Final Report Summary - ENVIROGENOMARKERS (Genomics biomarkers of environmental health)
The purpose of the ENVIROGENOMARKERS project was to evaluate the potential of omics technologies for the discovery of new biomarkers of exposure and disease risk through the analysis of human biosamples in the context of environmental health-related molecular epidemiology studies. Particular emphasis was given to the potential for analysis, using omics methodologies, of biosamples already in long-term storage in existing biobanks constructed in previous decades prior to the advent of omics and without the corresponding precautions nowadays considered as necessary for the proper collection and preservation of samples, especially in relation to RNA integrity.
Initially, the project developed methodology making possible the transcriptomics analysis of buffy coats stored for more than a decade without RNA preservative sand established cut-off criteria for the selection of biobank samples with minimum impact of the sample collection and storage conditions on transcriptomics, epigenomics, metabolomics and proteomics profiles. Subsequently the project generated the corresponding omic profiles from large numbers of human biosamples (blood fractions) in the context of prospective nested case-control studies to look for biomarkers predictive of future risks of breast cancer or non-Hodgkin's lymphoma as well as from samples collected at birth (cord blood) or during the first four years of life to look for biomarkers predictive of birth outcomes and developmental effects. In addition, measurements of chemical-specific biomarkers (including persistant organic chemicals, cadmium, phthalates) were conducted in order to look for omics-based biomarkers of exposure.
In most types of omics profiles obtained, large numbers of signals were found which associate strongly with specific environmental exposures as well as with disease risks (mainly B-cell chronic lymphatic leukemia in adults, preterm birth and neurodevelopmental effects in children). Bioinformatics analyses provide strong support for the biological relevance of these signals, demonstrating the potential of omics for the discovery of new biomarkers. Importantly, in some cases evidence of overlap between omics signals associating with exposure and disease risk was obtained, implying possible etiological links between exposure and disease.
Project context and objectives:
During the past few decades, extensive research has been conducted on the use of biomarkers of exposure, early effects and individual susceptibility in the study of the environmental aetiology of chronic diseases such as cancer. More recently, the revolutionary developments in the area of genomics have opened exciting opportunities for the discovery of a new generation of biomarkers to study this problem. Yet, exploitation of -omics technologies has so far been limited. The objective of the ENVIROGENOMARKERS project was to move this field forward through the utilisation of a range of -omics technologies in the context of relatively large, population-based molecular epidemiology studies to discover novel biomarkers of exposure and disease risk and use them to examine exposure-disease links.
Omics technologies and their potential in environmental health research
The term 'omics' refers to a collection of technologies which have emerged during the past decade or so and which permit the simultaneous, quantitative measurement of the totality, or a large fraction, of cellular features belonging to particular categories. For example, transcriptomics permits the measurement of the levels of expression of all genes through the measurement of the abundance of the corresponding RNA species; proteomics the measurement of large numbers of proteins; metabolomics, the measurement of hundreds or thousands of small molecules (metabolites); epigenomics the measurement of large numbers of epigenetic markers of different types. By making possible the simultaneous and untargeted search of enormous numbers of potential targets, -omics technologies provide unique opportunities for the discovery of biomarkers of exposure to environmental toxins as well as of biomarkers capable of predicting disease risk. Furthermore, the integration, utilising advanced bioinformatics and systems biology methods, of global profiling data measured at multiple biological levels (gene and protein expression, cell metabolism, and genetic and epigenetic status) provides exciting new opportunities for the development of a holistic view of the organism's responses under the influence of environmental stresses and progressing disease.
The objectives and structure of ENVIROGENOMARKERS
ENVIROGENOMARKERS is the first project in Europe (and one of very few worldwide) to conduct a large-scale application of a range of -omics technologies in a population study aiming at:
a) the discovery and validation of novel biomarkers predictive of increased risks of chronic diseases in which the environment may play an important role (breast cancer, non-Hodgkin's lymphoma - NHL, neurodevelopment and other childhood conditions);
b) the exploration of the association of such risk biomarkers with environmental exposures, including high-priority pollutants (carcinogens and immunotoxicants such as PCBs and PAHs, neurotoxicants such as cadmium, lead and ambient air pollution) and emerging exposures (such as phthalates and brominated flame retardants), many of which are also endocrine disruptors; and
c) the discovery and validation of biomarkers of exposure to the above and other high-priority environmental exposures (e.g. water disinfection byproducts).
The project was designed as a case-control study nested within three existing European prospective cohorts which contain biosamples collected prior to disease diagnosis, dietary, exposure and life style information and follow-up health histories: The Northern Sweden health and disease study, EPIC-Italy, and the Rhea mother-child cohort on the island of Crete.
i) The Northern Sweden health and disease study (NSHDS) has collected blood samples and exposure / health data and by January 2008 it contained 95 000 unique individuals with 172 600 sampling occasions. Blood samples were divided into aliquots of plasma, buffy coat and erythrocytes and stored in -80 oC freezers within 2 hours of collection. Lifestyle and environmental information was collected, while health status follow-up was conducted via different registries.
ii) EPIC-Italy is part of the European investigation into cancer and nutrition (EPIC) study, a prospective study, initiated in 1992 and recruiting subjects aged 35 - 70 years, aiming to investigate the relationship between diet, lifestyle and environmental factors and the incidence of cancer. EPIC-Italy represents approximately 47 700 subjects and, in addition to biosamples, includes detailed questionnaire-based information at enrolment on diet, lifestyle and personal history. Biological samples, including plasma, serum, white blood cells and erythrocytes were collected and have been stored in liquid nitrogen. The time interval from collection to -80 oC freezing was generally 2 - 5 h. Study subjects are followed up identify newly diagnosed cases of cancer at all sites and other outcomes of interest.
The Rhea mother-child cohort in Crete is a population-based cohort which enrolled mothers at the third month of pregnancy and followed them up until birth and after birth together with their children. By 2008 it had recruited 1700 pregnant mothers and nearly 600 children had been born. Data available include various maternal exposures during pregnancy (FFQ, exposure to dietary carcinogens and immunotoxins, air and water pollution) as well as various biomarkers of exposure and disease risk measured in cord blood, thus enabling the examination of the impact of in utero exposures on development and health in later life. Samples available include maternal (and in some cases paternal) blood and urine obtained during pregnancy and serum, buffy coat, erythrocytes from cord blood. The children are being followed up and subjected to clinical developmental tests at different ages.
In all cases, samples (serum, buffy coat for extraction of RNA and DNA, erythrocytes) collected 2 - 15 years prior to disease diagnosis in the case of the cancer studies, and corresponding samples collected at birth (cord blood) and at 2 and 4 years of age in the case of the children's study, were analysed for chemical-specific biomarkers of exposure as well as for omic profiles using the technologies mentioned. Subsequently advanced statistics was employed to look for features in the omic profiles correlating with:
a) disease risk, by comparing cases and controls; and
b) exposure to the chemicals of interest, by comparing subjects with different exposure levels.
Comparison of the lists of hits in each case allowed the search for biomarkers associated with both exposure and disease risk, thus providing evidence of aetiological links between the two ('meet-in-the-middle').
Omic technologies employed include microarray-based transcriptomics (whole-genome expression profiling using the Agilent 4x44k platform) and epigenomics (DNA methylation at more than 450 000 CpG sites using the Illumina HumanMethylation450 platform), metabolomics (ultraprecision liquid chromatography combined with mass spectrometry - UPLC/MS-MS) and wide-target proteomics (measurement of tens of immune-related proteins using the Luminex LabMAP technology). Importantly, because for the first time ever extensive use was made within ENVIROGENOMARKERS of biosamples stored in biobanks constructed prior to the emergence of omics and, hence, without the precautions nowadays used in relation to sample handling and storage (especially in relation to the preservation of RNA), the project also included a systematic technical validation of genomic technology's use with such samples.
Exposure assessment utilised chemical-specific biomarkers of exposure, food-frequency questionnaires, modelling and geographic information systems-technology. The main chemicals targeted in the project, and the corresponding biomarkers employed, were the following:
- persistent organic pollutants, including PCBs (118, 138, 153, 156, 170 and 180), hexachlorobenzene (HCB), pesticide p,p'-DDT and it's metabolite p.p'-DDE and the polybrominated dipheynyl ether BDE-47, measured in serum using GC-MS/MS;
- cadmium, measured in erythrocytes (ICP-MS);
- phthalates (children's study), 8 metabolites measured in urine using LC-MS/MS.
A plethora of additional data on exposure, biomarkers (including genome-wide SNP data) and health indices, available through other studies were also utilised.
Project results:
Main scientific and technological (S&T) results
a) Pilot study
The application of omics technologies in epidemiological studies raises certain practical issues of sample suitability, especially in relation to RNA quality for transcriptomics analysis, requiring that care be taken for blood samples to be collected and stored in the presence of RNA preservative. Yet, no study has evaluated systematically the influence on omic profiles of the handling and prolonged storage of blood samples and their components in these biobanks. As a first step in the project, a study was conducted to evaluate the reliability of omics data obtained from archived biosamples collected prior to the advent of omics technologies.
The study was conducted in two phases: During phase I, methods were established for the isolation of RNA of the desired quality from buffy coats isolated from blood freshly collected and processed without RNA preservative. Furthermore, the influence was examined on omics profiles of sample handling and storage-related parameters selected following scrutiny of the procedures employed at the project's partner biobanks. For this purpose, samples of fresh blood were collected from healthy volunteers using different anticoagulants (heparin, EDTA or citrate) and processed in different ways. After being allowed to stand at room temperature for different times up to 24hr ("bench-time"), the blood samples were separated into buffy coats, erythrocytes and plasma, aliquoted and stored at -80oC or in liquid nitrogen. Finally, samples representing different collection, processing and storage conditions were analysed on the different omics platforms employed in the project.
