Is FISH painting an appropriate biological marker for dose estimates of suspected accidental radiation overexposure? A review of cases investigated in France from 1995 to 1996.

From 1995 to 1996 about 15 people suspected of being overexposed to ionizing radiation were referred to the Institute for Nuclear Safety and Protection in Fontenay-aux-Roses, France, for investigation by chromosome aberration analysis. Biological estimates of accidental overexposure were first obtained by scoring radio-induced unstable structural chromosome aberrations (dicentrics, centric rings, and fragments) in peripheral blood lymphocytes. For dose estimates, the yield of these chromosomal aberrations observed in 500 metaphases was compared with the laboratory dose-response relationship established from human blood irradiated in vitro (gamma-rays, 60Co, 0.5 Gy/min). To extend the possibilities of detecting DNA damage from earlier exposures by visualizing stable chromosome aberrations, chromosome painting by fluorescence in situ hybridization (FISH painting) was developed using a cocktail of three composite whole human chromosome-specific DNA probes (numbers 2, 4, and 12). A laboratory calibration curve for scoring terminal and/or reciprocal translocations was established for the same radiation quality and dose rate as those used for conventional cytogenetics (gamma-rays, 60Co, 0.5 Gy/min). For dosimetry purposes, it was also important to verify whether FISH painting could be applied to each human blood sample assessed for conventional expertise. For each individual, 2000 metaphases were scored for the presence or absence of reciprocal and terminal translocations. We present here a comparison between the results obtained by the two technologies for each of the cases studied separately. We describe their similarities or differences and discuss the suitability of using FISH painting for routine expertise analysis.


Introduction
This paper considers the possibility of using at low doses. Suspected overexposure to the detection of stable chromosome aberraionizing radiation is usually estimated by tions by a fluorescence in situ hybridization the number of unstable chromosome abertechnique (FISH painting) to estimate doses rations [dicentric (Dic) and centric rings] in cases of various accidental overexposures in peripheral lymphocytes of exposed to ionizing radiation, which generally occur individuals (1). The data obtained are then Abbreviations used: BrdU, 5-bromodeoxyuridine, thymidine analogue; DAPI, 4',6-diamidine-2'-phenylindole dihydrochloride; Dic, dicentric; EC, cells containing visible complex exchange; FISH, fluorescence in situ hybridization; FITC, fluorescein isothiocyanate; FPG, fluorescence plus Giemsa technique; Gy, gray; Ins, insertion; IPSN, Institute for Nuclear Safety and Protection; RX, X-rays; 2xSSC, saline-sodium citrate buffer concentrated 2-fold; TR, reciprocal translocation; TT, terminal translocation; (TR + TT) eq, genomic estimated translocations. calibrated against a standard dose-response curve established after analysis of human lymphocytes exposed in vitro (2). However, Dic chromosomes are unstable with time after exposure (3,4) and a biological dosimetry based on their detection alone has limitations with regard to past overexposure. Problems may be encountered in dose reconstruction when the time between exposure and analysis is considerable or even unknown. On the other hand, it appears that translocations persist for many years after exposure and that their scoring may be an indication of past overexposure. FISH painting using whole human chromosome-specific DNA probes has opened new possibilities for detecting some interchromosomal exchanges (i.e., translocations, insertions) using a cocktail of composite DNA probes specific to some chromosomes (5,6). The data obtained by the analysis of only a few chromosomes (the painted ones) generally are scaled up to full genomic frequency by assuming a random distribution of break points. FISH painting, therefore, provides easy identification and classification of radiation-induced chromosome aberrations (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16).
In cases of retrospective dose estimation translocation frequencies must also be calibrated against standard dose-response curves established in vitro using the same experimental protocols. This is particularly important because interlaboratory variabilities are suspected and a recent report clearly shows that care must be taken when interpreting FISH data from more than one laboratory (17).
Consequently, it appears that the best way to study the utility of FISH painting for dose assessment if overexposure is suspected is: a) to compare the scoring data obtained using this technique with those obtained by conventional cytogenetics for each case of accidental overexposure; and b) to establish an in vitro standard curve for translocation scoring using a quality of radiation and dose rate similar to those used for the laboratory reference curve for Dic scoring. This paper presents our preliminary results in this area of research.

