Mutation Research/Reviews in Mutation Research

Transcription

1 Mutation Research 659 (2008) Contents lists available at ScienceDirect Mutation Research/Reviews in Mutation Research journal homepage: Community address: Review Low-dose ionizing radiation and chromosome translocations: A review of the major considerations for human biological dosimetry James D. Tucker * Department of Biological Sciences, 5047 Gullen Mall, Wayne State University, Detroit, MI 48202, USA ARTICLE INFO ABSTRACT Article history: Received 21 November 2007 Received in revised form 21 March 2008 Accepted 28 March 2008 Available online 10 April 2008 Keywords: Chromosome translocations Low-dose ionizing radiation Human Biological dosimetry Confounding effects Review Chromosome translocations are a molecular signature of ionizing radiation exposure. Translocations persist significantly longer after exposure than other types of chromosome exchanges such as dicentrics. This persistence makes translocations the preferred aberration type for performing radiation dosimetry under conditions of protracted exposure or when exposure assessments are temporally delayed. Low doses of radiation are inherently difficult to quantify because the frequency of induced events is low and the background level of translocations among unexposed subjects can show considerable variability. Analyses of translocation frequencies can be confounded by several factors, including age of the subject, lifestyle choices such as cigarette smoking, the presence of clones of abnormal cells, and possibly genotypic variability among subjects. No significant effects of gender or race have been observed, but racial differences have not been completely ruled out. Translocation analyses may be complicated by the presence of different types of exchanges, i.e., reciprocal or non-reciprocal, and because translocations sometimes occur as a component of complex exchanges that include other forms of chromosome rearrangements. Rates of radiation exposure, ranging from acute to chronic, are known to influence the accumulation of translocations and may also affect their persistence. The influences on translocation frequencies of low-dose radiation hypersensitivity as well as the bystander effect and the adaptive response remain poorly characterized. Thus, quantifying the relationship between radiation dose and the frequency of translocations in any given subject requires attention to multiple issues. Part of the solution to understanding the in vivo dose response relationship is to have accurate estimates of the baseline levels of translocations in healthy unexposed subjects, and some work in this area has been accomplished. Long-term cytogenetic follow-up of exposed subjects is needed to characterize translocation persistence, which is especially relevant for risk analyses. More work also needs to be done in the area of quantifying the role of known confounders. Characterizing the role of genotype will be especially important. Improvements in the ability to use translocation frequencies for low-dose biological dosimetry will require scoring very large numbers of cells per subject, which may be accomplished by developing a rapid automated image analysis system. This work would enhance our comprehension of the effects of low-dose radiation exposure and could lead to significant improvements in understanding the relationship between chromosome damage and human health. ß 2008 Elsevier B.V. All rights reserved. Contents 1. Introduction Types of translocations Low-dose radiation hypersensitivity Induction, accumulation and persistence of translocations The importance of cell type upon translocation frequencies Dose rates and translocation persistence Translocation frequencies in subjects not exposed to ionizing radiation Influences of confounding factors upon translocation frequencies * Tel.: ; fax: address: jtucker@biology.biosci.wayne.edu /$ see front matter ß 2008 Elsevier B.V. All rights reserved. doi: /j.mrrev

2 212 J.D. Tucker / Mutation Research 659 (2008) Improving the assessment of low doses of radiation Improving biological dosimetry in exposed people Population study of unexposed people Recommendations for the future Summary Acknowledgements References Introduction Chromosome aberrations have been widely accepted for many years as a biological marker of exposure for ionizing radiation. Recent evidence indicates that aberrations are also a biomarker of effect because groups of individuals with elevated aberration frequencies have increased risks of cancer. Ionizing radiation is a potent inducer of chromosome rearrangements, and high doses of radiation are carcinogenic. Less certainty exists concerning whether low doses of ionizing radiation are also carcinogenic. The past two decades have seen significant improvements in the ability to identify and quantify chromosome damage, of which the most notable development has been fluorescence in situ hybridization (FISH) with whole chromosome paints. FISH painting can identify translocations (symmetrical exchanges in which each resultant chromosome has one centromere) as readily as dicentrics (asymmetrical exchanges resulting in a chromosome with two centromeres and an acentric fragment). The ability to identify translocations with high accuracy and efficiency is significant because translocations have substantially greater persistence through cell division than dicentrics [1,2]. This stability of translocations makes them the preferred marker for radiation exposures that are chronic or that occurred many years previously. Dicentrics remain the aberration of choice when exposures are at least moderately acute and recent, i.e., within the last few weeks or months. Painting for translocation identification has a significant order-of-magnitude advantage compared to conventional cytogenetic methods with respect to analysis speed. The use of translocations for biodosimetry has increased over the past decade with the commercial availability of reliable whole chromosome painting probes and with investigators familiarity with in situ hybridization methods. A number of observations support the idea that translocations may be the most relevant cytogenetic endpoint for assessing cancer risks. These include the wide-spread