Darkness in El Dorado - Archived Document
Internet Source: The Genetic Effects of Ionizing Radiation on Humans, IX International Congress of Human Genetics, August 24, 1996
The Genetic Effects of Ionizing Radiation on Humans by James V. Neel
A presentation read at "Chornobyl: Implications of a Decade" held at the IX International Congress of Human Genetics, August 24, 1996
The Chernobyl Disaster in 1986 has created widespread concern regarding the somatic and genetic effects of the resulting exposures to ionizing radiation. Today I would like to provide a perspective to those genetic concerns by sharing with you the results of my 50 years of involvement with the most nearly comparable situation in human history, namely, the aftermath of the atomic bombings of Hiroshima and Nagasaki. First, I will describe the findings of these genetic studies carried out at the Atomic Bomb Casualty Commission (ABCC) and its successor agency, the Radiation Effects Research Foundation (RERF). Next, I will critique a recent study by Dubrova and colleagues on the effect of the fallout from the Chernobyl disaster on mutations in minisatellites, a study which I consider highly misleading. Finally, I will discuss the probable results of the Chernobyl exposures in light of the findings in Japan.
1. Brief summary of past studies on the genetic effects of the atomic bombs
The genetic studies of the children of survivors of the atomic bombings have been several times summarized in the past five years (Neel et al., 1990; Neel and Schull, 1991; Neel, 1995), and I propose to be very brief in the present summary. As of now, the reproductive performance of atomic bomb survivors living in Hiroshima and Nagasaki is complete. Most of the studies to be mentioned have been based on a cohort (or subset thereof) consisting of 31,150 children born to parents one or both of whom were within 2000 meters of the hypocenter at the time of the bombing (ATB) and a matched comparison cohort of 41,066 children born to survivors beyond this distance, or to parents now living in either city but not there ATB. Over the years, at the follow-up group in Japan, the Radiation Effects Research Foundation, data have been collected on these children (Fl) concerning untoward pregnancy outcomes (major congenital malformation and/or stillbirth and/or neonatal death); sex of child; malignant tumors with onset prior to age 20; death of liveborn infants through an average life expectancy of 26.2 years, exclusive of death from malignancy; growth and development of liveborn infants; cytogenetic abnormalities; and mutations altering the electrophoretic behavior or function of a selected battery of erythrocyte and blood plasma proteins. Most of the parents have been assigned radiation exposures based on a formula developed at the follow-up agency; where this was impossible, doses have been approximated by a less elaborate formula. All of the data have been analyzed on the basis of these exposures.
In a series of analyses that involved fitting a linear dose-response regression model to the occurrence of the various indicators of radiation damage, the model including up to six concomitant variables depending on the indicators in question, there was no statistically significant effect of combined parental exposure on any of these indicators. I wish to emphasize that relatively limited though our knowledge of mammalian radiation genetics was in 1947, when the decision to undertake this study was made, this possible outcome was anticipated (Genetics Conference, 1947), but the decision was made to proceed nevertheless with what has become the largest study in genetic epidemiology ever undertaken, on the basis that whatever the outcome, it was of profound interest.
In the planning stages of the study, inasmuch as most births were midwife-attended home deliveries, we were gravely concerned over the possibility that the birth of a child with congenital abnormalities might be concealed, since such births, in a society with open family (koseki) records, would stigmatize the family. We went to great pains to enlist the cooperation of the midwives and obstetricians, who readily grasped the significance of the study. The attendant at delivery was urged to report at once any abnormality, whereupon a physician employed by ABCC examined the baby as soon as possible; in any event, all newborns were examined by an ABCC physician within 2 or 3 weeks of birth, and a 30% sample reexamined at approximately 9 months of age. It is a source of confidence in the study that the data on congenital defects in the children of unexposed parents were very similar to world figures (Neel, 1958). This clinical program was terminated in 1953.
