Review
`
`For reprint orders, please contact reprints@future-drugs.com
`
`Diagnostic potential of circulating
`nucleic acids for oncology
`
`Carsten Goessl
`
`Approximately a decade ago, the PCR-based detection of extracellular, tumor-derived
`circulating nucleic acids in the plasma and serum of cancer patients was introduced as a
`noninvasive tool for cancer detection. Although the test criteria, sensitivity and specificity,
`compare favorably with conventional diagnostic measures, until now the methodical
`ponderousness of circulating nucleic acids in plasma and serum analysis prevented it from
`becoming a clinical routine application. However, with rapid technical improvement
`towards automated high-throughput platforms, it is expected that the next 5 years will see
`circulating nucleic acids in plasma and serum analysis integrated into the initial diagnosis
`and follow-up monitoring of cancer patients. The hope is that the use of circulating nucleic
`acids in plasma and serum as a molecular tumor marker and potential profiling tool will
`finally translate into a longer survival and better quality of life for cancer patients.
`
`Expert Rev. Mol. Diagn. 3(4), 431–442 (2003)
`
`Noninvasive nucleic acid-based molecular
`diagnosis of cancer using body fluids has
`emerged rapidly over the last 12 years. The
`first of these reports dealt with the detection of
`urothelial cancer by DNA-based analysis of
`exfoliated bladder tumor cells in voided urine
`[1]. In selected cases it has been repeatedly
`demonstrated that molecular detection of
`tumor cells in body secretions can precede
`conventional diagnosis of primary and
`[2,3]. Consequently,
`recurrent malignancy
`molecular tumor detection methods have
`been ascribed a great potential to impact on
`future therapeutic management of cancer
`patients [4].
`Beyond excreted body fluids such as urine
`and sputum, blood constitutes an attractive
`reservoir for nucleic acid-based detection of
`cancer [4]. Blood can be drawn in a minimally
`invasive manner (classified thereafter as non-
`invasive) and this enables repeated testing, a
`feature especially useful for longitudinal mon-
`itoring of responses to therapy and during fol-
`low-up of oncologic patients. Blood consists
`of two main fractions: a cellular and a
`plasma/serum fraction. Nucleic acid-based
`cancer detection analyzing the cellular blood
`fraction for circulating tumor cells has been
`
`extensively described elsewhere [5,6]. The focus
`of this review is the detection of cancer
`patients by analyzing their plasma/serum frac-
`tion for the existence of circulating extracellu-
`lar nucleic acids (both DNA and RNA)
`derived from tumoral sources. Cell-free circu-
`lating nucleic acids in the plasma and serum
`(CNAPS) of cancer patients have been investi-
`gated extensively as evidenced by more than
`200 articles on the subject. A recent review
`has summarized detailed results on colorectal,
`pancreatic, breast, lung and prostate cancer,
`and has concluded that detection and quanti-
`fication of viral DNA in virus-associated
`malignancies such as nasopharyngeal carci-
`noma (associated with Epstein–Barr virus
`[EBV]) might be closest to broad clinical
`application [7–8].
`This review is structured into four sections:
`first, the medical context for the detection of
`tumor-derived CNAPS;
`second, a brief
`description of the methodology used and the
`obstacles observed; third, a selection of the
`results obtained with different detection
`methods in various cancer entities; and finally,
`a 5-year outlook into the future, taking into
`account competing methods for noninvasive
`cancer detection in blood samples.
`
`CONTENTS
`CNAPS: from plant
`physiology to oncology
`Methods
`Overview of results
`Conclusion, expert opinion
`& five-year view
`Key issues
`References
`Affiliation
`
`Novartis Pharma AG,
`BU Oncology,
`WSJ 27.2.055,
`Lichtstr. 15,
`4002 Basel, Switzerland
`Tel.: +41 613 244 271
`Fax: +41 613 242 034
`carsten.goessl@pharma.novartis.com
`
`KEYWORDS:
`cancer detection, circulating
`nucleic acids in plasma and
`serum, tumor DNA, tumor RNA
`
`www.future-drugs.com
`
`© Future Drugs Ltd. All rights reserved. ISSN 1473-7159
`
`431
`
`00001
`
`EX1044
`
`

`

`Goessl
`
`CNAPS: from plant physiology to oncology
`The existence of circulating, noncellular nucleic acids (both
`RNA and DNA) was first described more than 50 years ago [9].
`These findings remained largely unnoticed by the oncological
`community until in the late 1970s, when elevated amounts of
`DNA were characterized in the serum of cancer patients [10].
