Analytical Biochemistry 433 (2013) 227–234
`
`Contents lists available at SciVerse ScienceDirect
`
`Analytical Biochemistry
`
`j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / y a b i o
`
`Mini-Review
`Mutation-based detection and monitoring of cell-free tumor DNA in peripheral
`blood of cancer patients
`L. Benesova a,b, B. Belsanova a,b, S. Suchanek c,d,e, M. Kopeckova a,b, P. Minarikova c,d,e, L. Lipska f,g, M. Levy f,g,
`V. Visokai f,g, M. Zavoral c,d,e, M. Minarik a,b,⇑
`
`a Laboratory of Molecular Genetics and Oncology, Genomac Research Institute, 155 41 Prague, Czech Republic
`b Center for Applied Genomics of Solid Tumors (CEGES), Genomac Research Institute, 155 41 Prague, Czech Republic
`c Institute for Postgraduate Medical Education, 100 05 Prague, Czech Republic
`d Department of Internal Medicine, 1st Faculty of Medicine, Charles University in Prague, 169 02 Prague, Czech Republic
`e Central Military Hospital, 169 02 Prague, Czech Republic
`f Department of Surgery, 1st Faculty of Medicine, Charles University in Prague, 140 59 Prague, Czech Republic
`g Thomayer Teaching Hospital, 140 59 Prague, Czech Republic
`
`a r t i c l e
`
`i n f o
`
`a b s t r a c t
`
`Article history:
`Received 23 March 2012
`Received in revised form 15 June 2012
`Accepted 20 June 2012
`Available online 28 June 2012
`
`Keywords:
`cfDNA
`Cancer
`Mutations
`Capillary electrophoresis
`BEAMing
`COLD–PCR
`Digital PCR
`
`Prognosis of solid cancers is generally more favorable if the disease is treated early and efficiently. A key
`to long cancer survival is in radical surgical therapy directed at the primary tumor followed by early
`detection of possible progression, with swift application of subsequent therapeutic intervention reducing
`the risk of disease generalization. The conventional follow-up care is based on regular observation of
`tumor markers in combination with computed tomography/endoscopic ultrasound/magnetic reso-
`nance/positron emission tomography imaging to monitor potential tumor progression. A recent develop-
`ment in methodologies allowing screening for a presence of cell-free DNA (cfDNA) brings a new viable
`tool in early detection and management of major cancers. It is believed that cfDNA is released from
`tumors primarily due to necrotization, whereas the origin of nontumorous cfDNA is mostly apoptotic.
`The process of cfDNA detection starts with proper collection and treatment of blood and isolation and
`storage of blood plasma. The next important steps include cfDNA extraction from plasma and its detec-
`tion and/or quantification. To distinguish tumor cfDNA from nontumorous cfDNA, specific somatic DNA
`mutations, previously localized in the primary tumor tissue, are identified in the extracted cfDNA. Apart
`from conventional mutation detection approaches, several dedicated techniques have been presented to
`detect low levels of cfDNA in an excess of nontumorous (nonmutated) DNA, including real-time polymer-
`ase chain reaction (PCR), ‘‘BEAMing’’ (beads, emulsion, amplification, and magnetics), and denaturing
`capillary electrophoresis. Techniques to facilitate the mutant detection, such as mutant-enriched PCR
`and COLD–PCR (coamplification at lower denaturation temperature PCR), are also applicable. Finally, a
`number of newly developed miniaturized approaches, such as single-molecule sequencing, are promising
`for the future.
`
`Ó 2012 Elsevier Inc. All rights reserved.
`
`Introduction
`
`Although significant progress has been made in the develop-
`ment of new therapy approaches, cancer remains the leading cause
`of death worldwide [1]. Despite the availability of a number of
`screening schemes, in most cases cancer remains undetected until
`its advanced stages [2,3]. In a typical course of disease develop-
`ment, defective cellular adhesion allows malignant cells to be re-
`leased and to travel to nearby structures or even migrate through
`the lymphatic or blood system to form malignant formations. If
`⇑ Corresponding author at: Laboratory of Molecular Genetics and Oncology,
`
`Genomac Research Institute, 155 41 Prague, Czech Republic. Fax: +420 220513448.
