DoReMi - Low Dose Research towards Multidisciplinary Integration

Transcription

1 DoReMi - Low Dose Research towards Multidisciplinary Integration Deliverable D5.17: Report for an integrated (biology-dosimetry-epidemiology) research project on occupational Uranium exposure (Task 5.8 CURE Final report) Due date: Month 60 Actual submission date: Month 63 (2 March 2015) Status: Final Nature Report Dissemination RR = Restricted to a group specified by the consortium (including the level Commission Services) Lead beneficiary IRSN Olivier LAURENT (IRSN), Maria GOMOLKA (BfS), Richard HAYLOCK (PHE), Eric BLANCHARDON (IRSN), Augusto GIUSSANI (BfS), Will Authors ATKINSON (Nuvia), Sarah BAATOUT (SCK CEN), Derek BINGHAM (AWE), Elisabeth CARDIS (CREAL), Janet HALL (Institut Curie), Ladislav TOMASEK (SURO), Dominique LAURIER (IRSN) Sophie ANCELET, Christophe BADIE, Gary BETHEL, Jean-Marc BERTHO, Richard BULL, Cécile CHALLETON de VATHAIRE, Rupert COCKERILL, Estelle DAVESNE, Damien DRUBAY, Teni EBRAHIMIAN, Hilde ENGELS, Nora FENSKE, Michael GILLIES, James GRELLIER, Stephane GRISON, Yann GUEGUEN, Sabine HORNHARDT, Chrystelle IBANEZ, Sylwia Contributors KABACIK, Lukas KOTIK, Michaela KREUZER, Anne-Laure LE BACQ, James MARSH, Dietmar NOSSKE, Jackie O'HAGAN, Eileen PERNOT, Matthew PUNCHER, Roel QUINTENS, Estelle RAGE, Tony RIDDELL, Laurence ROY, Eric SAMSON, Maamar SOUIDI, Michelle TURNER, Nina WEILAND, Sergey ZHIVIN Approval WP5 Leader Simon Bouffler: 27 February

2 SUMMARY The health effects of chronic exposure to uranium, as it occurs in populations of workers involved in the nuclear fuel cycle and in the general population, are not well known. Experimental studies on the effect of exposure to uranium have reported impairments of the cerebral function, genotoxic, nephrotoxic and other biological effects, but the implications of these findings to human health are not clear. Beside, a few epidemiological studies reported increases in lung, lymphatic/hematopoietic cancers and cardiovascular diseases associated with occupational exposure to uranium, but most available studies are descriptive or suffer from other limitations. New studies borrowing strengths from modern biology approaches on the one hand, and from large epidemiological datasets on the other hand, will have better potential to improve the characterization of the biological and health effects of chronic exposure to low doses of uranium. The CURE (Concerted Uranium Research in Europe) project was a 18-month concerted action funded by the European Network of Excellence DoReMi ( The aim of CURE was to elaborate a collaborative research project on the biological and health effects of uranium contamination, integrating epidemiology, biology/toxicology and dosimetry. It notably aimed to evaluate the feasibility of a molecular epidemiology approach in cohorts of workers exposed to uranium. Existing cohorts of workers exposed to uranium (miners and nuclear industry employees involved in uranium processing) in Belgium, Czech Republic, France, Germany, and the United Kingdom, plus existing competences in epidemiology, biology and dosimetry among CURE partners, provided a unique opportunity to develop a coherent multidisciplinary research project within a collaborative framework. Contacts have been established with other teams involved in similar research projects within and outside the European Union (e.g.: Russia, Kazakhstan, USA). Based on the results of this concerted action, a large scale integrated collaborative project will be proposed to improve the characterization of the biological and health effects associated with uranium internal contamination in Europe. In the future, it might be envisaged to extend collaborations with other countries outside the European Union, to apply the proposed approach to other internal emitters and other exposure situations of internal contamination, and to open the reflections to other disciplines interested in the effects of internal contaminations by radionuclides. 2

3 CONTENTS 1. BACKGROUND Health effects of internal contaminations by radionuclides Uranium: a major and widespread radionuclide Evidence of uranium effects from experimental studies Evidence of uranium effects from epidemiological studies Evidence of uranium effects from molecular epidemiology Room for improvement Epidemiology Molecular epidemiology OBJECTIVES OF THE CURE PROJECT PROJECT MANAGEMENT AND TASKS COMPLETED General characteristics Partners Work packages Tasks completed PROPOSED INTEGRATED PROTOCOL Epidemiologic cohorts and subsets for analyses Cohorts for broad base analyses Cohort subsets for dose response analyses Cohort subsets proposed for molecular epidemiology Biological protocol for a molecular epidemiology study Proposed biomarkers Biospecimen to be sampled Questionnaire Information sheet and consent form Logistic strategy for the collection of biospecimens in pilot cohorts Dosimetric protocol Introduction Biokinetic and dosimetric models Parameters of exposure Monitoring data Assessment procedure Methods for statistical analyses Broad base analyses Dose-response analyses Analysis of biological information Characterization and propagation of uncertainties Introduction

4 Characterizing the best estimates and probability distributions of model parameters and other relevant uncertainty sources Testing the feasibility to account for the uncertainty inherent in estimates of dose when estimating radiation-induced disease risk Creating a synthetic cohort as a tool for testing the proposed methodologies CONCLUSIONS REFERENCES LIST OF PARTICIPANTS TO THE CURE PROJECT LIST OF INVITED EXPERTS GLOSSARY ACKNOWLEDGEMENTS ANNEXES

