Estimates of the cancer burden in Europe from radioactive fallout from the Chernobyl accident
Article first published online: 20 APR 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 6, pages 1224–1235, 15 September 2006
How to Cite
Cardis, E., Krewski, D., Boniol, M., Drozdovitch, V., Darby, S. C., Gilbert, E. S., Akiba, S., Benichou, J., Ferlay, J., Gandini, S., Hill, C., Howe, G., Kesminiene, A., Moser, M., Sanchez, M., Storm, H., Voisin, L. and Boyle, P. (2006), Estimates of the cancer burden in Europe from radioactive fallout from the Chernobyl accident. Int. J. Cancer, 119: 1224–1235. doi: 10.1002/ijc.22037
- Issue published online: 7 JUL 2006
- Article first published online: 20 APR 2006
- Manuscript Accepted: 30 MAR 2006
- Manuscript Received: 8 MAR 2006
- Swiss Federal Office of Public Health
- Cancer Research UK
- ionizing radiation;
- Chernobyl accident;
- radiation risk;
- time trends;
- cancer incidence;
- cancer mortality
- Top of page
- Material and methods
- Supporting Information
The Chernobyl accident, which occurred April 26, 1986, resulted in a large release of radionuclides, which were deposited over a very wide area, particularly in Europe. Although an increased risk of thyroid cancer in exposed children has been clearly demonstrated in the most contaminated regions, the impact of the accident on the risk of other cancers as well as elsewhere in Europe is less clear. The objective of the present study was to evaluate the human cancer burden in Europe as a whole from radioactive fallout from the accident. Average country- and region-specific whole-body and thyroid doses from Chernobyl were estimated using new dosimetric models and radiological data. Numbers of cancer cases and deaths possibly attributable to radiation from Chernobyl were estimated, applying state-of-the-art risk models derived from studies of other irradiated populations. Simultaneously, trends in cancer incidence and mortality were examined over time and by dose level. The risk projections suggest that by now Chernobyl may have caused about 1,000 cases of thyroid cancer and 4,000 cases of other cancers in Europe, representing about 0.01% of all incident cancers since the accident. Models predict that by 2065 about 16,000 (95% UI 3,400–72,000) cases of thyroid cancer and 25,000 (95% UI 11,000–59,000) cases of other cancers may be expected due to radiation from the accident, whereas several hundred million cancer cases are expected from other causes. Although these estimates are subject to considerable uncertainty, they provide an indication of the order of magnitude of the possible impact of the Chernobyl accident. It is unlikely that the cancer burden from the largest radiological accident to date could be detected by monitoring national cancer statistics. Indeed, results of analyses of time trends in cancer incidence and mortality in Europe do not, at present, indicate any increase in cancer rates—other than of thyroid cancer in the most contaminated regions—that can be clearly attributed to radiation from the Chernobyl accident. © 2006 Wiley-Liss, Inc.
The Chernobyl nuclear power plant, in Ukraine approximately 10 km south of Belarus, experienced a major accident on April 26, 1986, resulting in the release of several types of radionuclides, including (1.2–1.8) × 1018 Bq of short-lived 131I and about 1.4 × 1017 Bq of long-lived 134Cs and 137Cs.1 This is the most severe radiological accident to date.2
Long-range transport of these and other radionuclides has caused serious contamination of the regions close to the Chernobyl power plant and dissemination of radionuclides throughout Europe, as well as to other continents. Deposition was highest in Belarus, Ukraine, and the western part of the Russian Federation, where ingestion of food contaminated with radioactive iodine resulted in inhabitants (particularly children) receiving considerable doses to the thyroid.
Epidemiological studies focusing on the most contaminated regions of the 3 most affected countries have confirmed a causal relationship between the observed increased risk of thyroid cancer and exposure to radioactive iodines from the Chernobyl fallout among those who were children or adolescents when the accident happened.3, 4, 5 Other types of cancer, including leukemia, have also been investigated,1, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 but as yet no association with radiation exposure has been clearly demonstrated. Recent studies suggest a possible doubling of the risk of leukemia among Chernobyl cleanup workers18 and a small increase in the incidence of premenopausal breast cancer19 in the most contaminated districts (with average whole-body doses above 40 mSv), both of which appear to be related to radiation dose. These findings need confirmation in further epidemiological studies with careful individual dose reconstruction.
The full extent of the health impact of Chernobyl on the population is difficult to gauge. Ten years ago, Cardis and collaborators20 estimated that about 9,000 deaths from cancers and leukemia might be expected over the course of a lifetime in the most exposed populations in Belarus, the Russian Federation and Ukraine. The objective of the present study was to evaluate the human cancer burden in Europe as a whole from radioactive fallout from the Chernobyl accident. The exposure of Chernobyl cleanup workers is not considered here.
Material and methods
- Top of page
- Material and methods
- Supporting Information
There are several approaches to estimating the cancer burden in Europe from Chernobyl, including using risk projection models and studying both incidence rates and mortality rates. We used all these approaches in the current study and based our overall assessment on comparisons of the 3.
Sources of data
The present analysis focused on 40 European countries*. These countries constitute the whole of what is defined geographically as Europe, except the Caucasus, Turkey, Andorra and San Marino. In the Russian Federation, only the 4 most contaminated regions (Bryansk, Kaluga, Orel and Tula, which make up only a small fraction of the territory of that country) are included. Information on population demographics, cancer incidence and mortality was compiled from a variety of sources, as described in Table A1 (electronic supplementary material).