A fact that we believe went far to ensure the quality of the data was that for some years after the end of WWII, the Japanese government maintained the wartime system of special rations for Japanese women who registered their pregnancy at the completion of the fifth lunar month. By incorporating a registration for the genetic studies into this ration system, we had the basis for a prospective study of pregnancy outcomes. This permitted us to determine the outcome of all pregnancies in the two cities that lasted at least five months, an aspect of the study design whose importance we cannot overemphasize. The registrations for the genetic studies were initiated in March of 1948; these children would have been conceived in October, 1947. There is thus a hiatus of some two years between the bombings and the first direct collection of data (but births in those first two years were incorporated into certain of the study groups).
A convenient metric for measuring the genetic effects of ionizing radiation is the estimated dose of radiation necessary to equal the impact of spontaneous mutation on the indicator in question, i.e., the doubling dose. If we assume that despite the absence of a statistically significant regression of any of the indicators on parental exposure, the regressions that were derived are nonetheless the best current indicators of the genetic effects of radiation on humans, then these regression can be employed to generate a doubling dose. In order to make such an estimate, one must have an estimate of the contribution of spontaneous mutation in the parents to each of the indicators. This is sometimes relatively easy, sometimes, as in the case of major congenital defect, more difficult. Also, the estimate is restricted to the effects of the mutation of single genes with discrete effects. This removed from usefulness to our efforts to calculate a doubling dose such indicators as physical growth and development and the sex ratio (although neither of these yielded findings suggestive of a radiation effect). For the remaining five indicators, the findings were as shown in Table 1. Since all the regressions evaluate the effects of essentially non-overlapping indicators (i.e., are independent of each other) and are based on the same cohorts or subsets thereof, it is legitimate to combine these additively. The sum of the five individual regressions is 0.00375/Sv equivalent with a considerable but difficult-to-define error. We estimated, as shown in Table 1, that each generation of newborns, something like 6.3 to 8.4 infants per 1000 would ultimately exhibit some one of these five indicators because of mutation in the parental generation. The estimate of the doubling dose thus falls between 0.00375/0.00632 Sv equivalents and 0.00375/0.00835 Sv equivalents, or 1.7 to 2.2 Sv equivalents, with a poorly-defined error term. This is a zygotic doubling dose. At the time of the study, the socioeconomic status of the exposed (and their children) was slightly below that of the unexposed. Inasmuch as a lower socioeconomic status should have the same impact on congenital defect and infant survival as induced mutations, we believe the doubling dose estimate is conservative, i.e., the doubling dose is apt to be even higher.
Most of the gonadal ionizing radiation humans will experience will be low-level, chronic, or intermittent, rather than the acute (single dose), relatively high exposures of the atomic bombs. I say 'relatively' because the average gonadal dose for both parents combined varied in the various different studies between 0.32 and 0.60 Sv equivalents, a high dose as human exposures go but small by the standards of Drosophila and mouse-based experiments. In the calculation of these doses the relatively low neutron component was assigned a relative biological effectiveness (RBE) of 20. Acute radiation at relatively high doses is on a per unit basis genetically more mutagenic than chronic radiation, the dose rate factor ranging from 3 to 10, depending on the precise indicator. The best studied system in mice, the Russell 7-locus test, yielded one of the lowest dose rate factors, of 3. Because of the linear-quadratic nature of the radiation dose-genetic damage curve, we suggest that the dose rate factor to be used in extrapolating from the effects of acute to chronic radiation on the basis of the Japanese data should be 2, so that the zygotic doubling dose for low level chronic ionizing radiation in humans becomes 3.4 to 4.4 Sv equivalents.
