IJE Advance Access originally published online on September 3, 2007
International Journal of Epidemiology 2007 36(5):1038-1046; doi:10.1093/ije/dym121
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Solid cancer incidence and low-dose-rate radiation exposures in the Techa River cohort: 1956–2002
1 Epidemiology Laboratory, Urals Research Center for Radiation Medicine, Chelyabinsk, Russia.
2 Division of Epidemiology and Biostatistics, University of Illinois at Chicago, Chicago, IL USA.
3 Biophysics Laboratory, Urals Research Center for Radiation Medicine, Chelyabinsk, Russia.
4 Hirosoft International Corporation, Eureka, CA USA.
5 Clinical Physiology Laboratory, Urals Research Center for Radiation Medicine, Chelyabinsk, Russia.
* Corresponding author. 1335 H St. Eureka, CA 95501, USA. E-mail: preston{at}hirosoft.net
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Background This is the first analysis of solid cancer incidence in the Techa River cohort, a general population of men and women of all ages who received chronic low-dose rate exposures from environmental radiation releases associated with the Soviet nuclear weapons programme. This cohort provides one of the few opportunities to evaluate long-term human health risks from low-dose radiation exposures.
Methods Cancer incidence rates in this cohort were analysed using excess relative risk (ERR) models. The analyses make use of individualized dose estimates that take into account residence history, age and other factors. Cases are identified on the basis of continuing, active follow-up of mortality and cancer incidence.
Results Based on 1836 solid cancer cases with 446 588 person years accrued over 47 years of follow-up, solid cancer incidence rates were found to increase with dose and about 3% of the cases were attributable to radiation exposure. The ERR was 1.0/Gy (P = 0.004 95% CI (0.3; 1.9) in a linear dose-response model. There was no significant non-linearity in the dose response and no indication of effect modification by gender, ethnicity, attained age or age at first exposure.
Conclusions The Techa River cohort provides strong evidence that low-dose, low-dose rate exposures lead to significant increases in solid cancer risks that appear to be linear in dose. The results do not suggest that risks associated with low-dose rate exposures are less than those seen following acute exposures such as were received by atomic bomb survivors.
Keywords Cohort studies, carcinogens, environmental, risk assessment, cancer, radiation-induced
Accepted 8 May 2007
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
In 1949, the Mayak Production Association located in the Southern Urals began producing plutonium for the Soviet nuclear weapons programme. Between 1949 and 1956, radioactive materials were released into the Techa River as a result of Mayak operations, with maximal releases in 1950 and 1951. The Techa is a small that originates near the Mayak complex and flows for almost 300 km through Chelyabinsk and Kurgan Oblasts before merging with Iset River. At the time of the Mayak releases, there were about 30 000 people living in 41 rural villages and settlements on the banks of the river. Residents of these villages received both external and internal low-dose rate radiation exposures. The external exposures were due to gamma-rays from contamination of the river shoreline and flood plain soil while the internal exposures were largely the result of the consumption of water, milk and food that contained 137Cs, 90Sr and other radionuclides.
Efforts to monitor the health of Techa River residents began in the 1950s. In the late 1960s and early 1970s, Soviet researchers initiated efforts to define a fixed cohort of people who lived on the Techa at some time between 1949 and 1956, to ascertain vital status and cause of death for cohort members, and to develop a dosimetry system that would allow the computation of individual dose estimates that would be useful in risk assessment. The initial open-literature publications on the Techa River cohort were published in the mid 1990s.1–5 These publications made it clear that the Techa River cohort has the potential to provide important, unique information on the long-term effects of protracted low-dose, low-dose rate, internal and external radiation exposures on the risk of cancer and other diseases in an unselected population of men and women of all ages at exposure.
Over the past 15 years, there has been an international effort to improve all aspects of the Techa River studies. This includes ongoing work to refine the cohort definition, to improve the quality and completeness of the vital status ascertainment and mortality follow-up data and improve the dosimetry system. Beginning about 10 years ago efforts were made to ascertain cancer incidence in a subset of the Techa River cohort. A detailed description of cohorts and the methods used for mortality and incidence follow-up in these cohorts are given in a recent article6 that was published together with an article presenting solid cancer and leukaemia mortality risk estimates7 based on 50 years of follow-up for the Techa River cohort using the most recent dosimetry system.8–10 The results provide clear evidence of radiation-associated increases in cancer mortality risks for the cohort. The results make it clear that studies of this population are providing quantitative information on environmental low-dose, low-dose-rate radiation effects that complements risk estimates based on the experience of atomic bomb survivors11–13 following acute exposures.
