R E V I E W S
`
`Cell-free nucleic acids as biomarkers
`in cancer patients
`
`Heidi Schwarzenbach*, Dave S. B. Hoon‡ and Klaus Pantel*
`
`Abstract | DNA, mRNA and microRNA are released and circulate in the blood of cancer
`patients. Changes in the levels of circulating nucleic acids have been associated with tumour
`burden and malignant progression. In the past decade a wealth of information indicating the
`potential use of circulating nucleic acids for cancer screening, prognosis and monitoring of
`the efficacy of anticancer therapies has emerged. In this Review, we discuss these findings
`with a specific focus on the clinical utility of cell-free nucleic acids as blood biomarkers.
`
`In 1948, Mandel and Métais1 described the presence of
`cell-free nucleic acid (cfNA) in human blood for the first
`time. This attracted little attention in the scientific com-
`munity and it was not until 1994 that the importance
`of cfNA was recognized as a result of the detection of
`mutated RAS gene fragments in the blood of cancer
`patients2,3 (TIMELINE). In 1996, microsatellite altera-
`tions on cell-free DNA (cfDNA) were shown in cancer
`patients4, and during the past decade increasing atten-
`tion has been paid to cfNAs (such as DNA, mRNA and
`microRNAs (miRNAs)) that are present at high concentra-
`tions in the blood of cancer patients (FIG. 1). Indeed, their
`potential value as blood biomarkers was highlighted in a
`recent editorial in the journal Science5.
`Detecting cfNA in plasma or serum could serve as
`a ‘liquid biopsy’, which would be useful for numer-
`ous diagnostic applications and would avoid the need
`for tumour tissue biopsies. Use of such a liquid biopsy
`delivers the possibility of taking repeated blood sam-
`ples, consequently allowing the changes in cfNA to
`be traced during the natural course of the disease or
`during cancer treatment. However, the levels of cfNA
`might also reflect physiological and pathological
`processes that are not tumour-specific6. cfNA yields
`are higher in patients with malignant lesions than in
`patients without tumours, but increased levels have
`also been quantified in patients with benign lesions,
`inflammatory diseases and tissue trauma7. The physi-
`ological events that lead to the increase of cfNA during
`cancer development and progression are still not well
`understood. However, analyses of circulating DNA
`allow the detection of tumour-related genetic and epi-
`genetic alterations that are relevant to cancer develop-
`ment and progression. In addition, circulating miRNAs
`have recently been shown to be potential cancer
`biomarkers in blood.
`
`This Review focuses on the clinical utility of cfNA,
`including genetic and epigenetic alterations that can
`be detected in cfDNA, as well as the quantification of
`nucleo somes and miRNAs, and discusses the relationship
`between cfNA and micrometastatic cells.
`
`Biology of cfNA
`The release of nucleic acids into the blood is thought to
`be related to the apoptosis and necrosis of cancer cells
`in the tumour microenvironment. Secretion has also
`been suggested as a potential source of cfDNA (FIG. 1).
`Necrotic and apoptotic cells are usually phagocytosed
`by macrophages or other scavenger cells8. Macrophages
`that engulf necrotic cells can release digested DNA into
`the tissue environment. In vitro cell culture experiments
`indicated that macrophages can be either activated or
`dying during the process of DNA release8. Fragments
`of cellular nucleic acids can also be actively released9,10.
`It has been estimated that for a patient with a tumour that
`weighs 100 g, which corresponds to 3 × 1010 tumour
`cells, up to 3.3% of tumour DNA may enter the blood
`every day 11. On average, the size of this DNA varies
`between small fragments of 70 to 200 base pairs and
`large fragments of approximately 21 kilobases12.
`Tumour cells that circulate in the blood, and micro-
`metastatic deposits that are present at distant sites, such
`as the bone marrow and liver, can also contribute to the
`release of cfNA13,14.
`Tumours usually represent a mixture of different cancer
`cell clones (which account for the genomic and epig-
`enomic heterogeneity of tumours) and other normal cell
`types, such as haematopoietic and stromal cells. Thus,
`during tumour progression and turnover both tumour-
`derived and wild-type (normal) cfNA can be released
`into the blood. As such, the proportion of cfNA that
`originates from tumour cells varies owing to the state
`
`microRNAs
`Small non-coding RNA
`molecules that modulate the
`activity of specific mRNA
`molecules by binding and
`inhibiting their translation into
`polypeptides.