2. A comparison of the findings in Hiroshima-Nagasaki with studies of the genetic effects of ionizing radiation on mice
This doubling dose estimate for humans is substantially higher than the extrapolation to humans from the results of experiments on mice that dominated thinking about genetic risks from approximately 1950 to 1985. The figure most often quoted for acute radiation was about .40 Gy. Accordingly, Susan Lewis and I (1990) undertook a review of the accumulated data on mouse radiation genetics. Because of the relative immaturity of mice at birth, as well as the intra-litter competition both before and after birth, we felt that the murine data on congenital defect and survival of liveborn offspring could not be compared with the human data. The most nearly comparable mouse data resulted from the specific locus-specific phenotype studies. For each of these indicators we derived an estimate of the doubling dose. The results are shown in Table 2. Note the wide range in the various estimates, to which we found it impossible to attach errors in the usual statistical sense. Not shown there (because the data do not lend themselves to the calculation of a doubling dose) are the important results of Roderick (1983), who estimated for mice a recessive lethal mutation rate in post-spermatogonial cells per locus from ionizing radiation of only 0.35 x 10-8/0.01 Gy, whereas for the Russell 7-locus system, the corresponding rate for all mutations was 45.32 x 10-8/0.01 Gy, approximately 80% of these mutations being homozygous lethal. As Roderick pointed out, this was about a 100-fold difference, although the error term to be attached to his estimate was large but difficult to calculate. The simple average of all the estimates in Table 2, unweighted because we could think of no good way to weight the individual studies, was 1.35 Gy, with an indeterminate error. Given the relatively high doses at which the mouse experiments had been performed, we felt that in extrapolating to the effects of low level, chronic or small intermittent exposures, a dose rate factor of 3 was appropriate (Russell et al., 1958). The estimate of the gametic doubling dose of chronic ionizing radiation thus became 4.05 Gy, and the zygotic doubling dose, approximately 8 Gy.
In retrospect, it is instructive to consider why the results of the Russell system so dominated the thinking of 20 and 30 years ago. It was a simple system yielding clear cut results for which a great deal of data became available. When other studies suggested lesser genetic effects of ionizing radiation, the results were often dismissed as reflecting the use of 'less sensitive' systems. Furthermore, the desire to be conservative in risk setting led to reliance on the systems that yielded the most striking results. However, I suggest that, for reasons now to be enumerated that Russell could not know at the time, the loci he selected probably had higher spontaneous and induced mutation rates than the average locus.
Russell in his very first papers recognized that the assumption that his loci were representative of the genome was key. There are now data for the mouse indicating a 7-fold range in the rate per locus with which spontaneous mutation results in phenotypic effects (Green et al., 1965; Schlager and Dickie, 1967). With respect to humans, Chakraborty and Neel (1989) from population data have suggested a 10-fold range in the spontaneous mutability of human genes. In Russell's data, radiation produced 18 times more mutations at the s locus than at the a locus, surely a signal to extrapolate with caution (rev. in Searle, 1974). Finally, in a somatic cell mutagenization experiment in our laboratory utilizing the TK6 line of human lymphocytoid cells and employing ethylnitrosourea as mutagen, the protein products of 263 loci were scored for the occurrence of mutants resulting in electrophoretic variants, employing a two-dimensional polyacrylamide gel system (Hanash et al., 1988). Ten of the 263 loci whose protein products were being scored were known to be associated with genetic polymorphisms, on the basis of family-oriented studies on gels derived from human peripheral lymphocytes. The induced mutation rate at these 10 loci was 3.6 times greater than at the monomorphic loci, an observation with a probability of occurring by chance of < 0.004. The relevance of all these observations to the possible bias in the Russell system was that to set the system up, Russell drew on loci characterized by genetic variation. There had to be at least two alleles known for each of the loci in his system, and it helped in creating the optimum phenotype for scoring if there were even more alleles available to choose among. This use of loci for which variants were readily available introduced the bias.
There are two additional reasons why the previous extrapolation from mouse to man was conservative. First, the mouse doubling-dose estimate was male based. The demonstration (Russell, 1965) that although in the first few litters post-treatment the offspring of radiated female mice exhibited about the same amount of genetic damage as the offspring of radiated male, there was no apparent damage in the later litters of these females, created a dilemma for risk setting. Was the human female similar to the mouse female in this respect? The estimates of Table 2 are male based. To be conservative, in extrapolating to the human situation, the mouse male-derived risks were applied to both sexes. In the Japanese data, however, radiated females contribute about half the dose. Second, it appears that Russell excluded some fraction of the mutations that occurred in clusters from his calculations. Since cluster mutations should be relatively more common in the control series, the net impact of this practice would be to bias the doubling dose estimate downward.