In this article, we present analyses of the risk of solid cancer for the subset of the full Techa River Cohort with incidence follow-up for the period from 1956 (the earliest date from which solid cancer incidence follow-up is feasible) to 2002. As discussed below, despite limitations on the follow-up period and study population there is roughly the same number of newly diagnosed solid cancer cases (1846) as there were cancer deaths (1842) for the larger study population and extended follow-up period used in the mortality analyses.
Leukaemia risks are not considered in this article because the study subcohort, follow-up period and catchment area differs from that for solid cancer incidence studies,6 and leukaemia mortality and incidence among Techa River residents has been considered in two recent publications.7,14
The conduct of this study is reviewed annually by the Urals Research Center for Radiation Medicine Institutional Review Board (IRB).
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Study population and incidence follow-up
The extended Techa River cohort currently includes 29 800 people born before 1950 who lived on the Techa for some time between 1950 and the end of 1960. The cohort includes 18 389 people who were residents of one of the 25 villages in Chelyabinsk Oblast (7–148 km from the release point) and 11 411 people who lived in one of the 16 riverside villages in Kurgan Oblast (155–237 km from the release point).
As recently described in detail by Kossenko et al.6 mortality follow-up for the full cohort begins in 1950 and cohort members are treated as lost to follow-up during periods in which they are not known to reside in either Chelyabinsk or Kurgan Oblasts. Data on non-fatal solid cancer cases are not available prior to 1956 (the year in which physicians were first required to inform the regional oncological centre about cancer cases) and even after 1956 comprehensive cancer morbidity data are only available for cohort members residing in the five raions of Chelyabinsk Oblast that constitute the original catchment area or in the city of Chelyabinsk. As a result of these limitations, current incidence analyses are restricted to the subcohort consisting of the 17 433 cohort members living in Chelyabinsk Oblast at the time of their initial exposure, who had neither died nor had a recorded cancer diagnosis (including leukaemia or other malignancies of the haematopoietic system) prior to January 1, 1956, and are known to have lived in the incidence catchment area which consists of the five Chelyabinsk Oblast raions through which the Techa River flows or Chelyabinsk city at some time between 1956 and the end of 2002.
There has been considerable migration during the 47-year follow-up period. As described previously,6,7 detailed residence histories that include data on residence location and migration dates have been constructed from a variety of sources and are updated as a part of the ongoing mortality screening for the full Techa River cohort. These records are used for dose computation and to determine the periods during which a cohort member is eligible for inclusion in risk analyses. Because cancer case ascertainment is limited to cases diagnosed among people residing in the incidence catchment area, people are considered in the analysis only for periods during which they are known to be living in these areas. When a person is known to have emigrated from the incidence catchment area (distal migrants) or when the current residence is unknown they are treated as lost to follow-up. As of January 2003, 3535 cohort members (20%) were known to have migrated from the catchment area and current vital status was unknown for 1397 cohort members (8%) whose last known address was in the catchment area.
Table 1 provides a summary of the state of solid cancer incidence follow-up in the cohort as of the end of the current follow-up. For people who are not known to have migrated (as determined from information provided by the Oblast address bureau) the date of last known vital status is used as the end of follow-up. This is the most recent date at which there was a record (from the address bureau or a personal contact) of the person being alive and resident in the study catchment area. This date is prior to December 31, 2002 for the 1397 non-migrants in the lost-to-follow-up group.
|
Table 2 presents some descriptive statistics on the cohort used for the cancer incidence analyses. The table includes information on some basic demographic characteristics of the incidence cohort stratified by residence in or out of the catchment area at the end of follow-up (i.e. the earliest of the date of cancer diagnosis, date of death, date of last known vital status or December 31, 2002). As discussed subsequently, both distant migrants and people whose current vital status is unknown are treated as lost-to-follow-up from the date of migration from the catchment area.
|
Both the gender and age distributions reflect the demographic impact of the Second World War. Almost 60% of the cohort members are women and, at the start of follow-up, the mean age for men was about 4 years younger than that for women. Almost one-third of the cohort members are identified as having Tartar/Bashkir ethnicity, the rest of cohort is assumed to have Slav ethnicity.
