Skip to main content

BARDiag - Biomarker tests for early cancer detection

Final Report Summary - BARDIAG (BARDiag - Biomarker tests for early cancer detection)

Executive Summary:
Executive Summary
Lung and colorectal cancers cause millions of death each year. For lung cancers in particular, there are worldwide 2 million persons diagnosed each year and 1.3 million who die from this cancer; -this makes lung cancer the most deadly cancer. Many of these deaths could be avoided if cancers would be diagnosed or suspected earlier. To be applicable to the most, an optimal test should be highly sensitive, reliable, affordable, easy to perform, and should probably be repeated over several years to follow-up patients at higher risks. Currently, such a test does not exist, ie there is no suitable non-invasive method for the early detection of these types of cancers.
The tumour suppressor gene BARD1 (BRCA1-associated RING domain protein) is aberrantly expressed in several types of cancer and could be a diagnostic target for early cancer diagnosis in blood samples. The overall objectives of the project were to develop blood tests for the early detection of colorectal and lung cancer based on cancer-specific BARD1 isoforms. The outlined tests were to analyse BARD1 isoforms at two levels: the expression of isoform-specific RNAs in circulating tumour cells (CTC) and the presence of autoantibodies in serum.
At the end of this project, BARD1AG owns a proprietary test for the early detection of lung cancers. This test is based on a proprietary signature of 20 peptides spotted on Mesoscale Multiplex plates; this peptide signature has been protected by BARD1AG through the filing of a patent application. Most importantly, the performance of the test has been evaluated using 178 lung cancer samples and 266 controls. The test performance is excellent with an AUC under the ROC curve of 0.97. BARD1AG is in advanced discussion with a pharma partner to further validate, manufacture and distribute its proprietary lung cancer screening test. This may represent a true breakthrough in lung cancer diagnostic, and indicates the level of success achieved in this proposal. In parallel, fewer colon cancer samples were included in the analysis, and thus the predictive statistical power of the final signature for colon cancers (which is different form the one for lung cancers) is weaker. However, and thanks to the first positive statistical results obtained, BARD1AG will pursue the statistical validation of a colorectal peptide signature. Finally, the progress of the BARDiag project allowed PluriSelect to develop and optimize its system for the fast and easy isolation of CTCs directly from whole blood. This has resulted in the setup of a Kit containing EpCAM or CK-19 specific pluriBeads and an optimized protocol. The target groups of this development are customers that intend to isolate CTC from whole blood for molecular analysis. The Kit is currently not suitable for cell analysis.


Project Context and Objectives:
Project context:
Lung and colorectal cancers cause millions of death each year. For lung cancers in particular, there are worldwide 2 million persons diagnosed each year and 1.3 million who die from this cancer; -this makes lung cancer the most deadly cancer. Many of these deaths could be avoided if lung cancer would be diagnosed or suspected earlier. To be applicable to the most, an optimal test should be highly sensitive, reliable, affordable, easy to perform, and should probably be repeated over several years to follow-up patients at higher risks. Currently, such a test does not exist, ie there is no suitable non-invasive method for the early detection of these types of cancers.
The tumour suppressor gene BARD1 (BRCA1-associated RING domain protein) is aberrantly expressed in several types of cancer and could be a diagnostic target for early cancer diagnosis in blood samples. The overall objectives of the project were to develop blood tests for the early detection of colorectal and lung cancer based on cancer-specific BARD1 isoforms. The outlined tests will analyse BARD1 isoforms at two levels: the expression of isoform-specific RNAs in circulating tumour cells (CTC) and the presence of autoantibodies in serum. To reach these objectives, several technological challenges had to be overcome. The BARDiag consortium includes 3 SMEs and 4 research centres, who have excellent expertise, specific knowledge, the required lab infrastructure and necessary clinical materials that will enable the conduct of this project. Within the frame of the project, innovative methods for the isolation of CTCs in colorectal and lung cancer patients were planned to be developed, specific signatures of the BARD1 isoforms at both mRNA and autoimmune levels should be defined, and assays for the detection of these isoforms should be established, validated with clinical samples and tested for their marketability. The results of the proposed project will have extensive impacts. Not only more scientific knowledge on the expression of BARD1 isoforms in colorectal and lung cancer will be obtained and therefore the understanding of these cancers will be improved, but also non-invasive methods for the early detection of colorectal and lung cancer will be made broadly available in the form of commercial test kits. The SMEs will extend their expertise and knowledge and therefore strengthen their economic power, which will contribute to increase European competitiveness.

