The British government undertook a series of nuclear tests (NTs) at various sites in the South Pacific between 1952 and 1958. Associated with these atmospheric tests was an experimental programme, conducted largely at Maralinga in Australia, in which radioactivity was dispersed into the environment. This programme ended in 1963 although clean-up operations continued through to 1967 [1]. Additionally, UK personnel participated in a series of American tests based at Christmas Island in 1962. According to the Ministry of Defence (MoD) 22 347 veterans participated in at least one of these British and American tests of which ∼7000 were alive in 2017. Concerns were first raised in the early 1970s that the health of veterans of this testing programme and that of their children may have been adversely affected. Epidemiological studies examining mortality and cancer incidence in test veterans, carried out up to 1998, showed limited evidence of any detectable effect although this has since been revised to show a small excess in mortality (RR = 1.02, 90% CI 1.00–1.05, p = 0.04), associated with similar increased risks for both cancer and non-cancer diseases [2–4]. Despite this, questions as to whether veterans could have received sufficient radiation exposure to cause harm and, worry about potential genetic risk to future generations of any historical radiation exposure, persist [5].

Dose estimates for British NT veterans were based on film badge measurements of external dose, where available. According to National Radiological Protection Board-R214, only ∼22% of the entire population were monitored of which 8% (1804 participants) recorded a 'non-zero' dose, with 44 NT veterans categorized as receiving between 50 and 100 mSv and 36 as receiving a dose of >100 mSv [6]. Based this, the vast majority of NT veterans were exposed to no or low dose exposures only (low dose defined as less than 100 mSv). The primary health concern for individuals exposed to low-moderate doses is cancer although the debate for non-cancer diseases such as cardiovascular disease and cataracts arising after doses of less than 500 mSv, is ongoing [7]. The psychological impact of real and/or perceived low dose exposure is also of concern [5, 8]. In total, 759 NT veterans were identified by the UK MoD as potentially receiving higher doses to that recorded and categorized into 'special groups', such as those veterans who were involved in air plume sampling, cleaning of 'sampling' aircraft or, crew of HMS Diana who were tasked to sail through a nuclear plume [1]. Many of those present at test sites were involved in support roles, such as construction, transport or catering, however additionally, were directly involved with the actual tests, including working in contaminated areas in the days, weeks and months following each test. Such roles may not have been accounted for by the formal categorization into a special group. Fallout from atmospheric tests (e.g. GRAPPLE series in the South Pacific) and, from radioactivity which was dispersed into the environment during the Maralinga experimental programme in South Australia includes long-lived radionuclides such as Caesium-137, Strontium-90, Uranium-235/238, Plutonium-239, which if inhaled, ingested or otherwise internalized within the body would contribute to chronic radiation exposure with potential relevance for human health risks [9]. Apart from limited autopsy analysis there is no public record of any historical internal dose measurements. For further information on the British atmospheric and experimental testing programme and, on the potential sources and routes of exposure at the varying geographical sites, please refer to [10].

Ionising radiation induces DNA double strand breaks which are the critical lesion for the formation of structural chromosome aberrations [11]. Fluorescence in situ hybridisation based techniques, which 'paint' individual chromosomes, enable the detection of structural rearrangements such as reciprocal translocations and has been validated for use in the assessment of radiation doses [12, 13]. As reciprocal translocations are capable of long-term cellular transmission (in otherwise stable cells) their quantification can be informative of historical radiation exposures, including where many decades have passed [14–18]. Damage to DNA is acquired throughout life from a range of endogenous and exogenous sources however meaning that reciprocal translocations will accumulate with age, thus, their quantification reflects a lifetime of all exposures. With the application of M-FISH, where all chromosomes in the genome are uniquely 'painted', a much more complete picture of the complexity of chromosomal interchange 'patterns' are being revealed [19, 20]. Complex chromosome aberrations (rearrangements involving three or more breaks in two or more chromosomes) have been shown to be characteristically induced after exposure to low doses of high-linear energy transfer (LET) radiation, such as α-particle emitters [21]. The frequency and type of chromosome aberrations observed by multiplex in situ hybridisation (M-FISH) are thus informative of radiation exposure, dose and radiation quality.

