Shared uncertainty in measurement error problems, with application to Nevada Test Site fallout data

In radiation epidemiology, it is often necessary to use mathematical models in the absence of direct measurements of individual doses. When complex models are used as surrogates for direct measurements to estimate individual doses that occurred almost 50 years ago, dose estimates will be associated with considerable error, this error being a mixture of (a) classical measurement error due to individual data such as diet histories and (b) Berkson measurement error associated with various aspects of the dosimetry system. In the Nevada Test Site(NTS) Thyroid Disease Study, the Berkson measurement errors are correlated within strata. This article concerns the development of statistical methods for inference about risk of radiation dose on thyroid disease, methods that account for the complex error structure inherence in the problem. Bayesian methods using Markov chain Monte Carlo and Monte-Carlo expectation-maximization methods are described, with both sharing a key Metropolis-Hastings step. Regression calibration is also considered, but we show that regression calibration does not use the correlation structure of the Berkson errors. Our methods are applied to the NTS Study, where we find a strong dose-response relationship between dose and thyroiditis. We conclude that full consideration of mixtures of Berkson and classical uncertainties in reconstructed individual doses are important for quantifying the dose response and its credibility/confidence interval. Using regression calibration and expectation values for individual doses can lead to a substantial underestimation of the excess relative risk per gray and its 95% confidence intervals.     

An estimate of the magnitude of random errors in the DS86 dosimetry from data on chromosome aberrations and severe epilation

An analysis of the proportion of cells with chromosome aberrations in cultured blood lymphocytes from A-bomb survivors in Hiroshima and Nagasaki reveals that the dose-response relationship using DS86 assigned dose is significantly steeper in the subsample of individuals who reported severe epilation after the bombings than in those who did not report severe epilation. This effect is due either to random errors in the DS86 dose assignments or to individual differences in sensitivity to radiation, or to both. In this paper, working within a class of dosimetry error models, we estimate the magnitude of random dosimetry errors which would be required to account for all of the difference in the observed dose response between people who did and did not report severe epilation under the assumption that random dosimetry error is the only cause of the effect. We conclude that random dosimetry errors in the range 45 to 50% of true dose are necessary to explain completely the difference in dose response between the two epilation groups. We discuss evidence that the contribution of individual differences in radiation sensitivity to the observed epilation effect is likely to be small, so that random dosimetry errors may be the major cause of this effect.     

The observed relationship between the occurrence of acute radiation effects and leukemia mortality among A-bomb survivors

In an analysis of a follow-up study of a fixed population of 73,330 atomic bomb survivors in Hiroshima and Nagasaki, the slope of an estimated dose response between ionizing radiation and leukemia mortality was found to be steeper (P less than 0.002), by a factor of 2.4, among those who reported epilation within 60 days of the bombings, compared to those who did not experience this sign of acute radiation exposure. The strength of this empirical finding as evidence of biological association in individual radiosensitivity for these two end points is studied here. The major factor complicating the interpretation of this finding as evidence of such an association is the degree of imprecision of the radiation dosimetry system used in assignment of radiation doses to the A-bomb survivors. Using models recently suggested for dealing with dosimetry errors in epidemiological analysis of the A-bomb survivor data, the sensitivity of the apparent association between leukemia mortality and severe epilation to the assumed level of dosimetry error is investigated.     

Cancer mortality following in utero exposure among offspring of female mayak worker cohort members