An assessment of radiation doses received during the first year after the accident was undertaken in most European countries shortly after the accident occurred in 1986. These estimates were summarized in the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 1988 report.21 For the purpose of the present article, these estimates needed to be updated for the following reasons. First, the results presented in the UNSCEAR 1988 report were based on information then available and were related to exposure only during the first year after the accident. A large number of radiation measurements conducted in the environment, foodstuffs, and humans since 1987 have provided important information on which to base estimates of doses accumulating between 1986 and 2005. Second, a comprehensive monitoring program was undertaken in Europe (except some Balkan states and Iceland) during 1992–1996 to reconstruct deposition of 137Cs and to create an Atlas on 137Cs deposition in Europe.22 Third, improved metabolic models describing the kinetics of radionuclides in the human body have become available since 1987.23, 24, 25, 26, 27, 28
Within the framework of the current project, average whole-body doses to adults as a result of external irradiation from radionuclides deposited on the ground surface and intake of long-lived isotopes, notably 134Cs and 137Cs, were estimated for 1986–2005 and projected to 2065. Age-dependent doses to the thyroid from inhalation and ingestion of 131I within the first 2 months after the accident (when nearly all the intake of 131I occurred) were also estimated. Doses to active bone marrow and to the breast from external irradiation and intake of long-lived isotopes were not estimated separately, as they are similar to the whole-body dose.24, 27 In this article doses are expressed in mSv as equivalent doses, with a radiation weighting factor of 1 for external γ exposure and for β- and γ-radiation emitted by 131I, 134Cs and 137Cs.
Dose reconstruction was based on available country-specific radiation monitoring data and estimates of exposure levels to populations. Dose evaluation for Belarus, Ukraine and the Russian Federation made use of detailed information provided by local experts or published in UNSCEAR and the UN Chernobyl Forum Report.1, 29 For less affected European countries, information was either obtained from local experts or authorities (Bulgaria, Czech Republic, Finland, Lithuania, and Switzerland) or from relevant publications.30, 31, 32, 33, 34, 35, 36, 37, 38 Data on 137Cs deposition density and time-integrated activity of 131I and 137Cs in foodstuffs in the first year after the accident were available in the Atlas on 137Cs Deposition in Europe22 and in the UNSCEAR 1988 report,21 respectively. This information, together with dose coefficients for inhalation26 and ingestion23, 24, 25 and conversion factors for external exposure27, 28 and data on population size and structure in 1986–2005 (http://globalis.gvu.unu.edu/), was used to derive country- and region-specific estimates of whole-body doses and of thyroid doses from 131I.
The following country- and region-specific input data were used to estimate doses: population-weighted 137Cs ground deposition density; radionuclide composition in depositions; time-integrated activity of radionuclides in foodstuffs, including milk and milk products, leafy vegetables, grain products, vegetables and fruits and meat; age-dependent consumption rates of milk and leafy vegetables; population structure; and protective actions applied in order to reduce the exposure of the local populations. In some countries (Albania, Bosnia and Herzegovina) insufficient data were available for dose reconstruction, and interpolation between neighboring countries was used to derive necessary data. Average doses were estimated separately for each country, for each of the most contaminated regions of Belarus, the Russian Federation and Ukraine, as well as for regions covered by the cancer registries included in the time trend analyses.
Uncertainties in doses were subjectively evaluated on the basis of availability and reliability of radiation data used for dose reconstruction. Uncertainty was considered smallest in Belarus, Ukraine and the Russian Federation, where a comprehensive dose reconstruction has been done based on results of intensive radiation monitoring. Higher degrees of uncertainty were assigned to other countries, particularly when obtained from very limited data (Albania, Bosnia and Herzegovina, Iceland, Macedonia).
Risk projection models
Projections of the number of cancer cases and of the number of cancer deaths possibly attributable to fallout from Chernobyl were based on risk models recently developed by the U.S. National Research Council's Committee on the Biological Effects of Ionizing Radiation (BEIR VII).39 The BEIR VII risk models are a combination of excess relative risk (ERR) and excess absolute risk (EAR) models, both of which are written as a linear function of dose, depending on sex, age at exposure and attained age. The BEIR VII risk models were derived from analyses of data on the Japanese atomic bomb survivors for all cancer sites except breast and thyroid; for the latter, they were based on published combined analyses of data on the atomic bomb survivors and medically exposed cohorts.40, 41 To estimate risks from exposure at low doses and dose rates, a dose and dose-rate effectiveness factor (DDREF) of 1.5 was used for all outcomes except leukemia.
Predictions were made for leukemia, breast cancer, thyroid cancer and all cancers other than leukemia, thyroid and nonmelanoma skin cancer for the 40 countries listed above; for the Russian Federation, predictions were made only for the most heavily contaminated regions. Estimates for all cancers other than leukemia, thyroid and nonmelanoma skin cancer were based on the BEIR VII approach for all solid cancers, as it was assumed that there was no excess risk of lymphoma or multiple myeloma.1 Cases arising in people born after the accident were included in the predictions based on doses from long-lived isotopes and from external exposures accumulated after birth.
The population distributions for 1986 and the life table, mortality and cancer rates for each country for 2000 were used in the analyses. For Belarus, Ukraine and the contaminated regions of the Russian Federation, where rates of thyroid cancer in 2000 reflected a large radiation-related increase, the baseline rates were estimated as the population-weighted average of rates in neighboring countries with established cancer registries (Czech Republic, Estonia, Finland and Lithuania).
