Final Report Summary - METACANCER (Identification and validation of new breast cancer biomarkers based on integrated metabolomics)
Executive summary:
Breast cancer is the most common cancer in women worldwide, and it is central to develop new technologies for the better understanding of the molecular changes that are involved in the progression of this type of cancer. Metabolomics - in analogy to the terms transcriptomics and proteomics - is defined as the study of all metabolites in a cell, tissue or organism for comprehensive understanding of a biological process. This is based on recently developed technologies that allow quantitative investigation of a multitude of different metabolites. Since small changes in enzyme concentrations or activities can lead to large changes in metabolite levels, the metabolome is regarded as the amplified output of a biological system.
In the METACANCER project, we have combined state-of-the-art technology for metabolic profiling to maximise the covering of the metabolome in human breast cancer. We have evaluated the hypothesis that alterations in the steady state concentrations of metabolites reflect amplified changes in metabolism, and that these can be used to classify breast cancer based on tumour biology, to identify new prognostic and predictive markers and to identify new targets for future therapeutic interventions.
The results of the METACANCER project are generally divided into three parts: validation of the methodological approach for classification of tumour types, combined analysis of metabolic, protein and gene expression data as well as pathway analysis, as well as development of new tools for integrated analysis of metabolic data.
The first major task of the METACANCER project was to introduce and evaluate metabolomics as a new method for cancer research. We have shown that the different metabolomics approaches Nuclear magnetic resonance (NMR) spectroscopy, Gas chromatography-mass spectrometry (GC-MS) and Liquid-chromatography-MS (LC-MS) are suitable for analysis of fresh-frozen human tumour tissue and provide complimentary information. An additional major task of the METACANCER project was the integration of metabolomics, proteomic and functional genomic data, which were generated using the same set of tumours. This resulted in a parallel dataset on different omic levels and enabled a systems biology evaluation of the pathology. The idea behind this approach is that it is more useful to combine information on different biological levels rather that to have the complete analysis solely based on metabolomics.
In addition the METACANCER consortium has developed new tools for the analysis of metabolomics data, such as MetaMapp, a tool to analyse metabolomics data in context of structural and functional relationship. MetaMapp visualises the relationships between identified metabolites but also unidentified expressed metabolite tags in a metabolic network. The tool METAtarget for linkage between enzyme and metabolite information been developed in the project.
The METACANCER project has shown that metabolomic analyses by GC-MS LC-MS and NMR spectroscopy are suitable for the analysis of tumour tissue and provide valid information that could be integrated with proteomic and transcriptomic data. This provides the possibility to study the changes in malignant tumours in a network that consists of combined Ribonucleic acid (RNA), protein and metabolite data. The European society will benefit from the knowledge gained through METACANCER resulting in new insights into the mechanisms of cancer progression as well as cancer patient management through improved molecular diagnostics leading to improvement of therapeutic concepts by selection of effective drugs.
Project context and objectives:
Breast cancer is the most common cancer in women worldwide with an incidence of more than 410 000 in the United States, Europe and Japan. In the Western world, approximately one in eight women will develop an invasive breast carcinoma in her lifetime. While the disease is curable in the early stages, about 50 % of patients have stage II or III tumours and are potential candidates for systemic therapy. This patient group would benefit from a patient-tailored therapy on the basis of biomarker testing. While genetic alterations have been extensively characterised in breast cancer, the changes in metabolism that occur downstream from genomic and proteomic alterations have not been analysed in detail before the start of the METACANCER project.
The metabolome reflects alterations in the pathophysiological state of biological systems. Metabolic alterations can be the consequences of changes in metabolic pathways, but also in signalling pathways, membrane turnover and other cellular networks. Since small changes in enzyme concentrations or activities can lead to large changes in metabolite levels, the metabolome is regarded as the amplified output of a biological system.
Metabolomics - in analogy to the terms transcriptomics and proteomics - is defined as the study of all metabolites in a cell, tissue or organism for comprehensive understanding of a biological process. This is based on recently developed technologies that allow quantitative investigation of a multitude of different metabolites. This emerging scientific discipline involves a multidisciplinary approach as well as the use of advanced analytical technologies such as NMR spectroscopy, GC-MS and LC-MS with elaborate deconvolution software to identify individual metabolites. Because of the diverse chemical nature of metabolites no single analytical tool can provide global coverage of the metabolome. Only by a combination of analytical approaches a maximal coverage of metabolism can be achieved.
In this project, we have combined state-of-the-art technology for metabolic profiling to maximise the covering of the metabolome. The METACANCER approach was the first-time application of these combined technologies to the large-scale analysis of tumour samples in the field of translational research in breast cancer.
It should be emphasised that metabolomics as a single method will not be able to explain the biological diversity and behaviour of tumour cells. Therefore we have combined the analysis of metabolites, proteins and gene expression in the METACANCER project. The methods used in METACANCER include DASL gene expression profiling, LC-MS / MS-based proteomics as well as in-situ proteomics by immunohistochemistry. These investigations were performed in the same tumours as the metabolic profiling, allowing the combination of different omic levels as part of a systems medicine approach to the disease.
The aim of this project was to characterise the metabolism of malignant tumours in order to identify new characteristic biomarkers and new targets for therapeutic interventions. The changes were further evaluated by mining of available databases including expression data at the transcriptional level as well as by additional investigations on protein and messenger RNA (mRNA) markers relevant for metabolic alterations.
The predictive biomarkers that were investigated in the METACANCER project were derived from different levels of the '-omic' network, starting with metabolites, but also including protein and RNA markers. Using a set of complementary technical approaches we have identified metabolic alterations in breast cancer tissue. This information can be used to determine the molecular tumour type in order to provide information on tumour progression, prognosis and treatment efficacy through these metabolic changes. The characterisation of tumours based on biological information can be used to guide patient tailored therapeutic approaches as a step towards personalised medicine.
We and others have shown that metabolomics can be used to investigate the changes in tumour tissue related to apoptosis, to hypoxia and energy metabolism. The most popular approaches for metabolomics involve either MS (GC-MS or LC-MS) or NMR spectroscopy. MS based approaches are typically more sensitive. NMR spectroscopy can be applied to intact tissue samples and even observe metabolites in vivo. Through the combination of different techniques as well as DASL gene expression analysis and LC-MS / MS based proteomics we have been able cover all metabolic pathways that have been described in the databases such as the Kyoto encyclopaedia of genes and genomes (KEGG) database, maximising the possibility of discovering metabolic biomarkers associated with disease.
METACANCER concept - hypothesis and approach
In the METACANCER project we have evaluated the hypothesis that alterations in the steady state concentrations of metabolites reflect amplified changes in metabolism, and that these can be used to classify breast cancer based on tumour biology, to identify new prognostic and predictive markers and to identify new targets for future therapeutic interventions. Based on previous investigations by members of the consortium and other groups, metabolic changes in energy metabolism, but also in other metabolic processes such as membrane turnover, apoptosis and lipid metabolism, are central for the malignant phenotype of breast cancer cells.
The overall objective of METACANCER was to identify and to validate metabolic profiles that are considered to reveal characteristic biomarkers relevant for progression of breast cancer and to translate this knowledge into clinical practice. Results are also the basis for the identification of new molecular targets involved in tumour progression to guide personalised therapies.
The METACANCER project has evaluated four hypotheses:
1. alterations of metabolites can be reliably measured in human tumour tissue and reflect amplified changes in metabolism;
2. these alterations can be used to classify breast cancer based on tumour biology;
3. to identify new prognostic predictive markers; and
4. to identify new targets for future therapeutic interventions.
To address these issues, the METACANCER research consortium combined expertise in the field of clinical and molecular tumour pathology and clinical breast cancer research with advanced technological approaches that cover the state-of-the-art of metabolomic research In addition, the consortium includes expertise on transcriptomic data mining as well as the investigation of protein and mRNA markers including new high-through-put bead based analysis techniques.
The partners have worked together to provide the research infrastructure, technological basis and biobank for this project. To be successful in such an ambitious task, the establishment of a European network for collaboration was essential and is a prerequisite to contribute to the development of new predictive tests and new anticancer strategies. The project was aimed at the identification of biomarkers on the metabolite level that are able to predict outcome as well as therapy response in patients with breast cancer. These metabolite biomarkers were linked to proteomic and transcriptomic data to identify molecular changes in tumours that may be used to develop new therapeutic strategies. Therefore, the consortium is focused on metabolomics, but metabolic data is integrated in a research network that uses state-of-the art-information from genomic and proteomic analysis as well as data mining to describe multidimensional cellular networks. The consortium has its main focus on tumour tissue, but some analyses were also performed using serum samples to investigate possible metabolic biomarkers in biofluids for the prediction of chemotherapy response.
Project results:
The results of the METACANCER project are generally divided into three parts:
1. Validation of the methodological approach for classification of tumour types.
2. Combined analysis of metabolic, protein and gene expression data as well as pathway analysis.
3. Development of new tools for integrated analysis of metabolic data.
Part 1: Metabolomics as a new method for analysis of tumour tissue
The first major task of the METACANCER project was to introduce and evaluate metabolomics as a new method for cancer research. We have shown that the different metabolomics approaches NMR spectroscopy, GC-MS and LC-MS are suitable for frozen tumour tissue.