Little is known about long-term cancer risks following in utero radiation exposure. We evaluated the association between in utero radiation exposure and risk of solid cancer and leukemia mortality among 8,000 offspring, born from 1948-1988, of female workers at the Mayak Nuclear Facility in Ozyorsk, Russia. Mother's cumulative gamma radiation uterine dose during pregnancy served as a surrogate for fetal dose. We used Poisson regression methods to estimate relative risks (RRs) and 95% confidence intervals (CIs) of solid cancer and leukemia mortality associated with in utero radiation exposure and to quantify excess relative risks (ERRs) as a function of dose. Using currently available dosimetry information, 3,226 (40%) offspring were exposed in utero (mean dose = 54.5 mGy). Based on 75 deaths from solid cancers (28 exposed) and 12 (6 exposed) deaths from leukemia, in utero exposure status was not significantly associated with solid cancer: RR = 0.94, 95% CI 0.58 to 1.49; ERR/Gy = -0.1 (95% CI < -0.1 to 4.1), or leukemia mortality; RR = 1.65, 95% CI 0.52 to 5.27; ERR/Gy = -0.8 (95% CI < -0.8 to 46.9). These initial results provide no evidence that low-dose gamma in utero radiation exposure increases solid cancer or leukemia mortality risk, but the data are not inconsistent with such an increase. As the offspring cohort is relatively young, subsequent analyses based on larger case numbers are expected to provide more precise estimates of adult cancer mortality risk following in utero exposure to ionizing radiation.     

A reanalysis of thyroid neoplasms in the Israeli tinea capitis study accounting for dose uncertainties

In the 1940s and 1950s, children in Israel were treated for tinea capitis by irradiation to the scalp to induce epilation. Follow-up studies of these patients and of other radiation- exposed populations show an increased risk of malignant and benign thyroid tumors. Those analyses, however, assume that thyroid dose for individuals is estimated precisely without error. Failure to account for uncertainties in dosimetry may affect standard errors and bias dose-response estimates. For the Israeli tinea capitis study, we discuss sources of uncertainties and adjust dosimetry for uncertainties in the prediction of true dose from X-ray treatment parameters. We also account for missing ages at exposure for patients with multiple X-ray treatments, since only ages at first treatment are known, and for missing data on treatment center, which investigators use to define exposure. Our reanalysis of the dose response for thyroid cancer and benign thyroid tumors indicates that uncertainties in dosimetry have minimal effects on dose-response estimation and for inference on the modifying effects of age at first exposure, time since exposure, and other factors. Since the components of the dose uncertainties we describe are likely to be present in other epidemiological studies of patients treated with radiation, our analysis may provide a model for considering the potential role of these uncertainties.     

Thyroid cancer following scalp irradiation: a reanalysis accounting for uncertainty in dosimetry

In the 1940s and 1950s, over 20,000 children in Israel were treated for tinea capitis (scalp ringworm) by irradiation to induce epilation. Follow-up studies showed that the radiation exposure was associated with the development of malignant thyroid neoplasms. Despite this clear evidence of an effect, the magnitude of the dose-response relationship is much less clear because of probable errors in individual estimates of dose to the thyroid gland. Such errors have the potential to bias dose-response estimation, a potential that was not widely appreciated at the time of the original analyses. We revisit this issue, describing in detail how errors in dosimetry might occur, and we develop a new dose-response model that takes the uncertainties of the dosimetry into account. Our model for the uncertainty in dosimetry is a complex and new variant of the classical multiplicative Berkson error model, having components of classical multiplicative measurement error as well as missing data. Analysis of the tinea capitis data suggests that measurement error in the dosimetry has only a negligible effect on dose-response estimation and inference as well as on the modifying effect of age at exposure.     

The 15-Country Collaborative Study of Cancer Risk among Radiation Workers in the Nuclear Industry: study of errors in dosimetry

To provide direct estimates of cancer risk after low-dose protracted exposure to ionizing radiation, a large-scale epidemiological study of nuclear industry workers was conducted in 15 countries. As part of this study, identification and quantification of errors in historical recorded doses was conducted based on a review of dosimetric practices and technologies in participating facilities. The main sources of errors on doses from "high-energy" photons (100-3000 keV) were identified as the response of dosimeters in workplace exposure conditions and historical calibration practices. Errors related to dosimetry technology and radiation fields were quantified to derive period- and facility-specific estimates of bias and uncertainties in recorded doses. This was based on (1) an evaluation of predominant workplace radiation from measurement studies and dosimetry expert assessment and (2) an estimation of the energy and geometry response of dosimeters used historically in study facilities. Coefficients were derived to convert recorded doses to H(p) (10) and organ dose, taking into account different aspects of the calibration procedures. A parametric, lognormal error structure model was developed to describe errors in doses as a function of facility and time period. Doses from other radiation types, particularly neutrons and radionuclide intake, could not be adequately reconstructed in the framework of the 15-Country Study. Workers with substantial doses from these radiation types were therefore identified and excluded from analyses. Doses from "lower-energy" photons (<100 keV) and from "higher-energy" photons (>3 MeV) were estimated to be small.     