`Successful therapy was associated with a decrease in serum DNA
`quantities, however, benign diseases, especially autoimmune dis-
`orders, had also been found to be associated with increased
`amounts of circulating DNA [11]. As a consequence, the malignant
`versus benign origin of the increased DNA concentrations found
`in the plasma/serum of cancer patients remained unknown.
`A group of Swiss plant physiologists associated with Philippe
`Anker and Maurice Stroun was the first to characterize malig-
`nant features of CNAPS in cancer patients [12]. This group had
`formerly investigated nucleic acid release from bacteria into
`plants [13], from frog auricles and from human lymphocytes [14].
`With another group [15], they further established the malignant
`nature of CNAPS in cancer patients by demonstrating gene
`mutations matching those of their primary tumors [16–17].
`Thus, as illustrated in FIGURE 1, translating the general biologi-
`cal phenomenon of DNA release from plant physiology to
`human oncology led to the establishment of a noninvasive
`molecular tumor detection method which has resulted in more
`than 200 articles on the subject to date [7].
`The initial targeting of genomic tumor DNA in CNAPS anal-
`ysis has now been expanded to circulating mitochondrial DNA,
`RNA and tumor-associated viral nucleic acids. The related use
`of CNAPS in nonmalignant indications [18], such as prenatal
`diagnosis, transplant rejection and infectious diseases, will not
`be covered by this review. CNAPS have been found to be parti-
`cle-associated (see Methods). The use of quantitative detection
`methods for CNAPS has emerged rapidly since 1999 and has
`further increased the diagnostic utility of this method [19–23].
`However, despite favorable test characteristics compared with
`conventional serum tumor markers and other molecular detection
`methods, so far CNAPS analysis has not left the research stage.
`Handling of the very small quantities of nucleic acids isolated
`from plasma or serum and dependence on comparably laborious
`PCR amplification processes are regarded as the main obstacles
`precluding this method from becoming clinical routine.
`
`Genometastasis hypothesis
`A horizontal transfer of circulating tumor DNA into tissues has
`been demonstrated in animal experiments and raises questions
`on the possible functional relevance of this so-called genome-
`tastasis, as opposed to conventional cellular metastatic spread
`[24]. Hypothetically, under the assumption that horizontal
`transfer of tumor DNA sequences has a transforming potential
`[25,26], stem cells in distant organ tissues would constitute possible
`targets for uptake.
`
`The other plasma: cell-free tumor DNA in bone
`marrow supernatants
`According to a recent report by Taback and coworkers, tumor
`DNA displaying tumor-specific genetic alterations is present in
`the plasma (supernatant) of bone marrow aspirates from
`patients with early breast cancer [27]. The sensitivity associated
`with this approach was determined to be superior to analysis of
`circulating tumor DNA in the blood serum of the same
`patients. Due to its invasive nature, bone marrow analysis is not
`the focus of this review, although the work by Taback clearly
`demonstrates that analysis of cell-free tumor DNA begins to
`enter fields considered to be the domain of RNA-based detection
`of micrometastatic tumor cells [6].
`
`Methods
`Isolation of nucleic acids
`The first step in CNAPS analysis consists of DNA isolation
`from plasma or serum. Careful two-step centrifugation has
`been recommended to ensure complete cell removal [28].
`Whereas higher absolute DNA concentrations have been
`observed in serum [29], the relative proportion of tumor DNA
`tends to be higher in plasma [30]. In vitro release of benign
`DNA from leukocytes during the coagulation process has been
`proffered as an explanation for this observation [30,31]. There is
`no agreement on the most efficient method of DNA isolation
`from plasma/serum. The commercially available DNA isolation
`kit from Qiagen (Venlo, The Netherlands) is widely used [31]
`and has compared favorably with manual isolation methods
`[32], although the superiority of that kit has been challenged by
`other authors [33]. Quantities of CNAPS have been determined
`using different methods yielding DNA concentrations of
`
`Bacteria [13]
`
`Lymphocytes [14]
`
`Tumor cells [12]
`
`Figure 1. Cellular DNA release: from bacteria to use as a molecular tumor marker in humans.