Project background 1: The BARD1 protein
The BRCA1 associated RING domain 1 protein, BARD1, is a fascinating protein with two highly relevant properties that makes it highly attractive for cancer diagnosis: first, it is secreted (ie outside the cell body) in large amount by cancer cells, and may thus give rise to the generation of large amount of circulating antibodies. These antibodies might represent early and specific circulating, and therefore easily accessible markers of different cancers. Second, cancer-specific isoforms are generated by alternate splicing of the BARD1 gene. This leads to RNA-based "signatures" that may lead to cancer-specific diagnosis; these signatures may also be used as predictive markers of treatments.
Based on the strength and the limitations of the use of the BARD1 protein to sustain the development of cancer diagnostic tools, we have herein proposed the development of two distinct, but potentially sequentially complementary tests. Importantly, and in contrast to many competing technologies, our aim is to provide for non-invasive testing based on the collection of blood samples.
1. a very simple, non-invasive, test based on an easy-to-perform small (that can be as low as a drop of blood) blood sampling. The analysis will then be based on a sophisticated "ELISA" test for improved detection and processing of a large number of samples simultaneously. Blood samples will be sent to a centralized facility for analysis. Depending on the actual data obtained (i.e. sensitivity, specificity), the test will be proposed as a general "screening" or "diagnostic" test
2. a non-invasive test based on the collection of a larger amount of blood (10 ml), and based on the analysis of Circulating Tumor Cells (CTCs). These cells, which are circulating markers of the development of solid tumors, will be analyzed for the expression of the different BARD1 isoforms. These will provide not only for diagnostic tools of cancers, but also for predictive tools to anticipate the potential outcomes to treatments, as well as to help doctors following the course of treatment.

Project background 2: Lung and colorectal cancers
Despite the numerous widespread applications of the BARD1 protein, we have proposed to address in priority lung and colorectal cancers.
Lung and colorectal cancers are very attractive niche to develop a test based on the BARD1 protein. These cancers are amongst the biggest killers of all cancers, and early detection correlates with improved positive prognosis; there is currently no available easy-to-perform screening tool applicable for mass usage. For lung cancer, there exists also a population easily identifiable to apply the test: the heavy smokers over 50. Thus, and because of its easiness, there is the possibility for our test to become a mass-screening test to detect lung cancers in the target population years before it can actually be detected by conventional diagnostic tools (CT-Scan, X-ray). We hope that our test will gain insurance-coverage when applied to its target population (a general screening test for the general population would certainly not reach reimbursement, due to negative cost issues).

Project main objectives:
The goal of the project was to produce non-invasive tests based on BARD1 isoforms and antibodies, and integrate them into one platform for the early diagnosis of lung and colorectal cancers. The final diagnosis was planned to be a two-step process, using two tests sequentially, each ones with different aims and specificities:
1. Screening test: Antibody-based test. The test will be ELISA-based, is simple, requires small blood sampling and is relatively cheap. However, it was expected to be suffering from a comparatively low sensitivity/specificity. This test is ideal for mass-screening
2. Diagnostic: RNA-based test. This test will be RT-PCR based, is more sophisticated, labor-intensive and costly, and requires more blood sampling. However, its excellent sensitivities and specificities will allow for diagnosis
3. Integration: positive screening with the first test (antibodies) would lead to confirmation/infirmation by the second diagnostic test (RNA signatures)

Project Results:
Major achievements:

1. BARD1AG at the end of this project owns a proprietary test for the early detection of lung cancers. This test is based on a signature of 20 peptides spotted on Mesoscale Multiplex plates. This peptide signature has been protected by BARD1AG through the filing of a patent. To achieve this goal, a set of more than 50 peptides have been synthesized, and all spotting and binding conditions carefully optimized through Genorama’s work. The performance of the test has been evaluated using 178 lung cancer samples and 266 controls. The data set consists of 379 unique samples for which at least one set of variables (peptides or polypeptides) has been measured. In total, the data set contains measurements for 40 variables, which are divided into four subsets of 10 variables each. The test performance is excellent with an AUC (Area under the curve) under the ROC curve of 0.97. This may represent a true breakthrough in lung cancer diagnostic, and indicates the level of success achieved in this proposal.
From the raw data, an independent analysis by the University of Vienna using the “Lasso” method to statistically evaluate different peptide selections, we conclude that using 20 selected peptides, as proposed by BARD1AG, a predictive performance of >97% (0.97 under the ROC curve) for lung cancer is attainable. The actual number of 20 is ideally suited for the Mesoscale equipment, that allows to spot and accurately measure 10 peptides per well in a 96-wells plate. This is a highly conclusive result, which gives all indications of accuracy (>95%) sufficient for predictions to the level of diagnosis of lung cancer in asymptomatic individuals with a long history of heavy smoking.
2. Fewer colon cancer samples were included in our analysis due to lack of resources, and thus the predictive statistical power of the final signature (which is different form the one for lung cancers) is weaker. Signature could/and will be improved in the future by analyzing more samples. Note that for commercial and market opportunities reasons as directed by BARD1AG, the consortium focused on delivering an accurate signature for lung cancer as: 1) there is today no method for screening this type of cancer early, and 2) risk groups are clearly defined. This is not the case for colon cancer, for which screening has been since made available and risk groups are not defined. As indicated earlier, due to the positive first statistical results obtained, BARD1 and HUG will pursue the statistical validation of a colorectal peptide signature.
Altogether, we believe that a screening test for lung cancer based on the BARD1 peptide-signature is ready to be developed and further marketed. The test has been validated using several hundred samples, the certainly best possible peptide signature has been extracted from the BARD1 protein, the test has shown excellent sensitivity, and we have optimized all conditions and methodologies (see in particular Genorama reports). We thus feel confident that the early development phase has been achieved. BARD1AG is actually in advanced discussion with a large pharma partner for the manufacturing and commercialization of the test. According to these advanced discussions, a further validation of the test with a different cohort of patients might be required. However, as this will be a retrospective study and samples are available, this independent validation should only marginally delay the process.
While at the time of starting the project, a peptide-based test followed by a CTC-based signature to confirm seemed an excellent solution, now, as the BARD1 peptide signature-based test has sensitivity and a specificity higher than expected, the CTC-based test will not contribute significantly. Whereas it was in our opinion important to develop both types of tests, the peptide-based test appears today and following this work the test of choice for early detection. In other words, the sensitivity of the antibody-based test appears to outscore what we could achieved with an RT-PCR based test following CTC isolation
3. The progress of the BARDiag project allowed PluriSelect to develop and optimize its system for the fast and easy isolation of CTCs directly from whole blood. This has resulted in the setup of a Kit containing EpCAM or CK-19 specific pluriBeads and an optimized protocol. The target groups of this development are customers that intend to isolate CTC from whole blood for molecular analysis. The Kit is currently not suitable for cell analysis