With each method it has been possible to evaluate several hundreds of metabolites, many of which could be identified and linked to biochemical pathways. Therefore it was possible to evaluate changes in metabolic pathways in different types of tissue. As expected, the largest differences were observed between normal breast tissue and malignant breast tumours. With all three metabolomics approaches it is possible to separate normal and malignant breast tissue in unsupervised and supervised analyses with a high sensitivity and specificity. In addition, we observed metabolic differences between different types of breast cancer. As known from other studies, the major groups of breast cancer are the hormone receptor positive and negative tumours. These tumour types have a different biological background, different clinical characteristics and are treated by different therapeutic strategies. In the metabolomics analysis, a set of metabolites was detected that was able to discriminate between hormone receptor positive and negative tumours, which had different metabolic profiles in our analysis. Other types of breast cancer are defined by the histological grade, which measures the dedifferentiation of the tumour and is one marker of tumour progression. By metabolomics, it was possible to detect differences between grade 1 or 2 and grade 3 tumours. Therefore we conclude that it is possible to generate meaningful biological information from tumour tissue by metabolic profiling, we have used this data to analyse the alterations of metabolic pathways in different types of tumour tissue. The results of the different metabolomics approaches for classification of tumour types and normal tissue are shown below.
Metabolite signatures for separation of breast cancer and normal tissues
In order to develop a metabolite based molecular approach for the separation of cancer and normal tissue we analysed each metabolite in the GC-MS data for its classifying power. A set of metabolites separated between tumour and normal breast tissues with sensitivity and specificity above 80 %. We found both, tumour markers and normal tissue markers. In order to enhance the separation, each of the tumours markers was divided by each of the normal tissues markers. Working with ratios instead of intensities has the advantage that these quantities are independent of data normalisation and robust against rescaling. For each classifier, we defined a cut-off value for the classification in cancer and normal tissues by maximising the sum of sensitivity and specificity.
There was a strong correlation between the votes of the different classifiers: For 217 tissues (58.8 %) there was concordant agreement among all classifiers, for 319 tissues (86.4 %) more than 90 % of the classifiers agreed on malignancy status.
Only 8 tumours and 6 normal tissues were classified wrong, leading to a sensitivity of 97 % and a specificity of 93.9 % of the molecular test.
Separation of healthy and normal tissue by 1H HR MAS NMR
Using multivariate statistical techniques on the entire data set the analysis by 1H HR MAS NMR was able to separate adipose tissue (classed as 100-50 % adipose) from healthy breast tissue (classed as 80 % or more breast tissue) and tumour samples. The separation was mostly driven by high concentrations of polyunsaturated and saturated fatty acid resonances in the adipose tissue and increased glucose resonances in the healthy and tumour samples compared to the adipose tissue. In addition normal healthy beast tissue was differentiated from tumour samples (ductal or lobular carcinoma) using stratified tumour samples with less than 10 % adipose tissue. Biochemical studies of breast cancer cells have suggested PCho as a biomarker of cell proliferation and cancer malignancy. PCho is a precursor, as well as a breakdown product of the major membrane component phosphatidylcholine, whereas GPC is solely a membrane breakdown product.
NMR-based metabolites separating ductal from lobular carcinoma, different tumour grades, different molecular subtypes
Invasive ductal carcinoma could also be separated from lobular carcinoma using a Partial least squares - Discriminant analysis (PLS-DA) model for the NMR data. Different tumour graded samples were also separated using multivariate data analysis. Grade 1 samples were readily separated from grade 2 and grade 3 tumours using Principal component analysis (PCA), PLS-DA and Orthogonal projections to latent structure (OPLS-DA) statistical models. In addition we could also separate grade 2 from grade 3 tissue.
The separation of Estrogen receptor (ER) and Progesterone receptor (PR) positive tissue by NMR was poor and only marginally better than the random predicted model from the negative samples. Similarly, we were not able to separate the Her2 classified samples. However, when using the combined classifications for Her2, ER and PR status, subtype 4 (Her2 + ER or PR pos) were separated from subtype 2 (Her neg, ER or PR neg) relatively robustly. We systematically tested multi-Block PCA and PLS, O2PLS-DA and correlation data fusion approaches on a combined dataset of 1H HR-MAS NMR, GC-MS and LC-MS measured on a cohort of 300 breast cancer patients with regard to various data pre-processing schemes and compared the outcomes. On the level of individual platforms it is apparent that the differentiation between healthy and diseased samples is often hampered by the heterogeneity of the tissue sample, specifically due to the fat content, with the individual platforms being affected to a different extent. Multi-block PCA differentiates the healthy and diseased subjects readily.
A metabolite index measured by GC-MS distinguishes between ER+ and ER- tumours
For construction of an ER status classifier from the GC-MS data, a feature selection step was combined with nearest centroid classification. First, metabolites were ranked by the capability to distinguish between training ER+ and ER- tumours based on Welch's t-test. Then, the centroids of the training ER+ and ER- tumours were calculated in the space of the top metabolites. Finally, test samples were predicted to belong to the class with the nearest centroid. A Metabolic index (MI) for prediction of ER status was constructed as linear combination of metabolites. To this end, the centroids of training ER+ and ER- tumours were determined in the metabolite space. The MI was defined as projection onto the difference vector between the centroids of ER+ and ER- tumours.
Using the training cohort, a MI was constructed as linear combination of the top metabolites. Receiver operating characteristic (ROC) analysis showed an excellent performance of the MI for prediction of the ER status in the training cohort. Validation of the MI in the validation cohort affirmed the excellent performance of the predictor.
Taken together, the results of METACANCER show that metabolites can be reliably measured in tumour tissue and that tumour metabolomics is able to provide useful information on biomarkers and biological pathways in different types of tissue and that this new method is a promising addition to transcriptomics, proteomics and sequencing.
Part 2 - Integration of metabolomics, transcriptomics and proteomics - analysis of tumour samples by combined multilevel -omics approaches
An additional major task of the METACANCER project was the integration of metabolomics, proteomic and functional genomic data which were generated using the same set of tumours. This resulted in parallel dataset on different omic levels and enabled a systems biology evaluation. The idea behind this approach was that it would be more useful to combine information on different biological levels rather that to have the complete analysis solely based on metabolomics.
We have used a data mining strategy using a system for evaluation of complex already published gene expression data sets to look for enzymes that might be interesting in breast cancer. The results of this approach have been linked to the metabolite alterations. From the combination of those approaches, we have focused on enzymes that might be relevant for the progression of breast cancer and the observed metabolic changes. We have analysed the relevant enzymes in breast cancer cohorts and functional models and linked the metabolic alterations to protein changes as well to changes in gene expression.
In addition to the data mining studies, we have also performed DASL gene expression analysis as well as LC-MS / MS-based proteomics. This allows us to study each tumour with different omic approaches which could then be combined to focus on the relevant pathway.
Within METACANCER, a large cohort of fresh-frozen breast tissue samples were analysed by metabolomics. The metabolomics analysis of the samples has been performed using three different methods, GC-MS, LC-MS and NMR. Additionally, genome-wide expression data of 150 out of these samples were generated using the DASL platform. Parallel to fresh-frozen tissues, Formalin-fixed paraffin-embedded (FFPE) tissues of patients in the METACANCER cohort were collected. Tissue microarrays (TMA)s were constructed and stained with a set of selected antibodies. Finally, proteomics analyses of 126 FFPE tissue sections were performed using LC-MS / MS.
All methods have been able to generate valid metabolomic, functional genomic and proteomic data from the samples. In the statistical analysis, it has been shown that it is possible to link the biological alterations of the samples with clinical pathological data. In particular, it has been possible to describe differences in metabolites between different subtypes of breast cancer. Based on METACANCER results it is now clear that metabolic profiling is a valid approach to analyse frozen tumour biopsies and that metabolomics data can be linked to other omic levels.
Therefore, at the end, metabolic profiling can be integrated in a systems' biology approach with other biomarkers derived from the other levels of biological organisation. By this strategy, we are able to go beyond the metabolite level and to identify and validate selected protein and mRNA biomarkers relevant for metabolic alterations. This results in a combined pathway dataset consisting of metabolites as well as key protein and mRNA markers as a basis for a validated diagnostic system to assess prognosis and to guide targeted therapies in breast cancer.
Global lipidomics for tumour tissue samples - changes in phospholipid synthesis are involved in the progression of breast cancer
Analysing the LC-MS lipidomics data, significant differences were observed in the tumour versus normal comparisons. In tumours, it was observed that tumour grade and ER status affect the lipid profiles most radically. The most significantly changing phospholipid was PC(14:0 / 16:0) which was different dependent on both ER status and grade. It is known that the majority of the ER tumours are of grade 3, and this was also the case in our patient population since only 7 % of the grade 1,2 tumours were ER, while 44 % of the grade 3 tumours were ER. Thus, either grade or ER status alone could explain our results. We therefore analysed the ER status only within grade 3 tumours and the grade only within ER+ tumours, and confirmed that both ER status and grade independently affected the same lipids, with the highest levels found in ER- grade 3 tumours.
Increased levels of phospholipids were observed during tumour progression, with the highest levels in hormone receptor negative grade 3 tumours. Using the GeneSapiens database for data mining, we have identified key enzymes of lipid biosynthesis. These enzymes were stained by immunohistochemistry in tumour tissue and analysed together with the LC-MS data of the same tumour cohort. We found that in particular FASN and ACACA, which are two major enzymes involved in phospholipid synthesis were increased on the protein level in those tumours that had high levels of phospholipids. This shows that it is possible to connect protein expression data by immunohistochemistry with LC-MS lipidomic profiles of the identical tumours.