`
`432
`
`Expert Rev. Mol. Diagn. 3(4), (2003)
`
`00002
`
`

`

`2–30 ng/ml in healthy controls [11,16,34–36] and 20–200 ng/ml
`in cancer patients [16,34–36]. However, absolute DNA amounts
`in plasma/serum do not reliably differentiate between cancer
`patients and controls, especially not in those with autoimmune
`diseases [11]. Fournie and coworkers found a correlation
`between advanced tumor stage and absolute DNA concentra-
`tions in lung cancer patients [35], however, no such associations
`were found in various other tumor entities including lung can-
`cer [31,34]. The relative proportion of tumor DNA within whole
`plasma/serum DNA is quite variable (3–93%) [36]. Colorectal
`cancer appears to be associated with comparatively low
`amounts of circulating tumor DNA as evidenced by the findings
`of Hibi and coworkers [37]. Conversely, in many other tumor
`types, circulating tumor DNA appears to be greatly enriched in
`the plasma/serum fraction [16,38]. The reasons for this phenome-
`non have been attributed to either increased release from malig-
`nant sources and/or reduced clearance from the circulation, the
`latter most likely attributable to its nucleoprotein structure
`[31,36,39]. The main bulk of DNA fragments in plasma/serum of
`cancer patients was found to be shorter than 200 bp [36]. Bote-
`zatu and coworkers reported on short mutated k-ras sequences
`secreted into the urine of cancer patients, indicating that at
`least short CNAPS might undergo renal filtration [40]. If con-
`firmed in a larger series of different tumor entities, urine may
`emerge as a second source for universal molecular diagnosis of
`malignancy based upon cell-free methodology [18,41].
`Recently, cell-free circulating tumor RNA also has been suc-
`cessfully isolated from plasma/serum of cancer patients [42–48].
`Given the significant RNAse activity of whole blood, this find-
`ing was unexpected. However, as demonstrated by Hasselmann
`and coworkers, circulating tumor RNA is embedded in apop-
`totic bodies and thus, as with circulating DNA, is protected
`against degradation. Although the exact mechanism of tumor
`DNA and RNA release into the circulation remains unknown,
`there is accumulating evidence that most tumor-derived
`CNAPS are from cancer cells undergoing apoptosis [31,36,39].
`
`Circulating tumor DNA: target selection for molecular diagnosis
`DNA mutations, microsatellite alterations and epigenetic DNA
`alterations manifested by gene promoter hypermethylation
`constitute the three main targets for analyzing circulating
`tumor DNA in cancer patients. The rationale behind this
`approach is the expected occurrence of the same alterations in
`primary tumor tissue. However, most likely due to the existence
`of heterogeneous tumor clones (and/or undetected micrometa-
`static spread), alterations found in a small tumor sample do not
`always match those found in plasma/serum DNA [50–51]. Only
`the main features of the three most commonly used detection
`methods for circulating tumor DNA in plasma/serum of cancer
`patients will be presented.
`
`DNA mutation analysis
`Most investigators choose ras mutations when analyzing circu-
`lating tumor DNA. The existence of ras mutations in solid
`malignancies is especially common in cancers of the pancreas
`
`Diagnostic potential of circulating nucleic acids
`
`and gastrointestinal tract, where they occur in 90 and 50% of
`the primary tumors, respectively [7,52]. Since these mutations
`are clustered in codon 12 of the k-ras gene, PCR-based meth-
`ods have been developed which enrich for common mutation
`variants in plasma/serum DNA [36,52]. However, the finding
`that k-ras mutations can occur in benign conditions and thus
`appear to be not entirely tumor-specific, confers some concern
`about their usefulness as molecular tumor markers. In addition,
`Ramirez and coworkers found that k-ras mutations in primary
`non-small cell lung carcinomas only rarely matched the consist-
`ent mutation pattern (i.e., TGT) in corresponding serum [53].
`Prospective, longitudinal studies are needed to eliminate specif-
`icity concerns associated with this molecular target [54]. Muta-
`tions of the tumor suppressor gene p53 have been investigated
`in plasma/serum of patients suffering from hepatocellular,
`breast and colorectal carcinoma [31], although the methodical
`workload associated with this approach does not favor routine
`clinical application.
`Recently, mitochondrial DNA mutations have been
`exploited as a target for analyzing plasma/serum DNA from
`cancer patients. Mitochondrial DNA is present in approxi-
`mately 200–10,000 copies per cell and should therefore provide
`a better sensitivity for detecting critically low quantities of tar-
`get DNA [31,55]. However, although three out of three prostate
`cancer patients with mitochondrial DNA mutations in their
`primary tumor exhibited the same mutations in serum [56], the
`detection rate in serum from ovarian cancer patients was 0%,
`perhaps due to the fact that no mutation enrichment method
`was applied [57]. However, even when using mutation enrich-
`ment PCR, the detection rate of tumor-derived mitochondrial
`DNA in colorectal cancer patients was disappointingly low
`(14% in primary tumors harboring target mitochondrial muta-
`tions) [58]. As mutations of genomic DNA (k-ras) can be
`detected in plasma/serum of more than 80% of colorectal cancer
`patients harboring the same mutations in the primary tumor
`[59], the high copy number of mitochondrial DNA does not
`compensate for decreased release efficacy and/or increased
`breakdown in plasma/serum of cancer patients.
