A statistical framework for radiation dose estimation with uncertainty quantification from the γ-H2AX assay

Identifying Bayesian optimal experiments for uncertain biochemical pathway models

Pharmacodynamic (PD) models are mathematical models of cellular reaction networks that include drug mechanisms of action. These models are useful for studying predictive therapeutic outcomes of novel drug therapies in silico. However, PD models are known to possess significant uncertainty with respect to constituent parameter data, leading to uncertainty in the model predictions. Furthermore, experimental data to calibrate these models is often limited or unavailable for novel pathways. In this study, we present a Bayesian optimal experimental design approach for improving PD model prediction accuracy. We then apply our method using simulated experimental data to account for uncertainty in hypothetical laboratory measurements. This leads to a probabilistic prediction of drug performance and a quantitative measure of which prospective laboratory experiment will optimally reduce prediction uncertainty in the PD model. The methods proposed here provide a way forward for uncertainty quantification and guided experimental design for models of novel biological pathways.

Evaluation of γ-H2AX foci distribution among different peripheral blood mononucleated cell subtypes

INTRODUCTION

The detection of γ-H2AX foci in peripheral blood mononucleated cells (PBMCs) has been incorporated as an early assay for biological dosimetry. However, overdispersion in the γ-H2AX foci distribution is generally reported. In a previous study from our group, it was suggested that overdispersion could be caused by the fact that when evaluating PBMCs, different cell subtypes are analyzed, and that these could differ in their radiosensitivity. This would cause a mixture of different frequencies that would result in the overdispersion observed.

OBJECTIVES

The objective of this study was to evaluate both the possible differences in the radiosensitivities of the different cell subtypes present in the PBMCs and to evaluate the distribution of γ-H2AX foci in each cell subtype.

MATERIALS AND METHODS

Peripheral blood samples from three healthy donors were obtained and total PBMCs, and CD3+, CD4+, CD8+, CD19+, and CD56+ cells were separated. Cells were irradiated with 1 and 2 Gy and incubated at 37 °C for 1, 2, 4, and 24 h. Sham-irradiated cells were also analyzed. γ-H2AX foci were detected after immunofluorescence staining and analyzed automatically using a Metafer Scanning System. For each condition, 250 nuclei were considered.

RESULTS

When the results from each donor were compared, no observable significant differences between donors were observed. When the different cell subtypes were compared, CD8+ cells showed the highest mean of γ-H2AX foci in all post-irradiation time points. The cell type that showed the lowest γ-H2AX foci frequency was CD56+. The frequencies observed in CD4+ and CD19+ cells fluctuated between CD8+ and CD56+ without any clear pattern. For all cell types evaluated, and at all post-irradiation times, overdispersion in γ-H2AX foci distribution was significant. Independent of the cell type evaluated the value of the variance was four times greater than that of the mean.

CONCLUSION

Although different PBMC subsets studied showed different radiation sensitivity, these differences did not explain the overdispersion observed in the γ-H2AX foci distribution after exposure to IR.

Biodose Tools: an R shiny application for biological dosimetry

INTRODUCTION

In the event of a radiological accident or incident, the aim of biological dosimetry is to convert the yield of a specific biomarker of exposure to ionizing radiation into an absorbed dose. Since the 1980s, various tools have been used to deal with the statistical procedures needed for biological dosimetry, and in general those who made several calculations for different biomarkers were based on closed source software. Here we present a new open source program, Biodose Tools, that has been developed under the umbrella of RENEB (Running the European Network of Biological and retrospective Physical dosimetry).

MATERIALS AND METHODS

The application has been developed using the R programming language and the shiny package as a framework to create a user-friendly online solution. Since no unique method exists for the different mathematical processes, several meetings and periodic correspondence were held in order to reach a consensus on the solutions to be implemented.

RESULTS

The current version 3.6.1 supports dose-effect fitting for dicentric and translocation assay. For dose estimation Biodose Tools implements those methods indicated in international guidelines and a specific method to assess heterogeneous exposures. The app can include information on the irradiation conditions to generate the calibration curve. Also, in the dose estimate, information about the accident can be included as well as the explanation of the results obtained. Because the app allows generating a report in various formats, it allows traceability of each biological dosimetry study carried out. The app has been used globally in different exercises and training, which has made it possible to find errors and improve the app itself. There are some features that still need consensus, such as curve fitting and dose estimation using micronucleus analysis. It is also planned to include a package dedicated to interlaboratory comparisons and the incorporation of Bayesian methods for dose estimation.