To evaluate the function of these lipid metabolising enzymes for breast cancer proliferation and apoptosis, we investigated breast cancer cell lines and inhibited the central enzymes by small interfering RNA (siRNA). This resulted in decreased cell viability and in lower phospholipid level measured by LC-MS. The increased de-novo lipid synthesis is typically found in tumour cells, but not in normal cells, which would make it a promising target for new tumour-specific therapeutic approaches. The detailed results of this project part have already been published, for additional information see Hilvo M, Denkert C, Lehtinen L, Müller B, Brockmöller S, Seppänen-Laakso T, Budzsies J, Bucher E, Yetukuri L, Castillo S, Berg E, Nygren H, Sysi-Aho M, Griffin JL, Fiehn O, Loibl S, Richter-Ehrenstein C, Radke C, Hyötyläinen T, Kallioniemi O, Iljin K & Oresic M. Novel theranostic opportunities offered by characterisation of altered membrane lipid metabolism in breast cancer progression. Cancer Res. (2011) 71: pp. 3236-3245.
Cross-correlation of metabolite and enzyme expression data
Interesting insights in the regulation of metabolic pathways in breast cancer can be obtained from a combined analysis of metabolite data and enzyme expression data. We have correlated GC-MS metabolomics data with expression data of the same METACANCER tumours that were generated using whole-genome DASL. In order to investigate the consequence of changes in the expression of glucose transporters in breast cancer, we correlated the family of glucose transporters with the concentrations of sugars in the cancer cells.
Combined markers GPAM analysis and metabolic profile
We investigated glycerol-3-phosphate acyltransferase (GPAM), a key enzyme in the lipid biosynthesis of triacylglycerols and phospholipids. Expression of GPAM in malignancies is of special interest because phospholipids are an important and major component of all cell membranes. To investigate the protein expression of GPAM 228 breast cancer samples were immunohistochemical stained, digitised and evaluated. The conclusion from the results was that in breast cancer GPAM affects more the level of phospholipids than triglycerides, and most of the changes are independent from those associated with ER status and tumour grade. This implies that at least in the context of breast cancer the function of GPAM is directed more towards phospholipid production rather than triglyceride synthesis. This is reasonable as it has been shown previously that increased de novo fatty acid synthesis is a hallmark of cancer cells and the products of this lipogenic pathway are directed mainly to the cell membrane phospholipids.
To extend the number of metabolic pathways covered by the analysis, an additional GC-MS based metabolic analysis was performed. Metabolomics analyses of breast cancer tissue led to the identification of 467 metabolites. Out of these, 161 metabolites had a known chemical structure and could by mapped to metabolite names. The analysis from the GPAM expression with the data showed 57 significantly changed metabolites (p < 0.05). Out of these 57 metabolites, 19 could be identified. The manuscript with the complete data is currently under submission to a scientific journal: Scarlet F. Brockmöller, Elmar Bucher, Berit M. Müller, Jan Budczies, Mika Hilvo, Julian L Griffin, Matej Oresic, Olli Kallioniemi, Kristiina Iljin, Sibylle Loibl, Silvia Darb-Esfahani, Bruno V. Sinn, Frederick Klauschen, Judith Prinsler, Nikola Bangemann, Fakher Ismaeel, Oliver Fiehn, Manfred Dietel, Carsten Denkert. Integration of metabolomics and expression of GPAM in breast cancer - link to patient survival, hormone receptor status and metabolic profiling.
Part 3 - New tools for analysis of metabolomics data
In the METACANCER consortium, new tools for the analysis of metabolomics data were developed:
UCD (Davis, California) has developed MetaMapp, a tool to analyse metabolomics data in context of structural and functional relationship. MetaMapp visualises the relationships between identified metabolites but also unidentified expressed metabolite tags in a metabolic network.
CHARITE (Berlin, Germany) has developed a metabolite based signature that distinguished between cancer and normal tissues at high sensitivity and specificity. Among the important clinicopathological factors of breast cancer, the status of ER had the strongest influence on the metabolite profile of a tumour. Meanwhile, a metabolite test has been developed that distinguished between ER+ and ER- breast cancer (for details see above, part 1). Cross-correlation between metabolite and enzyme expression have been studied using the whole-genome DASL data. GPAM, a key enzyme in the biosynthesis of triacylglycerols and phospholipids, was identified as a prognostic marker in breast cancer. The tool METAtarget for linkage between enzyme end metabolite information has already been published, for details see: Budczies J, Denkert C, Müller BM, Brockmüller S, Dietel M, Griffin JL, Oresic M & Fiehn O. METAtarget - extracting key enzymes of metabolic regulation from high-througput metabolomics data using KEGG reaction information. GI-Edition: Lecture Notes in Informatics (2010) P-173: pp. 103-112.
VTT (Helsinki, Finland) has analysed the lipid metabolism in breast cancer by a combined analysis of lipidomics data and enzymes involved in the regulation of fatty acid metabolism. Alterations in the membrane lipids were detected that correlated with cancer progression, prognosis, hormone receptor status and tumour grade. Gene silencing experiments indicated that silencing of multiple lipid metabolism regulating genes reduced the lipidomic profiles and viability of breast cancer cells.
UCAM (Cambridge, UK) has carried out 1H HR MAS NMR and showed that cancer and normal tissues could be separated using this platform. Further, invasive ductal cancer could be separated from lobular cancer and tumours of grade G1, G2 and G3 from another grade. In contrast, separation of ER+ from ER- as well as HER+ from HER- tumours was poor. Data from GC-MS, LC-MS and NMR were combined and analysed with multi-block PCA, with this technique not only identifying the metabolite changes previously identified in the analysis of the individual datasets but also produced metabolic correlates across the platforms that improved our mechanistic understanding of the data.
Identification of new metabolites by accurate MS
A total number of 391 samples were successfully investigated by GC followed by Time-of-flight (TOF) MS. Prior to GC-MS, the tissue samples were split in a training (2 / 3) and a validation cohort (1 / 3). Both cohorts had a comparable distribution of the most important clinicopathological parameters. Both sets were profiled at UCD (Davis, California), the training samples at the end of 2008, the validation samples in the beginning of 2009. For the METACANCER project, a new analysis mode of BinBase was implemented. In the first step of the new analysis mode, abundant metabolites are identified using the training samples. In the second step, these metabolites were measured in the validation samples. In the first data calculation performed in early 2009, this approach led to the initial detection of 136 identified metabolites, known by known chemical structures and metabolite names. After further identification of unknowns by accurate MS, depicted below for UDP-N-acetylglucosamine, further 64 novel unique metabolites were identified, plus 28 that were annotated by elemental formulas and best-hit structures. The results of this project part have already been published: Kumari S, Stevens D, Kind T, Denkert C, Fiehn O. Applying in-silico retention index and mass spectra matching for identification of unknown metabolites inaccurate mass GC-TOF MS. Anal Chem. 2011 Jun 16.
MetaMapp to highlight biochemical changes from GC-TOF profiling
Due to the large diversity of molecules in metabolism, no single instrumentation or technique is capable of capturing all metabolites simultaneously. As tumour progression depends on the operation of major metabolic pathways we have used a platform that yields the widest range of primary metabolites to date, CG coupled to TOF MS. We linked data acquisition to automatic metabolite annotation by BinBase, an in-house database of over 7 000 metabolite tags supported by over 1 200 authentic chemical standards with molecular weights below 600 Da.
Mapping metabolites to biochemical pathways appears to be the logical method of choice, supported by a range of tools and pathway databases supporting such as HumanCyc, EHMN and KEGG. However, our metabolomic analyses revealed that 19 of all the 185 structurally identified metabolites could not be retrieved from these repositories, especially for lipids, sugars and sugar conjugates. This result indicated that mammalian metabolism is more diverse than represented by current enzyme-specific databases.
In order to capture all identified metabolites plus additional 372 currently unidentified expressed metabolite tags, we have devised a novel tool to visualise structural and functional relationships: MetaMapp. MetaMapp combines the functional information of biochemical substrate/reaction pairs (obtained from the KEGG Rpair DB) with structural information obtained from chemical similarities between all identified compounds and mass spectral similarities for unidentified metabolites. The underlying axiom is that biochemistry refers to the conversion of chemically similar compounds by catalytic enzymes. Hence, it appears logical to associate all compounds directly by their chemical similarity. Clusters of chemically similar compounds should then resemble biochemical modules. We have used the public substructure decomposition tool in PubChem to calculate a chemical similarity matrix for all identified metabolites by the established Tanimoto coefficient.
MetaMapp provides several advantages. First, it is a platform independent approach meaning it does not matter what technology was utilised to detect or identify a metabolite. Cross-platform metabolomics datasets can be readily integrated and visualise to infer biological conclusions. The only requirement is that the chemical structures should be associated with machine encoded chemical structures. Second, MetaMapp is not constrained by genomics information. All the metabolite detected in a cross species manner can be straightforwardly mapped, enabling mapping of the metabolite that are originated from diet or gut microbes. Third, the layout is not static and it can be updated according to the input list of compounds. For clarity, metabolites of unknown structure have been excluded. The increase in metabolites was accompanied by increases in enzyme concentrations using the LC-MS / MS based proteomic results. A direct mapping of both enzymes and metabolites is feasible through KEGG Rpair and enzyme maps.