The genetic and cytogenetic family trio (GCFT) study is the first study to obtain blood samples from a group of British NT veterans and their families [10]. The aim was to recruit NT veteran family trios (veteran, child, child's mother) who had 'special group' status and/or who had participated in two or more operations' including the GRAPPLE series and at Maralinga test sites', to ask whether heritable genetic effects could exist due to historical participation in the British nuclear testing programme. The examination for any differences in the frequency and spectra of de novo germline DNA mutations compared to control veteran families is reported elsewhere [22], whilst future publications will report on the occurrence of chromosomal aberrations in 1st generation adult children of NT and control veterans. Here we report the M-FISH findings to ask firstly, if there is any cytogenetic evidence of historical radiation exposure in the NT veterans and secondly, if there is any relationship between the occurrence of chromosome aberrations in veteran fathers with the de novo germline mutations in these families.

2.1. No difference in chromosome aberration frequencies between control and test veteran cohorts

Blood was received from 91 NT and control veterans and processed to collect 1st in vitro metaphase cells for analysis. Cultures from five veterans (1 NT and 4 control) either failed to culture or generate sufficient numbers of metaphase cells for M-FISH analysis. In total, we analysed 9379 and 7698 metaphase cells using M-FISH (24-colour karyotyping) from 48 NT and 38 control veteran samples, representing veteran servicemen from the army, Royal Airforce (RAF) and Royal Navy. The number of metaphase cells per sample ranged from 78 to 390 (median = 196); 18 samples had less than 150 cells analysed, with no significant differences detected between the cohorts (figure 1).

Figure 1. Number of metaphase cells analysed by M-FISH for NT and control veterans.

Download figure:

Standard image High-resolution image

We observed stable and unstable simple- and complex-type chromosome aberrations in both NT and control veteran's samples, finding no statistically significant difference in yield of any aberration type between the two cohorts (figure 2, supplementary table 1, supplementary tables 6(a) and (b); p > 0.2 unless reported). Specifically, overall frequencies of 1.621 ± 0.167 and 1.585 ± 0.244 simple exchanges/100 cells (mean ± SEM), and 0.299 ± 0.075 and 0.351 ± 0.079 complex exchanges/100 cells were detected in the NT and control veteran groups, respectively. A total of 8 Robertsonian translocations were found (5 and 3 from NT and control veteran groups, respectively), including a constitutional Robertsonian rob(13;14) in the control group (supplementary table 1).

Figure 2. Structural chromosome aberrations observed in control and test veteran cohorts. Frequency of chromosome aberration types per 100 cells reported for control and NT cohorts and, for NT subgroups based upon geographic location (Christmas island, Maralinga range or on board a ship at time of test) and, potential for radiation exposure ranking (allocated blind to cytogenetic data as low [1], medium [2] or high [3] potential) [10]. (A). Total reciprocal translocations (complete and incomplete types), (B). simple exchanges (total reciprocal translocations, Robertsonian translocations and total dicentric chromosomes), (C). complex chromosome exchanges (complete and incomplete types), (D). total fragments (total associated with complex, dicentric, ring or break-only aberrations), E. total damage burden (total number of chromosome breaks irrespective of aberration type). The box-whisker plots show the values for each veteran (dots) together with median values (bar) and the 25%–75% interquartile range (whiskers).

Download figure:

Standard image High-resolution image

Recruitment of the NT veterans was group-matched with control veterans based on a number of criteria including age (79.9 and 80.3 years in the NT and control groups, respectively) (supplementary table 2). As shown in figure 3, we find the majority of veterans have similar translocation frequencies to that which is expected based upon their age with only 7 veterans (4 NT and 3 control) identified as having higher frequencies than expected (supplementary tables 3 and 4) [23]. As detailed in the Methods, of those NT veterans recruited, three had a record of dose however none of these individuals corresponded with the higher translocation frequencies observed.

Figure 3. Difference between observed reciprocal translocation frequency with what is expected according to age at time of sampling, for control and NT veterans. Expected frequency as reported by Sigurdsson et al [23] is detailed in supplementary tables 3 and 4. Total translocations include complete, incomplete and Robertsonian types.

Download figure:

Standard image High-resolution image

Further, a comparison with related studies examining reciprocal translocation frequencies in NT test veterans (figure 4) shows our 24-colour (full genome) findings to be consistent with full-genome converted frequencies reported in French, but below that of the New Zealand, NT veterans.

Figure 4. Reciprocal translocation frequencies in nuclear test cohorts. New Zealand control and NT veteran [24], French NT veteran (study lacked control group) [18] and GCFT British control and NT veteran's studies shown. All data presented as full genome (from 24-colour FISH) or full genome conversion (3-colour FISH data from Gregoire [18] converted based on [25]. British control and test veteran translocation equivalent frequency shown. Box-whisker plots shows the median value, the quartiles (bar), max/min values (whiskers) and outliers.