`E-mail address: mminarik@email.com (M. Minarik).
`
`0003-2697/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
`http://dx.doi.org/10.1016/j.ab.2012.06.018
`
`unnoticed, such micrometastases pose a serious risk for disease
`progression already in early stages of the primary tumor. Surgical
`treatment resulting in removal of the primary tumor, therefore,
`might not avert dissemination and generalization of the disease
`in the long term. Follow-up of cancer patients typically relies on
`computed tomography (CT),1 positron emission tomography
`
`1 Abbreviations used: CT, computed tomography; PET, positron emission tomogra-
`phy; cfDNA, cell-free DNA; PCR, polymerase chain reaction; LOH, loss of heterozy-
`gosity; DCE, denaturing capillary electrophoresis; ME–PCR, mutant-enriched PCR;
`BEAMing, beads, emulsion, amplification, and magnetics; CE, capillary electrophore-
`sis; DCE, denaturing capillary electrophoresis; dHPLC, denaturing high-performance
`liquid chromatography; COLD–PCR, coamplification at lower denaturation tempera-
`ture PCR.
`
`00001
`
`EX1041
`
`

`

`228
`
`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`(PET)–CT, or magnetic resonance imaging in combination with mon-
`itoring of serum tumor markers [4,5]. It is known that imaging meth-
`ods typically spot objects on a millimeter scale (containing tens to
`hundreds of millions of cells). In addition, the widely established
`utility of tumor markers is sometimes inefficient due to sensitivity
`and specificity issues [6,7]. Therefore, there is great expectation in
`finding new diagnostic markers for better management of all major
`cancers. Among the few alternatives, there is growing interest in
`molecular diagnostics directed at nucleic acids released directly
`from the tumor and circulating in peripheral blood of patients.
`The classic article on the occurrence of nucleic acids in human
`plasma was published back in 1948 by Mandel and Metais [8], fol-
`lowed by works of Bendich and coworkers in 1965 [9], Koffler and
`coworkers in 1973 [10], and Leon and coworkers in 1977 [11], who
`identified the importance of circulating tumor DNA as a vehicle of
`oncogenesis. With the use of then emerging methods such as
`radioimmunoassay, the presence of higher cell-free DNA (cfDNA)
`concentrations in serum in patients with carcinoma compared
`with healthy persons was observed along with a decrease after
`the administration of chemotherapy [10,11]. It was soon recog-
`nized that the circulating DNA could serve as a viable tool to mon-
`itor the efficiency of anticancer therapies by monitoring its levels
`in advanced cancers [12–14]. With the subsequent rapid develop-
`ment of modern polymerase chain reaction (PCR)-based tech-
`niques and their widespread availability, the interest in detecting
`circulating nucleic acids is steadily increasing, with colorectal,
`prostate, and lung cancers being the main focus of many published
`studies [15–19].
`
`Origin of free circulating nucleic acids
`
`Circulating nucleic acids, often referred to as cell-free DNA,
`emphasize their exogenous nature in comparison with DNA origi-
`nating from nuclei of the blood cells. Whereas cfDNA detection is
`currently at the forefront of molecular oncology community inter-
`est, the detail mechanism of cfDNA release from its native cell is
`yet to be fully elucidated. In 2001, a study by Jahr and coworkers
`revealed a combined contribution of apoptotic and necrotic pro-
`cesses to the overall production of cfDNA [20]. The idea was ex-
`tended in further detail by Diehl and coworkers [21]. These
`authors considered that DNA fragments present in the circulation
`originate from the necrotic neoplastic cells phagocytized by mac-
`rophages, and these also engulf nontumor (apoptotic) cells, which
`is the reason why a particular level of nontumor cfDNA occurs in
`healthy individuals, as confirmed by others [22]. The two proposed
`hypothetical mechanisms for necrotic and apoptotic release of
`DNA are illustrated in Fig. 1. Fig. 1A depicts a mucous membrane
`of the colon affected by a growing tumor with a layer of necrotic
`cells on the surface. The necrotic tumor cell is released, and its
`fragments are captured by the macrophage pseudopodia. An en-
`gulfed fragment forms a phagosome, which fuses with lysosome
`to form a phagolysosome. Subsequently, the ingested particles,
`including tumor DNA fragment of various lengths, are released into
`the environment. Fig. 1B shows an alternative mechanism with the
`mucous membrane of the colon with a normal epithelium layer
`releasing a cell undergoing apoptosis. The cell forms apoptotic par-
`ticles captured by the macrophage pseudopodia. The engulfed par-
`ticle forms a phagosome, which fuses with lysosome to form a
`phagolysosome. Subsequently, the ingested particles, including
`equally sized DNA, are released into the environment.