The results of phase I were used to establish minimum criteria that samples must satisfy in order to be suitable for reliable omics analysis. During phase II, historic samples from the partner biobanks, satisfying these criteria, were analysed so as to evaluate the influence of long-term storage.
RNA and DNA isolation
Several attempts have been made to extract RNA and DNA simultaneously; however, none of the methodologies applied were found to be applicable. Therefore it was decided that separate samples are needed for DNA and RNA extraction.
Transcriptomics-quality RNA could be isolated from phase I buffy coats frozen within eight hours of blood collection by thawing the samples fully immersed in RNAlater or Qiazol followed by immediate extraction of RNA. Application of the same method to samples obtained from the project's partner biobanks resulted in the isolation in good yield of high-quality RNA from most samples.
RNA yield and RIN for a set of EPIC samples stored in nitrogen (1 - 27) or at -80 oC (28 - 33)
Omics analysis of phase I samples showed that the main variable affecting the main parameters affecting the observed profiles were the subject (all omic platforms), anticoagulant (metabolomics, proteomics, transcriptomics) and bench-time (within two hours for transcriptomics, beyond eight hours for metabolomics and transcriptomics). Based on this, cut-off criteria ensuring minimum impact on biobank samples of processing conditions were defined as bench-time < 2 hours and use of the same anticoagulant.
To evaluate the impact of storage time on omics profiles of samples stored in the project biobanks, samples from a set of 31 subjects from each biobank were selected, coming only from healthy female donors and from the same collection centre per biobank to minimise the influence of other parameters. Their storage time prior to analysis was 13 - 17 years, while the collection-to-storage times for the EPIC-Italy sub-set were 100 - 198 min. No significant effect of storage time or collection-to-storage time could be detected for any of the omic profiles.
The overall conclusion of the pilot study is that a large fraction of the human blood-derived samples in storage in the project's biobanks is amenable to analysis using the omics technologies of the project, on condition that the time between blood collection and fractionation did not exceed 8 hr (4 hr for proteomics). These findings open the way to the application of omics technologies to biosamples collected over previous decades in the context of population-based or disease-oriented cohorts. The results of the pilot study have been published (Hebels et al., Performance in Omics Analyses of Blood Samples in Long-Term Storage: Opportunities for the Exploitation of Existing Biobanks in Environmental Health Research. Environ Health Perspect. 5 Feb. [Epub ahead of print] PubMed PMID:23384616]).
b) Cancer studies
Subjects for inclusion in the study were selected from the partner biobanks on the basis of the cases having a date of diagnosis 2 - 10 years after recruitment for breast cancer and 2 - 15 years for NHL. Controls were matched on the basis of sex, age, collection centre and years in storage. Based on these criteria and availability at the two biobanks, the following numbers of case-control pairs, per disease and cohort, from whom samples were to be analysed, were decided (totalling 900 pairs):
Cohort - case - control pairs
breast cancer - NSHDS - 325
EPIC - Italy - 294
NHL - NSHDS - 197
EPIC - Italy - 84
Initially full omic analyses were conducted on samples from 100 case-control pairs for each disease and the resulting omic profiles were analysed in terms of their relationships with disease status and exposures. For the statistical analysis, a mixed model was defined which included a random effects component (e.g. dates of RNA isolation, labelling and hybridisation for transcriptomics) which dealt effectively with nuisance variation. Data analysis using this model showed that, while a small number of transcriptomics, proteomics and epigenomics signals showed some association with disease risk (especially for NHL) and exposure, none reached statistical significance sufficient to survive multiple testing criteria.
From these results it was concluded that, while some hits appeared to emerge, no reliable selection of candidate biomarkers, to be subsequently validated through targeted analysis of larger numbers of samples, could be made. On the basis of this conclusion, the strategic decision was made to cancel the plans for targeted analysis and, instead, increase the number of samples on which to conduct full omics analysis so as to enhance the statistical power of the study. Furthermore, given that, in general, the significance of the top hits was better for NHL than for breast cancer, it was decided to focus the rest of the study on NHL by conducting full omics analyses on all available NHL samples, thus bringing the total number thus analysed from 100 pairs to 281 pairs (197 for NSHDS and 84 for EPIC Italy). After completion of these analyses, multiple statistical models and methodologies were used to evaluate to resulting data, as previously using mixed models to minimise the impact of nuisance (technical) variation. The outcome, summarised below, amply justified the choice made:
Transcriptomics:
Even with the strictest statistical criteria (Bonferoni), a number of signals associating with NHL risk, and some tens of signals associating with exposure to a number of PCBs (but not cadmium or lead), were found.
Numbers of transcriptomics signals associating with exposures or NHL risk at different statistical significance levels
As regards disease risk, using a (strict) criterion of FWER = 1 % and ENT=10 000, 12 significant signals were observed, almost all of which associate very strongly and exclusively with B-cell chronic lymphatic leukemia (BCLL). When the analysis was restricted to specific subtypes, more than 600 signals were found to associate significantly with BCLL risk and only approx. 10 to associate much more weakly with multiple myeloma (MM) or diffuse large B-cell lymphoma (DLBCL). Analysis by time-to-disease (TtD) showed approximately equal numbers of signals being statistically significant at TtD < 6 and TtD > 6, with the former category showing clearer patterns and a tendency to increase in expression while getting closer to disease.
Pathway analysis of the signals associating with BCLL revealed that the affected pathways were almost uniformly involved in lymphocyte-specific signalling such as B-cell receptor signalling and immune response and included some highly biologically relevant pathways such as aberrant B-cell receptor signalling which has been well documented for BCLL.
Pathway analysis of transcriptomics signals associating with BCLL
Turning to associations of transcriptomics signals with exposure, some tens of signals were found for a number of PCBs using stringent statistical criteria.
Transcriptomics signals associating with exposures at different levels of statistical stringency
Initial efforts at pathway analysis revealed only 1 pathway for most PCBs, namely 'calcineurin-regulated NFAT-dependent transcription in lymphocytes'. This may be a biologically relevant finding since expression of this gene is known to be mediated through the Ah receptor which is a target of some PCBs, however further work is required to evaluate fully this finding.
Finally, attempts have been made to search for 'meet-in-the-middle' biomarkers by looking for overlaps between the lists of signals (expanded by somewhat relaxing the statistical criteria) which associate with both disease risk and exposures. A few weak candidates have been found linking BCLL with cadmium and PCB-156, which require further evaluation.
Proteomics:
A total of 32 immune-related proteins were measured, of which 4 were excluded owing to their low levels. Of the remaining 28 signals which were included in the statistical analyses, 2 showed a statistically significant association with lymphoma risk in all subjects as well as in subjects with TtD both < and > 6 years. Analysis by sub-type revealed strong signals only for multiple myeloma, with 6 proteins showing downregulation mostly at TtD < 6 years.
Proteomics signals associating with multiple myeloma. Each column of symbols represents one protein signal measured. Vertical axis: -logp; dashed line: limit of statistical significance (Bonferoni correction); blue symbols: TtD < 6; red symbols: TtD > 6.
As regards exposure biomarkers, numerous significant associations were found for PCBs (mostly downregulation) and non-PCB chlorinated compounds (DDE, hexachlorobenzene; mostly upregulation) and fewer associations with cadmium. Some of these overlapped with signals associating with disease and are therefore possible meet-in-the-middle biomarkers, but further work is required to evaluate this possibility.
Proteomics signals associating with exposures
Metabolomics:
All samples from the subjects originally selected from the two cohorts (totalling 900 case-control pairs) were subjected to full UPLS-MS/MS metabolomics analysis. After quality evaluation and clean-up of the resulting datasets, statistical analysis did not reveal any signals associating with risk for either disease at the desired statistical significance. Small numbers of signals showed weak association (p < 0.05) with disease risk, 4 for NHL and 24 for breast cancer in NSHDS and 0 for NHL and 21 for breast cancer in EPIC Italy; however, there was no overlap between the two cohorts. Furthermore there was no strong association of the NHL hits with any particular lymphoma subtype. It was therefore concluded that no viable metabolomics biomarkers predictive for disease risk were found.
Further analysis of the metabolomics dataset in relation with other parameters suggests weakly significant associations with exposure to PCB156 as well as with sex, body burden, alcohol consumption and vegetable consumption, however these associations are substantially below the desired level of statistical stringency.
d) Epigenomics
Large numbers of signals significantly associating with NHL risk were found. As for transcriptomics, these signals clustered mainly with BCLL.
As regards epigenomics biomarkers of exposure, substantial numbers of signals were found to associate significantly with exposure to hexachlorobenzene (HCB) and, to a smaller degree, dioxin-like PCBs. GO term analysis of the genes associated with HCB exposure shows a substantial number of terms related to immune function, suggesting that HCB may alter the latter. Pathway analysis shows activation of multiple cancer-related pathways, with EGFR signalling being the most significant pathway.
c) Children's study
In view of the experience gained from the cancer study in relation with the need to maximise the statistical power of the project by conducting full omics analysis of the maximum possible number of samples, it was decided to modify the initial plans and
a) focus primarily on neurodevelopment, rather than multiple developmental end-points; and
b) instead of having a discovery phase using omics analyses of a limited number of samples and a validation phase based on targeted analyses of larger numbers of samples, expand the number of samples on which full omics would be conducted while cancelling the validation phase.