`
`*Institute of Tumour Biology,
`Center of Experimental
`Medicine, University Medical
`Center Hamburg-Eppendorf,
`Hamburg 20246, Germany.
`‡Department of Molecular
`Oncology, John Wayne
`Cancer Institute, Santa
`Monica, California 90404,
`USA.
`Correspondence to K.P.
`e-mail:
`pantel@uke.uni-hamburg.de
`doi:10.1038/nrc3066
`Published online 12 May 2011
`
`426 | JUNE 2011 | VOLUME 11
`
` www.nature.com/reviews/cancer
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
`00001
`
`EX1061
`
`

`

`R E V I E W S
`
` At a glance
`• Increased levels of circulating nucleic acids (DNA, mRNA and microRNA (miRNA)) in the blood reflect pathological
`processes, including malignant and benign lesions, inflammatory diseases, stroke, trauma and sepsis. During these
`processes nucleic acids are shed into the blood by apoptotic and necrotic cells.
`• In cancer patients, circulating DNA carries tumour-related genetic and epigenetic alterations that are relevant to
`cancer development, progression and resistance to therapy. These alterations include loss of heterozygosity (LOH)
`and mutations of tumour suppressor genes (such as TP53) and oncogenes (such as KRAS and BRAF).
`• Additional genetic alterations that are detectable on circulating DNA and used as biomarkers in cancer include the
`integrity of non-coding genomic DNA repeat sequences (such as ALU and LINE1). Although still in their infancy, DNA
`integrity assays have the potential to become a universal blood biomarker for multiple cancers.
`• Epigenetic alterations in genes (such as glutathione S-transferase P1 (GSTP1 and septin 9 (SEPT9)) and adenomatous
`polyposis coli (APC)) that are relevant to tumorigenesis and the progression of solid tumours have been detected on
`circulating DNA in cancer patients, and their potential clinical utility is indicated by the launch of commercial tests for
`cancer screening.
`• The detection of circulating nucleosomes in blood indicates that cell-free DNA (cfDNA) retains at least some features
`of the nuclear chromatin during the process of DNA release. Initial clinical studies have indicated that monitoring the
`abundance of nucleosomes has potential utility for monitoring the efficacy of therapy in cancer patients.
`• Dying tumour cells also discharge miRNAs, which circulate stably in the blood. The pivotal functions of miRNAs in
`cancer development and progression may explain the promising results of pilot studies on cancer patients using
`miRNA blood tests for tumour detection and prognosis.
`• The cellular source of tumour-derived circulating nucleic acids is still subject to debate. After complete removal of the
`primary tumour the detection of cfDNA may signal the presence of micrometastatic cells in distant organs, such as
`the bone marrow, which pose a risk of relapse.
`• Metastatic and primary tumours from the same patient can vary at the genomic, epigenomic and transcriptomic levels.
`Minimally invasive blood analyses of cell-free nucleic acid allow repetitive real-time monitoring of these events and
`will, therefore, gain clinical utility in the determination of prognosis and treatment efficacy.
`
`and size of the tumour. The amount of cfNA is also influ-
`enced by clearance, degradation and other physiological
`filtering events of the blood and lymphatic circulation.
`Nucleic acids are cleared from the blood by the liver and
`kidney and they have a variable half-life in the circulation
`ranging from 15 minutes to several hours7. Assuming an
`exponential decay model and plotting the natural loga-
`rithm of cfDNA concentration against time, serial DNA
`measure ments have shown that some forms of cfNA
`might survive longer than others. When purified DNA
`was injected into the blood of mice, double-stranded
`DNA remained in the circulation longer than single-
`stranded DNA15. Moreover, viral DNA as a closed ring may
`survive longer than linear DNA15. However, regardless
`of its size or configuration, cfDNA is cleared from the
`circulation rapidly and efficiently16. miRNAs seem to be
`highly stable, but their clearance rate from the blood has
`not yet been well studied in cancer patients owing to
`the novelty of this area of research. The nuclease activ-
`ity in blood may be one of the important factors for the
`turnover of cfNA. However, this area of cfNA physiology
`remains unclear and needs further examination.