CONCLUSION

Biodose Tools provides an open-source solution for biological dosimetry laboratories. The consensus reached helps to harmonize the way in which uncertainties are calculated. In addition, because each laboratory can download and customize the app's source code, it offers a platform to integrate new features.

Optimal Stein-type goodness-of-fit tests for count data

Common count distributions, such as the Poisson (binomial) distribution for unbounded (bounded) counts considered here, can be characterized by appropriate Stein identities. These identities, in turn, might be utilized to define a corresponding goodness-of-fit (GoF) test, the test statistic of which involves the computation of weighted means for a user-selected weight function f. Here, the choice of f should be done with respect to the relevant alternative scenario, as it will have great impact on the GoF-test's performance. We derive the asymptotics of both the Poisson and binomial Stein-type GoF-statistic for general count distributions (we also briefly consider the negative-binomial case), such that the asymptotic power is easily computed for arbitrary alternatives. This allows for an efficient implementation of optimal Stein tests, that is, which are most powerful within a given class  F $\mathcal {F}$ of weight functions. The performance and application of the optimal Stein-type GoF-tests is investigated by simulations and several medical data examples.

Radiation dose estimation with time-since-exposure uncertainty using the [Formula: see text]-H2AX biomarker

To predict the health effects of accidental or therapeutic radiation exposure, one must estimate the radiation dose that person received. A well-known ionising radiation biomarker, phosphorylated [Formula: see text]-H2AX protein, is used to evaluate cell damage and is thus suitable for the dose estimation process. In this paper, we present new Bayesian methods that, in contrast to approaches where estimation is carried out at predetermined post-irradiation times, allow for uncertainty regarding the time since radiation exposure and, as a result, produce more precise results. We also use the Laplace approximation method, which drastically cuts down on the time needed to get results. Real data are used to illustrate the methods, and analyses indicate that the models might be a practical choice for the [Formula: see text]-H2AX biomarker dose estimation process.

BPIFA2 as a Novel Early Biomarker to Identify Fatal Radiation Injury After Radiation Exposure

Background

Current dosimeters cannot cope with the two tasks of medical rescue in the early stage of nuclear accident, the accurate determination of radiation exposure and the identification of patients with fatal radiation injury. As radiation can cause alterations in serum components, it is feasible to develop biomarkers for radiation injury from serum. This study aims to investigate whether serum BPIFA2 could be used as a potential biomarker of predicting fatal radiation injury in the early stage after nuclear accident.

Methods

A rabbit anti-mouse BPIFA2 polyclonal antibody was prepared to detect the expression of BPIFA2. C57BL/6J female mice were exposed to total body radiation (TBI) at different dose and Partial body radiation (PBI) at lethal dose to detect the dynamic changes of BPIFA2 in serum at different time points after irradiation by Western blot assay.

Results

BPIFA2 in mice serum were significantly increased at 1–12 h post-irradiation at .5–10 Gy, and increased again significantly at 3 d after 10 Gy irradiation with associated with mortality closely. It also increased rapidly after PBI and was closely related to injury degree, regardless whether the salivary glands were irradiated.

Conclusions

The increase of serum BPIFA2 is a novel early biomarker not only for identifying radiation exposure, but also for fatal radiation injury playing a vital role in rational use of medical resources, and greater efficiency of medical treatment to minimize casualties.

Quantification of histone H2AX phosphorylation in white blood cells induced by ex vivo gamma irradiation of whole blood by both flow cytometry and foci counting as a dose estimation in rapid triage

A quick, reliable, and reproducible biological assay to distinguish individuals with possible life-threatening risk following radiological or nuclear incidents remains a quest in biodosimetry. In this paper, we examined the use of a γ-H2AX assay as an early dose estimation for rapid triage based on both flow cytometry and image analyses. In the experiment, whole blood from 11 donors was irradiated ex vivo inside a water phantom by gamma rays from Co-60 at 0.51 Gy/min. After the lysis of red blood cells, the white blood cells were collected for immunofluorescence labeling of γ-H2AX, CD45, and nuclear stained for signal collection and visualization. Analysis by flow cytometry showed that the relative γ-H2AX intensities of lymphocytes and granulocytes increased linearly with absorbed doses from 0 to 6 Gy with a large variation among individuals observed above 2 Gy. The relative γ-H2AX intensities of lymphocytes assessed by two different laboratories were highly correlated (ICC = 0.979). Using confocal microscopic images, γ-H2AX foci were observed to be discretely distributed inside the nuclei and to increase proportionally with doses from 0 to 2 Gy, whereas large plagues of merged foci appeared at 4 and 6 Gy, resulting in the saturation of foci counts above 4 Gy. The number of total foci per cell as well as the number of foci per plane were significantly different at 0 vs 1 and 2 vs 4 Gy doses (p < 0.01). Blind tests at 0.5 Gy and 1 Gy doses showed that dose estimation by flow cytometry had a mean absolute difference of less than 0.5 Gy from the actual value. In conclusion, while flow cytometry can provide a dose estimation with an uncertainty of 0.5 Gy at doses ≤ 1 Gy, foci counting can identify merged foci that are prominent at doses ≥ 4 Gy.