As a result of the studies on the genetic effects of the a-bombs, plus this evaluation of the totality of the murine data, the case for a major revision downward in the previous estimates of the genetic risks of radiation for humans must be very seriously considered.
3. New techniques for evaluating the genetic effects of the atomic bombs
In the 1980s, it became apparent that the techniques of molecular genetics might be brought to bear on the question of the genetic effects of the bombs. Accordingly, in 1985, the follow-up group in Japan began the task of establishing transformed cell lines appropriate to such studies, the goal being approximately 600 mother/father/child trios in which one or both parents had been exposed to the atomic bombs, and a similar set of comparison constellations in which neither parent had been exposed. Where more than one child was available, s/he was included in the study, and the mean number of children per parental set is 1.45. The task of establishing the cell lines is almost completed. Meanwhile, the Genetics Staff launched on an exploration of the technologies that might be employed. To date, three systems utilizing the DNA approach have been piloted out by RERF staff and their collaborators.
3.1. A DGGE system
The first of these pilot systems employed the denaturing gradient gel electrophoresis (DGGE) technique of Myers et al. (1985) and Lerman et al. (1986), in combination with other techniques to detect, primarily, unknown single nucleotide substitutions (Hiyama et al., 1990; Satoh et al., 1993; Takahashi et al., 1990). In terms of nucleotides examined per unit technician time, the most efficient of the several variations of this approach that were explored involved amplification of target sequences with PCR, digestion of these sequences to fragments of approximately 500 bp, followed by DGGE of the fragments (Satoh et al., 1993). A total of 6724 bp were examined per individual, using the human coagulation factor IX gene as substrate. In this pilot study, samples from 63 couples and 100 of their children were examined. Half of the children were born to parents one or both of whom had been exposed to the atomic bombs. Eleven previously undescribed nucleotide substitutions were detected in the parents. No mutations were detected in the approximately 672,000 nucleotides examined in the 100 children in the study, not surprising in view of the estimated spontaneous mutation rate per nucleotide per generation of 1 x 10-8 (Neel et al., 1986). This technique is most efficient in the detection of nucleotide substitutions.
3.2. The use of minisatellites
The second approach explored by the follow-up group in Japan (Kodaira et al. 1995), concentrates on studying mutation that alters the length of minisatellite loci, also known as VNTR loci (Variable Number of Tandem Repeats). Minisatellites are regions of the DNA characterized by a variable number of tandem repeats of identical units varying from 5 to 45 b.p. in length. They are relatively common in the genome and well known to exhibit a high rate of mutation, mutation in this case consisting primarily of the gain or loss of one or several of these repeat units. Employing Southern blot analysis of a battery of six minisatellites (Pc-1, l1M-18, ChdTC-15, plg3, IMS-1, and CEB-1), the investigators identified six mutations in 390 alleles from 65 parents whose gametes represented an average exposure of 1.9 Sv equivalents, a mutation rate of 1.5%, and 22 mutations in 1098 alleles of the 183 gametes from the unexposed parents, a mutation rate of 2.0%. The difference is not statistically significant and, in any event, is in a direction opposite to hypothesis. The authors calculate that given the observed spontaneous mutation rate, using standard power function statistics (a type I error of 0.05 and a type II error of 0.20), it would be necessary to survey with this technique two samples (exposed and unexposed) of 1188 germ cells each, to detect a significant difference at the 0.05 level. Certainly, given the need to extend our knowledge of the genetic effects of radiation, the series should be extended at least that far. Furthermore, given the general acceptance of the fact that ionizing radiation produces mutations, even without a significant difference between the two data sets, these data, like the data from the previous studies, can be taken at face value and used to produce a doubling dose estimate for this phenomenon, but only if, unlike the present situation, there is an excess of mutations in the children of exposed. However, even if the present deficiency of mutations in the children of exposed were to persist in an expanded series, the data can be used, at stated probability levels, to place a lower limit on the doubling dose.