acceptance of translocations as a biomarker of exposure to ionizing radiation [3 9], the evidence that chromosome aberration frequencies may be elevated as a result of exposure to chemicals such as cigarette smoke, e.g. [10 13], the recent findings that aberrations are associated with increased risks of cancer [14 16] and the observation that essentially all types of cancer cells bear translocations [17 19]. Low doses of ionizing radiation (<1 Gy) comprise the vast majority of exposures, which occur most often in occupational settings or as a result of contamination from environmental accidents such as Chernobyl, e.g. [20 26]. Such exposures are most often chronic or highly fractionated. Medical exposures such as diagnostic X-rays also occur and generally consist of very low doses (<1 cgy); with current technologies these have been regarded as nearly inconsequential in terms of the overall amount of radiation received and their effects on cytogenetic measurements. However, recent evidence suggests that the lifetime accumulation of personal diagnostic X-rays may result in detectable levels of translocations [27]. High exposure-doses (>1 Gy) are rare and will not be emphasized in this paper. 2. Types of translocations For many years, chromosome exchanges including translocations were thought to involve exactly two chromosomes. More recent evidence indicates that complex aberrations, which involve two or more breaks in three or more chromosomes [28], are surprisingly common. Complex aberrations arise from simultaneous double strand breaks in multiple chromosomes followed by a multi-way exchange [29 31]. Both the prevalence and the complexity of multi-way exchanges increase with dose, especially above 2 Gy [32], but complex aberrations have also been seen in cells exposed to low doses (1 Gy) and even in unexposed control cells. Complex exchanges have the pragmatic consequence of complicating biological dosimetry. What used to be a rather simple process, namely counting exchanges (dicentrics or translocations) has become much more complicated as researchers wrestle with the importance of three-way and higher level exchanges relative to two-way (reciprocal) exchanges. Fortunately for low-dose biodosimetry, the fraction of all complex exchanges is comparatively low, but generally not low enough to be ignored completely. Knowing that chromosome rearrangements can be complex, it should come as no surprise that different types of translocations occur [33]. Briefly, translocations may be as simple as a reciprocal exchange of material between two chromosomes. Translocations may also appear as non-reciprocal, i.e., chromosomal material has been transferred from one chromosome to another, but there is no evidence for the involvement of a second chromosome in the process. There are at least two possible explanations for the apparent existence of non-reciprocal translocations. First, the socalled exchange may actually be non-reciprocal, i.e., there is transfer of material from only one chromosome to only one other chromosome. The second possibility is that apparent nonreciprocal exchanges are in fact reciprocal at the molecular level, but that one of the participating chromosomal segments is too small to be seen cytologically. By labeling the telomeres of irradiated chromosomes, Boei and Natarajan [34] showed that many of the apparently non-reciprocal translocations were in fact reciprocal, and that the frequency of incomplete exchanges was low. This work suggests that almost all non-reciprocal events are in fact reciprocal at the molecular level. Some translocations may appear to be non-reciprocal only because of the chromosomes that were painted. Depending on which chromosomes are labeled, identical exchanges may appear to be different, and different exchanges may appear to be the same [35]. Unless all chromosomes are painted in their own unique color [36,37], or molecular banding methods are used [31], this problem is unavoidable. For reasons of cost as well as ease and speed of analysis, FISH biodosimetry studies are performed by painting only a few chromosome pairs, which means that some ambiguity with respect to aberration types is tolerated in exchange for convenience and affordability. Given the existence of different types of translocations, does it matter how translocations are categorized, analyzed and reported? Existing data suggest that there are several approaches which yield very similar results [33]. One approach, which we have called

3 J.D. Tucker / Mutation Research 659 (2008) classical, enumerates reciprocal translocations and non-reciprocal translocations each as one translocation. This first approach recognizes the work by Boei and Natarajan [34], mentioned above, that indicates that the vast majority of apparently non-reciprocal events are actually reciprocal. Edwards et al. [38] have taken an approach to scoring translocations that is consistent with this view. A second approach is to report cells with translocations. Here, each cell that contains one or more translocations of any type is considered to be a cell with a translocation. This approach ignores the presence of multiple translocations within a cell and thus underestimates the true translocation frequency, but it avoids problems associated with sub-classification of translocation types. A third approach reports the number of translocated chromosomes and ignores the possibility that some aberrations may share a common mechanistic origin. This approach reports a nonreciprocal translocation as one event, a reciprocal translocation as two events, a three-way exchange as three events, etc. This approach was first suggested when the PAINT system of scoring aberrations was published [35]. While this method has the benefits of being both simple and quantitative, it does not afford easy comparison with currently accepted methods of biodosimetry such as counting dicentrics [39] and for this reason can be problematic. All three of these methods, i.e., classical, cells with translocations, and PAINT, have been shown to yield similar results when quantifying the persistence of translocations in peripheral blood lymphocytes over time for the purposes of biological dosimetry [2,33]. This suggests that, for biodosimetry, the