`Results from different tumor entities with various disease
`stages are deemed necessary to better define the overall diagnostic
`sensitivity of this approach, irrespective of the problems asso-
`ciated with sequencing the whole or parts of the 16.5 kb
`mitochondrial genome.
`
`Microsatellite analysis
`Generally, microsatellite analysis is easy to perform, however, it
`requires tumor DNA within DNA originating from benign
`sources at a ratio of 0.5–5% when targeting microsatellite insta-
`bility (MIN) and more than 10–20% when targeting loss of
`heterozygosity (LOH) [16]. Since some tumor entities, such as
`bladder cancer, are not associated with consistent alterations at
`characteristic chromosomal locations, a panel of 17 markers
`was used by von Knobloch and coworkers for detection of
`tumor DNA in serum [60]. Despite the possibility of using
`multiplex PCR to reduce the workload, this level of complexity
`
`www.future-drugs.com
`
`433
`
`00003
`
`

`

`Goessl
`
`Table 1. DNA alterations in plasma/serum of cancer patients (at diagnosis, all stages if not otherwise indicated).
`Alteration
`Neoplasm
`Patients (n) Markers (n)
`Alterations in
`Specificity
`serum/plasma (%)
`(%)
`
`Ref.
`
`Microsatellite alterations
`LOH, MIN
`Non-small cell bronchial
`cancer
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH, MIN
`
`LOH
`(allelic imbalance)
`
`Gene mutations
`N-ras
`
`Lung cancer
`
`Neck/head tumors
`
`Colorectal cancer
`
`Colorectal cancer
`
`Clear-cell renal cancer
`
`Renal cell cancer
`
`Melanoma
`
`Breast cancer
`
`Breast cancer
`
`Bladder cancer
`
`Ovarian cancer
`
`Hematologic neoplasms
`
`21
`
`34
`
`21
`
`44
`
`27
`
`40
`
`40
`
`40
`
`21
`
`57
`
`39
`
`54
`
`10
`
`3
`
`6
`
`12
`
`8
`
`9
`
`4
`
`20
`
`10
`
`7
`
`2
`
`17
`
`8
`
`1
`
`71
`
`85
`
`29
`
`0 (!)
`
`59
`
`65
`
`87
`
`58
`
`48
`
`30
`
`85
`
`93
`
`50
`
`86
`
`100
`
`[16]
`
`100
`
`100
`
`nd
`
`nd
`
`100
`
`85
`(10 markers)
`
`100
`
`100
`
`100
`
`100
`
`100
`
`Nd
`
`100
`
`[51]
`
`[17]
`
`[37]
`
`[78]
`
`[38]
`
`[79]
`
`[80]
`
`[61]
`
`[81]
`
`[60]
`
`[64]
`
`[82]
`
`[83]
`
`Ig chain DNA
`
`B-cell leukemia
`
`k-ras
`
`k-ras
`
`k-ras
`
`k-ras
`
`p53
`
`Colorectal cancer§
`
`Colorectal cancer
`
`Colorectal cancer
`
`Pancreatic carcinoma
`
`Hepatocellular cancer
`
`Gene promoter hypermethylation§§§
`p16, MGMT, GSTP1,
`Non-small cell bronchial
`DAP kinase
`carcinoma
`
`p16 (Quant)
`
`Hepatocellular carcinoma
`
`p16
`
`p16, MGMT, GSTP1,
`DAP kinase
`
`APC (Quant)
`
`APC (Quant)
`
`GSTP1
`
`GSTP1 (Quant)
`
`Breast cancer
`
`Neck/head tumors
`
`Esophageal cancer
`
`Lung cancer
`
`Prostate cancer
`
`Prostate cancer
`(early stage)
`
`110
`
`8
`(adenoma: 62)
`
`14
`
`69
`(adenoma: 9)
`
`21
`
`20
`
`22
`
`22
`
`43
`
`50
`
`84
`
`89
`
`32
`
`69
`
`1
`
`1
`
`1
`
`1
`
`1
`
`1
`
`4
`
`1
`
`1
`
`4
`
`1
`
`1
`
`1
`
`1
`
`63
`(adenoma: 35)
`
`50
`
`41
`(adenoma: 44)
`
`81
`
`30
`
`52
`
`59
`
`14
`
`42
`
`18
`
`47
`
`72
`
`32
`
`86§§
`
`[54]
`
`100
`
`100
`
`100
`
`100
`
`[59]
`
`[52]
`
`[76]
`
`[84]
`
`Nd
`
`[85]
`
`100
`
`100
`
`100
`
`100
`
`100
`
`100
`
`100
`
`[23]
`
`[86]
`
`[87]
`
`[20]
`
`[21]
`
`[88]
`
`[22]
`
`434
`
`Expert Rev. Mol. Diagn. 3(4), (2003)
`
`00004
`
`

`

`Diagnostic potential of circulating nucleic acids
`
`Table 1. DNA alterations in plasma/serum of cancer patients (at diagnosis, all stages if not otherwise indicated) (cont.).
