Novel DNA methylation biomarkers show high sensitivity and specificity for blood-based detection of colorectal cancer—a clinical biomarker discovery and validation study
Early detection plays an essential role to reduce colorectal cancer (CRC) mortality. While current screening methods suffer from poor compliance, liquid biopsy-based strategies for cancer detection is rapidly gaining promise. Here, we describe the development of TriMeth, a minimal-invasive blood-based test for detection of early-stage colorectal cancer. The test is based on assessment of three tumour-specific DNA methylation markers in circulating cell-free DNA.
A thorough multi-step biomarker discovery study based on DNA methylation profiles of more than 5000 tumours and blood cell populations identified CRC-specific DNA methylation markers. The DNA methylation patterns of biomarker candidates were validated by bisulfite sequencing and methylation-specific droplet digital PCR in CRC tumour tissue and peripheral blood leucocytes. The three best performing markers were first applied to plasma from 113 primarily early-stage CRC patients and 87 age- and gender-matched colonoscopy-verified controls. Based on this, the test scoring algorithm was locked, and then TriMeth was validated in an independent cohort comprising 143 CRC patients and 91 controls. Three DNA methylation markers, C9orf50, KCNQ5, and CLIP4, were identified, each capable of discriminating plasma from colorectal cancer patients and healthy individuals (areas under the curve 0.86, 0.91, and 0.88). When combined in the TriMeth test, an average sensitivity of 85% (218/256) was observed (stage I: 80% (33/41), stage II: 85% (121/143), stage III: 89% (49/55), and stage IV: 88% (15/17)) at 99% (176/178) specificity in two independent plasma cohorts.
TriMeth enables detection of early-stage colorectal cancer with high sensitivity and specificity. The reported results underline the potential utility of DNA methylation-based detection of circulating tumour DNA in the clinical management of colorectal cancer.
KeywordsDNA methylation Epigenetic biomarkers Cancer Colorectal cancer Liquid biopsy Circulating tumour DNA Early detection
Infinium HumanMethylation450K BeadChip
Acute Myeloid Leukaemia
Area under the curve
Circulating cell-free DNA
Circulating tumour DNA
Droplet digital PCR
Faecal Immunochemical Test
Head and neck squamous cell carcinoma
Peripheral blood leukocytes
Receiver operating characteristics
Union for International Cancer Control
Colorectal cancer (CRC) claims more than 880,000 lives each year worldwide and is a major public health concern in the Western world [1, 2]. Much of the morbidity and mortality of CRC result from diagnosis at late stages, where the therapeutic intervention is less effective. CRC screening using faecal occult blood testing and bowel endoscopy has been shown to enable early detection and reduce CRC mortality [3, 4]. However, the compliance rates in CRC screening, based on either direct endoscopy or testing for occult blood in faeces and subsequent colonoscopy, are poor to modest . A recent study evaluated the sample preference, blood or faeces, for a CRC screening test among screening-aged individuals and found that 78% of the survey participants preferred to provide a blood sample . Hence, blood-based tests could potentially improve compliance in population-based screening programmes, given their minimally invasive nature and straightforward implementation in routine medical examinations . Blood contains numerous analytes, including circulating cell-free (cfDNA). We and others have previously shown that in individuals with cancer, some of the cfDNA may originate from the tumour (circulating tumour DNA, ctDNA) [8, 9, 10, 11]. Thus, cfDNA has the potential to distinguish healthy individuals from cancer patients. Recently, analyses using DNA mutation and methylation-based strategies for detection of ctDNA have suggested that such approaches may provide new avenues for early cancer diagnosis [12, 13, 14]. While both strategies have shown promises, mutation-based strategies are particularly challenged by the limited number of recurrent mutations available to distinguish tumour and normal cfDNA in a cost-efficient manner. By contrast, tumour-specific DNA hypermethylation occurs early in tumour development and is highly recurrent . Consistently, several promising DNA methylation markers have been reported [16, 17, 18], though none have yet shown sufficient clinical performance to be considered implemented in CRC screening . Here, we report the results of a combined discovery and validation study, aimed at identifying novel blood-based DNA methylation markers, and document their ability to efficiently discriminate healthy individuals from patients with early-stage CRC.