In vivo EPR tooth dosimetry for triage after a radiation event involving large populations

The management of radiation injuries following a catastrophic event where large numbers of people may have been exposed to life-threatening doses of ionizing radiation will rely critically on the availability and use of suitable biodosimetry methods. In vivo electron paramagnetic resonance (EPR) tooth dosimetry has a number of valuable and unique characteristics and capabilities that may help enable effective triage. We have produced a prototype of a deployable EPR tooth dosimeter and tested it in several in vitro and in vivo studies to characterize the performance and utility at the state of the art. This report focuses on recent advances in the technology, which strengthen the evidence that in vivo EPR tooth dosimetry can provide practical, accurate, and rapid measurements in the context of its intended use to help triage victims in the event of an improvised nuclear device. These advances provide evidence that the signal is stable, accurate to within 0.5 Gy, and can be successfully carried out in vivo. The stability over time of the radiation-induced EPR signal from whole teeth was measured to confirm its long-term stability and better characterize signal behavior in the hours following irradiation. Dosimetry measurements were taken for five pairs of natural human upper central incisors mounted within a simple anatomic mouth model that demonstrates the ability to achieve 0.5 Gy standard error of inverse dose prediction. An assessment of the use of intact upper incisors for dose estimation and screening was performed with volunteer subjects who have not been exposed to significant levels of ionizing radiation and patients who have undergone total body irradiation as part of bone marrow transplant procedures. Based on these and previous evaluations of the performance and use of the in vivo tooth dosimetry system, it is concluded that this system could be a very valuable resource to aid in the management of a massive radiological event.     

Re-evaluation of the RBE of 29 kV x-rays (mammography x-rays) relative to 220 kV x-rays using neoplastic transformation of human CGL1-hybrid cells

Neoplastic transformation of human CGL1-hybrid cells was examined after exposure to 29 kV x-rays (mammography x-rays) and conventional 220 kV x-rays. The study was designed to repeat, under well-defined irradiation and culture conditions, an earlier investigation by Frankenberg et al. (Radiat Res, 2002), and to assess the validity of the high RBE values of 29 kV x-rays that had been reported. The experiments with the two types of x-rays were performed simultaneously and shared the same controls. The transformation yields with both radiation qualities were fitted to the linear-quadratic dependence on absorbed dose, and a corresponding analysis was performed for the data earlier obtained by Frankenberg et al. The transformation yields in the present study exceed those in the earlier investigation substantially, and it appears that the difference reflects inadequate feeding conditions of the cell cultures in the early experiments. The standard error bands of the dose response curves are derived and are seen to be considerably more narrow in the present results. The lowest dose of the 29 kV x-rays was 1 Gy in both studies, and at this dose the RBE vs. the conventional x-rays has now been found to be 2 with a 95% confidence interval of 1.4-2.6. The previous result was about 3.2, but the 95% confidence is very broad for these data. The estimated limit at low doses is 3.4 in the present experiments with a confidence interval that extends from less than 2 to large values.     