The final arrangement adopted involved clinical testing, focussing on neurodevelopment, on children at ages two and four years, and omics analysis of samples from cord blood (so as to relate omic profiles to preceding in-utero exposures and future neurodevelopmental risks) and from blood collected at four years (to search for biomarkers associated with the clinical state and follow their development since birth). Taking into account the availability of cord blood material, pairs of cord- and 4-year-old-blood samples from a total of 118 children were subjected to full omics analysis. On the other hand, clinical evaluation for neurodevelopment and other clinical end-points, serum metabolomics and measurement of biomarkers of exposure (serum POPs, urine phthalates), were conducted on a total of approx. 600 children. Additional exposures for which data were available for all children include water disinfection byproducts, air pollution, biomarkers and estimates of exposure to dietary carcinogens, DR Calux and maternal smoking and passive smoking. Other data available include maternal psychological status, gestational hypertension, diabetes and metabolic symptoms, perinatal outcomes, cord leptin concentration, obesity, growth patterns (all, from birth onwards), depression (mother, pregnancy and postpartum), asthma / wheeze / allergies / atopic dermatitis / FeNO (first year and four years), lung function (fourth year), metabolic syndrome (mother, pregnancy), reproductive outcomes, birth weight and gestational age, anogenital distance, genotoxic outcomes-micronuclei (at birth and four years), thyroid function (mother, children), child cardiological condition (four years).
Transcriptomics
To look for biomarkers predictive of neurodevelopmental performance, buffy coats from cord blood were subjected to transcriptomics. After quality control and pre-processing, 44 019 probes were included in the statistical evaluation of correlation of expression levels versus the 6 neurodevelopment scores at age 4.
Bioinformatics analysis showed a number of pathways to be overrepresented among most of these gene lists; however, no striking links with developmental processes or nervous system function were obvious.
Metabolomics
For the children's study metabolomics analysis was conducted using a targeted approach based on the Biocrates p180 IDQ platform which provides information on 180 selected metabolites including 20 aminoacids, 18 biogenic amines, 40 acylcarnitines, 90 glycerophospholipids, 15 sphingolipids and hexose. PCA analysis showed a clear separation of the three types of samples (maternal, cord, four yearrs) analysed. Statistically significant differences between the sample types were observed for a number of aminoacids, glycerophospholipids and acylcarnitines.
PCA plot of metabolomics profiles of sera from maternal, cord and four-year old children's blood
Evaluation of the correlation between metabolomic profiles in serum and indices of neurodevelopment showed typically 10 - 20 statistical significant (Pearson p < 0.05) features for cord serum and 1 - 5 features in four-years serum. The metabolites concerned included carnitines, phosphatidylcholines, sphingolipids,sphingomyelins and symmertrical dimethylarginine. When each neurodevelopment variable was divided in two categories (below median and above median) and a combined measure of neurodevelopment scales was calculated by counting the number of 'high' category values across all six scales, moderately significant associations were found for four metabolites in four-years serum (a carnitine, tyrosine, valine and a phosphatidylcholine.
Significant differences in the levels of specific metabolites were observed between sera of mothers with normal and preterm births, including glycerophospholipids and aminoacids. As regards metabolomic profile of cord blood and birth outcomes, a number of metabolites were found to associate with preterm birth, gestational age, fetal weight growth restriction. Finally, significant associations were also found for small numbers of metabolites in sera collected at four-years with maternal passive or active smoking during pregnancy.
Epigenomics
Evaluation of the epigenomics profiles obtained in cord blood revealed a significant number of signals associating with different indices of neurodevelopment at four years.
There was minimal overlap between the hits observed for the different neurodevelopmental indices. No obvious connection of the hits observed with neurodevelopment-related biological features was noted. Strikingly, the association with the same neurodevelopment indices of epigenomics signals observed in samples collected at 4 years of age was much weaker, with signals significant at FDR < 0.05 being observed only for motor (536 hits). This is surprising in view of the fact that for this index no hits were found in the cord blood samples. Statistically weaker (FDR < 0.15) signals were observed in 4-year blood samples to associate with perceptual (15 hits) and cognitive (4 hits).
As regards exposure, significant numbers of signals were observed mainly in cord blood samples associating at FDR < 0.15 mainly with hexaclorobenzene (29 hits) and dioxin-like PCBs (126 hits), with no overlap with hits related to neurodevelopment.
Other exposures
As for the cancer studies, in addition to the main target exposures described above, exposure information generated through other projects has been collected to be utilised for the evaluation of its impact on omics profiles. This includes extensive information on exposure to air pollution including PM10, PM2.5 coarse particles, NO2, NO (ESCAPE project), disinfection-by-products (HiWate project), dietary carcinogens and immunotoxins including acrylamide, PAHs, N-nitroso compounds, dioxins and others, micronucleus data in cord four-year blood as well as detailed dietary intakes through FFQs (NewGeneris project), heavy metals, cotinine, lipids etc. These data have been added to the ENVIROGENOMARKERS dataset and will be utilised in the analysis of omics profiles.
d) Chemical-specific biomarkers of exposures versus disease risks
? detailed evaluation was conducted of the relationship between biomarkers of exposure (POPs, cadmium, lead) and disease risk for both breast cancer and lymphoma and in both cohorts. No evidence of an association between exposure to PCBs, DDE HCB or cadmium and overall risk of NHL was found, although weak evidence of positive associations for cadmium and NHL in females, and for metals / POPs and Follicular lymphoma, was observed. On the other hand, weak but consistent unexplained inverse associations with PCBs, DDE and HCB in males were observed. As regards breast cancer, unexplained, consistent evidence of a significant inverse association was observed between breast cancer and cadmium, HCB and DDE.
e) Exposures
The environmental exposures targeted by the project were estimated using biomarkers measured in archived or freshly collected material. The ranges of exposure levels observed are summarised below:
- Cadmium: Levels of cadmium measured in archived erythrocytes from the EPIC-Italy and NSHDS biobanks were obtained in the same analysis as measurement of lead levels.
- Persistant organic hydrocarbons: Levels of PCBs, DDT, DDE, hexachlorobezne and brominated diphenyl ether 47 were measured in archived sera in the adult cohorts as well as from pregnant mothers in the Rhea cohort.
- Phthalate metabolites: The levels of seven phthalate metabolites were measured in urine collected from pregnant mothers and children at age two years, as well as in urine of fathers.
Use and dissemination of foreground
Dissemination measures and plans
The project's activities and results have been publicised and disseminated using the following means:
- The website http://www.envirogenomarkers.net where fundamental information about the project's objectives, structure and partners was provided, along with updates of ongoing activities. During the period August 2009 - April 2013 the website received a total of 2227 unique visitors, corresponding to over 50 new visitors per month. During the last year of the project's formal duration (March 2012 - February 2013) the website was visited by a total of 824 visitors, corresponding to more than 68 new visitors per month.
- Production of a press release at the time of the project's start, as well as of an electronic brochure and two newsletters, which were circulated to hundreds of recipients and are also available for downloading from the project's website.
- Soon after its initiation, the project was highlighted in a feature in Nature magazine titled 'Prevention by numbers' (Nature, 458, 9 April 2009, 792 - 793) in which research efforts in the area of environmental health ongoing in the United States (US) and Europe were presented.
- Publication of the main scientific findings of the project in peer-reviewed international journals. So far 11 papers have been published (uploaded on ECAS), while a number of additional papers have been submitted or are in preparation. A publication plan has been put in place which foresees that some dozens of publications will be produced within two years of the project's formal completion. By extending the validity of their consortium agreement by 24 months the partners have committed themselves to fulfilling this plan.
- Presentation of the project and its results at a large number of international and national conferences targeting different audiences, including the research community, experts and authorities active in human biomonitoring, exposure assessment, chemicals risk assessment etc. In addition the project consortium maintained regular contact with other European projects in the area of environmental health, being presented at a number of their meetings (e.g. Cophes, ECNIS, NewGeneris, Exposomics).