actual types of translocations may be less important than their existence. Which type(s) of translocation should be counted, and which should be reported? All events should be tabulated in the initial collection of the data. No aberrant cell should be ignored, or worse, classified as normal, as long as the cell otherwise meets the criteria for inclusion in the study. This means that every cell with any rearrangement is recorded, without prejudice for the type(s) or the amount of damage present. Once the data collection is complete, decisions can be made concerning how the observations should be reported. The manner in which this is accomplished may depend on the nature of the exposure and on the dose. For example, following acute doses 2 Gy, some aberrant chromosomes will share a common mechanistic origin, i.e., will not be independently derived events. Under such circumstances, it may be necessary to report translocated chromosomes since multi-way events rather than reciprocal exchanges may predominate. For chronic exposures to photons, whether the total dose is high or low, the expectation is that most aberrations will be simple and that relatively few will be complex. In these cases, reporting either classical translocations or cells with translocations should yield answers that are very similar if not essentially identical, and either approach is acceptable. For exposures involving high-let radiation, even doses below 1 Gy may lead to complex aberrations, possibly necessitating the reporting of translocated chromosomes. How will an investigator know which reporting method to use? An empirical approach seems to be the most reasonable. If few complex aberrations are seen, then using either the classical method or reporting the number of cells with translocations should be adequate. If many complex aberrations are present, then the method used to report translocations should consider the complexity of the events seen, otherwise biases in estimating the dose may occur. 3. Low-dose radiation hypersensitivity Low doses of ionizing radiation are inherently more difficult to detect and quantify than high doses. Of interest to some investigators has been the question of whether there is a threshold below which ionizing radiation does not induce chromosome aberrations. Although the ultimate answer to this question can only be determined by evaluating the biological effects of the absorbance of a single photon or particle by a cell, the preponderance of evidence obtained from dose response curves provides little support for the existence of a threshold for aberration induction. However, there is now solid evidence that low doses (20 cgy) are capable of exhibiting hypersensitivity to ionizing radiation as measured by clonogenic cell survival [40,41]. The basic premise here is that cells die as a result of excessive sensitivity to small doses of radiation but become more resistant (per unit dose) as dose increases. Investigations into the molecular basis underlying this phenomenon indicate that low-dose radiation hypersensitivity does not depend on the recognition of DNA double strand breaks, even though residual breaks appear to persist based on presence of g-h2ax foci following exposure [42]. ATM activation does not appear to play a primary role in low-dose radiation hypersensitivity, although a downstream ATM-dependent G 2 checkpoint may be involved [43]. The existence of lowdose hyper-radiosensitivity in clonogenic survival experiments has been found to be associated with increased levels of apoptosis, and cell-cycle enrichment experiments suggest a relationship between low-dose hypersensitivity and the radioresponse of G 2 - phase cells. These observations suggest that low-dose hypersensitivity stems from the failure by ATM-dependent DNA repair processes to prevent progression of G 2 -phase cells with DNA damage from entering mitosis [44]. In practical terms, this means that at low doses the amount of damage induced is below that which is needed to stimulate the repair processes, so these low levels of damage are not repaired and as a result cells die. As dose increases, the amount of damage induced becomes sufficient to stimulate the repair processes, and this repair leads to diminished cell killing. Deciphering the complete mechanism for low-dose radiation hypersensitivity will have important implications for human health, and is particularly relevant to our knowledge of the low-dose effects of ionizing radiation upon translocation frequencies and the long-term risks stemming from such exposures. 4. Induction, accumulation and persistence of translocations The concepts of induction, accumulation and persistence are important for understanding the effects of exposure to ionizing radiation. Induction refers to the initial formation of translocations immediately following exposure. Each dose or dose fraction, no matter how small or large, has the potential for inducing double strand breaks that lead to translocations. In the case of multiple exposures, aberrations may be induced as a result of each exposure. Accumulation is the acquisition of translocations over time, and occurs if exposure is fractionated or prolonged for any reason. In the event of a single acute exposure, translocations will not accumulate but will nevertheless be present as a result of that exposure. Thus, the concept of accumulation assumes multiple exposures and, if sufficiently uniform, may have measurable kinetics. Persistence refers to the amount of time that translocations exist following their formation. In performing exposure assessments in humans, persistence typically refers to the continued presence of translocations in peripheral blood lymphocytes, although it must be emphasized that any persistence in these cells depends on the presence of translocations in the stem cells that give rise to them. All chromosome aberrations show at least some persistence in peripheral blood lymphocytes, even unstable events such as dicentrics. What distinguishes translocations from other aberration types is that their persistence is substantially greater. However, no type of aberration, not even translocations, shows complete persistence. Early studies