In projecting risks related to radiation exposure over time, it was assumed that the population demographics (including the age and sex distributions of the population and life table) and cancer incidence would remain constant from 1986 to 2065.
As in the BEIR VII report, the risk estimates presented in this article are accompanied by subjective uncertainty intervals that quantify the most important uncertainty sources: (i) sampling variability in risk model parameter estimates from the atomic bomb survivor data, (ii) uncertainty in the appropriate value of a DDREF and (iii) for all solid cancers and leukemia, uncertainty in using Japanese atomic bomb survivors to estimate risks in populations with different baseline risks. These uncertainties were evaluated as described in BEIR VII.39 In the current article, uncertainty in reconstructed dose estimates is also included.
Trends in cancer incidence and mortality
Annual cancer incidence rates were estimated using data for 1981–2002 from 22 registries (from 15 countries, including 11 national registries—see Table A1 and Figure A1, electronic supplementary material) using Poisson regression, adjusting for registry, age at diagnosis, sex and their interaction, taking into account extra Poisson variation with a scale parameter.42 Rates were standardized directly to the standard world population.43
Analysis of trends in mortality in 30 European countries (Table A1) was based on data from a recent study of mortality and smoking in developed countries 1950–2000.44 For ages 0–34, age-, sex- and cause-specific mortality rates were available for each year for 1980–2000. For ages 35–69, mortality rates were available for 1980, 1985, 1990, 1995 and 2000 by age, sex and cause, and, in addition, cancer mortality attributed and not attributed to smoking was available separately, calculated using the method of Peto et al.45 From these data, standardized mortality rates were calculated that were directly standardized to the world population.
Trends in mortality and cancer incidence when age at diagnosis was at least 70 years were not considered, as it is recognized that they are often subject to large artificial trends because of changes in the extent to which cancers in older people are diagnosed and recorded.46 Cancer groupings and types a priori of interest were: childhood cancers, thyroid cancer, all cancers excluding nonmelanoma skin cancer, leukemia and breast cancer. For mortality, cancers attributed to and not attributed to smoking45 were also considered, but not thyroid cancer, as this disease is rarely fatal.
The effect of dose on cancer incidence was examined using a Poisson regression model allowing for a registry effect, in addition to age at diagnosis (in 5-year categories), sex and their interactions. Whole-body dose was calculated annually as the cumulative dose since 1986, allowing for a lag time of 2 years for leukemia and 5 years for all other cancers, as has been done in studies of atomic bomb survivors.47 For thyroid cancer, age-specific (0–4, 5–9, 10–14, 15–29, 30–59 and ≥60 years) dose to the thyroid in 1986 was used, allowing for a 5-year lag.
Analyses adjusting for age at diagnosis, sex and registry were also conducted, testing for a change in the slope of the time trends after the Chernobyl accident using linear contrasts,48 in which the increase in cancer incidence during 1981–1990 (the period prior to and up to 5 years after the Chernobyl accident, allowing for a lag of 5 years after the accident before a possible effect on cancer rates could occur) was compared to that for 1991–2000.
To test for differing trends in mortality over time between countries, or groups of countries, with different average doses from Chernobyl, the numbers of deaths and of person-years were also stratified according to age in 5-year groups, sex, calendar year and country. For each country (or group of countries), the linear trend in the mortality rate over calendar year was estimated, assuming that the number of deaths in each age-sex group in each year had a Poisson distribution with mean depending on the relevant number of person-years. Tests for an interaction between the trend with calendar year and dose were carried out by an F test after carrying out linear regression of the estimated trend in each country on the country-specific mean dose. This method was used because it was found that there was substantial random variability between the calendar-year trends in the different countries, even after allowing for the extra-Poisson variation accounted for by a negative-binomial distribution.
The study did not include any contact with individual study subjects, and no individuals could be identified from the data used in the analyses.
- Top of page
- Material and methods
- Supporting Information
Radiation doses resulting from the Chernobyl accident
The spatial distribution of average country-specific whole-body doses from internal and external exposure from long-lived radionuclides for 1986 is shown in Figure 1a, and the cumulative dose from 1986 to 2005 is shown in Figure 1b. The latter represents, on average, about 85% of the lifetime dose from Chernobyl that would be accumulated by a person who lived until 2065. Country- and, where appropriate, region-specific average doses used in the risk predictions are shown in Figure A2 in the electronic supplementary material. The distribution of average doses in the geographical areas used in the analyses of cancer incidence trends and in the countries included in the analyses of mortality trends are shown in Figure A3. In 1986 doses were highest in Belarus and Ukraine, with the average cumulative whole-body dose exceeding 0.5 mSv. By 2005 the average cumulative country-specific whole-body doses were respectively, 2.8 mSv in Belarus, 5.1 mSv in the most contaminated areas of the Russian Federation and 2.1 mSv in Ukraine; it was 1.4 mSv in Finland. The average cumulative dose through 2005 in the Gomel region of Belarus and in the Bryansk region of the Russian Federation was estimated to be about 20 times higher (around 10 mSv) than that for Europe as a whole (0.5 mSv). For comparison, in Europe the average individual whole-body dose accumulated from natural background radiation, excluding radon, over that period is about 20 mSv. The geometric standard deviation (GSD) of the average cumulative whole-body dose estimates was evaluated to range from 1.1 in the most contaminated areas (Belarus, the 4 regions of the Russian Federation and Ukraine) to 1.6 in the least contaminated countries.
