The evidence base for circulating tumour DNA blood-based biomarkers for the early detection of cancer: a systematic mapping review
verfasst von:
Ian A. Cree, Lesley Uttley, Helen Buckley Woods, Hugh Kikuchi, Anne Reiman, Susan Harnan, Becky L. Whiteman, Sian Taylor Philips, Michael Messenger, Angela Cox, Dawn Teare, Orla Sheils, Jacqui Shaw, For the UK Early Cancer Detection Consortium
The presence of circulating cell-free DNA from tumours in blood (ctDNA) is of major importance to those interested in early cancer detection, as well as to those wishing to monitor tumour progression or diagnose the presence of activating mutations to guide treatment. In 2014, the UK Early Cancer Detection Consortium undertook a systematic mapping review of the literature to identify blood-based biomarkers with potential for the development of a non-invasive blood test for cancer screening, and which identified this as a major area of interest. This review builds on the mapping review to expand the ctDNA dataset to examine the best options for the detection of multiple cancer types.
Methods
The original mapping review was based on comprehensive searches of the electronic databases Medline, Embase, CINAHL, the Cochrane library, and Biosis to obtain relevant literature on blood-based biomarkers for cancer detection in humans (PROSPERO no. CRD42014010827). The abstracts for each paper were reviewed to determine whether validation data were reported, and then examined in full. Publications concentrating on monitoring of disease burden or mutations were excluded.
Results
The search identified 94 ctDNA studies meeting the criteria for review. All but 5 studies examined one cancer type, with breast, colorectal and lung cancers representing 60% of studies. The size and design of the studies varied widely. Controls were included in 77% of publications. The largest study included 640 patients, but the median study size was 65 cases and 35 controls, and the bulk of studies (71%) included less than 100 patients. Studies either estimated cfDNA levels non-specifically or tested for cancer-specific mutations or methylation changes (the majority using PCR-based methods).
Conclusion
We have systematically reviewed ctDNA blood biomarkers for the early detection of cancer. Pre-analytical, analytical, and post-analytical considerations were identified which need to be addressed before such biomarkers enter clinical practice. The value of small studies with no comparison between methods, or even the inclusion of controls is highly questionable, and larger validation studies will be required before such methods can be considered for early cancer detection.
Abkürzungen
14–3-3 s
14–3-3 sigma or tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein theta
ADAM
metallopeptidase with thrombospondin type 1 motif, 1
AIM1
absent in melanoma 1
ALU
Alu repeat/element 9e
APC
Adenomatous Polyposis Coli
ARF
alternate reading frame
BIN1
bridging integrator 1
BLU
zinc finger MYND-type containing 10
BM
biomarker
BNC1
basonuclin 1
bp
base pair
BRAF
B-Raf proto-oncogene, serine/threonine kinase
BRCA1
breast cancer 1, DNA repair associated
BRINP3
BMP/Retinoic Acid Inducible Neural Specific 3
CALCA
calcitonin related polypeptide alpha
CDH1
cadherin 1
CDH13
cadherin 13
CDO1
cysteine dioxygenase type 1
cfDNA
circulating cell-free DNA
CHD1
chromodomain helicase DNA binding protein 1
CHRM2
cholinergic receptor muscarinic 2
CINAHL
Cumulative Index to Nursing and Allied Health Literature
CLSI
Clinical & Laboratory Standards Institute
CRC
colorectal carcinoma
CST6
cystatin 6
ctDNA
circulating tumour DNA
CYCD2
cyclin D2
DAPK1
death-associated protein kinase 1
DCC
DCC Netrin 1 receptor
DCLK1
doublecortin like kinase 1
ddPCR
digital droplet polymerase chain reaction
DKK3
Dickkopf WNT signaling pathway inhibitor 3
DLEC1
deleted in lung and esophageal cancer 1
DNA
dexoxyribonucleic acid
ECDC
UK Early Cancer Detection Consortium
EGFR
epidermal growth factor receptor (HER1)
EP300
E1A binding protein P300
ERBB2
erb-B2 receptor tyrosine kinase 2 (HER2)
ESR
estrogen receptor 1
FAM5C
BMP/retinoic acid inducible neural specific 3 (BRINP3)
FDA
US Food and Drug Adminstration
FHIT
fragile histidine triad
FIT
faecal immunohistochemical testing
FoBT
faecal occult blood testing
GAPDH
glyceraldehyde-3-phosphate dehydrogenase
gCYC
cyclophilin A
GNA11
G protein subunit alpha 11
GNAQ
G protein subunit alpha Q
GPC3
glypican 3
GSTP1
glutathione S-transferase pi 1
HCC
hepatocellular carcinoma
HER1
human epidermal growth factor receptor 1
HER2
human epidermal growth factor receptor 2