`Alteration
`Neoplasm
`Patients (n)
`Markers (n)
`Alterations in
`Specificity
`Ref.
`serum/plasma (%)
`(%)
`
`Mitochondrial DNA mutations
`Prostate cancer (early stage)
`
`Ovarian cancer
`
`Colorectal cancer
`
`Hepatocellular carcinoma
`
`3
`
`14
`
`77 (7§§§§)
`
`10§§§§
`
`OML
`
`Seq
`
`OML§§§§
`
`OML
`
`100
`
`0
`
`1 (14§§§§)
`
`80§§§§
`
`Nd
`
`Nd
`
`Nd
`
`Nd
`
`[56]
`
`[57]
`
`[58]
`
`[89]
`
`§Prospective analysis: plasma DNA samples were taken before colonoscopy and pathohistological confirmation (screening situation).
`§§Control group consisted mainly of symptomatic or high-risk patients. §§§Review: [66]. §§§§Only patients with confirmed mitochondrial mutations in their primary tumor.
`APC: Adenomatous polyposis coli gene; DAP: Death-associated protein; GSTP1: Glutathione-S-transferase P1 gene; Ig: Immunoglobulin; LOH: Loss of heterozygosity;
`MGMT: O6-methylguanine-DNA methyltransferase gene; MIN: Microsatellite instability; Nd: Not determined; OML: Oligonucleotide mismatch ligation assay
`(mutation enriching PCR technique); Quant: Quantitative determination using real-time methylation-specific PCR; Seq: Determination by sequencing serum DNA.
`
`raises concerns regarding the practicality of broad clinical appli-
`cation in CNAPS analysis. In addition, microsatellite analysis
`using small DNA concentrations, such as those isolated from
`plasma/serum, is prone to technical artifacts [61–62]. Longer alle-
`les (>200 bp) are more likely to display false-positive LOH than
`their corresponding shorter counterparts [63]. In CNAPS analy-
`sis, the underlying difficulty to amplify longer microsatellite
`sequences can be ascribed to the highly fragmented nature of
`plasma/serum DNA [36]. Although true-positive LOH findings
`in plasma/serum DNA can usually be confirmed by compari-
`son with findings in primary tumor tissue and by use of addi-
`tional neighboring markers [38], specificity concerns, amongst
`others, are prohibiting broader routine clinical use of this
`method. According to Chang and coworkers, allelic imbalance
`in plasma/serum DNA can be analyzed using digital single-
`nucleotide polymorphism analysis without the potential risk of
`preferential amplification of shorter microsatellite alleles [64].
`This method is, however, associated with additional workload,
`leaving its suitability for routine CNAPS diagnosis doubtful.