`
`Circulating cfDNA
`DNA content. In patients with tumours of different histo-
`pathological types, increased levels of total cfDNA, which
`consists of epigenomic and genomic, as well as mito-
`chondrial and viral DNA, have been assessed by different
`fluorescence-based methods (such as, PicoGreen stain-
`ing and ultraviolet (UV) spectrometry) or quantitative
`PCR (such as, SYBR Green and TaqMan). Although
`cancer patients have higher cfDNA levels than healthy
`control donors, the concentrations of overall cfDNA
`
`vary considerably in plasma or serum samples in both
`groups17–19. A range of between 0 and >1,000 ng per ml
`of blood, with an average of 180 ng per ml cfDNA, has
`been measured20–23. By comparison, healthy subjects have
`concentrations between 0 and 100 ng per ml cfDNA of
`blood, with an average of 30 ng per ml cfDNA7. However,
`it is difficult to draw conclusions from these studies, as
`the size of the investigated patient cohort is often small
`and the techniques used to quantify cfDNA vary. A large
`prospective study assessed the value of plasma DNA
`levels as indicators for the development of neoplastic or
`pulmonary disease. The concentration of plasma DNA
`varied considerably between the European Prospective
`Investigation into Cancer and Nutrition (EPIC) centres
`that were involved in the study. This variation was pro-
`posed to be due to the type of population recruited and/or
`the treatment of the samples24. However, the quantifica-
`tion of cfDNA concentrations alone does not seem to be
`useful in a diagnostic setting owing to the overlapping
`DNA concentrations that are found in healthy individuals
`with those in patients with benign and malignant disease.
`The assessment of cfDNA concentration might prove to
`be useful in combination with other blood tumour bio-
`markers. Following surgery, the levels of cfDNA in cancer
`patients with localized disease can decrease to levels that
`are observed in healthy individuals25. However, when the
`cfDNA level remains high, it might indicate the presence
`of residual tumour cells17. Further studies are needed for
`the repeat assessment of quantitative cfNA in large cohorts
`of patients with well-defined clinical parameters. Such
`investigations will be crucial if we are to use cfDNA as a
`prognostic biomarker, as will the isolation and processing
`of cfNA to defined standards (discussed below).
`
`NATURE REVIEWS | CANCER
`
` VOLUME 11 | JUNE 2011 | 427
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
`00002
`
`

`

`R E V I E W S
`
`cfDNA is composed of both genomic DNA (gDNA)
`and mitochondrial DNA (mtDNA). Interestingly, the
`levels of cell-free mtDNA and gDNA do not correlate in
`some tumour types26,27, indicating the different nature of
`circulating mtDNA and gDNA. In contrast to two copies
`of gDNA, a single cell contains up to several hundred
`copies of mtDNA. Whereas gDNA usually circulates in a
`cell-free form, circulating mtDNA in plasma exists in both
`particle-associated and non-particle-associated forms28.
`Diverging results have been reported regarding whether
`cell-free mtDNA levels are increased and clinically
`relevant in cancer patients.
`The cfDNA can also include both coding and non-
`coding gDNA that can be used to examine microsatellite
`instability, loss of heterozygosity (LOH), mutations, poly-
`morphisms, methylation and integrity (size). In recent
`years, considerable attention has been paid to non-coding
`DNA, particularly repetitive sequences, such as ALU
`(which is a short interspersed nucleic element (SINE)) and
`as long interspersed nucleotide elements such as LINE1
`(REFS 29–31) (discussed below). ALU and LINE1 are dis-
`tributed throughout the genome and are known to be less
`methylated in cancer cells compared with normal cells32.