Fluorescence in Situ Hybrdization
The hybridization protocol was applied according to Pinkel et al. (5) with some modifications. In brief, slides containing target DNA were dehydrated in ethanol series before a ribonudease A treatment (1 mg/ml in 2-fold concentrated saline-sodium citrate buffer [2 x SSC], 1 hr at 37°C, Boehringer Mannheim, Meylan, France) followed by a pepsin digestion (0.005% in HCI 0.01 N, 10 min at 37°C, Boehringer). The hybridization mixtures containing human whole chromosome-specific DNA probes (chromosomes 2, 4. and 12, Vysis, Voisins le Bretonneux, France) were premixed with unlabeled human competitive placental DNA enriched for repetitive DNA sequences (Cot 1 DNA) and incubated at 37°C for 1 hr 30 min. Target DNA was denaturated 3 min at 70°C in solution (70% formamide, 2 x SSC). DNA probes were then deposited on the prewarmed (10 min, 56°C) denaturated slides and hybridization occurred overnight in a humid chamber at 37°C. Posthybridization washings were done according to standard Vysis protocol with shortened incubation times (3 min). Chromosomal DNA was counterstained with 4',6-diamidine-2'phenylindole dihydrochloride ( Estimated Genomic Frequency ofAberratons Estimation of the genomic translocation frequency was carried out using the formula proposed by Lucas et al. (6,7), (Fp=2.05 fp(l-fp)Fg), which links the observed translocation frequency on the painted chromosomes (Fp) to the total genomic translocation frequency (Fg) according to the fraction of the genome painted (jf). For the chromosomes painted in this study (2, 4, and 12), 18.6% of the genome was hybridized corresponding to a detection efficiency of 31%. The total number of metaphases scored at each dose was corrected (see "Cell Equivalent" in Table 1) to correspond with the amount of information that would have been available if aberrations had been scored by G banding. This process led to a genomic estimation of translocations.

Criteria for Aberration Scoring and Curve Representations
Conventional scoring was performed on an Optiphot microscope (Nikon, Micro-m&anique, Evry, France). Only complete cells (i.e., 46 centromeres) in first division were analyzed for the presence of Dics (Nikon x 100 objective, NA 1.25, oil). FISH slides were analyzed with a Microphot-FXA (Nikon) fluorescence microscope (mercury lamp, 100 W) equipped with Nikon PlanApo x 60 objective (NA 1.4, oil) combined with filter blocks for simultaneous observation of fluorescein isothiocyanate (FITC)/rhodamine (Nikon) or DAPI/FITC/rhodamine (Nikon). A simple pass DAPI filter (Nikon) was used to verify chromosome shapes. For FISH painting, only complete-looking metaphases that had complete painted patterns were considered.
Cells were analyzed according to the following criteria. A bicolored chromosome exhibiting a single centromere in the painted (red or green) part was classified as reciprocal translocation(s) (TR). This translocation was complete if its reciprocal bicolored counterpart was observed or it was incomplete (TRi) if only one bicolored monocentric chromosome was seen accompanied by the corresponding painted fragment. A bicolored chromosome with the centromere located in the nonpainted part of the chromosome (i.e., DAPI colored) was termed terminal translocation (TT). Generally, a TT was always accompanied by a painted truncated chromosome. Insertions (Ins) looked similar to a painted chromosome piece inserted in a nonpainted one. Twocolor chromosomes with two or more Table 1. Comparison between the yields of unstable and stable chromosome aberrations produced in blood peripheral lymphocytes by in vitro irradiation with y-rays from 60Co at a dose rate of 0.5 Gy/min. Chromosome aberrations were scored by conventional cytogenetics and FISH painting.
Conventional  To establish a dose-response reference curve, a linear-quadratic regression model was applied to each point and best fitted curves obtained using an iteratively reweighted regression calculation (Sigma Plot, Jandel Scientific, San Raphael, CA) assuming that chromosome aberrations follow a Poisson law distribution.