5 1. BACKGROUND 1.1. Health effects of internal contaminations by radionuclides The assessment of long term health effects resulting from exposure to ionizing radiation is primarily based on risk models developed from the Life Span study (LSS) of atomic bomb survivors who were exposed to external photon and neutron radiation in Hiroshima and Nagasaki in 1945 (ICRP, 2007; NRC, 2006; UNSCEAR, 2008a). By contrast, the major portion of the effective dose delivered to the general population results from radiation emitted by radionuclides located inside the body after internal contamination (UNSCEAR, 2008a). Many large populations of radiation workers are also exposed internally to alpha radiation emitted by radionuclides such as uranium or plutonium. Currently, estimates of risks associated with internal exposure to alpha radiation are derived from the risk models based on LSS data, by way of applying a radiation weighting factor (w R ) of 20 to the absorbed alpha dose to account for the experimentally observed difference in risk (per unit absorbed dose) for alpha particles as compared to photons (ICRP, 2007). However, there are concerns over the reliability and accuracy of the conversion of risk between these very different types of exposure. In 2009 the EURATOM-convened High Level and Expert Group (HLEG) considered the limited knowledge about the effects of internal contamination by radionuclides to be a key scientific issue for the purpose of radiation protection policy (HLEG, 2009). Subsequently, the effects of internal contamination have been considered as a cross-cutting issue by European platforms in the field of radiation research: MELODI ( EURADOS ( NERIS ( and ALLIANCE ( The DoReMi network of excellence supported by the European Commission ( also identified the gaps in the knowledge of the effects of internal contamination as key issues for radiation protection research. A workshop organized as part of DoReMi task 5.5 concluded that further radiobiological research on this topic as well as epidemiological studies in populations directly exposed to internal emitters are essential (Laurier et al., 2012). This research would allow the assumptions within the current radiation protection system with respect to risks from internal emitters (ICRP, 2007) to be evaluated and drive any future improvements to this system. This workshop also concluded that whenever possible, radiobiological and epidemiological research on the study of internal contamination should be integrated into the framework of molecular epidemiology studies nested within established cohorts. Uranium miners and workers were identified as populations of major interest for the characterization of risks associated with internal contamination (Laurier et al., 2012), notably to assess the validity of the radiation weighting factor for alpha emitters (e.g.: (Rage et al., 2012)) Uranium: a major and widespread radionuclide Uranium is a naturally occurring radioactive element (symbol U, atomic number 92). Natural uranium is a mixture of three isotopes: by mass, 238 U is the most abundant (99.275%), while 235 U and 234 U represent 0.72% and 0.005%, respectively. All three isotopes exhibit the same chemical properties, but have different radioactive characteristics: 238 U has low specific activity, with a half-life of 4.5x10 9 years, whereas 235 U and especially 234 U are more radioactive, with half-lives of 7.0x10 8 and 2.5x10 5 years, respectively (UNSCEAR, 2008b). In natural uranium approximately 49% of the activity is produced by 234 U and 49% by 238 U, with only 2% from 235 U. Food and drinking water are the main 5

6 sources of exposure to natural uranium isotopes among the general population (Ansoborlo et al., 2015; ATSDR, 2013). Uranium has important industrial applications since uranium ore is the base material for the production of nuclear fuel. Uranium is present in various chemical forms and isotopic compositions at all stages of the nuclear fuel cycle. This cycle includes industrial processes such as mining, milling, concentration, conversion and enrichment of uranium, fuel fabrication and reprocessing (see annex 1). Most civil reactors use enriched uranium, characterized by a higher proportion of 235 U (2 to 5%) than natural uranium (UNSCEAR, 2008b). Even higher degrees of enrichment can be encountered, for instance during the production of nuclear weapons (ATSDR, 2013). The industrial process of enrichment separates natural uranium into enriched uranium and depleted uranium, a by-product characterized by a lower percentage of 235 U (about 0.3%) than natural uranium. Depleted uranium has been used in various ways, notably as counterweights or ballast in aircraft, shielding and military applications (tank armor, munitions). Enriched uranium is more radioactive than natural uranium owing to its higher percentages of 235 U and 234 U. Conversely, depleted uranium is less radioactive than natural uranium since it contains a lower proportion of these two isotopes. Other, man-made (artificial), isotopes of uranium can also be encountered at specific stages of the fuel cycle, such as 232 U while handling reprocessed uranium (Guseva Canu et al., 2011). In summary, interest in the incorporation of uranium into the body is justified by the widespread exposure of the general population to natural uranium (Ansoborlo et al., 2015; ATSDR, 2013; Canu et al., 2011) and by the more localized exposure to depleted uranium in settings where conflicts have occurred (Krunic et al., 2005). In addition, there are many occupational settings in which workers are exposed to uranium in various forms: natural, depleted, enriched, reprocessed (Laurier et al., 2012). Since uranium is both a radionuclide and a heavy metal, it has the potential to exert both radiological and chemical toxicities Evidence of uranium effects from experimental studies Uranium has been found to be genotoxic in both in vitro and in vivo experiments (ATSDR, 2013), even in its depleted form which generates little alpha radiation (Bal et al., 2011; Stearns et al., 2005). A recent study showed that depleted and enriched uranium have different genotoxic profiles: depleted uranium has a low clastogenic but a high aneugenic potential, whereas enriched uranium has a high clastogenic potential, suggesting additional radiotoxicity on top of chemical toxicity (Darolles et al., 2010). These genotoxic properties clearly justify further investigations of the association between uranium exposure and cancer. In rats, inhalation of depleted uranium has been linked with the formation of DNA strand breaks in broncho-alveolar lavage cells, possibly as a consequence of uranium-induced inflammation and oxidative stress (Monleau et al., 2006). Also in rats, chronic inhalation of uranium ore dust was linked with primary malignant and non-malignant lung tumor formation and although the malignant tumor risk was not directly proportional to dose, it was directly proportional to dose rate (Mitchel et al., 1999). In dogs, uranium inhalation was related to neoplasia and fibrosis in the lungs (Leach et al., 1973). Although only limited experimental data are available on the effects of uranium inhalation on the lung, their findings suggest the need for further experimental and epidemiological studies. This is especially important since inhalation is a major route of exposure to uranium in the workplace. 6