Migrants tend to be younger than people who were last known to be living in the incidence catchment area. The data in this table and other analyses indicate that men and women were equally likely to have moved away from the incidence catchment area by the end of follow-up. However, while Tartar/Bashkir cohort members are on average somewhat younger than Slavs, it seems that they are less likely to emigrate.
The analyses in this report considered cases of solid cancer (ICD-9 codes 140–199) other than bone cancer (ICD-9 code 170). Forty-nine per cent of the 1846 solid cancers were diagnosed among men. The most common sites among men were lung (29%) and stomach (18%), while among women the most common sites were stomach (16%), cervix (15%) and breast (11%). The 10 cases of bone cancers were excluded because the dose to the bone surface is markedly higher due to exposure to 90Sr and other radionuclides than the soft-tissue doses that are relevant for most other solid cancers. Since portions of the lower gastro-intestinal tract may have received higher doses from exposure resulting from the passage of short-lived radionuclides, we also carried out some analyses in which the 49 colon cancers were excluded.
Data quality
As described in, Kossenko et al.6 while an active case ascertainment programme has been in place for much of the past decade, ascertainment of cases from the initial decades of follow-up was dependent on archival records from the Oblast oncology dispensary, the Urals Research Center for Radiation Medicine clinic and death certificates. As a result, the quality and completeness of the cancer incidence data in this cohort has varied over time. Table 3 presents information on common measures of the quality of and completeness of the cancer ascertainment; the proportion of cases with histological confirmation and the proportion of cases ascertained solely from death certificates. Only cases ascertained in the current incidence catchment area (the five riverside raions in Chelyabinsk Oblast and Chelyabinsk city) are considered in this table. Information is given for all solid cancers as a group (including bone cancers).
|
Over the full follow-up period less than half of the solid cancer cases were histologically confirmed, almost one-fifth of the cases were ascertained solely on the basis of death certificates, while the remaining 20% were either ascertained solely on the basis of clinical records or the method of diagnosis was not clearly specified. During the last 13 years of follow-up there have been marked increases in histological and visual verification rates and corresponding reductions in both the DCO and clinical verification rates. These verification rates were somewhat lower and the DCO rates higher than those seen for recent years in the world's best population-based cancer registries.15
As indicated in Table 1, cause of death was unknown for about 10% (784 of 8091) of the deaths among non-migrants over the full follow-up period. However, as with diagnostic confirmation, the proportion of deaths for which a death certificate could not be found has fallen to about 6% of the deaths occurring in the last 12 years of follow-up. Thus, some cases that would have been ascertained on the basis of death certificates have not been identified. Assuming that the proportion of cancer deaths among those with unknown cause of death is similar to that for reported causes of death, to date about 5% of the cancer cases were missed because of incomplete information on cause of death. The notification system and other sources used for case ascertainment also fail to detect some cases not mentioned on death certificates. This problem was particularly pronounced for the early years of follow-up when it is likely that 20% or more of the cases were not reported. However, the probability that a case is ascertained has improved over time and will continue to do so as the regional cancer registry managed by the Chelyabinsk Oblast Oncology Dispensary (ChOOD) improves. We believe that at present >90% of the cancer cases are being detected.
Dosimetry
These analyses are based on the current individualized annual cumulative dose estimates computed using the Techa River Dosimetry System (TRDS-) 2000.8,9 These are derived from annual village mean dose estimates that allow for dose rate in air at the river bank and in residence areas, representative behaviour patterns, intake of radionuclides with river water and food and other factors. These means are then individualized to account for age, residence history and other factors. Although there has been some criticism of the current Techa River dosimetry,16,17 a recent review of the TRDS-200010 has endorsed the basic dosimetric approach while indicating the need for additional efforts to clarify issues related to the source term and other aspects of the dosimetry.
The doses used in these analyses are time-dependent cumulative stomach doses computed with a 5-year lag. Stomach doses are viewed as a good approximation to the average soft-tissue dose. These doses arise from a combination of whole-body external gamma-ray exposures from contaminated flood plain soils and internal exposures arising largely from the ingestion of 137Cs and short-lived radionuclides in water, milk and food products. Internal exposure to beta particles from 90Sr has a major impact on doses to the bone surface and bone marrow but has little impact on the dose received by soft tissues other than the small and large intestine.
The mean (median) total cumulative stomach dose for cohort members used in these analyses is 0.04 (0.008) Gy. Approximately 55% of the stomach dose was associated with internal exposures.