Download figure:

Standard image High-resolution image

The frequencies of dicentrics are, as expected, lower than that of reciprocal translocations with no difference observed between the two veteran groups (0.149 ± 0.053, and 0.143 ± 0.041) dicentrics/100 cells in NT and control, respectively) (supplementary table 1).

For complex chromosome aberrations, indicative of exposure to low doses of high-LET radiation, or high doses of low-LET radiation, frequencies of 0.299 ± 0.075 (range 0–2.78)/100 cells and 0.351 ± 0.079 (range 0–1.69) were seen in both NT and control veteran cohorts (supplementary table 1).

2.2. Higher frequencies of complex chromosome aberrations in a small group of NT veterans

We examined sub-groups of the NT cohort to ask if the broader range of complex exchanges observed in the NT veteran cohort had any association with the exposure rank veterans were assigned or the test site location they attended. Figure 2 and supplementary table 1 shows the average complex aberration frequency of those present at Christmas Island to be similar or no higher (0.256 ± 0.094 and 0.198/100 cells) for veterans assigned into exposure ranks 1&2 combined or rank 3, respectively, than that for all exposure ranks combined (0.299 ± 0.075/100 cells). By contrast, although the numbers of veterans in each sub-group are small (meaning the statistical detection limit for identifying potential group differences were too high) there is a higher average frequency of complex aberrations in NT veterans who were on-board ships (0.350/100 cells for NT veterans on board ships; 3 out of the 4 personnel on HMS Diana) or, present at Maralinga (0.803 ± 0.191/100 cells; 4 of the 5 veterans had at least 1 complex/100 cells) and who were assigned into exposure rank 3, compared to those in respective locations but assigned into lower exposure ranks 1 or 2 (0.117 and 0.126 complex exchanges/100 cells) (figure 2, supplementary table 1).

No differences for any other chromosome aberration type were evident within the NT exposure rank sub-groups. The increase in overall damage burden seen in NT veterans assigned to exposure rank 3 who were onboard ships (5.957 ± 0.442/100 cells) or at Maralinga (6.881 ± 1.527/100 cells), is therefore likely dominated by the excess of complex aberrations noted above. The total damage burden accounts for all breaks necessary to result in the aberration pattern types categorised in supplementary table 1.

2.3. Newly arising unstable chromosome aberrations

Table 1 shows the frequency of stable and unstable exchange types, further categorised according to the completeness of each exchange. The vast majority of all incomplete stable exchanges were defined as incomplete due to unresolved 'ends' being below the limit of detection [26]. Similar to that seen for all aberration types, we find no difference in the frequency of stable or unstable exchanges, or cells, between the two veteran cohorts. When examining by the exposure rank and test site location sub-groups however, some differences are seen with an increase in the proportion of unstable exchanges for those in exposure rank 3 on board a ship (0.467/100 cells) or at Maralinga (0.459 ± 0.183/100 cells respectively), compared to other sub-groups, all exposure ranks combined or the control veterans (table 1). This is also reflected in a higher dicentric equivalent which is the total of all dicentric chromosomes identified in simple and complex exchanges (0.467 and 0.688 ± 0.305/100 cells, for exposure rank 3 onboard a ship and Maralinga respectively) when compared to all other subgroups, all NT veteran ranks combined and controls (table 1).

Table 1. Stability and completeness of simple and complex exchanges.