4 214 J.D. Tucker / Mutation Research 659 (2008) evaluating the persistence of dicentrics in cultured hamster V79 cells showed that their frequency declines by half each cell division [45,46] as a result of anaphase bridge formation and/or loss of the associated acentric fragment, both of which lead to cell death. Given that translocations have one centromere per chromosome and thus would not contribute to cell killing in the same manner as dicentrics and acentric fragments, it was originally expected by many investigators that translocations would not be cell-lethal. As a corollary, translocation frequencies were expected to show complete persistence, i.e., their frequencies would not decline over time following exposure. However, when tested experimentally, translocation frequencies have been shown repeatedly to decline over time [1 8,33,47]. However, when translocation frequencies have been evaluated only in stable cells (i.e., without unstable aberrations in any chromosome including those that are counterstained), the yield of translocations has been observed to be constant with time [48]. This suggests that the apparent partial instability of translocations is due not to the translocations themselves but to collateral damage inside the same cells. An alternative hypothesis is that at least some translocations are inherently unstable. These two explanations, which are not mutually exclusive, have been investigated by building models using actual chromosome aberration data. Both postulated mechanisms were found to contribute to the loss of both nonreciprocal and reciprocal translocations [49]. Some of the apparent discrepancies in translocation persistence may be explained by the way translocations are categorized and reported. For example, translocations that are simple, i.e., appear to be reciprocal and involve exactly two chromosomes, have greater persistence than translocations that are non-reciprocal [2,5,33,50 52]. Furthermore, translocations that are formed as part of a complex exchange (three or more breaks in two or more chromosomes [28]) show comparatively less persistence [53]. Unfortunately, it is not possible to obtain a population of radiationexposed cells in which translocations but no other aberration types have been induced. Thus, translocations are typically evaluated amidst a spectrum of aberration types. This breadth of the aberration spectrum increases with dose, especially if exposure is acute. At low doses (<1 Gy), whether acute or chronic, most aberrations will be simple, both in appearance and in actuality. The formation of simple aberrations corresponds to the linear portion of the linear-quadratic dose response curve, i.e., the ad term in the equation Y = k + ad + bd 2, where Y is the number of translocations per cell, D the absorbed dose in Gy, and k, a, and b are fitted coefficients. At doses above 2 Gy, where the quadratic (i.e., the bd 2 term) predominates, the fraction of all aberrations that involve more than two chromosomes increases dramatically, and some cells may contain multiple complex independent exchanges [29,30]. For a given dose, the more chronic (or fractionated) the exposure, the more simple the aberrations will be. The reason is that double strand breaks induced at the beginning of a chronic exposure period (or by one fraction of the total exposure) will be repaired or mis-repaired before the next exposure and will not be available to take part in the formation of chromosome exchanges with double strand breaks initiated at a later time. Thus, high doses that are administered chronically will have an aberration spectrum that is more reminiscent of low doses than would be true if the same high dose was administered acutely. 5. The importance of cell type upon translocation frequencies The persistence of translocations depends in part on the type of cells evaluated. Most human exposure evaluations are performed using peripheral blood T lymphocytes, which have a half-life in vivo of several months to a year [54,55]. Evaluating chromosomes in these cells is achieved by culturing whole blood (or occasionally isolated lymphocytes) in the presence of a mitogen, commonly phytohemagglutinin, which stimulates lymphocytes to divide in vitro. Without such stimulation the lymphocytes remain in the quiescent, non-dividing phase of the cell cycle known as G 0, where they normally reside in vivo. This is relevant to translocation frequencies because damage in cells that do not divide in vivo will not undergo the selective pressure of mitosis and so will persist longer than damage in cells that do divide regularly, e.g., stem cells. Evidence for the persistence of damage in lymphocytes comes from the presence of unstable chromosome rearrangements (e.g., dicentrics) that have been observed in peripheral blood lymphocytes months to years after radiation exposure, e.g. [56,57]. Such persistent unstable damage could be due to long-lived peripheral blood lymphocytes, or could have been present in stem cells that survived multiple cell divisions in spite of having dicentrics, or some combination of both. Stem cells are expected to last the lifetime of the individual provided they are not lethally damaged. For this reason, translocations induced in stem cells will show two fundamental characteristics: they will persist indefinitely because the progeny cells will inherit the rearrangement, and they will comprise a clone that may be recognizable. This importance of clones is discussed in detail below. Occasionally, tissues other than peripheral blood are used for exposure assessments. Translocations have been used as an endpoint in fibroblasts [58]. This is practical only for assessing exposures in a specific area of the body, which circulating cells obviously cannot do. Terminally differentiated fibroblasts are expected to divide much less often (if at all) compared to the stem cells that give rise to them. Evaluation of chromosome aberrations in bone marrow cells has also been performed, e.g. [59]. Due to the difficulties of obtaining bone marrow cells, this approach is used infrequently. Furthermore, the rapidly dividing nature of bone marrow cells is expected to diminish the half-life of chromosome damage, an observation that is consistent with translocation data obtained following acute whole-body exposures in mice [60]. For these reasons, bone marrow cells are not as useful as peripheral blood lymphocytes for evaluating exposure. 