Results
In ft,o Studies Table 1 shows data obtained by scoring unstable and stable chromosome aberrations induced in vitro in human lymphocytes by 7y-rays from a 60Co source at a dose rate of 0.5 Gy/min, as explained in "Methods." Results of Dic yields scored after conventional staining are listed in Table 1, which also shows the scoring by FISH painting of TR (complete and incomplete), TT, and Ins involving the painted chromosomes 2, 4, and 12. In this preliminary work, simplelooking complete and incomplete exchanges (i.e., bicolor chromosomes with only one color junction) or insertions (two color junctions) were considered. Exchanges between two painted chromosomes were scored as single events. Cells containing visible EC were scored separately and not included in the total data. To compare the yields of stable chromosome aberrations with those of the unstable aberrations, all observed frequencies were genomic estimated using the formula of Lucas et al. (6,7). So a full genome equivalent cell number (cell equivalent) was calculated and the corresponding genomic estimated translocation number was termed (TR+ TT)eq. Data from Table 1 were used to establish three dose-response fitted curves (Figure 1). In this figure the dose-effect relationship obtained from Dic yields scored by conventional cytogenetics is compared to the one obtained from the TReq yields for blood samples irradiated in vitro in the same conditions. The third dose-effect curve corresponds to the genomic estimated yield calculated for all types of translocations (TR+TT)eq. In this study, the level of genome equivalent TR scored by FISH painting is lower than the I scored by conventional cytoE when all types of translocation account, the two curves, i.4 translocations, are similar.

In Vivo Studies
The cases of suspected over ionizing radiation referred tc tory from 1995 to 1996 can roughly into two main catego sional (i.e., working with rac radioactive sources or close t public (usually not using ioi tion). Among the group of workers almost 40% carried physical dosimeter at the mc accident (data not shown). blood samples were coded according to the date of their laboratory ( Table 2). One c lyzed two times with blood sa rated by 9 weeks (cases L1 a quality of radiation exposure, this study were from diverse modes, as seen in Table 2. ' exposure to X-ray(s) (RX), 19 were external-acute and I cases A, E, F, 0, or recurren geneous for cases B, C, I. Tw Dic, fitted v TR eq, exp ---TR eq, fitted * (TR + TT) eq, exp -(TR + TT) eq, fitted 2 3 Dose, Gy Lsare taken in exposure for two people who stayed more e., Dics and than 30 months in geographical zones that possibly were contaminated. All the cases were different and it was impossible to arrange them by groups according to the rexposure to type of irradiation. Z our labora-Consequently, it was proposed to l be grouped classify these cases in four groups, I to IV, )ries-profesaccording to the delay between the suslioactivity or pected overexposure and the time of the *o them) and analysis. This classification was chosen to nizing radiatest the efficiency of translocation detection professional by FISH painting, which was used as a no personal bioindicator of past exposure. The results )ment of the obtained by FISH were compared with Individuals' those obtained using conventional cytoge-I by a letter netics (Dics scoring) for the same patients. arrival at the For this purpose, a minimum of 2000 cells ase was ana-per individual were scored by FISH paintmpling sepa-ing to detect stable chromosome aberra-.nd L2). The tions (except case A, where only 861 cells s involved in could be found). This number corresponds origins and to a full genome equivalent cell number of The cases of about 620 cells and the statistical uncer-)2Ir, or '37Cs tainty level became reliable with the 500 Localized for cells visually scored for Dics by convent but hetero-tional cytogenetics. In general, even for ro individuals conventional cytogenetics or for FISH iternal conta-painting, cells carrying only one aberration o other indiwere the most commonly found, especially uspected of for in vitro 7-ray irradiation at low doses. tion to 226Ra However, sometimes cells with multi-Lmination for aberrations were found in accidental overexposure cases, even at low doses estimated by physic dose reconstruction when possible. This was often seen in cases of radionuclides with beta-emission components. For this reason, some cases in Table 2 are scored two ways to distinguish two types of analysis: scoring excluding (single letter) or including (double letter) cells having many aberrations. This was true for case L, which , ' l was scored either taking into account one cell carrying many Dics (case LLI) or not (LI). The same procedure was used for cases that had complex exchanges detected by FISH painting (EC cells in Table 2). When possible, complex exchanges were 4 5 scored by transferring the complex patterns into a simple-looking base-type translocation (i.e., TR or TT). This was true for obtained by con-cases LL2-L2, E-EE, and A-AA. Case C had inting (60Co, dose one cell carrying an exchange that was too d using conven-complex; the cell was excluded from the e aberrations (TR ru. . f gthe FISH-paintresults. To facilitate comparison between the 12) Vertical bars Dic and the genomic estimated translocation are fitted curves yields, Table 2 gives the ratio between these square regression two types of aberrations per cell. These ratios listribution.