Data organization and statistical methods
For the primary analysis in this report the solid cancer incidence follow-up has been organized as a table of cases and person years (or rates) stratified on gender, ethnicity (Slav or Tartar/Bashkir), early (1950–52) or late (1953–60) entry into the affected area, age at the start of follow-up (seven categories: decades to age 59 and 60+), attained age (17 categories: five-year intervals to age 79 and 80+), years since start of follow-up (10 five-year intervals), calendar time (10 categories for the period from January 1, 1956 to December 31, 2002 with cut points on January 1 of 1956, 1960, 1965, 1970, 1975, 1980, 1985, 1990, 1995 and 2000), incidence catchment area subgroups (two groups: original catchment area raions and Chelyabinsk city) and time-dependent lagged cumulative stomach dose. The primary analyses were carried out using a 5-year lag, but the sensitivity of the risk estimates to the lag time was examined using doses with 2- and 10-year lags. The dose categories in the person-year table include a zero-dose category and positive-dose categories with lower bounds of 0, 2, 4, 8, 10, 50, 75, 100, 150, 200, 250 and 300 mGy. There are likely to be considerable classification errors, especially in the low-dose categories due to the use of village mean doses and additional uncertainties in individual dose estimates. Cumulative stomach doses were lagged five years. The maximum TRDS-2000 individual stomach dose estimate is 450 mGy.
The input data for the person year computations was a file that includes multiple records per person with one record for period in which residence was known along with indicators of the mortality and incidence study catchment area for the residence during the period. For the solid cancer analyses, there were 39 114 post-1956 residence periods including 31 773 in the incidence catchment area for the 17 433 cohort members who contribute some person years to the analyses. Follow-up for these analyses begins at January 1, 1956 for cohort members who were resident in the incidence catchment area at that date or the start of the first eligible residence interval with an entry date after this date for those who were not. Follow-up continues through the earliest of December 31, 2002, the date of last known vital status, the date of last known residence in the incidence catchment area and the date of diagnosis of the first primary cancer. As noted earlier, person years and cases were not counted for periods in which a person was known to be living outside of the incidence catchment area. Person years were also not counted during gaps when there is no information on residence.
The solid cancer person-year table contains counts of all solid cancers and person-year weighted mean values for age, time since start of follow-up, year, age at entry and lagged stomach dose.
Statistical methods and rate models
In this work, cancer rates in the cohort were analysed using simple parametric excess relative risk (ERR) models. The basic ERR model for age-specific cancer incidence rates can be written as 
(a,d,z) = 0(a,z0)(1 + 
(d)
(z1)) where a is age at diagnosis, d is dose (in Gy), and z0 represents factors (such as gender, birth cohort, ethnicity or time period) that can modify the baseline rates (0) and z1 represents factors that might modify the ERR.
The excess risk was described as a product of a dose-response function (d) and an effect modification function ((z1)). The dose-response function was generally taken as a linear function of dose (ß1d). Tests for non-linearity in the dose response were based on comparison of the linear and linear-quadratic dose-response models (ß1d + ß2d2). Because of the small number of radiation-associated cases in this cohort, there is little power to assess effect modification. Therefore, the primary analyses were based on dose-response models without effect modification. The nature of the effect modification was assessed by adding potential effect modifiers to the model. The effect modifiers considered in the analysis included gender, ethnicity, attained age and age at initial exposure. The attained age effect was taken to be proportional to attained age to a power, age at first exposure effects were described using a log-linear model, while gender and ethnicity were taken to have multiplicative effects on the dose-response slope.
The smoothed non-parametric fit presented in Figure 1 was based on running 5-point weighted average of dose-category-specific ERR estimates with weights proportional to a smoothing weight taken as 0.15, 0.2, 0.3 0.2, 0.15 and the inverse of the variance of the category-specific estimates. The plotting positions for the smoothed estimates were defined by applying the same set of weights to the category-specific person-year weighted mean doses. Standard errors are computed as the square root of the variance of the weighted sum that defines the smoothed values.
|
Baseline rates were described using a model in which the log-rates were modelled as gender-specific quadratic functions of log attained age with a multiplicative effect for ethnicity and log-linear variation in age-specific rates with birth cohort. We also considered a multiplicative effect to allow for differences in baseline rates for cohort members who remained in the riverside raions and those who moved to Chelyabinsk city, but this parameter was not included in the final baseline rate model.
