Recepción del artículo 25 de febrero, 2004 |
Aprobación 4 de marzo, 2004 |
Primera edición 10 de marzo, 2005 |
Segunda edición, ampliada y corregida 7 de junio, 2021 |
Abstract
Objective: To review the evidence for associations between population mixing, paternal preconceptional irradiation (PPI) and childhood cancer in offspring of male radiation workers. Population and setting: 266 710 children born during 1950-91 in the county of Cumbria, north west England, including 563 cancer cases; 9 859 of the children, including 31 cancer cases, had fathers employed as radiation workers at Sellafield nuclear installation. Methods: Cohort studies, analysed using generalised linear models, allowing for demographic variables. Results: We found (i) an association between population mixing and risk of childhood leukaemia and non-Hodgkin’s lymphoma, (ii) a weaker association between risk of some solid tumours and parental mobility, (iii) a dose-response relation between PPI and risk of leukaemia and non-Hodgkin’s lymphoma which was independent of population mixing, and (iv) no association between PPI and risk of solid tumours. Population mixing could account for the marked excess of leukaemia and lymphoma in young people in the village of Seascale, adjacent to Sellafield. Children of radiation workers had an increased risk of solid tumours, which was largely explained by parental mobility. Conclusions: The cluster of leukaemia and lymphoma in Seascale is likely to be largely due to population mixing, but PPI may also play a part.
Key words
Paternal preconceptional irradiation, childhood cancer, epidemiology, Sellafield
RADIATION AND POPULATION MIXING IN THE AETIOLOGY OF CHILDREN’S CANCER
Background In 1983 a regional television programme first drew attention to a cluster of childhood leukaemias and lymphomas in the village of Seascale, next to the Sellafield nuclear complex which had started operations in around 1950 (figure 1).1 This cluster became the subject of intense public interest and extensive scientific investigation.2,3 It was confirmed that since 1950, the incidence of leukaemia and non-Hodgkin’s lymphoma in children in Seascale had been over 10 times that expected from national rates.4 Concern that the excess could be caused by discharges of radioactive material from Sellafield was not supported by detailed radiological studies.2,3 In 1990, Gardner et al. reported the results of a case-control study of leukaemias and lymphomas in young people born and diagnosed in the west of Cumbria.5 They concluded that fathers’ exposure to external ionising radiation while employed at Sellafield was a risk factor for leukaemia and non-Hodgkin’s lymphoma in children whom they subsequently conceived and that this effect could explain the Seascale cluster. The implications of this study - that paternal preconceptional irradiation (PPI) could affect the health of children subsequently conceived - created great anxiety and stimulated much further research, including our program. Figure 1. The county of Cumbria in north west England. This figure uses Bartholomew digital map data owned and supplied by Bartholomew Ltd and is used with permission.After our work started, other researchers published important potential explanations, which we were able to evaluate within our program of research. Kinlen suggested that the excess of childhood leukaemias in Seascale may have been due to the high level of population mixing there.6 He postulated that the mixing of populations from different areas, bringing together infectious and susceptible individuals, can increase the incidence of childhood leukaemia and non-Hodgkin’s lymphoma. In a series of studies of rural areas with unusually high levels of inward migration, he showed that children living in such areas are at increased risk of leukaemia and non-Hodgkin’s lymphoma and that this effect is more marked in young children.7 The Health and Safety Executive (HSE) reported the findings of a case-control study of all cancers in the children of men employed at Sellafield, in relation to a wide range of possible occupational exposures and demographic characteristics.8,9 They concluded that there was an association between PPI and risk of leukaemia and non-Hodgkin’s lymphoma but that it was seen only in children born in Seascale. Program of work We considered the possible effects of PPI on the risk of leukaemia and non-Hodgkin’s lymphoma10,11 and solid tumours.12 In response to the work of Kinlen, we also carried out independent investigations of the possible effects of population mixing on the risk of leukaemia and non-Hodgkin's lymphoma among children born in Cumbria; in response to the HSE report, we assessed the effect of whether the place of birth was inside or outside Seascale.We also investigated the relation between PPI and the ratio of boys to girls at birth and the risk of stillbirth,14-16 but the findings of these studies are not presented in this review. Methods We used a robust cohort study design, thus avoiding the biases to which case-control studies are subject. The cohort comprised all 274 170 children born in Cumbria between 1950 and 1991, followed up to age 25 years or the end of 1991, thus extending the temporal and geographic