  Stable aberrations frequency per 100 cells (number)Unstable aberrations frequency per 100 cells (number)  Reciprocal translocationStable complexTotalDicentricUnstable complexTotalCohorts (N = 86)Cells scored C I C I Stable aberrationsStable cells C I C I Unstable aberrationsDicentric equivalentsUnstable cellsControl (N = 38)76981.000 ± 0.170 (77)0.390 ± 0.086 (30)0.156 ± 0.049 (12)0.065 ± 0.035 (5)1. 611 ± 0.267 (124)1.403 ± 0.250 (108)0.065 ± 0.030 (5)0.078 ± 0.028 (6)0.039 (3)0.091 ± 0.036 (7)0.273 ± 0.063 (21)0.260 ± 0.072 (20)1.234 ± 0.179 (95)BNTV (N = 48)93791.152 ± 0.140 (108)0.309 ± 0.066 (29)0.128 ± 0.039 (12)0.043 ± 0.032 (4)1.631 ± 0.185 (153)1.461 ± 0.172 (137)0.075 ± 0.044 (7)0.075 ± 0.0.25 (7)0.032 (3)0.096 ± 0.044 (9)0.277 ± 0.078 (26)0.288 ± 0.083 (27)1.034 ± 0.171 (97)NTV by exposure rankChristmas Island (N = 28)Rank 1 (N = 21)* & Rank 2 (N = 1)46851.174 ± 0.250 (55)0.299 ± 0.096 (14)0.149 ± 0.066 (7)0.064 ± 0.033 (3)1.686 ± 0.325 (79)1.588 ± 0.319 (73)0.064 (3)0.064 (3)0.021 (1)0.043 (2)0.192 ± 0.086 (9)0.171 ± 0.085 (8)0.832 ± 0.235 (39)Rank 3 (N = 6)*10100.693 ± 0.278 (7)0.495 ± 0.224 (5)001.188 ± 0.398 (12)0.990 ± 0.405 (10)00.099 (1)0.099 (1)0.099 (1)0.297 (3)0.297 (3)1.089 ± 0.369 (11)Onboard ship (N = 8)Rank 1 (N = 4)*8580.699 ± 0.369 (6)0.117 (1)0.117 (1)(0)0.932 ± 0.574 (8)0.816 ± 0.471 (7)0.117 (1)0.117 (1)(0)(0)0.233 (2)0.233 (2)1.166 ± 1.164 (10)Rank 3 (N = 4)*8561.051 ± 0.407 (9)0.234 (2)0.117 (1)(0)1.402 ± 0.308 (12)1.051 ± 0.165 (9)0.117 (1)0.117 (1)(0)0.234 (2)0.467 (4)0.467 (4)2.103 ± 0.683 (18)Maralinga (N = 18)Rank 1 (N = 9)* & Rank 2 (N = 4)23761.094 ± 0.143 (26)0.337 ± 0.147 (8)0.042 (1)0.042 (1)1.515 ± 0.210 (36)1.389 ± 0.151 (33)0.126 (3)0.0842 (2)(0)0.042 (1)0.253 ± 0.232 (6)0.253 ± 0.232 (6)1.010 ± 0.391 (24)Rank 3 (N = 5)8721.38 ± 0.600 (12)0.115 ± 0.089 (1)0.344 (3)(0)1.835 ± 0.848 (16)1.606 ± 0.683 (14)(0)(0)0.115 (1)0.344 (3)0.459 ± 0.183 (4)0.688 ± 0.305 (6)0.803 ± 0.399 (7)

C = Complete; I = incomplete; *includes veterans who attended more than one location. Frequencies are expressed as relative number of aberrations per total cells analysed, corresponding uncertainty is calculated as SEM (for N > 4 and with veteran as statistical unit) and absolute count number is provided in brackets.

2.4. Confounding exposures and chromosome aberrations of varying complexity

A similar exercise for examining sub-groups in control veterans to that carried out for the NT veteran groups is not possible, accordingly, more detailed comparisons of pertinent cytogenetic data and, associations between self-reported confounders (medical and occupational exposures etc) and chromosome aberrations observed, were examined for both veteran cohorts.

Firstly, a number of control veterans showed higher levels of complex chromosome aberrations than might be anticipated in non-radiation exposed individuals. To examine for any qualitative differences which may suggest differences in how these complex aberrations were formed, we looked at the complexity of each complex aberration finding the cohorts to be similar (average of 2.8 chromosomes & 2.7 breaks per stable complex and 3.4 chromosomes & 3.5 breaks per stable complex and, 4.0 chromosomes & 4.2 breaks per unstable complex and 4.8 chromosomes & 5.6 breaks per unstable complex, for control and NT veterans respectively). The number of insertions within each complex pattern was also similar averaging at 0.6/0.7 insertions/stable complex and, 0.5/1.0 insertions/unstable complex for control and NT veteran cohorts. To note, based upon the frequency of insertions detected (within either stable or unstable complex aberration types), the only difference seen was for rank 3 NT veterans who were present at Maralinga (frequency of insertions 0.195 ± 0.063, 0.245 ± 0.075 and 0.573 ± 0.247 for control, NT and rank 3 Maralinga veterans, respectively, supplementary table 1).