6. Dose rates and translocation persistence What do dose rates have to do with translocation persistence? As already noted, chronic (or fractionated or intermittent) exposures generally result in simpler aberrations than acute exposures. Since simpler aberrations are less likely to be cell-lethal, cells with translocations that were induced by chronic exposure may be more likely to survive than cells with translocations induced by acute exposure. Most human exposures involve chronic low doses, so it is reasonable to expect that aberrations induced under these conditions would have the greatest persistence. Consequently, biodosimetry estimates following chronic exposure may correlate closely with the true physical dose. However, since small doses of radiation are known to diminish the effects of subsequent larger doses, a phenomenon that is called the adaptive response [61 64], it is possible that chronic exposure may lead to a state in which some cells have certain cytoprotective genes that are perpetually expressed at elevated levels. The results of such an effect, if true, would mean that the actual dose is underestimated, perhaps by a significant amount. Dose and dose rate reduction effects have been observed when evaluating translocation frequencies in chronically exposed radiation workers relative to atomic bomb survivors [65], but it is not known how much, if any, of this effect is due to adaptive processes. In contrast, the bystander effect involves the induction of damage in cells that were not

5 J.D. Tucker / Mutation Research 659 (2008) directly hit by radiation. Bystander damage appears to occur in response to factors that are produced by cells that are directly hit by radiation. Such effects have been observed repeatedly following radiation exposure and have been the subject of recent reviews [66,67]. To the extent that the bystander effect occurs and induces the formation of translocations, biological dose estimates may be overestimated. For these reasons, reliable dosimetry of chronic exposure can be difficult. Biodosimetry almost always involves comparison of in vivo exposures with dose calibration curves established using acute, not chronic, exposures administered in vitro. Because low doses induce aberrations that are simpler and fewer in number than high doses, it is widely accepted that the low end of an acute dose response curve (or perhaps more specifically the ad term of the Y = k + ad + bd 2 equation) is an adequate model for chronic exposures, even those involving cumulative doses that may be quite high. The problem is that even the low-dose range of most acute dose response curves shows some upward trend in the number of translocations induced as dose increases. Assuming that the dose response curve has a sufficient number of treatment groups in the low-dose range, and that sufficient numbers of cells were scored to obtain reliable point estimates of the aberration frequency at each dose, the a coefficient should be known with an accuracy that is sufficient for estimating doses received chronically. 7. Translocation frequencies in subjects not exposed to ionizing radiation As with other types of genetic damage, cells with chromosome translocations are observed in somatic cells at low frequencies in unexposed humans. If sufficient numbers of cells are examined, virtually every adult has one or more peripheral blood lymphocytes with some type of translocation. Reasoning that the stability of translocations should be consistent with their accumulation with age, our laboratory set out some years ago to determine whether translocations would accumulate in normal, healthy subjects in the absence of exposure to ionizing radiation or to chemicals such as chemotherapeutic agents. In a study of nearly 100 subjects, half of whom were smokers and with equal numbers of men and women, a highly significant increase in translocation frequencies with age was observed [10,68]. This observation has since been made in other studies [6,51,69 72]. Cigarette smoking has been observed to lead to significant increases in translocation frequencies in some [10 12,21,70,73] but not all studies [6,71,74]. One reason may be due to differences in statistical power to detect an effect, assuming one exists. Other reasons may include the preferred types of cigarettes in each country, variations in the number of cigarettes smoked by each subject over time, and inaccurate or incomplete recall. Interestingly, the importance of age on translocation frequencies is more consistently observed and is a much stronger effect than smoking, at least in adults. In our original investigation on the relationship between translocation frequencies and age, we used FISH whole chromosome painting to evaluate newborns using blood obtained from the fetal side of the placenta, or from the umbilical cord. Translocation frequencies in these newborns were significantly lower than in even the youngest adults. Not surprisingly, babies born to women who smoked cigarettes during pregnancy had more translocations than those babies from women who did not smoke during pregnancy [10]. This prompted a much larger study, also with FISH whole chromosome painting, involving about 240 mother baby pairs; the cytogenetic results of this effort are still being evaluated. However, it is clear from this extended study that newborns have significantly fewer translocations than adults, and that many newborns had no translocations when 1000 cell equivalents were scored. Only a limited amount of information is available on translocation frequencies in children between birth and age 18 years [7,72], so the kinetics of the increase in translocation frequencies in this important age group and the relative importance of exposure to cigarette smoke during childhood remain largely unknown. Given the relatively strong effect of age as compared to cigarette smoke exposure, we hypothesized that translocation frequencies would increase with age in other species. To this end, we conducted a retrospective evaluation of translocation frequencies in 102 unexposed mice that had served as controls in studies