boundaries beyond those of Gardner et al.5 Within this cohort, we used a consistent method: first of all we compared the risk of leukaemia/non-Hodgkin’s lymphoma and solid tumours among the 9 859 children whose fathers were radiation workers at Sellafield when they were conceived with the risk among the 256 851 children whose fathers had never worked at Sellafield. Then we investigated whether, among the children of radiation workers, these risks increased as the dose of external radiation which the father received before conception increased (We excluded two small cohorts: children of Sellafield non-radiation workers and children whose fathers first worked at Sellafield after they were conceived). We allowed for possible effects of demographic variables at the level of the individual child - gender, birth order, singleton/multiple birth, paternal social class, parental mobility, time period of birth - and also for confounding with community-level population mixing.We considered risk in relation to all monitored radiation exposures, both external and internal. British Nuclear Fuels plc provided dosimetry data: for each father, the total dose of external ionising radiation up to the time of conception was estimated from annual summaries of the doses recorded prospectively on film badges. Urinalysis data were used to identify children whose fathers were monitored for exposure to natural uranium, plutonium and fission products before conception; these children were compared with children whose fathers were employed at Sellafield but not monitored.We used generalized linear models, both unadjusted and adjusted for demographic variables, to compare the risk for children of radiation workers and other Cumbrian-born children and, among children of radiation workers, to estimate trends in risk with increasing dose of external radiation. Findings By comparing details of the children born in Cumbria with details of registrations of cancers on regional and national children’s cancer registries, we identified 31 cancer cases among children of radiation workers and 532 cases among children whose fathers had never worked at Sellafield.We found that small areas with higher levels of community population mixing also had higher rates of leukaemia and non-Hodgkin’s lymphoma among children born there (rate ratio for trend from lowest to highest observed levels of population mixing = 11.7, 95%CI: 3.2-43).17 Even after allowing for community population mixing, children with both parents born outside Cumbria were at twice the risk (rate ratio = 1.9, 95%CI: 1.0-3.3). The association with population mixing was greater in children under 7 years of age. Population mixing could account for the Seascale cluster but did not exclude the possibility of other factors.Children of radiation workers had a two-fold risk of leukaemia and non-Hodgkin’s lymphoma compared to other Cumbrian-born children (rate ratio = 1.9, 95%CI: 1.0-3.1, p = 0.05), partly because radiation workers living in Seascale were much more mobile than other Cumbrian parents. Within the children of radiation workers, the risk of leukaemia and non-Hodgkin’s lymphoma increased significantly as external PPI increased (rate ratio per 100 mSv = 1.6, 95%CI: 1.0-2.2, p = 0.05), but we found no evidence that PPI was confounded with population mixing. As the dose-response inside and outside Seascale was not significantly different (rate ratio per 100 mSv = 2.0, 95%CI: 1.0-3.1, p = 0.05 and 1.5, 95%CI: 0.7-2.3, p = 0.26 respectively; p for interaction = 0.45), we could not exclude the possibility of an effect of PPI outside Seascale. We concluded that, although a substantial proportion of cases of leukaemia and non-Hodgkin’s lymphoma among children of radiation workers are probably due to population mixing, some may be due to PPI.11Children of radiation workers had a 50% higher risk of solid tumours than other children born in Cumbria, (rate ratio = 1.5, 95% CI: 0.9-2.4, p = 0.09), which was not statistically significant and which was partly explained by different patterns of migration among radiation workers and other Cumbrian parents.12 The risk of a major category of solid tumours (all such tumours except brain and spinal tumours, Hodgkin’s disease and gender-specific tumours) was higher among children whose parents were incomers to the area, and radiation workers included a higher proportion of incomers than other Cumbrian parents. Among the children of radiation workers, fathers’ occupational exposure to external radiation before conception was not a risk factor for these tumours.There was no significantly increased risk of cancer in any of the prespecified diagnostic groups for children whose fathers were monitored for plutonium or fission products before conception (4 880 and 2 165 children, respectively). Children whose fathers were monitored for exposure to natural uranium had no increased risk of solid tumours, but were at increased risk of leukaemia and non-Hodgkin’s lymphoma, even after adjustment for external PPI (adjusted rate ratio = 2.9, 95%CI: 0.6-9.8, p = 0.15, based on 3 cases among 667 children); this finding may be due to confounding of exposure to external radiation and internal uranium or to chance rather than