`Necrotic cells arise in invasive tumors, where tissue deteriora-
`tion occurs as a result of hypoxia [23]. Benign tumors do not have
`this property, and the amount of fragmented DNA produced is
`minimal [24,25]. This implies that malignity of the tumor leads
`to a higher degree of necrosis with a corresponding increase in cir-
`culating tumor DNA. Necrosis, however, affects surrounding non-
`
`tumor cells as well, leading to a parallel release of nontumorous
`cfDNA into the circulation, resulting in an increase in concentra-
`tions of both tumorous and nontumorous DNA in plasma. Thus,
`the fact that in patients with malignant disease the volume of free
`DNA is increased, regardless of its origin, can in some circum-
`stances be used for monitoring of cancer.
`
`Extraction of cfDNA
`
`To successfully detect the presence of tumor cfDNA in plasma or
`serum, a suitable methodology needs to be selected. This consists
`of several essential steps. The first step is to process collected
`blood while avoiding the rupture of blood cell membranes (i.e.,
`hemolysis) and subsequent plasma contamination with DNA
`derived from the blood cell nuclei. The next step is cfDNA isolation
`when it is necessary to select the most appropriate method to
`gain a sufficient amount of quality DNA for further analysis.
`Here, the essential tool is PCR, and the detection of amplified prod-
`ucts can be done either directly in real time (real-time PCR) or fol-
`lowing amplification by electrophoresis on a slab gel or capillary
`format.
`
`Blood collection, cfDNA isolation, and quantification
`
`Blood samples collected in an anticoagulant solution must be
`processed within 2 h after the collection to avoid damage to nucle-
`ated blood cells and release of their DNA [26]. In some studies sam-
`pling was performed in heparinized test tubes [27,28], and in
`others it was performed in tubes containing EDTA (ethylenedi-
`aminetetraacetic acid) solution [29]. Immediately after blood col-
`lection, plasma needs to be separated from the blood cells by
`centrifugation. The centrifugal speed must not be too high (to
`avoid causing cell lysis), and recentrifugation may be performed
`following primary elimination of blood cells at lower speeds [27].
`Alternatively, plasma may be extracted by filtration through mem-
`branes with a 0.45-lm pore size [29]. Extracted plasma may be
`stored at 20 °C for extended periods of time before subsequent
`processing. Some authors have reported the use of serum rather
`than plasma [30]. It was noted that serum is a less suitable mate-
`rial because it becomes readily contaminated with DNA from the
`leukocytes when blood coagulum is formed [26].
`The basis for successful cfDNA detection is selection of an isola-
`tion method that ensures extraction of a sufficient amount of qual-
`ity DNA. A classic phenol–chloroform extraction or commercial kits
`based on the principle of membrane columns can be applied. The
`advantage of the phenol–chloroform method is an unlimited
`amount of input material; thus, the yield can be higher with an in-
`creased volume of isolated plasma, whereas in commercial kits the
`amount of input material is limited. However, the use of commer-
`cial kits is considerably easier, and special kits designed specifically
`for cfDNA isolation are already available. Kuang and coworkers
`[28] compared three isolating kits: QIAamp DNA Micro Kit (Qia-
`gen), NucleoSpin Plasma XS (Macherey–Nagel), and Wizard (Pro-
`mega).
