Allowance for random dose estimation errors in atomic bomb survivor studies: a revision

Allowing for imprecision of radiation dose estimates for A-bomb survivors followed up by the Radiation Effects Research Foundation can be improved through recent statistical methodology. Since the entire RERF dosimetry system has recently been revised, it is timely to reconsider this. We have found that the dosimetry revision itself does not warrant changes in these methods but that the new methodology does. In addition to assumptions regarding the form and magnitude of dose estimation errors, previous and current methods involve the apparent distribution of true doses in the cohort. New formulas give results conveniently and explicitly in terms of these inputs. Further, it is now possible to use assumptions about two components of the dose errors, referred to in the statistical literature as "classical" and "Berkson-type". There are indirect statistical indications, involving non-cancer biological effects, that errors may be somewhat larger than assumed before, in line with recommendations made here. Inevitably, methods must rely on uncertain assumptions about the magnitude of dose errors, and it is comforting to find that, within the range of plausibility, eventual cancer risk estimates are not very sensitive to these.     

Mechanistic models for radiation carcinogenesis and the atomic bomb survivor data

Recently, Heidenreich et al. (Radiat. Res., 158, 607-617, 2002) suggested that the Radiation Effects Research Foundation (RERF) A-bomb survivor cohort study is not large enough to discriminate between various possible carcinogenic mechanisms. At least with the current follow-up, this is true to some extent, but I think the specific issues are rather different than they suggest. In particular, I do not think it is true-as they further indicate-that various models fit the data about equally well while estimating very different patterns of excess risk, which would imply that these patterns cannot be reasonably well characterized. I will point to specific criticisms of their approach to the data and offer some more general comments on mechanistic modeling approaches. Although there are important distinctions, I suggest on a very optimistic note that the two major approaches may be converging, and soon the main differences may not be in the assumptions made but in the aims of the modeling.     

Radiation dose dependences in the atomic bomb survivor cancer mortality data: a model-free visualization

Chomentowski, M., Kellerer, A. M. and Pierce, D. A. Radiation Dose Dependences in the Atomic Bomb Survivor Cancer Mortality Data: A Model-Free Visualization. The standard approach to obtaining nominal risk coefficients for radiation-related cancer involves fitting linear or linear-quadratic dose-response functions. This is usually complemented by a more direct visualization where the data are subdivided into distinct dose categories and the effect level is quantified for each of these categories. Such model-free computations, however, can be quite dependent on the arbitrary choice of the cutpoints in dose. The method proposed here largely avoids this arbitrariness by choosing a dose category width-constant on a log scale-to obtain the desired degree of smoothing, and then superimposing results for all placements of the resulting log-dose grids. The method is applied to leukemia and solid cancer mortality of the A-bomb survivors.     

Effect of recent changes in atomic bomb survivor dosimetry on cancer mortality risk estimates

The Radiation Effects Research Foundation has recently implemented a new dosimetry system, DS02, to replace the previous system, DS86. This paper assesses the effect of the change on risk estimates for radiation-related solid cancer and leukemia mortality. The changes in dose estimates were smaller than many had anticipated, with the primary systematic change being an increase of about 10% in gamma-ray estimates for both cities. In particular, an anticipated large increase of the neutron component in Hiroshima for low-dose survivors did not materialize. However, DS02 improves on DS86 in many details, including the specifics of the radiation released by the bombs and the effects of shielding by structures and terrain. The data used here extend the last reported follow-up for solid cancers by 3 years, with a total of 10,085 deaths, and extends the follow-up for leukemia by 10 years, with a total of 296 deaths. For both solid cancer and leukemia, estimated age-time patterns and sex difference are virtually unchanged by the dosimetry revision. The estimates of solid-cancer radiation risk per sievert and the curvilinear dose response for leukemia are both decreased by about 8% by the dosimetry revision, due to the increase in the gamma-ray dose estimates. The apparent shape of the dose response is virtually unchanged by the dosimetry revision, but for solid cancers, the additional 3 years of follow-up has some effect. In particular, there is for the first time a statistically significant upward curvature for solid cancer on the restricted dose range 0-2 Sv. However, the low-dose slope of a linear-quadratic fit to that dose range should probably not be relied on for risk estimation, since that is substantially smaller than the linear slopes on ranges 0-1 Sv, 0-0.5 Sv, and 0- 0.25 Sv. Although it was anticipated that the new dosimetry system might reduce some apparent dose overestimates for Nagasaki factory workers, this did not materialize, and factory workers have significantly lower risk estimates. Whether or not one makes allowance for this, there is no statistically significant city difference in the estimated cancer risk.     