Calculating confidence intervals for prediction error in microarray classification using resampling

Cross-validation based point estimates of prediction accuracy are frequently reported in microarray class prediction problems. However these point estimates can be highly variable, particularly for small sample numbers, and it would be useful to provide confidence intervals of prediction accuracy. We performed an extensive study of existing confidence interval methods and compared their performance in terms of empirical coverage and width. We developed a bootstrap case cross-validation (BCCV) resampling scheme and defined several confidence interval methods using BCCV with and without bias-correction. The widely used approach of basing confidence intervals on an independent binomial assumption of the leave-one-out cross-validation errors results in serious under-coverage of the true prediction error. Two split-sample based methods previously proposed in the literature tend to give overly conservative confidence intervals. Using BCCV resampling, the percentile confidence interval method was also found to be overly conservative without bias-correction, while the bias corrected accelerated (BCa) interval method of Efron returns substantially anti-conservative confidence intervals. We propose a simple bias reduction on the BCCV percentile interval. The method provides mildly conservative inference under all circumstances studied and outperforms the other methods in microarray applications with small to moderate sample sizes.     

A New Model of Biodosimetry to Integrate Low and High Doses

Biological dosimetry, that is the estimation of the dose of an exposure to ionizing radiation by a biological parameter, is a very important tool in cases of radiation accidents. The score of dicentric chromosomes, considered to be the most accurate method for biological dosimetry, for low LET radiation and up to 5 Gy, fits very well to a linear-quadratic model of dose-effect curve assuming the Poisson distribution. The accuracy of this estimation raises difficulties for doses over 5 Gy, the highest dose of the majority of dose-effect curves used in biological dosimetry. At doses over 5 Gy most cells show difficulties in reaching mitosis and cannot be used to score dicentric chromosomes. In the present study with the treatment of lymphocyte cultures with caffeine and the standardization of the culture time, metaphases for doses up to 25 Gy have been analyzed. Here we present a new model for biological dosimetry, which includes a Gompertz-type function as the dose response, and also takes into account the underdispersion of aberration-among-cell distribution. The new model allows the estimation of doses of exposures to ionizing radiation of up to 25 Gy. Moreover, the model is more effective in estimating whole and partial body exposures than the classical method based on linear and linear-quadratic functions, suggesting their effectiveness and great potential to be used after high dose exposures of radiation.     

Radiation dose verification of an X-ray based blood irradiator using EBT3 radiochromic films calibrated using Gamma Knife machine

AimBlood irradiators (BI) initial acceptance testing and routine annual dosimetry checks require radiation dose measurements in order to comply with regulatory requirements.BackgroundTraditionally thermo-luminescence dosimeters (TLD) have been used to measure the dose. The EBT3 film is reported to be a better dosimeter for low energy X-rays than its predecessors EBT2 and EBT. To the best of our knowledge, the use of EBT3 films to perform dosimetry on X-ray based BI has not been reported yet.Materials and methodsWe performed routine radiation dosimetry checks using EBT3 films on a new X-ray based BI and compared the results with TLD dosimetry. Calibration films were irradiated with radiation beam from a Co-60 Gamma Knife (GK) radiosurgery machine and, alternatively, using an Ir-192 high dose rate (HDR) brachytherapy device. The films were calibrated to cover a wide dose range from 1 to 40 Gy. Such a wide dose range has not been reported yet in BI film dosimetry.ResultsWe obtained a relative difference of about 6.6% between doses measured using TLD and those measured using EBT3 films. Both irradiation methods using GK or HDR were found to be adequate for the calibration of the EBT3 Gafchromic films.ConclusionsWe recommend the use of EBT3 films in routine X-ray based BI dosimetry checks. The presented method takes advantage of available radiotherapy equipment that can be efficiently used for EBT3 films calibration. The method is fast, reproducible and saves valuable medical physicist's time.     

