Abstract
Purpose: In the framework of RENEB, several biodosimetry exercises were conducted analyzing different endpoints. Among them, the analysis of translocations is considered the most useful method for retrospective biodosimetry due to the relative stability of their frequency with post irradiation time. The aim of this study was to harmonize the accuracy of translocation-based biodosimetry within the RENEB consortium.
Materials and methods: An initial telescoring exercise analyzing FISH metaphase images was done to harmonize chromosome aberration descriptions. Then two blind intercomparison exercises (IE) were performed, by sending irradiated blood samples to each partner. Samples were cultured and stained by each partner using their standard protocol and translocation frequency was used to produce dose estimates.
Results: The coefficient of variation in the 1st IE (CV = 0.34) was higher than in the 2nd IE (CV = 0.16 and 0.23 in the two samples analyzed), for the genomic frequency of total translocations. Z-score analysis revealed that eight out of 10 and 17 out of 20 dose estimates were satisfactory in the 1st and 2nd IE, respectively.
Conclusions: The results obtained indicate that, despite the problems identified in few partners, which can be corrected, the RENEB consortium is able to carry out retrospective biodosimetry analyzing the frequency of translocations by FISH.
Keywords:
Introduction
Biological dosimetry allows the estimation of dose from an exposure to ionizing radiation by the analysis of a biomarker. The gold standard method for recent and acute exposures is the analysis of the frequency of dicentric chromosomes in peripheral blood lymphocytes. In addition, other methodologies including the analysis of translocations in metaphases, micronuclei in binucleated cells or ring chromosomes by using premature chromosome condensation assays are also recognised as useful (International Atomic Energy Agency [IAEA] 2011). Moreover, other methodologies are recently being introduced into the biological dosimetry field, e.g. γH2AX foci assay, optically stimulated luminescence/electron paramagnetic resonance, gene expression analysis and others. Most of these methods are mainly accurate for recent and acute exposures, but their sensitivity decreases with time after exposure. The only method capable of dose estimation years after exposure or in follow-up studies is the analysis of translocations (IAEA Citation2011), because their frequency remains relatively stable with post irradiation time (Salassidis et al. Citation1995, Lindholm et al. Citation1998, Duran et al. Citation2009, IAEA Citation2011).
In the RENEB (Realising the European Network in Biodosimetry) project, several quality-controlled biodosimetry intercomparison assays were conducted analyzing different endpoints in order to establish the operational basis of the network. One of these endpoints was translocations, analyzed by fluorescent in situ hybridization (FISH). Therefore, the aim of the present exercise was to carry out this assay, including expert laboratories and others less experienced in FISH analysis, to determine the efficiency of the different laboratories in the analysis of translocations and in dose estimation. Before starting the practical part of the intercomparison, an initial harmonization exercise was carried out sending 50 FISH images of cells with different chromosome aberrations. These images were sent electronically to the participants for telescoring. Then, two FISH translocation intercomparisons were done. The first intercomparison exercise (1st IE), involved the shipment of an irradiated sample to the partners for processing, analysis and dose estimation. Further training was offered to the partners, where required, to provide a sound basis for future dose estimation. A second intercomparison exercise (2nd IE) served to improve the efficiency of dose estimation and harmonization of the network
Materials and methods
Participant laboratories
The 11 participant laboratories can be seen in . The harmonization exercise by telescoring images was carried out in 2012. The 1st IE was performed in 2013 and the 2nd IE in 2014.
Table 1. Participant laboratories
Telescoring exercise
As an exercise of harmonization, 50 FISH-coded images of cells containing chromosome aberrations were sent electronically to the participant laboratories. The images came from four different laboratories, and 40 cells showed one colour painting and 10 three colour painting. The laboratories were asked to score them and to describe the chromosome aberrations according to the modified PAINT nomenclature (Tucker et al. Citation1995, Knehr et al. Citation1998).
Samples, irradiation and delivery
Ten laboratories took part in the 1st IE. Peripheral blood samples were obtained, with informed consent, from a healthy male donor aged 53, in accordance with the Ethics Commission on Animal and Human Experimentation of the Universitat Autònoma de Barcelona (UAB) approved procedure. The samples were irradiated at 2 Gy with a Cs137 source (IBL437C, CIS Biointernational, GiF sur Yvette, France) located at the UAB, at a dose rate of 5.4 Gy/min. The samples were irradiated at 37 °C and maintained at this temperature for 2 h post irradiation. Then samples were packaged and delivered, at room temperature, by express services to the different laboratories, arriving approximately 24 h after irradiation. In this exercise, temperature changes and doses received by the samples during transport were not monitored. Shipments were declared as UN 3373 Biological Substance Category B.