List of scientific (peer reviewed) publications
1. Performance in Omics Analyses of Blood Samples in Long-Term Storage: Opportunities for the Exploitation of Existing Biobanks in Environmental Health Research, Hebels D.G.A.J. Envir. Hlth Perspectives, 121 (4), April 2013, NIEHS, US, 2013, 480-487, doi: dx.doi.org/10.1289/ehp.1205657
2. Metabolic profiling detects early effects of environmental and lifestyle exposure to cadmium in a human population, Ellis J. K., BMC Medicine, 19 June 2012, BMC, US, 2012, doi: 10.1186/1741-7015-10-61)
3. Effect of high doses of folic acid supplementation in early pregnancy on child neurodevelopment at 18 months of age: the mother-child cohort 'Rhea' study in Crete, Greece, Chatzi L., Public Health Nutrition, 15 (9), February 2012, Cambridge University Press, UK, 2012, 1728 - 1736, doi: 10.1017/S1368980012000067
4. Chiral Metabonomics:1H NMR-Based Enantiospecific Differentiation of Metabolites in Human Urine via Direct Cosolvation withß-Cyclodextrin, Pe´rez-Trujillo M., Anal. Chem., 84, American Chemical Society, US, 2012, 2868 - 2874, doi: 10.1021/ac203291d
5. Meeting-in-the-middle using metabolic profiling - a strategy for the identification of intermediate biomarkers in cohort studies, Chadeau-Hyam M., Biomarkers, 16 (1), Informa Healthcare, UK, 2011, 83 - 88, doi: 10.3109/1354750X.2010.533285
6. Integrating biomarkers into molecular epidemiological studies , Vineis P., Curr. Opin. Oncol., 23, Wolters Kluwer Health | Lippincott Williams & Wilkins, US, 2011, 100 - 105, doi: 10.1097/CCO.0b013e3283412de0
7. ESS++: a C++ objected-oriented algorithm for Bayesian stochastic search model exploration, Bottolo L., Bioinformatics, 27 (4), Oxford University Press, UK, 2011, 587 - 588, doi: 10.1093/bioinformatics/btq684
8. Advancing the application of omics-based biomarkers in environmental epidemiology, Vineis P., Environmental Molecular Mutagenesis, 21 March 2013, EMS, US, 2013, doi: 10.1002/em.21764
9. Methodological challenges for the derivation of -omics based biomarkers, Chadeau- Hyam M., Environmental Molecular Mutagenesis, in press, EMS, US, 2013
10. Making sense of OMICS data in population-based environmental health studies, Kyrtopoulos S. A., Environmental Molecular Mutagenesis, in press, EMS, US, 2013, doi: 10.1002/em.21778
11. A Composite Framework for the Statistical Analysis of Epidemiological DNA Methylation Data with the Infinium Human Methylation 450K BeadChip, Valavanis I., IEEJ Biomed Health Informatics, in press, Institute of Electrical and Electronics Engineers, UK, 2013
List of dissemination activitiesIST OF DISSEMINATION ACTIVITIES
(No., type of activities, main leader, title, date, place, type of audience, size of audience, countries addressed)
1. Website, Botsivali M., ENVIROGENOMARKERS , 2009-ongoing, http://www.envirogenomarkers.net all, global, on average 50 unique visitors per month, global
2. Electronic brochure, Botsivali M., ENVIROGENOMARKERS, 2009, available on project website and circulated by email, scientific, industry, policy, media, 200, global
3. Press release for CORDIS, Botsivali M., ENVIROGENOMARKERS, 2009, https://cordis.europa.eu/article/id/114224-envirogenomarkers-new-generation-of-biomarkers-to-study-environmental-agents-in-human-disease scientific, industry, policy, media, large, global
4. Nature magazine, news item, Kyrtopoulos S. A., 'Prevention by numbers', 9 April 2009, Nature, Vol. 458, pp 792 - 793, scientific, industry, policy, media, large, global
5. Press release, Botsivali M., ENVIROGENOMARKERS, 2009, European Press Agency, scientific, industry, policy, media, large, global
6. Conference, Keun H., International Conference on Environmental Mutagens, 2009, Florence, scientific, industry, policy, media, 250, Europe
7. Conference, Kyrtopoulos S. A., International Conference on Environmental Mutagens, 2009, Florence, scientific, industry, policy, media, 250, Europe
8. Conference, Kyrtopoulos S. A. , Workshop on Genomics in Cancer Risk Assessment, 2009,Venice, scientific, industry, policy, 100, Europe, US
9. Workshop, Kyrtopoulos S. A., American Association for Cancer Research, 2010, Miami, scientific, industry, policy, 200, US, Europe, Asia
10. Project factsheet in book, Botsivali M., ENVIROGENOMARKERS, 2010 , European research on environment and health funded by FP7, vol. 1, scientific, industry, policy, media, large, global
11. Electronic newsletter 1, Botsivali M., ENVIROGENOMARKERS , 2010, available on project website and circulated by email, scientific, industry, policy, media, 200, global
12. Conference, Keun H. , UK Environmental Mutagenesis Society Meeting, July 2010, United Kingdom, scientific, industry, policy, media, 200, United Kingdom
13. Conference, Keun H., American Association for Cancer Research, April 2010, Washington DC, scientific, industry, policy, media, 250, US, international
14. Conference, Kyrtopoulos S. A., European Environmental Mutagen Society, 2010, Oslo, scientific, industry, policy, media, 250, Europe
15. Conference, Kyrtopoulos S. A., International Union of Toxicology, 2010, Barcelona, scientific, industry, policy, media, 500, Europe, US, Asia
16. Conference, Kyrtopoulos S. A., ICCA-LRI and JRC Workshop, 2010, Stressa, Italy, industry, policy, scientific, media, 200, Europe, US
17. Conference, Botsivali M., European Environment and Health Conference, 2010, Brussels, policy, media, scientific, industry, 150, Europe
18. Conference, Kyrtopoulos S. A., European Environmental Mutagen Society, 2011, Barcelona, scientific, industry, policy, media, 250, Europe
19. Conference, Keun H. , American Chemistry Council / ICCA Long Range Initiative Meeting, 2011, Quebec, scientific, industry, policy, media, 500, Canada, US, Japan, Europe
20. Conference, Kyrtopoulos S. A., International Society for Environmental Epidemiology, 2011, Barcelona, scientific, industry, policy, media, 400, Europe, US, Asia
21. Conference, Vermeulen R. C., International Society for Environmental Epidemiology, 2011, Barcelona, scientific, industry, policy, media, 400, Europe, US, Asia
22. Workshop, Georgiadis P., Design of Future Molecular Epidemiology Studies and New Biomarkers, 2012, London, scientific, 100, Europe
23. Conference, Kyrtopoulos S. A., Human Biomonitoring - Linking Environment to Health and Supporting Policy, 2012, Larnaca, Cyprus, scientific, policy, 100, Europe
24. Conference, Chatziioannou A., 12th Intl Conf on BioInformatics and BioEngineering, 2012, Cyprus, scientific community, hundreds, European and other
25. Poster, de Kok T. M., Society of Toxicology, 2012, San Fransisco, scientific, large, global
26. Lecture, de Kok T. M., Molecular Epidemiology and Prevention, 11 December 2012, Masters program Biomedical Sciences Maastricht University (the Netherlands), scientific, 80, the Netherlands, Denmark, Bulgaria, Finland, United Kingdom, US, Estonia, China
27. Lecture / workshop, de Kok T. M., Genomics biomarkers and their application in molecular epidemiology, 17 April 2012, International Course on Molecular Epidemiology, Maastricht, the Netherlands, scientific, 25, multiple EU
28. Lecture / workshop, de Kok T. M., International Course on Molecular Epidemiology, 25 April 2012, Valkenburg, the Netherlands, scientific, large, the Netherlands, Estonia, Spain, Bulgaria, Denmark, Ireland, United Kingdom
29. Workshop, Keun H., Nutritional Genomics Annual Meeting, August 2012, Helsinki, scientific, 200, Europe-wide
30. Conference, Keun H. , European Association for Cancer Research, July 2012, Barcelona, scientific, industry, policy, media, 1000, Europe
31. Conference, Keun H., Frontiers in Cancer Nutrition, February 2012, Arizona Cancer Centre, scientific, policy, media, 500, US
32. Conference, Keun H., International Society for the Study of Xenobiotics, June 2012, Amsterdam, the Netherlands, scientific, industry, policy, media, 500, Europe
33. Conference, Keun H. , EUROTOX - Congress of the European Societies of Toxicology, June 2012, Stockholm, Sweden, scientific, industry, policy, media, 50, Europe
34. Conference, Keun H. , Hong Kong Society of Clinical Chemistry, January 2012, Hong Kong, scientific, policy, media, 500, China
35. Conference, Valavanis I., 12th International Conference on BioInformatics and BioEngineering, 2012, Cyprus, scientific, 2012, global
36. Lecture, de Kok T. M., Application of genomics in population studies, 9 May 2013, ECNIS2 course Molecular Epidemiology, Lódz, Poland, scientific, 20, multiple EU
37. Electronic newsletter 2, Botsivali M., ENVIROGENOMARKERS, 2013, available on project website and circulated by email, scientific, industry, policy, media, 200, global
38. Lecture, de Kok T. M., Symposium on ENVIROGENOMARKERS and NTU Exposomics, 2013, Taiwan, Scientific, 100, Asia
39. Conference, Vineis P., Omics in Epidemiology, 22 March 2013, Lyon Canceropole, Clinicians, 200, France
40. Conference, Vineis P., Omics in Epidemiology, 18 April 2013, University of Maastricht, Molecular biologists, 30, the Netherlands
41. Workshop, Vineis P., The exposome: the example of EGM, 1 September 2013, Interlaken, EUROTOX, toxicologists, 100, Europe
42. Workshop, Keun H., Exposome research, April 2013, College of Public Health, National Taiwan University, scientific, 100, Taiwan
43. Conference, Kyrtopoulos S. A., Symposium on ENVIROGENOMARKERS and NTU Exposomics, 2013, Taiwan, scientific, 100, Asia
Potential impact:
Overall conclusions from the ENVIROGENOMARKERS project - potential impact and use of the project's results
The results of the ENVIROGENOMARKERS project provide strong support for the power of omics technologies as tools for the discovery of new biomarkers of environmental exposures, prospective disease risks and links between exposure and disease. In particular, the project has shown that strong transcriptomic, epigenomic and proteomic signals predicting the occurrence of future BCLL can be detected in blood samples collected up to 12 years prior to clinical diagnosis of the disease, providing a possibility for the development of novel markers for the early disease diagnosis. Furthermore, omic profiles associating with exposure to specific environmental toxins have been discovered, which may serve as novel biomarkers of exposure. Many of these profiles shed light on the potential biological effects of these exposures, thus facilitating assessment of toxic risks associated with them and providing evidence of possible etiologic links with disease.
Analogous results have been obtained in relation to omic profiles obtained in cord blood and birth or neurodevelopmental effects.
Importantly, the project has demonstrated that omics technologies can be exploited in the analysis of biosamples already in long-term storage in existing biorepositories even if they were collected decades ago and without some of the precautions currently considered necessary, e.g. the use of RNA preservative. This opens the way for the exploitation in the context of environmental health research of millions of biosamples currently in storage throughout the world.