Potential impact:
Potential impact including socio-economic impact and wider societal implications
Global and European relevance
Breast cancer is worldwide the most common cancer among women. The chance of developing invasive breast cancer at some time in a woman's life is about 1 in 8 (13 % of women). According to research by the 'Breast health global' Initiative, breast cancer is the most prevalent cancer in the world today due to its high incidence and relatively good prognosis. An estimated 4.4 million women are alive today in whom breast cancer was diagnosed within the last five years. However, breast cancer is the most common cause of cancer-related deaths among women worldwide. More than 1.1 million women worldwide are newly diagnosed with breast cancer annually. This represents about 10 % of all new cancer cases and 23 % of all female cancers. With more than 410 000 deaths each year, breast cancer accounts for about 14 % of all female cancer deaths and 1.6 % of all female deaths worldwide. Incidence rates are climbing by as much as 5 % annually in low-resource countries. Women living in North America have the highest rate of breast cancer in the world. Death rates from breast cancer continue to decline, with larger decreases in women younger than 50. These decreases are believed to be the result of earlier detection through screening and increased awareness, as well as improved treatment.
The METACANCER project is focused on identification of biological markers on the metabolite level that aid in assessing the prognosis and the treatment response of patients with breast cancer. These validated biomarkers are a basis for improved strategies in translational and clinical research that will improve patient care on the one hand, as these biomarkers will be useful for prognosis, monitoring and treatment selection of patients with breast cancer. In addition, the metabolic profiling is aimed at the identification of metabolite and alterations in metabolic pathways, as well. This leads to a better understanding of the basic mechanisms underlying cancer which will lead - after translation to the clinical setting - to improved treatment strategies. As an example, we have shown that phospholipids in tumour tissue are synthesised de-novo and that this process is increased in tumour progression. This suggests that therapeutic approaches targeting lipid biosynthesis for cellular membranes might be a promising approach in breast cancer.
Impact for cancer research and clinical management of breast cancer patients
Treatment options for breast cancer are initially identified according to the stage of the disease. There are several modes of treatment including surgery, radiotherapy, endocrine therapy and cytotoxic chemotherapy. However, the management of an individual patient will depend on many factors including menopausal status, hormone receptor status and treatment preferences. The final treatment regimen selected, following discussion with the patient, will depend on the individual circumstances making it difficult to manage the disease by following a systematic treatment algorithm. To allow for a better stratification of patients, it is important to identify those pathways that are relevant for tumour progression and therapy response, and to determine biomarkers that could be used to monitor the activity of those pathways.
The METACANCER project has shown that metabolomic analyses by GC-MS, LC-MS and NMR spectroscopy are suitable for the analysis of tumour tissue and provide valuable information that could be integrated with proteomic and transcriptomic data. This gives the possibility to study the changes in malignant tumours by constructing a correlation network that consists of combined RNA, protein and metabolite data. The European society will benefit from the knowledge gained through METACANCER resulting in new insights into the mechanisms of cancer progression as well as cancer patient management through improved molecular diagnostics leading to improvement of therapeutic concepts by selection of effective drugs as part of systems medicine.
Breast cancer diagnosis, prognosis and treatment are of global relevance. Accordingly new biomarkers for breast cancer diagnosis or new therapeutic approaches are of interest for companies in the diagnostic and pharmaceutical field. For example, part of the METACANCER evaluations are focused on the analysis of PARP positive and negative tumours. PARP is a new promising target in the clinical setting, however, results of recent clinical trials have shown that the PARP activity might be restricted to tumour subsets, which need to be identified by new molecular markers.
Establishing European research networks
The METACANCER consortium has been strategically built to contain the most important European groups and one major United States group (UCD) with a high complementarity to ensure the strengthening of the European research area in the field of cancer metabolomics. Due to the advanced technology base, it is envisaged that advances in understanding of metabolic signatures as well as factors contributing to clinical response seen in cancer therapy will be made, leading to the development of new strategies to improve patient care providing new rationales for treating cancer patients. The proposed project contributes substantially to the competitiveness of European science and industry in the area of development of diagnostic tools, identification of biomarkers and cancer drug research. This offers an unprecedented opportunity to provide the European Community with a framework of innovate science that concurs with the major socio-economic and medical objectives to establish improved cancer therapies, and will be competitive on a global scale.
The development of protectable therapeutic, diagnostic, prognostic and drugable candidates by identifying new response-associated markers is potentially a further exploitable outcome as well as providing novel treatment approaches based on an improved understanding of metabolic changes in tumours that could put Europe ahead of the field.
The formation of unique data and material repositories will also benefit future expansion of cancer-related research in the European Community research providing a knowledge and resource base for further work.
Solving social problems - societal contribution
Cancer is a central health problem in Europe, therefore research that leads to a better understanding of the biological changes in cancer progression will have a wide-ranging and significant social impact, including quality of life. Research activities will result in a reduction of therapy-related suffering of cancer patients for the most common cancer in women world-wide and will reduce the amount of ineffective therapies, the side effects and thus the costs for the health care system. Finally it is aimed towards a decrease of cancer mortality in Europe.
Socially, many people are employed in the European biotech/pharmaceutical industry and their job security and career success depend increasingly on competitiveness in world markets. Thus, development must be streamlined and must become more efficient to ensure the evolution of better therapeutic strategies and products with fewer unwanted side effects. Additionally, the inclusion of research teams from different part of Europe allows for a wider participation in the project and greater dissemination of results for the benefit of a wider population. This also provides a basis for the exchange of technologies and the enlargement of a research group whose extensive experience in metabolic analysis will be significantly enhanced through the introduction of evolving research skills. The benefits will channel to society at large and develop a model for research activity in this field for future advances in metabolomics and translational cancer research.
The METACANCER project is focused on the identification of biological markers at the metabolite level that aid in assessing the prognosis and the treatment response of patients with breast cancer. These validated biomarkers will lead to a better understanding of the basic mechanisms underlying cancer which will lead - after translation to the clinical setting - to improved treatment strategies.
The project has a high public health impact, as - in the long run - the investment in this area will contribute to reaching the aims of reducing cancer's incidence and mortality and improving quality of life for patients.
Scientific impact
The METACANCER project is the first large-scale effort to combine the three major metabolomic technology platforms for combating a major human disease. We have combined efforts to improve the likelihood that genuine metabolic biomarkers for breast cancer tissues can be detected and validated, that would eventually lead to diagnostic toolkits which would facilitate a much more precise prediction and prognostic toolkit. Due to the high biological variability in human tissues, all publications so far have shown that mathematical combinations of variables ('multivariate statistics') will be needed to describe vectors that would facilitate such high quality diagnostics. In addition, this study was the first attempt to link high quality metabolomic data to proteomics and transcriptomics data in order to enable better insights into the cellular mechanisms that define the onset and progression of breast carcinoma. This provides further functional validation of the biomarkers and will thus add confidence into the statistical models built for the clinical applications. In summary, this is a first step into the exciting new area of individualised biomedicine that links personalised metabolic phenotypes into prediction cohorts and then renders therapeutic suggestions based on correlation with cellular mechanisms.
Main dissemination and exploitation activities
The consortium has an interdisciplinary combination of university research groups, Small and medium-sized enterprise (SME)s and clinical partners thus offering a wide range of dissemination and exploitation possibilities. The METACANCER results have been disseminated the results in different ways:
- publication of scientific results in peer reviewed journals and contribution to international conferences (scientific partners) - the publication process is still ongoing;
- transfer of knowledge into and beyond the consortium;
- presentation of the scientific and technical results to industry;
- to create awareness for novel technologies towards the public.
Internal dissemination of research results was realised via a restricted area of a public website. This platform, which can be accessed via the internet, served as a common knowledge sharing tool. Developments during the project were continuously monitored and evaluated via the progress surveys (deliverables in WP9 - Training and dissemination), reports on project's progress, data and manuscripts, information material etc. and are available on this website. There were a total of six semi-annual plenary progress meeting. The results of this project were regularly pre-screened for potential protection of the Intellectual property (IP) before they could enter the public domain. The evaluation of the market and existing patents / patent applications in the field was be the prime considerations in this initial evaluation.
Information exchange with the scientific community
Results were disseminated by presentations during national and international conferences, by publications in international peer-reviewed bio-medical journals. Several publications are currently under preparation. Project's interdisciplinary workshops have been used to discuss results between the various experts of the consortium and invited external guest speakers.
Public relation:
In order to provide general information on advances in breast cancer diagnosis and treatment field, METACANCER has published the project's background, novel scientific approach as well as the related impact for patients, clinicians and science on a public website, via a project flyer and press releases.
The METACANCER website www.METACANCER-fp7.eu was set up during the first project period by tp21 and maintained and updated until project end. The website will be kept open for several years after the project end. A list of major METACANCER publications on this site will demonstrate the successful implementation of the project. The METACANCER booklet was elaborated and implemented by tp21. The booklet was disseminated at conferences, meetings, at partners' organisations and submitted to interested parties on request. Target groups for the booklet are the general public and young scientists. The aim of the booklet's design was to demonstrate the cross disciplinary and the pan European novel approach in cancer metabolomics. The booklet is promoted on the METACANCER public website and is available on request.
Teaching and visibility: An important deliverable of this project was the further education of younger scientists (PhD and post-doctoral researchers). The partners represent a diverse range of metabolomic and bioinformatic tools and thus, those involved in the project will be exposed to state of the art approaches in metabolomics. This helps produce scientists with vital skills in analytical and mathematical analyses, improving the skills base of Europe. Interdisciplinary workshops and an exploitation workshop (deliverable in WP T&D) will actively support young scientists' education (all partners).
Strategies for the exploitation - management of intellectual property
In order to reflect the high importance of the issue, an experienced dissemination and exploitation manager has been integrated in the consortium i.e. TP21. Furthermore all partners in the project have devoted substantial effort to this task over the course of the project.
Training and awareness creation amongst the project staff is essential to identify results that could be target for protection and on the other side already protected know-how that could interfere with the own approaches. Six monthly progress surveys of the management have been performed to reveal any results that could be relevant for application.