`
`Tumour-specific LOH. Genetic alterations found in
`cfDNA frequently include LOH that is detected using
`PCR-based assays13,18,33–38 (TABLE 1). Although similar
`plasma- and serum-based LOH detection methods have
`been used, a great variability in the detection of LOH
`in cfDNA has been reported. Despite the concordance
`between tumour-related LOH that is present in cfDNA
`in blood and LOH that is found in DNA isolated from
`matched primary tumours, discrepancies have also been
`found7. These contradictory LOH data that have
`been derived from blood and tumour tissue and the low
`incidence of LOH in cfDNA have partly been explained by
`technical problems and the dilution of tumour-associated
`cfDNA in blood by DNA released from normal cells11,39–41.
`Moreover, the abnormal proliferation of benign cells,
`
`owing to inflammation or tissue repair processes, for
`example, leads to an increase in apoptotic cell death, the
`accumulation of small, fragmented DNA in blood and
`the masking of LOH42.
`Alternative approaches, such as the detection of
`tumour-specific deletions are needed to better address
`the inherent problems of LOH analyses.
`
`Tumour-specific gene mutations. The analysis of
`cfDNA for specific gene mutations, such as those in
`KRAS and TP53, is desirable because these genes have
`a high mutation frequency in many tumour types and
`contribute to tumour progression43. Additionally, clini-
`cally relevant mutations in BRAF, epidermal growth
`factor receptor (EGFR) and adenomatous polyposis coli
`(APC) have now been studied in cfDNA. Several thera-
`peutic agents in clinical trials target the KRAS, BRAF,
`EGFR or p53 pathways44,45, and require the identifica-
`tion of the mutation status of the patient’s tumour to
`predict response to treatment. In this regard, cfDNA
`provides a unique opportunity to repeatedly monitor
`patients during treatment. In particular, in stage IV
`cancer patients, biopsies are not possible or repeat sam-
`pling of primary tumour and metastatic samples is not
`practical or ethical.
`The major problem with this approach has been
`assay specificity and sensitivity. Assays targeting
`cfDNA mutations require that the mutation in the
`tumour occurs frequently at a specific genomic site.
`A major drawback of cfDNA assays is the low frequency of
`some of the mutations that occur in tumours. In general,
`wild-type sequences often interfere with cfDNA muta-
`tion assays. This is due to the low level of cfDNA
`mutations and the dilution effect of DNA fragments
`and wild-type DNA in circulation. In PCR-based assays
`technological design can significantly limit the assay
`sensitivity and specificity. An example is the KRAS muta-
`tion tissue assay that can frequently detect mutations in
`tumour tissues, such as the pancreas, colon and lung;
`
`Timeline | Detection of various forms of cfDNA in patients with different types of cancer
`
`• Melanoma and breast
`cancer microsatellite LOH
`• Breast cancer CpG island
`methylation
`
`• Lung cancer
`microsatellite LOH
`• HPV DNA in
`cervical cancer
`
`Ovarian and cervical
`cancer CpG island
`methylation
`
`• HBV DNA in
`hepatocellular cancer
`• Prostate cancer and
`melanoma CpG island
`methylation
`
`• Ovarian cancer DNA integrity
`• Lung cancer KRAS mutation and
`melanoma BRAF mutation
`• Oesophageal cancer CpG
` island methylation
`
`Nasopharyngeal
`carcinoma and
`bladder cancer
`DNA integrity
`
`1999
`
`2000
`
`2001
`
`2003
`
`2004
`
`2005
`
`2006
`
`2007
`
`2008
`
`• Pancreatic cancer
` KRAS mutation
`• Breast cancer
` TP53 mutation
`
`Detection of EBV DNA
`in nasopharyngeal
`carcinoma
`
`Colorectal cancer
`KRAS mutation
`
`• Lung cancer CpG island
`methylation
`• Hepatocellular and ovarian
`cancer TP53 mutation
`• Oral cancer microsatellite LOH
`
`• Hepatocellular cancer
`microsatellite LOH
`• Breast cancer DNA integrity
`• Lung cancer EGFR mutation
`
`• Hepatocellular cancer and
`neuroblastoma CpG island
`methylation
`• Prostate cancer
`microsatellite LOH
`
`The development of the detection of genetic and epigenetic alterations, as well as the measurement of DNA integrity and viral DNA, in blood from patients with different
`tumour types over the past decade is shown. We show only significant, prognostic findings from >40 patients with serum, plasma or bodily fluid detection of cell-free DNA
`(cfDNA) from individual cancers. This timeline is not meant to be comprehensive and is based on our own personal view of what have been important clinical
`translational events. EBV, Epstein–Barr virus; EGFR, epidermal growth factor receptor; HBV, hepatitis B virus; HPV, human papilloma virus; LOH, loss of heterozygosity.
