Heyn2012 PDF

REVIEWS
T R A N S L AT I O N A L G E N E T I C S
DNA methylation profiling in the clinic:

applications and challenges
Holger Heyn1 and Manel Esteller1,2,3
Abstract | Knowledge of epigenetic alterations in disease is rapidly increasing owing to

the development of genome-wide techniques for their identification. The ever-growing
number of genes that show epigenetic alterations in disease emphasizes the crucial role
of these epigenetic alterations particularly DNA methylation for future diagnosis,
prognosis and prediction of response to therapies. This Review focuses on epigenetic
profiling, which has started to be of clinical value in cancer and may in the future be
extended to other diseases, such as neurological and autoimmune disorders.
CpG islands
CpG-rich regions of DNA that
are often associated with the
transcription start sites of
genes and that are also found
in gene bodies and intergenic
regions.
Cancer Epigenetics and

Biology Program (PEBC),
Bellvitge Biomedical
Research Institute (IDIBELL),
08908LHospitalet de
Llobregat, Barcelona,
Catalonia, Spain.
2
Department of Physiological
Sciences II, School of
Medicine, University of
Barcelona, 08036 Barcelona,
Catalonia, Spain.
3
Instituci Catalana de
Recerca i Estudis Avanats
(ICREA), 08010 Barcelona,
Catalonia, Spain.
Correspondence to M.E.
e-mail: mesteller@idibell.cat
doi:10.1038/nrg3270
Published online
4 September 2012
1
Over the past 25 years, an increasing amount of evidence has supported a role for epigenetic processes in
cell biology and tissue physiology. The potential role of
these epigenetic processes in human disease is exemplified by aberrant DNA methylation in cancer, both at
individual genes and on a genome-wide scale. Initially,
studies focused on candidate gene approaches to identify alterations occurring in distinct diseases; however,
new powerful technologies, such as comprehensive DNA
methylation microarrays1,2 and genome-wide bisulphite
sequencing3, have recently emerged (BOX1). These have
reinforced the notion of epigenetic disruption at least as
a signature of human diseases4.
An indication that alterations in epigenetic marks
could be used as biomarkers specifically DNA
methylation provided by the use of the hypermethylation events at the genes O6methylguanine-DNA
methyltransferase (MGMT)5 in glioma and glutathione
Stransferase pi 1 (GSTP1)6 in prostate cancer; these
hypermethylation events have been shown to be effective for determining an appropriate treatment choice in
glioma and in the diagnosis of prostate cancer, respectively. The detection of epigenetic alterations is therefore
a promising tool for the diagnosis and prognosis of disease and for the prediction of drug response.
This Review focuses on epigenetic profiling at the
level of DNA methylation7,8. There is an emphasis on
neoplasia, as this is the area in which using these epigenetic markers is best characterized, but many of these
concepts are applicable to other disorders, such as
neurological, metabolic and cardiovascular disorders
and autoimmune diseases. DNA methylation marking
repressed genes is a frequent event in development, differentiation and tissue homeostasis. Cancer cells present
a global loss of DNA methylation centred in hypomethylated blocks and regional hypermethylation, especially
in CpG-rich contexts8,9. For cancer, recent evidence has
given an indication of the altered DNA methylation
events beyond classical promoter hypermethylation of
CpG islands of tumour suppressor genes 10. The early
signs are that this knowledge will be useful in practical
clinical situations, such as in the diagnosis of cancers of
unknown primary origin, the screening for malignancies in high-risk populations or biomarker-based selection of the patients that should receive treatment with
epigenetic drugs. Furthermore, there is a growing list of
mutated chromatin remodeller genes that contribute to
the tumorigenesis process and that may be targeted with
drugs in the future for cancer treatment11 (BOX2). In the
future, studying chromatin structure may provide useful
insights for diagnosis and prognosis (but this requires
further research and is not discussed here).
This Review provides an insight into epigenetic
alterations implicated in diagnostics, prognostics and
prediction to drug response and their current use in
clinical settings. First, an overview of what makes a suitable DNA-methylation-based biomarker is presented,
including a consideration of the non-invasive tissues that
may be used to profile these. Then the applications of
DNA profiling in disease diagnosis followed by disease
prognosis and prediction to drug response are presented,
including a consideration of relevant examples at every
stage. Finally, future directions for the development of
relevant biomarkers and their applications are discussed.
DNA methylation profiling for disease diagnosis

The dynamic nature of DNA methylation. DNA methylation is a dynamic modification that is put in place and
NATURE REVIEWS | GENETICS
VOLUME 13 | O CTOBER 2012 | 679

2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
removed by a range of enzymes that may be targeted for
disease treatment (BOX2); this is reviewed elsewhere4,11,12.
Given their dynamic nature, epigenetic disease biomarker genes have to be determined by considering
Box 1 | The technology revolution in epigenetic profiling

Whole-genome bisulphite sequencing
Whole-genome bisulphite sequencing (WGBS) allows for an unbiased assessment of
the profile of DNA methylomes. Using bisulphite-treated DNA converting
unmethylated cytosines into thymidines82, next-generation sequencing (NGS)
technology is used to obtain a complete overview of CpG methylation level at base-pair
resolution. Originally applied in Arabidopsis thaliana83, WGBS has been used to analyse
DNA methylation variation between embryonic stem cells (ESCs)84,85, induced
pluripotent stem cells (iPSCs)86 and differentiated cells87. Subsequent studies have
revealed detailed information about healthy somatic tissues and associated cancer
methylomes for breast7 and colorectal cancers8,9.
Restriction enzyme-enriched sequencing techniques
Although less comprehensive than whole-genome bisulphite sequencing, reduced
representation bisulphite sequencing (RRBS) is an efficient technique that is suitable
for obtaining information from most CpG islands and information about sequences
outside CpG-rich regions88. This technique reduces the number of non-informative
reads by DNA digestion using a methylation-insensitive restriction enzyme (MspI). As
MspI contains a CpG site in its recognition sequence motif, each sequencing read
contains at least one informative position. RRBS has been used to identify changes
during cell differentiation and was also intensively applied in the Roadmap project89,
providing detailed information about ESCs and iPSCs90.
Restriction-enzyme-based comprehensive high-throughput arrays for relative DNA
methylation (CHARM) uses the methylation-dependent specificity of the methylationsensitive restriction enzyme McrBC with subsequent analysis on tiling arrays. Using
genome-weighted smoothing, CHARM allows quantitative analysis of genomic regions
rather than single base pairs91. CHARM technology was carried out to establish the
impact of CpG island shores on tumorigenesis of colorectal cancer92 but also for intense
studies of iPSCs9395 and the epithelialmesenchymal transition96.
Affinity-enrichment-based sequencing techniques
Methylated DNA binding domain sequencing (MBD-seq)97,98 and methylation DNA
immunoprecipitation sequencing (MeDIP-seq)99,100 combine the advantages of NGS and
enrichment of methylated regions by immunoprecipitation. Both techniques were
successfully applied to identify aberrantly methylated disease-related CpG sites101.
Precipitation-based techniques are suitable for covering large parts of the genome in a
quantitative manner, especially in combination with methylation-sensitive restriction
enzyme sequencing (MRE-seq)102,103. A detailed comparison of sequencing-based
technologies showed all methods to yield comparable methylation calls, although they
differed substantially in terms of resolution and costs3.
DNA methylation arrays
CpG-specific array technology is an alternative option for determining a genome-wide
DNA methylation profile. The latest HumanMethylation 450 beadchip assay (Illumina)
allows for the high-resolution, genome-wide DNA methylation profiling of human
samples to be carried out, covering 99% of all RefSeq genes and approximately
450,000CpGs overall1,2,104. The array analyses CpG islands, shores (the 2kb flanking
the CpG islands) and shelves (the 2kb flanking the shores) as well as island-independent
CpGs. Although focused on gene promoters and intragenic regions, there are also
additional sites of potential interest, such as DNaseI-sensitive and enhancer regions.
Compared with sequencing approaches, DNA methylation arrays are a low-cost
alternative, which allows the profiling of a large number of samples, although at a
reduced resolution.
Locus-specific DNA methylation analysis
In addition to genome-wide technologies, locus-specific identification of the DNA
methylation level is a cost-effective strategy, especially if single genes are already
established as biomarkers for diagnosis or prognosis. Sequencing-based technologies
such as direct bisulphite and pyrosequencing but also alternatives such as methyl
ation-specific-PCR-based and matrix-assisted laser desorption/ionization (MALDI)mass-spectrometry-based analysis105108 are suitable for reliably assessing DNA
methylation levels.
inter-individual and intra-individual variations. In this