Project website: http://www.envirogenomarkers.net
Dissemination officer:
Dr Maria Botsivali,
National Hellenic Research Foundation,
Institute of Biology, Medicinal Chemistry and Biotechnology,
48 Vassileos Constantinou Avenue,
Athens 11635,
Greece
Tel: +30-210-7273740
Fax: +30-210-7273677
Email: mbotsi@eie.gr
Initially, the project developed methodology making possible the transcriptomics analysis of buffy coats stored for more than a decade without RNA preservative sand established cut-off criteria for the selection of biobank samples with minimum impact of the sample collection and storage conditions on transcriptomics, epigenomics, metabolomics and proteomics profiles. Subsequently the project generated the corresponding omic profiles from large numbers of human biosamples (blood fractions) in the context of prospective nested case-control studies to look for biomarkers predictive of future risks of breast cancer or non-Hodgkin's lymphoma as well as from samples collected at birth (cord blood) or during the first four years of life to look for biomarkers predictive of birth outcomes and developmental effects. In addition, measurements of chemical-specific biomarkers (including persistant organic chemicals, cadmium, phthalates) were conducted in order to look for omics-based biomarkers of exposure.
In most types of omics profiles obtained, large numbers of signals were found which associate strongly with specific environmental exposures as well as with disease risks (mainly B-cell chronic lymphatic leukemia in adults, preterm birth and neurodevelopmental effects in children). Bioinformatics analyses provide strong support for the biological relevance of these signals, demonstrating the potential of omics for the discovery of new biomarkers. Importantly, in some cases evidence of overlap between omics signals associating with exposure and disease risk was obtained, implying possible etiological links between exposure and disease.
Project context and objectives:
During the past few decades, extensive research has been conducted on the use of biomarkers of exposure, early effects and individual susceptibility in the study of the environmental aetiology of chronic diseases such as cancer. More recently, the revolutionary developments in the area of genomics have opened exciting opportunities for the discovery of a new generation of biomarkers to study this problem. Yet, exploitation of -omics technologies has so far been limited. The objective of the ENVIROGENOMARKERS project was to move this field forward through the utilisation of a range of -omics technologies in the context of relatively large, population-based molecular epidemiology studies to discover novel biomarkers of exposure and disease risk and use them to examine exposure-disease links.
Omics technologies and their potential in environmental health research
The term 'omics' refers to a collection of technologies which have emerged during the past decade or so and which permit the simultaneous, quantitative measurement of the totality, or a large fraction, of cellular features belonging to particular categories. For example, transcriptomics permits the measurement of the levels of expression of all genes through the measurement of the abundance of the corresponding RNA species; proteomics the measurement of large numbers of proteins; metabolomics, the measurement of hundreds or thousands of small molecules (metabolites); epigenomics the measurement of large numbers of epigenetic markers of different types. By making possible the simultaneous and untargeted search of enormous numbers of potential targets, -omics technologies provide unique opportunities for the discovery of biomarkers of exposure to environmental toxins as well as of biomarkers capable of predicting disease risk. Furthermore, the integration, utilising advanced bioinformatics and systems biology methods, of global profiling data measured at multiple biological levels (gene and protein expression, cell metabolism, and genetic and epigenetic status) provides exciting new opportunities for the development of a holistic view of the organism's responses under the influence of environmental stresses and progressing disease.
The objectives and structure of ENVIROGENOMARKERS
ENVIROGENOMARKERS is the first project in Europe (and one of very few worldwide) to conduct a large-scale application of a range of -omics technologies in a population study aiming at:
a) the discovery and validation of novel biomarkers predictive of increased risks of chronic diseases in which the environment may play an important role (breast cancer, non-Hodgkin's lymphoma - NHL, neurodevelopment and other childhood conditions);
b) the exploration of the association of such risk biomarkers with environmental exposures, including high-priority pollutants (carcinogens and immunotoxicants such as PCBs and PAHs, neurotoxicants such as cadmium, lead and ambient air pollution) and emerging exposures (such as phthalates and brominated flame retardants), many of which are also endocrine disruptors; and
c) the discovery and validation of biomarkers of exposure to the above and other high-priority environmental exposures (e.g. water disinfection byproducts).
The project was designed as a case-control study nested within three existing European prospective cohorts which contain biosamples collected prior to disease diagnosis, dietary, exposure and life style information and follow-up health histories: The Northern Sweden health and disease study, EPIC-Italy, and the Rhea mother-child cohort on the island of Crete.
i) The Northern Sweden health and disease study (NSHDS) has collected blood samples and exposure / health data and by January 2008 it contained 95 000 unique individuals with 172 600 sampling occasions. Blood samples were divided into aliquots of plasma, buffy coat and erythrocytes and stored in -80 oC freezers within 2 hours of collection. Lifestyle and environmental information was collected, while health status follow-up was conducted via different registries.
ii) EPIC-Italy is part of the European investigation into cancer and nutrition (EPIC) study, a prospective study, initiated in 1992 and recruiting subjects aged 35 - 70 years, aiming to investigate the relationship between diet, lifestyle and environmental factors and the incidence of cancer. EPIC-Italy represents approximately 47 700 subjects and, in addition to biosamples, includes detailed questionnaire-based information at enrolment on diet, lifestyle and personal history. Biological samples, including plasma, serum, white blood cells and erythrocytes were collected and have been stored in liquid nitrogen. The time interval from collection to -80 oC freezing was generally 2 - 5 h. Study subjects are followed up identify newly diagnosed cases of cancer at all sites and other outcomes of interest.
The Rhea mother-child cohort in Crete is a population-based cohort which enrolled mothers at the third month of pregnancy and followed them up until birth and after birth together with their children. By 2008 it had recruited 1700 pregnant mothers and nearly 600 children had been born. Data available include various maternal exposures during pregnancy (FFQ, exposure to dietary carcinogens and immunotoxins, air and water pollution) as well as various biomarkers of exposure and disease risk measured in cord blood, thus enabling the examination of the impact of in utero exposures on development and health in later life. Samples available include maternal (and in some cases paternal) blood and urine obtained during pregnancy and serum, buffy coat, erythrocytes from cord blood. The children are being followed up and subjected to clinical developmental tests at different ages.
In all cases, samples (serum, buffy coat for extraction of RNA and DNA, erythrocytes) collected 2 - 15 years prior to disease diagnosis in the case of the cancer studies, and corresponding samples collected at birth (cord blood) and at 2 and 4 years of age in the case of the children's study, were analysed for chemical-specific biomarkers of exposure as well as for omic profiles using the technologies mentioned. Subsequently advanced statistics was employed to look for features in the omic profiles correlating with:
a) disease risk, by comparing cases and controls; and
b) exposure to the chemicals of interest, by comparing subjects with different exposure levels.
Comparison of the lists of hits in each case allowed the search for biomarkers associated with both exposure and disease risk, thus providing evidence of aetiological links between the two ('meet-in-the-middle').
Omic technologies employed include microarray-based transcriptomics (whole-genome expression profiling using the Agilent 4x44k platform) and epigenomics (DNA methylation at more than 450 000 CpG sites using the Illumina HumanMethylation450 platform), metabolomics (ultraprecision liquid chromatography combined with mass spectrometry - UPLC/MS-MS) and wide-target proteomics (measurement of tens of immune-related proteins using the Luminex LabMAP technology). Importantly, because for the first time ever extensive use was made within ENVIROGENOMARKERS of biosamples stored in biobanks constructed prior to the emergence of omics and, hence, without the precautions nowadays used in relation to sample handling and storage (especially in relation to the preservation of RNA), the project also included a systematic technical validation of genomic technology's use with such samples.
Exposure assessment utilised chemical-specific biomarkers of exposure, food-frequency questionnaires, modelling and geographic information systems-technology. The main chemicals targeted in the project, and the corresponding biomarkers employed, were the following:
- persistent organic pollutants, including PCBs (118, 138, 153, 156, 170 and 180), hexachlorobenzene (HCB), pesticide p,p'-DDT and it's metabolite p.p'-DDE and the polybrominated dipheynyl ether BDE-47, measured in serum using GC-MS/MS;
- cadmium, measured in erythrocytes (ICP-MS);
- phthalates (children's study), 8 metabolites measured in urine using LC-MS/MS.
A plethora of additional data on exposure, biomarkers (including genome-wide SNP data) and health indices, available through other studies were also utilised.
Project results:
Main scientific and technological (S&T) results
a) Pilot study
The application of omics technologies in epidemiological studies raises certain practical issues of sample suitability, especially in relation to RNA quality for transcriptomics analysis, requiring that care be taken for blood samples to be collected and stored in the presence of RNA preservative. Yet, no study has evaluated systematically the influence on omic profiles of the handling and prolonged storage of blood samples and their components in these biobanks. As a first step in the project, a study was conducted to evaluate the reliability of omics data obtained from archived biosamples collected prior to the advent of omics technologies.
The study was conducted in two phases: During phase I, methods were established for the isolation of RNA of the desired quality from buffy coats isolated from blood freshly collected and processed without RNA preservative. Furthermore, the influence was examined on omics profiles of sample handling and storage-related parameters selected following scrutiny of the procedures employed at the project's partner biobanks. For this purpose, samples of fresh blood were collected from healthy volunteers using different anticoagulants (heparin, EDTA or citrate) and processed in different ways. After being allowed to stand at room temperature for different times up to 24hr ("bench-time"), the blood samples were separated into buffy coats, erythrocytes and plasma, aliquoted and stored at -80oC or in liquid nitrogen. Finally, samples representing different collection, processing and storage conditions were analysed on the different omics platforms employed in the project.