`
`428 | JUNE 2011 | VOLUME 11
`
` www.nature.com/reviews/cancer
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
`00003
`
`

`

`R E V I E W S
`
`Tumour
`
`Blood vessel
`
`Mutation or
`polymorphism
`
`CG
`
`TA
`
`C
`G
`
`A
`G
`
`G
`C
`
`T
`A
`TA
`
`CH3
`
`T A
`
`C G
`
`Methylation
`CH3
`C
`G
`C
`G
`
`C
`G
`
`A A
`T T
`
`Viral DNA
`
`DNA integrity
`
`Microsatellite
`alteration
`
`Mitochondria
`
`Nucleosome
`
`Exosome
`
`Necroptosis
`
`Golgi
`
`Apoptosis
`
`Lysosome
`
`Secretion
`
`Figure 1 | Cell-free nucleic acids in the blood. Mutations, methylation, DNA integrity, microsatellite alterations and
`Nature Reviews | Cancer
`viral DNA can be detected in cell-free DNA (cfDNA) in blood. Tumour-related cfDNA, which circulates in the blood of
`cancer patients, is released by tumour cells in different forms and at different levels. DNA can be shed as both
`single-stranded and double-stranded DNA. The release of DNA from tumour cells can be through various cell
`physiological events such as apoptosis, necrosis and secretion. The physiology and rate of release is still not well
`understood; tumour burden and tumour cell proliferation rate may have a substantial role in these events. Individual
`tumour types can release more than one form of cfDNA.
`
`however, cfDNA mutation assays using blood sam-
`ples have not yet been concordantly successful46–48.
`New approaches are needed, such as cfDNA sequenc-
`ing. The BRAF mutation V600E, which is present in
`>70% of metastatic melanomas, can be detected
`in cfDNA and has been shown to be useful in monitor-
`ing patients with melanoma who are receiving ther-
`apy 49. This mutation has been detected in different
`stages of melanoma (according to the American Joint
`Committee on Cancer (AJCC) Cancer Staging Manual)
`using a quantitative real-time clamp PCR assay, with the
`highest levels found in the more advanced stages49. This
`is one of the first major studies to demonstrate that
`cfDNA mutation assays have the sensitivity to monitor
`patient responses before and after treatment. The util-
`ity of a cfDNA BRAF mutation assay has gained more
`importance, as new anti-BRAF drugs, such as PLX4032
`(Roche)50 and GSK2118436 (GlaxoSmithKline)51, have
`shown substantial responses in patients in early clinical
`trials. EGFR mutations that occur in a specific subset of
`patients with lung cancers52–54 make these tumours sen-
`sitive to EGFR-targeted therapies; however, the detec-
`tion of EGFR mutations in cfDNA has not been well
`developed owing to issues with sensitivity and specifi-
`city. Patients whose tumours have a specific gene muta-
`tion would be strong candidates for monitoring of their
`cfDNA in blood for the respective specific mutation.
`However, sensitivity, specificity and validation need
`to be carried out in multicentre settings to determine
`true clinical utility. Alternatively, cfDNA assays might
`be more appropriate when used with other biomarker
`
`assays, and this might be applicable to personalized
`medicine, rather than diagnostic screens that can be
`used across a wide group of cancer patients.