HIC1
HIC ZBTB transcriptional repressor 1
HNSCC
head and neck squamous cell carcinoma
HOXA7
Homeobox A7
HOXA9
Homeobox A9
HOXD13
Homeobox D13
hTERT
human telomerase reverse transcriptase DNA
IgH
immunoglobulin heavy locus
INK4A
cyclin dependent kinase inhibitor 2A (CDKN2A/P16)
ISO
International Standards Organization
ITIH5
inter-alpha-trypsin inhibitor heavy chain family member 5
KLK10
kallikrein related peptidase 10
KRAS
KRAS Proto-Oncogene, GTPase
LCH
Langerhans cell histocytosis
LINE1
long interspersed nuclear element 1
LoH
loss of heterozygosity
Max
maximum
MDG1
microvascular endothelial differentiation gene 1
MGMT
O(6)-methyl-guanine-DNA methyltransferase
Min
minimum
MLH1
MutL Homolog 1
mtDNA
mitochondrial DNA
MYC
MYC proto-oncogene
MYF3
myogenic differentiation 1 (MYOD1)
MYLK
myosin light chain kinase
NGS
next generation sequencing
NICE
UK National Institute for Health and Care Excellence
NOS
not otherwise specified
OPCML
opioid binding protein/cell adhesion molecule like
Standards for Reporting of Diagnostic Accuracy Studies
TAC1
tachykinin precursor 1
TFPI2
tissue factor pathway inhibitor 2
THBD-M
thrombomodulin
TIMP3
tissue inhibitor of metalloproteinase 3
TMS
tumor differentially expressed protein 1
UCHL1
Ubiquitin C-Terminal Hydrolase L1
V600E
Mutation resulting in an amino acid substitution at position 600 in BRAF, from a valine (V) to a glutamic acid (E)
VHL
Von Hippel Lindau gene
ZFP42
ZFP42 Zinc Finger Protein
Background
The early detection of cancers before they metastasise to other organs allows definitive local treatment, resulting in excellent survival rates. This is particularly true for breast cancer, but also others, including lung and colorectal cancer [1]. Early detection and diagnosis has therefore been a major goal of cancer research for many years, and the concept of early detection from a blood sample has been the focus of considerable effort. However, to date no blood biomarkers have had sufficient sensitivity and specificity to warrant their clinical use for early cancer detection, and their potential remains unrealised [2]. Hanahan and Weinberg [3] identified the major biological attributes of cancer, and it is apparent that most if not all of these biological processes give rise to biomarkers present in blood [4]. Circulating cell free DNA produced from cancers is known as circulating tumour DNA (ctDNA), and represents a subset of the circulating DNA (cfDNA) normally present at low levels in the blood of healthy individuals.
Since the first description of circulating cfDNA in blood [5, 6], it has become clear that total ctDNA levels rise in a number of disorders in addition to cancer including myocardial infarction [7], serious infections, and inflammatory conditions [8], as well as pregnancy where it can be used for prenatal diagnosis [9]. The source of this DNA appears to be mainly the result of cell death – either by necrosis or apoptosis [5, 9‐11]. A raised ctDNA level is therefore non-specific, but may indicate the presence of serious disease. In blood, ctDNA is always present as small fragments, which makes assay design challenging [12]. Nevertheless, many analytical methods are available to measure ctDNA, and the field is rapidly maturing to the point where it may be clinically relevant to many patients.
Anzeige
In 2014, the UK Early Cancer Detection Consortium (ECDC) conducted a rapid mapping review of blood biomarkers of potential interest for cancer screening [13], and identified 814 biomarkers, including 39 ctDNA biomarkers. This paper uses the list generated from the mapping review, updated with relevant publications published since its completion to discuss the candidacy of ctDNA markers for early detection of cancer.
Methods
Our mapping review [13] conducted comprehensive searches of the electronic databases Medline, Embase, CINAHL, the Cochrane library, and Biosis to obtain relevant literature on blood-based biomarkers for cancer detection in humans (PROSPERO no. CRD42014010827). The search period finished in July 2014, therefore the searches have been updated to December 2016 using the same search terms. The abstracts of the publications retrieved were reviewed to identify those with validation data (usually indicated by case-control design) and to determine what ctDNA biomarkers had been measured in serum or plasma. Full details of the methods used are published elsewhere [13], and described briefly here. English language publications of any sample size were eligible and the full eligibility criteria used are provided in Table 1.
