Article Text

Download PDFPDF

Original article
Does uranium induce circulatory diseases? First results from a French cohort of uranium workers
  1. Irina Guseva Canu1,
  2. Jerome-Philippe Garsi1,
  3. Sylvaine Caër-Lorho1,
  4. Sophie Jacob1,
  5. Philippe Collomb2,
  6. Alain Acker3,
  7. Dominique Laurier1
  1. 1Institute of Radiological Protection and Nuclear Safety, Division of Radiological Protection and Human Health, Department of Radiobiology and Epidemiology
  2. 2AREVA NC, Department of Occupational Medicine, Pierrelatte, France
  3. 3AREVA group, Medical Coordination Section, Paris, France
  1. Correspondence to Dr Irina Guseva Canu, Institut de Radioprotection et de Sûreté Nucléaire, IRSN/DRPH/SRBE/LEPID, BP 17, Fontenay-aux-Roses 92262, France; irina.canu{at}


Objectives Increased risk of circulatory system diseases (CSDs) was observed in nuclear workers handling uranium and plutonium in Russia and the UK. This work examines the CSD mortality after chronic intake of uranium among 2897 workers (79 892 person-years) at a uranium processing plant (1960–2006) in France.

Methods Cumulative exposure to different uranium compounds, classified by their isotopic composition and solubility type, was quantified on the basis of a plant-specific job-exposure matrix and individual job histories. HRs and associated 95% CI for CSD (n=111) and specific CSD categories were estimated using Cox regression models, stratified on sex and birth cohort and adjusted for potential confounders. The effect of smoking was analysed among 260 smokers (42 CSD deaths).

Results Compared to unexposed workers, CSD mortality was increased among workers exposed to slowly soluble reprocessed uranium (RPU) (HR=2.13, 95% CI=0.96 to 4.70) and natural uranium (HR=1.73, 95% CI=1.11 to 2.69). The risk increased with cumulative exposure and exposure duration. In the subgroup of smokers, the risk estimates were higher but with larger CIs: HR=1.91 (95% CI=0.92 to 3.98) for natural uranium and HR=4.78 (95% CI=1.38 to 16.50) for RPU.

Conclusions The authors observed that exposure to slowly soluble uranium, namely RPU, may increase the risk of CSD mortality. However, these results are preliminary since the study is lacking statistical power and many other biological and lifestyle-related factors may cause CSD. More detailed investigations are necessary to confirm these findings and analyse in depth the effects of internal radiation exposure on the circulatory system.

  • Uranium
  • internal radiation
  • inhalation
  • mortality
  • circulatory system diseases

Statistics from

What this paper adds

This study addresses the risk of mortality from CSD among a cohort of French uranium workers by considering the physical–chemical form of industrial uranium compounds and their isotopic composition. The study suggests that exposure to slowly soluble uranium compounds, notably reprocessed uranium compounds, may increase the risk of CSD mortality.


There is compelling evidence of a causal link between therapeutic acute high-dose radiation exposure (>5 Sv) and mortality from circulatory system disease (CSD)1–3 More recently, an association between external radiation exposure and CSD mortality has been studied among Japanese A-bomb survivors (receiving single whole-body irradiation, which resulted in doses below 4 Sv)4 and nuclear workers (exposed to protracted low-dose radiation).5 In general, there is little information on the risk of CSD after protracted exposure to low-dose external radiation, and very little information in relation to internal exposure from incorporated radionuclides. In the nuclear fuel industry, internal exposure to uranium and plutonium is of major concern for radiation protection measures. An increase in cardiovascular mortality and incidence was reported among Mayak plutonium workers.6–8 A strong association between mortality from CSD and radiation exposure was also observed in British Nuclear Fuels plc (BNFL) men.9 Although part of this cohort consisted of uranium workers (37% of the cohort were workers employed at Springfields uranium processing plant and 6.7% at Capernhurs uranium enrichment plant), no formal study of the effects of uranium on the circulatory system has been performed to date.