Because the function of these minisatellites is so poorly understood, they are not a very satisfactory marker of radiation damage for a public interested in the phenotypic impact of an increased mutation rate on its children, and such studies should only be undertaken in conjunction with other DNA-based studies. However, it occurred to me, in writing an invited editorial to accompany the paper of Kodaira et al. (Neel, 1995), that these studies might have a value that could not have been anticipated a few years ago. Since 1991, eleven diseases have been recognized for which the mutational basis is an increase in the numbers of a specific trinucleotide repeats embedded in the gene (rev. in Ashley and Warren, in press). Some of these are quite well known causes of mutational morbidity (e.g., the fragile X syndrome, myotonic dystrophy, Huntington's chorea). The basic repeat unit in the minisatellites employed in the study of Kodaira et al. varies from 5 to 43 nucleotides. If mutation at these loci can be shown to follow the same principles as mutation affecting the length of minisatellites, then the latter may in the future serve as surrogates for the former.
3.3. Two-dimensional DNA gels
We come now to a third approach, in which our group in Ann Arbor has been collaborating with Dr. Asakawa of RERF. While RERF was establishing the cell lines, we in Ann Arbor devoted considerable effort to exploring the utility of two-dimensional polyacrylamide gel electrophoresis of protein solutions in the study of mutation (Neel et al., 1984; Neel et al., 1989). Figure 1 illustrates a typical preparation based on peripheral lymphocytes, in which approximately 200 proteins can be visualized with the clarity necessary for unequivocal classification. We obtained good identification of electrophoretic variants of these proteins, but, as a result of the many steps between gene and protein quantity, only a small subset of these proteins was sufficiently reproducible quantitatively that there was satisfactory discrimination between those in which a specific protein was represented in half normal amounts and those for which the normal amount was present -- a vital prerequisite to the study of null mutations. However, we were successful in developing an algorithm that, with a minimum of operator intervention, would compare the gels of a father, mother, and child, with the objective of identifying characteristics of the child's gel not present in either parent, and so putative mutations (Kuick et al., 1991; Skolnick and Neel, 1986). This approach to mutation requires that gels be run on the child and both parents.
Beginning in 1979, techniques for visualizing the DNA fragments resulting from genomic digests on a two-dimensional gel began to become available (Fischer and Lerman, 1979; Hatada et al., 1991; Uitterlinden et al., 1989; Yi et al., 1990). We are now exploring the applicability of this approach to studying the genetic effects of the atomic bombs, employing a modified version of the technique described by Hatada, et al (1991), as developed by Asakawa (1995). Hatada, et al. (1991) have termed this technique "restriction landmark genomic scanning" (RLGS). A gel based on a lymphocytoid cell line in the RERF collection in shown in Figure 2. For a diploid organism such as our species, in the absence of sex-linkage or genetic variation, each spot is the product of two homologous DNA fragments. [For these preparations, genomic DNA was digested with NotI and EcoRV restriction enzymes and the Notl-derived 5' protruding ends were a-32P labeled. These fragments were electrophoretically separated in an agarose disc gel, which was subsequently treated with HinfI to further cleave the fragments in situ. The resulting fragments are separated perpendicularly in a 5.25% polyacrylamide gel (33 cm x 46 cm x 0.05 cm). Autoradiograms are then obtained (Asakawa et al., 1994, 1995; Kuick et al., 1995).]