`
`DNA integrity. Another assay that is applicable to cfDNA
`that has gained interest in recent years is the integrity
`of non-coding gDNA, such as the repeat sequences of
`ALU and LINE1. The ALU and LINE1 sequences have
`been referred to as ‘junk DNA’; however, in recent
`years evidence has indicated their importance in
`various physiological events, such as DNA repair,
`transcription, epigenetics and transposon-based activ-
`ity55,56. Approximately 17–18% of the human genome
`consists of LINE1. In normal cells LINE1 sequences
`are heavily methylated, restricting the activities of
`these retrotransposon elements and thus preventing
`genomic instability. LINE1 sequences are moderately
`CpG-rich, and most methylated CpGs are located
`in the 5′ region of the sequence that can function as
`an internal promoter 23. These forms of DNA can be
`detected as cfDNA of different sizes, but also as methyl-
`ated and unmethylated DNA. Studies on these types of
`cfDNA are still in their infancy; however, recent studies
`have shown potential prognostic and diagnostic util-
`ity23,29–31. The assays are based on the observation that
`common DNA repeat sequences are preferentially
`released by tumour cells that are undergoing non-
`apoptotic or necrotic cell death, and these fragments
`can be between 200 bp and 400 bp in size. The ALU and
`LINE1 sequences are well interspersed throughout the
`genome on all chromosomes, so although specificity
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
` VOLUME 11 | JUNE 2011 | 429
`
`Quantitative real-time
`clamp PCR assay
`A technique that uses a
`peptide nucleic acid clamp and
`locked nucleic acid probes,
`which are DNA synthetic
`analogues that hybridize to
`complementary DNA and are
`highly sensitive and specific for
`recognizing single base pair
`mismatches.
`
`NATURE REVIEWS | CANCER
`
`00004
`
`

`

`R E V I E W S
`
`for an individual cancer type is lost in these assays, sen-
`sitivity is enhanced. Using a PCR assay, the integrity
`of cfDNA ALU sequences in blood has been shown to
`be sensitive for the assessment of the early stages of
`breast cancer progression, including micro metastasis30.
`DNA integrity cfDNA assays have also been used in
`
`testicular, prostate, nasopharyngeal and ovarian
`cancer31,57–59. These assays are still in their infancy and
`address an important challenge of whether a ‘universal’
`blood biomarker for multiple cancers can be of clinical
`utility. Further validation of these assays will also
`determine their clinical utility in specific cancers.
`
`Table 1 | Detection of cfDNA and its alterations in patients with different tumour types*
`Cancer
`cfDNA
`Diagnostic
`Prognostic
`Bladder
`DNA integrity
`•
`•
`Methylation
`•
`Microsatellite alterations
`•
`Methylation
`•
`Microsatellite alterations
`•
`DNA integrity
` 
`
`Breast
`
` 
`•
`•
`•
`•
` 
`•
` 
`•
` 
`•
`•
`•
`•
`•
` 
`•
`•
`•
`•
`•
`•
` 
`•
` 
`•
`•
`•
` 
`•
`•
`•
`•
`•
` 
`•
` 
`•
` 
`•
` 
`•
` 
`•
`•
`•
`•
`•
`13,38
` 
`Microsatellite alterations
`•
`180
`DNA integrity
`•
`•
` 
`Mitochondrial
`26,181
`•
`*This table represents different forms of cell-free nucleic acid (cfNA) that have been detected in patients with the most prevalent
`cancers in both males and females182. This table is not meant to be comprehensive and is based on our own view of studies that
`offer substantial clinical insight. cfDNA, cell-free DNA.
`
`Cervical
`
`Colorectal
`
`Hepatocellular
`carcinoma
`
`Lung
`
`Non-Hodgkin’s
`lymphoma
`
`Melanoma
`
`Ovarian
`
`Pancreatic
`
`Prostate
`
` 
`•
`•
`•
`•
`•
`•
`•
` 
`•
`•
`•
` 
`•
`•
`
`Mutation
`Mitochondrial
`Methylation
`Viral DNA
`Mutation
`DNA integrity
`Methylation
`Methylation
`Microsatellite alterations
`Mutation
`DNA integrity
`Viral DNA
`Mutation
`Methylation
`Microsatellite alterations
`Mutation
`Viral DNA
`Methylation
`DNA integrity
`Mutation
`Methylation
`Microsatellite alterations
`Methylation
`DNA integrity
`Mutation
`Mitochondrial
`Methylation
`DNA integrity
`Mutation
`Methylation
`
`Refs
`123
`124
`125
`126–130
`33–35
`30,131
`
`34
`132
`133,134
`135
`47,136–139
`31
`136,140–143
`144–146
`147
`148,149
`29
`150
`48,53,151,152
`153–157
`36,37
`158
`159–161
`162
`162
`49,163,164
`111,115
`165–168
`169,170
`59
`171
`172
`173,174
`31
`46
`38,175–179
`
`430 | JUNE 2011 | VOLUME 11
`
` www.nature.com/reviews/cancer
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
`00005
`
`

`

`Epigenetic alterations. Epigenetic alterations can have
`a substantial effect on tumorigenesis and progression
`(BOX 1). Several studies have revealed the presence of
`methylated DNA in the serum or plasma of patients with
`various types of malignancy (TABLE 1). The detection of
`methylated cfDNA represents one of the most promising
`approaches for risk assessment in cancer patients.