Parameter estimates were estimated using Poisson regression maximum likelihood analyses of rates in the detailed rate tables described above. Significance tests and confidence intervals were determined directly from the likelihood. All P-values refer to two-sided tests.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Baseline rates
Baseline cancer rates were estimated jointly with the linear radiation effect discussed below. As illustrated in Figure 2, these rates increased in the expected manner with increasing attained age with a statistically significant (P < 0.001) gender-specific age patterns. Under a simple proportional hazards model rates for females were about 55% of those for males (P < 0.001; 95% CI 50%; 60%).
|
Baseline rate parameter estimates and 95% confidence bounds are presented in Table 4. Age- and gender-specific rates for cohort members identified as having Tartar/Bashkir ethnicity were 80% of those for Slavs (P < 0.001; 95% CI 73%; 88%). There was a suggestion (P = 0.08) of a small birth cohort effect with age-specific rates increasing by about 3% per decade increase in year of birth (95% CI –0.4%; +7%). There was no evidence (P = 0.5) that the age- and gender-adjusted rates for Chelyabinsk city residents differed from those for cohort members who continued to live in the five riverside raions (SIR = 1.05, 95% CI 0.92; 1.2).
|
Radiation effects
The dose response was significant (P = 0.004) in a linear ERR model. The estimated ERR per Gy was 1.0 (95% CI 0.3; 1.9). The dose-response estimate was virtually unchanged when a 2-year lag was used but decreased by about 10% when the analyses were carried out using 10-year lagged cumulative dose Table 5 summarizes the distribution of cases, person years and fitted values in 5-year lagged cumulative dose categories.
|
Under a linear dose-response model, it was estimated that about 3% of the cases are associated with the radiation exposure (95% CI 1%; 6%). A formal test for non-linearity based on a linear-quadratic dose-response model was not statistically significant (P = 0.4). The fitted linear-quadratic model is essentially a pure quadratic dose response with an ERR of 2.9/Gy2 (95% CI 0.92; 5.3). Under this model, the predicted risk at 1 Gy was considerably greater than that for the linear model but lower than the linear model for doses below about 0.3 Gy. The non-parametric smooth suggests that the response over the low-dose region (<0.1 Gy) falls in between the linear and pure quadratic fits. The fitted parametric and non- parametric dose-response functions are shown in Figure 1.
Under the fitted quadratic dose-response model, about 2% (32) cases appeared to be associated with the radiation exposure with 95% confidence bounds of 0.5 to 3%. As these results indicate, while there is rather compelling evidence of dose response in these data, there is considerable uncertainty about the magnitude of the risk at very low doses.
In view of the small number of radiation-associated cases, it is not surprising that there is no indication of significant modification by gender, ethnicity, age at first exposure or attained age (P > 0.5 in all cases). As indicated in Table 6, point estimates suggested that women had slightly lower ERRs than men and that Tartars and Bashkirs had slightly lower risks than Slavs. The point estimate of the age-at-first-exposure effect was positive while the ERR was estimated to increase with attained age
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The risk estimates from this first analysis of solid cancer incidence rates in a subset of the Techa River cohort complement the results of previously published analyses of cancer mortality in this cohort. While the incidence and mortality analyses result in similar estimates of the radiation-associated ERR, there are several important differences in the analyses. First, since comprehensive data on cancer incidence are only available for cohort members initially exposed in Chelyabinsk Oblast who reside in either Chelyabinsk City or the five Chelyabinsk Oblast raions near the Techa River at the time of diagnosis, the incidence cohort contains only about 60% of the full cohort and the problem of loss-to-follow-up due to migration was more severe than in the mortality analyses. Second, cancer registration in Chelyabinsk Oblast began in 1956 (at the time of the introduction of a mandatory nationwide cancer reporting system) so incidence follow-up began 6 years later than mortality follow-up for most cohort members. However, these incidence analyses included 3 years more follow-up (2000–02) than were available at the time of the mortality analyses in.7 During the common follow-up period (1956–99), there were about 35% more solid cancer cases than deaths among people included in the incidence analyses. About 8% of the solid cancer cases occurred during the final 3 years of follow-up. Thus, despite the reduced cohort size and follow-up time for these solid cancer incidence data the analyses were based on approximately the same number of cases (1836 cases) as the previous mortality analyses (1860 deaths). Because the primary analyses were based on 5-year lagged cumulative doses, because doses for the Kurgan residents excluded from these analyses were quite low, and because of the extended follow-up time for these incidence analyses, the number of person years for cumulative doses in >0.1 Gy in these analyses were similar to those in the mortality analyses.7
Like the mortality data, the Techa River cohort incidence data provide clear evidence of increases in solid cancer rates and a strong dose-response relationship associated with exposure to radiation from the contaminated Techa River. Under a linear dose-response model, risk estimates for the two analyses were almost identical: 0.92/Gy for mortality and 1.0/Gy for incidence. While the estimated number of radiation-associated incident cases was greater than for the mortality data, the estimated proportion of cases due to radiation exposure (attributable risks) were similar for the incidence (3.2%) and mortality (2.5%) data. As with the mortality data, there is no evidence against a simple linear-dose-response relationship for solid cancer incidence rates.