7 Once absorbed, uranium is deposited throughout the body and the highest levels are found in the bones and kidneys (ATSDR, 2013). Many animal studies have reported damage to the kidney caused by acute exposure to uranium (Gueguen and Rouas, 2012), which is considered to be the main target organ for uranium toxic effects under such exposure conditions (ATSDR, 2013). Experimental studies on the effect of chronic exposure to uranium in kidney have reported mixed findings. Some reported nephrotoxic effects (Gilman et al., 1998; Zhu et al., 2009), whereas others unexpectedly did not find such effects (Poisson et al., 2014). The brain does not accumulate large quantities of uranium. However, some studies in rats have reported associations between uranium exposure and impairments in cerebral function, suggesting that the brain is an organ at risk for this nuclide (Houpert et al., 2005; Lestaevel et al., 2005). The potential effects of uranium on the cardiovascular system have rarely been evaluated in experimental studies. However, this question is of importance because of the relationships that have been demonstrated between impaired renal function (which again, may be caused by uranium (ATSDR, 2013)) and cardiovascular diseases (Currie and Delles, 2013). In addition, the inhalation of particulate matter of less than 10 µm aerodynamic diameter is an established risk factor for cardiopulmonary diseases (Araujo and Nel, 2009). Uranium bearing particles of such diameters are encountered in uranium mines (Marsh et al., 2012) and at all stages of the nuclear fuel cycle (Ansoborlo et al., 2002). Further biological effects of uranium exposure have been identified by experimental research, including oxidative stress, damage to the skin, impaired bone formation or reproductive function (ATSDR, 2013). A recent study in rats chronically exposed to low doses of uranium via drinking water showed that a metabolomics approach using urine samples allowed the correct classification of exposed and nonexposed animals (Grison et al., 2013). This suggests that metabolomics might be a useful tool to identify new biomarkers ( fingerprints ) for uranium exposure in humans, as well as to identify metabolic pathways activated by uranium exposure. This might in turn lead to the identification of further biological functions and systems perturbed by uranium exposure, which would justify complementary investigations of potentially related biological and health effects in epidemiological cohorts. In summary, it is important to emphasize that experimental studies on natural or depleted uranium (both being only weakly radioactive, i.e. showing low specific activity) have reported significant biological effects including genotoxicity, renal toxicity, lung damage and neuro-physiological perturbations. This suggests that uranium, as a heavy metal, may cause harm by chemical toxicity, aside from any radiation effects. In addition, experimental research confirmed that radiotoxicity can occur in addition to chemical toxicity, especially when exposure is to enriched or reprocessed uranium (Grignard et al., 2008; Houpert et al., 2005). There are however, considerable uncertainties in the extrapolation of the biological effects at low doses observed in animals, to possible disease risks in humans (ATSDR, 2013). These uncertainties are related 1) to possibly different biological responses induced by uranium in animals and in humans and 2) to the relations of such biological responses with disease risk quantification (e.g., the predictive value of early biological effects on subsequent disease risk). 7