`In this study, the authors used a method described
`previously [31] based on a principle that the majority of tumor
`DNA in plasma derived from necrotic cells occurs in unequally
`sized fragments ranging from 185 to 926 bp, whereas the DNA
`from apoptotic (i.e., nontumor) cells is usually present in relatively
`uniform sizes ranging from 185 to 200 bp. Based on these facts,
`they examined the amounts of both cfDNA types using real-time
`PCR by amplifying two different length fragments of Alu se-
`quences, namely, 115 and 247 bp. Alu 115 captured the concentra-
`tion of short DNA fragments derived from apoptotic cells as well as
`DNA fragments from tumor cells, whereas Alu 247 captured only
`the concentration of tumor DNA. From a subsequent Alu 247/115
`ratio, they calculated the concentration of DNA derived only from
`
`00002
`
`

`

`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`229
`
`Fig.1. Alternative mechanisms of cfDNA release during phagocytosis. Unequally sized DNA fragments result from phagocytosis of a necrotic cell (A), whereas uniformly sized
`DNA fragments are released by macrophage from apoptotic cell (B).
`
`tumor cells. This led to a finding that although the greatest concen-
`tration of total cfDNA was obtained using the NucleoSpin Plasma
`kit, for extracting fragments derived from tumor cells, the QIAamp
`DNA Micro Kit was more suitable. The results are summarized in
`Table 1. In another study, a QIAamp MinElute Virus Vacuum Kit
`
`(Qiagen) was used for cfDNA extraction. Its advantage was a great-
`er input volume of isolated plasma [29]. In some other studies, col-
`umn kits were used for
`isolation from blood [32]. Their
`disadvantage was a loss of small cfDNA fragments through mem-
`brane pores, leading to reduced detection sensitivity [33].
`
`Table 1
`Comparison results of isolation kits for cfDNA detection.
`
`DNA extraction protocol
`
`Total DNA concentration (ng/ll)
`
`Fraction of tumor cfDNA (% ratio of Alu 247/115)
`
`QIAamp DNA Micro Kit (Qiagen)
`NucleoSpin Plasma XS (Macherey–Nagel)
`Wizard (Promega)
`
`Source. Taken from Ref. [28].
`
`0.064
`0.086
`0.021
`
`50.9
`10.9
`59.4
`
`00003
`
`

`

`230
`
`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`Analysis of cfDNA based on tumor-specific mutations
`
`There are two basic approaches to cfDNA analysis: quantitative
`analysis and analysis based on DNA-specific mutations. The first
`approach is based solely on quantification of cfDNA, including both
`tumor and nontumorous cfDNA [34]. Increased DNA levels in plas-
`ma of cancer patients compared with healthy controls indicate the
`presence of tumor cfDNA. The actual cfDNA amount is typically
`determined by amplification of Alu sequences or other specific
`markers (e.g., b-globin, b-actin) [35]. This method yields high sen-
`sitivity but has rather low specificity because both tumor and non-
`tumorous DNA is amplified. The specificity can, alternatively, be
`enhanced by relating amplification of two unequally sized frag-
`ments, one of which is expressed in both tumor and nontumorous
`cfDNA and the other of which is only reflecting tumor cfDNA [31].
`An alternative approach is based on locating a tumor-specific
`mutation in the primary tumor, followed by detection of the same
`marker in isolated cfDNA. The most commonly used tumor-specific
`mutations include single-point substitutions or short deletions of
`proto-oncogenes and tumor suppressor genes as well as extensive
`deletions and DNA hypermethylation of tumor suppressor gene
`promoters. Detection of extensive deletions is performed by anal-
`ysis of microsatellite markers using the LOH (loss of heterozygos-
`ity) method. LOH has considerable limitations, primarily due to
`the need for simultaneous amplification of multiple microsatellite
`markers from a limited material and a relatively complicated re-
`sults interpretation [36]. Hypermethylation can be analyzed in sev-
`eral ways, including methylation-specific PCR [37], quantitative or
`fluorescent methylation-specific PCR [38,39], and methylation-
`specific restriction analysis [40]. Although the sensitivity of cfDNA
`detection based on methylation is relatively high, its specificity is
`limited. One of the reasons is the dynamic process of DNA methyl-
`ation, which may lead to a variation in methylation status among
`DNA molecules derived from the same tumor or, indeed, a primary
`tumor versus metastasis [41] or a primary tumor versus circulating
`cfDNA [42]. Another cause or a lower specificity in this case is a
`possible coincidence of a given methylation in healthy cells or in
`cells stemming from other defects [43].