Table 1
Search criteria for ctDNA publications
Inclusion Criteria
Exclusion Criteria
English language studies
Studies published in non-English language
Studies within last seven years (2010–2016)
Studies published in 2009 or earlier
Controlled studies
Citation titles without abstracts
Validation Studies (comparison with controls)
Parallel publications and reviews based on the same or overlapping patient populationsa
Cancer detection/ diagnosis/screening
Prognosis or prediction (treatment response) associated markers
Biomarkers measured in blood plasma or serum
(markers or biomarkers)
Tissue, blood cells, or other bodily fluid samples
DNA (including cfDNA and ctDNA)
Abstracts of panels which do not state which biomarkers are studied
Human DNA
Viral and microbial DNA
aReviews and meta-analyses are cited, but not considered as evidence, but studies were included if they appeared to contain new data
The search strategy was deliberately inclusive, using keywords and subject headings as follows, to provide a comprehensive list of those ctDNA candidate biomarkers that had been used to identify cancers from blood samples. The search terms included ‘cancer’ ‘diagnosis’, ‘markers’, ‘blood’, and ‘screening’ with ‘DNA’, ‘cfDNA’, or ‘ctDNA’. Keywords and subject headings were determined by members of the ECDC working with the review team at the University of Sheffield. The results of the searches were collated in an Endnote database and results tabulated, with references, size of study, and methods used. To avoid bias, two reviewers conducted screening; references identified by either as relevant were included for further inspection. Those featuring ctDNA with data related to diagnosis or detection of three or more types of cancer were identified and retained for closer scrutiny to determine their potential utility.
Results
Following the updated searches and study selection, a total of 84 ctDNA markers were identified from 94 individual publications (Table 2 and Fig. 1).
Table 2
Individually identified markers with detection ability in ctDNA
CRC colorectal cancer, HNSCC head and neck squamous cell carcinoma, HCC hepatocellular carcinoma, LCH Langerhans cell histocytosis, SCLC small cell lung cancer
×
Anzeige
The ctDNA biomarkers divided naturally into two groups:
I.
those with potential specificity for neoplasia (ctDNA - usually mutations or DNA alterations such as methylation), and
II.
those designed to measure DNA levels, which may not be specific to neoplasia.
Figure 2 shows the distribution of studies by cancer type, including two publications on amplification [12, 14], and one on clonality [15]. One of the amplification papers looked at HER2 [14], while the other examined multiple targets by NGS [12].
×
Of the 94 publications included, 72 publications (77%) were case-control design diagnostic validation studies, and 22 were case series. The size and design of the studies varied widely. The largest study included 640 cancer patients [16]. The median study size was 65 cases, with a mean of 98 cases (range 12–640 cancer patients), indicating that the bulk of studies (67/94, 71%) included <100 patients (Fig. 3).
×
Most publications were focussed on ctDNA in plasma (n = 67) rather than serum (n = 25) with 2 comparing both. Plasma was used for 38 markers, and serum for 28 markers, and either for 18 markers (Fig. 4). Two comparative studies of serum and plasma were conducted: one for BRAF mutations, and the other for PIK3CA mutations [17, 18].
×
The target of ctDNA studies and the methods used to measure these targets varied considerably (Figs. 5 and 6 respectively). Non-specific total ctDNA levels (quantitation) were usually estimated by size distribution assays based on repeats: LINE1, and ALU were used in 3 [19‐21] and 6 publications respectively [20‐25]. However, some single genes were also used to measure DNA levels – particularly GAPDH in a series of 4 publications on breast cancer [26‐29], and hTERT in 4 publications [30‐33]. The majority of publications examined gene methylation markers (n = 49), though most examined methylation of multiple target genes for a particular tumour type (Fig. 5). Genes commonly mutated in cancer were also markers of interest, namely APC, BRAF, EGFR, HER2, GNAQ, GNA11, KRAS, P53, and PIK3CA. Only one gene, APC, was studied for both methylation and mutation. Few markers were used to identify particular tumour types, but some are particularly likely to occur in certain tumour types. GNAQ and GNA11 mutations have been identified in the plasma of uveal melanoma patients and are rare in other tumour types [34]. Other mutations are not tumour type-specific, and mutations in 6 of the 9 genes listed above were reported in multiple tumour types.