Patient characteristics and demographics of plasma cohorts
Sex n (%)
UICC stage, n (%)
Tumour diameter (mm)
Histological type n (%)
Signet ring cell adenocarcinoma
Unspecified or missing
Localization n (%)
Right (cecum, ascending, transverse)
Left (descending, sigmoid)
DNA methylation biomarker discovery and validation of candidate marker regions by bisulfite sequencing
Biomarker candidate assay development and technical validation
An average of two methylation-specific ddPCR assays, targeting bisulfite-converted DNA, were designed, tested, and optimized for the 29 CpG candidate sites, 58 assays in total (Additional file 7). The validation tests and the order in which they were performed are described in Fig. 2. The technical sensitivity of the assays was evaluated using a 7-point dilution series, where 0, 8, 16, 32, 64, 128, and 256 methylated DNA copies were mixed with 20,000 human unmethylated DNA copies (data not shown). The best performing assay, based on linearity and sensitivity, was selected for each candidate CpG site, leaving one assay per CpG site (Additional file 9: Table S3). All selected assays were able to detect 8 copies of methylated DNA in a background of 20,000 copies of unmethylated DNA. None of the assays amplified unmethylated DNA.
Biological validation of biomarker candidates in clinical tissue and plasma samples
Biomarker evaluation in plasma from early-stage CRC patients and matched controls
Independent validation of TriMeth
Quantities of methylated DNA fragments in cases and controls
The basis of the present CRC detection approach is that the methylated DNA templates detected in plasma are derived from dying cancer cells. In agreement, methylated DNA was rarely detected in cfDNA from controls (Figs. 4 and 5). Only 1% (2/178) of the controls were positive and most often by one marker only. Moreover, these positive control samples contained a median of only 0.1 (95% CI 0.1–0.2) methylated DNA fragments/ml plasma. By contrast, methylated DNA was detected, by two or more markers, in 85% (218/256) of samples from CRC patients, with a median of 1.3 (95% CI 0.9–2.0) methylated fragments/ml plasma.
Early detection is key to increase eligibility for curative intervention and thereby reduce CRC mortality. CRC screening has proven efficient for early detection of cancerous lesions, but current screening strategies suffer from low compliance rates. Here, we report the identification of three CRC-specific DNA methylation markers C9orf50, KCNQ5, and CLIP4 and demonstrate their utility (the TriMeth test) for detection of CRC-specific ctDNA in human blood samples. TriMeth was applied to plasma from two independent cohorts and on average detected ctDNA in 85% (218/256) of CRC patients. At this sensitivity, the specificity was 99% in both cohorts, only 2 of 178 controls scored positive. We cannot be certain that the two “false positive” individuals did not have a CRC that was missed by the colonoscopy, but classifying them as false positives provides the most conservative approach to interpretation of the data. One of the most important attributes of a screening test is the ability to detect early-stage cancers. In the present study, the TriMeth test achieved high sensitivity for all stages, with average sensitivities of 80% for stage I, 85% for stage II, 89% for stage III, and 88% for stage IV. While plasma from patients with colorectal adenomas, the benign counterpart to adenocarcinomas, were not investigated in this study, our analyses of adenoma tumour tissues revealed the same TriMeth signals as found in colorectal cancer tissues (Fig. 3a). Hence, if adenomas shed tumour DNA to the circulation then they can potentially be detected. In the future, TriMeth-based identification and removal of adenomas may be a path to reduce CRC incidence.