DeepFoci: Deep learning-based algorithm for fast automatic analysis of DNA double-strand break ionizing radiation-induced foci

DNA double-strand breaks (DSBs), marked by ionizing radiation-induced (repair) foci (IRIFs), are the most serious DNA lesions and are dangerous to human health. IRIF quantification based on confocal microscopy represents the most sensitive and gold-standard method in radiation biodosimetry and allows research on DSB induction and repair at the molecular and single-cell levels. In this study, we introduce DeepFoci - a deep learning-based fully automatic method for IRIF counting and morphometric analysis. DeepFoci is designed to work with 3D multichannel data (trained for 53BP1 and γH2AX) and uses U-Net for nucleus segmentation and IRIF detection, together with maximally stable extremal region-based IRIF segmentation. The proposed method was trained and tested on challenging datasets consisting of mixtures of nonirradiated and irradiated cells of different types and IRIF characteristics - permanent cell lines (NHDFs, U-87) and primary cell cultures prepared from tumors and adjacent normal tissues of head and neck cancer patients. The cells were dosed with 0.5-8 Gy γ-rays and fixed at multiple (0-24 h) postirradiation times. Under all circumstances, DeepFoci quantified the number of IRIFs with the highest accuracy among current advanced algorithms. Moreover, while the detection error of DeepFoci remained comparable to the variability between two experienced experts, the software maintained its sensitivity and fidelity across dramatically different IRIF counts per nucleus. In addition, information was extracted on IRIF 3D morphometric features and repair protein colocalization within IRIFs. This approach allowed multiparameter IRIF categorization of single- or multichannel data, thereby refining the analysis of DSB repair processes and classification of patient tumors, with the potential to identify specific cell subclones. The developed software improves IRIF quantification for various practical applications (radiotherapy monitoring, biodosimetry, etc.) and opens the door to advanced DSB focus analysis and, in turn, a better understanding of (radiation-induced) DNA damage and repair.

Establishment and validation of surface model for biodosimetry based on γ-H2AX foci detection

INTRODUCTION

In the event of a radiation accident detecting γ-H2AX foci is being accepted as fast method for triage and dose assessment. However, due to their disappearance kinetics, published calibrations have been constructed at specific post-irradiation times.

OBJECTIVES

To develop a surface, or tridimensional, model to estimate doses at times not included in the calibration analysis, and to validate it.

MATERIALS AND METHODS

Calibration data was obtained irradiating peripheral mononucleated cells from one donor with radiation doses ranging from 0 to 3 Gy, and γ -H2AX foci were detected microscopically using a semi-automatic method, at different post-irradiation times from 0.5 to 24 h. For validation, in addition to the above-mentioned donor, blood samples from another donor were also used. Validation was done within the range of doses and post-irradiation times used in the calibration.

RESULTS

The calibration data clearly shows that at each analyzed time, the γ-H2AX foci frequency increases as dose increases, and for each dose this frequency decreases with post-irradiation time. The γ-H2AX foci nucleus distribution was clearly overdispersed, for this reason to obtain bidimensional and tridimensional dose-effect relationships no probability distribution was assumed, and linear and non-linear least squares weighted regression was used. In the two validation exercises for most evaluated samples, the 95% confidence limits of the estimated dose were between ±0.5 Gy of the real dose. No major differences were observed between donors.

CONCLUSION

In case of a suspected overexposure to radiation, the surface model here presented allows a correct dose estimation using γ-H2AX foci as biomarker. The advantage of this surface model is that it can be used at any post-irradiation time, in our model between 0.5 and 24 h.

The effect of data aggregation on dispersion estimates in count data models

For the modelling of count data, aggregation of the raw data over certain subgroups or predictor configurations is common practice. This is, for instance, the case for count data biomarkers of radiation exposure. Under the Poisson law, count data can be aggregated without loss of information on the Poisson parameter, which remains true if the Poisson assumption is relaxed towards quasi-Poisson. However, in biodosimetry in particular, but also beyond, the question of how the dispersion estimates for quasi-Poisson models behave under data aggregation have received little attention. Indeed, for real data sets featuring unexplained heterogeneities, dispersion estimates can increase strongly after aggregation, an effect which we will demonstrate and quantify explicitly for some scenarios. The increase in dispersion estimates implies an inflation of the parameter standard errors, which, however, by comparison with random effect models, can be shown to serve a corrective purpose. The phenomena are illustrated by γ-H2AX foci data as used for instance in radiation biodosimetry for the calibration of dose-response curves.

