Search

Purpose: To present Poisson exact goodness-of-fit tests as alternatives and complements to the asymptotic u-test, which is the most widely used in cytogenetic biodosimetry, to decide whether a sample of chromosomal aberrations in blood cells comes from an homogeneous or inhomogeneous exposure., Materials and Methods: Three Poisson exact goodness-of-fit test from the literature are introduced and implemented in the R environment. A Shiny R Studio application, named GOF Poisson, has been updated for the purpose of giving support to this work. The three exact tests and the u-test are applied in chromosomal aberration data from clinical and accidental radiation exposure patients., Results: It is observed how the u-test is not an appropriate approximation in small samples with small yield of chromosomal aberrations. Tools are provided to compute the three exact tests, which is not as trivial as the implementation of the u-test., Conclusions: Poisson exact goodness-of-fit tests should be considered jointly to the u-test for detecting inhomogeneous exposures in the cytogenetic biodosimetry practice.

Response to a large-scale radiological incident could require timely medical interventions to minimize radiation casualties. Proper medical care requires knowing the victim's radiation dose. When physical dosimetry is absent, radiation-specific chromosome aberration analysis can serve to estimate the absorbed dose in order to assist physicians in the medical management of radiation injuries. A mock exercise scenario was presented to six participating biodosimetry laboratories as one individual acutely exposed to Co under conditions suggesting whole-body exposure. The individual was not wearing a dosimeter and within 2-3 h of the incident began vomiting. The individual also had other medical symptoms indicating likelihood of a significant dose. Physicians managing the patient requested a dose estimate in order to develop a treatment plan. Participating laboratories in North and South America, Europe, and Asia were asked to evaluate more than 800 electronic images of metaphase cells from the patient to determine the dicentric yield and calculate a dose estimate with 95% confidence limits. All participants were blind to the physical dose until after submitting their estimates based on the dicentric chromosome assay (DCA). The exercise was successful since the mean biological dose estimate was 1.89 Gy whereas the actual physical dose was 2 Gy. This is well within the requirements for guidance of medical management. The exercise demonstrated that the most labor-intensive step in the entire process (visual evaluation of images) can be accelerated by taking advantage of world-wide expertise available on the Internet.

The bottleneck in data acquisition during biological dosimetry based on a dicentric assay is the need to score dicentrics in a large number of lymphocytes. One way to increase the capacity of a given laboratory is to use the ability of skilled operators from other laboratories. This can be done using image analysis systems and distributing images all around the world. Two exercises were conducted to test the efficiency of such an approach involving 10 laboratories. During the first exercise (E1), the participant laboratories analysed the same images derived from cells exposed to 0.5 and 3 Gy; 100 images were sent to all participants for both doses. Whatever the dose, only about half of the cells were complete with well-spread metaphases suitable for analysis. A coefficient of variation (CV) on the standard deviation of ∼15 % was obtained for both doses. The trueness was better for 3 Gy (0.6 %) than for 0.5 Gy (37.8 %). The number of estimated doses classified as satisfactory according to the z-score was 3 at 0.5 Gy and 8 at 3 Gy for 10 dose estimations. In the second exercise, an emergency situation was tested, each laboratory was required to score a different set of 50 images in 2 d extracted from 500 downloaded images derived from cells exposed to 0.5 Gy. Then the remaining 450 images had to be scored within a week. Using 50 different images, the CV on the estimated doses (79.2 %) was not as good as in E1, probably associated to a lower number of cells analysed (50 vs. 100) or from the fact that laboratories analysed a different set of images. The trueness for the dose was better after scoring 500 cells (22.5 %) than after 50 cells (26.8 %). For the 10 dose estimations, the number of doses classified as satisfactory according to the z-score was 9, for both 50 and 500 cells. Overall, the results obtained support the feasibility of networking using electronically transmitted images. However, before its implementation some issues should be elucidated, such as the number and resolution of the images to be sent, and the harmonisation of the scoring criteria. Additionally, a global website able to be used for the different regional networks, like Share Points, will be desirable to facilitate worldwide communication.