Detailed S&T results: Milestones and achievements

Final Goal 1: antibody-based, “screening “test
As described above, a test requiring 20 peptides spotted on Mesoscale Multiplex plates has been designed and its performance evaluated using 178 lung cancer samples and 266 controls. The data set consists of 379 unique samples for which at least one set of variables (peptides or polypeptides) has been measured. In total, the data set contains measurements for 40 variables, which are divided into four subsets of 10 variables each. The test performance is excellent with an AUC (Area Under the Curve) under the ROC curve of 0.97. This may represent a true breakthrough in lung cancer diagnostic, and indicates the level of success achieved in this proposal.
We conclude that using 20 selected peptides, as proposed by BARD1AG, a predictive performance of >95% (0.97 under the ROC curve) for lung cancer is attainable. The actual number of 20 is ideally suited for the Mesoscale equipment, that allows to spot and accurately measure 10 peptides per well in a 96-wells plate. This is a highly conclusive result, which gives all indications of accuracy (>95%) sufficient for predictions to the level of diagnosis of lung cancer in asymptomatic individuals with a long history of heavy smoking.
Final milestones (Goal 1), with comments about completeness and potential deviations:
✓ Spotting of 30 different peptides using different conditions for optimization of binding and availability of epitope (Identification of optimal conditions). Milestone achieved as optimal conditions were defined. 50 peptides have been synthesized and tested
✓ Screening of 30 different peptides for antibody recognition in 100-200 lung cancer samples, including full-length recombinant BARD1 protein as positive control. Milestone fully achieved: a signature of 20 peptides has been extracted and validated on 178 lung cancer samples and 266 controls. 50 peptides and 4 recombinant BARD1 constructs have been synthesized and tested
✓ Screening of 30 different peptides for antibody recognition in 100-200 colorectal cancer samples, including full-length recombinant BARD1 protein as positive control. Milestone essentially achieved with 80 colorectal samples analyzed together with 266 controls. A signature of 20 peptides has been extracted, but its validation lacks statistical power due to the more limited number of cases analyzed.
Final Goal 2, nucleic-acid based, ”diagnostic” test:
The major challenge for this part of the project was to unambiguously demonstrate the feasibility of CTC isolation and the ability of the consortium to detect and analyze RNA-based isoforms of BARD1,-this at levels of less than 10 cancer cells per 10 ml of blood. These very stringent requirements represent what is generally accepted in the field, and means that the successful completion of this milestone will lead to one of the best tool for CTC isolation available today. As described in the report form Vienna University and pluriSelect, the consortium is indeed able to detect less than 10 “spiked” CTCs in 10 ml of blood, and to detect the expression of different BARD1 isoforms in 10 CTCs per 10 ml of blood sample. This means that both CTC isolation conditions and the RT-PCR conditions have been mastered, and the analysis of the RNA samples has been completed. However, the results are largely disappointing as none of the BARD1 markers were detected in clinical samples. Reasons for this failure are discussed in the reports from MUW and PluriSelect
Thus, as a general conclusion to the work performed, whereas the simple antibody-based test can unambiguously be used to as an early screening tool to detect lung and certainly colorectal cancers, the RNA-based test does not appear to be easily transferable as a useful diagnostic test.
Final milestones (Goal 2), with comments about completeness and potential deviations:
✓ Development of a method for CTC isolation (Ability to detect one CTC cell per ml sample). Milestone successfully completed as we are able to detect 10 CTCs spiked into 10 ml blood. Please see PluriSelect/MedUniWien reports
✓ Screening of different oligonucleotide primer sets for amplification of wt, lung and colorectal mutant BARD1 isoforms (Validated amplimers obtained using established lung-derived cell-lines/micro-dissected tumors. Best sets of primers identified for BARD1 lung-specific isoforms); Screening of 10 lung and 10 colorectal samples for CTC isolation coupled to RT-PCR amplification using best sets of primers (Validated amplimers obtained using CTCs isolated from lung samples). First part of the milestone achieved with 27 small cell lung cancer cell lines analyzed for BARD1 isoforms; blood samples from 14 colon carcinoma patients, 5 lung cancer patients, 2 pancreatic cancer patients and 10 healthy controls however failed to demonstrate the presence of isoform-specific BARD1 transcripts
✓ Note that other genes of interest were included as well in the analysis. Please see report from MUW.
✓ Discussion on CTC issues has been presented above

Potential Impact:
Potential impact (including the socio-economic impact and the wider societal implications of the project so far and the main dissemination activities and the exploitation of results
The development of a screening test for lung cancer has huge potential impact on the health of people affected by this cancer. Today, there is no method to detect this cancer early on asymptomatic individuals. Rather, CTC scans are prescribed when a clear suspicion of lung cancer has been established. Due to the potential damages incurred by the CTC scan, the scan cannot be repeated without danger. Thus a 100% safe and affordable screening test might be used repeatedly to detect early signs of cancers in fully asymptomatic individuals. By detecting early, the chances of a cure increase markedly. Thus, the test that we have designed will have a real socio-economic impact.
Dissemination activities consist of scientific publications in peer-reviewed journals. A teaser describing the results has alsom been prepared by BARD1AG and distributed to major pharmas. For exploitation of the results, BARD1AG is currently partnering with a pharma company to further validate, manufacture and market its test.

List of Websites:
web site address: http://cordis.europa.eu/search/index.cfm?fuseaction=proj.document&PJ_RCN=11657766

Contact:

Christophe BONNY, CEO
BARD1AG
RUE DE LA FONTAINE 7
1204 GENEVE
Switzerland

christophe@bard1ag.com