8 It is therefore important to evaluate if exposure to uranium is associated with biological and health effects in humans. The influence of different mixtures of uranium isotopes also appears worth examining in this frame, in order to disentangle chemical from radiological effects (Zhivin et al., 2014) Evidence of uranium effects from epidemiological studies Cohorts of uranium miners and other workers have been identified as populations of priority interest for the study of uranium health effects (Laurier et al., 2012). Therefore the present section will focus on evidence from epidemiological studies conducted in occupational settings. Studies conducted in non-occupational settings (e.g., in populations living around uranium processing sites) have mostly employed ecological designs, and are therefore minimally informative. Molecular epidemiology studies incorporating biomarkers will be reviewed in section 1.5. Underground uranium miners are exposed to uranium ore dust and are therefore of interest to study the health effects of uranium exposure. However, previous analyses in these miners predominantly investigated the health effects related to the exposure to radon (a decay product of uranium or thorium) (NRC, 1999), whereas the associations between uranium exposure and health outcomes have been more rarely studied (e.g.: (Kreuzer et al., 2006; Rage et al., 2014)). The doses from alpha radiation to the lung from uranium and other long-lived radionuclides (LLR) present in uranium ore are indeed negligible when compared to those from radon exposure (Marsh et al., 2012). This is particularly true for miners who worked before the implementation of radiation protection standards (especially ventilation) in mines (Kreuzer et al., 2013; Rage et al., 2014; Tomasek, 2012). In addition, potentially high correlations between lung doses from radon and from long-lived radionuclides raise concerns about the possibility of disentangling the effect of uranium exposure on lung cancer from that of radon exposure. Non-lung cancer health outcomes are less likely to be affected by any confounding effect of radon exposure, because doses from radon to the non-respiratory organs where uranium may accumulate are very low (Marsh et al., 2012). By comparison, doses from LLR to some non-respiratory organs or tissues (e.g.: kidney, red bone marrow, liver) may be non-negligible (Laurier et al., 2009). In addition, the chemical toxicity of the uranium dust or other toxic effects resulting from the inhalation of uranium bearing particles (which might be related to their chemical composition or to other, nonradiological physical properties such as size (Schlesinger et al., 2006)) might increase the risk of diseases other than lung cancer (e.g. including, but not necessarily limited to, non-cancer diseases of the lung and kidney, and circulatory diseases). The associations between uranium exposure and outcomes other than lung cancer have been evaluated in a few studies of uranium miners (Kreuzer et al., 2006; Rage et al., 2014), but clearly deserve further investigation. Epidemiological studies focusing on the health effects of uranium exposure at later stages of the fuel cycle, where radon exposures are extremely low, have also been performed (ATSDR, 2013; Canu et al., 2008; Zhivin et al., 2014). These studies included individuals engaged in uranium mills (Boice et al., 2008; Kreuzer et al., 2014; Pinkerton et al., 2004)) and other uranium processing nuclear industries such as those located at Port Hope, Canada (Zablotska et al., 2013), Fernald, Ohio, USA (Ritz, 1999a; Silver et al., 2013), Mallinckrodt, Missouri, USA (Dupree-Ellis et al., 2000), Oak Ridge National Laboratory, Tennessee, USA (Richardson and Wing, 2006; Yiin et al., 2009), Paducah, Kentucky, USA (Chan et al., 2010; Figgs, 2013), Rocketdyne, California, USA (Boice et al., 2011), 8

9 AREVA NC Pierrelatte, France (Guseva Canu et al., 2012; Guseva Canu et al., 2011), Springfields and Capenhurst, UK (McGeoghegan and Binks, 2000a; McGeoghegan and Binks, 2000b). However, most of these previous studies were limited by various factors (Canu et al., 2008; Zhivin et al., 2014): The statistical power of individual studies conducted in specific settings was limited by relatively low numbers of workers included (usually a few hundred or thousands) and sometimes short follow-up times at the time of publication. Because of the often moderate to low level of exposures (and related low expected excess incidences of relevant diseases), the numbers of workers and length of follow-up might not have been sufficient to provide the power needed to detect statistically significant effects. Reconstructing retrospective exposures for large populations of workers is challenging and as a consequence many studies used indirect approaches for quantifying exposure. Individual dose assessment based on urine analysis and other results of individual monitoring for uranium (faeces, lung or whole body counts) would provide more accurate estimates and thus increase the ability of studies to detect excess risk (Boice et al., 2006). In addition, most analytical studies could not take into account the physical and chemical properties of uranium (isotopic composition, chemical speciation, aerodynamic diameter of particles, and resulting solubility). These properties may modify the mass and activity of the uranium reaching different organs, and therefore might determine whether toxic effects of radiological or chemical nature occur (Guseva Canu et al., 2011). Finally, only a few studies have collected information on occupational exposures other than uranium (e.g. (Chan et al., 2010; Guseva Canu et al., 2011; Silver et al., 2013) or on lifestyle risk factors such as smoking (e.g.(dupree et al., 1995)). If such factors are correlated with uranium exposure, they might have biased risk estimates in studies which did not consider them in the analysis. As a result, individual epidemiological studies conducted so far do not provide reliable evidence on the potential health risks associated with uranium exposure (Zhivin et al., 2014). Nevertheless, some studies are suggestive of biologically plausible positive associations between uranium exposure and lung cancer (Guseva Canu et al., 2011; Ritz, 1999b), lymphatic and hematopoietic tumours (Guseva Canu et al., 2011; Yiin et al., 2009), circulatory diseases (Guseva Canu et al., 2012), and intestinal cancer (Silver et al., 2013). Larger epidemiological studies with adequate individual dosimetry and control for confounders are clearly required Evidence of uranium effects from molecular epidemiology So far, research on the biological and health effects of uranium has mostly been disconnected between experimental and epidemiological studies. However, a few pioneering molecular epidemiology studies have also been conducted. Some of these studies investigated biological effects of exposure to depleted uranium in human populations. These populations include Gulf war veterans exposed through friendly-fire incidents via wound contamination and inhalation (McDiarmid et al., 2011a; McDiarmid et al., 2011b; McDiarmid et al., 2013). In these veterans, only relatively weak genotoxic, pulmonary or renal adverse effects have been associated with depleted uranium exposure so far. Other studies were carried out in workers involved in the clean-up of conflict zones contaminated with depleted uranium munitions in 9