respect, the definition of reference DNA methylation
data sets will be of a high value, facilitating biomarker
selection by initially defining suitable sites that is,
sites that show consistent levels of DNA methylation
in healthy individuals. Comprehensively studying the
profiles of different healthy individuals and tissue types
enables us to estimate the variance of a particular CpG
site or of regions such as promoters. Reference data sets
are now being created in consortia such as Blueprint, the
International Human Epigenome Consortium (IHEC)
and Roadmap using high-resolution technologies that are
summarized in BOX1 (REFS 13,14). Focusing on normal
tissue types, these joint efforts aim freely to release reference data sets of integrated epigenomic profiles of stem
cells, as well as developmental and somatic tissue types to
the research community. As a paradigm, the estimation of
genetic variance in the human population and the identification of SNPs greatly improved mutational analysis
for disease identification by excluding false-positive
hits for disease-linked mutations before screening.
Concordantly, filtering for loci that are unstable in
DNA methylation between individuals excludes unsuitable CpG sites before biomarker candidate selection
approaches. Systematic screening of reference data sets
obtained from different individuals will enable us to identify and to exclude variable CpG sites and regions, highly
facilitating future biomarker selections. Furthermore,
owing to tissue-specific DNA methylation, particular
regions are unsuitable markers in biological fluids, as
the cancer-specific hypermethylation event might be
occluded by hypermethylation of the cells that are present in thefluid.
The suitability of epigenetic markers versus genetic
markers as biomarkers. Although aberrant epigenetic
modifications affect a broad range of disease types, cancer presents the best characterized examples: in addition
to genetic mutations, epigenetic alterations have been
noted during tumour formation and progression15. In
particular, changes in DNA methylation have been correlated with the alteration of gene expression of some
tumour-suppressor genes, suggesting that alterations of
DNA methylation can be exploited in cancer diagnosis16
(TABLE1). In comparison with genetic approaches, such as
mutational analysis, DNA-methylation-based diagnostic
techniques present advantages with regard to their clinical application. For a given marker gene, DNA hypermethylation analysis focuses on a rather small promoter
region as hypermethylation only occurs at CpG islands,
facilitating detection, whereas genetic studies often have
to cope with large regions of interest, as numerous different point mutations occur throughout the length of the
gene. In general, DNA methylation alterations occur at
higher percentages of tumours than genetic variations,
resulting in a higher sensitivity even in single-gene DNA
methylation studies17.
Also, diseases other than cancer, such as neuro
developmental and metabolic disorders and autoimmune diseases, reveal profound alterations in DNA
methylation profiles, whereas gene mutations are rarely
680 | O CTOBER 2012 | VOLUME 13
www.nature.com/reviews/genetics
REVIEWS
Box 2 | Epigenetic modifiers: potential drug targets
Epigenetic drugs in clinical use and trials
Tumours that display global epigenomic alterations might also benefit from therapies that restore these global patterns.
Some of the modifying enzymes that put in place the mark are mutated in disease and can be targeted by specific
molecules, representing the first examples of the development of personalized medicine therapies that combine
genomic and epigenomic knowledge. Considering the dynamic and reversible nature of epigenetic marks, they
represent an attractive target for personalized therapy. However, treatment with currently approved epigenetic drugs
(DNA methyl transferase (DNMT) and histone deacetylase (HDAC) inhibitors) is rather broad, and yet to be defined
epigenetic cancer subtypes might respond differently. Here, high-resolution profiling might improve the treatment
efficiency by guiding the therapy decision. Currently approved epigenetic treatments are described below, although
readers are referred elsewhere for further detailed discussion of these11,12,109.
To date, four epigenetic drugs have been approved by the US Food and Drug Administration (FDA): two DNMT
inhibitors and two HDAC inhibitors. 5azacytidine (5azaCR; manufactured by Celgene as Vidaza) was approved in
2004 specifically to inhibit DNA methylation, and two years later its variant 5aza2deoxycytidine was approved
(5aza-CdR; manufactured by Eisai as Dacogen). Both were approved for the treatment of higher-risk myelodysplastic
syndrome. In addition, S110 (REF. 110), which is a dinucleotide containing 5aza-CdR and is suggested to have enhanced
stability and efficiency, complements the list of DNMT inhibitors that are in use or that are being tested in clinical trials.
Vorinostat (Merck) an FDA-approved HDAC inhibitor for the treatment of cutaneous T cell lymphoma (CTCL)111,112
was also confirmed to induce complete response or haematological improvement in AML patients113. Besides
Vorinostat, Romidepsin (Celgene), another HDAC inhibitor, revealed remarkable efficacies in the treatment of
cutaneous T cell lymphoma (CTCL) patients114. Two additional HDAC inhibitors Panobinostat (Novartis) and CI994
(Pfizer) are currently being tested in clinical phaseIII trials for the treatment of lymphomas and non-small-cell lung
cancer (NSCLC), respectively.
A new generation of targeted epigenetic drugs
Owing to an increased resolution of genomic and epigenetic profiling methods, targeted epigenetic treatments, such as
the specific inhibition of enzymes by small molecules, are being developed, adding to the drug arsenal to improve drug
performance further while reducing toxicity to healthy tissue. Specific treatments that target epigenetic modifiers are
the subject of current investigation, and the most recent examples are briefly considered in the following section.
Bromodomain and extra-terminal domains (BETs) are adaptor molecules that are involved in chromatin-dependent
signal transduction, resulting in transcriptional initiation and elongation. The bromodomain family member BRD4 has
been shown to be involved in gene fusions in squamous carcinomas and leukaemia. The small molecule JQ1 has been
established as a potent inhibitor of BRD4, and JQ1 treatment showed strong antiproliferative effects in cell line and
xenograft models that harboured BRD4 fusion115. Chromosomal translocation involving mixed lineage leukaemia (MLL),
which is a histone lysine methyl transferase, is an initiator of aggressive forms of leukaemia with extremely poor
prognosis. MLL fusion partners, especially members of the super elongation complex (SEC), such as AF4, AF6, AF9, AF10
and ENL, have been identified as interaction partners with BET proteins. Using the small-molecule inhibitor of the BET
family I-BET151 causes displacement of BRD3BRD4 and SEC from chromatin, followed by inhibition of coactivator
function and subsequent repression of key oncogenes, such as MYC, B cell CLL/lymphoma 2 (BCL2) and
cyclin-dependent kinase 6 (CDK6)116.
Fusion products of translocated MLL genes also initiate aberrant recruitment of DOT1L, which is a histone
methyltransferase that is involved in histone H3 lysine 79 (H3K79) methylation, to MLL target promoters, resulting in
mislocalized enzymatic activity117,118. In patients harbouring the translocated MLL genes, altered DOT1L activity
functions as a downstream event associated with the activation of leukaemogenic genes such as homeobox A9 (HOXA9)
and Meis homeobox 1 (MEIS1). Recently, Daigle and co-workers119 developed the small molecule, EPZ004777, which
specifically inhibits the H3K79 methylation activity of DOT1L. EPZ004777 treatment results in selective killing of
leukaemia cells that harbour the MLL translocation in mouse xenograft models.
Personalized medicine
Therapeutic decisions based
on genetic and epigenetic
information of individual
patients
detected, which makes epigenetic analysis applicable to a

broader range of diseases1820. Most importantly, a combination of DNA methylation and genetic analyses could
exploit the advantages of both techniques, as has recently
been shown for the diagnosis of colorectal cancers21.
Although DNA hypomethylation is also a major epigenetic alteration in cancer cells, its diagnostic use is limited
because a reduction of DNA methylation is technically
more complicated to detect than a gain of a signal. Broadly
used techniques, such as methylation-specific PCR (MSP),
reliably identify a hypermethylation event; however, a
quantitative detection, as is required for DNA hypomethylation analysis, is technically more challenging and
more difficult to standardize. However, DNA hypomethylation of the repetitive element LINE1 has frequently
been observed and is associated with drug response to
fluoropyrimidines in colorectal cancer patients22. In addition, loss of imprinting at the insulin-like growth factor 2
(IGF2) gene locus is a frequent event in cancer and has
potential use for colon cancer diagnosis23,24.
Early tumour detection is one of the major factors in
successful cancer treatment. Using non-invasive methods, such as the analysis of blood, stool, saliva or urine
samples, has clear advantages over invasive techniques,
such as biopsies. Tumour DNA can enter biological
fluids in different ways: tumour cells can be directly
released from the tissue of origin; necrotic tumour cells
are engulfed by macrophages by phagocytosis and subsequently the tumour cell enters the blood or urethra; or
free tumour DNA directly reaches the fluids by cell lysis.
Analysis of epigenetic alterations can then be carried out
by these non-invasive methods25.

REVIEWS
Table 1 | Hypermethylated genes in cancer and their associated tissue types
Gene name
Gene function
Cancer type
APC
WNT signalling
Prostate, colon, lung, bladder
AR
Androgen receptor signalling
Prostate
BMAL1
AHR signalling
Leukaemia, lymphoma
BRCA1
DNA damage response
Breast, ovarian
CDH1
Cellcell adhesion
Breast, prostate
CDH11
Cellcell adhesion
Colon, breast, oesophagus, gastric, liver
CDH13
Cellcell adhesion
Lung, head and neck
CDKN2A
Cell cycle control
Lymphoma, colon, stomach, prostate
CDKN2B
Cell cycle control
Leukaemia
DAPK1
Programmed cell death control
Lung, head and neck, bladder
EMP3
Signal transduction
Glioma
ESR1
Oestrogen receptor signalling
Breast
GSTP1
Detoxification
Prostate, liver, lung
IGFBP3
Signal transduction
Colon, lung, ovarian, prostate
LGALS3
Extracellular matrix protein
Prostate
MASPIN
Peptidase inhibitor
Pancreas
MGMT
DNA repair
Colon, glioma, lymphoma, prostate, lung
miR148a
Metastasis suppression
Metastasis
miR34b and miR-34c
Metastasis
miR9
Metastasis
miR200s
Epithelialmesenchymal transition
Colon, bladder, squamous cell carcinoma
MLH1
DNA repair
Colon, endometrium, stomach
NORE1A
Cell growth control
Colon, liver, lung, thyroid
NSD1
Nuclear receptor
Glioma, neuroblastoma
PYCARD
Apoptosis
Glioma, breast, colon, gastric, lung
RARB
Retinoic acid receptor
Breast, colon, prostate
The diagnostic potential of histone marks faces

more technical challenges compared with the diagnostic potential of DNA methylation, as histone modifications are less-stable modifications and have been
shown to be more dynamic26,27. The assessment of histone marks is more difficult to apply and to standardize, because it relies on using antibodies, which vary in
performance.
Prostate-specific antigen
(PSA). A serine protease of the
kallikrein gene family that is
secreted into seminal fluid by
prostatic epithelial cells and is
found in the serum. As it is
almost exclusively a product of
prostate cells, measurement in
blood has proved to be useful
as a tumour marker for
diagnosis of prostate cancer
and monitoring the
effectiveness of treatment.
GSTP1 hypermethylation: a prostate cancer diagnosis