`Examination of DNA variations of single-point mutations or
`short indels is challenging due to the presence of high levels of nor-
`mal (wild-type) DNA originating either in apoptotic or necrotic tis-
`sues lacking the detected mutation or in an inherent background
`presence of wild-type DNA from leukocytes. The situation with a
`mutated DNA copy screened by an overwhelming background of
`wild-type DNA is often referred to as a needle in a haystack
`[44,45]. The protocol usually consists of initial PCR followed by
`detection of amplified products. This can be performed using sev-
`eral methods, beginning with sequencing with limited sensitivity,
`through methods based on conformational changes and electro-
`phoresis (e.g., denaturing capillary electrophoresis, DCE), alterna-
`tively increasing the mutant fraction by mutant-enriched PCR
`(ME–PCR), and opting for a dedicated approach such as ‘‘BEAMing’’
`(beads, emulsion, amplification, and magnetics), digital PCR, or sin-
`gle-molecule sequencing.
`
`DCE
`
`during a temperature gradient [48]. Following a variety of abbrevi-
`ations from the original CDCE (constant denaturant capillary elec-
`trophoresis)
`[47] and TGCE (temperature gradient capillary
`electrophoresis) [49] through further improvements by cycling
`the temperature gradient in CGCE (cycling gradient capillary elec-
`trophoresis) [50] or CTCE (cycling temperature capillary electro-
`phoresis)
`[51],
`the terminology recently settled on DCE
`(denaturing capillary electrophoresis) [52] in an apparent analogy
`to dHPLC (denaturing high-performance liquid chromatography)
`widely adapted in routine DNA testing laboratories [53]. Similar
`to dHPLC, the basic principle of DCE lies in separation of partially
`denatured double-stranded PCR fragments. Homoduplexes are
`fragments with perfect sequence complementarity, whereas het-
`eroduplexes have sequence complementarity except at the muta-
`tion position containing a mismatch due to a presence of
`mutation. Both homo- and heteroduplexes are formed by a random
`combination of individual single strands during a process of full
`denaturation (melting) and a slow reannealing performed at the
`very end of PCR amplification. Each homo- and heteroduplex frag-
`ment has an individual melting temperature reflecting its actual
`composition of bases at the mutation site. When the fragment mix-
`ture is electrophoresed at an optimal temperature in a gel matrix,
`fragments with higher melting temperatures (homoduplexes) will
`remain in nondenatured (double-stranded) conformation, and thus
`migrate faster, compared with fragments with lower melting tem-
`peratures (heteroduplexes), which will adopt a denatured struc-
`ture [54]. A subtle change in melting temperature of ±0.1 °C is
`translated into a significant difference in electrophoretic migration,
`giving DCE great separation power for resolving all mutations
`within a given target sequence [47]. Unlike in dHPLC, where indi-
`vidual mutations are recognized based on different shapes of peaks
`in chromatograms, each fragment in DCE is usually observed as an
`individual peak. Naturally, DCE was applied to detect somatic
`mutations in cancerous tissues [54–56]. The potential of high
`mutation sensitivity of the approach, usable in detecting mutated
`DNA fragments circulating in patients’ plasma, has long been rec-
`ognized [57,58].