`
`Gene promoter hypermethylation analysis
`Gene promoter hypermethylation has become an attractive target
`for molecular cancer detection as promoter hypermethylation
`often constitutes an early neoplastic change, occurring even in
`premalignant or morphologically yet benign lesions [65]. A
`comprehensive review on the use of gene promoter hypermeth-
`ylation analysis to detect exfoliated tumor cells in secreted body
`fluids as well as circulating tumor DNA in blood fractions from
`cancer patients was recently published by Laird [66]. Detection
`of tumor-associated gene promoter hypermethylation
`in
`plasma/serum DNA by methylation specific PCR (MSP) [67]
`significantly enhances the analytical sensitivity compared with
`microsatellite analysis. The methodical sensitivity of such assays
`equals approximately 1 x 104 [67] to 1 x 105
`[68]. Since three
`methylated sites are usually targeted by each of the corresponding
`two MSP-primers, false-positive results due to technical artifacts
`are not considered to be of concern. Age-related gene promoter
`hypermethylation (e.g., of APC and p16 in normal gastric
`mucosa) is a well-known event with possible implications for
`
`analysis of plasma/serum DNA [20]. Fortunately, because normal
`tissues affected by gene promoter hypermethylation apparently
`do not secrete DNA into the blood circulation, this problem
`appears to be theoretical in nature as p16 and APC gene pro-
`moter hypermethylation has not yet been observed in the
`serum/plasma of controls [20,66]. To further protect against
`false-positive scoring, suitable cut-off levels can be defined
`when using quantitative MSP for tissue and/or plasma/serum
`samples [20–23]. Quantification of promoter hypermethylation
`provides a tool to distinguish low-level hypermethylation due
`to age-related processes and/or biological noise from high-level
`hypermethylation associated with neoplastic growth.
`
`Circulating tumor RNA
`The same rationale applies for analysis of circulating tumor
`cells as for analysis of circulating tumor RNA [5,6]. It is assumed
`that transcripts of epithelial origin do not normally occur in the
`blood, thus their existence is thought to indicate malignancy
`which has gained access to the circulation. This concept has
`been challenged by the observation of illegitimate transcription;
`the secretion – albeit at low levels – of epithelial cell transcripts
`by any cell including normal blood cells and/or endothelial
`cells. Certain transcripts used as tumor targets are more likely
`to be produced by normal nonepithelial cells (e.g., albumin and
`CK-19) than others whose occurrence is known to be more
`closely tumor-associated [69].
`
`Circulating viral nucleic acids
`In addition to tumor nucleic acids from endogenous sources,
`analysis of tumor-associated viral DNA has emerged rapidly as
`a molecular tool for initial diagnosis and follow-up monitoring
`in defined populations of cancer patients. Chan and coworkers
`found that patients with EBV-associated nasopharyngeal carci-
`noma had persistently elevated EBV-DNA levels in their
`plasma/serum, whereas patients after recovery from benign
`mononucleosis (caused by EBV) did not [19,70]. Quantitative
`determination of EBV-DNA also yielded prognostic information
`and decreasing levels were strongly correlated with favorable
`response to therapy [70]. Determination of viral DNA levels is
`
`www.future-drugs.com
`
`435
`
`00005
`
`

`

`Goessl
`
`close to becoming routine application in these patients [7,8,70].
`The methodology may also have clinical application potential
`in other virus-associated malignancies, such as EBV-associated
`lymphoma and gastric cancer [71,72], as well as Human Papil-
`loma virus (HPV)-associated cervical cancer and head and neck
`cancer [73–75]. A limitation to quantitative determination of
`viral DNA as part of CNAPS analysis lies in the fact that strong
`association with viral infection is limited to few cancer entities
`and to certain geographic areas (mostly Asia–Pacific region).
`
`Overview of results
`Circulating tumor DNA
`TABLE 1 provides data on detection rates of circulating tumor
`DNA in various cancer entities at initial diagnosis. This selection
`is incomplete for reasons of space. The detection rates are given
`for calculation of clinical sensitivity; prior finding of a target
`DNA alteration in the primary tumor was not considered a pre-
`requisite. A less conservative estimation, yielding numerically
`higher detection rates, is the calculation of methodical sensitivity
`which, however, includes only patients with confirmed target
`DNA alterations in their primary tumors [66].
`All the patients in TABLE 1 were diagnosed using clinically
`established tumor detection methods and conventional patho-
`histological tumor confirmation was the gold standard against
`which positive CNAPS findings were compared. As different
`tumor stages were included in most studies, the table data do
`not specifically address the question of how well CNAPS analysis
`performs as a detection method specifically for early-stage cancer.
`In some but not all tumor entities, the finding of circulating
`
`tumor DNA may indicate poor prognosis and/or advanced
`tumor stage [7,31]. However, the remarkable detection rate of
`80% k-ras mutations found by Kopreski and coworkers in
`plasma of patients with k-ras positive colorectal adenoma pro-
`vides proof-of-principle evidence that CNAPS analysis is able to
`detect early tumors, including preinvasive stages [54]. These find-
`ings have been recently confirmed by Ryan and coworkers who
`found 44% of patients with tubulovillous adenoma to be posi-
`tive for serum k-ras mutations (57% when k-ras positive adeno-
`mas were selected) [52]. In contrast to this, molecular detection
`of circulating tumor cells is per definition dependent on invasive
`growth. Thus, CNAPS analysis has the potential to detect pre-
`malignancy and malignancy earlier than RT-PCR-based meth-
`ods for detection of circulating tumor cells (FIGURE 2). In addi-
`tion, the early release and circulation of oncogenic nucleic acids
`from preinvasive malignancies could have a signal function for
`facilitating tumor growth both locally and at distant sites.