Secondly, analysis to examine for associations between all known confounders and the chromosome aberrations detected was performed for all veterans, for the control veterans only and, the NT veterans only. Considering only those associations that were identified as highly significant by both statistical approaches (p < 0.01) for a trend or difference (supplementary table 7), we detected one out of all 154 associations when considering all 86 veterans: specifically, self-reported exposures to 'other' medical sources of radiation (DEXA, radionuclide, fluoroscopy, or coronary angiogram) was related to higher chromosome aberration frequencies for total damage burden (supplementary tables 5(a), (b) and 7). When considering only the NT veteran cohort (N = 48), none of the potential confounders showed any statistical evidence for a difference or trend (supplementary tables 5(e) and (f)). However, a different picture was seen for the control veterans (supplementary tables 5(c) and (d)) where, despite the relatively small sample size (N = 38), we identified a total of 5 associations as being statistically highly significant (associations between alcohol consumption, or exposure to 'other' medical sources of radiation with higher chromosome aberration frequencies (reciprocal translocations and chromosome breaks)). More statistical details can be found in supplementary table 7.

2.5. Examination for association between potential for exposure in veteran father and germline mutation frequency

The frequency of de novo germline DNA mutations was determined for a sub-set of the NT and control veteran family trios (veteran father, mother, biological child) and reported elsewhere [22]. Using this information, we look here to identify any relationship between chromosome aberration frequency in the veteran father and, the germline mutation frequency for the adult child sampled, if such a relationship exists. As shown in figures 5(A) and (B), no evidence for any association between the overall chromosome damage burden or complex chromosome aberrations and DNA germline mutations in their adult child was seen (Spearman correlation coefficient, p > 0.1).

Figure 5. Association between veteran father's chromosome aberration frequency and germline de novo mutations. Analysis was carried out for all families where both M-FISH (as reported here) and WGS data [22], was available (28 control and 30 NT families). (A). Total chromosome aberration frequency in veteran verses total de novo germline mutations, (B). complex chromosome exchange frequency in veteran verses total de novo germline mutations, (C). total aberration frequency in veteran verses de novo SNV germline mutations allocated to SBS16, (D). complex chromosome exchange frequency in veteran verses de novo SNV germline mutations allocated to SBS16. Total chromosome damage frequencies were categorised according to their quartiles, and complex exchanges according to their median (0.68) and a separate zero group, for all veterans (control and NT). The box-whisker plots show the values for each offspring (dots) together with median values (bar) and the 25%–75% interquartile range (whiskers).

Download figure:

Standard image High-resolution image

We repeated this analysis with a focus on the mutation pattern termed as tumour mutation signature single base substitution (SBS)16. As shown in figures 5(C) and (D), no evidence for any statistically significant trend was observed between the germline single nucleotide variant (SNV) mutations allocated to SBS16 and veteran father's chromosome aberration frequency (Spearman correlation coefficients, p > 0.1). However, when those families who were previously identified as having an over-representation of germline SNVs allocated to SBS16, defined as the cluster of >40 SNVs compared to <40, were related to complex aberration groups (none, below and above median frequencies), a potential difference was seen (Kruskal–Wallis test, p = 0.026) (figure 5(D)). Specifically, of the 8 families with >40 germline SNVs allocated to SBS16, 2/30, 1/14 and 5/14 veterans have none, below and above medium frequencies of complex chromosome aberrations, respectively.

To explore in more detail, families were further grouped as [1] those who had the highest proportion of SNVs allocated to SBS16 [2], those families who self-reported a health effect and [3], all NT veterans assigned to exposure rank 3 [10]. Table 2 summarises the averages for translocation equivalent in stable cells, complex aberration frequency and, the overall damage burden detected by M-FISH in the veteran father with the DNA germline mutation averages (total mutations, SNV, indels, SV, clustered and mutations allocated to SBS16) for these family groupings. Chromosome aberration frequencies for this smaller sub-set of control (N = 28) veterans is also shown and is consistent with those derived from the analysis of 91 veterans (supplementary table 1). As expected, those families categorised as having the highest SBS16, average around 2–3 times higher for this signature (46.3) than all other family categories examined here (∼15–16). Further, both the total of all de novo mutations (88.5) and all SNVs (79) are raised compared to any other sub-group, whilst the frequencies of complex aberration (0.686 ± 0.315/100 cells) and overall damage burden (7.546 ± 1.82/100cells) in the veteran father are also raised, again relative to any other sub-group examined. Families of veterans who were assigned into exposure rank 3 were associated with a slightly higher proportion of SNVs allocated to SBS16 (24.3 compared to average ∼15–16 for e.g. exposure ranks 0 (control), 1 and 2). For those families who self-reported a health effect, no elevation in aberration burden in the veteran father or germline mutations relative to those families who did not, was seen. No other differences were evident although it is noted that significantly more veterans who were allocated into exposure rank 3, reported health concerns compared to all other exposure ranks (0.014 (5/48 families), 0.31 (11/35 families) and 0.429 (6/14 families) reporting at least one child/grandchild with health issues for exposure rank 0 (control), 1 + 2 combined and rank 3, respectively, p < 0.1) [10].

