Effects on Populations of Exposure to Low Levels of Ionizing Radiation (1972)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Chapter V GENETIC EFFECTS OF IONIZING RADIATION I. Introduction and Brief History 42 A. Historical Basis for Radiation Protection Guides for the General Population 42 B. Early Genetic Risk Estimates 42 II. The Present Situation 44 A. How Has the Problem Changed Since 1956 44 B. What Has Been Learned Since the 1956 Report 44 III. What Kind of Genetic Damage Does Radiation Cause? 46 IV. Could an Increased Mutation Rate Possibly Be Beneficial? 48 V. Sources of Genetically Significant Radiation Exposure 49 VI. Risks Versus Benefits 51 VII. Methods of Estimating the Genetic Risks from Radiation 51 VIII. Risk Estimates 51 A. The Risk Relative to That From Natural Background Radiation 51 B. Risk Estimates for Specific Genetic Conditions 52 C. Risk Relative to Current Incidence of Serious Disabilities .. 55 D. Estimation of the Cost in Terms of Ill-Health 56 IX. Discussion 57 X. Summary and Conclusion 58 Appendices to Chapter V Explanatory Notes 60 References 70 41

Chapter V GENETIC EFFECTS OF IONIZING RADIATION I. Introduction and Brief History This chapter reviews briefly the information now available on the genetic risk to the human population from low levels of ionizing radia- tion and gives the Subcommittee,s conclusions and recommendations. Supporting evidence and further information are given in a series of Explanatory Notes. Our task was made much easier by the volu- minous reports and extensive bibliographies prepared by the United Nations Scientific Committee on the Effects of Atomic Radiation (See Explanatory Note 1). A. The Historical Basis for Radiation Pro- tection Guides for the General Population Although the discovery that radiation can cause mutations was reported by H. J. Muller in 1927, it was not until after World War II that genetic risks to the population were re- garded as a major factor in determining maxi- mum permissible doses. The emphasis instead had been on the protection of the individual who, for occupational or other reasons, might receive a radiation exposure that would be harmful to himself. In 1956, the National Academy of Sciences - National Research Council Committee on the Biological Effects of Atomic Radiation (the BEAR Committee) introduced a new concept, the regulation of the over-all average dose to the population. Because of the genetic risk to future generations, the BEAR Genetics Com- mittee recommended that man-made radiation be kept at such a level that the a verage individ- ual exposure be less than 10 R (roentgens) be- fore the mean age of reproduction, a period of time taken to be 30 years. Simultaneously, a report with a similar recommendation was is- sued by the British Medical Research Council. The 10 R numerical limitation was accepted by the National Committee of Radiation Protec- tion (NCRP) and included in its 1957 recommen- dation. The present Radiation Protection Guides for the general population grew out of these recommendations (See Explanatory Note 2). The BEAR Genetics Committee included med- ical radiation in its recommendations, the ge- netically significant medical dose (65 GSD) (pre- reproductive gonad exposure) at that time being estimated to be about half of the recom- mended 10 R limit. The Federal Radiation Coun- cil did not include medical radiation and, there- fore, took 5 R as the 30-year limit for the popu- lation average in the Radiation Protection Guides. This is 0.17 R per year, or 170 milli- roentgens, the value now in effect. There is at present no stated limitation on population ex- posure from medical practice. The Radiation Protection Guides are stated in rems rather than roentgens, since the rem takes into account differences in biological effectiveness of different kinds of radiation. We shall also use rems and millirems as the units of our discussion, and thereby assume that when radiations of different biological effectiveness are used the exposures have been converted into roentgen equivalents. B. Early Genetic Risk Estimates The 1956 Genetics report relied mainly on data from Drosophila and the laboratory mouse, as there were almost no relevant human data. According to the BEAR report, "the best one can do is to use the excellent information on such lower forms as fruit flies, the emerging information for mice, the few sparse data we 42

have for man. . .and then use the kind of biol- ogical judgment which has, after all, been so generally successful in interrelating the prop- erties of forms of life which superficially ap- pear so unlike but which turn out to be so re- markably similar in their basis aspects." The general principles that guided the com- mittee at that time were: (1) Mutations, sponta- neous or induced, are usually harmful; thus, the harm from an increased mutation rate greatly outweighs any possible benefit. (2) Any dose of radiation, however small, that reaches the reproductive cells entails some genetic risk. (3) The number of mutations produced is pro- portional to the dose, so that linear extrapola- tion from high dose data provides a valid esti- mate of the low-dose effects. (4) The