10 the Balkans (Milacic, 2008; Milacic and Simic, 2009). In these workers, DNA alterations were found at higher rates immediately after decontamination than before (Milacic, 2008). Damages to chromosomes and cells were higher in workers exposed to depleted uranium than in controls groups of workers (Milacic and Simic, 2009). Other studies were conducted in populations drinking water naturally contaminated with uranium (Kurttio et al., 2002; Kurttio et al., 2006; Mao et al., 1995; Selden et al., 2009; Zamora et al., 2009). In these populations, several biomarkers of kidney function indicated some alterations associated with uranium concentrations in drinking water (ATSDR, 2013). No indication of serious renal injury was reported since values of biomarkers generally remained in the normal range (Ansoborlo et al., 2015). However, a cautious interpretation is warranted since the cross sectional design of these studies did not allow for a detailed characterization of the chronic effects of protracted, cumulated exposure to uranium (Canu et al., 2011). Similar studies also reported a positive association between uranium concentrations in water and bone resorption (as indicated by levels of serum type I collagen carboxyterminal telopeptide) (Kurttio et al., 2005) or with blood pressure (Kurttio et al., 2006). Several molecular epidemiology studies were conducted in uranium miners, but generally with a focus on radon rather than uranium effects (Brandom et al., 1972; Gomolka et al., 2012; Leng et al., 2013; Li et al., 2014; Zolzer et al., 2012a; Zolzer et al., 2012b). A few studies were conducted in other uranium workers (Clarkson and Kench, 1952; Dounce et al., 1949; Eisenbud and Quigley, 1956; Luessenhop et al., 1958; Martin et al., 1991; Prat et al., 2011; Thun et al., 1985). Most of them focused on renal biomarkers. The earliest studies failed to detect nephrotoxic effects of uranium exposure, possibly since the biomarkers investigated were not sensitive enough (ATSDR, 2013). However a study in uranium millers identified reduced renal proximal tubular reabsorption in this population, as compared to other workers (Thun et al., 1985). Another study in uranium workers reported a decrease of urinary osteopontin associated with uranium exposure, suggesting kidney damage (Prat et al., 2011). Another study focused on cytogenetic markers in blood lymphocytes and identified increases in asymmetrical chromosome aberrations and sister chromatid exchanges in workers occupationally exposed to uranium (Martin et al., 1991). An interesting biobank is being set up in uranium and plutonium workers in Russia (Takhauov et al., 2015) but no result on the association between uranium exposure and biomarkers is available yet. So far, published molecular epidemiology studies conducted in uranium workers have focused only on a limited number of biomarkers (mainly related to kidney function or cytogenetics) and generally lacked proper organ dosimetry. Most of these studies have been cross sectional and therefore have not allowed for a detailed assessment of the long-term effects of protracted exposure to uranium Room for improvement Overall, in spite of experimental results demonstrating chemical and radiological toxicity of uranium on living organisms, the biological and health effects of chronic uranium exposure in humans, particularly those due to occupational exposure to uranium, are not well known. However, through 10

11 conducting improved epidemiological and molecular epidemiology studies it should be possible to improve the understanding and quantification of these effects Epidemiology Despite the limitations of previous epidemiological studies regarding the quantification of the health effects of uranium exposure, cohorts of uranium miners and uranium processing workers were identified as part of DoRemi Task 5.5 as among the most potentially informative source of data for the characterization of effects of internal contamination (Laurier et al., 2012). The strengths of these studies include regular collection of dosimetric data and long-term follow-up of worker populations. In order to allow for such cohorts to provide better insight into the health effects of uranium exposure, several limitations of previous studies need to be addressed: First, existing cohorts should be pooled after verification of their compatibility, in order to increase the statistical power available for analyses and therefore improve the detection and quantification of potential uranium-related health effects. Similar pooling and joint analyses of cohorts have successfully been conducted in other populations of radiation workers and miners (Cardis et al., 2005; Leuraud et al., 2011; NRC, 1999; Tomasek, 2014). Second, the doses of cohort subjects should be estimated according to state-of-the-art methods. Methodologies to enable uncertainty in the dose estimates to be incorporated in dose response analyses should be developed. Finally, available information on risk factors other than uranium exposure should be carefully collected in order to adequately control for potential confounding or to take potential interactions into account in subsequent epidemiological analyses. It has been acknowledged that setting up multi-country joint epidemiological studies with reliable dose estimates and using harmonized methods is a highly desirable, but complex, approach. Such an approach requires preparatory work between researchers (epidemiologists, dosimetrists and statisticians) in charge of the cohorts to describe and compare available datasets and methods, and to define harmonized procedures for successfully pooling their data Molecular epidemiology DoReMi task 5.5. has identified that research on the effects of internal contamination would benefit from the integration of epidemiology and biology within the framework of molecular epidemiology. Such integration would have a strong potential to improve the understanding and the quantification of the biological and health effects of radiation exposure, notably resulting from internal contamination by uranium and other radionuclides (Laurier et al., 2012; Pernot et al., 2012). The rationale for integrating measurements of biomarkers in a large cohort study on the effects of internal contamination by uranium is based upon three specific points: 1. to refine the assessment of exposure to uranium by using biomarkers of exposure (e.g.: (Grison et al., 2013)) 2. to similarly evaluate possible confounders by the use of confounder specific biomarkers, if available and 3. to elucidate the different biological response mechanisms induced by uranium exposure in a human population by using biomarkers of early and late biological effects (Pernot et al., 2012). 11