marker. Prostate cancer is an example in which noninvasive diagnostics is showing promise. The most
promising candidate for a biomarker for prostate cancer
diagnosis and prognosis is the DNA hypermethylation
at an enzyme that is involved in cellular detoxification of xenobiotics and carcinogens: the glutathione
Stransferase pi1 (GSTP1) (FIG.1). Having first been
identified as being hypermethylated in cancer in 1994
(REF. 28), GSTP1 was confirmed to be mainly repressed in
prostate cancer6,2932 and also in additional cancer types,
albeit to a lesser extent6. The gain of DNA methylation
was consistently validated in over 30 independent studies, and a meta-analysis pooling these ~3,500 subjects
established GSTP1 as a promising biomarker of prostate cancer with a sensitivity of 82% and a specificity of
95%33. As GSTP1 hypermethylation was detected in 69%
of proliferative inflammatory atrophy lesions (precursor lesions for the development of prostate cancer and/
or high-grade prostatic intraepithelial neoplasia) in an
analysis of 27 patient samples, it is thought to be an early
event in tumorigenesis, underlining its importance for
diagnosis34.
Initially, Goessl and colleagues35 confirmed the presence of GSTP1 hypermethylation in urine (36%; 4 out
of 11 subjects), serum (72%; 23 out of 32 subjects) and
ejaculate (50%; 4 out of 8 subjects) from patients with
prostate cancer. Later, the GSTP1 hypermethylation was
confirmed in independent studies that detected 32%,
42% and 13% of subjects displaying hypermethylation
by analysing 62, 168 and 83 serum samples from patients
with prostate cancer, respectively3638. Here, a combination with testing of the serum-based biomarker prostatespecific antigen (PSA) could increase the sensitivity of
both methods in the diagnosis of prostate cancer38.
GSTP1 hypermethylation in urine was further confirmed with sensitivities of 27% and 50% in studies
REVIEWS
Table 1 (cont.) | Hypermethylated genes in cancer and their associated tissue types
Gene name
Gene function
Cancer type
RASSF1A
Cell cycle, DNA repair
Breast, ovarian, lung, prostate, colon
RBP1
Cell growth control
Lymphoma, gastric, squamous cell carcinoma
RIZ1
Histone methyltransferase
Leukaemia, liver, thyroid, gastric, prostate
S100P
Cell cycle control
Pancreatic
SEPT9
Cell cycle control
Colon
SFRP1
WNT signalling
Lung, liver, kidney, leukemia
SFRP2
WNT signalling
Colon
SNCG
Cell growth control
Breast, ovarian
SOCS1
Cytokine signalling
Liver
TFPI2
Colon
THBS1
Glioma
TIG1
Retinoic acid receptor responder
Prostate
TIMP2
Metallopeptidase inhibitor
Prostate
TP73
Stress response
Lymphoma
TSHR
TSH receptor
Thyroid
VHL
Hypoxia response
Kidney
WIF1
WNT signalling
Colon
WRN
DNA helicase
Colon
For further details, please see REF. 16. APC, adenomatous polyposis coli; AR, androgen receptor; BMAL1, brain and muscle
ARNT-like 1; BRCA1, breast cancer 1, early onset; CDH1, cadherin 1, type1, Ecadherin; CDH11, cadherin 11, type2,
OBcadherin; CDH13, cadherin 13, Hcadherin (heart); CDKN2A, cyclin-dependent kinase inhibitor 2A; DAPK1,
death-associated protein kinase 1; EMP3, epithelial membrane protein 3; ESR1, oestrogen receptor 1; GSTP1, glutathione
Stransferase pi 1; IGFBP3, insulin-like growth factor binding protein 3; LGALS3, lectin, galactoside-binding, soluble, 3;
MASPIN, also known as SERPINB5; MGMT, O6methylguanine-DNA methyltransferase; MLH1, mutL homologue 1, colon cancer,
nonpolyposis type2; NORE1A, also known as RASSF5; NSD1, nuclear receptor binding SET domain protein 1; PYCARD, PYD
and CARD domain containing; RARB, retinoic acid receptor beta; RASSF1A, RAS association (RalGDS/AF6) domain family
member 1; RBP1, retinol-binding protein 1; RIZ1, also known as PRDM2; S100P, S100 calcium binding protein P; SEPT9,
septin9; SFRP1, secreted frizzled-related protein 1; SNCG, synuclein, gamma (breast cancer-specific protein 1); SOCS1,
suppressor of cytokine signalling 1; TFPI2, tissue factor pathway inhibitor 2; THBS1, thrombospondin 1; TIG1, retinoic acid
receptor responder (tazarotene-induced) 1; TIMP2, TIMP metallopeptidase inhibitor 2; TP73, tumour protein p73; TSHR, TSH
receptor; VHL, von Hippel-Lindau tumour suppressor, E3 ubiquitin protein ligase; WIF1, WNT inhibitory factor 1; WRN, Werner
syndrome, RecQ helicase-like.
that analysed 22 and 36 prostate cancer patient samples, respectively39,40, and diagnosis was technically
improved by prostatic massage41 and the combination
of analysis of hypermethylation at different markers,
such as cyclin-dependent kinase inhibitor 2A (CDKN2A,
which encodes the protein p16), ARF and MGMT42,43.
Meta-analysis of 22 studies that analysed the value of
GSTP1 as a biomarker for prostate cancer in biological
fluids, including plasma, serum and urine, revealed a
high specificity of 89% but showed modest sensitivity
(52%)44. Analysing GSTP1 hypermethylation in biological fluid presents a promising alternative to conventional
methods of detection of prostate cancer. In particular,
the combination of GSTP1 with additional biomarker
genes and PSA testing could highly improve the sensitivity of diagnosis.
In addition to GSTP1, other genes exhibit a high incidence of hypermethylation in prostate cancer and can be
detected in biological fluids37,38,42. In particular, the following genes were confirmed to gain DNA methylation
to similar extents as those detected for GSTP1 (promoter
hypermethylation levels given in brackets): the tumour
suppressor adenomatous polyposis coli (APC; 8390%),
RAS association (RalGDS/AF6) domain family member 1 (RASSF1; 6895%), prostaglandin-endoperoxide

synthase 2 (PTGS2; 6888%), mediator complex subunit
15 (MED15; also known as TIG1; 96%) and multidrugresistance protein 1 (MDR1; also known as ABCB1;
8088%)29,31,32,45. Especially in combination, these markers were able to distinguish reliably between primary
cancer tissue and benign tissue29,31,45. Combining DNA
hypermethylation of GSTP1, APC, RASSF1, PTGS2 and
MDR1 resulted in both sensitivity and specificity of up to
100% of 73 tested prostate cancer specimens29. Using only
a subset of hypermethylated markers (namely, GSTP1,
APC and PTGS2) led to a sensitivity of 7191% and to a
specificity of 93100% in the detection of prostate cancer in 53 specimens31. Even the combination of only
methylation analysis of GSTP1 and TIG1 in 80 cancer
samples resulted in 93% sensitivity and 85% specificity
of diagnosis and prognosis of prostate cancer, respectively45. Specifically, the combination of hypermethylation of the above mentioned markers presents significant
correlation with pathological grade and stage: for example, hypermethylation of GSTP1, APC and PTGS2
gene loci correlated with prognostic indicators such

REVIEWS
Healthy tissue
Cancer
Prognosis and diagnosis
MGMT
CpG island
hypermethylation
MGMT
MGMT
Transcription
Transcription
MGMT
Translation
DNA
repair
Translation
DNA
damage
Prediction
of drug
response
Brain biopsy
GSTP1
CpG island
CpG island
hypermethylation
Urine system biopsy
GSTP1
GSTP1
Transcription
Transcription
GSTP1
Translation
Tumour
detection
Translation
6
Figure 1 | GSTP1 and MGMT: case examples for epigenetic profiling in diagnosis and prognosis.
methylguanineNatureOReviews
| Genetics
DNA methyltransferase (MGMT) protects normal cells against transition mutations by removing alkyl groups (red
squares), which have been introduced by carcinogens such as nitrosamides, from guanine bases. Hypermethylation of
MGMT is correlated with blocked gene expression, leading to the accumulation of alkylated guanines and subsequent
DNA damage. However, hypermethylated MGMT in glioblastomas is a predictor of good response to DNA-alkylating
drugs, such as carmustine and temozolomide, as an active MGMT is able to repair modified bases introduced by the
therapy. Hypermethylation of MGMT is used as biomarker for drug response in glioblastoma biopsies but can also be
detected in serum. Glutathione Stransferase pi1 (GSTP1) has been established as biomarker for prostate cancer
diagnosis and prognosis. It is involved in cellular detoxification of xenobiotics and carcinogens (red circles). In a normal
context, GSTP1 actively removes damaging agents from the cell. Hypermethylation of GSTP1 in tumours is correlated
with gene repression and subsequent cellular accumulation of xenobiotics and carcinogens, thereby favouring
tumorigenesis. Hypermethylation of GSTP1 is used as a biomarker for malignant cells in prostate biopsies but can also be
detected in urine and serum.
as pathologic stage31. Although these studies analysed

hundreds of cancer samples and clearly suggest the great
potential of these markers, large-scale studies and comprehensive meta-analysis, as carried out for GSTP1, are
needed to establish the potential markers as being suitable for clinical use in primary tissues29,31,45 and biological fluids37,38,42.
Non-invasive DNA methylation markers for diagnosis
of other cancers. In addition to prostate cancer, other
biomarkers from cancers of diverse origins have been
identified in biological fluids. Colorectal cancer (CRC)
frequently displays hypermethylation of APC, MGMT,
RASSF2 and WNT inhibitory factor 1 (WIF1). Analysing
the hypermethylation levels of these genes in 243 primary

tissues and corresponding plasma samples of sporadic
colorectal cancer patients estimated a detection sensitivity of 87% and a specificity of 92% for colorectal cancer diagnosis in patient plasma46. Another study that
screened for DNA methylation candidate markers differentially methylated between primary CRC and nonpathologic tissue identified the hypermethylation of
transmembrane protein with EGF-like and two follistatin-like domains 2(TMEFF2), nerve growth factor receptor (NGFR) and septin 9 (SEPT9) in cancer samples47.
Their potential as blood-based biomarkers was validated
in 133 CRC samples and 179 healthy control individuals.
The use of SEPT9 hypermethylation as a biomarker for
REVIEWS
blood-based CRC diagnosis was later confirmed in an
independent study that screened 50 CRC patients; cancer
was detected with a sensitivity of 90% and a specificity
of 88% using the methylation at this gene as a marker48.
Using stool as a non-invasive screening material for 84
CRC patients revealed hypermethylation of TFPI2 as
a CRC marker. The presence of hypermethylation at
TFPI2 could detect CRCs in samples from patients with
cancer with a sensitivity of 7689%49.
Gene promoter hypermethylation has also been established as a biomarker candidate for patients with glioblastoma50 and patients with non-small-cell lung cancer
(NSCLC)51. Glioblastomas could be detected with a sensitivity of 95% and a specificity of 60% by analysing MGMT
promoter methylation in the serum of 37 patients50. Here
the patient serum DNA hypermethylation levels of
MGMT were highly correlated with the primary tissue
levels and were able to predict a lack of cancer progression and overall survival of the patient. The detection of
MGMT hypermethylation in combination with CDKN2A
and GSTP1 hypermethylation allowed for the detection of
NSCLC with 73% sensitivity but analysed a small cohort
of 15 samples51. Using saliva as a source of biomarkers
in 30 patients suffering from head and neck cancer identified the combination of MGMT, CDKN2A and deathassociated protein kinase 1 (DAPK1) hypermethylation
as powerful biomarkers, detecting 65% of tumours52. In
followup studies, the number of potential biomarkers was
complemented by homeobox A9 (HOXA9) and nidogen 2
(NID2) hypermethylation, showing detection sensitivities
of 75% and 87%, respectively, and specificities of 53% and
21%, respectively, giving a higher sensitivity when tested
in combination53.
The detection of DNA methylation alteration in
biological fluids represents a promising tool for noninvasive cancer detection. However, many candidates
present a low specificity, which results in a false-positive
diagnosis. Reference data sets allow for the initial exclusion of genes that show high inter-individual variation
or high levels of basic methylation, making it difficult
to detect a gain in cancer patients. Improving candidate
selection by using reference DNA methylomes could
accelerate biomarker discovery and specificity by minimizing the number of unsuitable markers for prognostic DNA methylation approaches selected for further
investigation.
Genome-wide epigenetic technologies might also
facilitate the identification of cancers of unknown primary origin. Following the detection of metastatic diseases, it is crucial for the success of the treatment and, in
turn, for the prognosis to identify the primary origin of
the tumour. In a comprehensive study, analysing DNA
methylation of 1,505 CpG sites in 1,628 human samples, Fernandez etal.20 determined tissue- and cancerspecific fingerprints and were able to assign a tissue of
origin for 69% of the cancers of unknown primary origin analysed (FIG.2).
non-cancer context54 (TABLE2). Twin studies are a powerful approach as, despite their identical genetic background, twins have differences in DNA methylation55.
Three studies of monozygotic twins who were discordant
for disease development revealed surprising epigenetic
differences. Monozygotic twins who were discordant for
systemic lupus erythematosus (SLE), which is an autoimmune disorder, revealed differences in DNA methylation
of blood cells in a number of genes that are involved in
immune function18. A similar study setup using blood
from monozygotic twins who were discordant for childhood-onset type1 diabetes identified a disease-specific
DNA methylation signature, which was subsequently validated in a cohort outside the twin context56. Differential
DNA methylation markers may also help in the diagnosis
of diseases in which only a few genetic mutations have
been linked to the development of the disease, such as
Alzheimers disease57. Comparing DNA methylation levels
in neurons of monozygotic twins who were discordant for
the disease unravelled global loss of methylation levels in
patients compared with their healthy relatives, suggesting
a role for an altered epigenetic landscape in Alzheimers
disease58. In this regard, the incidence of neurodegenerative disorders increase with advanced age, and recent data
obtained by whole-genome bisulphite sequencing show
important DNA methylation changes in nonagenarians
and centenarians59. In all of these studies, how these
changes might contribute to future diagnosis has to be
clarified in followup studies.
DNA methylation in the diagnosis of non-cancer diseases. It is of note that DNA methylation profiling is
also able to identify differentially methylated genes in a
Single-loci DNA methylation profiling for disease

