In August 2008, an incidental gaseous release of I-131 occurred at the Institute for Radio-Elements (IRE) situated in the nuclear site of Fleurus. Following the incident, the Belgian Minister of Social Affairs and Public Health commissioned a study to assess, by means of an epidemiological study at the national level, the possible health risks for populations living in proximity of nuclear power plants or other facilities that can be at the origin of a leak of radioactive material. An ecological study was performed over the period 2000–2008 to assess the possible risk of thyroid cancer for people residing near nuclear sites [1
]. For two of the four Belgian sites (Mol-Dessel and Fleurus), the incidences of thyroid cancer within the 20-km proximity area were higher than expected.
Exposure to ionizing radiation, particularly during childhood, is the best-established risk factor associated with thyroid cancer [3
]. Significantly increased risks for developing thyroid carcinomas following radiation exposure during childhood were observed among people treated by radiotherapy or exposed in nuclear accidents or among survivors of the atomic bombings. For example, a direct consequence of the Chernobyl disaster was an increase in the number of thyroid cancers in children during the years following the accident [4
]. Ionizing radiation can also cause thyroid cancer in adults, but will occur after exposures to higher doses [5
]. The risk in Japanese A-bomb survivors is marginally elevated [6
]. The radionuclide I-131 is known to be a major contributor to thyroid cancer risk in case of nuclear accidents [7
The present study follows the recommendations resulting from the previous one, which were: “To repeat the epidemiological monitoring within five years, since by that time more cancer data will have become available.” The previous analyses were then replicated for the period 2000–2014. The investigation particularly focused on the question of whether thyroid cancer incidence is higher than expected in the vicinity of the nuclear sites in Belgium. In addition, the hypothesis of a gradient in thyroid cancer incidence with increasing levels of surrogate exposures (proximity, prevailing wind from the nuclear site, and I-131 doses estimated by mathematical modeling) were explored.
The present study investigated the incidence of thyroid cancer around the four Belgian nuclear sites at the level of the communes over the period 2000–2014. No increased incidence of thyroid cancer was observed around the nuclear power plants of Doel and Tihange. The focused tests and the estimated exposure response curves did not suggest a gradient for thyroid cancer incidence. In contrast, increases in thyroid cancer incidence were found around the nuclear sites of Mol-Dessel and Fleurus, but risk ratios were not significant. For Mol-Dessel, the focused tests and the estimated exposure response curves showed a gradient for thyroid cancer incidence, irrespective of the surrogate exposure (proximity, prevailing wind direction frequency, or radioactive discharge estimates). For Fleurus, a gradient for thyroid cancer incidence was observed with increasing prevailing dominant wind from the site and less clearly with increasing exposure to I-131.
This study is a follow-up study of Bollaerts et al. 2014 [2
] and Bollaerts et al. 2015 [1
] over a prolonged period. In Bollaerts et al. 2014, over the period 2000–2008, no increased thyroid cancer incidence was found within the 20 km proximity area around the nuclear power plants of Doel and Tihange. For the sites of Mol-Dessel and Fleurus, the incidences of thyroid cancer within the 20 km proximity area were higher than expected [2
]. The present 15-years follow up study confirms the previous results. For Mol-Dessel, the results of the focused hypothesis tests and estimated exposure-response curves were far from significant [1
]. The present study which includes a larger number of cases over a longer period showed evidence for a gradient in thyroid cancer incidence with surrogate exposures (proximity, prevailing winds, and radioactive discharges). In the past analyses, for Fleurus, the focused hypothesis tests suggested an increased incidence using radioactive discharge estimates as surrogate and less clearly using prevailing winds [1
]. This is in line with the results of the present study.
We used three measures as surrogate exposures leading to different assumptions and limitations. The first measure was the residential proximity to the source, assuming that the risk of thyroid cancer is decreasing monotonously from the source and that this decrease is independent of the direction (isotropic). The second measure was the prevailing wind direction frequency, assuming that the risk of thyroid cancer is higher when the wind is blowing more often towards the residential commune. The distance is hence not taken into account. Finally, we have further combined both the effect of distance and the effect of wind by modeling the estimated discharges from the plants. This method is the radio-ecologically most plausible, but the potential bias of measurement error may be more pronounced because it combines distance and wind direction misclassifications. In addition to the different surrogates of exposure, different statistical methods were exploited. In particular, three different focused hypothesis tests (i.e., Stone’s test, Bithell’s LRS test with levels of surrogate exposure, and Bithell’s LRS test with ranks) were used, all ranging differently with respect to the trade-off between power and the need to correctly specify the exposure–response relationship. The use of different statistical methods reduces the dependence of the results on the tests assumptions. In the present study, conclusions were strengthened since the methods and globally, the surrogate exposures give similar results. The nuclear power plants of Chooz and Borssele are situated close to the Belgian border, and parts of Belgian territory are within the 20 km proximity area around these installations. Both sites were not included in the present study. The Belgian territory around the nuclear site of Borssele was not included because no Belgian commune has its centroid within the 20 km proximity area. Some Belgian communes have their centroid within the proximity area of Chooz. However, previous analyses showed that results were unstable because of the small population size [2
In the present study, data are compared at the population-level rather than at the individual level so that conclusions cannot be transferred at the individual level. We used cancer incidence data aggregated at the level of the commune (ecological design). Data in the Belgian Cancer Registry are available for the year of cancer diagnosis and the place of residence where the incident case lives at the moment of diagnosis. Hence, measurement error could have occurred because migration phenomena were not taken into account. In general, the misclassification of exposure due to migration is assumed to be non-differential. The aggregation of data causes a loss of information that could lead to an ecological bias. Ecological bias arises from the inability of ecological data to characterize within-area variability in confounders and exposures [16
]: we did not account for individual characteristics which could potentially confound the associations. Moreover, due to the size of the geographical areas, exposure misclassification may arise. Further analyses will be performed at the level of the statistical sectors, the smallest basic areal unit defined by the Federal Public Service Economy, Directorate-General Statistics and Economic Information for which population data and cancer cases become available. Such analyses would reduce ecological bias due to within-area variation of the exposure.