The results of phase I were used to establish minimum criteria that samples must satisfy in order to be suitable for reliable omics analysis. During phase II, historic samples from the partner biobanks, satisfying these criteria, were analysed so as to evaluate the influence of long-term storage.
RNA and DNA isolation
Several attempts have been made to extract RNA and DNA simultaneously; however, none of the methodologies applied were found to be applicable. Therefore it was decided that separate samples are needed for DNA and RNA extraction.
Transcriptomics-quality RNA could be isolated from phase I buffy coats frozen within eight hours of blood collection by thawing the samples fully immersed in RNAlater or Qiazol followed by immediate extraction of RNA. Application of the same method to samples obtained from the project's partner biobanks resulted in the isolation in good yield of high-quality RNA from most samples.
RNA yield and RIN for a set of EPIC samples stored in nitrogen (1 - 27) or at -80 oC (28 - 33)
Omics analysis of phase I samples showed that the main variable affecting the main parameters affecting the observed profiles were the subject (all omic platforms), anticoagulant (metabolomics, proteomics, transcriptomics) and bench-time (within two hours for transcriptomics, beyond eight hours for metabolomics and transcriptomics). Based on this, cut-off criteria ensuring minimum impact on biobank samples of processing conditions were defined as bench-time < 2 hours and use of the same anticoagulant.
To evaluate the impact of storage time on omics profiles of samples stored in the project biobanks, samples from a set of 31 subjects from each biobank were selected, coming only from healthy female donors and from the same collection centre per biobank to minimise the influence of other parameters. Their storage time prior to analysis was 13 - 17 years, while the collection-to-storage times for the EPIC-Italy sub-set were 100 - 198 min. No significant effect of storage time or collection-to-storage time could be detected for any of the omic profiles.
The overall conclusion of the pilot study is that a large fraction of the human blood-derived samples in storage in the project's biobanks is amenable to analysis using the omics technologies of the project, on condition that the time between blood collection and fractionation did not exceed 8 hr (4 hr for proteomics). These findings open the way to the application of omics technologies to biosamples collected over previous decades in the context of population-based or disease-oriented cohorts. The results of the pilot study have been published (Hebels et al., Performance in Omics Analyses of Blood Samples in Long-Term Storage: Opportunities for the Exploitation of Existing Biobanks in Environmental Health Research. Environ Health Perspect. 5 Feb. [Epub ahead of print] PubMed PMID:23384616]).
b) Cancer studies
Subjects for inclusion in the study were selected from the partner biobanks on the basis of the cases having a date of diagnosis 2 - 10 years after recruitment for breast cancer and 2 - 15 years for NHL. Controls were matched on the basis of sex, age, collection centre and years in storage. Based on these criteria and availability at the two biobanks, the following numbers of case-control pairs, per disease and cohort, from whom samples were to be analysed, were decided (totalling 900 pairs):
Cohort - case - control pairs
breast cancer - NSHDS - 325
EPIC - Italy - 294
NHL - NSHDS - 197
EPIC - Italy - 84
Initially full omic analyses were conducted on samples from 100 case-control pairs for each disease and the resulting omic profiles were analysed in terms of their relationships with disease status and exposures. For the statistical analysis, a mixed model was defined which included a random effects component (e.g. dates of RNA isolation, labelling and hybridisation for transcriptomics) which dealt effectively with nuisance variation. Data analysis using this model showed that, while a small number of transcriptomics, proteomics and epigenomics signals showed some association with disease risk (especially for NHL) and exposure, none reached statistical significance sufficient to survive multiple testing criteria.
From these results it was concluded that, while some hits appeared to emerge, no reliable selection of candidate biomarkers, to be subsequently validated through targeted analysis of larger numbers of samples, could be made. On the basis of this conclusion, the strategic decision was made to cancel the plans for targeted analysis and, instead, increase the number of samples on which to conduct full omics analysis so as to enhance the statistical power of the study. Furthermore, given that, in general, the significance of the top hits was better for NHL than for breast cancer, it was decided to focus the rest of the study on NHL by conducting full omics analyses on all available NHL samples, thus bringing the total number thus analysed from 100 pairs to 281 pairs (197 for NSHDS and 84 for EPIC Italy). After completion of these analyses, multiple statistical models and methodologies were used to evaluate to resulting data, as previously using mixed models to minimise the impact of nuisance (technical) variation. The outcome, summarised below, amply justified the choice made:
Transcriptomics:
Even with the strictest statistical criteria (Bonferoni), a number of signals associating with NHL risk, and some tens of signals associating with exposure to a number of PCBs (but not cadmium or lead), were found.
Numbers of transcriptomics signals associating with exposures or NHL risk at different statistical significance levels
As regards disease risk, using a (strict) criterion of FWER = 1 % and ENT=10 000, 12 significant signals were observed, almost all of which associate very strongly and exclusively with B-cell chronic lymphatic leukemia (BCLL). When the analysis was restricted to specific subtypes, more than 600 signals were found to associate significantly with BCLL risk and only approx. 10 to associate much more weakly with multiple myeloma (MM) or diffuse large B-cell lymphoma (DLBCL). Analysis by time-to-disease (TtD) showed approximately equal numbers of signals being statistically significant at TtD < 6 and TtD > 6, with the former category showing clearer patterns and a tendency to increase in expression while getting closer to disease.
Pathway analysis of the signals associating with BCLL revealed that the affected pathways were almost uniformly involved in lymphocyte-specific signalling such as B-cell receptor signalling and immune response and included some highly biologically relevant pathways such as aberrant B-cell receptor signalling which has been well documented for BCLL.
Pathway analysis of transcriptomics signals associating with BCLL
Turning to associations of transcriptomics signals with exposure, some tens of signals were found for a number of PCBs using stringent statistical criteria.
Transcriptomics signals associating with exposures at different levels of statistical stringency
Initial efforts at pathway analysis revealed only 1 pathway for most PCBs, namely 'calcineurin-regulated NFAT-dependent transcription in lymphocytes'. This may be a biologically relevant finding since expression of this gene is known to be mediated through the Ah receptor which is a target of some PCBs, however further work is required to evaluate fully this finding.
Finally, attempts have been made to search for 'meet-in-the-middle' biomarkers by looking for overlaps between the lists of signals (expanded by somewhat relaxing the statistical criteria) which associate with both disease risk and exposures. A few weak candidates have been found linking BCLL with cadmium and PCB-156, which require further evaluation.
Proteomics:
A total of 32 immune-related proteins were measured, of which 4 were excluded owing to their low levels. Of the remaining 28 signals which were included in the statistical analyses, 2 showed a statistically significant association with lymphoma risk in all subjects as well as in subjects with TtD both < and > 6 years. Analysis by sub-type revealed strong signals only for multiple myeloma, with 6 proteins showing downregulation mostly at TtD < 6 years.
Proteomics signals associating with multiple myeloma. Each column of symbols represents one protein signal measured. Vertical axis: -logp; dashed line: limit of statistical significance (Bonferoni correction); blue symbols: TtD < 6; red symbols: TtD > 6.
As regards exposure biomarkers, numerous significant associations were found for PCBs (mostly downregulation) and non-PCB chlorinated compounds (DDE, hexachlorobenzene; mostly upregulation) and fewer associations with cadmium. Some of these overlapped with signals associating with disease and are therefore possible meet-in-the-middle biomarkers, but further work is required to evaluate this possibility.
Proteomics signals associating with exposures
Metabolomics:
All samples from the subjects originally selected from the two cohorts (totalling 900 case-control pairs) were subjected to full UPLS-MS/MS metabolomics analysis. After quality evaluation and clean-up of the resulting datasets, statistical analysis did not reveal any signals associating with risk for either disease at the desired statistical significance. Small numbers of signals showed weak association (p < 0.05) with disease risk, 4 for NHL and 24 for breast cancer in NSHDS and 0 for NHL and 21 for breast cancer in EPIC Italy; however, there was no overlap between the two cohorts. Furthermore there was no strong association of the NHL hits with any particular lymphoma subtype. It was therefore concluded that no viable metabolomics biomarkers predictive for disease risk were found.
Further analysis of the metabolomics dataset in relation with other parameters suggests weakly significant associations with exposure to PCB156 as well as with sex, body burden, alcohol consumption and vegetable consumption, however these associations are substantially below the desired level of statistical stringency.
d) Epigenomics
Large numbers of signals significantly associating with NHL risk were found. As for transcriptomics, these signals clustered mainly with BCLL.
As regards epigenomics biomarkers of exposure, substantial numbers of signals were found to associate significantly with exposure to hexachlorobenzene (HCB) and, to a smaller degree, dioxin-like PCBs. GO term analysis of the genes associated with HCB exposure shows a substantial number of terms related to immune function, suggesting that HCB may alter the latter. Pathway analysis shows activation of multiple cancer-related pathways, with EGFR signalling being the most significant pathway.
c) Children's study
In view of the experience gained from the cancer study in relation with the need to maximise the statistical power of the project by conducting full omics analysis of the maximum possible number of samples, it was decided to modify the initial plans and
a) focus primarily on neurodevelopment, rather than multiple developmental end-points; and
b) instead of having a discovery phase using omics analyses of a limited number of samples and a validation phase based on targeted analyses of larger numbers of samples, expand the number of samples on which full omics would be conducted while cancelling the validation phase.