prognosis. Single-loci strategies amplify the knowledge obtained from tumour diagnosis by combining
Moving towards clinical epigenetic tests. Although none

of the assays for epigenetic diagnostics has yet achieved
approval by authorities, there are commercial tests for
laboratory usage and clinical trials available. In particular, these take advantage of the predictive potential
of hypermethylation of GSTP1 and APC for prostate
cancer diagnosis. In addition, assays that detect loss
of imprinting of IGF2 as a risk factor for colon cancer
and aberrant DNA methylation patterns, causing the
neurodevelopmental disorders of genomic imprinting
PraderWilli syndrome and Angelman syndrome, are
commercially available. Note, however, that the alterations in methylation here occur in the germline and are
inherited rather than somatic mutations. Furthermore,
in moving to genome-wide DNA methylation profiling
with ever-decreasing costs per sample, DNA methylation
screening will probably soon prove to have great advantages over current diagnostic methods, such as prostatespecific antigen testing for prostate cancer diagnosis.
DNA methylation profiling for disease prognosis

In addition to its diagnostic potential, DNA methylation
is informative for patient prognosis in terms of tumour
recurrence and overall survival. This is certainly true for
single-gene loci; however, recent genome-wide profiling
has also indicated a strong potential for genome-wide
profiles to predict outcome (FIG.2).

REVIEWS
a Disease diagnosis
Malignant tissue
Patient with cancer
c Therapy response
Poor response
Benign tissue
Good response
Patient sample
b Disease prognosis
Short-term survival
Long-term survival
Biopsy
Saliva
Blood
Stool
Urine
Cancers of
unknown
primary origin
d Tumour type identication from cancer of

unknown primary origin
Colon Lung Breast Kidney
Liver
Skin
Brain
Figure 2 | High-resolution screening of patient samples in disease diagnosis, prognosis, prediction of drug
Nature Reviews | Genetics
response and tumour-type identification. Future cancer diagnosis, prognosis and therapy will benefit from
epigenetic high-resolution screening technologies. Profiling and subsequent classification can be carried out on
primary tissue or biological fluids, such as saliva, blood, stool and urine. Profiles in panels a, b and c are representative
profiles that give an indication of what may be possible with genome-wide profiling in the future. a | The assessment of
reference DNA methylome data sets for normal tissues and cancer types enables us to classify subtypes on the basis
of their DNA methylation signatures. These newly defined methylation-profile-based subtypes are suitable for the
differentiation between malignant and benign tissue. b | Methylation signatures can also be used for the differentiation
between patients with short-term or long-term survival. c | Different methylation profiles can be used to predict the
different response to drug treatment. d | Furthermore, the unique and tissue-specific DNA methylation profiles of
normal cells can help in the identification of the tissue of origin of cancers with unknown primary origin. Panel d is
modified, with permission, from REF. 20 (2011) CSH Press.
tumour markers with clinical followup data to try and

identify correlations between epigenetic markers
and prognosis. Using a single-loci approach and a
cohort of 51 patients with recurrence within 40months
and 116 patients without tumour recurrence in that
period, the hypermethylation of four genes CDKN2A,
cadherin 13 (CDH13), RASSF1 and APC was able
to predict tumour progression in stage 1 NSCLC independently of other clinical features, such as stage, age
and smoking history; in a multivariate model, promoter
DNA hypermethylation of these genes was associated

with early tumour recurrence60. The results were confirmed in an independent patient cohort of 20 samples60.
Further examples of single-gene loci that indicate outcome include hypermethylation of the genes encoding
NSD1, DAPK1, epithelial membrane protein 3 (EMP3)
and CDKN2A; promoter hypermethylation of these
genes has been linked to poor outcome in neuroblastomas for NSD1, lung cancer for DAPK1, brain tumours
for EMP3 and CRC for CDKN2A11.
REVIEWS
Table 2 | Epigenetic alterations in non-cancer diseases
Epigenetic
modification
Disease
Alteration
DNA
methylation
Alzheimers disease
Hypermethylation (NEP)
Angelman syndrome
Imprinting defect (15q11.2q13)
Atherosclerosis
Aberrant methylation
ATRX syndrome
Aberrant methylation (subtelomeric repeats)
Diabetes type1
Aberrant methylation
Friedrichs ataxia
Hypermethylation (FXN)
Immunodeficiency, centromeric region

instability and facial anomalies syndrome
Aberrant methylation (mutated DNMT3B)
Multiple sclerosis
Hypomethylation (PADI2)
PraderWilli syndrome
Imprinting defect (15q11.2q13)
Retts syndrome
Mutation (MECP2)
Rheumatoid arthritis
Aberrant methylation (DR3 and L1)
Systemic lupus erythematosus
Hypomethylation (PRF1, CD70, CD154 and AIM2)
Alzheimers disease
Aberrant phosphorylation
Atherosclerosis
Histone H4 lysine 20 (H4K20) hypermethylation; H4

deacetylation
Diabetes type1
Aberrant methylation (CLTA4 and IL-6)
Fragile X syndrome
Aberrant histone binding (FMR1)
Friedrichs ataxia
Aberrant acetylation (FXN)
Huntingtons disease
Aberrant methylation and acetylation
Parkinsons disease
Aberrant methylation and acetylation
Histone
modification
Nucleosome
positioning
RubensteinTaybi syndrome
Aberrant acetylation (mutated CREBBP, EP300)
Systemic lupus erythematosus
Aberrant phosphorylation (NFB1 targets)
Congenital myotonic dystrophy
Mispositioning
Diabetes type1
SNP-dependent distribution (CLTA4 and IL-6)
Rheumatoid arthritis
Histone variant replacement (NFB1 targets)
For further details, see REF.53. AIM2, absent in melanoma 2; CD154, also known as CD40LG; CLTA4, clathrin, light chain A;
CREBBP, CREB-binding protein; DNMTB, DNA (cytosine5)-methyltransferase 3 beta; DR3, also known as TNFRSF25; EP300,
E1A-binding protein p300; FMT1, fragile X mental retardation 1; FXN, frataxin; IL-6, interleukin 6; L1, also known as FAM49B;
MECP2, methyl-CpG-binding protein 2 (Retts syndrome); NEP, also known as MME; NFB1, nuclear factor- B p105 subunit;
PADI2, peptidyl arginine deiminase, typeII; PRF1, perforin 1 (pore-forming protein).
DNA methylation profiling for metastasis identification. Crucial to prognosis is the detection and treatment of tumour cells that have disseminated from the
point of origin. DNA methylation alterations in metastatic cancer cells were first reported at Ecadherin for
breast cancer61 and extended to an additional member
of the cadherin family (cadherin 11) in head and neck
cancer and melanoma samples62. Another study identified microRNA gene hypermethylation profiles that are
characteristic of human metastasis63.
DNA methylomes in prognosis. Cancer cells display a
global loss of DNA methylation; however, CpG-rich
gene promoters are frequently targets of DNA hypermethylation associated with tumour-suppressor gene
silencing and cancer formation and progression. Thus,
comprehensive DNA methylation profiling presents a
potent tool to profile regions with possible prognosis
predictive potential (BOX1; FIG.2). As high-resolution
data serve not only as a source for cancer-subtype-specific profiles but also for the identification of powerful
single epigenetic biomarker genes and gene combinations at various stages of disease, many types of clinical
applications can benefit from this advanced screening
strategy.
The prognostic potential of DNA methylation profiling was initially reported for childhood acute lymphoblastic leukaemia samples64. Determining the DNA
methylation level of 401 patients and 1,320CpG sites
(using GoldenGate from Illumina) allowed the classification of the patients into acute lymphoblastic leukaemia subtypes. Furthermore, DNA methylation profiling
allowed the stratification of patients with high hyperdiploidy and translocation t(12;21) into two subgroups
with different probabilities of relapse. In particular, DNA
methylation profiling of 40 genes permitted the identification of pathologically distinct disease subtypes and the
identification of 20 genes at which methylation predicted