`The data in TABLE 1 demonstrate that CNAPS analysis of cir-
`culating tumor DNA is associated with a remarkable high spe-
`cificity, reaching 100% in most cases. This is in sharp contrast
`to all pressently used protein-based serum tumor markers.
`However, in most studies on CNAPS in oncology indications,
`the number of controls used was small and, in many cases, did
`not include inflammatory or other benign disease conditions.
`Data on using serial CNAPS analysis for follow-up monitoring
`are even more sparse. In pancreatic, colorectal and small-cell
`lung cancer, longitudinal analysis of DNA alterations in the
`plasma/serum of patients was found to be useful for monitoring
`during follow-up [52,76,77].
`
`Review
`reference
`
`CNAPS
`
`[4–6,18,31,69]
`
`CTC
`
`[4,7,8,69]
`
`Tumor
`
`Blood/lymph vessel
`
`Figure 2. Origin of circulating tumor DNA/RNA in the serum/plasma of cancer patients (CNAPS)
`as opposed to circulating tumor cells (CTC). Detection of malignant cell growth using CNAPS
`analysis is possible even at preinvasive stages [52,54]. At present, the role of CNAPS as a cellular signal
`and its possible role in facilitating tumor spread and metastasis remains unknown [24–26].
`
`Circulating tumor RNA
`TABLE 2 provides data on cell-free circulat-
`ing tumor RNA in cancer patients. In
`contrast to analysis of circulating tumor
`DNA, the targets are not alterations of
`nucleic acids but rather the mere detection
`of transcripts from epithelial origin is
`taken as a surrogate for malignancy in
`affected patients. In addition with telom-
`erase RNA, a tumor-specific RNA target
`has also become available for detection in
`plasma and serum of cancer patients
`[44,90]. Due to a certain rate of false-posi-
`tive findings in controls, specificity with
`some RNA markers (e.g., CK-19) appears
`to be considerably less than 100% and
`lower than DNA-based detection methods
`in plasma/serum. Extensive comparisons
`between DNA- and RNA-based CNAPS
`analysis using
`identical plasma/serum
`samples from cancer patients and controls
`have, however, not yet been reported [69].
`There is only one report on CNAPS anal-
`ysis comparing sensitivity of two markers
`for cell-free RNA versus two microsatellite
`
`436
`
`Expert Rev. Mol. Diagn. 3(4), (2003)
`
`00006
`
`

`

`Diagnostic potential of circulating nucleic acids
`
`Ref.
`
`[42]
`
`[43]
`
`[90]
`
`[44]
`
`[45]
`
`[45]
`
`[46]
`
`[47]
`
`[47]
`
`[48]
`
`Table 2. RNA of epithelial origin in plasma/serum of cancer patients (at diagnosis, all stages).
`Target RNA
`Neoplasm
`Patients (n)
`Sensitivity (%)
`Tyrosinase
`Malignant melanoma
`6
`67
`
`Specificity (%)
`100
`
`100
`
`100
`100§§
`
`100
`
`80
`
`88
`
`Nd
`
`96
`
`80
`
`0
`
`60
`
`100§
`89§
`
`25–28§§
`
`60
`
`49
`
`73
`
`32
`
`74
`
`Nd
`
`10
`
`99
`
`18
`
`45
`
`45
`
`45
`
`53
`
`53
`
`5
`
`Tyrosinase
`
`Telomerase
`
`Telomerase
`
`CK-19
`
`Mammaglobin
`
`CK-19 and mammaglobin
`
`CEA
`
`CK-19
`
`Malignant melanoma
`
`Colorectal cancer
`Lymphoma
`
`Breast cancer
`
`Breast cancer
`
`Breast cancer
`
`Breast cancer
`
`Colorectal cancer
`
`Colorectal cancer
`
`Lung cancer
`
`CK-19
`
`HER2/neu + hnRNP-B1
`
`Lung cancer
`§Sensitivity and specificity obtained with a cut-off level higher than maximum value obtained in ten healthy volunteers.
`§§Depending on telomerase subunit.
`CEA: Carcinoembryonic antigen; Nd: Not determined.