×
×
Discussion
The number of publications on ctDNA is increasing rapidly [35, 36], and a recent review emphasises the potential of the field [37]. Most (71%) are small case control studies with less than 100 patients, and in our view very few studies meet the requirements of analytical validation allowing their use within accredited (ISO:15,189) clinical laboratories, though some may have unpublished commercially-held analytical validation data. The stage and size of the tumours included is variable, and few studies are large enough to give robust subgroup assessments. Larger tumours produce more ctDNA, though tumour type also has an impact [16]. The value of small studies with no comparison between methods, or even the inclusion of controls is highly questionable. Most include a statement that ‘larger studies are required’, but larger trials rarely result due to the necessary cost implications. Unless well-designed prospective studies based on sample size calculations are performed, there is little likelihood of such methods reaching clinical practice for the detection of cancer at an early stage. There is also a likelihood of bias in that negative results for these markers are rarely if ever reported, and unlike clinical trials, there is no requirement for the registration of diagnostic validation studies. The use of ctDNA for early cancer detection comes under existing molecular pathology guidance, which emphasises the requirements for careful pre-analytical preparation, analysis, and reporting of results [38]. It is important that studies adhere to the Standards for Reporting of Diagnostic Accuracy Studies (STARD) guidance [39], and regional guidance (e.g. US Food and Drug Adminstration (FDA); UK National Institute for Health and Care Excellence (NICE); Clinical & Laboratory Standards Institute (CLSI)). It is hardly surprising then that, to date, no ctDNA markers have made it into screening programmes, due in part to the economic feasibility of completing the necessary stages of validation [40]. Nevertheless, there is encouraging evidence that ctDNA can be used to detect cancers of many types [16], and the poor quality of many studies should not detract from this fact.
A plethora of methods are available for ctDNA measurement, which have been well reviewed elsewhere [41]. BEAMing, PCR clamping methods, and deep sequencing using NGS are now the most commonly used [42, 43] and are widely regarded as the most sensitive methods currently available. A recent report of copy number variation (CNV) in breast cancer is not surprising given the ability of this method to detect such changes in pregnancy [15]. However, it should be noted that many of these methods are expensive. The development of highly sensitive NGS methods for ctDNA may prove necessary to obtain the best results [44], but large blood samples (> 10 ml may be needed as the number of DNA molecules present in small samples is often low) [45]. This may be at odds with the key requirement of cost effectiveness for screening programmes, and in our view this represents a real challenge for ctDNA. The problem is probably not insuperable if automation allows the integration of such methods into large blood sciences laboratories, but this is not as yet the case.
As ctDNA is composed largely of short fragments, short amplicons are required for maximum sensitivity of PCR reactions, particularly if mutations are being detected [46]. This is compounded by DNA loss in some reactions, particularly bisulphite modification of DNA, and it may be preferable to use nuclease protection assays [47, 48]. Methylation of key genes involved in carcinogenesis can be found in ctDNA, and has been studied by many groups, but it should be noted that substantial numbers of normal controls also have methylation of ctDNA for these genes [49].
Anzeige
It is clear that high sensitivity methods will be needed if ctDNA is to be used for early cancer detection. Several factors affect the sensitivity of ctDNA measurement. The first is the extraction method, and there are as yet too few studies which have compared the different options available, which now include automated instruments as well as manual extraction systems [50, 51]. The proportion of tumour derived DNA (ctDNA) in total cfDNA is greater in plasma than serum, and the higher ctDNA levels in serum are due to leakage from leukocytes during clotting [17]. The dilution effect for ctDNA in serum results in a reduced ability to detect mutations, particularly by methods with low analytical sensitivity [50]. Most groups working in the field realise this, and the majority of publications now look at plasma rather than serum.
Several publications were noteworthy, including one influential study which did not include healthy controls [16]. However, the comparison of DNA levels and multiple mutations in plasma from many different tumours types is helpful [44], and makes it clear that some tumours (e.g. gliomas) do not have high ctDNA levels in plasma, as previously found when comparing CSF with plasma [52]. This is also one of several publications that examines early stage disease, and shows that patients with localised disease have lower ctDNA levels [16]. Few publications have examined the ability of ctDNA to detect smaller tumours, though all agree that ctDNA levels increase as tumours enlarge [42].
Choice of target also influences results: the use of LINE1 and ALU repeats allows quantitative size distribution of DNA to be measured. Several publications suggest that this can distinguish cancer, and even pre-cancerous conditions from controls [30]. The size distribution of CRC appears to be different from other tumours due to first pass hepatic metabolism [20, 53]. Absolute quantitation by single gene methods such as GAPDH or hTERT will result in lower estimates of DNA content, and it is likely that this is due to the higher sensitivity of the ALU and LINE1 assays [30].
The use of mutations common within cancers is attractive, and the use of ctDNA to provide companion diagnostic information in patients in whom biopsy material is not available is now entering practice [54]. However, it should be noted that such mutations in P53 can occur in the blood of healthy controls, and could give rise to substantial numbers of false positive results [55].