12 If any association is observed between uranium exposure and the risk to develop a disease, point 3) may provide insight into the physio-pathological processes involved in disease onset. This will help assess the biological plausibility of the associations observed in epidemiological studies and inform related judgment about causality. This might also help to identify pre-clinical biomarkers useful for the surveillance of people exposed to uranium. Again, it should be possible to overcome the main limitations of previous molecular epidemiology studies, and to build on recent, as well as future, progresses of techniques for biological analyses: Although most previous molecular epidemiology studies have been cross sectional, regular medical check-ups conducted by the occupational medicine service during the workers careers have been identified as a unique opportunity for the prospective and repeated collection of biological samples among workers (Laurier et al., 2012). A biobank centralizing such samples over many years from a same cohort of individuals would be a valuable resource for a detailed assessment of the long-term effects of protracted exposure to uranium, as it is the case for other exposures (Palmer, 2007; Zins et al., 2010). Such a prospective scheme is also of great interest to identify, or to validate, biomarkers of recent exposures (e.g.: (Grison et al., 2013)) and their persistence or modifications over time. In addition, the immediate collection of biological material from workers exposed to uranium or other sources of radiation years to decades ago is of great interest in order to directly investigate any persistent biomarkers of past exposures (Gomolka et al., 2012). If specific diseases occurred in some of these workers, pathological archives (e.g. formalin fixed paraffin embedded tissue from target organs (Wiethege et al., 1999)) may also be of great interest to decipher uranium specific fingerprints in diseased tissues, for instance in tumor tissues from cancer cases. The need to use the best possible dosimetric estimates was mentioned above for epidemiology, and this applies to molecular epidemiology as well. An efficient solution to address this issue is to nest molecular epidemiology studies within established cohorts for which dosimetric data have already been carefully collected over many years to decades (Gomolka et al., 2012), or will be. This consideration also applies to many other factors for which data are already available within cohorts (e.g., potential confounders). Novel and already established biomarkers, notably originating from animal experiments, may prove useful to better understand the biological effects of uranium exposure on its different target organs, tissues and systems in humans. Some of these biomarkers have never been tested in humans exposed to uranium, but appear worth investigating for instance in uranium workers, as they might prove to be of interest to occupational health surveillance before the onset of any disease (e.g., biomarkers of brain damage, inflammation or kidney dysfunction). Non-targeted (OMIC) approaches are also promising for the purpose of identifying not only new biomarkers of exposure, but also biological (e.g., metabolic) pathways potentially affected by uranium exposure (Grison et al., 2013), and help direct research towards related pathways, functions/processes, organs and tissues. Overall, modern biology approaches are developing fast and should lead to the identification of new biomarkers of high interest in the forthcoming years or decades. Modern biobanking methods allow for the long term conservation of biological material in suitable conditions for later analyses of emerging biomarkers (Palmer, 2007; Zins et al., 2010). Diseases potentially 12

13 related to current uranium exposure might only be diagnosed many years, even decades, following exposure. Being able to investigate biological samples collected from the time of exposure to possible disease onset using the latest available techniques for biological analyses would be an invaluable asset for future research into potential mechanisms of disease onset. As for the preparation of pooled epidemiological studies, cooperation and coordination between researchers (biologists, epidemiologists, dosimetrists and statisticians) is essential, in order to undertake any successful and informative pilot molecular epidemiology study in a population of uranium miners or workers. These discussions should result in a good understanding of the expectations from the study field by each partner and related justifications, as well as the mutual understanding of each partner s (and discipline s) role within the frame of molecular epidemiology studies. To set up a molecular epidemiology study, the identification of biomarkers useful to study the effects of uranium exposure in humans is necessary as well as the identification of populations suitable for the collection of appropriate biospecimens. Valid biomarker typing can only been done in biological samples of good quality. Therefore the definition of proper study instruments (standard operating procedures, questionnaire, information and consent sheets), and the setting-up of an optimal sampling scheme are essential requirements. These have to be a priori verified and tested in a smaller feasibility study. The agreement and support of all stakeholders (workers, worker representatives, employers, national ethics committees,) must be obtained before undertaking the study. The experiences and feedback of other European initiatives in the field must be used especially from the European Biobanking and Biomolecular Resources Research Infrastructure (BBMRI) project (bbmri-eric.eu). 13

14 2. OBJECTIVES OF THE CURE PROJECT The objective of the concerted action CURE (Concerted Uranium Research in Europe) was to elaborate a multidisciplinary and collaborative European research project, integrating epidemiology, biology/toxicology and dosimetry to improve understanding and quantification of long-term biological and health effects (including both risks of cancer and non-cancer diseases) associated with uranium contamination. This general objective included two specific aims: To prepare a common protocol for pooled epidemiological analyses of uranium miners and workers in Europe, that would overcome the limitations of previous studies in the field, in order to directly estimate the potential health risks associated with uranium exposure. To verify the feasibility of a molecular epidemiology approach (building biobanks and measuring biomarkers) for integration into the risk assessment process and if feasible, to elaborate a common protocol for the development of a molecular epidemiology study. A parallel objective of the project was to identify and establish contact with similar research programs, within and outside Europe. 3. PROJECT MANAGEMENT AND TASKS COMPLETED 3.1. General characteristics CURE was a concerted action supported by the DoReMi EU FP7 network of excellence ( and coordinated by Institut de Radioprotection et de Sûreté Nucléaire (IRSN, France). It was integrated as task 5.8 of DoReMi. The project duration was 18 months, from July 1, 2013 to December 31, Partners CURE gathered together the main organizations in charge of conducting and/or analyzing cohort studies of miners and other workers occupationally exposed to uranium in Europe, as well as organizations with recognized expertise in internal dosimetry and/or in radiobiology. In total 9 partners from 6 Countries (France, UK, Germany, Belgium, Czech Republic, Spain) participated in the CURE project: Institut de Radioprotection et de Sûreté Nucléaire (IRSN, France), Bundesamt für Strahlenschutz (BfS, Germany), Public Health England (PHE, United Kingdom), Nuvia limited (United Kingdom), Atomic Weapons Establishment (AWE, United Kingdom), StudieCentrum voor Kernenergie Centre d'étude de l'energie Nucléaire (SCK CEN, Belgium), Státní ústav radiační ochrany (SURO, Czech Republic), Centre de Recerca en Epidemiologia Ambiental (CREAL, Spain), Institut Curie (IC, France). All partners were involved in the reviews conducted in DoReMi WP4 on pertinent cohorts for radiation protection, in WP5 task 5.5 on cancer risk and internal contamination (Laurier et al., 2012), 14