In the marker discovery phase of the present study, extensive efforts were made to ensure that the selected DNA methylation markers were specifically hypermethylated in CRC compared to other cancer types and normal cells of haematopoietic origin. However, we cannot exclude that other normal tissues might have a DNA methylation profile similar to that observed in CRC. One can imagine that such tissues, in some situations (e.g. disease-related), may shed DNA into the circulation and cause single markers to become positive, similar to what we observed. To overcome this issue, we included three synchronously methylated CRC markers in the TriMeth test. We hypothesized that while normal tissues might have a DNA methylation pattern similar to CRC at a single site, it was unlikely to happen at multiple sites. Consequently, we expected to be able to discriminate methylated DNA released from cancer cells and other cells, by requiring at least two out of three markers to be methylated, and indeed this was confirmed. TriMeth showed a specificity of 99%. The sensitivity and, particularly the specificity, of TriMeth are favourable to that of the frequently used FIT test. In a recent meta-analysis, FIT was reported to have a sensitivity of 71% (95% CI 58–81%) and a specificity of 94% (95% CI 91–96%) when the reference standard was colonoscopy, as in the present study . This suggests that TriMeth may have potential as a screening test. As TriMeth is blood-based, a patient compliance higher than for FIT may be expected , which could potentially lead to detection of a larger proportion of the CRCs in the screening population. TriMeth could potentially also be used to supplement existing FIT-based CRC screening programmes, e.g. as an option for the invitees that refuse the FIT test, or for triaging FIT-positives to colonoscopy to reduce the number of colonoscopies needed to detect one CRC . Potentially, TriMeth could also be used in a postoperative setting, to identify patients with minimal residual disease and relapse. We observed significant inter-patient variation in the cfDNA quantity per millilitre plasma. To minimize the risk of falsely classifying a sample as negative due to insufficient cfDNA input, equal quantities of cfDNA input were used in the test cohort. However, analysing equal amounts of cfDNA is not practically feasible. Hence, 16 ml of plasma was used in the validation cohort, which ensured that a minimum of 5000 cfDNA copies were analysed per ddPCR for > 90% of samples. This volume of plasma exceeds what previous studies have used, but it is indeed feasible in a clinical setting . However, the ability to detect very early-stage tumours and adenomas will ultimately be limited by the level of tumour DNA shedding and the presence of ctDNA fragments in the collected blood volume. It may very well be that small and early-stage tumours, which are shedding only limited amounts of DNA to the circulation, may be undetectable at standardly collected blood volumes. If the number of tumour DNA fragments in the available blood volume falls below the detection threshold of the used methods, the lesion will go undetected. A few limitations of the study should be acknowledged. Firstly, the CRC patients were individuals with known cancers, most of which were diagnosed on the basis of symptoms. The fraction of stage I tumours will probably be higher among asymptomatic, screened individuals, and consequently the sensitivity of detection in a screening population might be less than reported here. To limit this bias, the inclusion was focused on CRC patients with early-stage disease to mimic a true screening setting. Secondly, while the controls were recruited among asymptomatic individuals participating in the Danish national CRC screening programme, they were selected among the FIT-positive and colonoscopy-negative subset. Consequently, they may not reflect the screening population in all details. Thirdly, because our markers show weak to moderate DNA methylation signals in other gastrointestinal cancers, particular gastric cancer (Fig. 3a), there is a risk that TriMeth might become positive in a fraction of non-CRC gastrointestinal cancer patients. Consequently, in a clinical setting it will be important to consider the clinical follow-up procedure after a positive TriMeth test. For instance, if the follow-up colonoscopy is negative, it may be advisable to do a gastroscopy.
In summary, we have developed and validated a sensitive and specific DNA methylation marker TriMeth test for the detection of ctDNA released by CRCs. TriMeth awaits validation in an asymptomatic setting, but the findings reported here emphasize the potential utility of our DNA methylation markers as a basis for minimally invasive, blood-based, sensitive, and specific early tumour detection for cancer interception.
This study presents a multi-phased marker discovery and validation study with retrospective analysis of cfDNA using Locked Nucleic Acid™-enhanced methylation-specific ddPCR assays, to detect CRC-specific DNA methylation alterations in plasma from CRC patients and controls. We analysed plasma from 434 individuals, including 178 controls (FIT-positive and colonoscopy-negative) and 256 stage I-IV CRC patients with most patients exhibiting early-stage disease (Table 1). The sensitivity and specificity of three DNA methylation markers were evaluated in two independent plasma cohorts. No statistical methods were used to predetermine sample size.