Project website: http://www.METACANCER-fp7.eu
Breast cancer is the most common cancer in women worldwide, and it is central to develop new technologies for the better understanding of the molecular changes that are involved in the progression of this type of cancer. Metabolomics - in analogy to the terms transcriptomics and proteomics - is defined as the study of all metabolites in a cell, tissue or organism for comprehensive understanding of a biological process. This is based on recently developed technologies that allow quantitative investigation of a multitude of different metabolites. Since small changes in enzyme concentrations or activities can lead to large changes in metabolite levels, the metabolome is regarded as the amplified output of a biological system.
In the METACANCER project, we have combined state-of-the-art technology for metabolic profiling to maximise the covering of the metabolome in human breast cancer. We have evaluated the hypothesis that alterations in the steady state concentrations of metabolites reflect amplified changes in metabolism, and that these can be used to classify breast cancer based on tumour biology, to identify new prognostic and predictive markers and to identify new targets for future therapeutic interventions.
The results of the METACANCER project are generally divided into three parts: validation of the methodological approach for classification of tumour types, combined analysis of metabolic, protein and gene expression data as well as pathway analysis, as well as development of new tools for integrated analysis of metabolic data.
The first major task of the METACANCER project was to introduce and evaluate metabolomics as a new method for cancer research. We have shown that the different metabolomics approaches Nuclear magnetic resonance (NMR) spectroscopy, Gas chromatography-mass spectrometry (GC-MS) and Liquid-chromatography-MS (LC-MS) are suitable for analysis of fresh-frozen human tumour tissue and provide complimentary information. An additional major task of the METACANCER project was the integration of metabolomics, proteomic and functional genomic data, which were generated using the same set of tumours. This resulted in a parallel dataset on different omic levels and enabled a systems biology evaluation of the pathology. The idea behind this approach is that it is more useful to combine information on different biological levels rather that to have the complete analysis solely based on metabolomics.
In addition the METACANCER consortium has developed new tools for the analysis of metabolomics data, such as MetaMapp, a tool to analyse metabolomics data in context of structural and functional relationship. MetaMapp visualises the relationships between identified metabolites but also unidentified expressed metabolite tags in a metabolic network. The tool METAtarget for linkage between enzyme and metabolite information been developed in the project.
The METACANCER project has shown that metabolomic analyses by GC-MS LC-MS and NMR spectroscopy are suitable for the analysis of tumour tissue and provide valid information that could be integrated with proteomic and transcriptomic data. This provides the possibility to study the changes in malignant tumours in a network that consists of combined Ribonucleic acid (RNA), protein and metabolite data. The European society will benefit from the knowledge gained through METACANCER resulting in new insights into the mechanisms of cancer progression as well as cancer patient management through improved molecular diagnostics leading to improvement of therapeutic concepts by selection of effective drugs.
Project context and objectives:
Breast cancer is the most common cancer in women worldwide with an incidence of more than 410 000 in the United States, Europe and Japan. In the Western world, approximately one in eight women will develop an invasive breast carcinoma in her lifetime. While the disease is curable in the early stages, about 50 % of patients have stage II or III tumours and are potential candidates for systemic therapy. This patient group would benefit from a patient-tailored therapy on the basis of biomarker testing. While genetic alterations have been extensively characterised in breast cancer, the changes in metabolism that occur downstream from genomic and proteomic alterations have not been analysed in detail before the start of the METACANCER project.
The metabolome reflects alterations in the pathophysiological state of biological systems. Metabolic alterations can be the consequences of changes in metabolic pathways, but also in signalling pathways, membrane turnover and other cellular networks. Since small changes in enzyme concentrations or activities can lead to large changes in metabolite levels, the metabolome is regarded as the amplified output of a biological system.
Metabolomics - in analogy to the terms transcriptomics and proteomics - is defined as the study of all metabolites in a cell, tissue or organism for comprehensive understanding of a biological process. This is based on recently developed technologies that allow quantitative investigation of a multitude of different metabolites. This emerging scientific discipline involves a multidisciplinary approach as well as the use of advanced analytical technologies such as NMR spectroscopy, GC-MS and LC-MS with elaborate deconvolution software to identify individual metabolites. Because of the diverse chemical nature of metabolites no single analytical tool can provide global coverage of the metabolome. Only by a combination of analytical approaches a maximal coverage of metabolism can be achieved.
In this project, we have combined state-of-the-art technology for metabolic profiling to maximise the covering of the metabolome. The METACANCER approach was the first-time application of these combined technologies to the large-scale analysis of tumour samples in the field of translational research in breast cancer.
It should be emphasised that metabolomics as a single method will not be able to explain the biological diversity and behaviour of tumour cells. Therefore we have combined the analysis of metabolites, proteins and gene expression in the METACANCER project. The methods used in METACANCER include DASL gene expression profiling, LC-MS / MS-based proteomics as well as in-situ proteomics by immunohistochemistry. These investigations were performed in the same tumours as the metabolic profiling, allowing the combination of different omic levels as part of a systems medicine approach to the disease.
The aim of this project was to characterise the metabolism of malignant tumours in order to identify new characteristic biomarkers and new targets for therapeutic interventions. The changes were further evaluated by mining of available databases including expression data at the transcriptional level as well as by additional investigations on protein and messenger RNA (mRNA) markers relevant for metabolic alterations.
The predictive biomarkers that were investigated in the METACANCER project were derived from different levels of the '-omic' network, starting with metabolites, but also including protein and RNA markers. Using a set of complementary technical approaches we have identified metabolic alterations in breast cancer tissue. This information can be used to determine the molecular tumour type in order to provide information on tumour progression, prognosis and treatment efficacy through these metabolic changes. The characterisation of tumours based on biological information can be used to guide patient tailored therapeutic approaches as a step towards personalised medicine.
We and others have shown that metabolomics can be used to investigate the changes in tumour tissue related to apoptosis, to hypoxia and energy metabolism. The most popular approaches for metabolomics involve either MS (GC-MS or LC-MS) or NMR spectroscopy. MS based approaches are typically more sensitive. NMR spectroscopy can be applied to intact tissue samples and even observe metabolites in vivo. Through the combination of different techniques as well as DASL gene expression analysis and LC-MS / MS based proteomics we have been able cover all metabolic pathways that have been described in the databases such as the Kyoto encyclopaedia of genes and genomes (KEGG) database, maximising the possibility of discovering metabolic biomarkers associated with disease.
METACANCER concept - hypothesis and approach
In the METACANCER project we have evaluated the hypothesis that alterations in the steady state concentrations of metabolites reflect amplified changes in metabolism, and that these can be used to classify breast cancer based on tumour biology, to identify new prognostic and predictive markers and to identify new targets for future therapeutic interventions. Based on previous investigations by members of the consortium and other groups, metabolic changes in energy metabolism, but also in other metabolic processes such as membrane turnover, apoptosis and lipid metabolism, are central for the malignant phenotype of breast cancer cells.
The overall objective of METACANCER was to identify and to validate metabolic profiles that are considered to reveal characteristic biomarkers relevant for progression of breast cancer and to translate this knowledge into clinical practice. Results are also the basis for the identification of new molecular targets involved in tumour progression to guide personalised therapies.
The METACANCER project has evaluated four hypotheses:
1. alterations of metabolites can be reliably measured in human tumour tissue and reflect amplified changes in metabolism;
2. these alterations can be used to classify breast cancer based on tumour biology;
3. to identify new prognostic predictive markers; and
4. to identify new targets for future therapeutic interventions.
To address these issues, the METACANCER research consortium combined expertise in the field of clinical and molecular tumour pathology and clinical breast cancer research with advanced technological approaches that cover the state-of-the-art of metabolomic research In addition, the consortium includes expertise on transcriptomic data mining as well as the investigation of protein and mRNA markers including new high-through-put bead based analysis techniques.
The partners have worked together to provide the research infrastructure, technological basis and biobank for this project. To be successful in such an ambitious task, the establishment of a European network for collaboration was essential and is a prerequisite to contribute to the development of new predictive tests and new anticancer strategies. The project was aimed at the identification of biomarkers on the metabolite level that are able to predict outcome as well as therapy response in patients with breast cancer. These metabolite biomarkers were linked to proteomic and transcriptomic data to identify molecular changes in tumours that may be used to develop new therapeutic strategies. Therefore, the consortium is focused on metabolomics, but metabolic data is integrated in a research network that uses state-of-the art-information from genomic and proteomic analysis as well as data mining to describe multidimensional cellular networks. The consortium has its main focus on tumour tissue, but some analyses were also performed using serum samples to investigate possible metabolic biomarkers in biofluids for the prediction of chemotherapy response.
Project results:
The results of the METACANCER project are generally divided into three parts:
1. Validation of the methodological approach for classification of tumour types.
2. Combined analysis of metabolic, protein and gene expression data as well as pathway analysis.
3. Development of new tools for integrated analysis of metabolic data.
Part 1: Metabolomics as a new method for analysis of tumour tissue
The first major task of the METACANCER project was to introduce and evaluate metabolomics as a new method for cancer research. We have shown that the different metabolomics approaches NMR spectroscopy, GC-MS and LC-MS are suitable for frozen tumour tissue.