represent the disappearance of unstable  aCell with more than one dicentric (Multi Dic). Multi Dic is the number of cells having more than one unstable chromosome aberration. Each individual is coded by a letter. A single letter represents scoring data that do not take into account those cells having more than one aberration. Double letters represent scoring taking these kind of cells into account. One case, L, was analyzed for two periods; a number near the letter distinguishes the two scorings. chromosome aberrations with time after exposure, but the data obtained are often dissimilar. In groups I and II, where the delay between suspicion of overexposure and analysis is theoretically short enough so there is no Dic loss, the genomic estimated total translocation yield (TT+TR)eq often exceeds the Dic frequency. However, case LI surprisingly presents more Dics than translocations, even when the cell carrying many Dics was misscored (Li). When a new analysis was performed with a new blood sample 9 weeks later (L2-LL2), however, the observed ratio was the opposite, i.e., there were more translocations than Dics. Conversely, except for case C, no Dics were found for groups III (long delay) and IV (protracted), although some stable aberrations persisted at a level higher than background level (see Table 1).  Figure 2A shows a complete TR of chromosome 12 with a nonpainted one found in an overexposed patient. Figure 2B is an example of a complete TR between two painted chromosomes of another overexposed individual (numbers 4 and 12). Figure 2C illustrates an incomplete TRi between chromosomes 2 and 4.

Discussion
The primary purpose for using biological dosimetry in cases of suspected radiation overexposure is to help the medical staff devise a therapeutic strategy. Therefore, it is necessary to ascertain as quickly as possible the answers to such crucial questions as, "Was this person really overexposed to ionizing radiation?" and, "What was the dose estimation?" This paper describes the procedures we used in an attempt to answer these questions. Actually, the yield of unstable chromosome aberrations in phytohemagglutinin-stimulated human peripheral blood lymphocytes provide the most reliable biological indicator, provided the delay between irradiation and analysis does not exceed a few weeks to a few months (1). A gradual decrease in the frequency of cells carrying Dics generally is reported with increasing time after exposure (4). Some contrary examples were reported; for example, the presence of cells carrying unstable aberrations detected decades postirradiation in atomic survivors (3) or the case of some Thorotrast patient analysis [a-particle emission (19)], but these cases are rare. Thus, the presence of several Dics certainly is useful information in analyzing exposure status. The correspondence between Dic yield and dose estimates could be obtained through a calibration curve, which must be established in vitro under the same experimental conditions as those for analysis. This was the first step of our analysis and Figure 1 shows the laboratory reference curve obtained for 7-rays of 60Co at 0.5 Gy/min. In more than half the cases we analyzed, the delay between exposure and the cytogenetic analysis was more than two months so Dic scoring by conventional cytogenetics might present some limitations. The introduction of FISH techniques in our laboratory has allowed us to extend the spectrum of chromosome-type aberrations to stable ones that can be analyzed. It is reasonable to suggest that this Environmental Health Perspectives * Vol 105, Supplement 6 * December 1997 new technology could improve dose reconstruction in cases of past overexposure and probably be useful in new and old dose estimates as well.