Between May 2014 and December 2014, pre-operative plasma was collected from 256 patients diagnosed with stage I-IV CRC at the Surgical Departments of Aarhus University Hospital and the Regional Hospitals in Randers and Herning. In the same period, control plasma was collected from 178 age- and gender-matched FIT-positive participants with colonoscopy-verified clean colons, no previous cancer diagnosis, and no comorbidities except for hypertension in the Danish colorectal cancer screening programme . The samples were organized in two cohorts and general demographic information is presented in Table 1. Informed consent was obtained from all participating patients and controls. Tumour samples from CRC patients were collected at the Surgical Departments of Aarhus University Hospital, Randers Regional Hospital and Herning Hospital in Denmark. The tissue was snap-frozen in liquid nitrogen within 30 min from end of tumour resection and stored at − 80 °C. PBLs were isolated from 10 ml of blood collected from presumed healthy Danish blood donors at the Blood Bank, Aarhus University Hospital, Denmark.
Biomarker discovery and filter criteria
For the marker discovery, we used DNA methylation datasets generated by 450K arrays, which quantifies the DNA methylation levels of 482,421 CpG loci by calculating the ratio (β-value) of intensities between methylated and unmethylated alleles. The 450K data was either available in-house [24, 25] or through “Marmal-aid”, a public database for Infinium HumanMethylation450 datasets . The public data were primarily generated by “The Cancer Genome Atlas” project (https://portal.gdc.cancer.gov/). All datasets were processed (β-value calling and normalization) using standard settings by the ChAMP R-package . CRC-specific marker candidate CpG sites were identified using the filter steps shown in Fig. 2, and 50 CpG sites were selected for assay design and further evaluation.
Primers flanking selected candidate CpG sites were designed using Bisearch . The primers were placed 100–250 nucleotides upstream and downstream of the index CpG site, to ensure that the amplicon covered several CpG sites. In order to amplify the candidate regions of interest, ranging from ~ 200–500 nucleotides, PCR reaction mixes were prepared containing 1.5 μl Tempase Key Buffer (Amplicon), 0.2 μl Tempase Hot Start DNA Polymerase (Amplicon), 1.5 μl dNTP mix (1.25 mM) (Roche), 0.5 forward primer (10 μM), 0.5 reverse primer (10 μM), and 9.8 μl AccuGENE™ Molecular Biology Water (Lonza). One microlitre bisulfite-converted template DNA from colorectal tumours, PBLs, or normal colorectal mucosa was added to a final volume of 15 μl and PCR reactions were run on a C1000 thermal cycler (Bio-Rad Laboratories). Primers are listed in Additional file 9: Table S2. Following PCR amplification, 5 μl of each amplicon were visualized by agarose gel electrophoresis in Tris-acetate-EDTA (TAE) buffer (Fagron) with ethidium bromide (EtBr) to test for correct band size. GeneRuler™ 100 bp Plus DNA Ladder (Thermo Scientific) was used as a molecular marker. To remove excess primers and dNTPs, 2 μl PCR products were treated with 1 μl FastAP Thermosensitive Alkaline Phosphatase (1 μmol/μl) (Life Technologies) and 1 μl EXO1 (20,000 U/ml) (New England Biolabs) diluted 1:4 in 10x Exonuclease I Reaction buffer (New England Biolabs). The mix was incubated on a C1000 thermal cycler (Bio-Rad) at 37 °C for 15 min and 85 °C for 15 min. Sanger sequencing was performed using 0.5 μl Ready Reaction mix (Life Technologies), 1.5 μl Big-Dye sequencing buffer (Life Technologies), 1 μl primer (2 pmol/μl), 2 μl purified PCR product, and RNase-free water to a total volume of 10 μl. The product was ethanol/EDTA/na-acetate precipitated and sequenced on a 3130x Genetic Analyser (Applied Biosystem). Results were analysed using Sequencer 5.1 software (Gene Codes Cooperation).