`
`5
`
`100
`
`100
`
`[48]
`
`markers for cell-free DNA [46]. The authors concluded that
`RNA-based analysis was superior to DNA-based analysis in
`plasma samples of breast cancer patients, however, specificity
`was not analyzed and, given the findings of Fleischhacker and
`coworkers (0% specificity for serum CK-19 RNA) [48], the over-
`all clinical superiority of analyzing cell-free RNA versus DNA
`in plasma/serum of cancer patients remains to be determined.
`Positive findings of circulating tumor mRNA were correlated
`with poor prognosis in breast cancer and advanced-stage in
`colorectal cancer [46,47]. Furthermore, simultaneous analysis of a
`few transcripts from serum is already possible, thereby allowing
`a noninvasive miniprofiling of malignancy [48]. In addition,
`
`Dasi and coworkers reported on the possibility of quantifying
`plasma levels of circulating tumor RNA in cancer patients [90].
`They found that all nine patients suffering from lymphoma and
`eight out of nine patients suffering from colorectal cancer had
`higher plasma human telomerase reverse transcriptase (hTERT)
`mRNA levels than the maximum value determined in ten
`healthy volunteers.
`
`Circulating viral DNA & RNA
`TABLE 3 summarizes data on the use of viral DNA (and RNA) as
`a nucleic acid-based tumor marker. Quantitative analysis has
`further improved the diagnostic specificity for initial diagnosis
`
`Ref.
`[91]
`
`[19]
`
`[92]
`
`[71]
`
`[72]
`
`[73]
`
`[74]
`
`[75]
`
`437
`
`Table 3. Circulating viral DNA (and RNA) in plasma/serum of cancer patients (at diagnosis, all stages).
`Target
`Neoplasm
`Patients (n)
`Sensitivity (%)
`Specificity (%)
`EBV-DNA
`Nasopharyngeal cancer
`42
`31
`100
`
`96
`
`89
`
`94
`
`93
`
`79
`
`100
`
`93–100
`
`77–96§§
`
`100
`
`98
`
`Nd
`
`12
`
`72
`
`6–9
`
`57
`
`26
`
`18
`
`19
`
`50
`
`175
`57
`
`70
`
`EBV-DNA§
`
`EBV-RNA
`
`EBV-DNA§
`
`EBV-DNA
`
`HPV-DNA
`
`HPV-DNA
`
`Nasopharyngeal cancer
`
`Nasopharyngeal cancer
`
`EBV-associated T-cell lymphoma
`
`EBV-associated gastric cancer
`
`Cervical cancer
`
`Cervical cancer
`Cervical cancer in situ
`
`HPV-DNA
`
`Head and neck squamous cancer
`
`§Quantitative PCR
`§§Controls: detection rate in gastritis patients 23%; in healthy controls 3.6%.
`EBV: Epstein-Barr virus; HPV: Human papilloma virus; Nd: Not determined.
`
`www.future-drugs.com
`
`00007
`
`

`

`Goessl
`
`and follow-up of treated cancer patients [8,19,70]. In nasopha-
`ryngeal carcinoma, postradiation levels of plasma EBV DNA
`were correlated with survival (progression-free and overall)
`and found to accurately reflect residual tumor load [70].
`Detection rates in HPV-associated cancer (cervical cancer and
`a few cases of head and neck cancer) are disappointingly low
`(<15%), thus limiting the clinical value of CNAPS analysis
`targeting HPV [73–75].
`
`Conclusion, expert opinion & five-year view
`Unfortunately, detection of tumor-derived signals encoded by
`nucleic acids in plasma/serum of cancer patients is usually
`dependent upon laborious amplification by PCR or RT-PCR.
`Each DNA or RNA marker requires unique primer sequences
`for PCR. Although a few target sequences may be amplified in
`parallel (multiplex PCR), until now, PCR-based techniques are
`rarely integrated into routine oncological decision making. An
`exception to this observation is the use of molecular techniques
`in virus-associated malignancies, for example, detection of
`HPV DNA in cervical smears already aids differential diagnosis
`of early cervical cancer [93]. With regard to CNAPS analysis,
`quantitative analysis of plasma/serum EBV DNA in virus-asso-
`ciated cases of nasopharyngeal carcinoma is on the verge of
`becoming a routine clinical application in geographic areas
`with high viral prevalence [7,8,70].
`What, however, will be the role of CNAPS-based detection
`of malignancy in nonvirus-associated tumors? It can be
`expected that within 5 years an increasing deg