With each method it has been possible to evaluate several hundreds of metabolites, many of which could be identified and linked to biochemical pathways. Therefore it was possible to evaluate changes in metabolic pathways in different types of tissue. As expected, the largest differences were observed between normal breast tissue and malignant breast tumours. With all three metabolomics approaches it is possible to separate normal and malignant breast tissue in unsupervised and supervised analyses with a high sensitivity and specificity. In addition, we observed metabolic differences between different types of breast cancer. As known from other studies, the major groups of breast cancer are the hormone receptor positive and negative tumours. These tumour types have a different biological background, different clinical characteristics and are treated by different therapeutic strategies. In the metabolomics analysis, a set of metabolites was detected that was able to discriminate between hormone receptor positive and negative tumours, which had different metabolic profiles in our analysis. Other types of breast cancer are defined by the histological grade, which measures the dedifferentiation of the tumour and is one marker of tumour progression. By metabolomics, it was possible to detect differences between grade 1 or 2 and grade 3 tumours. Therefore we conclude that it is possible to generate meaningful biological information from tumour tissue by metabolic profiling, we have used this data to analyse the alterations of metabolic pathways in different types of tumour tissue. The results of the different metabolomics approaches for classification of tumour types and normal tissue are shown below.
Metabolite signatures for separation of breast cancer and normal tissues
In order to develop a metabolite based molecular approach for the separation of cancer and normal tissue we analysed each metabolite in the GC-MS data for its classifying power. A set of metabolites separated between tumour and normal breast tissues with sensitivity and specificity above 80 %. We found both, tumour markers and normal tissue markers. In order to enhance the separation, each of the tumours markers was divided by each of the normal tissues markers. Working with ratios instead of intensities has the advantage that these quantities are independent of data normalisation and robust against rescaling. For each classifier, we defined a cut-off value for the classification in cancer and normal tissues by maximising the sum of sensitivity and specificity.
There was a strong correlation between the votes of the different classifiers: For 217 tissues (58.8 %) there was concordant agreement among all classifiers, for 319 tissues (86.4 %) more than 90 % of the classifiers agreed on malignancy status.
Only 8 tumours and 6 normal tissues were classified wrong, leading to a sensitivity of 97 % and a specificity of 93.9 % of the molecular test.
Separation of healthy and normal tissue by 1H HR MAS NMR
Using multivariate statistical techniques on the entire data set the analysis by 1H HR MAS NMR was able to separate adipose tissue (classed as 100-50 % adipose) from healthy breast tissue (classed as 80 % or more breast tissue) and tumour samples. The separation was mostly driven by high concentrations of polyunsaturated and saturated fatty acid resonances in the adipose tissue and increased glucose resonances in the healthy and tumour samples compared to the adipose tissue. In addition normal healthy beast tissue was differentiated from tumour samples (ductal or lobular carcinoma) using stratified tumour samples with less than 10 % adipose tissue. Biochemical studies of breast cancer cells have suggested PCho as a biomarker of cell proliferation and cancer malignancy. PCho is a precursor, as well as a breakdown product of the major membrane component phosphatidylcholine, whereas GPC is solely a membrane breakdown product.
NMR-based metabolites separating ductal from lobular carcinoma, different tumour grades, different molecular subtypes
Invasive ductal carcinoma could also be separated from lobular carcinoma using a Partial least squares - Discriminant analysis (PLS-DA) model for the NMR data. Different tumour graded samples were also separated using multivariate data analysis. Grade 1 samples were readily separated from grade 2 and grade 3 tumours using Principal component analysis (PCA), PLS-DA and Orthogonal projections to latent structure (OPLS-DA) statistical models. In addition we could also separate grade 2 from grade 3 tissue.
The separation of Estrogen receptor (ER) and Progesterone receptor (PR) positive tissue by NMR was poor and only marginally better than the random predicted model from the negative samples. Similarly, we were not able to separate the Her2 classified samples. However, when using the combined classifications for Her2, ER and PR status, subtype 4 (Her2 + ER or PR pos) were separated from subtype 2 (Her neg, ER or PR neg) relatively robustly. We systematically tested multi-Block PCA and PLS, O2PLS-DA and correlation data fusion approaches on a combined dataset of 1H HR-MAS NMR, GC-MS and LC-MS measured on a cohort of 300 breast cancer patients with regard to various data pre-processing schemes and compared the outcomes. On the level of individual platforms it is apparent that the differentiation between healthy and diseased samples is often hampered by the heterogeneity of the tissue sample, specifically due to the fat content, with the individual platforms being affected to a different extent. Multi-block PCA differentiates the healthy and diseased subjects readily.
A metabolite index measured by GC-MS distinguishes between ER+ and ER- tumours
For construction of an ER status classifier from the GC-MS data, a feature selection step was combined with nearest centroid classification. First, metabolites were ranked by the capability to distinguish between training ER+ and ER- tumours based on Welch's t-test. Then, the centroids of the training ER+ and ER- tumours were calculated in the space of the top metabolites. Finally, test samples were predicted to belong to the class with the nearest centroid. A Metabolic index (MI) for prediction of ER status was constructed as linear combination of metabolites. To this end, the centroids of training ER+ and ER- tumours were determined in the metabolite space. The MI was defined as projection onto the difference vector between the centroids of ER+ and ER- tumours.
Using the training cohort, a MI was constructed as linear combination of the top metabolites. Receiver operating characteristic (ROC) analysis showed an excellent performance of the MI for prediction of the ER status in the training cohort. Validation of the MI in the validation cohort affirmed the excellent performance of the predictor.
Taken together, the results of METACANCER show that metabolites can be reliably measured in tumour tissue and that tumour metabolomics is able to provide useful information on biomarkers and biological pathways in different types of tissue and that this new method is a promising addition to transcriptomics, proteomics and sequencing.
Part 2 - Integration of metabolomics, transcriptomics and proteomics - analysis of tumour samples by combined multilevel -omics approaches
An additional major task of the METACANCER project was the integration of metabolomics, proteomic and functional genomic data which were generated using the same set of tumours. This resulted in parallel dataset on different omic levels and enabled a systems biology evaluation. The idea behind this approach was that it would be more useful to combine information on different biological levels rather that to have the complete analysis solely based on metabolomics.
We have used a data mining strategy using a system for evaluation of complex already published gene expression data sets to look for enzymes that might be interesting in breast cancer. The results of this approach have been linked to the metabolite alterations. From the combination of those approaches, we have focused on enzymes that might be relevant for the progression of breast cancer and the observed metabolic changes. We have analysed the relevant enzymes in breast cancer cohorts and functional models and linked the metabolic alterations to protein changes as well to changes in gene expression.
In addition to the data mining studies, we have also performed DASL gene expression analysis as well as LC-MS / MS-based proteomics. This allows us to study each tumour with different omic approaches which could then be combined to focus on the relevant pathway.
Within METACANCER, a large cohort of fresh-frozen breast tissue samples were analysed by metabolomics. The metabolomics analysis of the samples has been performed using three different methods, GC-MS, LC-MS and NMR. Additionally, genome-wide expression data of 150 out of these samples were generated using the DASL platform. Parallel to fresh-frozen tissues, Formalin-fixed paraffin-embedded (FFPE) tissues of patients in the METACANCER cohort were collected. Tissue microarrays (TMA)s were constructed and stained with a set of selected antibodies. Finally, proteomics analyses of 126 FFPE tissue sections were performed using LC-MS / MS.
All methods have been able to generate valid metabolomic, functional genomic and proteomic data from the samples. In the statistical analysis, it has been shown that it is possible to link the biological alterations of the samples with clinical pathological data. In particular, it has been possible to describe differences in metabolites between different subtypes of breast cancer. Based on METACANCER results it is now clear that metabolic profiling is a valid approach to analyse frozen tumour biopsies and that metabolomics data can be linked to other omic levels.
Therefore, at the end, metabolic profiling can be integrated in a systems' biology approach with other biomarkers derived from the other levels of biological organisation. By this strategy, we are able to go beyond the metabolite level and to identify and validate selected protein and mRNA biomarkers relevant for metabolic alterations. This results in a combined pathway dataset consisting of metabolites as well as key protein and mRNA markers as a basis for a validated diagnostic system to assess prognosis and to guide targeted therapies in breast cancer.
Global lipidomics for tumour tissue samples - changes in phospholipid synthesis are involved in the progression of breast cancer
Analysing the LC-MS lipidomics data, significant differences were observed in the tumour versus normal comparisons. In tumours, it was observed that tumour grade and ER status affect the lipid profiles most radically. The most significantly changing phospholipid was PC(14:0 / 16:0) which was different dependent on both ER status and grade. It is known that the majority of the ER tumours are of grade 3, and this was also the case in our patient population since only 7 % of the grade 1,2 tumours were ER, while 44 % of the grade 3 tumours were ER. Thus, either grade or ER status alone could explain our results. We therefore analysed the ER status only within grade 3 tumours and the grade only within ER+ tumours, and confirmed that both ER status and grade independently affected the same lipids, with the highest levels found in ER- grade 3 tumours.
Increased levels of phospholipids were observed during tumour progression, with the highest levels in hormone receptor negative grade 3 tumours. Using the GeneSapiens database for data mining, we have identified key enzymes of lipid biosynthesis. These enzymes were stained by immunohistochemistry in tumour tissue and analysed together with the LC-MS data of the same tumour cohort. We found that in particular FASN and ACACA, which are two major enzymes involved in phospholipid synthesis were increased on the protein level in those tumours that had high levels of phospholipids. This shows that it is possible to connect protein expression data by immunohistochemistry with LC-MS lipidomic profiles of the identical tumours.