The visual comparison of the gel of a child with those of its parents, to detect attributes of a child's gel not present in either parent, i.e., a potential mutation, would be extremely demanding, the type of activity guaranteed to lead to a high turn-over rate in technicians. Fortunately, the computer algorithm developed for the analysis of protein gels did just as well with these complex DNA images (Asakawa et al., 1994). Among the approximately 2000 DNA fragments to be visualized on these preparations, we initially identified a subset of approximately 500 for which the coefficient of variation (CV) of spot intensity is < 0.12, this reproducibility permitting the distinction with high accuracy between spots of normal intensity and spots with 50% intensity (i.e., two-fragment or one-fragment spots). Already we have identified mendelizing genetic variation involving some 10% of these fragments (See Fig. 3). We believe that with impending technical developments, the battery of fragments suitable for quantitative scoring may increase to 600 or 700. Other enzyme combinations can be used for the genomic digests that precede the gel runs, and, by altering the electrophoretic conditions, larger DNA fragments can be visualized. Currently we are working on three different types of gels for which we believe there is little overlap in the DNA fragments visualized. Furthermore, we have demonstrated the feasibility of recovering (and characterizing the nucleotide sequence of) specific fragments from the gels (Asakawa et al., 1994). Thus, a mutant fragment can be precisely studied. For the study of mutation, this approach, as in the study of proteins, requires running gels on both parents as well as the child.
We would not like to seem to lead you to believe that there is something magic in these new approaches. Their implementation will require a major effort. For each of these approaches, RERF staff and ourselves have attempted to make calculations of the magnitude of the effort required to reach a significant difference between the children of exposed and controls, based on a variety of assumptions considering the doubling dose for the mutational endpoint. There is no time in this presentation to go into the excruciating details of these calculations; let me say only that given what we now know, only a very major effort can be expected to demonstrate a significant difference between the children comprising the two samples. I think it only appropriate to emphasize that these DNA technologies are demanding and that creating the amount of data to reach a reasonably firm conclusion, even in a situation such as the atomic bomb exposures, will be quite laborious.
4. Can a bridge be built between somatic cell genetic studies in a-bomb survivors and germline studies in their offspring?
For many years now, the desirability of somatic cell indicators of genetic damage has been obvious, and several possible systems have been explored at RERF. Four such systems deserve mention: 1) frequency of cytogenetic damage in cultured lymphocytes; 2) frequency of mutations in the glycophorin system; 3) frequency of mutations in the HGPRT system; and 4) frequency of mutant T lymphocytes defective in the expression of the T-cell antigen receptor gene. Time does not permit a discussion of the pros and cons of each of these systems. Each of these systems has been shown to provide evidence of genetic radiation effects, but each has its limitations as a barometer of germline damage. For instance, one of the standbys in cytogenetic studies, the dicentric chromosome, is unstable and either would not be transmitted at gametogenesis or, if transmitted, would probably be incompatible with fetal survival. The nature of the genetic variation revealed by the glycophorin system cannot be studied because the erythrocyte is enucleate, but the phenotypic findings suggest that many of the changes detected are the result of somatic cell crossing over rather than a mutation in the usual sense of the term. For each of these indicators, there is the question of how well it represents the genome as a whole. For all of these, there is also the question of the nature of the damage decay curve following the initial genetic damage. For these and other reasons, the investigators working with these systems have been properly cautious in suggesting doubling dose estimates, albeit these are the type of estimates needed for comparison with germinal rates and for guidelines regarding permissible exposures.
We would like to suggest a fresh approach to the matter of a-bomb-induced somatic cell damage. It is possible to clone directly individual lymphocytes transformed by the Epstein-Barr virus (See Fig. 4). The detailed similarity to the preparation based on the Japanese cell lines is striking. Although our studies are still preliminary, we suggest that this technique may represent a new and powerful multi-locus approach to the study of radiation-induced somatic cell damage, the technique having the additional advantage that the results of a study on a-bomb survivors would be directly comparable to the results of a study of their children that used the same technique. We remind you that just as is possible for the studies in the F1 cell lines, in principle, any apparent somatic mutation that is detected in this system can be precisely characterized. However, as with germline studies using this technique, somatic cell studies of this nature will be laborious.