15 in WP6 on useful biomarkers for epidemiology (Pernot et al., 2012) and in WP7 Task7.2 on pertinent studies for low dose cardiovascular risk. The respective experiences and fields of expertise of the partners involved in CURE and directly relevant to this project are summarized below: IRSN has experience in European collaborative projects (e.g.: DoReMi WP7 and WP5-Task5.5, coordination of the FP6 Alpha-Risk project), in the conduction of epidemiological cohorts studies of uranium miners and of uranium workers, in internal dosimetry of uranium miners and of uranium workers, in biodosimetry, in the investigation of biological and toxicological effects of uranium exposure (Envirhom program) and in advanced biostatistics. Two PhD students at IRSN (Drubay D, Zhivin S) participated to the CURE project. BfS has experience in European collaborative projects (Alpha-Risk, DoReMi, Multibiodose, RENEB, etc), in epidemiological cohorts of uranium miners, in biobanking of radiation exposed cohorts and radiation sensitive individuals (German Uranium Miners Biobank), in the analyses of biological effects (Chromosomal analyses [mfish, Giemsa], gammah2ax, Comet assay, Proteomics) of radiation exposure, of radiation sensitivity, biological effectiveness of radiation, and in dosimetry of internal emitters. PHE has experience of European collaborative projects (Alpha-Risk, SOUL, EpiRadBio, SOLO). As part of the SOLO project, PHE performed coordination, dosimetric reconstruction, biological and epidemiological risk analysis tasks. PHE has expertise in epidemiological cohorts of nuclear workers (PHE manages both the UK National Registry for Radiation Workers comprising of over 200 thousand workers and a cohort of over 60 thousand former employees of British Nuclear Fuels Limited), in internal dosimetry for uranium and plutonium exposed workers and for uranium miners. PHE has expertise in radiation biology (RISK-IR), for instance in the field of cancer genetics and cytogenetics, and carries out research into the fundamental mechanisms by which radiation causes cancer. Nuvia and AWE have experience in European collaborative projects (Alpha-Risk), in epidemiological cohorts of uranium workers, in the monitoring of workers exposures and in the assessment of internal doses. Nuvia and AWE are collaborators in the AIRDoseUK project within DoReMi (task ), which will improve the organ doses calculation among United Kingdom Atomic Energy Authority (UKAEA) workers. SCK CEN has experience in European collaborative projects (NOTE, EPI-CT, Alpha-Risk, CEREBRAD, PROCARDIO, GENRISK-T), in the analysis of biological/molecular effects of radiation exposure, radiation sensitivity as well as on cancer susceptibility markers and non-cancer diseases using up-todate infrastructures. SCK CEN has the long-term commitment to develop research in low dose radiation issues, and in epidemiological cohorts of nuclear workers. SURO has experience in European collaborative projects (Alpha-Risk). SURO has expertise in epidemiological cohorts of uranium miners, internal dosimetry, radiochemistry, statistics and occupational medicine. Within DoReMi, SURO is leading the IntEmitUM project (task 5.5.1), which aims to the integration of dosimetry, biology and epidemiology to estimate cancer risks related to internal exposure among miners using measurement of uranium in urine. 15

16 CREAL has experience in European collaborative projects (e.g.: Alpha-Risk, Int-Thyr), in the epidemiology of nuclear workers and in risk modelling, in molecular epidemiology and biomarkers. CREAL has experience in the consideration of dose uncertainties and their propagation into risk estimates (nuclear workers, Chernobyl liquidators). In DoReMi, CREAL is coordinating the crosscutting molecular epidemiology group and the reflexions on molecular epidemiology in WP4 and WP6. IC has considerable experience in molecular epidemiology studies and biomarkers. In the framework of DoReMI, IC participates in the cross-cutting molecular epidemiology group. 16

17 3.3. Work packages The CURE project included four work packages (WPs): WP1 (leader: Richard Haylock, PHE) dedicated to epidemiology WP2 (leader: Eric Blanchardon, IRSN) dedicated to dosimetry WP3 (leader: Maria Gomolka, BfS) dedicated to biology WP4 (leader: Dominique Laurier, IRSN) dedicated to the management and general coordination of the project In addition, it was decided in March 2014 to create an uncertainty work group (UWG, leader: Augusto Giussani, BfS), including members from all the main Work Packages of CURE (epidemiology, dosimetry and biology). The UWG was dedicated to the coordination and harmonization of the methods for addressing the cross-cutting issue of uncertainties at different steps of the project Tasks completed The main tasks of each work package and interactions between them are briefly summarized in Figure 1. Figure 1. Work packages of the CURE project, main tasks and interactions between work packages (WPs) and the Uncertainty Work Group (UWG). A more detailed list of the tasks performed as part of the CURE project is provided below. Because of the strong interactions between the different WPs and disciplines, many tasks were conducted 17