To evaluate the function of these lipid metabolising enzymes for breast cancer proliferation and apoptosis, we investigated breast cancer cell lines and inhibited the central enzymes by small interfering RNA (siRNA). This resulted in decreased cell viability and in lower phospholipid level measured by LC-MS. The increased de-novo lipid synthesis is typically found in tumour cells, but not in normal cells, which would make it a promising target for new tumour-specific therapeutic approaches. The detailed results of this project part have already been published, for additional information see Hilvo M, Denkert C, Lehtinen L, Müller B, Brockmöller S, Seppänen-Laakso T, Budzsies J, Bucher E, Yetukuri L, Castillo S, Berg E, Nygren H, Sysi-Aho M, Griffin JL, Fiehn O, Loibl S, Richter-Ehrenstein C, Radke C, Hyötyläinen T, Kallioniemi O, Iljin K & Oresic M. Novel theranostic opportunities offered by characterisation of altered membrane lipid metabolism in breast cancer progression. Cancer Res. (2011) 71: pp. 3236-3245.
Cross-correlation of metabolite and enzyme expression data
Interesting insights in the regulation of metabolic pathways in breast cancer can be obtained from a combined analysis of metabolite data and enzyme expression data. We have correlated GC-MS metabolomics data with expression data of the same METACANCER tumours that were generated using whole-genome DASL. In order to investigate the consequence of changes in the expression of glucose transporters in breast cancer, we correlated the family of glucose transporters with the concentrations of sugars in the cancer cells.
Combined markers GPAM analysis and metabolic profile
We investigated glycerol-3-phosphate acyltransferase (GPAM), a key enzyme in the lipid biosynthesis of triacylglycerols and phospholipids. Expression of GPAM in malignancies is of special interest because phospholipids are an important and major component of all cell membranes. To investigate the protein expression of GPAM 228 breast cancer samples were immunohistochemical stained, digitised and evaluated. The conclusion from the results was that in breast cancer GPAM affects more the level of phospholipids than triglycerides, and most of the changes are independent from those associated with ER status and tumour grade. This implies that at least in the context of breast cancer the function of GPAM is directed more towards phospholipid production rather than triglyceride synthesis. This is reasonable as it has been shown previously that increased de novo fatty acid synthesis is a hallmark of cancer cells and the products of this lipogenic pathway are directed mainly to the cell membrane phospholipids.
To extend the number of metabolic pathways covered by the analysis, an additional GC-MS based metabolic analysis was performed. Metabolomics analyses of breast cancer tissue led to the identification of 467 metabolites. Out of these, 161 metabolites had a known chemical structure and could by mapped to metabolite names. The analysis from the GPAM expression with the data showed 57 significantly changed metabolites (p < 0.05). Out of these 57 metabolites, 19 could be identified. The manuscript with the complete data is currently under submission to a scientific journal: Scarlet F. Brockmöller, Elmar Bucher, Berit M. Müller, Jan Budczies, Mika Hilvo, Julian L Griffin, Matej Oresic, Olli Kallioniemi, Kristiina Iljin, Sibylle Loibl, Silvia Darb-Esfahani, Bruno V. Sinn, Frederick Klauschen, Judith Prinsler, Nikola Bangemann, Fakher Ismaeel, Oliver Fiehn, Manfred Dietel, Carsten Denkert. Integration of metabolomics and expression of GPAM in breast cancer - link to patient survival, hormone receptor status and metabolic profiling.
Part 3 - New tools for analysis of metabolomics data
In the METACANCER consortium, new tools for the analysis of metabolomics data were developed:
UCD (Davis, California) has developed MetaMapp, a tool to analyse metabolomics data in context of structural and functional relationship. MetaMapp visualises the relationships between identified metabolites but also unidentified expressed metabolite tags in a metabolic network.
CHARITE (Berlin, Germany) has developed a metabolite based signature that distinguished between cancer and normal tissues at high sensitivity and specificity. Among the important clinicopathological factors of breast cancer, the status of ER had the strongest influence on the metabolite profile of a tumour. Meanwhile, a metabolite test has been developed that distinguished between ER+ and ER- breast cancer (for details see above, part 1). Cross-correlation between metabolite and enzyme expression have been studied using the whole-genome DASL data. GPAM, a key enzyme in the biosynthesis of triacylglycerols and phospholipids, was identified as a prognostic marker in breast cancer. The tool METAtarget for linkage between enzyme end metabolite information has already been published, for details see: Budczies J, Denkert C, Müller BM, Brockmüller S, Dietel M, Griffin JL, Oresic M & Fiehn O. METAtarget - extracting key enzymes of metabolic regulation from high-througput metabolomics data using KEGG reaction information. GI-Edition: Lecture Notes in Informatics (2010) P-173: pp. 103-112.
VTT (Helsinki, Finland) has analysed the lipid metabolism in breast cancer by a combined analysis of lipidomics data and enzymes involved in the regulation of fatty acid metabolism. Alterations in the membrane lipids were detected that correlated with cancer progression, prognosis, hormone receptor status and tumour grade. Gene silencing experiments indicated that silencing of multiple lipid metabolism regulating genes reduced the lipidomic profiles and viability of breast cancer cells.
UCAM (Cambridge, UK) has carried out 1H HR MAS NMR and showed that cancer and normal tissues could be separated using this platform. Further, invasive ductal cancer could be separated from lobular cancer and tumours of grade G1, G2 and G3 from another grade. In contrast, separation of ER+ from ER- as well as HER+ from HER- tumours was poor. Data from GC-MS, LC-MS and NMR were combined and analysed with multi-block PCA, with this technique not only identifying the metabolite changes previously identified in the analysis of the individual datasets but also produced metabolic correlates across the platforms that improved our mechanistic understanding of the data.
Identification of new metabolites by accurate MS
A total number of 391 samples were successfully investigated by GC followed by Time-of-flight (TOF) MS. Prior to GC-MS, the tissue samples were split in a training (2 / 3) and a validation cohort (1 / 3). Both cohorts had a comparable distribution of the most important clinicopathological parameters. Both sets were profiled at UCD (Davis, California), the training samples at the end of 2008, the validation samples in the beginning of 2009. For the METACANCER project, a new analysis mode of BinBase was implemented. In the first step of the new analysis mode, abundant metabolites are identified using the training samples. In the second step, these metabolites were measured in the validation samples. In the first data calculation performed in early 2009, this approach led to the initial detection of 136 identified metabolites, known by known chemical structures and metabolite names. After further identification of unknowns by accurate MS, depicted below for UDP-N-acetylglucosamine, further 64 novel unique metabolites were identified, plus 28 that were annotated by elemental formulas and best-hit structures. The results of this project part have already been published: Kumari S, Stevens D, Kind T, Denkert C, Fiehn O. Applying in-silico retention index and mass spectra matching for identification of unknown metabolites inaccurate mass GC-TOF MS. Anal Chem. 2011 Jun 16.
MetaMapp to highlight biochemical changes from GC-TOF profiling
Due to the large diversity of molecules in metabolism, no single instrumentation or technique is capable of capturing all metabolites simultaneously. As tumour progression depends on the operation of major metabolic pathways we have used a platform that yields the widest range of primary metabolites to date, CG coupled to TOF MS. We linked data acquisition to automatic metabolite annotation by BinBase, an in-house database of over 7 000 metabolite tags supported by over 1 200 authentic chemical standards with molecular weights below 600 Da.
Mapping metabolites to biochemical pathways appears to be the logical method of choice, supported by a range of tools and pathway databases supporting such as HumanCyc, EHMN and KEGG. However, our metabolomic analyses revealed that 19 of all the 185 structurally identified metabolites could not be retrieved from these repositories, especially for lipids, sugars and sugar conjugates. This result indicated that mammalian metabolism is more diverse than represented by current enzyme-specific databases.
In order to capture all identified metabolites plus additional 372 currently unidentified expressed metabolite tags, we have devised a novel tool to visualise structural and functional relationships: MetaMapp. MetaMapp combines the functional information of biochemical substrate/reaction pairs (obtained from the KEGG Rpair DB) with structural information obtained from chemical similarities between all identified compounds and mass spectral similarities for unidentified metabolites. The underlying axiom is that biochemistry refers to the conversion of chemically similar compounds by catalytic enzymes. Hence, it appears logical to associate all compounds directly by their chemical similarity. Clusters of chemically similar compounds should then resemble biochemical modules. We have used the public substructure decomposition tool in PubChem to calculate a chemical similarity matrix for all identified metabolites by the established Tanimoto coefficient.
MetaMapp provides several advantages. First, it is a platform independent approach meaning it does not matter what technology was utilised to detect or identify a metabolite. Cross-platform metabolomics datasets can be readily integrated and visualise to infer biological conclusions. The only requirement is that the chemical structures should be associated with machine encoded chemical structures. Second, MetaMapp is not constrained by genomics information. All the metabolite detected in a cross species manner can be straightforwardly mapped, enabling mapping of the metabolite that are originated from diet or gut microbes. Third, the layout is not static and it can be updated according to the input list of compounds. For clarity, metabolites of unknown structure have been excluded. The increase in metabolites was accompanied by increases in enzyme concentrations using the LC-MS / MS based proteomic results. A direct mapping of both enzymes and metabolites is feasible through KEGG Rpair and enzyme maps.
Potential impact:
Potential impact including socio-economic impact and wider societal implications
Global and European relevance
Breast cancer is worldwide the most common cancer among women. The chance of developing invasive breast cancer at some time in a woman's life is about 1 in 8 (13 % of women). According to research by the 'Breast health global' Initiative, breast cancer is the most prevalent cancer in the world today due to its high incidence and relatively good prognosis. An estimated 4.4 million women are alive today in whom breast cancer was diagnosed within the last five years. However, breast cancer is the most common cause of cancer-related deaths among women worldwide. More than 1.1 million women worldwide are newly diagnosed with breast cancer annually. This represents about 10 % of all new cancer cases and 23 % of all female cancers. With more than 410 000 deaths each year, breast cancer accounts for about 14 % of all female cancer deaths and 1.6 % of all female deaths worldwide. Incidence rates are climbing by as much as 5 % annually in low-resource countries. Women living in North America have the highest rate of breast cancer in the world. Death rates from breast cancer continue to decline, with larger decreases in women younger than 50. These decreases are believed to be the result of earlier detection through screening and increased awareness, as well as improved treatment.