`cfDNA detection in plasma of colorectal cancer patients using
`DCE has been presented over the past 3 years [59,60]. It was clearly
`demonstrated that for the advanced stages of the disease, a muta-
`tion found in primary tumor tissue can be readily confirmed in lo-
`cal lymph nodes and distant metastases as well as detected in
`cfDNA. Fig. 2 demonstrates such a result. The DCE electrophero-
`gram of a PIK3CA amplification control from healthy tissue con-
`tains a single peak from the nonmutated DNA, as seen in line A
`of Fig. 2. A PIK3CA mutation is detected in the presence of addi-
`tional fragment peaks in the mutation-specific region. In the cur-
`rent case, the mutation was detected in rectal adenocarcinoma
`from a primary tumor biopsy collected during an endoscopic poly-
`pectomy (Fig. 2, line B). A subsequent CT scan uncovered progres-
`sion, with the disease spreading into lymph nodes and liver
`metastases, both also containing the PIK3CA mutation (Fig. 2, lines
`C and D, respectively). The presence of high levels of cfDNA was
`subsequently confirmed in peripheral blood of the patient (Fig. 2,
`line E). A relative amount of tumor DNA can be estimated from
`the ratio of the normal PCR fragment (Fig. 2, N) and the muta-
`tion-specific fragments (Fig. 2, M).
`
`The separation power of capillary electrophoresis (CE) has long
`been extensively used in DNA analyses. During the 1990s, a family
`of mutation detection techniques was introduced combining a
`classic principle of differential melting known from classic DGGE
`(denaturing gradient gel electrophoresis) [46] with the separation
`power of CE [47]. The separation was performed either at the pre-
`cise temperature optimum, where mutated and nonmutated PCR
`fragments adopted different electrophoretic properties [47], or
`
`ME–PCR and COLD–PCR approaches
`
`ME–PCR methods are also useful in mutation-based detection of
`cfDNA. They are based on suppression of nonmutated (wild-type)
`DNA or, alternatively, a preferential amplification of mutated
`DNA during PCR. An important distinction from a number of al-
`lele-specific techniques is that the ME–PCR is performed in a con-
`ventional way (i.e., using a standard pair of primers enclosing the
`
`00004
`
`

`

`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`231
`
`Fig.2. Use of denaturing capillary electrophoresis (DCE) technique to detect somatic mutations (KRAS and TP53) in multiple sample types on a patient with advanced colon
`adenocarcinoma. A normal tissue with only the wild-type peak and no mutant peaks (A), tissue from tumor biopsy (B), lymph nodes (C), liver metastases and plasma
`containing cell-free DNA (E) all showing wild-type peak followed by mutant homoduplex peak and two mutant heteroduplex peaks.
`
`desired target sequence). A traditional mutant enrichment is
`achieved by restriction of the wild-type sequence prior to or during
`the amplification. In a typical experiment, after initial PCR cycles, a
`specific restriction enzyme is applied to cut the wild-type frag-
`ments, resulting in enrichment of the mutated amplicons in the
`reaction mixture. The PCR then proceeds, alternatively, with addi-
`tional subsequent restriction steps. From 2004 through 2008, such
`an approach was shown to enable detection of KRAS mutated
`cfDNA in plasma, serum, and urine of colorectal cancer patients
`[27,39,61–65]. A BstNI restriction enzyme targeting wild-type
`KRAS sequence was used in the process. The aim of the study
`was to compare the ideal amount of input volume of plasma, ser-
`um, or urine for cfDNA isolation. Better results were evaluated
`from a greater volume of isolated material. However, positive find-
`ings in plasma in patients did not correspond in any way with the
`patients’ stage of the disease, and the conclusion focused only on a
`comparison of the DNA yields.
`Recently an alternative technique was introduced in which the
`mutant fragments are preferably amplified by lowering the tem-
`perature of the PCR annealing step [66]. At a lower annealing tem-
`perature, nonmutated fragments remain as DNA double strands
`inaccessible for annealing of primers. At the same time, mutant
`fragments, randomly forming imperfect mismatch duplexes, will
`be partly melted at the site of the mutated nucleotide, hence
`exposed to primer annealing. A result of this ingenious approach,
`
`referred to as COLD–PCR (coamplification at lower denaturation
`temperature PCR), is enrichment of the mutated fragments over
`nonmutated wild-type fragments. The technique has been shown
`to detect low levels of KRAS and BRAF mutations in colorectal
`cancer, suggesting it as a suitable tool for cfDNA [67]. From a prac-
`tical point of application in routine diagnostics and cfDNA moni-
`toring, the above mutant enrichment methods are relatively
`simple and attainable to a standard molecular diagnostic labora-
`tory [68,67,69].