The spatial distribution of doses to the thyroid from 131I is shown in Figure 2 for children younger than 5 years and for adults at least 30 years old in 1986. The country- or, where appropriate, region-specific doses used in the risk predictions are shown in Figure A4, whereas Figure A5 shows the distribution of average doses in the geographical areas used in the analyses of cancer incidence trends. Doses to the thyroid from 131I were considerably higher than whole-body doses from external and internal exposure to long-lived radionuclides. The highest average doses to the thyroid were received in the Gomel region of Belarus (630 and 150 mSv for young children and adults, respectively), in the Bryansk region of the Russian Federation (180 and 25 mSv, respectively) and in the Zhytomir region of Ukraine (150 and 40 mSv respectively). Doses to young children were consistently higher than doses received by adults. The GSD of the average thyroid dose estimates was evaluated to range from 1.2, in the most contaminated areas, where they were based on direct thyroid measurement (the Gomel region of Belarus, the Bryansk region of the Russian Federation and the Kiev and Zhytomir regions of Ukraine) to 2.0, in the least contaminated countries.
Comparisons of the doses with those estimated and published by UNSCEAR21 will be discussed in detail in another article. The largest changes concerned thyroid doses in the 3 most affected countries. Finland, Austria, Greece and Liechtenstein now appear to have had the highest average whole-body doses outside Belarus, the Russian Federation and Ukraine.
Cancer risk projections
The numbers of cases and deaths predicted to occur between 1986 and 2065 (the latter will be 80 years after the accident) from radiation exposure from the Chernobyl accident are shown in Table I for all cancers (excluding leukemia, thyroid cancer and nonmelanoma skin cancer), leukaemia and breast cancer and in Table II for thyroid cancer. Corresponding estimates for 1986–2005 are shown in Tables III and IV. Also shown are the numbers of cancer cases predicted to occur from causes unrelated to the Chernobyl accident, as well as the estimated attributable fractions, defined as the percentage of the total cancer burden that could be expected from the Chernobyl accident over the same periods.
|Country group||Average whole-body dose (mSv) 1986–2005||Population (in millions) in 1986||All cancers other than leukemia, thyroid and nonmelanoma skin cancers||Leukemia||Breast cancer|
|From radiation||95% UI||From other causes||AF to 2065||From radiation||95% UI||From other causes||AF to 2065||From radiation||95% UI||From other causes||AF to 2065|
|Country group||Average thyroid dose (mSv)1||Exposure at all ages||Exposure before age 15|
|Population (in millions) in 1986||From radiation||95% UI||From other causes||AF to 2065||Population in 1986||From radiation||95% UI||From other causes||AF to 2065|
|Country group||Average whole-bodydose (mSv) 1986–2005||Population (in millions) in 1986||All cancers other than leukemia, thyroid and nonmelanoma skin cancer||Leukemia||Breast cancer|
|From radiation||95% UI||From other causes||AF to 2005||From radiation||95% UI||From other causes||AF to 2005||From radiation||95% UI||From other causes||AF to 2005|
|Country group||Average thyroid dose (mSv)1||Exposure at all ages||Exposure before age 15|
|Population (in millions) in 1986||From radiation||95% UI||From other causes||AF to 2005||Population (in millions) in 1986||From radiation||95% UI||From other causes||AF to 2005|
The total predicted number of cases possibly attributable to Chernobyl in Europe (whose population was more 570 million people in 1986) up to 2065 is large in absolute terms, about 23,000 for all cancers excluding leukemia, thyroid cancer and nonmelanoma skin cancer (including 4,500 cases of breast cancer) and 2,400 for leukemia (Table I). An additional 16,000 cases of thyroid cancer are predicted from 131I exposure (Table II). The predicted number of deaths up to 2065 is about 14,000 for all cancers excluding leukemia, thyroid cancer and nonmelanoma skin cancer (including 2,000 from breast cancer) and about 1,700 for leukemia. These estimates are subject to substantial uncertainty, as reflected by the 95% uncertainty intervals.
The estimated attributable fractions (AFs) are very small compared to the background number of cases expected in the absence of exposure from Chernobyl. For all cancers excluding leukemia, thyroid cancer and nonmelanoma skin cancer, they range from 0.002%, in the least contaminated countries (with a cumulative whole-body dose of less than 0.2 mSv by 2005), to 0.23%, in the most contaminated countries (with a cumulative whole-body dose of at least 3 mSv). For leukemia, they range from 0.01%, in the least contaminated countries, to 0.66%, in the most contaminated countries.
For all cancers combined (excluding leukemia, thyroid cancer and nonmelanoma skin cancer), leukemia and breast cancer, more than half the possible excess of cases is expected in countries where the average dose is at least 1 mSv, which represent only 11% of the entire population under consideration. Of the total number of excess cases predicted to 2065, about 14% of all cancers, 40% of leukemia cases and 6% of breast cancers are predicted to have occurred in the first 20 years after the accident.
For thyroid cancer, the estimated AFs range from 0.08%, in the countries least contaminated by 131I (with an average thyroid dose of less than 5 mSv), to 12%, in the most contaminated countries (with an average thyroid dose of at least 100 mSv), and 31%, in the Gomel region of Belarus (not shown). The uncertainty intervals for the estimates are particularly wide, ranging from about 3,400 to 72,000 excess thyroid cancer cases in Europe by 2065. About half the possible excess cases are predicted to occur in countries with an average dose of at least 25 mSv, which represent only 3% of the population under study. Of the total excess cases predicted up to 2065, about 6% are predicted to have occurred in the first 20 years after the accident. The vast majority (90%) of the predicted radiation-induced cases of thyroid cancer are expected among those who were younger than 15 years of age when the Chernobyl accident occurred.