In France, since 1960, the AREVA NC Pierrelatte plant has enriched uranium, carried out uranium chemical conversion and performed all associated logistical activities. Since uranium is the only nuclear material ever to have been used at the plant, a cohort of AREVA NC Pierrelatte plant workers was formed in 2005 as a pilot study of French uranium workers. It included 2709 male workers employed at the plant for at least 6 months between 1960 and 2005, who had never been uranium miners. The mortality pattern and association between internal uranium exposure and cancer mortality in this cohort had been studied and reported previously.10 11

The present study examines the possible impact of uranium exposure on CSD mortality. Since uranium presents both chemical (heavy metal) and radiological (α-radiation emission) toxicity, exposure to different uranium compounds handled at the plant was considered according to their solubility and isotopic composition.

Materials and methods

Study population and follow-up

In 2010, the AREVA NC Pierrelatte cohort was updated to include female workers and to extend the follow-up. Meanwhile, missing data on socioeconomic status (SES) at hire was completed and validated based on the archived employee rosters. Person-time of follow-up was computed for each subject from the date of employment at the AREVA NC Pierrelatte plant plus 6 months or from 1 January 1968 if employment started before this date, until the date of death or the end of follow-up (31 December 2006). Vital status was updated using data from the National Natural Persons Identification Index (date and place of death of deceased workers). Causes of death were obtained from the National Cause of Death Registry, which contains anonymised records of all deaths in France since 1968 and their causes. Death records were matched to cohort members by date of birth, gender and date and place of death. Because causes of death were unavailable on the national level before 1968, the follow-up between 1960 and 1968 was not possible. The causes of death were coded according to the 9th and 10th revisions of the International Classification of Diseases (ICD-9 and ICD-10).

In accordance with available literature,12 we defined the overall group of deaths from all-CSD as ICD-9 ‘390–459’ and ICD-10 ‘I00–I99’ and two major subgroups: ischaemic heart diseases (IHD) as ICD-9 ‘410–414’ and ICD-10 ‘I20–I25’ and cerebrovascular diseases (CVD) as ICD-9 ‘430–438’ and ICD-10 ‘I60–I69’.

Exposure assessment

Internal uranium exposure

Individual exposure to different uranium compounds was assessed using a plant-specific job-exposure matrix (JEM).13 14 In this JEM, uranium compounds were classified according to their isotopic composition, by distinguishing reprocessed uranium compounds from the natural uranium compounds (NU) and their solubility in biological tissues (table 1). Three solubility types (F-fast, M-moderate and S-slow) were determined, based on the Human Respiratory Tract Model15 and specific workstation analytical studies.16 17 For each type of uranium, the JEM provides the frequency of exposure and the amount of compound handled by workers when exposure occurs, both assessed on a four-level relative scale for 232 job period pairs. The individual cumulated exposure scores were calculated as the product of frequency, extent of exposure and duration of employment for each job period and each type of exposure and for each of the jobs in the worker's career at the AREVA NC Pierrelatte plant. Details regarding the construction of the JEM and cumulative exposure scores are described in supplementary data, supplement-1 (S1) and previous publications.13 14 18

Table 1

Classification of uranium compounds handled at the AREVA NC Pierrelatte plant

Non-radiation-related risk factors

Exposure to solvents, shift work and noise were suggested as potential risk factors of CSD.19–22 Moreover, heat (in degree Celsius) and trichloroethylene (TCE) exposures were found to be correlated with uranium exposure.14 To account for their potential confounding effect, exposure to these factors was assessed using the JEM,14 as binary (ever vs never) variables. Workers' shift work experience was reconstructed using the JEM's job definition. An experienced ergonomist flagged job periods with exposure to noise (>80 dB) using industrial hygiene measurement records from the Pierrelatte plant.

Information on smoking (smoker vs non-smoker) was obtained for a sample of 345 (12%) workers. It was collected in 2007 using a standardised self-administered questionnaire mailed to AREVA NC Pierrelatte retired workers randomly selected from the company retiree record. The questionnaires of deceased retired workers were answered by their relatives.