These analyses were based on the average stomach dose. This should be a reasonable representative dose for solid cancer risk estimation except for bone cancers, which were excluded from these analyses, and, possibly, colorectal cancers, which were included in most analyses. Bone doses for Techa River residents are elevated because bone receives extensive exposure from internally deposited 90Sr. As indicated in,18 doses to the lower GI tract may have been substantially higher than those to other organs because of the relatively long passage times for short-lived radionuclides and 90Sr. However, excluding the 49 colon cancers had little impact on the results (ERR/Gy = 1.1, P = 0.004, 95% CI 0.3; 1.9).
As noted in the Methods section, a substantial portion of the cohort is lost to follow-up because they are known to have migrated from the catchment area or because they were last known to be alive and living in the catchment area sometime prior to end of follow-up. While losses due to migration or tracing failures serve to reduce the effective sample size and hence the power of the study, there is no reason to think that migration acts as a confounding factor (which would require migration rates to be correlated with both the likelihood of cancer diagnosis and dose). The relatively high DCO rate and the fact that the cause of death is unknown for some deaths in the catchment area imply that some cancer cases were missed. Therefore, baseline cancer rate estimates are biased downward. Provided the probability of missing a case does not depend on dose (which it does not) under-ascertainment of cases will not affect ERR estimates. However, excess rate estimates will be biased downward, which was the primary reason we chose not to present excess rate estimates in this article.
Unfortunately, in view of the small number of radiation-associated solid cancers in this cohort, the data have little power to detect or describe how the excess risk varies with gender, ethnicity, attained age or age at initial exposure. As noted in the Results section, there is no evidence of significant variation in the ERR with any of these factors (P > 0.5 for each). The mortality analyses7 provided a weak (P = 0.1) suggestion of an increase in the solid cancer ERR with increasing attained age, however the point estimate of the change is essentially zero for the incidence data. Further follow-up of the current cohort should lead to some increase in the ability to characterize effect modification, however larger increases in power are likely to result from extension of the catchment area to include cohort members living in regions of Chelyabinsk Oblast that are not included in the current incidence catchment area. We anticipate that this will be possible as cancer registration data from the ChOOD are computerized and procedures for linking information in the ChOOD registry and the Techa River cohort are improved.
The TRDS-2000 dose estimates 8,9,18,19 used for these analyses should not be regarded as definitive since efforts are currently underway to improve the dosimetry system and better characterize uncertainties in individual dose estimates. However, an international review group10 has endorsed the basic approach. As the dosimetry is improved to provide increased individualization of dose estimates, there may be substantial changes in dose estimates for specific individuals but we do not anticipate significant changes in average doses within villages. We expect that future versions of the dosimetry system will provide more information on the nature of shared and unshared uncertainties20 and as the necessary information becomes available we will carry out analyses that allow for the impact of these uncertainties on risk estimates.
Following publication of the mortality results for this study7 a question was raised about the possibility of confounding by chemical exposures. As stated in our published response to that letter,21 there is no evidence of significant chemical exposures in this cohort and even if such exposures did exist, it is difficult to see how they would seriously distort the radiation dose response and even more difficult to see how they would induce a spurious radiation dose-response.