conducted for other purposes [75]. The animals ranged in age from 8 weeks (young adulthood) to 2 years, which is close to their average life expectancy. Interestingly, translocation frequencies in the peripheral blood lymphocytes of these mice showed an increase with age that closely resembles observations made in peripheral blood lymphocytes of humans. However, the mice showed strain-dependent differences, and the increase with age was observed only in the peripheral blood, not the bone marrow. In humans, bone marrow cells have not been systematically examined for baseline translocations. Taken together, these results obtained from unexposed humans and mice shed light on the process of translocation formation that is relevant to understanding the effects of low-dose radiation exposure in humans. Several things seem clear. First, every adult has some non-zero baseline level of chromosome damage against which the frequency of translocations induced by exposure must be compared. Second, this baseline level of damage arises from as yet unknown sources. That is, not all chromosome damage is induced by known, exogenous exposures. While it is clear that cigarette smoke can induce translocations, most translocations arise from other sources, even in heavy smokers. While it is beyond the scope of this paper to evaluate the source(s) of all translocations, one leading possibility is oxidative damage that arises in the course of cellular metabolism [76,77]. Another non-exclusive possibility is diminished DNA repair capacity with age [78]. Third, susceptibility to radiation-induced DNA damage may increase with age. Although based on only two subjects, in vitro radiation dose response curves from two men aged 20 and 49 years indicated that the older subject had significantly more translocations induced than the younger person [2,33]. These two individuals also exhibited differences in the persistence of radiation-induced translocations in cultures of their whole blood. A systematic study of the effects of aging on the susceptibility to and persistence of radiation-induced chromosome damage is needed to determine whether these observations are applicable to people in general. 8. Influences of confounding factors upon translocation frequencies Since it is clear that translocations arise from a variety of known and unknown sources, it is relevant to consider the variables that could confound the analyses of subjects potentially exposed to low doses (<1 Gy) of ionizing radiation. As indicated above, aging is the variable with the largest effect. In the study by Ramsey et al. [10], aging accounted for 70% of the statistical variation. Other studies have not explained this much of the variance, probably because a narrower range of ages was evaluated. Nevertheless, the effect of aging upon translocation frequencies is real and much too large to be ignored. Cigarette smoking must also be considered. Different studies have considered the effects of smoking in various ways. Ideally, one should determine the number of cigarettes smoked during the

6 216 J.D. Tucker / Mutation Research 659 (2008) lifetime of the individual, since translocations measure the accumulation of exposure. Simply asking potential subjects if they currently smoke, or if they have ever smoked, is not adequate as the effects of previous cigarette smoking may still be seen in their chromosomes. It is possible that the effects of smoking in light smokers, and/or in people who have smoked for only a few years, may not be detectable given the number of cells per subject that are typically evaluated. This is not unexpected as the total number of cigarettes smoked may be relatively small. Some effects of diet upon translocation frequencies have been reported, e.g. [10], but in general these findings have not been replicated, perhaps due to the absence of a large, systematic study designed to evaluate this complex set of variables. However, considerable work has now been performed in humans using micronuclei as the endpoint and the results suggest that some vitamins increase the frequency of micronuclei while others decrease their frequency [79,80]. Genotypes are also known to influence translocation frequencies [11,12,81,82]. It is quite possible that genetic polymorphisms in genes that are known to be involved in metabolism as well as in DNA damage recognition and repair [83] affect chromosome aberration frequencies. However, deciphering the effects of individual polymorphisms can be difficult due to the rarity of specific alleles, the multiplicity of alleles for some genes, and the large number of subjects needed to detect small differences in translocation frequencies. These issues have not kept investigators from performing genotyping assays. However, given the small sample sizes seen in most translocation studies only relatively large effects would be visible. It should not be surprising that analyses of allelic differences among healthy subjects have not revealed any major effect genes. However, cells from patients with clinically recognizable conditions such as Fanconi anemia, ataxia telangiectasia, and Bloom syndrome have increased frequencies of chromosomal aberrations [84]. A major and as yet largely unanswered question is whether people who are heterozygous for any of these conditions have phenotypic effects [85]. If so, a fraction of the population may have recognizably elevated translocation frequencies. Other variables that are potentially important include gender and race. To date, no effect of gender has been observed in spite of the fact that females have slightly more DNA than males due to the larger size of the X compared to the Y chromosome. The difference in the amount of DNA is relatively small and it is quite likely that existing studies have lacked the statistical power to detect a difference if one exists. A large systematic study designed to examine race has not been published, although our as-yetunpublished study on the effects of smoking during pregnancy evaluated roughly equal numbers of African Americans and Caucasian Americans. The results of these analyses are pending. Clones of cells with translocations or insertions have been observed in some healthy control subjects. Given that lymphocytes are the cell