Statistical analyses

Exposure variables

The distribution of exposure scores was highly skewed in the cohort,14 23 with a few deaths in the high exposure categories. To analyse the relationship between uranium exposure and CSD mortality in the cohort, uranium exposure was modelled in three different ways: (1) a time-dependent binary variable corresponding to exposure status was created to separate workers into two categories: exposed workers with at least one annual uranium compound exposure score >0 and unexposed workers with no exposure to the uranium compound of interest. The cumulative exposure score was used as (2) a three-class categorical variable with cut-points at 0, at the 75th percentile for NU compound exposure and at the 95th percentile for RPU compound exposure and as (3) a log-transformed continuous variable to analyse the risk per step of cumulative exposure. Moreover, duration of exposure to each of the uranium compounds was considered in order to examine risk trends per year of exposure. Each of these variables was treated as time-dependent.

Statistical model

Cox regressions with age as the main time variable24 were run to estimate HRs and 95% CI for the association between each exposure variable for each type of uranium compound and mortality from CSD, IHD and CVD. Potential confounding problems were addressed by stratification of each model on sex and 10-year birth cohort and by adjustment on, four-class SES at hire in order to account for differential baseline risk across the SES categories.10 Furthermore, each model was adjusted on exposure to heat, aromatic solvents (eg, solvents containing aromatic hydrocarbons, such as benzene, toluene, xylene or styrene), TCE and shift-work, which were considered as time-dependent covariates in models, without lag. The selection of adjustment covariates was done (1) a priori, based on the knowledge on CSD risk factors, notably aromatic solvents, TCE, heat and shift work and (2) based on the results of the unidimensional models for specific exposures with effects of confounding (20% change in risk coefficient). Since no CSD death occurred among workers with exposure to noise (supplementary data, table S1), the latter was not included into the model. The final model selection was done based on the Akaike information criterion.25 To account for a latency time between exposure and disease, the cumulative exposure score was lagged by 10 years.

To check the sensitivity of the results, models with 0-year, 5-year and 10-year lag were run. Moreover, models with additional adjustment for employment status and employment duration were tested. Since heat is known to modify respiratory function, we also modelled individual uranium exposure by multiplying annual exposure score by 10% and 15% for workers employed at jobs with exposure to 35–60°C and >60°C temperature, respectively.

Although the analysis of available smoking data (n=345, 12% of the cohort) showed no association with all types of uranium exposures18 (supplementary data, table S2), we examined the effect of smoking in addition to uranium exposure. Since only one CSD death occurred among those who had never smoked, we restricted analysis to smokers (n=260), using the model with binary time-dependent exposure variable and similar adjustment as described above.

Statistical analyses were conducted using STATA-V.10 (Stata Corporation) software. This study met all local ethical recommendations. The use of the individual data was approved by the French Data Protection Authority (CNIL) and Pierrelatte Plant Committee of Hygiene, Safety and Working Conditions (CHSCT).


Cohort description

The AREVA NC Pierrelatte cohort is characterised by a long follow-up period (27.6 years, on average), a high percentage of skilled workers (65%) and a young age at the end of follow-up (table 2). Nobody was lost to follow-up and at the end of follow-up, 16% of the cohort members were deceased (460 deaths). Causes of deaths were available for 455 deaths (99%). CSD was an underlying cause for only a quarter of observed deaths (111 deaths), including 48 deaths from IHD and 31 deaths from CVD.

Table 2

Demographic and occupational characteristics among a cohort of workers employed at the AREVA NC Pierrelatte uranium processing plant

Exposure to soluble NU compounds was a major concern at the plant, whereas exposure to insoluble compounds was less common (table 3). Since industrial processing of RPU only began in the 1980s, only a limited proportion of the workers was exposed to RPU dusts.

Table 3

Distribution of exposed workers, person-years and number of observed deaths from circulatory system diseases by type of uranium compounds in the cohort AREVA NC Pierrelatte plant workers (1968–2006)

Exposure to uranium of different isotopic composition and solubility did not always arise at the same time in a worker's career. In total, 2331 workers (81%) were exposed to at least one type of uranium compound; 945 workers (33%) among them were exposed to NU compounds of three types of solubility (F, M and S), while 390 workers (14%) handled RPU compounds of type F, M and S. About 13% (377 workers) were exposed to all types of uranium available at the plant. Workers exposed to NU type S compounds were always co-exposed to more soluble uranium compounds, whereas eight workers were exposed solely to RPU type S compounds. No workers were exposed to both RPU type M and RPU type S compounds without being exposed to RPU type F. We observed no correlation between exposures to the different uranium compounds. Most of workers were exposed to the heat (81%) and TCE (57%) (table S1).