Our estimate of the radiation-associated solid cancer incidence ERR in the Techa River cohort (1.0/Gy), is essentially the same as that (0.9/Gy) seen in analyses of mortality in a cohort of about 500 000 nuclear workers from 15 countries, >90% of whom had cumulative doses of <50 mSv.22 The overall risk in the 15-country study appears to be influenced fairly strongly by the large risk estimate for the subset of the Canadian nuclear worker studies used in those analyses. Recent analyses of the Canadian worker data suggest ERRs on the order of 2.5/Sv, associated with low-dose exposures.23,24 Using the solid cancer risk model given in the recent analysis of cancer incidence in the Life Span Study, the gender-averaged ERR per Gy estimate at age 65 for a person exposed at age 25 (which correspond roughly to the mean age at initial exposure and age at diagnosis in the current Techa River cohort data) is about 0.6/Gy. Although this is lower than what is seen in these analyses, the atomic bomb survivor estimate is within the CI for the Techa River cohort incidence risk estimate. The direction of the estimated gender, age and age at initial exposure effects on the ERR in the Techa River cohort are different from those in the atomic bomb survivors. However, since the number of radiation-associated cases in this cohort is relatively small, these effect modification parameters are poorly estimated and the CIs include the estimates seen for similar effects in the atomic bomb survivors.11–13
While the results of these analyses do not provide definitive, precise quantitative estimates of the increase in solid cancer risks following protracted low-dose, low-dose-rate exposures, we feel that they do provide strong evidence that such exposures lead to significant increases in risk that are roughly proportional to dose. At this time these data, like the pooled nuclear study data,22 do not provide any evidence to support the notion that low dose rate exposures are less effective (per unit dose) than acute exposures such as those received by the atomic bomb survivors. Continued follow-up of these relatively low-dose cohorts is needed to improve the precision, better characterize the uncertainties in these risk estimates and provide some insights into effect modification and dose-rate effects on the magnitude of the risks. This article does not include risk estimates for specific cancer sites. While the fitting of site-specific risks is fairly straightforward, the site-specific risk estimates are quite uncertain, which makes interpretation of the risks for specific sites extremely difficult. We are currently working on an article that will present some site-specific risk estimates and discuss the problems of interpretation of these risks. As the dosimetry system evolves, we will incorporate additional information on dosimetric uncertainty into the characterization of these excess risks.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
As with any long-term large-scale project, many people have made important contributions to this project. We are grateful for the support of the Chelyabinsk Oblast Oncology Dispensary. Catherine Zhidkova has played many important roles in our collaboration. We would also like to acknowledge the contributions of Mira Kossenko, Nickolai Startsev, Daniel Hoffman and the late Terry Thomas. Financial support has been provided by the Russian Ministry of the Health, the Russian Federal Medical-Biological Agency and the United States Department of Energy (Contract DE-FC02-04EH04014).
Conflict of interest: None declared.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Degteva MO, Kozheurov VP, Vorobiova MI. General approach to dose reconstruction in the population exposed as a result of the release of radioactive wastes into the Techa River. Sci Total Environ (1994) 142:49–61.[CrossRef][Medline]
2 Kossenko MM, Degteva MO. Cancer mortality and radiation risk evaluation for the Techa River population. Sci Total Environ (1994) 142:73–89.[CrossRef][Medline]
3 Akleyev AV, Lyubchansky ER. Environmental and medical effects of nuclear weapon production in the southern Urals. Sci Total Environ (1994) 142:1–8.[CrossRef][Medline]
4 Kossenko MM. Cancer mortality in the exposed population of the Techa River area. World Health Stat Q (1996) 49:17–21.[Medline]
5 Kossenko MM, Degteva MO, Vyushkova OV, Preston DL, Mabuchi K, Kozheurov VP. Issues in the comparison of risk estimates for the population in the Techa River region and atomic bomb survivors. Radiat Res (1997) 148:54–63.[Web of Science][Medline]