REVIEWS
disease relapse64. Here, DNA methylation profiling
therefore resulted in subtype-specific and predictive
signatures with potential use as future prognostic biomarkers for childhoodacute lymphoblastic leukaemia.
A more comprehensive study that analysed ~14,000
genes (using Infinium 27K from Illumina) classified
154 normal and CRC samples according to their DNA
methylation profile65. Strikingly, this study identified
subgroups correlating with prognostic markers such as
mutL homologue 1, colon cancer, nonpolyposis type2
(MLH1) methylation and v-raf murine sarcoma viral
oncogene B1 (BRAF), TP53 or v-Ki-ras2 Kirsten rat
sarcoma viral oncogene (KRAS) mutation, suggesting
that DNA methylation analysis might contribute to
CRC classification, and a combination of epigenetic and
genetic analysis might further improve the accuracy of
prognosis.
Another study, focusing on the analysis of developmental regulated Polycomb target genes (PCTGs)
and loci heavily methylated in embryonic stem
cells (MESCs) in 1,475 samples using the Infinium
27k technology, identified both regions to be differentially methylated in various cancer types 66.
Hypermethylation of PCTGs was observed as an early
event in tumorigenesis even detectable before neoplastic changes occur, hence suggesting PCTGs as potential
diagnostic markers. In contrast, hypomethylation of
MESCs progressed concomitantly with tumour stage
and correlated with poor prognosis in breast, ovarian,
cervical and endometrial cancers.
The DNA methylation profiling of 248 breast cancer
samples identified a previously unrecognized subtype
associated with T lymphocyte infiltration. Importantly,
profiling immune genes determined a prognostic value
of the profile. In particular, the promoter hypermethylation of lymphocyte transmembrane adaptor 1(LAX1)
and CD3D significantly correlated with survival in certain breast cancer subtypes67.
Aiming to identify a DNA methylation signature
related to breast cancer metastasis, 39 primary breast
tumour specimens were analysed using the Infinium
27K platform68. Strikingly, DNA methylation profiling
determined two distinct groups with a substantially different association to metastasis-free survival. Assessing
the DNA methylation of only three genes of the methylation signature rho guanine nucleotide exchange
factor (ARHGEF7), ALX homeobox 4 (ALX4) and
RAS-protein-specific guanine nucleotide-releasing factor 2 (RASGRF2) allowed the prediction of metastasis-free cancer and overall survival. In particular, the
authors further suggest a common signature of alteration
in DNA methylation between gliomas, colon and breast
cancers.
Analysing 272 glioblastoma samples with the
GoldenGate and Infinium 27K technology, a subgroup
of cancers (8.8%) was identified, showing high levels of
CpG island methylation69. Furthermore, the authors
detected a high correlation between hypermethylation
and mutation of isocitrate dehydrogenase 1 (IDH1).
Mutant IDH1 protein converts ketoglutarate into
the oncometabolite (R)-2hydroxyglutarate (2HG),
which is a potent inhibitor of the TET methylcytosine

dioxygenase (TET2), which is in turn involved in active
DNA demethylation. Using an independent technique
and only eight hypermethylated biomarker genes, they
confirmed the glioblastoma subtype in an additional
cohort of 208 samples, detecting 7.8% to be highly
methylated. The novel subtype was even more detectable in low-grade gliomas, with the highest abundance
in oligodendrogliomas (93%). Strikingly, DNA methylation status of eight biomarker genes could predict
patient survival and hence might be a useful clinical
tool for estimating disease prognosis. The results could
be confirmed using the higher-resolution Infinium
450K technology analysing more than 450,000CpG
sites70 (BOX1). Seventy-two low-grade gliomas displayed
a high correlation between CpG island hypermethylation and patient survival. In addition, the association
between high levels of CpG island DNA methylation
and IDH1 mutation was empirically and experimentally confirmed70.
These examples give an outlook on the potential of
DNA methylation profiling as an important tool for
fine-tuning diagnosis and prognosis. In combination
with currently used cytogenetic and genetic markers,
DNA methylation might highly improve the prognostic accuracy in the future. Here, classification of newly
profiled patient samples by comparing with publicly
available reference data sets as discussed above is a possible future scenario (FIG.2). Owing to rapidly decreasing
sequencing costs, a large number of cancer methylomes
will be available soon, which in addition to array-based
discovery should allow for the identification of novel
prognostic markers in an unbiased manner even outside
currently suspected regions of importance.
DNA methylation profiling for disease response

Sensitivity to chemotherapeutic drugs is variable
between patients and tumour types. However, epigenetic profiling has identified tumour-specific
drug-response markers that are capable of predicting
response to chemotherapeutic treatment. TABLE3 provides a demonstrative list of differentially methylated
genes in human cancer that are candidate markers of
drug response.
DNA methylation predictors of the response to chemotherapeutic drugs. The best example of a methylation
marker indicating a response to a chemotherapeutic
agent is hypermethylation at MGMT in glioma. Its promoter hypermethylation and associated transcriptional
repression lead to greater sensitivity to carmustine- and
temozolomide-based therapies5,71,72 (FIG.1). MGMT protects cells against transition mutations by removing
alkyl groups from guanine bases, which are introduced
by carcinogens such as nitrosamides. As guanine bases
are also alkylated by chemotherapeutic drugs such as
carmustine and temozolomide, MGMT expression
in tumour cells leads to resistance to these drugs. In
cancers that harbour hypermethylated MGMT, the
repair mechanism is disabled, leading to sensitivity to
alkylating agents5,71.
REVIEWS
Table 3 | Hypermethylated genes predict drug sensitivity
Gene name
Gene function
Therapeutical consequences
Tumour type
application
Example
refs
ABCB1
Protein transport
Sensitivity to doxorubicin
Breast
120
APAF1
Apoptotic activator
Resistance to adriamycin
Melanoma
121
BRCA1
DNA damage response
Sensitivity to PARP inhibitors and

alkylating agents
Breast, ovary
CDK10
Cell cycle control
Resistance to anti-oestrogens
Breast
122
CHFR
Ubiquitin protein ligase
Sensitivity to paclitaxel and

docetaxel
Ovary, endometrium,
stomach
123
ESR1
ER signalling
Resistance to anti-oestrogens
Breast
124
FANCF
DNA damage response
Sensitivity to cisplatin
Ovarian
125
GSTP1
Detoxification
Sensitivity to doxorubicin
Prostate, breast, kidney
126
IGFBP3
Signal transduction
Resistance to cisplatin
Lung
127
LINE1
Repetitive element
Resistance to fluoropyrimidines
Colon
22
MGMT
DNA repair
Sensitivity to temozolomide,
BCNU, ACNU, procarbazine
Glioma, colon, lung,

lymphoma
MLH1
DNA repair
Resistance to cisplatin
Colon, stomach,
endometrium, ovary
128
MT1E
Antioxidant
Melanoma
129
PITX2
Transcriptional regulator
Resistence to tamoxifen
Breast
130
PLK2
Cell division
Sensitivity to paclitaxel and

carboplatin
Ovary
131
PRKCDBP
Signal transduction
Resistance to TNF
Colon
132
SFN
Signal transduction
Sensitivity to cisplatin and

gemcitabine
Lung
133
SLC19A1
Folate transporter
Resistance to methotrexate
Lymphomas
134
SULF2
Heparin signalling
Sensitivity to camptothecin
Lung
135
TFAP2E
Transcriptional regulator
Sensitivity to fluorouracil
Colon
136
TGM2
Apoptosis
Resistance to doxorubicin and

cisplatin
Lung, breast, ovary
137
TP73
Stress response
Renal, melanoma
138
WRN
DNA helicases
Sensitivity to irinotecan
Colon
139
ERCC5
DNA repair
Resistance to nemorubicin
Ovary
140
81
ABCB1, ATP-binding cassette, subfamily B (MDR/TAP), member 1; ACNU, (1-4-amino-2-methyl-5-pyrimidinyl)-methyl-3-(2chloroethyl)-3-nitrosourea hydrochloride; APAF1, apoptotic activator; BCNU, bis-chloroethylnitrosourea; BRCA1, breast cancer 1,
early onset; CDK10, cyclin-dependent kinase 10; CHFR, checkpoint with forkhead and ring finger domains, E3 ubiquitin protein
ligase; ERCC5, excision repair cross-complementing rodent repair deficiency, complementation group 5; ESR1, oestrogen
receptor1; FANCF, Fanconi anaemia, complementation group F; GSTP1, glutathione Stransferase pi 1; IGFBP3, insulin-like growth
factor binding protein 3; MGMT, O6methylguanine-DNA methyltransferase; MLH1, mutL homologue 1, colon cancer, nonpolyposis
type2; MT1E, metallothionein 1E; PITX2, paired-like homeodomain 2; PLK2, polo-like kinase 2; PRKCDBP, protein kinase C, delta
binding protein; SFN, stratifin; SLC19A1, solute carrier family 19; SULF2, sulphatase 2; TFAP2E, transcription factor AP2 epsilon
(activating enhancer binding protein 2 epsilon); TGM2, transglutaminase 2 (C polypeptide, protein-glutamine-gamma-glutamyltransferase); TP73, tumour protein p73; TNF, tumour necrosis factor-; WRN, Werner syndrome, RecQ helicase-like.
Hypermethylation of MGMT was first identified

in 40% of gliomas and colorectal cancer samples in a
study of 500 primary human tumours72. Furthermore,
NSCLC, lymphomas and head and neck carcinomas
revealed hypermethylation of MGMT72. A subsequent
study of 47 patients with glioma detected 40% of
hypermethylated promoters and determined MGMT
promoter methylation as a predictor of responsiveness to the alkylating agent carmustine5. In this study,
MGMT hypermethylation was found to be associated
with regression of the tumour and prolonged overall
and disease-free survival following treatment with
carmustine. Subsequently, MGMT promoter hypermethylation has been associated with treatment
outcome of additional alkylating agents, such as
temozolomide71, and other chemotherapeutic drugs,
including cilengitide73 and cyclophosphamide74, as well
as radiation75.
A further example of a DNA methylation event
being indicative of successful chemotherapeutic intervention is that of hypermethylation at breast cancer 1,
early onset (BRCA1). Mutations in BRCA1, which is
a gene involved in DNA repair, have frequently been
observed in patients with inherited breast tumours76.