`Assays for the detection of promoter hypermethylation
`may have a higher sensitivity than microsatellite analyses,
`and can have advantages over mutation analyses. In gen-
`eral, aberrant DNA methylation, which seems to be com-
`mon in cancer, occurs at specific CpG dinucleotides60.
`The acquired hypermethylation of a specific gene can be
`detected by sodium bisulphite treatment of DNA, which
`converts unmethylated (but not methylated) cytosines
`to uracil. The modified DNA is analysed using either
`methylation-specific PCR, with primers that are specific
`for methylated and unmethylated DNA, or DNA sequenc-
`ing61. Nevertheless, to improve the assay conditions
`and the clinical relevance, the selection of appropriate
`tumour-related genes from a long list of candidate genes
`that are known to be methylated in neoplasia is essential.
`Although epigenetic alterations are not unique for a single
`tumour entity, there are particular tumour suppressor
`genes that are frequently methylated and downregulated
`in certain tumours62,63. For example, important epigenetic
`events in carcinogenesis include the hypermethylation
`of the promoter region of the genes pi-class glutathione
`S-transferase P1 (GSTP1) and APC, which are the most
`common somatic genome abnormalities in prostate
`and colorectal cancer, respectively62,63. Other important
`methylated genes that have shown prognostic utility using
`cfDNA assays in significant numbers of patients include
`RAS association domain family 1A (RASSF1A), retinoic
`acid receptor-b (RARB), septin 9 (SEPT9), oestrogen
`receptor-a (ESR1) and cyclin-dependent kinase inhibitor
`2A (CDKN2A) (TABLE 1). The first commercial real-time
`PCR plasma test for the detection of early colorectal cancer
`(developed by Epigenomics AG and Abbott Molecular) is
`for the detection of SEPT9. This biomarker is still under-
`going validation, but it demonstrates the potential diag-
`nostic screening utility of methylated tumour-related
`cfDNA to differentiate cancer patients from healthy
`individuals and to identify the tumour type.
`It is also possible to detect tumour-related alterations
`in histone modifications in the blood. By monitoring
`changes in the circulating histones and DNA methylation
`pattern, the antitumour effects of histone deacetylase
`
` Box 1 | Epigenetics
`Epigenetic changes can include the methylation of gene promoter regions and histone
`modifications. In chromosomal regions where tumour-associated genes reside,
`epigenetic modifications may affect important regulatory mechanisms that normally
`limit malignant transformation60. Inactivation of tumour suppressor genes by promoter
`hypermethylation is thought to have a crucial role in this process117. DNA methylation of
`the cytosine base in CpG dinucleotides, which are found as isolated or clustered CpG
`islands, induces gene repression by inhibiting the access of transcription factors to their
`binding sites, and by recruiting methyl-CpG-binding proteins (MBDs) to methylated
`DNA together with histone-modifying enzymes118. Epigenetic modifications also alter
`the packing of nucleosomes that are implicated in transcriptional regulation119,120.
`
`R E V I E W S
`
`and histone methyltransferase inhibitors may be evalu-
`ated and consequently allow a better screening of cancer
`patients64,65.