6 Kossenko MM, Thomas TL, Akleyev AV, et al. The Techa River Cohort: Study Design and Follow-up Methods. Radiat Res (2005) Nov;164:591–601.[CrossRef][Web of Science][Medline]
7 Krestinina LY, Preston DL, Ostroumova EV, et al. Protracted radiation exposure and cancer mortality in the Techa River cohort. Radiat Res (2005) Nov;164:602–11.[CrossRef][Web of Science][Medline]
8 Degteva MO, Kozheurov VP, Tolstykh EI, et al. The Techa River dosimetry system: methods for the reconstruction of internal dose. Health Phys (2000) 79:24–35.[Web of Science][Medline]
9 Degteva MO, Vorobiova MI, Kozheurov VP, Tolstykh EI, Anspaugh LR, Napier BA. Dose reconstruction system for the exposed population living along the Techa River. Health Phys (2000) 78:542–54.[CrossRef][Web of Science][Medline]
10 Balonov M, Alexakhin R, Bouville A, Liljinzin JO. Report from the Techa River dosimetry review workshop held on 8-10 December 2003 at The State Research Centre Institute of Biophysics, Moscow, Russia. Health Phys (2006) 90:97–113.[CrossRef][Web of Science][Medline]
11 Thompson DE, Mabuchi K, Ron E, et al. Cancer incidence in atomic bomb survivors. Part II: Solid tumors, 1958-1987. Radiat Res (1994) 137(Suppl. 2):S17–67.[Web of Science][Medline]
12 Preston DL, Pierce DA, Shimizu Y, et al. Effect of recent changes in atomic bomb survivor dosimetry on cancer mortality risk estimates. Radiat Res (2004) 162:377–89.[CrossRef][Web of Science][Medline]
13 Preston DL, Shimizu Y, Pierce DA, Suyama A, Mabuchi K. Studies of mortality of atomic bomb survivors. Report 13: Solid cancer and noncancer disease mortality: 1950-1997. Radiat Res (2003) 160:381–407.[CrossRef][Web of Science][Medline]
14 Ostroumova E, Gagniere B, Laurier D, et al. Risk analysis of leukaemia incidence among people living along the Techa River: a nested case-control study. J Radiol Prot (2006) 26:17–32.[CrossRef][Web of Science][Medline]
15 Parkin DM, Whelan SL, Ferlay J, Teppo L, Thomas DB, eds. Cancer Incidence in Five Continents, Vol. VIII. In: IARC Scientific Publications No. 155. (2002) Lyon: IARC.
16 Mokrov Y. Internal and external dose assessments for the population of the Metlino and Muslyumovo settlements living at the Techa river during 1949-1954. (2003) 41–52. [in Russian]. Radiation Safety Problems (Special Issue devoted to the 50th Anniversary of the Southern Urals Biophysics Institute).
17 Mokrov YG. Dose assessment for the Metlino and Muslyumovo populations who lived along the Techa river from 1949 to 1954. Radiat Environ Biophys (2004) 43:209–18.[CrossRef][Web of Science][Medline]
18 Degteva MO, Vorobiova MI, Tolstykh EI, et al. Development of an improved dose reconstruction system for the Techa River population affected by the operation of the Mayak Production Association. Radiat Res (2006) 166(Pt 2):255–70.[CrossRef][Web of Science][Medline]
19 Napier BA, Shagina NB, Degteva MO, Tolstykh EI, Vorobiova MI, Anspaugh LR. Preliminary uncertainty analysis for the doses estimated using the Techa River dosimetry system—2000. Health Phys (2001) 81:395–405.[Web of Science][Medline]
20 Stram DO, Kopecky KJ. Power and uncertainty analysis of epidemiological studies of radiation-related disease risk in which dose estimates are based on a complex dosimetry system: some observations. Radiat Res (2003) 160:408–17.[CrossRef][Web of Science][Medline]
21 Ulsh BA. Comments on ‘Protracted radiation exposure and cancer mortality in the Techa River cohort’ by Krestinina et al. Radiat Res. (2006) 166. 814;author reply-5.
22 Cardis E, Vrijheid M, Blettner M, et al. Risk of cancer after low doses of ionising radiation: retrospective cohort study in 15 countries. Brit Med J (2005) 9;331:77.
23 Sont WN, Zielinski JM, Ashmore JP, et al. First analysis of cancer incidence and occupational radiation exposure based on the National Dose Registry of Canada. Am J Epidemiol (2001) 15(153):309–18.
24 Zablotska LB, Ashmore JP, Howe GR. Analysis of mortality among Canadian nuclear power industry workers after chronic low-dose exposure to ionizing radiation. Radiat Res (2004) 161:633–41.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P Jacob, W Ruhm, L Walsh, M Blettner, G Hammer, and H Zeeb Is cancer risk of radiation workers larger than expected? Occup. Environ. Med., December 1, 2009; 66(12): 789 - 796. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E J HALL and D J BRENNER Cancer risks from diagnostic radiology Br. J. Radiol., May 1, 2008; 81(965): 362 - 378. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E Cardis Commentary: Low dose-rate exposures to ionizing radiation Int. J. Epidemiol., October 1, 2007; 36(5): 1046 - 1047. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||