`
`BEAMing
`
`In 2005, Diehl and coworkers introduced a dedicated approach
`for detection of mutations in plasma of colorectal cancer patients
`[21]. PCR products formed by amplification of target DNA sequence
`containing a specified mutation were mixed with magnetic beads,
`and the mixture was dispersed into the trillions of microparticles
`in water/oil emulsion. Then, a second PCR was performed using
`the primers bound to magnetic beads, followed by hybridization
`of resulting PCR products with two types of specific fluorescently
`labeled probes: mutated and nonmutated sequences, each labeled
`with a different fluorescent dye. Finally, the beads were analyzed
`using flow cytometry. The technique, referred to as BEAMing,
`was applied for monitoring of cfDNA in colorectal cancer patients
`before, during, and after surgery, and the results were found to
`
`00005
`
`

`

`232
`
`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`Table 2
`Overview of techniques used for detection of cfDNA in plasma of cancer patients.
`
`Method
`
`Sample
`costa
`
`Instrumentation
`
`Percentage of CRC tumors
`with KRAS mutation
`
`Estimated detectable fraction
`of mutated alleles (%)
`
`References
`
`DCE (denaturing capillary
`electrophoresis)
`ME-PCR (mutant-enriched PCR)
`
`Cold-PCR (co-amplification at lower
`denaturation temperature-PCR)
`BEAMing (beads emulsion amplification
`and magnetics)
`Digital PCR
`Single-molecule DNA sequencing
`Targeted (tagged-amplicon) deep
`sequencing
`
`low
`
`low
`
`low
`
`high
`
`PCR cycler and capillary sequencer
`
`(real-time) PCR cycler + arbitrary
`method to detect amplicons
`(real-time) PCR cycler + arbitrary
`method to detect amplicons
`PCR and flow cytometry systems
`
`Dedicated digital PCR system
`high
`Single-molecule sequencer
`high
`medium Next-generation sequencer
`
`30–35% (tissue) 28%
`(plasma)
`35% (tissue) 39% (plasma)
`62% (stool)
`37.5% (tissue)
`
`32% (tissue) 52% (plasma)
`63% (stool)
`38% (tissue) 47% (stool)
`18–30% (tissue)
`59% (tissue)
`
`0.30%
`
`0.10%
`
`0.8–2.5%
`
`0001%–1.7%
`
`0005%–0.1%
`0015%–3%
`2%
`
`a Low (<$10/sample); medium (<$100/sample); high (>$100/sample) (in U.S. dollars).
`
`[1–3]
`
`[38,61–
`64]
`[67,68]
`
`[20,28,69]
`
`[78,80,81]
`[90–92]
`[94,95]
`
`be highly correlated with the success of surgical treatment or dis-
`ease progression [29]. BEAMing clearly represents the highest sen-
`sitivity of all published methods dealing with tumor cfDNA
`detection so far [21,29,70]. A demonstrable fraction of mutated
`cfDNA was detected in all patients with advanced colorectal can-
`cer, in a majority of patients with nonmetastatic colorectal cancer,
`and even in one patient with benign disease (adenoma). Unfortu-
`nately, due to dedicated instrumentation as well as a rather com-
`plicated workflow, practical implementation of the technique in
`routine practice is currently limited.
`
`Digital PCR
`
`The principle of digital PCR is that a single sample is split into
`many aliquots by the simple process of dilution so that each frac-
`tion contains approximately one copy of DNA template or less per
`picoliter droplet. Then, the PCR is performed under optimal condi-
`tions designed to amplify a single copy of PCR template. Individual
`aliquots can only be positive or negative for the target sequence.
`Thus, after PCR amplification, nucleic acids may be quantified by
`counting the regions that contain PCR product (positive reactions).