Exposure risk estimates

CSD mortality was associated with exposure to insoluble compounds of NU and RPU, though the latter was borderline significant (table 4). For both types of uranium, the risk was highly significant among the most exposed workers and increased with cumulative exposure and exposure duration. The risk remained stable whatever the lagged cumulative exposure to NU type S, and increased with increasing lag of RPU type S cumulative exposure. Whatever the exposure variable considered, for both NU and RPU isotopic composition, we observed an increase in mortality risk with decreasing solubility of uranium.

Table 4

HRs* and 95% CIs for circulatory system, ischaemic and cerebrovascular disease mortality associated with exposure to different types of uranium compounds

Analysis of IHD mortality was based on 48 deaths. It provided a similar pattern of results (table 4), with the highest, though not always significant, HRs corresponding to type S NU and RPU compounds exposure.

Analysis of CVD mortality was based on 31 deaths, and results are less consistent (table 4). A significant increase of mortality per step of 10-year lagged cumulative exposure score was observed only among workers exposed to soluble RPU. No trend according to uranium solubility was observed for any of the exposure variables.

Analyses restricted to 260 smokers (table 5) showed a three- to fivefold increased risk of the CSD mortality among smokers exposed to RPU compared to unexposed smokers. A similar trend to that observed in the entire cohort with decreasing solubility was demonstrated. However, HRs among RPU-exposed smokers in the smoking subcohort were twice those observed in the entire cohort.

Table 5

Risk of circulatory system disease mortality by exposure to different types* of uranium in the subcohort of 260 smoking workers


In this study, we consider the CSD mortality risk in relation to protracted low-level internal exposure to different types of uranium generated by the nuclear fuel industry. Our results suggest that exposure to slowly soluble compounds may increase the risk of CSD and that this increase might be higher after RPU exposure. This association seems to persist after accounting for smoking.

This study is the first epidemiological study regarding the effect of industrial uranium on the circulatory system to date and has several limitations. Its statistical power is limited, with few observed deaths in specific disease categories, thus the resulting risk estimates are not very precise, except those for CSD mortality. The unavailability of bioassay data and resulting internal uranium dose for risk quantification per gray or sievert is another limitation, though other relatively accurate exposure metrics23 enabled quantitative exposure-effect analyses. Since no national registry of CSD incidence exists in France, only mortality analysis was performed. A significant effort was done to address the issue of potential bias23 and confounders. According to the sensitivity analysis results, additional adjustment of models for employment duration or employment status (results not shown) does not modify our results. Another potential selection issue related to differences in baseline risk between workers assigned to more versus less exposed jobs, with more sedentarily and less smoking restrictions (eg, in clerical workers) was controlled by adjustment for SES at hire.

Only a partial set of data (12%) could be obtained for smokers. Contrary to our cancer mortality study,18 CSD mortality risk after uranium exposure could not be addressed with an adjustment for smoking because only one death occurred among unexposed non-smokers. Nevertheless, the results from the smoker subgroup suggest that smoking modifies rather than confounds the effect. This is concordant with our previous study of lung cancer18 and studies of Mayak plutonium workers and uranium miners26–28 which also reported a joint (synergy) effect between smoking and radiation exposure. Besides smoking, there are numerous other well-known risk factors of CSD, such as hypertension and a sedentary lifestyle, for which we had no information in this study. Therefore, our results should be interpreted with caution, before their impact on the uranium–CSD mortality relationship can be assessed.