5. 'If we had it to do over again'
Looking back on this long-running complex study, it is important to ask, given all the amazing developments in genetic science these past 50 years, how differently should such a study be designed if it were beginning today? The most obvious change in the research design would be to include studies at the DNA level from the outset. However, it will be some years before it is possible to extrapolate with the desired precision from damage at the DNA level to gross phenotypic effects, and these latter are what the public which ultimately supports such studies really wants to know. Accordingly, we would suggest that any future study should still include most of the components of the study in Japan: frequency of congenital malformations and still births, death rates among liveborn children, growth and development of surviving children, cancer and chromosomal abnormalities in children of exposees. The one study that would probably not be repeated would be a search for electrophoretic and activity variants in proteins. These studies were undertaken at the time as the best available approach to nucleotide substitutions and small deletions in DNA, which comprise roughly one-third of the spectrum of radiation-induced damage in DNA. Now the ability to examine DNA directly has the potential to be much more efficient in the detection of such lesions. Otherwise, however, looking both backward and forward, we suggest that if ever again a study were undertaken of the genetic effects of an ionizing or chemical mutagen, if finances permitted, the study should include all but one of the components of the Japanese study plus, now, a DNA component.
6. A critique of the recent Dubrova-Jeffreys study.
I hope that by now I have persuaded you that a proper genetic study of the effects of a radiation exposure upon a population is highly laborious and demands impeccable controls. In such a situation, it is tempting to look for a short-cut, some kind of easily applied litmus paper test. That is exactly the trap into which Dubrova and colleagues seem to have fallen in their recent report in Nature (25 April 1996). Studying germline mutation rates in human minisatellite loci in the fallout contaminated areas of the Mogilev district of Belarus and in a control population, they report that: "The frequency of mutation was found to be twice as high in the exposed families as in the control group." As noted earlier, minisatellites are regions of the DNA characterized by a variable number of tandem repeats of, usually, 5-45 bp, identical units. They are relatively common in the genome, and well known to exhibit a high rate of mutation, mutation in this case consisting primarily in the gain or loss of one of several of these repeat units. The mechanisms responsible for these mutations is not well understood. These mutations are not accompanied by any detectable phenotypic effect, i.e., appear to be of no consequence to the person to whom they are transmitted. This report, in a highly respected journal, can only bring great and unwarranted distress to an already badly stressed population. There are at least five reasons why that report must be viewed with great caution.
1. Choice of controls. The control group of parents and children were drawn from the United Kingdom. This is a violation of one of the first principles of genetic epidemiology, since these controls are neither matched by environment nor by ethnicity. With respect to the former, Belarus and surrounding regions are known to be subject to a high level of industrial pollution. With respect to the latter, strain differences in mutation rates are very well known in mice and Drosophila, and, especially for such labile markers as minisatellites, certainly cannot be excluded for the human species.
2. Dosimetry. There are very serious problems with the dosimetry. The principle source of radiation is 137Cs, the average individual dose for external and internal chronic exposure to this element estimated "to be less than 5 mSv per year". Since the children studied were born in 1994, both their parents had been exposed to this dose for approximately 8 years, for a total maximum dose of 8 x 2 x 5 = 80 mSv, or 8 rem, and the average may be half of that! However, what is important is the gonad dose, and this is probably less than half the external dose. So -- we are talking about gonadal doses of radiation only a few rem above background.
3. Lack of perspective. Since the estimates of the doubling dose of chronic radiation for mice and humans is some 4.0 Sv equivalents, their finding suggests sensitivities some 200 times greater than the literature. The authors recognize the discrepancy, and invoke the induction by the radiation of some unknown intermediary mechanism which then causes the mutation. Since the basis for the gain or loss of repeat units which they are measuring in these minisatellites is unknown, it is difficult to discuss this suggestion critically, but it is difficult for me to imagine a "special mechanism" specifically for this type of mutation which has not previously come to light.