effect is independent of the rate at which the radiation is delivered and of the spacing between expo- sures. The last of these principles has turned out to be incorrect, as will be discussed later. The BEAR Committee estimated that the amount of radiation required to produce a mutation rate equal to that which occurs spon- taneously (a "doubling dose") was almost sure- ly between 5 R and 150 R and probably between 30 and 80 R. It also assumed that about 2 per- cent of all live-born children are or will be seri- ously affected by defects with "a simple genetic origin." Under the assumption that for this fraction of human defects the incidence is pro- portional to the mutation rate, the effect at equilibrium after a continuing exposure to the recommended 10 R limit of radiation per gener- ation was computed. Taking 40 R as a reasona- ble value for the doubling dose, the BEAR Committee calculated that 10 R per generation continued indefinitely would lead to about 5,000 new instances of "tangible inherited defects" per million births, with about one-tenth this number in the first generation after radiation begins. The BEAR Committee also estimated the to- tal number of mutations which would be pro- duced at all gene loci by 10 R of radiation. The principles listed above made these calculations relatively simple. The number of mutations produced is (the number of genes in the popula- tion) x (the dose) x (the mutation rate per gene per unit dose). For the last quantity, mouse data were available. But there was no evidence from any mammal as to the number of genes per cell. For this, the Committee used Drosophi- ]a data, dividing the total mutation rate by that for individual genes. So the estimates of the number of mutations induced were for a hypothetical organism whose mutation rate per gene is that of the mouse and whose gene number is that of Drosophila (See Explanatory Note 3). The Committee then used the principle that each harmful mutant gene is eventually elimi- nated from the population and that this occurs by reduced viability or fertility. Thus, in a sta- tistical sense each new mutant gene, in a popu- lation of stable size, must eventually be bal- anced by a gene extinction. This extinction oc- curs through pre-reproductive death or re- duced fertility. The BEAR Committee was div- ided as to the usefulness of this kind of calcula- tion. It was noted that the death of an early embryo is much less traumatic than the death . of a child or adult and that the failure to re- produce cannot be equated to premature death in any tangible way. How is a single major de- fect to be judged in comparison with a number of minor risks? As stated in the report: "This kind of estimate is not a meaningful one to cer- tain geneticists. Their principal reservation is doubtless a feeling that, hard as it is to to esti- mate numbers of mutants, it is much harder still, at the present state of knowledge, to translate this over into a recognizable state- ment of harm to individual persons. Also, they recognize that there is a risk involved in ex- trapolating from mouse and Drosophila to the human case." But the group concluded that "in spite of all the difficulties and complications and ranges in numerical estimates, the result is nevertheless very sobering." Based on these estimates and other consider- ations which it regarded as germane, the BEAR Genetics Committee made two recommenda- tions that are related to our present purposes: "That for the present it be accepted as a uni- form national standard that x-ray installa- tions (medical and nonmedical), power installa- tions, disposal of radioactive wastes, experi- mental installations, testing of weapons, and all other human controllable sources of radia- tion be so restricted that members of our gener- al population shall not receive from such sources an average of more than 10 roentgens, in addition to background, of ionizing radia- tion as a total accumulated dose to the repro- ductive cells from conception to age 30." 43

"The previous recommendation should be reconsidered pepodically with the view to keep- ing the reproductive cell dose at the lowest practicable level. If it is feasible to reduce med- ical exposures, industrial exposures, or both, the total should be reduced accordingly." The present subcommittee concurs with this recommendation for periodic review and it is in this spirit that the present study has been un- dertaken. II. The Present Situation A. How Has the Problem Changed Since 1956? There have been three major changes: (1) As discussed elsewhere, somatic as well as genetic risks must now be taken into consideration in the setting of radiation protection guides for the general population, (2) the potential sources of population exposure have changed somewhat, and (3) new genetic knowledge ne- cessitates revision of some of the earlier ideas. When this problem was evaluated earlier, in 1956, the chief source of man-made radiation, aside from medical uses, was fallout from the testing of nuclear explosives. Exposure from nuclear power plants, although it was dis- cussed, was not yet an immediate concern. Now the situation is quite different. There is very