Whole-blood immunoassay for γH2AX as a radiation biodosimetry assay with minimal sample preparation

The current state of the art in high-throughput minimally invasive radiation biodosimetry involves the collection of samples in the field and analysis at a centralized facility. We have developed a simple biological immunoassay for radiation exposure that could extend this analysis out of the laboratory into the field. Such a forward placed assay would facilitate triage of a potentially exposed population. The phosphorylation and localization of the histone H2AX at double-stranded DNA breaks has already been proven to be an adequate surrogate assay for reporting DNA damage proportional to radiation dose. Here, we develop an assay for phosphorylated H2AX directed against minimally processed sample lysates. We conduct preliminary verification of H2AX phosphorylation using irradiated mouse embryo fibroblast cultures. Additional dosimetry is performed using human blood samples irradiated ex vivo. The assay reports H2AX phosphorylation in human blood samples in response to ionizing radiation over a range of 0–5 Gy in a linear fashion, without requiring filtering, enrichment, or purification of the blood sample.     

Estimating Growth Reduction Dose for Polynomial Response Relationships

Mutation breeders often estimate an irradiation dose which causes a specified reduction in growth of a biological material. In this paper, we estimate the growth reduction dose (GRD) and its confidence interval, when the dose-response relationship is of a general polynomial form. We make a realistic assumption that the response at the control dose is a random variable. We compare the estimates, standard error and confidence intervals of GRD with those obtained under a situation where response at control dose is a known constant. We illustrate the procedure with data on the effect of gamma irradiation and Ethylmethane sulphate on shoot lengths of chickpea genotypes.     

Detection of partial-body exposure to ionizing radiation by the automatic detection of dicentrics

In accidental exposure to ionizing radiation, it is essential to estimate the dose received by the victims. Currently dicentric scoring is the best biological indicator of exposure. The standard biological dosimetry procedure (500 metaphases scored manually) is suitable for a few dose estimations, but the time needed for analysis can be problematic in the case of a large-scale accident. Recently, a new methodology using automatic detection of dicentrics has greatly decreased the time needed for dose estimation and preserves the accuracy of the estimation. However, the capability to detect nonhomogeneous partial-body exposures is an important advantage of dicentric scoring-based biodosimetry, and this remains to be tested with automatic scoring. Thus we analyzed the results obtained with in vitro blood dilutions and in real cases of accidental exposure (partial- or whole-body exposure) using manual scoring and automatic detection of dicentrics. We confirmed that automatic detection allows threefold quicker dicentric scoring than the manual procedure with similar dose estimations and uncertainty intervals. The results concerning partial-body exposures were particularly promising, and homogeneously exposed samples were correctly distinguished from heterogeneously exposed samples containing 5% to 75% of blood irradiated with 2 Gy. In addition, the results obtained for real accident cases were similar whatever the methodology used. This study demonstrates that automatic detection of dicentrics is a credible alternative for recent and acute cases of whole- and partial-body accidental exposures to ionizing radiation.     

Ischemic Heart Disease in Workers at Mayak PA: Latency of Incidence Risk after Radiation Exposure

We present an updated analysis of incidence and mortality from atherosclerotic induced ischemic heart diseases in the cohort of workers at the Mayak Production Association (PA). This cohort constitutes one of the most important sources for the assessment of radiation risk. It is exceptional because it comprises information on several other risk factors. While most of the workers have been exposed to external gamma radiation, a large proportion has additionally been exposed to internal radiation from inhaled plutonium. Compared to a previous study by Azizova et al. 2012, the updated dosimetry system MWDS-2008 has been applied and methods of analysis have been revised. We extend the analysis of the significant incidence risk and observe that main detrimental effects of external radiation exposure occur after more than about 30 years. For mortality, significant risk was found in males with an excess relative risk per dose of 0.09 (95% CI: 0.02; 0.16) while risk was insignificant for females. With respect to internal radiation exposure no association to risk could be established.     

Cytogenetic damage in lymphocytes for the purpose of dose reconstruction: a review of three recent radiation accidents

The analysis of chromosomal aberrations in peripheral blood of radiation accident victims is an established method of biological dosimetry. The dose estimate on the basis of an in vitro calibration curve is straightforward when the radiation exposure is homogeneous and the analysis not delayed. In recent years three radiation accidents occurred, where the irradiation or sampling conditions precluded a simple estimation of the dose. During the Georgian accident soldiers carried in their pockets small sources of 137Cs leading to partial and protracted body exposures. During the Tokai-mura accident, three employees involved in the process of 235U enrichment were exposed to very high doses of gamma rays and neutrons. During the Bialystok accident, five patients with breast cancer undergoing radiotherapy were exposed to a single dose of electrons which reached about 100 Gy. In the present paper the approaches chosen to estimate, by cytogenetic methods, the doses absorbed by the people involved in the accidents are described.     