4. Treatment of results from Japan. At the time the authors submitted their paper to Nature, they were unaware of a comparable study at the RERF, referred to earlier in this presentation. This is understandable, since the results of the study in Japan were published in December of 1995, and the Nature paper appeared in April, 1996. The Japanese study did come to their attention in time to add a note in proof, namely, "A similar study of minisatellite mutation in a relatively small number of families of atomic-bomb survivors from Hiroshima and Nagasaki has failed to show evidence for mutation induction following acute exposure." This is a quite misleading statement. The average gonadal dose of the parents of the 64 Japanese children who were studied was 1.93 Sv equivalents, so that the total dose represented by these children was 64 x 1.93 = 123.5 Sv equivalents of acute radiation. The exposure of the parents of the Belarus children was ~7.5 yrs x 5 mSv/year x 2 parents x 79 children = 5.93 Sv equivalents of chronic radiation. Acute exposures to ionizing radiation, as in Japan, are commonly held to be some 2-3 times as effective in inducing mutation as chronic exposures, as in Byelorussia. Therefore, the Japanese study should have (123.5 x 3)/5.9 = 62.8 times the statistical power of the Byelorussian study.
5. Choice of indicator. Finally, we must consider the choice of the DNA indicator for this study. Certainly a variety of types of DNA should be examined if studies of the genetic effects of radiation move to the DNA level. The choice of minisatellites was perhaps dictated by their relatively high spontaneous rate, and the thought that they might be an unusually sensitive indicator of radiation effects. In my view, however, we should not be concentrating on presumably sensitive indicators, especially of unknown function, but on a representative battery, of which this might be only one.
There is an urgent need to repeat this study, and I presume such an effort or efforts is under way. However, one must be very aware of the apparently deteriorating environmental situation in that region of Eastern Europe, and the possibility that this affects the mutation rate of minisatellites.
6. What, if any, further genetic studies should be taken on the aftermath of Chernobyl? There are two groups of people exposed to radiation in consequence of the Chernobyl disaster who should have a sufficient number of children to provide the basis for a genetic study, namely, the "clean up" workers and those only exposed to fallout. At this point in time, certainly a properly designed effort to confirm the findings of Dubrova et al. (1996) is indicated, and, at the same time, the corresponding study in Japan needs to be substantially extended. But what further genetic studies, if any, are called for? There are two dominant considerations that should enter into a decision. First, by comparison with the Japanese situation, the epidemiological circumstances do not favor a study. The two exposed populations are widely dispersed; it would be very difficult to define a control, and 10 years have already elapsed since the disaster, so that there has been a major loss of data. I would judge any attempt to reconstruct a control and exposed birth cohort of the last 10 years to be ill-advised. Second, the radiation exposures are, from the genetic standpoint, really very low. The gonadal doses to the 'clean-up' workers apparently can't average more than .1 5 Sv equivalents. The maximum accumulated dose of chronic radiation from fallout to the parents combined, as mentioned above, is now about 5 x 2 x 10 = 100 mSv equivalents. This is 10/400 = 1/40th of the doubling dose for chronic radiation as we have estimated it. Even if the genetic effects of radiation are linear to dose to such very low levels of radiation, which implies that someplace in the former Soviet Republics there are a few children carrying mutations induced by parental exposures incidental to the Chernobyl disaster, no conceivable study can reveal this effect at a statistically significant level. Let's be very clear -- I am not suggesting there is no genetic effect from the Chernobyl fallout, but rather, that it is quite small and not to be detected by the epidemiological approach.
There will be those who argue that a study should be undertaken 'for reassurance'. I see that as a 'slippery slope'. Perhaps the simplest study would be a registry of sentinel phenotypes, such as Dr. Czeizel will describe. However, if such a study has a sufficient number of indicators, there is the likelihood of one or more false positives, and these are difficult to deal with. Two independent studies in parallel are a partial answer to the problem of false positives, but defining two equivalent study and control populations and finding the funding might be difficult. I look forward to a lively discussion of this matter later this morning, but will stop for now.
(To view graphics click ==> Table 1 *** Table 2 *** Figures)
Asakawa, J., Kuick, R., Neel, J.V., Kodaira, M., Satoh, C., Hanash, S.M. Genetic variation detected by quantitative analysis of end-labeled genomic DNA fragments. Proc. Natl. Acad.Sci. USA. 91:9052-9056 (1994).
|Content is copyright © by the authors, websites, or companies that originally published and/or wrote the text of this document.|
|Page design and layout is copyright © 2015, Douglas W. Hume.|