REVIEWS
Furthermore, hypermethylation and associated transcriptional repression of BRCA1 have also been observed
in sporadic breast and ovarian cancers77. Cancers that
are marked by altered BRCA1 expression can be treated
with two distinct treatment strategies: conventional
chemotherapy or specific inhibition of DNA repair proteins. Triple-negative breast tumours (a breast cancer
subtype that lacks the expression of v-erb-b2 erythroblastic leukaemia viral oncogene homologue 2 (ERBB2;
also known as HER2/neu) as well as the oestrogen and
progesterone receptors), showing frequent hypermethylation of BRCA1 (REF. 78), were associated with an
improved response to cytotoxic drugs such as cisplatin79.
In addition, BRCA1 gene mutation carriers benefit from
a targeted therapy using inhibitors of PARP, which area
proteins that are involved in base excision repair (that is,
a DNA repair mechanism that is distinct from homologous recombination, in which BRCA1 is involved) 80.
Two PhaseII trials are currently ongoing that aim to
establish the efficiency of PARP inhibitors in the treatment of BRCA1-mutated breast and ovarian cancers.
Consequently, it is tempting to speculate that PARP
inhibitors might also successfully target sporadic cases
that are positive for promoter BRCA1 methylation81.
Perspectives in personalized medicine

We now have ever-growing numbers of reported epigenetic alterations in disease, and this offers a chance to
increase sensitivity and specificity of future diagnostics
and therapies.
In order for this to be realized, further reference
disease genomes and DNA methylomes are required,
and this is currently on the way to being met. Consortia
such as the 1000 Genomes Project for genome sequencing and the Blueprint, IHEC and Roadmap projects for
methylome mapping13,14 currently aim to generate reference data sets for research and clinical use. Dealing
with the huge amount of data produced is the most obvious immediate challenge. Modern cancer management
1.
2.
3.
4.
5.
6.
7.
8.
Sandoval, J. etal. Validation of a DNA methylation

microarray for 450,000CpG sites in the human
genome. Epigenetics 6, 692702 (2011).
Bibikova, M. etal. High density DNA methylation
array with single CpG site resolution. Genomics 98,
288295 (2011).
Harris, R.A. etal. Comparison of sequencing-based
methods to profile DNA methylation and identification
of monoallelic epigenetic modifications. Nature
Biotech. 28, 10971105 (2010).
Baylin, S.B. & Jones, P.A. A decade of exploring the
cancer epigenome biological and translational
implications. Nature Rev. Cancer 11, 726734 (2011).
Esteller, M. etal. Inactivation of the DNA-repair gene
MGMT and the clinical response of gliomas to alkylating
agents. N.Engl. J.Med. 343, 13501354 (2000).
Esteller, M. etal. Inactivation of glutathione
Stransferase P1 gene by promoter hypermethylation
in human neoplasia. Cancer Res. 58, 45154518
(1998).
Hon, G.C. etal. Global DNA hypomethylation coupled
to repressive chromatin domain formation and gene
silencing in breast cancer. Genome Res. 22, 246258
(2011).
Berman, B.P. etal. Regions of focal DNA
hypermethylation and long-range hypomethylation in
colorectal cancer coincide with nuclear laminaassociated domains. Nature Genet. 44, 4046
(2011).
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
strategies will be required to combine and integrate

data from various omics approaches (such as genomics, transcriptomics and epigenomics) to determine
profiles predicting disease outcome in terms of patient
prognosis and treatment response. The integration of
data from different studies and the comparison of epigenetic profiles from normal and diseased tissue will
allow the production of predictive models for disease
outcomes using epigenetic biomarkers and signatures.
As we advance from studying alterations at single loci
to entire genomes, groups of disease-specific biomarkers will soon be able to diagnose and to predict disease
prognosis more accurately than is possible at present.
For certain diseases, such as prostate33 and lung cancers60, combinations of DNA hypermethylated biomarkers have achieved high sensitivities in the detection of
cancer cells and in the prediction of tumour progression, respectively. However, before being of clinical
value, in particular detection sensitivity and specificity
of biomarkers in biological fluids have to be improved.
Here, the combination of different epigenetic, genetic
and molecular biomarkers (for example, PSA) could
increase diagnostic reliability. Using high-throughput
epigenetic-screening approaches, the identification
of combinations of biomarkers is now achievable for
other less well-characterized cancer types. In addition,
the increased resolution of DNA methylation profiling
might further increase the number of reliable biomarkers for well-studied diseases. However, owing to high
costs of high-resolution techniques, these approaches
are currently only applicable to marker identification,
but for clinical assays, single-gene approaches will still
be the methods ofchoice.
The key to successful therapy lies in combinatorial
approaches. If we can integrate candidate-altered epigenetic profile gene lists into potent predictive models and
combine conventional treatments with novel drugs, we
will achieve the low-dose, customized and high-impact
treatment weseek.
Hansen, K.D. Increased methylation variation in

epigenetic domains across cancer types. Nature
Genet. 43, 768775 (2011).
Jones, P.A. Functions of DNA methylation: islands,
start sites, gene bodies and beyond. Nature Rev.
Genet. 13, 484492 (2012).
Rodrguez-Paredes, M. & Esteller, M. Cancer
epigenetics reaches mainstream oncology. Nature
Med. 17, 330339 (2011).
Dawson, M.A. & Kouzarides, T. Cancer epigenetics:
from mechanism to therapy. Cell 150, 1227 (2012).
Adams, D. etal. BLUEPRINT to decode the epigenetic
signature written in blood. Nature Biotech. 30,
224226 (2012).
Bernstein, B.E. The NIH roadmap epigenomics
mapping consortium. Nature Biotech. 28,
10451048 (2010).
Esteller, M. Epigenetics in cancer. N.Engl. J.Med.
358, 11481159 (2008).
Esteller, M. Epigenetic gene silencing in cancer:
the DNA hypermethylome. Hum. Mol. Genet. 16,
R50R59 (2007).
Esteller, M., Corn, P.G., Baylin, S.B. & Herman, J.G.
A gene hypermethylation profile of human cancer.
Cancer Res. 61, 32253229 (2001).
Javierre, B.M. etal. Changes in the pattern of DNA
methylation associate with twin discordance in
systemic lupus erythematosus. Genome Res. 20,
170179 (2010).
19. Volkmar, M. etal. DNA methylation profiling identifies

epigenetic dysregulation in pancreatic islets from type
2 diabetic patients. EMBO J. 31, 14051426 (2012).
20. Fernandez, A.F. etal. A DNA methylation fingerprint
of 1628 human samples. Genome Res. 22, 407419
(2011).
21. Ahlquist, D.A. etal. Next-generation stool DNA test
accurately detects colorectal cancer and large
adenomas. Gastroenterology 142, 248256; quiz
e25e26 (2012).
22. Kawakami, K. etal. Long interspersed nuclear
element1 hypomethylation is a potential biomarker
for the prediction of response to oral fluoropyrimidines
in microsatellite stable and CpG island methylator
phenotype-negative colorectal cancer. Cancer Sci.
102, 166174 (2011).
23. Cui, H. etal. Loss of imprinting in colorectal cancer
linked to hypomethylation of H19 and IGF2. Cancer
Res. 62, 64426446 (2002).
24. Ito, Y. etal. Somatically acquired hypomethylation of
IGF2 in breast and colorectal cancer. Hum. Mol. Genet.
17, 26332643 (2008).
25. Maxwell, A., McCudden, C.R., Wians, F. & Willis, M.S.
Recent advances in the detection of prostate cancer
using epigenetic markers in commonly collected
laboratory samples. Lab. Med. 40, 171178 (2009).
26. Feinberg, A.P. Phenotypic plasticity and the
epigenetics of human disease. Nature 447, 433440
(2007).
REVIEWS
27. Berdasco, M. & Esteller, M. Aberrant epigenetic
landscape in cancer: how cellular identity goes awry.
Dev. Cell 19, 698711 (2010).
28. Lee, W.H. etal. Cytidine methylation of regulatory
sequences near the piclass glutathione Stransferase
gene accompanies human prostatic carcinogenesis.
Proc. Natl Acad. Sci. USA 91, 1173311737 (1994).
29. Yegnasubramanian, S. etal. Hypermethylation of CpG
islands in primary and metastatic human prostate
cancer. Cancer Res. 64, 19751986 (2004).
30. Jernimo, C. etal. Quantitation of GSTP1 methylation
in non-neoplastic prostatic tissue and organ-confined
prostate adenocarcinoma. J.Natl Cancer Inst. 93,
17471752 (2001).
31. Bastian, P.J. etal. Diagnostic and prognostic
information in prostate cancer with the help of a small
set of hypermethylated gene loci. Clin. Cancer Res. 11,
40974106 (2005).
32. Padar, A. etal. Inactivation of cyclin D2 gene in
prostate cancers by aberrant promoter methylation.
Clin. Cancer Res. 9, 47304734 (2003).
33. Van Neste, L. etal. The epigenetic promise for
prostate cancer diagnosis. Prostate 72, 12481261
(2011).
34. Nakayama, M. etal. Hypermethylation of the human
glutathione Stransferase gene (GSTP1) CpG island is
present in a subset of proliferative inflammatory
atrophy lesions but not in normal or hyperplastic
epithelium of the prostate: a detailed study using
laser-capture microdissection. Am. J.Pathol. 163,
923933 (2003).
35. Goessl, C. etal. Fluorescent methylation-specific
polymerase chain reaction for DNA-based detection of
prostate cancer in bodily fluids. Cancer Res. 60,
59415945 (2000).
36. Reibenwein, J. etal. Promoter hypermethylation of
GSTP1, AR, and 1433 in serum of prostate cancer
patients and its clinical relevance. Prostate 67,
427432 (2007).
37. Ellinger, J. etal. CpG island hypermethylation in
cell-free serum DNA identifies patients with localized
prostate cancer. Prostate 68, 4249 (2008).
38. Sunami, E. etal. Multimarker circulating DNA assay
for assessing blood of prostate cancer patients.
Clin. Chem. 55, 559567 (2009).
39. Gonzalgo, M.L., Pavlovich, C.P., Lee, S.M. &
Nelson, W.G. Prostate cancer detection by GSTP1
methylation analysis of postbiopsy urine specimens.
Clin. Cancer Res. 9, 26732677 (2003).
40. Cairns, P. etal. Molecular detection of prostate cancer
in urine by GSTP1 hypermethylation. Clin. Cancer Res.
7, 27272730 (2001).
41. Goessl, C. etal. DNA-based detection of prostate
cancer in urine after prostatic massage. Urology 58,
335338 (2001).
42. Hoque, M.O. etal. Quantitative methylation-specific
polymerase chain reaction gene patterns in urine
sediment distinguish prostate cancer patients from
control subjects. J.Clin. Oncol. 23, 65696575
(2005).
43. Rogers, C.G. etal. High concordance of gene
methylation in post-digital rectal examination and
post-biopsy urine samples for prostate cancer
detection. J.Urol. 176, 22802284 (2006).
44. Wu, T. etal. Measurement of GSTP1 promoter
methylation in body fluids may complement PSA
screening: a meta-analysis. Br. J.Cancer 105, 6573
(2011).
45. Ellinger, J. etal. CpG island hypermethylation at
multiple gene sites in diagnosis and prognosis of
prostate cancer. Urology 71, 161167 (2008).
46. Lee, B.B. etal. Aberrant methylation of APC, MGMT,
RASSF2A, and Wif1 genes in plasma as a biomarker
for early detection of colorectal cancer. Clin. Cancer
Res. 15, 61856191 (2009).
47. Lofton-Day, C. DNA methylation biomarkers for bloodbased colorectal cancer screening. Clin. Chem. 54,
414423 (2008).
48. Warren, J.D. etal. Septin 9 methylated DNA is a
sensitive and specific blood test for colorectal cancer.
BMC Med. 9, 133 (2011).
49. Glockner, S.C. Methylation of TFPI2 in stool DNA: a
potential novel biomarker for the detection of colorectal
cancer. Cancer Res. 69, 46914699 (2009).
50. Balana, C. etal. Tumour and serum MGMT promoter
methylation and protein expression in glioblastoma
patients. Clin. Transl. Oncol. 13, 677685 (2011).
51. Esteller, M. etal. Detection of aberrant promoter
hypermethylation of tumor suppressor genes in serum
DNA from non-small cell lung cancer patients. Cancer
Res. 59, 6770 (1999).
52. Rosas, S.L. etal. Promoter hypermethylation patterns