Biological dosimetry intercomparison exercise: an evaluation of triage and routine mode results by robust methods

Well-defined protocols and quality management standards are indispensable for biological dosimetry laboratories. Participation in periodic proficiency testing by interlaboratory comparisons is also required. This harmonization is essential if a cooperative network is used to respond to a mass casualty event. Here we present an international intercomparison based on dicentric chromosome analysis for dose assessment performed in the framework of the IAEA Regional Latin American RLA/9/054 Project. The exercise involved 14 laboratories, 8 from Latin America and 6 from Europe. The performance of each laboratory and the reproducibility of the exercise were evaluated using robust methods described in ISO standards. The study was based on the analysis of slides from samples irradiated with 0.75 (DI) and 2.5 Gy (DII). Laboratories were required to score the frequency of dicentrics and convert them to estimated doses, using their own dose-effect curves, after the analysis of 50 or 100 cells (triage mode) and after conventional scoring of 500 cells or 100 dicentrics. In the conntional scoring, at both doses, all reported frequencies were considered as satisfactory, and two reported doses were considered as questionable. The analysis of the data dispersion among the dicentric frequencies and among doses indicated a better reproducibility for estimated doses (15.6% for DI and 8.8% for DII) than for frequencies (24.4% for DI and 11.4% for DII), expressed by the coefficient of variation. In the two triage modes, although robust analysis classified some reported frequencies or doses as unsatisfactory or questionable, all estimated doses were in agreement with the accepted error of ±0.5 Gy. However, at the DI dose and for 50 scored cells, 5 out of the 14 reported confidence intervals that included zero dose and could be interpreted as false negatives. This improved with 100 cells, where only one confidence interval included zero dose. At the DII dose, all estimations fell within ±0.5 Gy of the reference dose interval. The results obtained in this triage exercise indicated that it is better to report doses than frequencies. Overall, in both triage and conventional scoring modes, the laboratory performances were satisfactory for mutual cooperation purposes. These data reinforce the view that collaborative networking in the case of a mass casualty event can be successful.     

Optimized matching of film dosimetry with calculated doses for IMRT quality assurance

We develop a new method with a global optimization for registering films to calculate doses for intensity-modulated radiation therapy (IMRT) and intensity-modulated radiosurgery (IMRS) quality assurance (QA). Both absolute point dosimetry and two-dimensional (2D) film dosimetry are performed through the IMRT and IMRS using Clinac 21EX's 120 millenium MLC and BrainLab's micro-MLC, respectively. The measured and calculated dose distributions are superimposed by coincidence of their origins, followed by comparison of the point doses at all matched positions. Then, with the optimization algorithm the setup error of the dosimeter is corrected. An example of IMRT cases shows that the average percentage showing 3% of dose difference for 10 patients has been reduced from 19% to 9%, before and after optimization and weight, respectively. Similar results are obtained for IMRS. This method dramatically reduces the difference between measured and calculated dose distributions in all cases investigated.     

Likelihood-based experimental design for estimation of ED50

In selecting the best dosage choice for the estimation of ED50, it is natural to try to minimize the length of the confidence intervals. In this presentation, the dose allocation that minimizes the length of the likelihood-based confidence intervals is presented and compared with alternative allocations that have been proposed based on the length of different types of confidence intervals, such as those based on the asymptotic variance or on Fieller's Theorem. Effective strategies to deal with the parameter dependence of these allocations are explored. A series of experiments to evaluate the effect of small doses per fraction on the radiation tolerance of the rat cervical spinal cord provide the motivation and an illustration for the proposed procedures.