What are the limitations in cases of low-dose exposure? To find answers to these questions we established a calibration curve using three-color painting for three chromosomes (2, 4, and 12) that differed enough in length to be easily distinguishable. This choice of chromosomes represents a mean of chromosome radiosensitivity: one seems to be more often involved in the formation of symmetrical exchanges (number 4), another seems to be less frequently involved (number 2), and the third (number 12) seems to have a medium involvement, as shown recently by Knehr et al. (16). Figure  1 presents data for this curve, which needs more scoring in the low-dose range to be really usable for dose estimation assay. Nevertheless, preliminary results show that the number of genomic estimated total translocations [i.e., (TR+ TT)eq] is similar to the number of dicentric ones but also that, in our case, the number of TR is lower. Even though the theoretical prediction on the equality of radiation-produced dicentrics and translocations (20) is not supported by many publications, further analyses are being done in our laboratory to resolve the apparent discrepancy between the number of TR scored by FISH painting and the number of dicentrics scored by conventional cytogenetics.
To eliminate any statistical bias, the applicability of FISH painting on in vivo accidental exposure was then tested using the same cocktail of DNA probes as the one used in establishing a reference curve. Our experience in the field of accidental overexposure shows that overexposure cases are rarely similar (Table 2). Differing radiation qualities ('y-rays, X-rays, P emission), modes of exposure (acute, heterogeneous), and possibly dose rates were involved. Consequently, the findings given in this study must be considered preliminary because the number of expertise cases is too small to draw clear conclusions. Nevertheless, classification of the scoring data in the four groups according to the supposed delay between suspicion of irradiation and the analysis ( Table 2) leads to some interesting observations, which are discussed below. First, even though the delay was short between suspicion of overexposure and analysis ( Table 2, Group I), no conclusion was possible because the number of cases was too small. It must be noted that case L apparently presented more Dics than translocations just after irradiation (LL1-L1). Two months later, however, the number of Dics decreased and the number of translocations appeared to have increased (LL2-L2). This internal exposure resulted from an accidental injection of strontium and the decrease of Dics might be because of the delay between elimination of the strontium (more than a week) and the moment of the second analysis (9 weeks later). Conversely, the apparent increase of stable chromosome aberrations as translocations might have occurred while the strontium was still in the body (more than 1 week) and may not have decreased in the 2 months before the Environmental Health Perspectives * Vol 105, Supplement 6 * December 1997 second analysis. The second group (Table 2) seems more homogeneous and, as expected with a longer delay period, the translocation level generally is higher than the Dic level. However, there are two exceptions. First is case A, which exhibits many fewer TR than TT, giving a higher value for Dics when compared with those for the TR. We cannot explain this observation. The second case (D) had localized irradiation to both hands. If the cell carrying a lot of Dics is taken into account (DD), the number of Dics is higher than the number of translocations. Nevertheless, our dose estimate supports the known discrepancy between a very heterogeneous irradiation to the hands and the whole-body integrated dose given by blood lymphocytes. In the third group Dics seem to have disappeared if, indeed, they ever existed, except for case C. The history of this patient was unclear and no conclusion could be drawn. Moreover, the presence of translocations in this third group cannot be explained by a simple effect of age because case B is still young but has a high translocation level. In fact, the effect of age on translocation frequency remains a confounding variable, as explained in the report of Chung et al. (21). Group IV gathers two protracted cases of overexposure in a contaminated area over 3 years. Analyses were done 5 months after return to the noncontaminated area and show a translocation level apparently higher than the background one. Whereas no Dics were observed, a significant yield of stable aberrations was found. It is difficult to ascertain whether this level is attributable to the 3 years period in the contaminated area because we do not know the background translocation frequencies of these people before the suspicion of overexposure. Note that with our present reference curve no translocation was found in control samples. All these first observations point out the necessity of stable chromosome aberration analysis when the delay between exposure and analysis increases.
A second observation is that a better understanding of the population background with regard to such factors as life habits, working conditions, and environmental situations is essential before using FISH painting as a biodosimeter. It is also necessary to solve the problem of age before validating translocation scoring as a biological indicator of suspicion of in vivo exposure. Table 2 shows examples of 40-to 50-yearold people with only a few translocations.
This study constitutes a preliminary step in our process of defining the possibilities of FISH painting for biological dosimetry expertise. Because of the limitations of a number of cases, no clear conclusion could be reached. In actuality these data provide more questions than answers in the case of varied suspected accidental overexposure.