Again 10 laboratories took part in the 2nd IE. Nine had taken part in the 1st IE, but one laboratory was no longer able to continue and a new partner was recruited. Blood was taken from two healthy donors, a 31-year-old female (Re5) and a 22-year-old male (Re6). The irradiation was performed at the BfS laboratory, with a 137-Cs gamma ray source (HWM D2000). Doses of 0.85 Gy (Re5) and 2.7 Gy (Re6) were given at a dose rate of 0.478 Gy/min. The samples were irradiated at 37 °C and kept for 2 h at 37 °C. Coded blood samples were then sent to the participants by express service. Shipments were again declared as UN 3373 Biological Substance Category B. Each shipment contained one sample of a low dose (Re5) and a high dose (Re6). Furthermore, in this second exercise, each package included a temperature logger and dosemeter (glass dosemeter, Typ SC-2) to monitor the temperature and record any dose received by the samples during transport.
Aberrations scoring and dose estimates
Each laboratory set up blood cultures following standard protocols, and the FISH technique was applied according to the individual laboratory’s procedure. The chromosomes painted by each laboratory are shown in . The different painting strategies cover different proportions of the genome, therefore to compare the frequencies of translocations between laboratories the observed frequencies were converted to genomic frequencies (FG), according to Lucas and co-workers proposed model (Lucas et al. Citation1992, Lucas and Deng Citation2000). The chromosome aberrations involving painted chromosomes can be described by different nomenclatures. This allows to score different types of chromosome aberrations (i.e. one- or two-way, apparently simple etc.), in all cells or in stable cells (those without dicentrics, rings or acentric fragments). In order to compare the same data, it was intended that all laboratories scored the frequency of total translocations in 500 cells. However, as can be seen in , some laboratories carried out dose estimates using other types of translocations which were not considered in the present study.
Table 2. FISH strategy of each partner.
The partners produced dose estimates for the coded samples using their own FISH dose-effect calibration curves for translocations. However, three partners in the 1st IE (partners 2, 5 and 8) and two laboratories in the 2nd IE (partners 5 and 8) did not have their own FISH dose-effect curves. For these partners, the FG of total translocations was interpolated to their dicentric dose-effect curve, a procedure recommended when no FISH dose-effect curve is available (IAEA Citation2011). In the dose-effect curve coefficients used in the present intercomparison are shown. The majority of the dose estimates were done blindly, but it must be pointed out that, due to various circumstances, the work was delayed in the laboratories of partners 7 (1st IE) and 3 (2nd IE) and their dose estimates were produced after the sample codes were uncovered. However, the translocation frequency and dose estimation from partners 3 and 7 were in line with those from the other participants and their results are included in this report. The lack of a full blind procedure is noted for the sake of transparency.
Table 3. Total translocations dose-effect curve coefficients from all partners except 5 and 8, whose coefficients correspond to their dicentrics dose-effect curves.
Statistics
To compare the FG of translocations, two analyses were carried out. First, the robust z-score (Algorithm A, ISO 5725-5:1998-6.2-) was used to analyze deviations from the mean genomic frequency. Values of the z-score between 2.0 and 3.0 are regarded as questionable and higher than 3 or below −3 as unsatisfactory. Second, to determine if there were differences between partners a one-way ANOVA with Tukey’s correction for multiple comparisons was performed in the 1st IE. In the 2nd IE, where the distribution of translocations among cells was known, a non-parametric ANOVA (Kruskal-Wallis) was used.
The expected Poisson distribution of translocations among cells was checked using the normalised unit u of the dispersion index (Rao and Chakravarti Citation1956, Savage Citation1970).
Dose estimates were mainly carried out using ‘Dose estimation’ software (Ainsbury and Lloyd Citation2010). To compare dose estimates between laboratories, the robust z-score was also used.