type most commonly evaluated in humans, and that these cells are part of the immune system and subsets of these cells may expand from stem cells to combat specific immunological challenges, it should not be a surprise that clones exist and that some will contain aberrations such as translocations and insertions that are capable of surviving cell division. A number of individuals with clones of chromosomally abnormal cells have been identified [86 88]. Clearly these clones have arisen from proliferative expansion of stem cells. Some of these clones have as few as three cells (the minimum number we have considered in order to be counted as a clone) and as many as several dozen cells per whole genome cell equivalents. The fraction of all cells that bear clones is thus small. However, recognizing clones when they exist and properly accounting for them when performing biodosimetry is essential, especially for low-dose radiation exposure where the presence of even a small number of abnormal cells that includes clones may overly influence biological dosimetry. There are two possible approaches for handling translocation frequencies when cells with clonal aberrations are observed. In the first approach, the analysis could consider the clones as a single mutagenic event and report all clonal cells as a single aberrant cell. This approach, which is equivalent to estimating the mutation frequency, recognizes that there was just one original chromosome translocation. With this approach the frequency of translocationbearing cells is obtained by reducing both the numerator and the denominator by one less than the number of clonal cells. With the second approach, the number of cells with clones is not taken into consideration, i.e., this approach measures the mutant frequency, where each abnormal cell is reported as an independent event. In this case, no mathematical adjustment of the translocation frequency is needed. The reasoning here involves the realization that all lymphocytes are clonal and that only a small fraction of clones will have translocations. Ignoring the clonal nature of the aberrant cells recognizes that the chromosomally normal cells are also clonal. As with the clones of abnormal cells, the clones of normal cells also expand and contract over time. Which approach for handling clones of abnormal cells should be used? In short, when performing dosimetry on an individual it is probably best to adjust the translocation frequency, as this estimates the amount of induced damage. When calculating the average dose to a population it is probably best not to adjust the translocation frequency, as there should be enough subjects evaluated that the effect of any one clone on the population mean and standard deviation will be minimal. Indeed, ignoring clones at the population level would lead to an underestimate of the average translocation frequency. To obtain a better understanding of the relative importance of age, smoking, race, and gender on translocation frequencies, an effort was undertaken to collect and evaluate the world s available data on translocation frequencies in people who had not been exposed to ionizing radiation except for routine diagnostic evaluations [72]. Sixteen laboratories from around the world contributed data on a total of 1933 subjects in which whole chromosome painting had been performed. Age was associated with translocation frequencies and this relationship showed significant upward curvature at older ages compared to a linear increase. Current and former smokers had significantly higher translocation frequencies than non-smokers, and smoking modified the effect of age on translocation frequencies with a steeper age-related increase among those who have ever smoked compared to non-smokers. Translocation frequencies did not differ by gender. The racial distribution of subjects by laboratory was markedly non-random, precluding a definitive analysis of this variable. This study has provided the most reliable estimates to date of background translocation frequencies by age, gender and smoking status. These data are important for designing future studies of low-level environmental genotoxins or for retrospective evaluations of radiation exposure following an accident. 9. Improving the assessment of low doses of radiation Exposure to low doses of ionizing radiation is a fact of life in certain occupational settings. Radiation accidents, while unfortunate at the minimum and devastating in the worst cases, will no doubt continue to occur. Fortunately most radiation exposures involve low doses (1 Gy) and as such do not have immediate lifethreatening effects. However, long-term effects of low-dose exposures, especially the potential for increased risks of cancer,

7 J.D. Tucker / Mutation Research 659 (2008) may be real and should be given serious consideration. Epidemiological approaches are useful for estimating risks to populations but are not useful at the level of individuals. Cytogenetic analysis offers the hope of establishing risk estimates for individuals but the relatively large number of known confounding variables together with statistical uncertainties associated with cytogenetic measurements currently preclude reliable risk estimates for individuals. Is there any realistic basis for believing that risks following exposures below 1 Gy can be reliably assessed? For populations, the answer will need to be based on epidemiological studies. Just as the tools of epidemiology have been used to estimate the relative importance of risks due to factors other than radiation, these same methods hold promise that the risks of low-dose radiation may be reliably established. For individuals, the answer is not as clear. Obtaining reliable risk estimates will require more precise measures of chromosome damage. This will require counting large numbers of translocations which is currently a timeconsuming process due to their rarity. Until or unless rapid measurement techniques become available (e.g., accurate and reliable computerized scoring of chromosome aberrations), individual risk assessments are likely to remain elusive. With the present set of technological capabilities, what can be done now to improve our understanding of the effects of low-dose