of p16, O6methylguanine-DNA-methyltransferase,
and death-associated protein kinase in tumors and
saliva of head and neck cancer patients. Cancer Res.
61, 939942 (2001).
53. Guerrero-Preston, R. etal. NID2 and HOXA9
promoter hypermethylation as biomarkers for
prevention and early detection in oral cavity squamous
cell carcinoma tissues and saliva. Cancer Prev. Res. 4,
10611072 (2011).
54. Portela, A. & Esteller, M. Epigenetic modifications and
human disease. Nature Biotech. 28, 10571068
(2010).
55. Fraga, M.F. etal. Epigenetic differences arise during
the lifetime of monozygotic twins. Proc. Natl Acad. Sci.
USA 102, 1060410609 (2005).
56. Rakyan, V.K. etal. Identification of type 1 diabetesassociated DNA methylation variable positions that
precede disease diagnosis. PLoS Genet. 7, e1002300
(2011).
57. Goate, A. & Hardy, J. Twenty years of Alzheimers
disease-causing mutations. J.Neurochem. 120
(Suppl. 1), 38 (2012).
58. Mastroeni, D., McKee, A., Grover, A., Rogers, J. &
Coleman, P.D. Epigenetic differences in cortical
neurons from a pair of monozygotic twins discordant
for Alzheimers disease. PLoS ONE 4, e6617 (2009).
59. Heyn, H. etal. Distinct DNA methylomes of newborns
and centenarians. Proc. Natl Acad. Sci. USA 109,
1052210527 (2012).
60. Brock, M.V. DNA methylation markers and early
recurrence in stage I lung cancer. N.Engl. J.Med.
358, 11181128 (2008).
61. Graff, J.R., Gabrielson, E., Fujii, H., Baylin, S.B. &
Herman, J.G. Methylation patterns of the Ecadherin
5 CpG island are unstable and reflect the dynamic,
heterogeneous loss of Ecadherin expression during
metastatic progression. J.Biol. Chem. 275,
27272732 (2000).
62. Carmona, F.J. etal. Epigenetic disruption of
cadherin11 in human cancer metastasis. J.Pathol.
26Jul 2012 (doi:10.1002/path.4011).
63. Lujambio, A. etal. A microRNA DNA methylation
signature for human cancer metastasis. Proc. Natl
Acad. Sci. USA 105, 1355613561 (2008).
64. Milani, L. etal. DNA methylation for subtype
classification and prediction of treatment outcome in
patients with childhood acute lymphoblastic leukemia.
Blood 115, 12141225 (2010).
65. Hinoue, T. etal. Genome-scale analysis of aberrant
DNA methylation in colorectal cancer. Genome Res.
22, 271281 (2011).
66. Zhuang, J. etal. The dynamics and prognostic potential
of DNA methylation changes at stem cell gene loci in
womens cancer. PLoS Genet. 8, e1002517 (2012).
67. Dedeurwaerder, S. etal. DNA methylation profiling
reveals a predominant immune component in breast
cancers. EMBO Mol. Med. 3, 726741 (2011).
68. Fang, F. etal. Breast cancer methylomes establish an
epigenomic foundation for metastasis. Sci. Transl.
Med. 3, 75ra25 (2011).
69. Noushmehr, H. etal. Identification of a CpG island
methylator phenotype that defines a distinct subgroup
of glioma. Cancer Cell 17, 510522 (2010).
70. Turcan, S. etal. IDH1 mutation is sufficient to establish
the glioma hypermethylator phenotype. Nature 483,
479483 (2012).
71. Hegi, M.E. etal. MGMT gene silencing and benefit
from temozolomide in glioblastoma. N.Engl. J.Med.
352, 9971003 (2005).
72. Esteller, M., Hamilton, S.R., Burger, P.C., Baylin, S.B.
& Herman, J.G. Inactivation of the DNA repair gene
O6methylguanine-DNA methyltransferase by promoter
hypermethylation is a common event in primary
human neoplasia. Cancer Res. 59, 793797 (1999).
73. Stupp, R. etal. Phase I/IIa study of cilengitide and
temozolomide with concomitant radiotherapy followed
by cilengitide and temozolomide maintenance therapy
in patients with newly diagnosed glioblastoma. J.Clin.
Oncol. 28, 27122718 (2010).
74. Esteller, M. etal. Hypermethylation of the DNA repair
gene O6-methylguanine DNA methyltransferase and
survival of patients with diffuse large Bcell lymphoma.
J.Natl Cancer Inst. 94, 2632 (2002).
75. Chakravarti, A. etal. Temozolomide-mediated
radiation enhancement in glioblastoma: a report
on underlying mechanisms. Clin. Cancer Res. 12,
47384746 (2006).
76. King, M.C., Marks, J.H. & Mandell, J.B. Breast and
ovarian cancer risks due to inherited mutations in
BRCA1 and BRCA2. Science 302, 643646 (2003).
77. Esteller, M. etal. Promoter hypermethylation and

BRCA1 inactivation in sporadic breast and ovarian
tumors. J.Natl Cancer Inst. 92, 564569 (2000).
78. Stefansson, O.A. etal. CpG island hypermethylation
of BRCA1 and loss of pRb as cooccurring events in
basal/triple-negative breast cancer. Epigenetics 6,
638649 (2011).
79. Silver, D.P. etal. Efficacy of neoadjuvant cisplatin in
triple-negative breast Cancer J.Clin. Oncol. 28,
11451153 (2010).
80. Fong, P.C. etal. Inhibition of poly(ADP-ribose)
polymerase in tumors from BRCA mutation carriers.
N.Engl. J.Med. 361, 123134 (2009).
81. Veeck, J. etal. BRCA1 CpG island hypermethylation
predicts sensitivity to poly(adenosine diphosphate)ribose polymerase inhibitors. J.Clin. Oncol. 28,
e563e564; author reply e565e566 (2010).
82. Frommer, M. etal. A genomic sequencing protocol
that yields a positive display of 5methylcytosine
residues in individual DNA strands. Proc. Natl Acad.
Sci. USA 89, 18271831 (1992).
83. Lister, R. etal. Highly integrated single-base resolution
maps of the epigenome in Arabidopsis. Cell 133,
523536 (2008).
84. Lister, R. Human DNA methylomes at base resolution
show widespread epigenomic differences. Nature 462,
315322 (2009).
85. Laurent, L. etal. Dynamic changes in the human
methylome during differentiation. Genome Res. 20,
320331 (2010).
86. Lister, R. etal. Hotspots of aberrant epigenomic
reprogramming in human induced pluripotent stem
cells. Nature 471, 6873 (2011).
87. Li, Y. etal. The DNA methylome of human peripheral
blood mononuclear cells. PLoS Biol. 8, e1000533
(2010).
88. Meissner, A. etal. Reduced representation bisulfite
sequencing for comparative high-resolution DNA
methylation analysis. Nucleic Acids Res. 33,
58685877 (2005).
89. Bernstein, B.E. etal. The NIH Roadmap Epigenomics
Mapping Consortium. Nature Biotech. 28,
10451048 (2010).
90. Bock, C. etal. Reference maps of human ES and iPS
cell variation enable high-throughput characterization
of pluripotent cell lines. Cell 144, 439452 (2011).
91. Irizarry, R.A. etal. Comprehensive high-throughput
arrays for relative methylation (CHARM). Genome Res.
18, 780790 (2008).
92. Irizarry, R.A. etal. The human colon cancer
methylome shows similar hypo- and hypermethylation
at conserved tissue-specific CpG island shores. Nature
Genet. 41, 178186 (2009).
93. Kim, K. etal. Epigenetic memory in induced
pluripotent stem cells. Nature 467, 285290 (2010).
94. Kim, K. etal. Donor cell type can influence the
epigenome and differentiation potential of human
induced pluripotent stem cells. Nature Biotech. 29,
11171119 (2011).
95. Doi, A. Differential methylation of tissue- and cancerspecific CpG island shores distinguishes human
induced pluripotent stem cells, embryonic stem cells
and fibroblasts. Nature Genet. 41, 13501353 (2009).
96. McDonald, O.G., Wu, H., Timp, W., Doi, A. &
Feinberg, A.P. Genome-scale epigenetic reprogramming
during epithelial-tomesenchymal transition. Nature
Struct. Mol. Biol. 18, 867874 (2011).
97. Serre, D., Lee, B.H. & Ting, A.H. MBD-isolated
genome sequencing provides a high-throughput and
comprehensive survey of DNA methylation in the
human genome. Nucleic Acids Res. 38, 391399
(2010).
98. Taylor, B.S. etal. Frequent alterations and epigenetic
silencing of differentiation pathway genes in
structurally rearranged liposarcomas. Cancer Disc. 1,
587597 (2011).
99. Down, T.A. etal. A Bayesian deconvolution strategy
for immunoprecipitation-based DNA methylome
analysis. Nature Biotech. 26, 779785 (2008).
100. Jacinto, F.V., Ballestar, E. & Esteller, M. Methyl-DNA
immunoprecipitation (MeDIP): hunting down the
DNA methylome. BioTechniques 44, 3543 (2008).
101. Feber, A. etal. Comparative methylome analysis of
benign and malignant peripheral nerve sheath tumors.
Genome Res. 21, 515524 (2011).
102. Ball, M.P. etal. Targeted and genome-scale strategies
reveal gene-body methylation signatures in human
cells. Nature Biotech. 27, 361368 (2009).
103. Maunakea, A.K. Conserved role of intragenic DNA
methylation in regulating alternative promoters.
Nature 466, 253257 (2010).