Trends in cancer mortality
Mortality rates from all cancers combined have tended to decrease in most European countries in the age group 0–14 years during the last 2 decades (Fig. A6). Decreasing trends with calendar year occurred in all 4 dose groups for all cancers combined, as well as for leukemia and for all other cancers considered separately (Fig. 3, top panels). An analysis of the trend over 1985–2000 in all 30 countries, taking into account average dose in each country, showed no significant association between the rate of decrease and average dose (p > 0.05).
Substantial decreases in the mortality rate from all cancers combined have also occurred in many European countries among those 15–34 years of age at death (Fig. A7). Decreases were seen in the groups of countries with average doses of <0.2, 0.2- and 0.5- mSv; in the group with average doses of 1.0 mSv or more, however, mortality from all cancers combined showed no decreasing trend with calendar year from 1985 to 2000 (Fig. 3, bottom left-hand panel). An analysis of the data for 1985–2000 from all 30 countries, taking into account average dose in each country, showed a significant interaction between calendar year and dose, with countries with higher mean doses having a smaller decrease in mortality with succeeding calendar year (p = 0.03). Further analyses show that this interaction was not driven by leukemia (Fig. 3, bottom middle panel) or by any of the specific cancers for which mortality data were available for analysis (lymphomas or colorectal, stomach or lung cancers; data not shown). When the trends over time among individual countries were examined, it could be seen that the interaction was entirely due to Ukraine, where cancer mortality rates in this age group tended to increase over this period. In all the other countries included in the analyses they either decreased or tended to decrease. Mortality rates in this age group from causes other than cancer have also been increasing over the same period in Ukraine.
Among those 35–69 years of age at death, cancer mortality rates from all cancers combined differed substantially between males and females in many European countries and also differed substantially from country to country, with large trends in some countries, especially in males (Fig. A8). These differences appear to be chiefly a result of changes in the effects of smoking, which in many countries cause about half of all cancer deaths in men in this age range.49 When cancer mortality attributed and not attributed to smoking was considered separately, it was clear that the large trends are chiefly in cancer mortality attributed to smoking, with trends in mortality not attributed to smoking being much smaller (Figs. A9 and A10). Nevertheless, a small increase in mortality from cancer not attributed to smoking was seen in countries with an average dose of 1.0+ mSv in both men and women during the 1985–2000 period (Fig. 4, right-hand panels). When the trends in cancer not attributed to smoking were examined for each of the 30 countries, it was apparent that the relative increase was a result of increases in both men and women in Ukraine and Belarus. For both of these countries all-cause mortality not attributed to smoking also increased from 1985 to 2000.
Trends in cancer incidence
Incidence rates for all cancers excluding nonmelanoma skin cancer have tended to increase in most European countries since 1981 (Fig. A11). Figure 5 shows temporal trends in the incidence of all cancers excluding thyroid and nonmelanoma skin cancers for 1981–2002 classified by age at diagnosis into 3 groups (<15, 15–34 and 35–69 years) and by registry according to average cumulative whole-body dose. Although increasing trends can be seen over time in all age at diagnosis and dose groups, the rate of increase appears higher in the 2 highest-dose groups. In these dose groups, however, the incidence in those 15 years and older at diagnosis was lowest in 1986 and by 2002, despite the steeper increase, had remained lower than or comparable to rates in the lower-dose groups. Similar trends can be seen for the incidence of all cancers excluding nonmelanoma skin cancer (Fig. A12). Increases in the incidence of breast cancer and, less consistently, in the incidence of leukemia also were observed in Europe between 1981 and 2002 (Figs. A13 and A14).
Figure 6 shows the trends in the incidence of thyroid cancer by age at diagnosis and by registry grouped by thyroid dose (trends in incidence by country are shown in Fig. A15). Increasing trends can be seen in the 3 higher-dose groups. The increase was greatest among those exposed in childhood and was most pronounced in registries experiencing the highest dose. A decrease in the incidence of cancer in those who were 0–14 years at diagnosis was noted after 1995, when most of those who had been children when the Chernobyl accident occurred had become adolescents and young adults.
An analysis of the data from all registries for 1981–2002 indicated statistically significant associations (p < 0.05) between the average dose in each registry and the incidence of all cancers, of leukemia, of breast cancer, of thyroid cancer and of all cancers excluding thyroid and nonmelanoma skin cancer, once the effects of age at diagnosis, sex, registry and calendar year were taken into account. Analyses of linear contrasts, however, indicated that although the incidence of all cancer groupings of interest has been increasing in Europe since 1981, the slope of this rise actually decreased after 1991 for all cancers, breast cancer and leukemia. This reduction was statistically significant except for leukemia. Only for thyroid cancer was a statistically significant increase in the slope of the trend noted after 1991 (p < 0.0001).
- Top of page
- Material and methods
- Supporting Information
This article provides an assessment of the cancer burden in Europe due to radioactive fallout from the Chernobyl accident. Strengths of our analysis include having performed a systematic review and synthesis of available data on radionuclide exposures, culminating in an evaluation of the spatial distribution of whole-body and thyroid doses in Europe. Our analysis includes predictions of the number of cancer cases and cancer deaths due to radiation from the Chernobyl accident to 2065 using risk projection models based on the long-term experience of other populations exposed to radiation,39 as well as an evaluation of trends in cancer incidence and mortality before and after the Chernobyl accident and by radiation dose. Weaknesses include the ecological nature of the trend analyses, where exposure to radiation and to potential confounding factors are not known at the individual level and where cancer registration has been increasing in quality over time in some countries. In addition, the risk models used for predictions were based on different populations exposed to higher doses for brief periods and may not be generalizable to circumstances of very low cumulative doses delivered over decades.