exposures? The answer has two parts. First, it will be important to obtain much more accurate biological measurements of radiation doses in exposed people. Second, a large prospective population study should be conducted to quantify and characterize baseline translocation frequencies in unexposed people across a broad spectrum of societies. These two answers will now be discussed in more detail Improving biological dosimetry in exposed people There are two components associated with improving the dose estimates for biological dosimetry. The first is to obtain more precise measures of translocation frequencies in individual subjects exposed to <1 Gy. This can be achieved by increasing the number of cell equivalents that are scored, which can be accomplished by counting more cells as well as by optimizing the number of chromosomes that are painted and the number of colors used to achieve painting. This can be accomplished with current technology. The idea here is to identify a sufficient number of translocations per subject (e.g., N = 30) that the statistical counting errors are small. This increased precision would require cytogenetic analyses of thousands of cell equivalents per subject rather than the hundreds typically scored now. For statistical purposes, the goal for each subject should be to identify a specific number of translocations (or perhaps cells with translocations) rather than to score a pre-determined number of cells. The reason is that the statistical confidence provided by each measurement is dependent upon the number of translocations observed, not the number of cells scored. The actual number of translocations that should be counted will of course vary depending upon the desired level of statistical power for estimating the doses. In situations where clones of abnormal cells are observed, it is recommended that each clone be counted as a single abnormal cell. The low-dose biodosimetry effort should include the long-term cytogenetic follow-up of exposed subjects. The primary question is the issue of persistence in peripheral blood lymphocytes of translocations induced by low-dose exposure. Translocations induced in vivo by high doses decline in frequency over time [4], not unlike translocations acutely induced in mice in vivo [53,89]. Translocations induced by acute exposure to as little as 20 cgy in human cells in vitro also decline over time [2,33]. However, the persistence of translocations induced in vivo by low doses has not been evaluated in humans, in spite of the fact that this is the most common type of exposure. Quantifying the persistence of low-dose induced translocations will be important for determining whether it is the amount of translocations that are initially induced, or the persistence of those translocations, that best predict long-term cancer risks. A secondary question that should be addressed by this study pertains to the presence of clones of cytogenetically abnormal cells. As described above, clones of cells with translocations and even insertions have been observed in healthy human subjects. What is not known is the extent to which the frequency of these clones varies over time, or whether the presence of clones themselves might be indicative of increased cancer risks. A study of the persistence of translocations in peripheral blood lymphocytes following exposure would help address this question. Yet a third question that could be addressed by this study is the persistence of complex aberrations, and whether the frequency of these events shows an association with cancer. The second component needed to improve biological dosimetry involves obtaining accurate prospective measurements of translocation frequencies in a very broadly based set of radiationunexposed populations. This is needed to provide a robust set of baseline or control values against which the observations made in the exposed subjects can be compared. The currently available data indicate the importance of age, cigarette smoking, and possibly genotype and diet in baseline levels of translocation frequencies. The importance of other important variables such as race and nationality on translocation frequencies has not been established, as noted above. Only by determining the normal range of translocation frequencies can the effects of low radiation doses ever be identified and characterized. The issue here is one of detecting a small signal against a background noise of similar cytogenetic events induced by exposures other than to radiation. In vitro dose response curves and subsequent careful statistical analyses of those curves have been useful for determining the signal that is expected following a given dose. Where our knowledge is lacking is in understanding and characterizing the background levels against which any increase due to exposure must be compared. This problem is unique to low-dose exposures, as the signal-to-noise problem is not nearly as relevant for highdose exposures. Careful characterization of the background levels will involve an epidemiologically based study which is described in the following section Population study of unexposed people Population characterization of baseline translocation frequencies in unexposed subjects is absolutely essential for understanding and characterizing the biological effects of low-level exposures. This effort should be international in scope to capture the maximum possible information from as many ethnic groups and nationalities as possible. Each subject should be analyzed for chromosome aberrations, including translocations, by FISHpainting with the same set of probes used throughout the entire effort. Furthermore, the study should seek to sample the full spectrum of subjects involving the widest possible range of ages including children, and would involve obtaining information detailing demographic and lifestyle factors including but not limited to gender, national origin, race, tobacco use, dietary preferences, and alcohol consumption. Some consideration should be given to determining the altitude above sea level at which the subjects have lived, as the higher altitudes afford less protection from solar and cosmic radiation that may damage chromosomes. Assessments of physical exercise should also be considered as DNA