a real effect. Debates Population mixingWe confirmed the findings of Kinlen7 –that population mixing increases the risk of childhood leukaemia and non-Hodgkin’s lymphoma– but in a different population. Our study estimated the increased risk to be much greater than had previously been thought, possibly because we considered exposure to population mixing around the time of birth which may be a more critical time than later in life and possibly because we had accurate estimates of the birth populations in small areas. Recently, high rates of childhood leukaemia around the La Hague nuclear reprocessing plant in northern France have been related to population mixing.18 As in our study, the risk was stronger in younger children. Like Sellafield, this installation attracted a continuing influx of construction workers and nuclear workers to an isolated rural area over a period of several decades.19In contrast, our finding of an association between parental mobility and risk of solid tumours has not been confirmed in other populations20 and may have been a chance finding or a consequence of confounding with an unmeasured exposure.Paternal preconceptional irradiationLike us, other researchers have not found any association between PPI and risk of solid tumours.5,21,22 However, the debate over a possible association between PPI and risk of leukaemia and non-Hodgkin’s lymphoma remains unresolved. Although we used a cohort rather than a case-control design and studied a wider area over a longer time period, our findings were not independent of those of Gardner et al., as 10 of the 13 cases of leukaemia and non-Hodgkin’s lymphoma in children of radiation workers in our study were also included in Gardner’s study.5 When we excluded these Gardner cases, the dose-response remained non-significantly elevated, based on 3 cases.HSE studyOur findings were not consistent with those of the HSE case-control study, which concluded that the statistical association between risk of leukaemia and non-Hodgkin’s lymphoma and PPI was confined to children born in the village of Seascale, where the dose-response was extremely high and very significant.8,9 In contrast, our birth cohort study, investigating largely the same cases, concluded that the dose-response was much lower and was not confined to Seascale.10 We therefore made a detailed comparison of the design, data and analyses of the two studies.23 In their interpretation, HSE put most emphasis on analyses which treated PPI dose as categorical (4 categories). In contrast, our interpretation relied on analyses of dose as a continuous variable, since categorical analyses ignore detailed information about the known distribution of dose within categories.The distribution of offspring-years over categories of dose and residence at birth (inside or outside Seascale) was very different in the HSE study and among the children of radiation workers in our study. The HSE study was based on a sample of 179 controls, whereas our study was based on all 9 859 births to radiation workers. In particular, although neither study had any cases over 200 mSv, the Cumbrian study had 512 controls (4% of person-years) in this dose range, whereas the HSE study had none. Further, the HSE study excluded a high-dose case born outside Seascale in 1988. These differences resulted in the HSE estimate of dose-response being higher inside Seascale and lower outside Seascale (odds ratio per 100 mSv = 40.5, 95%CI: 6.3-263, p ≤ 0.001 and 0.6, 95%CI: 0.2-2.3, p = 0.44 inside and outside Seascale respectively, based on categorical analysis) than the estimates of the Cumbrian study (rate ratio per 100 mSv = 2.0, 95%CI: 1.0-3.1, p = 0.05 and 1.5, 95%CI: 0.7-2.3, p = 0.26 inside and outside Seascale respectively, based on continuous dose). Overall, we believe that the results of our cohort study are more robust than those of the HSE case-control study.The COMARE seventh reportThe United Kingdom government Committee on Medical Aspects of Radiation in the Environment (COMARE) recently updated its review of the evidence from both animal and human studies for an association between PPI and cancer risk in offspring.24 They concluded that carcinogenic effects of PPI have been demonstrated in mice25-28 but that there was substantial variation between strains and the magnitude of the effect was not sufficient to account for the associations reported in the Sellafield workforce.They also assessed three major epidemiological studies which reported shortly before our study was completed. The Record Linkage Study was a case-control study in England, Scotland and Wales, which identified parents of children who developed cancer and of matched control children.22 Record linkage techniques were then used to determine which of these parents were included on the National Registry for Radiation Workers. Hence the study largely included the same cases as ours. Consistent with our findings, it reported an increased risk of leukaemia and non-Hodgkin’s lymphoma among children of radiation workers. However, when the Gardner cases were excluded, it found no association with PPI. In the Nuclear Industry Families Study, the children of 36 050 men and women employed in the nuclear industry in the United