Design and optimization of methylation-specific droplet digital PCR assays
Methylation-specific ddPCR primers and probes were designed to cover candidate regions validated by bisulfite sequencing and be exclusively specific for methylated, bisulfite-treated DNA. To increase assay specificity and reduce amplicon lengths, LNA™, that have an increased affinity for complementary DNA bases , were incorporated into primers and probes. In the assay design, the LNA™ Oligo Optimizer tool (https://www.exiqon.com/ls/Pages/ExiqonOligoOptimizerTool.aspx) was used to ensure LNA™ oligo designs with a self-complementarity and secondary structure score below 40. The annealing temperature (Tm) of the LNA™-enhanced oligos was predicted using the LNA™ Oligo Tm Prediction tool (https://www.exiqon.com/ls/Pages/ExiqonTMPredictionTool.aspx). Primers were designed to have Tm = 59–61 °C and probe Tm = primer Tm + 5–10 °C. Primers were manufactured by Qiagen and probes manufactured by LGC biosearch technology. All assays were optimized for ddPCR according to the guidelines for minimum information for publication of quantitative digital PCR experiments  (digital MIQE checklist shown in Additional file 9: Table S6). Primer and probe sequences and assay details are shown in Additional file 9: Table S3.
Tissue and blood processing, including DNA isolation
DNA was extracted from fresh frozen tissue using the Gentra Puregene Tissue Kit (Qiagen) as specified by manufacturer. DNA from PBLs was purified on a QIAsymphony robot (Qiagen) using the QiaSymphony DSP DNA mini kit (Qiagen) as specified by manufacturer, eluted in 1.5 ml Eppendorf tubes (Eppendorf AG) and stored at − 80 °C until use (< 2 months). Whole blood was collected in BD Vacutainer K2 EDTA tubes (Becton Dickinson) and processed within 2 h from venipuncture. Blood from controls was collected after bowel cleansing but prior to colonoscopy, and blood from CRC patients was collected prior to surgery. Blood samples from CRC patients and controls were processed identically. To separate plasma from cellular components, plasma was double centrifuged at 3000g for 10 min at 20 °C and stored in cryotubes (TPP) at − 80 °C until the time of DNA extraction (< 3 years). Plasma was thawed at room temperature and cfDNA from 8 to 24 ml of plasma was extracted using a QIAsymphony robot and the QIAamp® Circulating Nucleic Acids kit (Qiagen) as specified by manufacturer. Purified cfDNA was eluted in LoBind tubes or LoBind 96-well plates (Eppendorf AG) and stored at − 80° until further use (< 2 months). Purification efficiency and analysis for contamination with DNA from lysed lymphocytes were assessed by ddPCR as previously described . In brief, a fixed amount of soybean CPP1 DNA fragments was added to each plasma sample prior to extraction. Purification efficiency was calculated as the percent recovery of CPP1 fragments following cfDNA extraction (CPP1 assay). Lymphocyte DNA contamination was estimated by an assay targeting the VDJ rearranged IGH locus specific for B cells (PBC assay). The median purification efficiency was 74.4% (interquartile range 65.8–84.0%). A minor contamination with lymphocyte DNA, was observed in 6.8% of samples, but since their cfDNA levels did not deviate from the rest, these samples were flagged rather than excluded.
Prior to bisulfite conversion, cfDNA was dried using vacuum centrifugation (speedVac, Concentrator plus 5350, Eppendorf AG) at 30 °C and resuspended in 20 μl AccuGENE™ Molecular Biology Water (Lonza). All DNA samples were bisulfite-converted using the EZ-96 DNA Methylation-Direct™ MagPrep kit (Zymo Research) according to manufacturer’s instructions, but with the following modifications to the volumes used: 60 μl CT conversion reagent, 280 μl M-Binding Buffer, 5 μl MagBinding Beads, 185 μl M-Wash Buffer, 93 μl M-Desulphonation Buffer, and 25 μl M-Elution Buffer. Methylated and unmethylated DNA standards (Zymo Research) were included in each bisulfite conversion batch, as positive and negative controls. Reactions were performed on a S1000 Thermal cycler (Bio-Rad). The bisulfite-converted DNA samples were analysed using ddPCR immediately after completed bisulfite conversion or stored at − 20 °C until use (< 2 months).