Improving the accuracy of dose estimates from automatically scored dicentric chromosomes by accounting for chromosome number

PURPOSE

The traditional workflow for biological dosimetry based on manual scoring of dicentric chromosomes is very time consuming. Especially for large-scale scenarios or for low-dose exposures, high cell numbers have to be analyzed, requiring alternative scoring strategies. Semi-automatic scoring of dicentric chromosomes provides an opportunity to speed up the standard workflow of biological dosimetry. Due to automatic metaphase and chromosome detection, the number of counted chromosomes per metaphase is variable. This can potentially introduce overdispersion and statistical methods for conventional, manual scoring might not be applicable to data obtained by automatic scoring of dicentric chromosomes, potentially resulting in biased dose estimates and underestimated uncertainties. The identification of sources for overdispersion enables the development of methods appropriately accounting for increased dispersion levels.

MATERIALS AND METHODS

Calibration curves based on in vitro irradiated (137-Cs; 0.44 Gy/min) blood from three healthy donors were analyzed for systematic overdispersion, especially at higher doses (>2 Gy) of low LET radiation. For each donor, 12 doses in the range of 0-6 Gy were scored semi-automatically. The effect of chromosome number as a potential cause for the observed overdispersion was assessed. Statistical methods based on interaction models accounting for the number of detected chromosomes were developed for the estimation of calibration curves, dose and corresponding uncertainties. The dose estimation was performed based on a Bayesian Markov-Chain-Monte-Carlo method, providing high flexibility regarding the implementation of priors, likelihood and the functional form of the association between predictors and dicentric counts. The proposed methods were validated by simulations based on cross-validation.

RESULTS

Increasing dose dependent overdispersion was observed for all three donors as well as considerable differences in dicentric counts between donors. Variations in the number of detected chromosomes between metaphases were identified as a major source for the observed overdispersion and the differences between donors. Persisting overdispersion beyond the contribution of chromosome number was modeled by a Negative Binomial distribution. Results from cross-validation suggested that the proposed statistical methods for dose estimation reduced bias in dose estimates, variability between dose estimates and improved the coverage of the estimated confidence intervals. However, the 95% confidence intervals were still slightly too permissive, suggesting additional unknown sources of apparent overdispersion.

CONCLUSIONS

A major source for the observed overdispersion could be identified, and statistical methods accounting for overdispersion introduced by variations in the number of detected chromosomes were developed, enabling more robust dose estimation and quantification of uncertainties for semi-automatic counting of dicentric chromosomes.

The 2019 Bo Lindell Laureate Lecture: On the use of interdisciplinary, stakeholder-driven, radiation protection research in support of medical uses of ionising radiation

Medical exposures form the largest manmade contributor to total ionising radiation exposure of the UK population. In recent years, new technologies have been developed to improve treatment and prognosis of individuals treated with radiation for diseases such as cancer. However, there is evidence of public, patient, and medical professional concern that radiation protection regulations and practices, as well as understanding of potential long-term adverse health effects of radiation exposure (in the context of other health risks), have not always 'kept pace' with technological developments in this field. This is a truly complex, multi-disciplinary problem for the modern world.The 'Radiation Theme' of the Public Health England and Newcastle University Health Protection Research Unit on 'Chemical and Radiation Threats and Hazards' is addressing this need, with a key focus on a genuinely interdisciplinary approach bringing together world-leading epidemiologists, radiation biologists, clinicians, statisticians, and artists. In addition, the project has a strong grounding in public, patient, and medical professional involvement in research. Similarly, the EU-CONCERT-funded LDLensRad project seeks to understand the mechanisms of action of low-dose ionising radiation in the lens of the eye, and the potential contribution to the development of cataract - in contemporary research, such projects will only be considered successful when they make use of expertise from a variety of fields and when they are able to demonstrate that the outputs are not only of benefit to society, but that society understands and welcomes the benefits. Finally, successful engagement, training, and retention of early career scientists within this field is crucial for sustainability of the research. Herein, the contribution of embedded interdisciplinary working, stakeholder involvement, and training of early career scientists to recent advancements in the field of medical (and wider) radiation protection research is discussed and considered.