We found no specific human study on cardiovascular effect after occupational uranium exposure. Blood pressure and pulse were reported unchanged in a man accidentally exposed to uranium powder for 5 min,29 while environmental exposure to uranium was positively correlated with diastolic30 and systolic blood pressure.31 No animal study reported cardiovascular damage after uranium inhalation exposure, probably because they aimed at exposure limit establishment and damage in uranium-target organs (kidney>bone>liver) and the lungs appears earlier.32 However, in human and animal studies of uranium ingestion, absorbed uranium was shown to alter renal function due to its particular tropism for kidney and its heavy metal properties.33 It is likely that these properties enable uranium to act upon renin–angiotensin–aldosterone system regulating blood pressure and to trigger changes in metabolism. Some data suggest that protracted exposure to uranyl-nitrate increases plasma fibrinogen and prothrombin in rabbits34 and alter cholesterol metabolism in rats.35 36 Protracted inhalation of uranium dioxide was shown to increase macrophage activity in rats.37

Experimental in vivo and in vitro studies were reviewed2 and suggest two potential radiation-related mechanisms for CSD. The first is an inflammatory response mediated by the nuclear factor kappa B (NF-κB) with activation of macrophages and endothelial cells and release of cytokines, oxygen and nitrogen radicals, E-selectin, PECAM-1 and interleukin 8 in the vessels of the myocardium. This leads to a loss of capillary density and subsequent ischaemic myocardial degeneration.2 The second pathway involves the innate immune reaction to oxidised LDL in the intima of large blood vessels with expression of interleukin 6, C reactive protein, serum amyloid A and fibrinogen, related to atherosclerotic damage observed in cerebrovascular injuries.2 Moreover, radiation was shown to induce prothrombotic effects via release and deposition of von Willebrand factor in the arteries of rat myocardium, in mouse kidney glomerulus and in the plasma of monkeys.38 This induces a subsequent increase in platelet adherence and thrombus formation in irradiated capillaries and arteries.39

All uranium isotopes, except 237U, emit α radiation (∼50 μm range in soft tissue), six emit γ rays, and three of them are β particle emitters. RPU contains various isotopes of uranium and other actinides (table 1) as well as contaminants from fission and activation reactions.40 This particular isotopic composition makes it more radioactive than NU, especially due to its higher potential as a β and γ ray emitter.11 We may suppose that the accumulation of insoluble NU and especially RPU in the lung and regional lymph nodes would create a radioactive environment for the heart because of long retention of slowly soluble uranium particles (weeks for type M to many years for type S11). Thus, internal irradiation continues even after exposure cessation. Soluble uranium compounds entering into systemic circulation might induce a systemic inflammatory response, oxidative stress and intimal damage in large vessels, leading to CVD. The inflammatory effect is probably aggravated by the toxicity of inhalable dust particles (the form in which uranium is inhaled) because most studies recognise the cardiovascular effect of particulate matter.41

In the light of these data, our results seem biologically relevant. Moreover, CSD increasing risk was also observed among plutonium workers6–8 and uranium miners.42 However, many other factors, biological (eg, heredity), physiological (eg, hypertension, diabetes) and lifestyle related (eg, physical activity, smoking) are known to cause CSD, according to complex mechanisms with lots of interactions. In our study, we could only assess some of them, focusing on the uranium and associated occupational exposures. Therefore, the results should be interpreted as preliminary and require further investigation of the internal uranium exposure effect. We hope this will be achievable by the means of a nested case–control study, in which we expect to consider all available data from the individual workers' medical follow-up, to reconstruct the uranium internal dose and to quantify the dose–response relationship.


We observed that exposure to slowly soluble uranium increases the risk of CSD and that this increase might be higher after exposure to RPU. However, these results should be considered as preliminary since many other factors are known to cause circulatory diseases. Furthermore, more detailed investigations are necessary to confirm our findings and analyse in depth the pathogenesis of internal radiation exposure for the circulatory system.


The authors thank Dr Rage, from the IRSN, and our colleagues from the ‘Alpha risk’ project network for their advice and their review of the manuscript.



  • IGC and J-PG contributed equally to this paper.

  • Funding This work was funded by the IRSN and AREVA (PIC-Epidemiology 2009/2012 grant), with partial financial support from the EC (EURATOM FIP6-516483 grant).

  • Competing interests None.

  • Patient consent Only anonymous individual data were used in the study, with approval of the French Data Protection Authority (CNIL) and Pierrelatte Plant Committee of Hygiene, Safety and Working Conditions (CHSCT).

  • Ethics approval The ethics approval was provided by French Data Protection Authority (CNIL) and Pierrelatte Plant Committee of Hygiene, Safety and Working Conditions (CHSCT).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.