REVIEWS
104. Dedeurwaerder, S. etal. Evaluation of the Infinium
Methylation 450K technology. Epigenomics 3,
771784 (2011).
105. Tost, J., Schatz, P., Schuster, M., Berlin, K. & Gut, I.G.
Analysis and accurate quantification of CpG
methylation by MALDI mass spectrometry. Nucleic
Acids Res. 31, e50 (2003).
106. Cocklin, R.R. & Wang, M. Identification of methylation
and acetylation sites on mouse histone H3 using
matrix-assisted laser desorption/ionization time
offlight and nanoelectrospray ionization tandem
mass spectrometry. J.Protein Chem. 22, 327334
(2003).
107. Ehrich, M. etal. Quantitative high-throughput analysis
of DNA methylation patterns by base-specific cleavage
and mass spectrometry. Proc. Natl Acad. Sci. USA
102, 1578515790 (2005).
108. Schatz, P., Dietrich, D. & Schuster, M. Rapid analysis
of CpG methylation patterns using RNase T1
cleavage and MALDI-TOF. Nucleic Acids Res. 32,
e167 (2004).
109. Kelly, T.K., De Carvalho, D.D. & Jones, P.A.
Epigenetic modifications as therapeutic targets.
Nature Biotech. 28, 10691078 (2010).
110. Yoo, C.B. etal. Delivery of 5aza2deoxycytidine to
cells using oligodeoxynucleotides. Cancer Res. 67,
64006408 (2007).
111. Kelly, W.K. etal. Phase I study of an oral histone
deacetylase inhibitor, suberoylanilide hydroxamic acid,
in patients with advanced cancer. J.Clin. Oncol. 23,
39233931 (2005).
112. Duvic, M. etal. Phase 2 trial of oral vorinostat
(suberoylanilide hydroxamic acid, SAHA) for refractory
cutaneous Tcell lymphoma (CTCL). Blood 109, 3139
(2007).
113. Garcia-Manero, G. etal. Phase 1 study of the histone
deacetylase inhibitor vorinostat (suberoylanilide
hydroxamic acid [SAHA]) in patients with advanced
leukemias and myelodysplastic syndromes. Blood 111,
10601066 (2008).
114. Whittaker, S.J. etal. Final results from a multicenter,
international, pivotal study of romidepsin in refractory
cutaneous Tcell lymphoma. J.Clin. Oncol. 28,
44854491 (2010).
115. Filippakopoulos, P. etal. Selective inhibition of BET
bromodomains. Nature 468, 10671073 (2010).
116. Dawson, M.A. etal. Inhibition of BET recruitment to
chromatin as an effective treatment for MLL-fusion
leukaemia. Nature 478, 529533 (2011).
117. Bernt, K.M. etal. MLL-rearranged leukemia is
dependent on aberrant H3K79 methylation by
DOT1L. Cancer Cell 20, 6678 (2011).
118. Krivtsov, A.V. etal. H3K79 methylation profiles define
murine and human MLLAF4 leukemias. Cancer Cell
14, 355368 (2008).
119. Daigle, S.R. Selective killing of mixed lineage leukemia
cells by a potent small-molecule DOT1L inhibitor.
Cancer Cell 20, 5365 (2011).
120. Chekhun, V.F. etal. Role of DNA hypomethylation in

the development of the resistance to doxorubicin in
human MCF7 breast adenocarcinoma cells. Cancer
Lett. 231, 8793 (2006).
121. Soengas, M.S. etal. Inactivation of the apoptosis
effector Apaf1 in malignant melanoma. Nature 409,
207211 (2001).
122. Iorns, E. etal. Identification of CDK10 as an important
determinant of resistance to endocrine therapy for
breast cancer. Cancer Cell 13, 91104 (2008).
123. Toyota, M. etal. Epigenetic inactivation of CHFR
in human tumors. Proc. Natl Acad. Sci. USA 100,
78187823 (2003).
124. Fan, M. etal. Diverse gene expression and DNA
methylation profiles correlate with differential
adaptation of breast cancer cells to the antiestrogens
tamoxifen and fulvestrant. Cancer Res. 66,
1195411966 (2006).
125. Taniguchi, T. etal. Disruption of the Fanconi anemiaBRCA pathway in cisplatin-sensitive ovarian tumors.
Nature Med. 9, 568574 (2003).
126. Dejeux, E. etal. DNA methylation profiling in
doxorubicin treated primary locally advanced breast
tumours identifies novel genes associated with survival
and treatment response. Mol. Cancer 9, 68 (2010).
127. Ibanez de Caceres, I. etal. IGFBP3 hypermethylationderived deficiency mediates cisplatin resistance in
non-small-cell lung cancer. Oncogene 29, 16811690
(2010).
128. Strathdee, G., MacKean, M.J., Illand, M. & Brown, R.
A role for methylation of the hMLH1 promoter in loss
of hMLH1 expression and drug resistance in ovarian
cancer. Oncogene 18, 23352341 (1999).
129. Faller, W.J. etal. Metallothionein 1E is methylated in
malignant melanoma and increases sensitivity to
cisplatin-induced apoptosis. Melanoma Res. 20,
392400 (2010).
130. Maier, S. etal. DNA-methylation of the homeodomain
transcription factor PITX2 reliably predicts risk of
distant disease recurrence in tamoxifen-treated, nodenegative breast cancer patientstechnical and clinical
validation in a multi-centre setting in collaboration
with the European Organisation for Research and
Treatment of Cancer (EORTC) PathoBiology group.
Eur. J.Cancer 43, 16791686 (2007).
131. Syed, N. etal. Polo-like kinase Plk2 is an epigenetic
determinant of chemosensitivity and clinical
outcomes in ovarian cancer. Cancer Res. 71,
33173327 (2011).
132. Lee, J.H. etal. Epigenetic alteration of PRKCDBP in
colorectal cancers and its implication in tumor cell
resistance to TNF-induced apoptosis. Clin. Cancer
Res. 17, 75517562 (2011).
133. Ramirez, J.L. etal. 1433 methylation in
pretreatment serum circulating DNA of cisplatin-plusgemcitabine-treated advanced non-small-cell lung
cancer patients predicts survival: the Spanish Lung
Cancer Group. J.Clin. Oncol. 23, 91059112 (2005).
134. Ferreri, A.J.M. etal. Aberrant methylation in the

promoter region of the reduced folate carrier gene is
a potential mechanism of resistance to methotrexate
in primary central nervous system lymphomas.
Br. J.Haematol. 126, 657664 (2004).
135. Tessema, M. etal. SULF2 methylation is prognostic for
lung cancer survival and increases sensitivity to
topoisomeraseI inhibitors via induction of ISG15.
Oncogene 12Dec 2011 (doi:10.1038/onc.2011.577).
136. Ebert, M.P.A. etal. TFAP2EDKK4 and
chemoresistance in colorectal cancer. N.Engl. J.Med.
366, 4453 (2012).
137. Ai, L. etal. The transglutaminase 2 gene (TGM2), a
potential molecular marker for chemotherapeutic drug
sensitivity, is epigenetically silenced in breast cancer.
Carcinogenesis 29, 510518 (2008).
138. Shen, L. etal. Drug sensitivity prediction by CpG
island methylation profile in the NCI60 cancer cell line
panel. Cancer Res. 67, 1133511343 (2007).
139. Agrelo, R. etal. Epigenetic inactivation of the premature
aging Werner syndrome gene in human cancer. Proc.
Natl Acad. Sci. USA 103, 88228827 (2006).
140. Sabatino, M.A. etal. Down-regulation of the
nucleotide excision repair gene XPG as a new
mechanism of drug resistance in human and murine
cancer cells. Mol. Cancer 9, 259 (2010).
Acknowledgements
The authors are supported by the European Research Council

Advanced Grant EPINORC under the agreement no. 268626,
the MICINN ProjectSAF201122803, the European
C o m m u n i t y s S eve n t h F ra m ewo r k P ro g ra m m e
(FP7/20072013) by the grant HEALTH-F52011282510
BLUEPRINT, Fondo de Investigaciones Sanitarias Grant
PI081345, the Dr. Josef Steiner Cancer Research
Foundation Award, Botin Foundation, Cellex Foundation and
the Health Department of the Catalan Government
(Generalitat de Catalunya). M.E. is an Institucio Catalana de
Recerca i Estudis Avanats (ICREA) Research Professor.
Competing interests statement
The authors declare no competing financial interests.
FURTHER INFORMATION
Groups homepage: http://www.pebc.cat
BLUEPRINT project: http://www.blueprint-epigenome.eu
Computational Epigenetics: http://www.computationalepigenetics.de
CURELUNG project: http://www.curelung.eu
Epigenomics at UCL: http://www.ucl.ac.uk/cancer/medicalgenomics/medgenhome
International Human Epigenome Consortium: http://www.
ihec-epigenomes.org
Roadmap Epigenomics Project: http://www.
roadmapepigenomics.org
ALL LINKS ARE ACTIVE IN THE ONLINE PDF

Heyn2012 PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Heyn2012 PDF

Uploaded by

Copyright:

Available Formats

REVIEWS

DNA methylation profiling in the clinic:

Abstract | Knowledge of epigenetic alterations in disease is rapidly increasing owing to

Cancer Epigenetics and

DNA methylation profiling for disease diagnosis

NATURE REVIEWS | GENETICS

VOLUME 13 | O CTOBER 2012 | 679

Box 1 | The technology revolution in epigenetic profiling

inter-individual and intra-individual variations. In this

680 | O CTOBER 2012 | VOLUME 13

detected, which makes epigenetic analysis applicable to a

NATURE REVIEWS | GENETICS

VOLUME 13 | O CTOBER 2012 | 681

Prostate, colon, lung, bladder

Androgen receptor signalling

DNA damage response

Colon, breast, oesophagus, gastric, liver

Lung, head and neck

Cell cycle control

Lymphoma, colon, stomach, prostate

Cell cycle control

Programmed cell death control

Lung, head and neck, bladder

Oestrogen receptor signalling

Prostate, liver, lung

Colon, lung, ovarian, prostate

Extracellular matrix protein

Colon, glioma, lymphoma, prostate, lung

miR34b and miR-34c

Colon, bladder, squamous cell carcinoma

Colon, endometrium, stomach

Cell growth control

Colon, liver, lung, thyroid

Glioma, breast, colon, gastric, lung

Retinoic acid receptor

Breast, colon, prostate

The diagnostic potential of histone marks faces

GSTP1 hypermethylation: a prostate cancer diagnosis

682 | O CTOBER 2012 | VOLUME 13

Cell cycle, DNA repair

Breast, ovarian, lung, prostate, colon

Cell growth control

Lymphoma, gastric, squamous cell carcinoma

Leukaemia, liver, thyroid, gastric, prostate

Cell cycle control

Cell cycle control

Lung, liver, kidney, leukemia

Cell growth control

Extracellular matrix protein

Extracellular matrix protein

Retinoic acid receptor responder

RAS association (RalGDS/AF6) domain family member 1 (RASSF1; 6895%), prostaglandin-endoperoxide

NATURE REVIEWS | GENETICS

VOLUME 13 | O CTOBER 2012 | 683

Prognosis and diagnosis

as pathologic stage31. Although these studies analysed

the hypermethylation levels of these genes in 243 primary

684 | O CTOBER 2012 | VOLUME 13

Single-loci DNA methylation profiling for disease

Moving towards clinical epigenetic tests. Although none

DNA methylation profiling for disease prognosis

NATURE REVIEWS | GENETICS