limited atmospheric testing of nuclear explo- sives. In the future, there will probably be ei- ther no appreciable exposure from fallout, or the amount will be so staggering that any rea- sonable standard is breached. We are also faced with an energy problem. An expanding world population with ever-rising expectations for a higher standard of living, which implies more energy consumption, is confronted by dwindling supplies of accessible, economically feasible fossil fuels. There is, furthermore, in- creasing concern over pollution, and the fossil fuels are clearly major sources of atmospheric pollutants. Thus, not only does there appear to be a need for more energy, but the main alterna- tive to atomic energy, fossil fuels, is accompa- nied by newly recognized health and other haz- ards of its own. For the United States population in 1956, exposure from medical radiation was much greater than that from weapons testing or nu- clear installations. The genetically significant exposure to radiation during a 30-year period at the rates estimated at that time were 3R from medical uses and 0.1 R from fallout, with no measurable exposure of the general public from the nuclear power industry. Technical developments have made it possible to reduce the amount of medical radiation per procedure without significant loss of diagnostic efficiency. Although the number of radiological examina- tions per capita has gone up, the average ge- netically significant dose from this source has not increased proportionately and may have been reduced. Data are given below (Section V). It is important to stress that man-made ra- diation to which we are exposed is self-imposed by our demands for medical care and for ener- gy. Furthermore, as mentioned above, alterna- tives to radiation also involve risks. Accord- ingly, all calculations of possible genetic conse- quences of ionizing radiation from nuclear power developments and from medical practice must be set against our needs and the risks of the alternatives. The risk-benefit equation is particularly hard to balance in the case of ge- netic risks, for those who receive the benefits and those who run the risk are not the same people—those at risk may be many generations in the future. To summarize: The problem has not changed greatly since 1956, Medical radiation is still the major man-made contributor to the genetic risk. Nuclear power has become a reality and the radiation effect of this must be taken into account. Fallout from nuclear testing is re- duced. We are much more aware of other kinds of environmental genetic risks. Yet, the diffi- culties still remain. Despite the new knowledge that will be discussed in the next section, the assessment of genetic risks in any meaningful quantitative terms is still very uncertain. B. What Has Been Learned Since the 1956 Report? Since 1956, our knowledge of genetics has been revolutionized. The chemical structure of the gene and the nature of the mutation proc- ess are now understood in great detail. The number of recognized genetic diseases has in- creased by more than four-fold. One disease 44

after another is being understood in molecular and chemical terms. Human chromosomes can now be studied with great precision, whereas at the time of the first BEAR report not even the correct number of human chromosomes was known. Chromosome aberrations have been shown to be an important cause of human mal- formation and embryonic death. With such deep fundamental knowledge one might expect that estimation of radiation risks could now be made with considerable precision. Unfortun- ately, there are serious gaps in our knowledge. Most serious are: (1) almost complete absence of information on radiation-induced mutation in man; and (2) our inability to quantify the relation between an increased mutation rate and deleterious effects on human well-being. Recent studies of radiation genetics have brought out new, and in some cases unexpected, complexities. As mentioned before, the princi- ple of dose-rate independence has turned out to be wrong; it is now known that, for germ cell stages in the mouse that correspond to the human stages of the longest duration, radia- tion given at a very low rate or at widely spaced intervals produces fewer mutations than the same total dose given more rapidly. More mechanisms are now known by which ra- diation damage may be repaired in the cell. Furthermore, radiation effects differ among species, among strains of the same species, between the sexes, and among different cell stages (See Explanatory Note 4). For these reasons, although we have much information not available to the writers of the 1956 BEAR Report, we cannot be any more cer- tain than they were about quantitative assess- ments. A further difficulty is that, despite the increase in knowledge regarding the molecular nature of some chemically induced mutation processes, the molecular and cellular bases of genetic damage from ionizing radiation remain less well understood. Our knowledge is, there- fore, insufficient to provide a strong theoretical basis for extrapolation from one biological system to another, and to man in particular. Nevertheless, in the absence of