DNA quantification before and after bisulfite conversion
Native DNA samples were quantified by ddPCR using assays targeting two reference regions located on chromosome 1 (CF assay) and chromosome 3 (Chr3 assay) (Additional file 9: Table S5). Both assays are located in regions that only rarely show copy number aberrations in cancer. Reported quantities are the average of the two assays. The CF assay was furthermore designed to amplify a cytosine-free region of the genome, thereby, enabling the use of the same assay for quantification of both native and bisulfite-converted DNA. The CF assay was used for DNA quantification and recovery assessments after bisulfite conversion. Recovery was calculated as the CF quantity after bisulfite conversion divided by the CF quantity before. Using the same assay before and after bisulfite treatment facilitates an unbiased recovery estimate.
Droplet digital PCR
All reagents, except from template DNA, were prepared in an isolated pre-PCR room to avoid contamination. The reaction master mix was prepared as follows: 2–9 μl template DNA, 18 pmol forward primer, 18 pmol reverse primer, 5 pmol probe, 2xSupermix for Probes no UTP (Bio-Rad), and AccuGENE™ Molecular Biology Water (Lonza) to a final volume of 22 μl. Complete lists of applied ddPCR assays are provided in Additional file 9: Tables S3 and S5. One-nanoliter droplets were generated on the QX200 AutoDG Droplet Generator (Bio-Rad). The median number of droplets (partitions) was 16,218 (interquartile range 14,896–17,235). After droplet generation, samples were amplified by PCR in a S1000 Thermal cycler (Bio-Rad) at 95 °C for 10 min and 45 cycles of 95 °C for 30 s, 56 °C for 1 min, and 98 °C for 10 min. Amplified samples were stored at 4 °C for up to 17 h before analysis on the QX200 reader (Bio-Rad). Positive and no-template controls were included for each assay in each plate. Furthermore, for methylation-specific assays a negative control was also included. For methylation-specific assays, the positive and negative controls were 5 ng human methylated and 66 ng non-methylated DNA standards (Zymo Research), respectively. For the CF, Chr3, and PBC assays the positive control was 5 ng human leukocyte DNA. For the CPP1 assay, the positive control was 7000 CPP1 DNA fragments. For fresh frozen tumour tissue and PBL test samples, the DNA input was 5 ng and 66 ng, respectively (quantified prior to bisulfite conversion). Quantasoft v1.7 software (Bio-Rad) with standard settings was used for analysis of ddPCR data from all, but the plasma samples. Plasma samples were analysed using a custom analysis pipeline (see the “Data analysis” section below).
Methylation-specific droplet digital PCR
Plasma samples were analysed on the Droplet Digital PCR System (Bio-Rad) according to manufacturer’s instructions (Bio-Rad) and performed in accordance with the Minimum Information for Publication of Quantitiative Digital PCR Experiments  (dMIQE) guidelines (Additional file 9: Table S6). ddPCR assay information is provided in Additional file 9: Table S3. Each plate included a positive control (5 ng human methylated DNA), a negative control (66 ng human unmethylated DNA), and a no-template control.
The raw fluorescence intensity data for all individual droplets in each well was extracted using Quantasoft and analysed plate-wise. We used fluorescence data from a fully methylated positive control sample on each plate to identify fluorescence maxima (for the negative and positive droplet populations) and minimum. This was done using a Gaussian kernel density estimator with the smallest possible bandwidth that identified exactly two maxima and one minimum (Additional file 8: Figure S8). All test samples on the plate were subsequently normalized to the median fluorescence of the negative population from the positive control. The fluorescence threshold for calling droplets positive or negative was finally set for all wells, at the minimum point between the negative and positive populations as defined by the positive control sample. The concentration c (copies per well) of methylated DNA calculated as c = − N × ln(1 − P/N), where N is the total number of droplets and P is the number of positive droplets . The code in the R language is available at GitHub (https://github.com/MOMA-CRC/ddanalyzor.git).
The predictive accuracy of the individual markers C9orf50, KCNQ5, and CLIP4 was estimated by ROC analysis using the R package ROCR. The sensitivity and specificity of the TriMeth test were estimated with corresponding 95% confidence intervals.