Results
Telescoring exercise
In total, 400 metaphase chromosome aberrations descriptions (40 cells ×10 laboratories) were done for one colour FISH. Forty metaphase descriptions, those from the laboratories that supplied the images, were considered as reference, so 360 new descriptions were carried out. A total of 86.1% of these descriptions exactly matched those of the reference laboratory. For three colour FISH, 90 new metaphase descriptions were done, and 81.1% were in agreement with the reference description. Globally, the metaphase description of aberrations was 84.7% consistent for all laboratories and the discrepancies were discussed in an annual RENEB meeting, before the start of the 1st IE
First intercomparison exercise
The results of this exercise are summarized in . In relation to the FG of total translocations, the mean frequency was 0.438, ranging from 0.214–0.696. The robust z-score analysis showed satisfactory results for all partners, indicating that no frequency differed significantly from the mean. However, a wide range of FG was detected by a coefficient of variation (CV) of 0.34. This variability was also seen when a one-way ANOVA was applied. Significant differences with the highest FG (partner 2), were observed for partners 4, 5, 7, 8, 9 and 11. On the other hand, significant differences with the lowest FG (partner 7) were observed for partners 1, 2, 4, 6 and 10 (in all cases p < .05).
Table 4. Frequency of translocations in 1st intercomparison exercise.
In relation to dose estimates (), the CV (0.20) was lower than for translocation frequencies. Two of 10 dose estimates were questionable after the robust z-score analysis. Both were higher than 3 Gy, which is at least a 50% higher than the real dose of 2 Gy.
Figure 1. Dose estimates in the first intercomparison exercise. The sample was irradiated at 2 Gy. Empty diamonds indicate agreement between the dose of irradiation and the dose estimates (robust z-score analysis). Half filled diamonds indicate a questionable dose estimates. Error bars show 95% confidence intervals.
Initially, in the 1st IE, the partners presented their results in different ways, e.g. some laboratories sent the aberrations scored but did not calculate the frequencies, u or FG, making it difficult to carry out comparisons and dose estimates. A conclusion of the 1st IE exercise was that a better harmonization in the management of results was necessary. As some laboratories were not as experienced in the management of FISH data, an excel file was designed to automatically obtain other data of interest such as the variance, calculating the u statistic from the distribution of translocations among cells and the genomic frequency, for use in the 2nd IE.
Second intercomparison exercise
In the 2nd IE, two irradiated blood samples were analyzed. One received a dose of 0.85 Gy (Re5) and the other 2.7 Gy (Re6) and and show the results of the two samples, respectively.
Table 5. Frequency and distribution of translocations among cells of the Re5 sample (2nd intercomparison exercise).
Table 6. Frequency and distribution of translocations among cells of the Re6 sample (2nd intercomparison exercise).
The robust z-score analysis for the FG of total translocations showed satisfactory results for all the partners, with a mean FG of 0.117 and a range from 0.083–0.140 for sample Re5. In this sample the coefficient of variation was clearly lower than in the 1st IE (0.16 vs. 0.34, respectively), and the non-parametric ANOVA analysis did not show significant differences between the FG of the partners. In the Re6 sample, the mean FG of translocations was 0.652 (range 0.478–0.921). The robust z-score analysis also showed satisfactory results for all the partners, with a coefficient of variation (0.22) slightly higher than in Re5, but again clearly lower than in the 1st IE. However, in contrast to Re5, the non-parametric ANOVA (Kruskal-Wallis) analysis for Re6 showed significant differences between the FG of some partners. The partner with the highest FG (partner 8) showed significant differences with partners 2 (p < .05) and 4, 5, 6, 9, 10 and 11 (p < .01). The partner with the lowest FG (partner 9) showed significant differences with partners 1 and 2 (p < .01) and 3 and 8 (p < .05).
Dose estimates for the Re5 sample () showed questionable results for two partners (6 and 8), using the robust z-score analysis. In both cases dose estimates were higher than the real dose. For the Re6 sample (), these two labs also showed the highest dose estimates, but the robust z-score analysis only showed an unsatisfactory result for partner 8.
Figure 2. Dose estimates in the second intercomparison exercise. The samples were irradiated at: (A) 0.85 Gy (Re5) and (B) 2.7 Gy (Re6). Empty diamonds indicate agreement between the dose of irradiation and the dose estimates (robust z-score analysis). Half filled diamonds indicate questionable dose estimates and full filled diamonds indicate unsatisfactory dose estimates. Error bars show 95% confidence intervals.