The final arrangement adopted involved clinical testing, focussing on neurodevelopment, on children at ages two and four years, and omics analysis of samples from cord blood (so as to relate omic profiles to preceding in-utero exposures and future neurodevelopmental risks) and from blood collected at four years (to search for biomarkers associated with the clinical state and follow their development since birth). Taking into account the availability of cord blood material, pairs of cord- and 4-year-old-blood samples from a total of 118 children were subjected to full omics analysis. On the other hand, clinical evaluation for neurodevelopment and other clinical end-points, serum metabolomics and measurement of biomarkers of exposure (serum POPs, urine phthalates), were conducted on a total of approx. 600 children. Additional exposures for which data were available for all children include water disinfection byproducts, air pollution, biomarkers and estimates of exposure to dietary carcinogens, DR Calux and maternal smoking and passive smoking. Other data available include maternal psychological status, gestational hypertension, diabetes and metabolic symptoms, perinatal outcomes, cord leptin concentration, obesity, growth patterns (all, from birth onwards), depression (mother, pregnancy and postpartum), asthma / wheeze / allergies / atopic dermatitis / FeNO (first year and four years), lung function (fourth year), metabolic syndrome (mother, pregnancy), reproductive outcomes, birth weight and gestational age, anogenital distance, genotoxic outcomes-micronuclei (at birth and four years), thyroid function (mother, children), child cardiological condition (four years).
Transcriptomics
To look for biomarkers predictive of neurodevelopmental performance, buffy coats from cord blood were subjected to transcriptomics. After quality control and pre-processing, 44 019 probes were included in the statistical evaluation of correlation of expression levels versus the 6 neurodevelopment scores at age 4.
Bioinformatics analysis showed a number of pathways to be overrepresented among most of these gene lists; however, no striking links with developmental processes or nervous system function were obvious.
Metabolomics
For the children's study metabolomics analysis was conducted using a targeted approach based on the Biocrates p180 IDQ platform which provides information on 180 selected metabolites including 20 aminoacids, 18 biogenic amines, 40 acylcarnitines, 90 glycerophospholipids, 15 sphingolipids and hexose. PCA analysis showed a clear separation of the three types of samples (maternal, cord, four yearrs) analysed. Statistically significant differences between the sample types were observed for a number of aminoacids, glycerophospholipids and acylcarnitines.
PCA plot of metabolomics profiles of sera from maternal, cord and four-year old children's blood
Evaluation of the correlation between metabolomic profiles in serum and indices of neurodevelopment showed typically 10 - 20 statistical significant (Pearson p < 0.05) features for cord serum and 1 - 5 features in four-years serum. The metabolites concerned included carnitines, phosphatidylcholines, sphingolipids,sphingomyelins and symmertrical dimethylarginine. When each neurodevelopment variable was divided in two categories (below median and above median) and a combined measure of neurodevelopment scales was calculated by counting the number of 'high' category values across all six scales, moderately significant associations were found for four metabolites in four-years serum (a carnitine, tyrosine, valine and a phosphatidylcholine.
Significant differences in the levels of specific metabolites were observed between sera of mothers with normal and preterm births, including glycerophospholipids and aminoacids. As regards metabolomic profile of cord blood and birth outcomes, a number of metabolites were found to associate with preterm birth, gestational age, fetal weight growth restriction. Finally, significant associations were also found for small numbers of metabolites in sera collected at four-years with maternal passive or active smoking during pregnancy.
Epigenomics
Evaluation of the epigenomics profiles obtained in cord blood revealed a significant number of signals associating with different indices of neurodevelopment at four years.
There was minimal overlap between the hits observed for the different neurodevelopmental indices. No obvious connection of the hits observed with neurodevelopment-related biological features was noted. Strikingly, the association with the same neurodevelopment indices of epigenomics signals observed in samples collected at 4 years of age was much weaker, with signals significant at FDR < 0.05 being observed only for motor (536 hits). This is surprising in view of the fact that for this index no hits were found in the cord blood samples. Statistically weaker (FDR < 0.15) signals were observed in 4-year blood samples to associate with perceptual (15 hits) and cognitive (4 hits).
As regards exposure, significant numbers of signals were observed mainly in cord blood samples associating at FDR < 0.15 mainly with hexaclorobenzene (29 hits) and dioxin-like PCBs (126 hits), with no overlap with hits related to neurodevelopment.
Other exposures
As for the cancer studies, in addition to the main target exposures described above, exposure information generated through other projects has been collected to be utilised for the evaluation of its impact on omics profiles. This includes extensive information on exposure to air pollution including PM10, PM2.5 coarse particles, NO2, NO (ESCAPE project), disinfection-by-products (HiWate project), dietary carcinogens and immunotoxins including acrylamide, PAHs, N-nitroso compounds, dioxins and others, micronucleus data in cord four-year blood as well as detailed dietary intakes through FFQs (NewGeneris project), heavy metals, cotinine, lipids etc. These data have been added to the ENVIROGENOMARKERS dataset and will be utilised in the analysis of omics profiles.
d) Chemical-specific biomarkers of exposures versus disease risks
? detailed evaluation was conducted of the relationship between biomarkers of exposure (POPs, cadmium, lead) and disease risk for both breast cancer and lymphoma and in both cohorts. No evidence of an association between exposure to PCBs, DDE HCB or cadmium and overall risk of NHL was found, although weak evidence of positive associations for cadmium and NHL in females, and for metals / POPs and Follicular lymphoma, was observed. On the other hand, weak but consistent unexplained inverse associations with PCBs, DDE and HCB in males were observed. As regards breast cancer, unexplained, consistent evidence of a significant inverse association was observed between breast cancer and cadmium, HCB and DDE.
e) Exposures
The environmental exposures targeted by the project were estimated using biomarkers measured in archived or freshly collected material. The ranges of exposure levels observed are summarised below:
- Cadmium: Levels of cadmium measured in archived erythrocytes from the EPIC-Italy and NSHDS biobanks were obtained in the same analysis as measurement of lead levels.
- Persistant organic hydrocarbons: Levels of PCBs, DDT, DDE, hexachlorobezne and brominated diphenyl ether 47 were measured in archived sera in the adult cohorts as well as from pregnant mothers in the Rhea cohort.
- Phthalate metabolites: The levels of seven phthalate metabolites were measured in urine collected from pregnant mothers and children at age two years, as well as in urine of fathers.
Use and dissemination of foreground
Dissemination measures and plans
The project's activities and results have been publicised and disseminated using the following means:
- The website http://www.envirogenomarkers.net where fundamental information about the project's objectives, structure and partners was provided, along with updates of ongoing activities. During the period August 2009 - April 2013 the website received a total of 2227 unique visitors, corresponding to over 50 new visitors per month. During the last year of the project's formal duration (March 2012 - February 2013) the website was visited by a total of 824 visitors, corresponding to more than 68 new visitors per month.
- Production of a press release at the time of the project's start, as well as of an electronic brochure and two newsletters, which were circulated to hundreds of recipients and are also available for downloading from the project's website.
- Soon after its initiation, the project was highlighted in a feature in Nature magazine titled 'Prevention by numbers' (Nature, 458, 9 April 2009, 792 - 793) in which research efforts in the area of environmental health ongoing in the United States (US) and Europe were presented.
- Publication of the main scientific findings of the project in peer-reviewed international journals. So far 11 papers have been published (uploaded on ECAS), while a number of additional papers have been submitted or are in preparation. A publication plan has been put in place which foresees that some dozens of publications will be produced within two years of the project's formal completion. By extending the validity of their consortium agreement by 24 months the partners have committed themselves to fulfilling this plan.
- Presentation of the project and its results at a large number of international and national conferences targeting different audiences, including the research community, experts and authorities active in human biomonitoring, exposure assessment, chemicals risk assessment etc. In addition the project consortium maintained regular contact with other European projects in the area of environmental health, being presented at a number of their meetings (e.g. Cophes, ECNIS, NewGeneris, Exposomics).