The METACANCER project is focused on identification of biological markers on the metabolite level that aid in assessing the prognosis and the treatment response of patients with breast cancer. These validated biomarkers are a basis for improved strategies in translational and clinical research that will improve patient care on the one hand, as these biomarkers will be useful for prognosis, monitoring and treatment selection of patients with breast cancer. In addition, the metabolic profiling is aimed at the identification of metabolite and alterations in metabolic pathways, as well. This leads to a better understanding of the basic mechanisms underlying cancer which will lead - after translation to the clinical setting - to improved treatment strategies. As an example, we have shown that phospholipids in tumour tissue are synthesised de-novo and that this process is increased in tumour progression. This suggests that therapeutic approaches targeting lipid biosynthesis for cellular membranes might be a promising approach in breast cancer.
Impact for cancer research and clinical management of breast cancer patients
Treatment options for breast cancer are initially identified according to the stage of the disease. There are several modes of treatment including surgery, radiotherapy, endocrine therapy and cytotoxic chemotherapy. However, the management of an individual patient will depend on many factors including menopausal status, hormone receptor status and treatment preferences. The final treatment regimen selected, following discussion with the patient, will depend on the individual circumstances making it difficult to manage the disease by following a systematic treatment algorithm. To allow for a better stratification of patients, it is important to identify those pathways that are relevant for tumour progression and therapy response, and to determine biomarkers that could be used to monitor the activity of those pathways.
The METACANCER project has shown that metabolomic analyses by GC-MS, LC-MS and NMR spectroscopy are suitable for the analysis of tumour tissue and provide valuable information that could be integrated with proteomic and transcriptomic data. This gives the possibility to study the changes in malignant tumours by constructing a correlation network that consists of combined RNA, protein and metabolite data. The European society will benefit from the knowledge gained through METACANCER resulting in new insights into the mechanisms of cancer progression as well as cancer patient management through improved molecular diagnostics leading to improvement of therapeutic concepts by selection of effective drugs as part of systems medicine.
Breast cancer diagnosis, prognosis and treatment are of global relevance. Accordingly new biomarkers for breast cancer diagnosis or new therapeutic approaches are of interest for companies in the diagnostic and pharmaceutical field. For example, part of the METACANCER evaluations are focused on the analysis of PARP positive and negative tumours. PARP is a new promising target in the clinical setting, however, results of recent clinical trials have shown that the PARP activity might be restricted to tumour subsets, which need to be identified by new molecular markers.
Establishing European research networks
The METACANCER consortium has been strategically built to contain the most important European groups and one major United States group (UCD) with a high complementarity to ensure the strengthening of the European research area in the field of cancer metabolomics. Due to the advanced technology base, it is envisaged that advances in understanding of metabolic signatures as well as factors contributing to clinical response seen in cancer therapy will be made, leading to the development of new strategies to improve patient care providing new rationales for treating cancer patients. The proposed project contributes substantially to the competitiveness of European science and industry in the area of development of diagnostic tools, identification of biomarkers and cancer drug research. This offers an unprecedented opportunity to provide the European Community with a framework of innovate science that concurs with the major socio-economic and medical objectives to establish improved cancer therapies, and will be competitive on a global scale.
The development of protectable therapeutic, diagnostic, prognostic and drugable candidates by identifying new response-associated markers is potentially a further exploitable outcome as well as providing novel treatment approaches based on an improved understanding of metabolic changes in tumours that could put Europe ahead of the field.
The formation of unique data and material repositories will also benefit future expansion of cancer-related research in the European Community research providing a knowledge and resource base for further work.
Solving social problems - societal contribution
Cancer is a central health problem in Europe, therefore research that leads to a better understanding of the biological changes in cancer progression will have a wide-ranging and significant social impact, including quality of life. Research activities will result in a reduction of therapy-related suffering of cancer patients for the most common cancer in women world-wide and will reduce the amount of ineffective therapies, the side effects and thus the costs for the health care system. Finally it is aimed towards a decrease of cancer mortality in Europe.
Socially, many people are employed in the European biotech/pharmaceutical industry and their job security and career success depend increasingly on competitiveness in world markets. Thus, development must be streamlined and must become more efficient to ensure the evolution of better therapeutic strategies and products with fewer unwanted side effects. Additionally, the inclusion of research teams from different part of Europe allows for a wider participation in the project and greater dissemination of results for the benefit of a wider population. This also provides a basis for the exchange of technologies and the enlargement of a research group whose extensive experience in metabolic analysis will be significantly enhanced through the introduction of evolving research skills. The benefits will channel to society at large and develop a model for research activity in this field for future advances in metabolomics and translational cancer research.
The METACANCER project is focused on the identification of biological markers at the metabolite level that aid in assessing the prognosis and the treatment response of patients with breast cancer. These validated biomarkers will lead to a better understanding of the basic mechanisms underlying cancer which will lead - after translation to the clinical setting - to improved treatment strategies.
The project has a high public health impact, as - in the long run - the investment in this area will contribute to reaching the aims of reducing cancer's incidence and mortality and improving quality of life for patients.
Scientific impact
The METACANCER project is the first large-scale effort to combine the three major metabolomic technology platforms for combating a major human disease. We have combined efforts to improve the likelihood that genuine metabolic biomarkers for breast cancer tissues can be detected and validated, that would eventually lead to diagnostic toolkits which would facilitate a much more precise prediction and prognostic toolkit. Due to the high biological variability in human tissues, all publications so far have shown that mathematical combinations of variables ('multivariate statistics') will be needed to describe vectors that would facilitate such high quality diagnostics. In addition, this study was the first attempt to link high quality metabolomic data to proteomics and transcriptomics data in order to enable better insights into the cellular mechanisms that define the onset and progression of breast carcinoma. This provides further functional validation of the biomarkers and will thus add confidence into the statistical models built for the clinical applications. In summary, this is a first step into the exciting new area of individualised biomedicine that links personalised metabolic phenotypes into prediction cohorts and then renders therapeutic suggestions based on correlation with cellular mechanisms.
Main dissemination and exploitation activities
The consortium has an interdisciplinary combination of university research groups, Small and medium-sized enterprise (SME)s and clinical partners thus offering a wide range of dissemination and exploitation possibilities. The METACANCER results have been disseminated the results in different ways:
- publication of scientific results in peer reviewed journals and contribution to international conferences (scientific partners) - the publication process is still ongoing;
- transfer of knowledge into and beyond the consortium;
- presentation of the scientific and technical results to industry;
- to create awareness for novel technologies towards the public.
Internal dissemination of research results was realised via a restricted area of a public website. This platform, which can be accessed via the internet, served as a common knowledge sharing tool. Developments during the project were continuously monitored and evaluated via the progress surveys (deliverables in WP9 - Training and dissemination), reports on project's progress, data and manuscripts, information material etc. and are available on this website. There were a total of six semi-annual plenary progress meeting. The results of this project were regularly pre-screened for potential protection of the Intellectual property (IP) before they could enter the public domain. The evaluation of the market and existing patents / patent applications in the field was be the prime considerations in this initial evaluation.
Information exchange with the scientific community
Results were disseminated by presentations during national and international conferences, by publications in international peer-reviewed bio-medical journals. Several publications are currently under preparation. Project's interdisciplinary workshops have been used to discuss results between the various experts of the consortium and invited external guest speakers.
Public relation:
In order to provide general information on advances in breast cancer diagnosis and treatment field, METACANCER has published the project's background, novel scientific approach as well as the related impact for patients, clinicians and science on a public website, via a project flyer and press releases.
The METACANCER website www.METACANCER-fp7.eu was set up during the first project period by tp21 and maintained and updated until project end. The website will be kept open for several years after the project end. A list of major METACANCER publications on this site will demonstrate the successful implementation of the project. The METACANCER booklet was elaborated and implemented by tp21. The booklet was disseminated at conferences, meetings, at partners' organisations and submitted to interested parties on request. Target groups for the booklet are the general public and young scientists. The aim of the booklet's design was to demonstrate the cross disciplinary and the pan European novel approach in cancer metabolomics. The booklet is promoted on the METACANCER public website and is available on request.
Teaching and visibility: An important deliverable of this project was the further education of younger scientists (PhD and post-doctoral researchers). The partners represent a diverse range of metabolomic and bioinformatic tools and thus, those involved in the project will be exposed to state of the art approaches in metabolomics. This helps produce scientists with vital skills in analytical and mathematical analyses, improving the skills base of Europe. Interdisciplinary workshops and an exploitation workshop (deliverable in WP T&D) will actively support young scientists' education (all partners).
Strategies for the exploitation - management of intellectual property
In order to reflect the high importance of the issue, an experienced dissemination and exploitation manager has been integrated in the consortium i.e. TP21. Furthermore all partners in the project have devoted substantial effort to this task over the course of the project.
Training and awareness creation amongst the project staff is essential to identify results that could be target for protection and on the other side already protected know-how that could interfere with the own approaches. Six monthly progress surveys of the management have been performed to reveal any results that could be relevant for application.
Project website: http://www.METACANCER-fp7.eu