It is estimated that to date, exposure to radiation from Chernobyl may have caused about 3,000 cases of cancer other than leukemia, thyroid and nonmelanoma skin cancer in Europe (with an uncertainty range of 1,400–7,700 cases), about 0.008% of the total number of cancer cases in Europe during this period. The number of cases of leukemia that are possibly a result of the Chernobyl accident is estimated as about 950, similar to the estimated number of cases of thyroid cancer. Apart from leukemia, for which 40% of the cases are expected to have occurred already, only a small proportion of the cases that can possibly be attributed to Chernobyl are predicted to have occurred to date. This is because of the relatively long average latency (10–20 years or more) between exposure and occurrence of most types of radiation-related cancers. Further, based on the experience of other populations exposed to radiation in other settings, risk appears to be largest after exposure at young ages and does not become apparent until later in life.
Because of the very small doses generally received outside Belarus, the Russian Federation and Ukraine, the small proportion of cancers estimated to be attributable to these exposures and the fact that most radiation-related cases are expected to occur 20 years or more after exposure, epidemiological studies to date (including analyses of time trends) have had very little power to detect an effect of the Chernobyl accident on cancer risk. An exception is thyroid cancer in young people, where, because of the high 131I thyroid doses received in the most contaminated areas and the low background incidence rates, epidemiological studies have demonstrated an association between radiation dose to the thyroid and thyroid cancer risk in the general population.6, 7
Cancer mortality rates for children and young adults tended to decrease throughout Europe from 1985 to 2000, except among those 15–34 years of age at death in countries with average whole-body doses of 1 mSv or greater. It is notable that decreases in all-cancer and leukemia mortality were seen in those 0–14 years of age at death where, if a radiation effect were detectable, it would be expected to be largest. Survival from childhood cancer has improved substantially in Europe in the last 30 years,50 however, especially for leukemia. Hence, a small increase in the risk of childhood cancer mortality related to Chernobyl will be difficult to detect. Among those 35–69 years of age at death, trends throughout Europe vary, depending on sex and whether the cancer is smoking related. Although trends appear to differ in countries that have average whole-body doses of 1 mSv or more (because of trends in Belarus and Ukraine), increases in cancer mortality in these countries were accompanied by similar increases in noncancer mortality; therefore, these trends are unlikely to reflect an effect of radiation from the accident.
Our analyses showed a trend of increasing cancer incidence as recorded by cancer registries throughout Europe since 1986, particularly in regions with average whole-body doses from Chernobyl of 0.5 mSv or more. Analyses of linear contrasts, however, showed that for cancers other than thyroid, incidence has in fact been increasing since 1981 and that the slope of this rise decreased (for all cancers, breast cancer and leukemia) after the accident. Only for thyroid cancer did the increase become more marked after the accident.
The interpretation of trends in cancer incidence should be made with caution, as cancer registration data are subject to a number of potential biases.46 These include variations in the extent to which cancers are diagnosed and registered, improvements in diagnostic standards and changes in screening practices that may have increased the apparent incidence of some types of cancer because of earlier diagnosis and of identification of tumors that may never have become clinically manifest. These considerations are particularly important in the interpretation of time trends in the registries with doses above 0.5 mSv, as these are dominated by trends observed in Belarus, where since 1990 considerable effort has gone into improving the completeness and accuracy of cancer registration1, 51 and where in the most contaminated areas particular attention is paid to the early detection and treatment of various cancers.1, 19
Trends in cancer mortality are also subject to artifacts: cause of death assignments may be subject to misclassification,52 and reduced cancer mortality may be related to early detection (which is related to screening) and to improved treatment. Indeed, the observation that decreases in mortality from all cancers among those 15–34 years of age are less pronounced in countries with average whole-body doses of 1 mSv, and the increasing mortality trends in those 35–69 years of age in the same countries may well reflect changes in cause-of-death certification and differential improvements in diagnosis and treatment between Eastern European countries and the rest of Europe. For many types of cancer, furthermore, the length of time between diagnosis of cancer and death from that cancer can be substantial. Therefore, it is unlikely that analyses of time trends in mortality up to the year 2000 would have much power to detect radiation–related increases in the rates of cancers other than leukemia, as it is expected that most radiation-induced cancers will take at least 10–20 years after exposure to appear.
Taken together, the results of analyses of trends in cancer incidence and mortality do not appear to indicate (except for thyroid cancer) a measurable increase in cancer incidence in Europe to date, related to radiation from the Chernobyl accident. These results are compatible with predictions of the expected number of cancer cases, which suggest AFs of about 0.01% for cancers other than leukemia, thyroid and nonmelanoma skin cancer and 0.8% for leukemia.