List of scientific (peer reviewed) publications
1. Performance in Omics Analyses of Blood Samples in Long-Term Storage: Opportunities for the Exploitation of Existing Biobanks in Environmental Health Research, Hebels D.G.A.J. Envir. Hlth Perspectives, 121 (4), April 2013, NIEHS, US, 2013, 480-487, doi: dx.doi.org/10.1289/ehp.1205657
2. Metabolic profiling detects early effects of environmental and lifestyle exposure to cadmium in a human population, Ellis J. K., BMC Medicine, 19 June 2012, BMC, US, 2012, doi: 10.1186/1741-7015-10-61)
3. Effect of high doses of folic acid supplementation in early pregnancy on child neurodevelopment at 18 months of age: the mother-child cohort 'Rhea' study in Crete, Greece, Chatzi L., Public Health Nutrition, 15 (9), February 2012, Cambridge University Press, UK, 2012, 1728 - 1736, doi: 10.1017/S1368980012000067
4. Chiral Metabonomics:1H NMR-Based Enantiospecific Differentiation of Metabolites in Human Urine via Direct Cosolvation withß-Cyclodextrin, Pe´rez-Trujillo M., Anal. Chem., 84, American Chemical Society, US, 2012, 2868 - 2874, doi: 10.1021/ac203291d
5. Meeting-in-the-middle using metabolic profiling - a strategy for the identification of intermediate biomarkers in cohort studies, Chadeau-Hyam M., Biomarkers, 16 (1), Informa Healthcare, UK, 2011, 83 - 88, doi: 10.3109/1354750X.2010.533285
6. Integrating biomarkers into molecular epidemiological studies , Vineis P., Curr. Opin. Oncol., 23, Wolters Kluwer Health | Lippincott Williams & Wilkins, US, 2011, 100 - 105, doi: 10.1097/CCO.0b013e3283412de0
7. ESS++: a C++ objected-oriented algorithm for Bayesian stochastic search model exploration, Bottolo L., Bioinformatics, 27 (4), Oxford University Press, UK, 2011, 587 - 588, doi: 10.1093/bioinformatics/btq684
8. Advancing the application of omics-based biomarkers in environmental epidemiology, Vineis P., Environmental Molecular Mutagenesis, 21 March 2013, EMS, US, 2013, doi: 10.1002/em.21764
9. Methodological challenges for the derivation of -omics based biomarkers, Chadeau- Hyam M., Environmental Molecular Mutagenesis, in press, EMS, US, 2013
10. Making sense of OMICS data in population-based environmental health studies, Kyrtopoulos S. A., Environmental Molecular Mutagenesis, in press, EMS, US, 2013, doi: 10.1002/em.21778
11. A Composite Framework for the Statistical Analysis of Epidemiological DNA Methylation Data with the Infinium Human Methylation 450K BeadChip, Valavanis I., IEEJ Biomed Health Informatics, in press, Institute of Electrical and Electronics Engineers, UK, 2013
List of dissemination activitiesIST OF DISSEMINATION ACTIVITIES
(No., type of activities, main leader, title, date, place, type of audience, size of audience, countries addressed)
1. Website, Botsivali M., ENVIROGENOMARKERS , 2009-ongoing, http://www.envirogenomarkers.net all, global, on average 50 unique visitors per month, global
2. Electronic brochure, Botsivali M., ENVIROGENOMARKERS, 2009, available on project website and circulated by email, scientific, industry, policy, media, 200, global
3. Press release for CORDIS, Botsivali M., ENVIROGENOMARKERS, 2009, https://cordis.europa.eu/article/id/114224-envirogenomarkers-new-generation-of-biomarkers-to-study-environmental-agents-in-human-disease scientific, industry, policy, media, large, global
4. Nature magazine, news item, Kyrtopoulos S. A., 'Prevention by numbers', 9 April 2009, Nature, Vol. 458, pp 792 - 793, scientific, industry, policy, media, large, global
5. Press release, Botsivali M., ENVIROGENOMARKERS, 2009, European Press Agency, scientific, industry, policy, media, large, global
6. Conference, Keun H., International Conference on Environmental Mutagens, 2009, Florence, scientific, industry, policy, media, 250, Europe
7. Conference, Kyrtopoulos S. A., International Conference on Environmental Mutagens, 2009, Florence, scientific, industry, policy, media, 250, Europe
8. Conference, Kyrtopoulos S. A. , Workshop on Genomics in Cancer Risk Assessment, 2009,Venice, scientific, industry, policy, 100, Europe, US
9. Workshop, Kyrtopoulos S. A., American Association for Cancer Research, 2010, Miami, scientific, industry, policy, 200, US, Europe, Asia
10. Project factsheet in book, Botsivali M., ENVIROGENOMARKERS, 2010 , European research on environment and health funded by FP7, vol. 1, scientific, industry, policy, media, large, global
11. Electronic newsletter 1, Botsivali M., ENVIROGENOMARKERS , 2010, available on project website and circulated by email, scientific, industry, policy, media, 200, global
12. Conference, Keun H. , UK Environmental Mutagenesis Society Meeting, July 2010, United Kingdom, scientific, industry, policy, media, 200, United Kingdom
13. Conference, Keun H., American Association for Cancer Research, April 2010, Washington DC, scientific, industry, policy, media, 250, US, international
14. Conference, Kyrtopoulos S. A., European Environmental Mutagen Society, 2010, Oslo, scientific, industry, policy, media, 250, Europe
15. Conference, Kyrtopoulos S. A., International Union of Toxicology, 2010, Barcelona, scientific, industry, policy, media, 500, Europe, US, Asia
16. Conference, Kyrtopoulos S. A., ICCA-LRI and JRC Workshop, 2010, Stressa, Italy, industry, policy, scientific, media, 200, Europe, US
17. Conference, Botsivali M., European Environment and Health Conference, 2010, Brussels, policy, media, scientific, industry, 150, Europe
18. Conference, Kyrtopoulos S. A., European Environmental Mutagen Society, 2011, Barcelona, scientific, industry, policy, media, 250, Europe
19. Conference, Keun H. , American Chemistry Council / ICCA Long Range Initiative Meeting, 2011, Quebec, scientific, industry, policy, media, 500, Canada, US, Japan, Europe
20. Conference, Kyrtopoulos S. A., International Society for Environmental Epidemiology, 2011, Barcelona, scientific, industry, policy, media, 400, Europe, US, Asia
21. Conference, Vermeulen R. C., International Society for Environmental Epidemiology, 2011, Barcelona, scientific, industry, policy, media, 400, Europe, US, Asia
22. Workshop, Georgiadis P., Design of Future Molecular Epidemiology Studies and New Biomarkers, 2012, London, scientific, 100, Europe
23. Conference, Kyrtopoulos S. A., Human Biomonitoring - Linking Environment to Health and Supporting Policy, 2012, Larnaca, Cyprus, scientific, policy, 100, Europe
24. Conference, Chatziioannou A., 12th Intl Conf on BioInformatics and BioEngineering, 2012, Cyprus, scientific community, hundreds, European and other
25. Poster, de Kok T. M., Society of Toxicology, 2012, San Fransisco, scientific, large, global
26. Lecture, de Kok T. M., Molecular Epidemiology and Prevention, 11 December 2012, Masters program Biomedical Sciences Maastricht University (the Netherlands), scientific, 80, the Netherlands, Denmark, Bulgaria, Finland, United Kingdom, US, Estonia, China
27. Lecture / workshop, de Kok T. M., Genomics biomarkers and their application in molecular epidemiology, 17 April 2012, International Course on Molecular Epidemiology, Maastricht, the Netherlands, scientific, 25, multiple EU
28. Lecture / workshop, de Kok T. M., International Course on Molecular Epidemiology, 25 April 2012, Valkenburg, the Netherlands, scientific, large, the Netherlands, Estonia, Spain, Bulgaria, Denmark, Ireland, United Kingdom
29. Workshop, Keun H., Nutritional Genomics Annual Meeting, August 2012, Helsinki, scientific, 200, Europe-wide
30. Conference, Keun H. , European Association for Cancer Research, July 2012, Barcelona, scientific, industry, policy, media, 1000, Europe
31. Conference, Keun H., Frontiers in Cancer Nutrition, February 2012, Arizona Cancer Centre, scientific, policy, media, 500, US
32. Conference, Keun H., International Society for the Study of Xenobiotics, June 2012, Amsterdam, the Netherlands, scientific, industry, policy, media, 500, Europe
33. Conference, Keun H. , EUROTOX - Congress of the European Societies of Toxicology, June 2012, Stockholm, Sweden, scientific, industry, policy, media, 50, Europe
34. Conference, Keun H. , Hong Kong Society of Clinical Chemistry, January 2012, Hong Kong, scientific, policy, media, 500, China
35. Conference, Valavanis I., 12th International Conference on BioInformatics and BioEngineering, 2012, Cyprus, scientific, 2012, global
36. Lecture, de Kok T. M., Application of genomics in population studies, 9 May 2013, ECNIS2 course Molecular Epidemiology, Lódz, Poland, scientific, 20, multiple EU
37. Electronic newsletter 2, Botsivali M., ENVIROGENOMARKERS, 2013, available on project website and circulated by email, scientific, industry, policy, media, 200, global
38. Lecture, de Kok T. M., Symposium on ENVIROGENOMARKERS and NTU Exposomics, 2013, Taiwan, Scientific, 100, Asia
39. Conference, Vineis P., Omics in Epidemiology, 22 March 2013, Lyon Canceropole, Clinicians, 200, France
40. Conference, Vineis P., Omics in Epidemiology, 18 April 2013, University of Maastricht, Molecular biologists, 30, the Netherlands
41. Workshop, Vineis P., The exposome: the example of EGM, 1 September 2013, Interlaken, EUROTOX, toxicologists, 100, Europe
42. Workshop, Keun H., Exposome research, April 2013, College of Public Health, National Taiwan University, scientific, 100, Taiwan
43. Conference, Kyrtopoulos S. A., Symposium on ENVIROGENOMARKERS and NTU Exposomics, 2013, Taiwan, scientific, 100, Asia
Potential impact:
Overall conclusions from the ENVIROGENOMARKERS project - potential impact and use of the project's results
The results of the ENVIROGENOMARKERS project provide strong support for the power of omics technologies as tools for the discovery of new biomarkers of environmental exposures, prospective disease risks and links between exposure and disease. In particular, the project has shown that strong transcriptomic, epigenomic and proteomic signals predicting the occurrence of future BCLL can be detected in blood samples collected up to 12 years prior to clinical diagnosis of the disease, providing a possibility for the development of novel markers for the early disease diagnosis. Furthermore, omic profiles associating with exposure to specific environmental toxins have been discovered, which may serve as novel biomarkers of exposure. Many of these profiles shed light on the potential biological effects of these exposures, thus facilitating assessment of toxic risks associated with them and providing evidence of possible etiologic links with disease.
Analogous results have been obtained in relation to omic profiles obtained in cord blood and birth or neurodevelopmental effects.
Importantly, the project has demonstrated that omics technologies can be exploited in the analysis of biosamples already in long-term storage in existing biorepositories even if they were collected decades ago and without some of the precautions currently considered necessary, e.g. the use of RNA preservative. This opens the way for the exploitation in the context of environmental health research of millions of biosamples currently in storage throughout the world.
Project website: http://www.envirogenomarkers.net
Dissemination officer:
Dr Maria Botsivali,
National Hellenic Research Foundation,
Institute of Biology, Medicinal Chemistry and Biotechnology,
48 Vassileos Constantinou Avenue,
Athens 11635,
Greece
Tel: +30-210-7273740
Fax: +30-210-7273677
Email: mbotsi@eie.gr