Flexible dose-response models for Japanese atomic bomb survivor data: Bayesian estimation and prediction of cancer risk

Generalised absolute risk models were fitted to the latest Japanese atomic bomb survivor cancer incidence data using Bayesian Markov Chain Monte Carlo methods, taking account of random errors in the DS86 dose estimates. The resulting uncertainty distributions in the relative risk model parameters were used to derive uncertainties in population cancer risks for a current UK population. Because of evidence for irregularities in the low-dose dose response, flexible dose-response models were used, consisting of a linear-quadratic-exponential model, used to model the high-dose part of the dose response, together with piecewise-linear adjustments for the two lowest dose groups. Following an assumed administered dose of 0.001 Sv, lifetime leukaemia radiation-induced incidence risks were estimated to be 1.11 x 10(-2) Sv(-1) (95% Bayesian CI -0.61, 2.38) using this model. Following an assumed administered dose of 0.001 Sv, lifetime solid cancer radiation-induced incidence risks were calculated to be 7.28 x 10(-2) Sv(-1) (95% Bayesian CI -10.63, 22.10) using this model. Overall, cancer incidence risks predicted by Bayesian Markov Chain Monte Carlo methods are similar to those derived by classical likelihood-based methods and which form the basis of established estimates of radiation-induced cancer risk.     

An analysis of various aspects of atomic bomb dose estimation at RERF using data on acute radiation symptoms

The dose-response curves for acute radiation symptoms reported by atomic bomb survivors are compared by dose estimation method (the method used to calculate the transmission factor), shielding category, and city. Circular symmetry is also investigated. It is found that response rates for acute symptoms differ considerably by dose estimation method and shielding category even after controlling for both gamma and neutron exposure as well as for city, sex, and age at the time of the bomb. One explanation of these results is that the doses of survivors in Japanese type houses estimated by the nine parameter method are subject to less random measurement error, while doses of those survivors who were in the open and shielded by terrain, who were totally shielded by concrete buildings, and who were in factories are subject to especially large random errors. The degree to which systematic bias contributes to these differences could not be determined. These results have important implications for comparisons between cities since Nagasaki includes a far greater proportion of survivors in shielding categories showing weak dose-response relationships than does Hiroshima. The hypothesis that doses might be higher in the westerly direction in Hiroshima is not supported by acute effects analyses, but excess acute effects are found in the north of Hiroshima.     

Postoperative cataract cases among atomic bomb survivors: radiation dose response and threshold

Recent evidence argues against a high threshold dose for vision-impairing radiation-induced cataractogenesis. We conducted logistic regression analysis to estimate the dose response and used a likelihood profile procedure to determine the best-fitting threshold model among 3761 A-bomb survivors who underwent medical examinations during 2000-2002 for whom radiation dose estimates were available, including 479 postoperative cataract cases. The analyses indicated a statistically significant dose-response increase in the prevalence of postoperative cataracts [odds ratio (OR), 1.39; 95% confidence interval (CI), 1.24-1.55] at 1 Gy, with no indication of upward curvature in the dose response. The dose response was suggestive when the restricted dose range of 0 to 1 Gy was examined. A nonsignificant dose threshold of 0.1 Gy (95% CI, <0-0.8) was found. The prevalence of postoperative cataracts in A-bomb survivors increased significantly with A-bomb radiation dose. The estimate (0.1 Gy) and upper bound (0.8 Gy) of the dose threshold for operative cataract prevalence was much lower than the threshold of 2-5 Gy usually assumed by the radiation protection community and was statistically compatible with no threshold at all.     