We thank Birgitte Trolle, Lisbet Kjeldsen, Maria Engtoft Skjøtt, Margaret Elizabeth Gellett, Gitte Glistrup Nielsen, Jesper Boulund Kristensen, Mie Aarup, Lotte Gernyx, Marianne Lysdahl, and Mette Carlsen Mohr for highly skilled technical assistance. The research nurses, technicians, and secretaries employed at the Endoscopy III study and “The Danish Collaborative Research Group on Early Detection of Colorectal Cancer” are thanked for their skillful work with inclusion, blood collection and handling, recording of data, and final audits. The Danish Collaborative Research Group on Early Detection of Colorectal cancer includes the following:
Lars Nannestad Jørgensen and Morten Rasmussen (Bispebjerg Hospital, Copenhagen), Jakob W. Hendel (Herlev Hospital, Herlev), Mogens R. Madsen and Anders H. Madsen (Herning Hospital, Herning), Jesper Vilandt and Thore Hillig (Herning Hospital, Herning), Søren Brandsborg (Horsens Hospital, Horsens), Linnea Ferm and Eva Rømer (Hvidovre Hospital, Hvidovre), Tobias Boest (Randers Hospital, Randers), and Ali Khalid (Viborg Hospital, Viborg). We thank Christina Demuth for critically commenting the intellectual content of the manuscript. We thank the blood donors at the Blood Bank, Aarhus University Hospital, Denmark, participants in the Danish national CRC screening programme, and CRC patients for contribution of clinical material. The Danish Cancer Biobank is acknowledged for providing access to tissue and blood materials.
SØJ, NØ, MWØ, JBB, and CLA conceived and designed the study. SØJ, NØ, MWØ, MHR, HK, PM, JBB, and CLA acquired, analysed, and interpreted the data. MRM, AHM, KGS, LHI, SL, IJC, and HJN obtained informed consent and acquired patient samples and clinical information. SØJ wrote the first draft of the manuscript with contributions from MWØ and CLA. all authors critically read and revised the manuscript and are accountable for all aspects of the work. CLA supervised the study. All authors read and approved the final manuscript.
This work was supported by The Danish Cancer Society (R107-A7035, R133-A8520-00-S41, and R146-A9466-16-S2), The Danish Council for Strategic Research (1309-00006B), The Danish Council for Independent Research (4183–00619), The Novo Nordisk Foundation (NNF14OC0012747 and NNF17OC0025052), The Neye Foundation, Dansk Kræftforskningsfond, and Harboefonden. The Endoscopy III study received research grants from The Augustinus Foundation, The Kornerup Fund, The Axel Muusfeldt Fund, The KID Fund, The Toyota Fund, The Aage and Johanne Louis-Hansen Fund, The Orient Fund, The Walter and O. Kristiane Christensen Fund, The A.P. Moeller and Chastine Mc-Kinney Moeller Foundation, The P.M. Christiansen Family Fund, The Aase and Ejnar Danielsen Fund, The Inger Bonnén Fund, The Hans and Nora Buchard Fund, The Elna and Jørgen Fagerholdt Fund, The Sofus C.E. Friis Fund, The Eva and Henry Fraenkel Fund, The Sven and Ina Hansen Fund, The Henrik Henriksen Fund, The Jørgen Holm Fund, The Humanitarian Fund, The IMK Fund, Foundation Jochum, The Obel Family Fund and Krista, and Viggo Petersen Fund.
Ethics approval and consent to participate
The prospective collection of plasma samples from participants in the Danish national CRC screening programme have been approved by the Scientific Ethical Committee of the Capital Region Denmark (j. no. H-4-2013-050) and the Danish Data protection Agency (j. nos. HVH-2013-022/2007-58-0015). The Committee on Health Research Ethics (j. no. 1999/4678 and H-3-2009-110) and the Danish Data Protection Agency (j. no. 2007-58-0010 and 2008-41-2252) have approved the use of the existing tissue collections.
Consent for publication
The authors declare that they have no competing interests.
- 2.GLOBOCAN: World Health Organization; 2019 [Available from: https://gco.iarc.fr/.
- 23.Rasmussen L, Wilhelmsen M, Christensen IJ, Andersen J, Jørgensen LN, Rasmussen M, et al. Protocol outlines for parts 1 and 2 of the prospective endoscopy III study for the early detection of colorectal cancer: validation of a concept based on blood biomarkers. JMIR Res Protoc. 2016;5(3):e182.CrossRefGoogle Scholar
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.