Discussion
Dicentrics and translocations are induced at a similar frequency by ionizing radiation, so both can be used for dose estimates if appropriate dose-effect curves are available. However, translocations cannot be usually detected by uniform stain and chromosome painting by FISH becomes a useful tool because an easy detection of exchange chromosome aberrations is possible.
It is assumed that translocations are more persistent with post irradiation time (Salassidis et al. Citation1995, Lindholm et al. Citation1998, Citation2002) than dicentrics, making the analysis of translocations the method to choose for retrospective biological dosimetry. However, the persistence of translocations is influenced by some factors, i.e. the type of translocations scored (Duran et al. Citation2009), if the exposure is whole body or partial (Guerrero-Carbajal et al. Citation1998, Duran et al. Citation2002), the initial dose (Duran et al. Citation2002, Sevan’kaev et al. Citation2005), the dose fractionation (Muller et al. Citation2005, Xunclà et al. Citation2008), the completeness and complexity of the aberration (Hande and Natarajan Citation1998, Lindholm et al. Citation1998, Xiao et al. Citation1999, Pala et al. Citation2001, Duran et al. Citation2002, Gardner and Tucker Citation2002) and the co-occurrence of translocations and unstable aberrations in the same cells (Rodríguez et al. Citation2004, Romm and Stephan Citation2004). Nevertheless, the analysis of translocations has been useful in retrospective biodosimetry studies (Salassidis et al. Citation1994, Citation1998, 2000, Finnon et al. Citation1995, Darroudi and Natarajan Citation1996, Citation2000, Snigiryova et al. Citation1997, Montoro et al. Citation2005).
To our knowledge, no quality-controlled intercomparison biodosimetry studies using the FISH translocation assay have been performed until now. Some participating partners of the present study had little or no experience with FISH until taking part in the RENEB project. It is for this reason that, after a telescoring exercise to harmonize the nomenclature to describe chromosome aberrations, the exercise was scheduled in two stages. In the first (1st IE), the laboratories received an irradiated blood sample and proceeded according to standard laboratory procedures. The partners with experience in FISH technique have dose-effect curves for different types of translocations, but other partners only have dose-effect curves for dicentrics. When using the latter type of curve, the total translocation frequency must be converted to FG to produce a dose estimate, therefore, in the present article we have only used total translocation data. Moreover, after the 1st IE a meeting to harmonize the FISH analysis was organized, with the participation of six partners. Only then was the 2nd IE carried out.
In relation to the FG of total translocations, in the 1st IE the coefficient of variation (0.34) was clearly higher than in the 2nd assay (0.16 and 0.22 for Re5 and Re6 samples, respectively), indicating that in general the score of translocations improved and hence the usefulness of training and intercomparison exercises. Partners 2 and 6, with the highest FG in the 1st IE showed FG closer to the mean in the 2nd IE. When the partners were ranked from highest to lowest FG, no trend was observed between the 1st and 2nd IE.
In the 1st IE, the two partners with the highest FG were those with questionable dose estimates. Partner 2 used their dicentric calibration curve to estimate the dose of the 1st IE sample. However, before the 2nd IE this partner constructed a FISH dose-effect curve, and in the 2nd IE, their dose estimates were much more precise. Partner 6 showed FG very close to the mean, indicating successful translocation scoring in the 2nd IE. However, their dose estimates remained higher than the true dose, especially for the Re5 sample where the estimated dose was questionable. In this case, the reason for the elevated dose estimates can be attributed to their dose-effect curve, established at the beginning of FISH dosimetric studies with slightly different scoring criteria. This partner is now establishing a new dose-response curve by FISH.
In the 2nd IE, partner 8 showed questionable (Re5) and unsatisfactory (Re6) dose estimates. The FG was in line with the other partners for the Re5 sample, but it was clearly higher than the other partners for the Re6 sample. Partner 8 also produced the highest doses estimates for both samples compared to the other partners. All the doses estimated by partner 8 were made using their dicentric calibration curve and a similar problem has occurred with dicentric dose estimates from this lab (data not shown). The combination of the effect of the calibration curve and the high frequency of translocations scored in the Re6 sample gave a dose estimate of 5.48 Gy, 2-fold higher than the real one.