18 jointly by two or more WPs. Therefore, many tasks completed could not strictly be grouped by WP, but the contributions of each WP to the listed tasks are mentioned: Review of regulatory issues in each country (agreements from ethics committees and workers representatives), both for epidemiology and molecular epidemiology studies. See (Haylock et al., 2014a) for details (WP1+WP3). Detailed description of the cohorts of miners and workers to verify their compatibility (nature of the variables collected, availability, quality and extent of follow-up, quality and completeness of dosimetric information) and identify the specific strengths of some subcohorts (e.g.: data on potential confounders available either from Job exposure matrices (JEMs) or from medical files, cancer incidence data). See (Haylock et al., 2014b) for details (WP1+WP2) Definition of cohorts or cohort subsets suitable for different kind of epidemiological analyses, see section 4.1. (WP1 + WP2) Definition of a common general strategy for molecular epidemiology (i.e.: to perform biological sampling, biomarker measurements and analyses in a cohort of uranium exposed workers and proposition of cohort subsets for pilot molecular epidemiology studies, see section (WP1 + WP3) Identification of relevant biomarkers of exposure or of potential early or late effects, of interest to study the biological and health effects of uranium (Gomolka et al., 2014). Both targeted biomarkers (selected notably on the basis of previous experimental studies) and non-targeted biomarkers (OMICs) were considered (WP3). Identification of biological specimens to be collected for the measurement of the selected biomarkers, see section (WP3) Definition of standard operation procedures (SOPs) for biological samples and biomarkers (collection, processing, storage, analysis) (Gomolka et al., 2014) (WP3) Development of instruments for a pilot molecular epidemiology study (Gomolka et al., 2014): questionnaire (see section ), information sheet and consent form (see section ) (WP1+ WP3) Verification of the feasibility of pilot studies in the proposed cohort subsets for molecular epidemiology: onsite visits, collection of information on field conditions, and comparison with SOPs, see section (WP1+ WP3) Development and adaptation of the logistic strategy in the pilot cohorts to inform the medical doctors, workers, to collect and transport the biospecimens and to isolate and store the biological material of interest, see section (WP1+ WP3) Identification of centers able to provide or assist in biospecimen sampling, storage and/or specific biomarker testing (Gomolka et al., 2014)(WP3) Evaluation of the costs of biospecimen sampling, storage and biomarker testing (Gomolka et al., 2014) (WP3) Definition of the relevant biokinetic and dosimetric target organs, relevant to the diseases (see section 4.4.) (WP2+ WP1) and the biomarkers (see section ) of interest (WP2+ WP3) Detailed evaluation of the availability and quality of monitoring data available in each cohort for dosimetric calculations (Blanchardon et al., 2014b) (WP1+ WP2) Analysis of similarities and differences in exposure reconstruction approaches for miners and workers (Blanchardon et al., 2014a) (WP2) 18

19 Definition of a dosimetry protocol, harmonizing the dose assessment procedures for similarly exposed subjects across different countries and sites (Blanchardon et al., 2014a) to allow for joint epidemiological and molecular epidemiology analyses, see section 4.3. (WP2) Creation of the uncertainty work group, to address the cross-cutting issue of uncertainties (UWG = WP1+ WP2 + WP3) Identification of the potential sources of uncertainties at several steps of the project via the collective building of a matrix listing identified sources of uncertainties at each step (Giussani et al., 2014): data collection, dose calculation process, epidemiological and biomarker analyses (WP1+ WP2 + WP3 within UWG) Case studies to assess the relative impact of certain sources of uncertainties on dose calculations (WP2 within UWG) Prioritization of the sources of uncertainties to be considered (e.g.: to be further characterized or reduced, whenever feasible) (WP1+ WP2 + WP3 within UWG) Propositions of methodologies for the propagation of uncertainties in epidemiological analyses (WP1 within UWG) Establishment of contacts with other researchers involved in studies of uranium workers or miners inside or outside Europe, in order to identify their potential synergies with CURE for future projects (WP4 + WP1 + WP3) Establishment of contacts with researchers outside the radiation field for input of already existing knowledge in the biobanking field and identifications of potential synergies with other projects (WP3+WP4) Setting up of a CURE web space hosted by DoReMi website (WP4) General project coordination (WP4) Early dissemination work : presentations at several workshops and conference (WP4), see annex 2 19

20 4. PROPOSED INTEGRATED PROTOCOL The current protocol was designed with two specific aims: to set up a joint epidemiological study in order to directly estimate risks from occupational exposure to uranium, by pooling data from established nuclear worker and miner/miller cohorts from five countries (United Kingdom, France, Belgium, Czech Republic, Germany). This would generate datasets with the highest possible statistical power to provide evidence for the appropriateness of radiation protection standards for people occupationally exposed to uranium. More generally, this would contribute to improve the knowledge of the health effects (both cancer and non-cancer) of internal contamination and help assess the appropriateness of current radiation protection standard regarding internal exposure. to investigate the feasibility of setting up molecular epidemiology studies. Prospective molecular epidemiology studies would allow for an improved understanding of the mechanisms through which uranium might generate health effects and possibly to a better quantification of related health risks. It might help to identify biomarkers useful to the health surveillance of workers exposed to uranium. Thus it is important to test the feasibility to integrate pertinent biomarkers into an epidemiological study. For each aim, importance has been given to the production and use of the best possible dosimetric estimates, to the identification of sources of uncertainties at each step of the research protocol, and to the identification of possible methods for propagating such uncertainties into the analyses. As a result, the outline of the protocol is divided into the following sections: Epidemiologic cohorts and subsets for analyses Biological protocol for a molecular epidemiology study Dosimetric protocol Methods for statistical analyses Characterization and propagation of uncertainties 20