Anzeige
Septin 9 methylation is often regarded as a model for future work [56, 57], and it is notable that there are some large studies [58] within the evidence base for the use of this marker in colorectal cancer, often used in addition to other markers, such as faecal occult blood testing (FoBT) or faecal immunohistochemical testing (FIT). Pre-analytical factors have been examined for this marker [59], including diurnal variation [60]. Plasma methylation of Septin 9 is now available as a commercial test (Epi proColon 2.0; Epigenomics AG, Berlin, Germany) which has recently obtained FDA approval for colorectal cancer screening (April 2016). This is the first blood test to be approved for cancer screening, and represents an encouraging milestone.
Other methylation targets have been studied in depth and show considerable promise. These include APC for colorectal cancer, with a large number of studies (Table 2), and SHOX3, for which a recent meta-analysis suggests that it could have an important role in the diagnosis of lung cancer [61].
There is an encouraging trend towards larger, more ambitious studies, supported by the commercial sector (e.g. (https://clinicaltrials.gov/ct2/show/NCT02889978, and https://clinicaltrials.gov/ct2/show/NCT03085888). Case control studies (particular retrospective ones) can give biased results, and prospective studies in at-risk cohorts would be more useful in examining the predictive capability of these markers. Such prospective studies should include controls proven not to have cancer. The comparison of new with existing methods (e.g. tumour markers, radiology), and competing technologies, is recommended, and often required by regulators. This has cost implications for funding bodies, but is essential if the field is to progress rapidly.
Conclusions
While ctDNA analysis may provide a viable option for the early detection of cancers, not all cancers are detectable using current methods. However, improvements in technology are rapidly overcoming some of the issues of analytical sensitivity, and it is likely that mutation and methylation analysis of ctDNA will improve specificity for the diagnosis of cancer.
Anzeige
Acknowledgements
We are grateful to the wider Early Cancer Detection Consortium for their assistance in putting together this paper, and for the many discussions which underpin it. Patient and Public representatives were involved in this work.
Funding
This work was conducted on behalf of the Early Cancer Detection Consortium, within the programme of work for work packages & 2. The Early Cancer Detection Consortium is funded by Cancer Research UK under grant number: C50028/A18554. It was subsequently supported by an unrestricted educational grant from PinPoint Cancer Ltd. (www.pinpointcancer.co.uk), following cessation of the grant in 2016. Neither of the two funding bodies had any input or influence over the design, study, collection, analysis, or interpretation of the data.
Availability of data and materials
The papers quoted are publically available from the publishers, and many are now open access.
Authors’ information
IC is a pathologist and has recently moved to a post with the International Agency for Research on Cancer of the World Health Organisation in Lyon. LU, and SH are Research Fellows in systematic review and HBW is an Information Specialist working at the University of Sheffield, UK. HK is a scientist and PhD student working on early cancer detection. AR is a Lecturer in Biomedical Science working at Coventry University, UK. STP is an associate professor with a NIHR Career Development Fellowship using quantitative research methods to assess new screening programmes. MM is a healthcare scientist at the University of Leeds with expertise in biomarker and in vitro diagnostic (IVD) development, validation and clinical evaluation. AC is Professor of Cancer Genetic Epidemiology at the University of Sheffield, UK. DT is Reader in Epidemiology and Biostatistics at the University of Sheffield, UK. OS is Director of the Trinity Translational Medicine Institute (TTMI) and Professor in Molecular Pathology at Trinity College Dublin, Eire. JS is Professor of Translational Cancer Genetics at Leicester University, UK, with a particular interest in cfDNA.
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The ECDC has grant funding for early cancer biomarker research from Cancer Research UK who funded this work. The ECDC involves several companies as follows: GE Healthcare, Life Technologies, NALIA Systems Ltd., and Perkin-Elmer. Individual ECDC members have declared their interests to the ECDC secretariat. IC was formerly chairman and CEO of PinPoint Cancer Ltd., a spin-out company from ECDC which in part funded the completion of this work though provision of staff time (IC). MM is supported by the National Institute for Health Research Diagnostic Evidence Co-operative Leeds. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the UK Department of Health.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
The evidence base for circulating tumour DNA blood-based biomarkers for the early detection of cancer: a systematic mapping review
verfasst von
Ian A. Cree Lesley Uttley Helen Buckley Woods Hugh Kikuchi Anne Reiman Susan Harnan Becky L. Whiteman Sian Taylor Philips Michael Messenger Angela Cox Dawn Teare Orla Sheils Jacqui Shaw For the UK Early Cancer Detection Consortium