Table 2. Summary of chromosome aberration burden in veterans observed by M-FISH and de novo germline mutations detected by whole-genome sequencing (WGS) analysis.

 Veteran M-FISH data Frequency/100 cells (number)Germline mutation frequency/offspring Cells (stable)ComplexTotal damage burdenTranslocation equivalent in stable cellsTotalSNVInDelSVCluster (10bp /100bp)SBS16SBS161Families with >40 SNV mutations allocated to SBS16 (N = 8)1312 (1288)0.6862 ± 0.315 (9)7.546 ± 1.82 (99)2.562 ± 0.717 (33)88.5798.11.40.9/1.446.3Families with <40 SNV mutations allocated to SBS16 (N = 50)9885 (9772)0.303 ± 0.061 (30)5.53 ± 0.573 (547)1.770 ± 0.265 (175)6759.766.260.980.8/1.212.5Families who self-reported health effect in offspring3Families reporting effect (N = 16)3355 (3318)0.268 ± 0.107 (9)4.978 ± 0.731 (167)1.567 ± 0.299 (52)71.562.447.751.30.8/1.414.5None (N = 42)7844 (7742)0.382 ± 0.089 (30)6.107 ± 0.711 (479)2.015 ± 0.320 (156)69.462.46.050.930.8/1.218.2Veterans allocated to Rank 33Rank 3 (N = 11)2037 (2005)0.540 ± 0.156 (11)6.284 ± 0.900 (128)1.546 ± 0.535 (31)70.162.26.910.6/1.124.3Ranks 1 + 2 (N = 19)3525 (3493)0.312 ± 0.163 (11)5.702 ± 0.952 (201)2.262 ± 0.430 (79)69.562.36.210.950.63/0.8916.05Rank 0 (controls) (N = 28)5637 (5562)0.302 ± 0.085 (17)5.624 ± 0.913 (317)1.762 ± 0.370 (98)70.362.66.571.110.93/1.515.2Ranks 0,1 + 2 (M = 47)9160 (9055)0.306 ± 0.083 (28)5.655 ± 0.200 (518)1.932 ± 0.279 (177)69.962.56.431.040.81/1.315.5

Includes all families (NT and control combined) where both veterans father M-FISH data and adult child's whole genome sequence data [22], were available.1 Moorhouse et al [22], 2statistical significance for difference: p = 0.054 (Wilcoxon rank-sum/KruskalWallis test) and p = 0.032 (Negative binomial regression), 3 Rake et al [10].

The GCFT study is the first study to obtain blood samples from a group of British NT veterans and their families for the purposes of identifying genetic alterations in offspring which may have arisen as a consequence of historical paternal exposure to ionising radiation [10]. The available information on radiation dose received by veterans, if any, is limited due to approximately only ∼22% being monitored at the time, therefore, the purpose of the work carried out in this part of the study is to ascertain if there is any cytogenetic evidence of historical exposure to ionising radiation in veterans of the nuclear testing programme. For this, 24-colour karyotyping M-FISH was used to detect the occurrence of stable and unstable chromosome exchanges of varying complexity [27]. By doing this, aberrations which have persisted over time and those which may be more recently induced, arising as a consequence of lifestyle/medical/occupational factors, ongoing internalised radiation exposure or through other mechanisms including delayed genomic instability, may be compared between the NT and control veteran cohorts. As shown in figure 2, table 1 and supplementary table 1, we observed stable and unstable chromosome type aberrations of varying complexity to occur in both cohorts, however for all aberration types, no difference in frequencies between the NT and control veteran's cohorts, was seen.

Those exposed to radiation, even in the distant past, may be expected to have more aberrations (particularly stable types) than someone not exposed to radiation. For instance, the occurrence of reciprocal translocations after radiation exposure is routinely applied for the retrospective assessment of radiation dose [12]. However, other factors including occupational, medical and lifestyle exposures also contribute to the induction of translocations and as a consequence, a person's translocation burden increases with increasing age [23]. As it is not possible to distinguish translocations induced by radiation from those arising by other causes, detectable frequencies potentially attributable to radiation, in a controlled study such as this, need to be higher than would be expected for their respective