Effect of radiation dose on the height of atomic bomb survivors: a longitudinal study

We conducted a longitudinal analysis of height after age 20 for atomic bomb survivors in the Adult Health Study (AHS) cohort. The measurements we used were made from July 1958 to June 1998 (AHS examination cycles 1-20). We analyzed only the subjects with known atomic bomb radiation doses, excluding those who were not in the city at the time of bombing (ATB) and those exposed in utero. We also excluded from the analysis measurements made after the occurrence of vertebral fracture. The total number of subjects was 11,862, and the total number of measurements was 109,770; the mean number of measurements per subject was 9.25. Assuming that stature after age 20 is approximately constant, a simple mixed-effects model was fitted to stature after age 20, and linear dose effects for young ATB subjects were modeled for both sexes. The estimated mean heights for subjects born in 1945 in Hiroshima were 166.0 cm for men and 155.4 cm for women. The sex difference in height was 10.6 cm, with men significantly taller than women (P < 0.001). The difference between the cities was not significant (P = 0.162). The birth cohort effects per decade were -1.7 cm for men (P < 0.001) and -2.1 cm for women (P < 0.001). A reduction of stature due to radiation exposure was observed for individuals of both sexes who were below 19 years of age ATB (95% confidence interval, 17-21 years), and the dose effect was larger for women than for men (P = 0.028). The estimated effects per gray for those who were age 0 ATB were -1.2 cm for men and -2.0 cm for women and for those who were age 10 ATB were-0.57 cm for men and -0.96 cm for women.     

Organ doses from radiation therapy in atomic bomb survivors

Previous surveys of radiation therapy among the Life Span Study (LSS) population at the Radiation Effects Research Foundation (RERF) revealed that 1,670 (1.4%) of the LSS participants received radiation treatments before 1984. The data on therapeutic radiation doses are indispensable for studying the relationship between radiation treatments and subsequent cancer occurrences. In this study, the radiation treatments were reproduced experimentally to determine the scattered radiation doses. The experiments were conducted using a female human phantom and various radiation sources, including a medium-voltage X-ray machine and a (60)Co gamma-ray source. Doses were measured using thermoluminescence dosimetry and ionization chambers. Radiation doses were determined for the salivary glands, thyroid gland, breast, lung, stomach, colon, ovary and active bone marrow. The results have been used for documenting the organ doses received by patients in previous surveys. The contribution of therapeutic irradiation to the occurrence of chromosome aberrations was studied using data on doses to active bone marrow from both radiation treatments and atomic bomb exposures in 26 RERF Adult Health Study participants. The results suggest that radiation treatments contributed to a large part of their frequencies of stable-type chromosome aberrations. The therapeutic radiation doses determined in the present study are available for investigating the effects of therapeutic irradiation on the subsequent primary cancers among atomic bomb survivors who received radiation treatments.     

Analysis of the DS86 atomic bomb radiation dosimetry methods using data on severe epilation