`This method differs from standard PCR, in which the signal is gen-
`erated and measured from the amplification of multiple copies of
`the same locus template, and allows the detection of DNA muta-
`tions that are presented at very low levels compared with high
`background of normal DNA. Digital PCR was conceived by Sykes
`and coworkers in 1992 [71] and has quickly found its main appli-
`cation in noninvasive prenatal diagnosis of fetal chromosomal
`aneuploidies and monogenic diseases [72–75] as well as detection
`of low-level pathogens [76]. It has also been used in a number of
`clinical cancer research applications, including quantification of
`mutant alleles and detection of allelic imbalance in many types
`of cancer [77–79], with the main emphasis on colorectal cancer
`[80]. This approach can provide reliable information not only from
`tissue but also from body fluid samples [72,75,79,81,82]. Moreover,
`the multiwell plate method and chip enabling wide analysis are
`now emerging [83].
`
`Single-molecule sequencing
`
`In classical sequencing, there must be DNA amplified many
`times and then the copies are read. Single-molecule sequencing
`may skip the copying step; thus, it is an attractive approach due
`to its simplicity. The method was proposed by Keller and cowork-
`ers in 1989 [84] and was demonstrated for the first time by Bra-
`slavsky and coworkers in 2003 [85]. Since then, great progress
`has been made in this field. Single-molecule sequencing has been
`realized in many laboratories through several approaches, includ-
`
`ing exonuclease sequencing, scanning probe microscopy, and
`sequencing by synthesis [86–88]. Two novel technologies, trans-
`mission electron microscopy and nanopore sequencing within
`zeptoliter volumes, seem to be promising as well. Single-cell
`sequencing provides an unequaled view of genomic diversity with-
`in tumors and allows detection and analysis of the genomes of rare
`cancer cells [89–93].
`
`Targeted deep sequencing
`
`Indeed, with most of the above-mentioned mutation-based ap-
`proaches, there is a requirement of capturing a mutation in the pri-
`mary tumor tissue. The basic portfolio consisting of the most
`commonly observed somatic mutations in major oncogenes and
`tumor suppressors allows for tracking of most, but not all, tumors
`[94]. In an alternative approach, hundreds of preselected ampli-
`cons and entire genes from cfDNA are sequenced without any prior
`knowledge about the presence of any mutations in the primary tu-
`mor [95]. So-called targeted deep sequencing of preamplified re-
`gions is then performed on a next-generation sequencer [96],
`resulting in specificity of more than 97%. With the constantly
`declining cost of new-generation sequencing, such an approach
`has significant potential.
`An overview of all techniques discussed above is found in Ta-
`ble 2, which lists approximate levels of cost per sample, required
`instrumentation, clinical sensitivity (expressed as the percentage
`of KRAS mutants detected in a group of colorectal cancer patients),
`and relative mutant detection sensitivity (expressed at the per-
`centage of mutant alleles detectable within an excess of wild-type
`alleles).
`
`Conclusions
`
`The need for efficient tools to detect early stages of cancer pro-
`gression prompts further development of molecular tests directed
`at cfDNA. The main factor is minimal invasivity, which is often key
`in long-term follow-up of patients undergoing challenging antican-
`cer therapy. The routine adaptation of these tests, now frequently
`referred to as ‘‘liquid biopsy’’, is expected in the near future. The
`pace of such adaptation will depend on the availability of easily
`adoptable and low-cost methodologies.
`
`Acknowledgments
`
`This work was supported by the Czech Ministry of Health Inter-
`nal Grant Agency (Project NS 9809). This is contribution number 7
`from the Center for Applied Genomics of Solid Tumors (CEGES).
`
`00006
`
`

`

`cfDNA detection in cancer / L. Benesova et al. / Anal. Biochem. 433 (2013) 227–234
`
`233
`
`References
`
`[1] J.J. Ott, A. Ullrich, M. Mascarenhas, G.A. Stevens, Global cancer incidence and
`mortality caused by behavior and infection, J. Public Health (Oxf.) 33 (2011)
`223–233.
`[2] R.E. Schoen, J.L. Weissfeld, J.M. Trauth, B.S. Ling, M. Hayran, A population-based
`community estimate of total colon examination: the impact on compliance
`with screening for colorectal ca