The increased thyroid cancer incidences were observed around Mol-Dessel and Fleurus, the two nuclear sites with research and industrial activities, and not around the nuclear power plants. These differences could be associated with differences in exposure or risk between the two types of sites. Fleurus is one of Europe’s major production sites of radio-iodines. Regarding the level of exposure during the incident (International Nuclear and Radiological Event Scale, INES-rating: 3), an estimated gaseous amount of 48 GBq of I-131 was released to the environment. Regarding earlier exposures, the post-incident investigation indicated points of serious concern with regard to both the operational safety and the management of the Fleurus’ site [2
]. Radio-iodines releases before the incident cannot be ruled out. The Mol-Dessel site, in combining research and industrial nuclear and radiological activities, is and always has been characterized by a great variety of possible radioisotope releases, radio-iodines being one of them. Significant releases of radio-iodines have never been reported to the authorities, but cannot be completely excluded, in particular before the 1990s, when an active surveillance network was not yet available. Illegal radioactive waste storage and treatment activities have taken place in the past in Dessel and might have been at the origin of unauthorized historical releases.
Most of the studies on thyroid cancer incidence and nuclear sites investigated the risk for thyroid cancer among residents living near nuclear power plants. Knowledge about the degree of risk nevertheless remains uncertain. A recent review and meta-analysis found no increased risk of thyroid cancer associated with living near nuclear power plants [17
]. The pooled estimates did not reveal different patterns of risk by gender, exposure definition, or reference population. However, sensitivity analysis by exposure definition showed that living less than 20 km from nuclear power plants was associated with a significant increase in the risk of thyroid cancer in well-designed studies (summary OR = 1.75; 95% CI = 1.17–2.64). Thyroid cancer risk caused by radioactive discharges from nuclear power plants might be too small to be statistically detectable. Although the thyroid is one of the organs most sensitive to ionizing radiation, the measured radiation dose has been reported to be very low near nuclear power plants under normal operating conditions in most countries [18
It was of interest to investigate whether the observed incidence pattern could be compatible with levels of radiation exposure from the sites. However, these levels are often below the detection limit of the routine environmental monitoring in Belgium, by which natural background radiation is predominantly measured. Therefore, we considered emission data and used a model for computing I-131 radioactive discharges estimates since measurements were not available. It should be noted that it is based on strictly hypothetical releases.
A substantial increase in papillary thyroid carcinoma among children exposed to the radioiodine fallout has been one of the main consequences of the Chernobyl reactor accident [4
]. Recently, the investigation of papillary thyroid carcinoma from a cohort revealed an overexpression of the CLIP2
gene among young patients exposed to the post-Chernobyl radioiodine fallout at a very young age [19
]. More recently, mechanistic models were proposed to explore the impact of radiation on the molecular landscape of papillary thyroid carcinoma [20
]. The findings point to a function of CLIP2
as a driver gene in radiation-induced papillary thyroid carcinoma. These models constitute a potential promising interface between molecular biology and radiation epidemiology.
The incidental gaseous release of I-131 at the IRE in Fleurus occurred in 2008. Whereas the study period of the previous study [1
] stopped at the time of the incident, the time window of the available data now allows for the investigation of the health effects of the Fleurus 2008 incident itself. The age- and sex-standardized rates in the 0–20 km proximity area of Fleurus over the periods 2004–2007 and 2008–2014 were not significantly different. Nevertheless, it should be noted that radiation-induced thyroid cancers are characterized by a long latency period, and individual variation in latency time exists [22
]. Thyroid cancer in young children is characterized by a shortened latency period, between 10 and 15 years [23
], and in the very young age (neonates, babies, toddlers), latency could be some five years or even less. However, the thyroid cancer incidence among people less than 20 years of age is not sufficiently high to capture a potential beginning of increase, and the age- and sex-standardized rates in the 0–20 km proximity area of Fleurus over the periods 2004–2007 and 2008–2014 were not significantly different.