This report presents a reanalysis of the Hiroshima and Nagasaki data on severe epilation as an acute radiation effect using both the new DS86 and the old T65D dosimetries. The focus of the report is on several aspects of the data which have previously been examined by Jablon et al (ABCC TR 12-70, 1970) and Gilbert and Ohara [Radiat. Res. 100, 124-138 (1984)]. The report examines the uniformity of epilation response across shielding category, across sex and age, and in terms of interactions between city, sex, age, and shielding category; it also investigates the apparent relative biological effectiveness (RBE) of neutrons in the DS86 dose compared with the T65D dose, using both within- and between-city information. In addition the report discusses evidence for nonlinearity in epilation response. The epilation response function exhibits nonlinearity in terms of both a marked increase in slope at about 0.75 Gy, and then, beginning at about 2.5 Gy, a leveling off and eventual decrease in response. The principal conclusions of the report are as follows. The use of the DS86 dosimetry rather than T65D increases the apparent RBE of neutrons compared with gamma dose from approximately 5 to 10. At these values of RBE the slope of the dose response, in a middle range from 0.75-2.5 Gy, is about 165% greater using DS86 than T65D. With respect to the interactions of sex, city, and shielding method, the size and significance of virtually all nonuniformities in epilation response seem using T65D are also evident with DS86. Additionally it seems difficult to find any evidence that DS86 is an improved predictor of epilation response over T65D. Finally, the fact that the nonlinearity in dose response and apparent actual downturn in epilation occurrence rate at the high end of dose is more striking with DS86 than with T65D is found to be due primarily to the common practice of truncating all T65D doses to 600 rad.     

Standard errors for EM estimates in generalized linear models with random effects

A procedure is derived for computing standard errors of EM estimates in generalized linear models with random effects. Quadrature formulas are used to approximate the integrals in the EM algorithm, where two different approaches are pursued, i.e., Gauss-Hermite quadrature in the case of Gaussian random effects and nonparametric maximum likelihood estimation for an unspecified random effect distribution. An approximation of the expected Fisher information matrix is derived from an expansion of the EM estimating equations. This allows for inferential arguments based on EM estimates, as demonstrated by an example and simulations.     

The effect of changes in dosimetry on cancer mortality risk estimates in the atomic bomb survivors

In the spring of 1986 the Radiation Effects Research Foundation (RERF) received a new atomic bomb dosimetry system. This report presents the comparisons of leukemia and nonleukemia cancer mortality risk estimates under the old and new dosimetries. In terms of total kerma (essentially whole-body gamma plus neutron exposure), risk estimates for both classes of cancer are 75-85% higher with the new dosimetry. This and other summary comparisons allow for possible nonlinearity at high estimated doses. Changes are also considered in relation to organ doses and assumptions about the relative biological effectiveness (RBE) of neutrons. Without regard to RBE, the risk estimates for total organ dose are essentially unchanged by the dosimetry revision. However, with increasing assumed values of RBE, the estimated low-LET risk decreases much less rapidly under the new dosimetry, due to the smaller neutron component. Thus at an assumed constant RBE of 10, for example, the effect of the dosimetry revision is to increase organ dose risk estimates, relative to those based on the old dosimetry, by 30% for nonleukemia and 80% for leukemia. At an RBE of 20 these increases are 72 and 136%, respectively. A number of other issues are discussed. The city difference in dose is no longer statistically significant, even at an RBE of one. Estimation of RBE is even less feasible with new dosimetry. There is substantial question of the linearity in dose response, in the sense of a leveling off at higher doses. Finally, some indication is given of how risks estimated from this dosimetry and the current data may compare to widely used estimates based largely on the RERF data with the previous dosimetry.     

No evidence for increased tumor rates below 200 mSv in the atomic bomb survivors data

We investigated for which doses a significantly increased tumor rate can be seen in the RERF Life Span Study data sets on mortality or incidence of solid tumors. No significant increase was found below about 200 mSv. Received: 28 April 1997 / Accepted in revised form: 2 June 1997     

