4. Brodeur GM (2003) Neuroblastoma: biological insights into a clinical enigma. Nat Rev Cancer 3: 203–216.
- 5. Jones PA, Baylin SB (2002) The fundamental role of epigenetic events in cancer. Nat Rev Genet 3: 415–428.
- 6. Portela A, Esteller M (2010) Epigenetic modifications and human disease. Nat Biotechnol 28: 1057–1068.
- 7. Toyota M, Suzuki H, Yamashita T, Hirata K, Imai K, et al. (2009) Cancer epigenomics: implications of DNA methylation in personalized cancer therapy. Cancer Sci 100: 787–791.
- 8. Banelli B, Gelvi I, Di Vinci A, Scaruffi P, Casciano I, et al. (2005) Distinct CpG methylation profiles characterize different clinical groups of neuroblastic tumors. Oncogene 24: 5619–5628.
- 9. Banelli B, Di Vinci A, Gelvi I, Casciano I, Allemanni G, et al. (2005) DNA methylation in neuroblastic tumors. Cancer Lett 228: 37–41.
- 10. Banelli B, Bonassi S, Casciano I, Mazzocco K, Di Vinci A, et al. (2010) Outcome prediction and risk assessment by quantitative pyrosequencing methylation analysis of the SFN gene in advanced stage, high-risk, neuroblastic tumor patients. Int J Cancer 126: 656–668.
- 11. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, et al. (1999) CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 96: 8681–8686.
- 12. Issa JP (2004) CpG island methylator phenotype in cancer. Nat Rev Cancer 4: 988–993.
- 13. Teodoridis JM, Hardie C, Brown R (2008) CpG island methylator phenotype (CIMP) in cancer: causes and implications. Cancer Lett 268: 177–186.
- 14. Abe M, Ohira M, Kaneda A, Yagi Y, Yamamoto S, et al. (2005) CpG island methylator phenotype is a strong determinant of poor prognosis in neuroblastomas. Cancer Res 65: 828–834.
- 15. Abe M, Watanabe N, McDonell N, Takato T, Ohira M, et al. (2008) Identification of genes targeted by CpG island methylator phenotype in neuroblastomas, and their possible integrative involvement in poor prognosis. Oncology 74: 50–60.
- 16. Abe M, Westermann F, Nakagawara A, Takato T, Schwab M, et al. (2007) Marked and independent prognostic significance of the CpG island methylator phenotype in neuroblastomas. Cancer Lett 247: 253–258.
- 17. Tost J, Gut IG (2007) DNA methylation analysis by pyrosequencing. Nat Protoc 2: 2265–2275.
- 18. Banelli B, Brigati C, Di Vinci A, Casciano I, Forlani A, et al. (2012) A pyrosequencing assay for the quantitative methylation analysis of the PCDHB gene cluster, the major factor in neuroblastoma methylator phenotype. Lab Invest 92: 458–465.
- 19. Gonzalez-Gomez P, Bello MJ, Lomas J, Arjona D, Alonso ME, et al. (2003) Aberrant methylation of multiple genes in neuroblastic tumours. relationship with MYCN amplification and allelic status at 1p. Eur J Cancer 39: 1478–1485.
- 20. Hoebeeck J, Michels E, Pattyn F, Combaret V, Vermeulen J, et al. (2009) Aberrant methylation of candidate tumor suppressor genes in neuroblastoma. Cancer Lett 273: 336–346.
- 21. Dallosso AR, Hancock AL, Szemes M, Moorwood K, Chilukamarri L, et al. (2009) Frequent long-range epigenetic silencing of protocadherin gene clusters on chromosome 5q31 in Wilms' tumor. PLoS Genet 5: e1000745.
- 22. Novak P, Jensen T, Oshiro MM, Watts GS, Kim CJ, et al. (2008) Agglomerative epigenetic aberrations are a common event in human breast cancer. Cancer Res 68: 8616–8625.
- 23. Fang F, Turcan S, Rimner A, Kaufman A, Giri D, et al. (2011) Breast cancer methylomes establish an epigenomic foundation for metastasis. Sci Transl Med 3: 75ra25.
- 24. Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, et al. (2010) Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 17: 510–522.
- 25. Alaminos M, Davalos V, Cheung NK, Gerald WL, Esteller M (2004) Clustering of gene hypermethylation associated with clinical risk groups in neuroblastoma. J Natl Cancer Inst 96: 1208–1219.
- 26. Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, et al. (2010) Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol 28: 3506–3515.