Overall, an improvement was seen between the 1st and 2nd IE in the dose estimates produced by RENEB partners. In the 1st IE, two out of 10 dose estimates were questionable and in the 2nd IE, two were questionable and one unsatisfactory out of a total of 20. Those classified as questionable or unsatisfactory were all over estimates of the true dose, indicating that in these cases a correction must be done for accurate dose estimation in cases of real exposures. As possible reasons for these results have been established, they can be corrected with further training and repeating the dicentric dose-response curve or by establishing a FISH calibration curve. A final consideration is that part of the variability in dose estimates is inherent to this type of studies. First, because the different laboratories established their own dose-response curves using different types of radiation (i.e. gamma rays from Co60 or Cs137, or X-rays of different energies) and different dose rates. Second, because the type of radiation and/or the dose rate received by the sample can be different from the one used in the establishment of the dose-effect curve.
Conclusion
In summary, despite the problems identified in few partners, the results obtained indicate that analysis of translocations gives reproducible biodosimetry results with a good level of accuracy. This means that the RENEB consortium is able to carry out retrospective biodosimetry using the FISH translocation assay in case of large-scale studies.
Acknowledgements
This work received financial support from the European Seventh Framework Programme (295513). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Disclosure statement
The authors report no conflicts of interest. The authors alone are response for the content and writing of the paper.
References
- Ainsbury EA, Lloyd DC. 2010. Dose estimation software for radiation biodosimetry. Health Phys. 98:290–295.
- Darroudi F, Natarajan AT. 1996. Biological dosimetric studies in the Chernobyl radiation accident, on populations living in the contaminated areas (Gomel regions) and in Estonian clean-up workers, using FISH technique, The Radiological Consequences of the Chernobyl Accident (Proc. Mtg Minsk, 1996). Eur 16544 EN:1067–1072.
- Darroudi F, Natarajan AT. 2000. Application of FISH chromosome painting assay for dose reconstruction: state of the art and current views. Radiat Prot Dosim. 88:51–58.
- Duran A, Barquinero JF, Caballín MR, Ribas M, Barrios L. 2009. Persistence of radiation-induced chromosome aberrations in a long-term cell culture. Radiat Res. 171:425–437
- Duran A, Barquinero JF, Caballín MR, Ribas M, Puig P, Egozcue J, Barrios L. 2002. Suitability of FISH painting techniques for the detection of partial-body irradiations for biological dosimetry. Radiat Res. 157:461–468.
- Finnon P, Lloyd DC, Edwards AA. 1995. Fluorescence in situ hybridization detection of chromosomal aberrations in human lymphocytes: applicability to biological dosimetry. Int J Radiat Biol. 68:429–435.
- Gardner SN, Tucker J D. 2002. The cellular lethality of radiation radiation-induced chromosome translocations in human lymphocytes. Radiat Res. 157:539–552.
- Guerrero-Carbajal YC, Moquet JE, Edwards, AA, Lloyd DC. 1998. The persistence of FISH translocations for retrospective biological dosimetry after simulated whole or partial body irradiation. Radiat Prot Dosim. 76:159–168.
- Hande MP, Natarajan AT. 1998. Induction and persistence of cytogenetic damage in mouse splenocytes following whole-body X irradiation analysed by fluorescence in situ hybridization. IV. Dose response. Int J Radiat Biol. 74:441–448.
- International Atomic Energy Agency (IAEA). 2011. Cytogenetic dosimetry: applications in preparedness for and response to radiation emergencies. Vienna: IAEA.
- Knehr S, Zitzelsberger H, Bauchinger M. 1998. FISH-based analysis of radiation-induced chromosomal aberrations using different nomenclature systems. Int J Radiat Biol. 73:135–141.
- Lindholm C, Tekkel M, Veidebaum T, Ilus T, Salomaa S. 1998. Persistence of translocations after accidental exposure to ionizing radiation. Int J Radiat Biol. 74:565–571.
- Lindholm C, H. Romm, G. Stephan, E. Schmid, J. Moquet, A. Edwards. 2002. Intercomparison of translocation and dicentric frequencies between laboratories in a follow-up of the radiological accident in Estonia. Int J Radiat Biol. 78:883–890.