`
`Circulating nucleosomes. Circulating gDNA that is
`derived from tumours seems to predominantly exist as
`mononucleosomes and oligonucleosomes, or it is bound
`to the surface of blood cells by proteins that harbour
`specific nucleic acid-binding properties66. A nucleosome
`consists of a histone octamer core wrapped twice by a
`200 bp-long DNA strand. Under physiological condi-
`tions these complexes are packed in apoptotic particles
`and engulfed by macrophages67. However, an excess of
`apoptotic cell death, as occurs in large and rapidly pro-
`liferating tumours or after chemotherapy treatment, can
`lead to a saturation of apoptotic cell engulfment and
`thus increased nucleosome levels in the blood68. The
`detection of circulating nucleosomes that are associ-
`ated with cfDNA suggests that DNA in blood retains at
`least some features of the nuclear chromatin during the
`process of release.
`Enzyme-linked immunosorbent assays (ELISAs) have
`been developed to quantify circulating nucleosomes.
`As increased concentrations are found in both benign
`and malignant tumours, high nucleosome levels in blood
`are not indicators of malignant disease69. However, the
`observed changes in apoptosis-related deregulation of
`proteolytic activities along with the increased serum
`levels of nucleosomes have been linked to breast cancer
`progression70. As typical cell-death products, the quan-
`tification of circulating nucleosomes seems to be valu-
`able for monitoring the efficacy of cytotoxic cancer
`therapies71. For example, platinum-based chemotherapy
`induces caspase-dependent apoptosis of tumour cells
`and an increase in circulating nucleosomes in the blood
`of patients with ovarian cancer17. Moreover, the outcome of
`therapy can be indicated by nucleosome levels during the
`first week of chemotherapy and radiotherapy in patients
`with lung, pancreatic and colorectal cancer, as well as in
`patients with haematological malignancies71.
`
`Viral DNA. Viral cfDNA can also be detected in some
`tumour types. Viruses, such as human papillomavirus
`(HPV), hepatitis B virus (HBV) and Epstein–Barr virus
`(EBV), are aetiological factors in various malignancies,
`such as nasopharyngeal, cervical, head and neck, and
`hepatocellular cancer and lymphoma72–75. Their specific
`DNA may have the potential to be used as molecular
`biomarkers for neoplastic disease. Associations between
`circulating viral DNA and disease have been reported
`for EBV with Hodgkin’s disease, Burkitt’s lymphoma and
`nasopharyngeal carcinoma; for HBV with some forms
`of hepatic cell carcinoma; and for HPV with head and
`neck, cervical and hepatocellular cancers (TABLE 1). The
`clinical utility of EBV cfDNA in diagnosis and prognosis
`of nasopharyngeal carcinoma has been demonstrated in
`multiple studies with large cohorts of patients76–80, and the
`use of this cfDNA has became one of the leading cfDNA
`blood tests for the assessment of nasopharyngeal carci-
`noma progression in Hong Kong, Taiwan and China,
`where this cancer is highly prevalent77,78,81. The limitation
`
`NATURE REVIEWS | CANCER
`
` VOLUME 11 | JUNE 2011 | 431
`
`© 2011 Macmillan Publishers Limited. All rights reserved
`
`00006
`
`

`

`R E V I E W S
`
`of most viral cfNA assays is that benign viral infections
`that are caused by the same viruses can complicate the
`interpretation of results, particularly in diagnostic screen-
`ing. Establishing clinically meaningful cut-off levels is
`important to move these screens into the clinic.
`
`Genometastasis. The genometastasis hypothesis
`describes the horizontal transfer of cell-free tumour
`DNA to other cells that results in transformation. If true,
`metastases could develop in distant organs as a result of a
`transfection-like uptake of dominant oncogenes that are
`released from the primary tumour82. García-Olmo et al.83
`showed that plasma isolated from patients with colon
`cancer is able to transform NIH-3T3 cells and that these
`cells can form carcinomas when injected into non-obese
`diabetic-severe combined immunodeficient mice83.
`Whether this biological function of circulating DNA has
`relevance in human blood is an aspect to be considered
`in the future.
`
`cfDNA assay issues
`One of the problems in evaluating cfNA is the standard-
`ization of assays, such as isolation technologies,
`standards, assay conditions, and specificity and sensitivity7.
`It remains controversial whether plasma