Effect of radiation on age at menopause among atomic bomb survivors

Exposure to ionizing radiation has been thought to induce ovarian failure and premature menopause. Proximally exposed female atomic bomb survivors were reported to experience menopause immediately after the exposure more often than those who were distally exposed. However, it remains unclear whether such effects were caused by physical injury and psychological trauma or by direct effects of radiation on the ovaries. The objective of this study was to see if there are any late health effects associated with the exposure to atomic bomb radiation in terms of age at menopause in a cohort of 21,259 Life Span Study female A-bomb survivors. Excess absolute rates (EAR) of natural and artificial menopause were estimated using Poisson regression. A linear threshold model with a knot at 0.40 Gy [95% confidence interval (CI): 0.13, 0.62] was the best fit for a dose response of natural menopause (EAR at 1 Gy at age of 50 years = 19.4/1,000 person-years, 95% CI: 10.4, 30.8) and a linear threshold model with a knot at 0.22 Gy (95% CI: 0.14, 0.34) was the best fit for artificial menopause (EAR at 1 Gy at age of 50 years for females who were exposed at age of 20 years = 14.5/1,000 person-years, 95% CI: 10.2, 20.1). Effect modification by attained age indicated that EARs peaked around 50 years of age for both natural and artificial menopause. Although effect modification by age at exposure was not significant for natural menopause, the EAR for artificial menopause tended to be larger in females exposed at young ages. On the cumulative incidence curve of natural menopause, the median age at menopause was 0.3 years younger in females exposed to radiation of 1 Gy compared with unexposed females. The median age was 1 year younger for combined natural and artificial menopause in the same comparison. In conclusion, age at menopause was thought to decrease with increasing radiation dose for both natural and artificial menopause occurring at least 5 years after the exposure.     

The children of parents exposed to atomic bombs: estimates of the genetic doubling dose of radiation for humans.

The data collected in Hiroshima and Nagasaki during the past 40 years on the children of survivors of the atomic bombings and on the children of a suitable control population are analyzed on the basis of the newly revised estimates of radiation doses. No statistically significant effects emerge with respect to eight different indicators. Since, however, it may confidently be assumed some mutations were induced, we have taken the data at face value and calculated the minimal gametic doubling doses of acute radiation for the individual indicators at various probability levels. An effort has also been made to calculate the most probable doubling dose for the indicators combined. The latter value is between 1.7 and 2.2 Sv. It is suggested the appropriate figure for chronic radiation would be between 3.4 and 4.5 Sv. These estimates suggest humans are less sensitive to the genetic effects of radiation than has been assumed on the basis of past extrapolations from experiments with mice.     

Type 2 diabetes,obesity, and breast cancer risk among Japanese women of the atomic bomb survivor cohort

BackgroundMuch less is known about diabetes than obesity as a predictor of breast cancer incidence and most previous studies have been conducted in white populations. Therefore, this project within the Radiation Effects Research Foundation’s cohort of Japanese atomic bomb survivors aimed to determine the independent contributions of obesity and diabetes to develop breast cancer.MethodsAfter excluding women with unknown A-bomb radiation dose, a radiation dose of ≥100 mGy, a pre-existing history of breast cancer, and missing body mass index (BMI), the analysis included 29,818 women. Breast cancer status and deaths until 2009 were identified from cancer registries and vital records. Cox regression with age as the time metric was applied to estimate hazard ratios (HR) and 95% confidence intervals (CI) for BMI and diabetes status as time-varying exposures alone and in combination while adjusting for known confounders.ResultsDiabetes prevalence increased from 2.6% to 5.3% and 7.5% from the first to the second and third data collection. During 27.6 ± 12.2 years of follow-up, 703 women had developed breast cancer (mean age of 66.0 ± 12.9 years) and 31 (4.4%) had been diagnosed with diabetes. A diagnosis of diabetes was not significantly associated with breast cancer incidence without (HR 1.12, 95% CI 0.77–1.64) and with BMI (HR 1.01, 95% CI 0.69–1.49) as a covariate. The respective HRs for overweight and obesity were 1.61 (95% CI 1.34–1.93) and 2.04 (95% CI 1.40–2.97).ConclusionsAmong a long-time Japanese cohort, excess body weight but not a diabetes diagnosis was significantly associated with breast cancer risk.