Kingdom during 1993-1996 were identified by a postal questionnaire.21 This study included cancers identified during the later part of our study, but not the earlier cases who tended to have higher PPI doses. Overall, they found a significant association between PPI and risk of leukaemia and non-Hodgkin’s lymphoma which was not significant when Cumbrian-born cases were excluded. The Three Site Study was a case-control study of 233 cases of childhood cancer around three Department of Energy facilities in the United States.29 For analyses in relation to PPI, these cases were compared with control children whose fathers had worked at one of the facilities before the child was conceived. The relative risk for leukaemia in relation to PPI was above one only among children whose fathers worked at one site (Hanford) and this estimate had a wide confidence interval (relative risk per 100 mSv = 1.93, 95%CI: < 0.01-29). However, overall the cases and controls had an average dose of 11 mSv –much lower than the average dose of 58 mSv in the Sellafield studies.30 In the United Kingdom, an association between PPI and leukaemia risk in offspring has been found only in those analyses which included the Sellafield workforce, possibly because other occupationally exposed workers have had much lower exposures than the Sellafield workforce. COMARE concluded that the explanation for the Seascale cluster remains uncertain – it may relate to lifestyle factors such as population mixing.The Sellafield workforce has received high exposure to radiation compared to others in the nuclear industry. Further, the doses received by Sellafield employees, especially in the 1960’s and 1970’s, were considerably higher than current radiation doses received by employees. The power of a study to detect an effect depends on the number of subjects and the distribution of the exposure among them; larger studies with more highly exposed subjects are more likely to detect an effect if one exists. In the absence of studies of nuclear workforces with collective exposures comparable to those at Sellafield, it may be impossible to perform an informative independent assessment of the relationship between PPI and the risk of childhood leukaemia. Updates Children of atomic bomb survivorsThe only other studies of heritable effects of radiation exposure in man of the scale of our studies are those of the children of the survivors of the atomic bombs in Hiroshima and Nagasaki. A recent study analysed mortality among children born between 1 month and 38 years after the bombings who had one or both parents in Hiroshima or Nagasaki at the time of both the bombings and child’s birth.31 It found that the rate of death from cancer, both before and after 20 years of age, was no higher among children of exposed parents (either fathers or mothers) than among other subjects and that mortality did not increase with increasing dose. Nevertheless, the authors concluded that an increase in cancer mortality among children of exposed parents could not be ruled out at this stage.While these children were more highly exposed than children of Sellafield radiation workers in terms of collective dose, the number of offspring reported to have PPI over 10 mSv was lower. For several reasons, direct comparison of studies of these cohorts is difficult. The Japanese study considered cancer mortality whereas our studies considered incidence of cancer. Preconceptional radiation dose estimates for Sellafield employees were measured prospectively using film badges which were worn by the men and changed about once a month, whereas dose estimates for the atomic bomb survivors were modelled retrospectively from reports of the position of the survivor at the time of bombing. Doses of Sellafield employees typically accumulated over several years, whereas doses of atomic bomb survivors were assumed to have been received within 100 seconds and so may have had different effects. The Sellafield employees were all of reproductive age when they were irradiated, whereas some of the atomic bomb survivors were pre-pubescent in 1945. For these reasons, it is difficult to draw any strong scientific conclusions from the studies of offspring of the atomic bomb survivors.Genetic mechanismsEvidence is accumulating for an elevated germline mutation rate in parents directly exposed to ionizing radiation and a number of chemical mutagens. Radiation may also have a direct effect on genome stability which is transmitted through the germ line of irradiated parents to their offspring, who show elevated mutation rates.32,33 Conclusions Evidence from many different populations confirms a robust association between exposure to population mixing and an increased risk of childhood leukaemia and non-Hodgkin’s lymphoma, especially in young children. The cluster of childhood leukaemia and lymphoma in Seascale is likely to be largely due to population mixing, but paternal preconceptional irradiation may also play a part. Although children of radiation workers at Sellafield had an increased risk of solid tumours compared to other children born in Cumbria, most of this excess could be explained by parental mobility. Acknowledgements: We thank Dr. Trevor Dummer for drawing the map. |