- Lucas JN, Awa A, Straume T, Poggensee M, Kodama Y, Nakano M, Ohtaki K, Weier HU, Pinkel D, Gray J, Littlefield G. 1992. Rapid translocation frequency analysis in humans decades after exposure to ionizing radiation. Int J Radiat Biol. 62:53–63.
- Lucas JN, Deng W. 2000. Views on issues in radiation biodosimetry based on chromosome translocations measured by FISH. Radiat Protect Dosim. 88:77–86.
- Montoro A, Rodríguez P, Almonacid M, Villaescusa JI, Verdú G, Caballín MR, Barrios L, Barquinero JF. 2005. Biological dosimetry in a group of radiologists by the analysis of dicentrics and translocations. Radiat Res. 164:612–617.
- Muller I, Geinitz H, Braselmann H, Baumgartner A, Fasan A, Thamm R, Molls M, Meinke V, Zitzelsberger H. 2005. Time-course of radiation-induced chromosomal aberrations in tumor patients after radiotherapy. Int J Radiat Oncol Biol Phys. 63:1214–1220.
- Pala FS, Moquet JE, Edwards AA, Lloyd DC. 2001. In vitro transmission of chromosomal aberrations through mitosis in human lymphocytes. Mutat Res. 474:139–146.
- Rao CR, Chakravarti IM. 1956. Some small sample tests of significance for a Poisson distribution. Biometrics. 12:264–282.
- Rodríguez P, Montoro A, Barquinero JF, Caballin MR, Villaescusa I, Barrios L. 2004. Analysis of translocations in stable cells and their implications in retrospective biological dosimetry. Radiat Res. 162:31–38.
- Romm H, Stephan G. 2004. Dose dependency of FISH-detected translocations in stable and unstable cells after Cs gamma irradiation of human lymphocytes in vitro. Cytogenet Genome Res. 104:162–167.
- Salassidis K, Schmid E, Peter RU, Braselmann H, Bauchinger M. 1994. Dicentric and translocation analysis for retrospective dose estimation in humans exposed to ionising radiation during the Chernobyl power plant accident. Mutat Res. 311:39–48.
- Salassidis K, Georgiadou-Schumacher V, Braselmann H, Müller P, Peter RU, Bauchinger M. 1995. Chromosome painting in highly irradiated Chernobyl victims: a follow-up study to evaluate the stability of symmetrical translocations and the influence of clonal aberrations for retrospective dose estimation. Int J Radiat Biol. 68:257–262.
- Salassidis K, Braselmann H, Okladnikova ND, Pressl S, Stephan G, Snigiryova G, Bauchinger M. 1998. Analysis of symmetrical translocations for retrospective biodosimetry in radiation workers of the Mayak nuclear-industrial complex (Southern Urals) using FISH-chromosome painting. Int J Radiat Biol. 74:431–439.
- Savage JRK. 1970. Sites of radiation induced chromosome exchanges. Curr Top Radiat Res. 6:129–194.
- Sevan’kaev AV, Lloyd D, Edwards AA, Khvostunov IK, Mikhailova GF, Golub EV, Shepel NN, Nadejina NM, Galstian IA, Barrios L, et al. 2005. A cytogenetic follow-up of some highly irradiated victims of the Chernobyl accident. Radiat Prot Dosim. 113:152–161.
- Snigiryova G, Braselmann H, Salassidis K, Shevchenko V, Bauchinger M. 1997. Retrospective biodosimetry of Chernobyl clean-up workers using chromosome painting and conventional chromosome analysis. Int J Radiat Biol. 71:119–127.
- Tucker JD, Morgan WF, Awa AA, Bauchinger M, Blakely D, Cornforth MN, Littlefield LG, Natarajan AT, Shaserre CA. 1995. Proposed system for scoring structural aberrations detected by chromosome painting. Cytogenet Cell Genet. 68:211–222.
- Xiao Y, Darroudi F, Grigorova M, Natarajan AT. 1999. Induction and persistence of chromosomal exchanges in mouse bone marrow cells following whole-body exposure to X-rays. Int J Radiat Biol. 75:1119–1128.
- Xunclà M, Barquinero JF, Caballín MR, Craven-Bartle J, Ribas M, De Vega JM, Barrios L. 2008. Cytogenetic damage induced by radiotherapy. Evaluation of protection by amifostine and analysis of chromosome aberrations persistence. Int J Radiat Biol. 84:243–251.