The situation is somewhat different for thyroid cancer. It is known that screening for thyroid cancer, either through formal screening campaigns or through closer attention by medical professionals, occurred in many European countries, in particular in the most contaminated areas of Belarus, the Russian Federation and Ukraine.1 It is therefore possible that the observed increasing trends in thyroid cancer incidence over time and in the regions with the highest doses are attributable at least in part to a screening bias. Much of the increased incidence of thyroid cancer observed in the most contaminated regions, however, relates to childhood and adolescent thyroid cancer. Analyses of the cases in these countries showed that most of these cases were fairly aggressive, with a large proportion showing extracapsular invasion and distant metastases,1, 6 cases that are likely to have been diagnosed even in the absence of screening. However, this may be different in European countries that were less contaminated by Chernobyl. When analyses of thyroid cancer incidence in relation to radiation dose were restricted to Belarus (not shown) 50% of the cases of thyroid cancer diagnosed in 1986–2002 were attributed to radiation in those exposed before age 15 (23% of cases regardless of age at exposure). This is consistent with recent analyses that indicated that approximately 60% of the cases diagnosed in Belarus between 1986 and 2001 among those who were children or adolescents in 1986 can be attributed to radiation.5 These estimates are greater than our predictions using the BEIR VII model (170 cases, with an AF of 25% for the slightly wider period of 1986–2005), although these results are consistent within their uncertainty limits.
Given the lack of demonstrated increases in cancer risk (other than for thyroid cancer) in the most affected countries and the relatively short follow-up for cancers other than thyroid and leukemia (diseases with a very long latent period), it is clear that any assessment of the cancer burden from the Chernobyl accident must be based on risk prediction models derived from other populations exposed to radiation, most notably the atomic bomb survivors and patients exposed to medical sources of radiation. Major questions relate to the choice of models for transport of risk between populations with different background cancer rates, for projection of risk over time and for extrapolation of risks following high-dose and high dose-rate exposure to primarily external radiation to low-dose and low dose-rate exposure involving a mixture of external and internal radiation. These factors limit the accuracy and precision of such projections. Uncertainties related to transport and DDREF have been taken into account in the current article. Further uncertainty arises, however, from the varying quality of cancer incidence and mortality data throughout Europe, from the assumption of constant population demographics and incidence and mortality rates over time and, in particular, from the unknown shape of the dose–response relationship at the very low doses of radiation that many countries received from Chernobyl. Although uncertainties in incidence and mortality rates might be important for individual countries, they are of lesser importance for overall predictions because such uncertainties are averaged over countries. The predictions made in our study from extrapolations to low doses are based on models developed by the BEIR VII Committee39 which, after a comprehensive and critical review of available epidemiological, biological and biophysical data, concluded that the risk would continue in a linear fashion at lower doses without a threshold and that even the smallest dose has the potential to cause a small increase in the risk to humans.
Despite uncertainties in the predictions of the cancer burden due to Chernobyl, they are useful for public health reasons. Such predictions provide an idea of the order of magnitude of the likely impact of the accident, estimated to be several tens of thousands of cancer cases in Europe as a whole. Although this number is far from negligible, it should be viewed in the context of the considerably larger number of cancer cases expected to occur in Europe from other causes. Our predictions suggest that of all the cancer cases expected to occur in Europe between 1986 and 2065, around 0.01% may be related to radiation from the Chernobyl accident. The largest AF, about 1%, is predicted for thyroid cancer, with the vast majority of these cases (close to 70%) expected to occur in the most contaminated regions of Belarus, the Russian Federation and Ukraine.
The predictions presented here are difficult to verify because, except for thyroid cancer, the AF is too low to be detected epidemiologically. For thyroid cancer, predictions based on the BEIR VII model are consistent with the number of cases observed in the most contaminated regions of Belarus, the Russian Federation and Ukraine among those exposed in childhood6 (3,800 reported vs. 3,600 predicted). In Belarus, however, the predictions are lower than, but statistically compatible with, the number of cancers that have been diagnosed in Belarus among those exposed in childhood: our prediction of the total number of cases (attributable to radiation and to other causes) is of the order of 700 (95% UI: 500–1,200) up to the end of 2005, whereas in fact 1,700 cases have been diagnosed through the end of 2002.6 The reasons for the discrepancies between the predicted and observed number of thyroid cancer cases include instability in baseline thyroid cancer rates in young people,53 a possible acceleration of the incidence of thyroid tumors that would otherwise have been diagnosed later because of the promoting effect of iodine deficiency in the more heavily exposed areas of Belarus, the Russian Federation and Ukraine3 and widespread thyroid screening in the affected countries.1 Predictions of the number of cancer deaths in the most contaminated areas of Belarus, the Russian Federation and Ukraine (8,200 – not shown) are similar to those published previously.20
Further follow-up of the most exposed populations in Belarus, the Russian Federation and Ukraine, and careful analytical studies of specific outcomes, in particular breast cancer in the most contaminated areas, will provide important information for assessing the adequacy of risk models developed from other populations exposed to radiation in other settings for the evaluation of cancer burden from Chernobyl in the future.
- Top of page
- Material and methods
- Supporting Information
We thank representatives of the cancer registries of Belarus, Latvia, Lithuania and Ukraine who provided data specifically for use in the present analysis, the dosimetrists (Drs. I. Malátová, T. Ilus, T. Nedveckaite, V. Filistovic, G. Bruk, I. Zvonova, H. Völkle, I. Likhtarev, L. Kovgan, N. Chobanova) who provided detailed information used in the dose reconstruction reported here, Dr. J. Boreham for providing the data for analysis of the mortality trends, Mr. T. Boulet for assistance in the production of the maps and Dr. J. Boice, Jr., for helpful suggestions on a previous version of the manuscript. Dr. D. Krewski is the NSERC/SSHRC/McLaughlin Chair in Population Health Risk Assessment at the University of Ottawa. Dr. S. Darby is supported by Cancer Research UK.
Albania, Austria, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Macedonia, Malta, Moldova, Netherlands, Norway, Poland, Portugal, Romania, the Russian Federation, Serbia and Montenegro, Slovakia, Slovenia, Spain, Sweden, Switzerland, Ukraine and the United Kingdom (see Figures 1 and 2).
- Top of page
- Material and methods
- Supporting Information
- 1United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). Sources and effects of ionizing radiation—Volume II, Effects. New York: United Nations, 2000.
- 2United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). Sources and effects of ionizing radiation. New York: United Nations, 1993.
- 6UN Chernobyl Forum Expert Group Health (EGH). Health effects of the Chernobyl accident and special health care programmes. Geneva, Switzerland: World Health Organization, 2006.
- 10Health status and follow-up of liquidators in Russia. In: KaraoglouA, DesmetG, KellyGN, MenzelHG. First International Conference of the European Commission, Belarus, the Russian Federation and the Ukraine on the Radiological Consequences of the Chernobyl Accident (Minsk, Belarus, 18–22 March 1996); EUR 16544 EN. Brussels, Belgium: European Commission, 1996. p. 861–70..
- 11Medical radiological effects of the Chernobyl catastrophe on the population of Russia: Estimation of radiation risks. Moscow: Meditsina, 2002., .
- 13Chernobyl NPP accident consequences cleaning up participants in Ukraine— health status epidemiologic study—main results. In: KaraoglouA, DesmetG, KellyGN, MenzelHG. First International Conference of the European Commission, Belarus, the Russian Federation and the Ukraine on the Radiological Consequences of the Chernobyl Accident. EUR 16544 EN (Minsk, Belarus, 18–22 March 1996). Brussels, Belgium: European Commission; 1996. p. 871–8., , , , , .
- 16Childhood leukaemia in Belarus, Russia, and Ukraine following the Chernobyl power station accident: results from an international collaborative population-based case-control study. Int J Epidemiol 2005; 3 [Epub ahead of print]., , , , , , , , , , , , et al.
- 18Elevated leukemia rates in Chernobyl accident liquidators [electronic letter]. Br Med J 2003; http://bmj.com/cgi/eletters/319/7203/145/a#31231., , , , , , .
- 20Estimated long term health effects of the Chernobyl accident. In: One decade after Chernobyl. Summing up the consequences of the accident. Proceedings of the EU/IAEA/WHO International Conference. Vienna: International Atomic Energy Agency, 1996. p. 241–79., , , , , , .
- 21United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). Sources and Effects of Ionizing Radiation. New York: United Nations, 1988.
- 22European Communities (EC). Atlas on Cs137 deposition on Europe after the Chernobyl accident. Luxembourg: Office for Official Publications of the European Communities, 1998.
- 23International Commission on Radiological Protection (ICRP). Age-dependent doses to members of the public from intakes of radionuclides: Part 1—Ingestion dose coefficients. ICRP Publication 56. Oxford, UK: Pergamon Press, 1990.
- 24International Commission on Radiological Protection (ICRP). Age-dependent doses to members of the public from intakes of radionuclides: Part 2—Ingestion dose coefficients. ICRP Publication 67. Oxford, UK: Pergamon Press, 1993.
- 25International Commission on Radiological Protection (ICRP). Age-dependent doses to members of the public from intakes of radionuclides: Part 3—Ingestion dose coefficients. ICRP Publication 69. Oxford, UK: Pergamon Press, 1995.
- 26International Commission on Radiological Protection (ICRP). Age-dependent doses to members of the public from intakes of radionuclides: Part 4—Inhalation dose coefficients. ICRP Publication 71. Oxford: Pergamon Press, 1995.
- 27International Commission on Radiological Protection ICRP. Conversion coefficients for use in radiological protection against externalradiation. ICRP Publication 74. Oxford, UK: Pergamon Press, 1996.
- 28External exposures to radionuclides in air, water and soil. Federal Guidance Report 12. Washington, DC: EPA, 1993., .
- 29United Nations (UN). Environmental consequences of the Chernobyl accident and their remediation: Twenty years of experience. IAEA Report of the UN Chernobyl Forum expert group “Environmental” (EGE). Vienna, Austria: IAEA, 2005.
- 39US National Research Council. Health risks from exposure to low levels of ionizing radiation. BEIR, VII Report, phase II. Washington, DC: National Academy of Science, 2005.
- 43Cancer mortality for selected sites in 24 countries (1950–57). Sendai, Japan: Tohoku University of Medicine, 1960..
- 44Mortality from smoking in developed countries 1950–2000: Indirect estimates from national vital statistics. Oxford, UK: Oxford University Press, 1994., , , , .
- 48A method for judging all contrasts in the analysis of variance. Biometrika 1953; 40: 87–104..
- 49Mortality from smoking in developed countries 1950–2000. The MRC/Cancer Research UK/BHF Clinical Trial Service Unit & Epidemiological Studies Unit (CTSU). 2005. http://www.ctsu.ox.ac.uk/˜tobacco/, , , .
- 51Cancer registration techniques in the new independent states of the former Soviet Union. IARC Scientific Publication 35. Lyon, France: IARC Press, 1998., , , , , .
- 53Cancer incidence in five continents. Vol. VIII. International Agency for Research on Cancer IARC Scientific Publication 155. Lyon, France: IARC Press, 2002., , , , .
- Top of page
- Material and methods
- Supporting Information
This article contains supplementary material that can be found on the International Journal of Cancer website.
|jws-ijc.22037.pdf||1305K||Supporting Information file jws-ijc.22037.pdf|
Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.