accurate hu- man data we have no choice but to rely mainly on information from experimental animals. Assessing the changes in our knowledge since 1956, we now ask: Are the earlier estimates of risk too high or too low in view of current infor- mation? Several reasons suggest that the early risk assessments were, if anything, too high; others point in the opposite direction. Among those new findings suggesting that the 1956 estimates were too high are the follow- ing. It is now known that exposure of male mice at low dose rates produces considerably fewer mutations per rem than the same dose at the high rates on which earlier estimates were based. It is also known that those stages in the male that are most susceptible to genetic dam- age, spermatids and spermatozoa, make up a very short part of the human reproductive life span. Furthermore, transmission of genetic damage in these sensitive cells can be prevent- ed by postponement of conception until the mature sperm cells are derived from cells that were in a less sensitive stage at the time of the radiation. It is also known that the female mouse is, for most of her pre-reproductive life, much less sensitive to genetic effects of radia- tion than the male. Another reason for suspecting that the ear- lier estimates may have been too high comes from empirical studies on the descendants of irradiated mice. These studies have revealed substantially fewer harmful effects than might have been expected from mutation rates for single genes. These animals were exposed to high doses of radiation for many generations (more than 40 in one case) and yet the offspring showed no demonstrable effect on viability, fer- tility, or growth, nor were there any detected abnormalities attributable to the radiation (See Explanatory Note 5). On the other hand, there are reasons for thinking that the earlier estimates might have been too low. One is the increased realization that chromosome aberrations, structural and numerical, cause substantial genetic damage in man. A second reason is that, in addition to producing genetic effects by emitting radia- tion, radioactive isotopes may produce genetic changes as a direct result of the chemical change caused by transmutation. The latter, however, is believed to be a far smaller risk than the radiation effects (See Explanatory note 7). A third reason for thinking that the risk may have been underestimated in the earlier report comes from recent Drosophila studies. There is now strong evidence that mutations with very small effects occur with a much high- er frequency than do those causing a conspi- 46

cuous or lethal effect. That this class of muta- tions exists was realized earlier, but whereas at the time of the 1956 report they were esti- mated to occur with a frequency two or three times that of lethals it is now estimated that they are at least 10 times as frequent as lethals among spontaneous mutations. Furthermore, these mutations are expressed in the heterozy- gous state, so the effects would be manifest in the early generations after the occurrence of the mutation. There is supporting evidence for both of these conclusions from studies on bac- teria and yeast. These conclusions are mitigat- ed somewhat by Drosophila data suggesting that mutants with small effects expressed in the heterozygous state are less common, rela- tive to lethals, among radiation-induced than among spontaneous mutants (See Explanatory Note 8). The reasons for thinking that the earlier es- timates of risk were too high are probably stronger than those for thinking that they were too low, especially if we consider the next half-dozen generations (which, of course, is much farther in the future than is ordinarily considered in policy determinations). The opin- ion of the Subcommittee is that the genetic risk estimates in 1956 were probably on the high and therefore conservative side, but there are far too many uncertainties to be dogmatic. III. What Kind of Genetic Damage Does Radia- tion Cause? The genetic effect of radiation is to produce gene mutations and chromosome aberrations. Some of the ways in which radiation produces such effects are given in Note 9. The effect of radiation on the well-being of the future popu- lation is a consequence of these changes. Be- cause mutations and chromosome aberrations occur spontaneously, it follows that the conse- quences of radiation are not something new but rather an increase in frequency of various dele- terious traits with which we are already beset. Since almost every aspect of the living organ- ism is determined to some extent by its genes, the range of possible mutational effects encom- passes virtually every aspect of our physical and mental well-being. The major exception is infectious disease, but even here inherited sus- ceptibilities play a role. Some results of genetic change are conspicu- ous, others are invisible; some are tragic, oth- ers so mild as to be trivial; some occur in the first generation following the gene or chromo- some change, others are postponed tens or hundreds of generations into the future. Fur- thermore, most of the effects that are produced by mutation are mimicked by others, of nonge- netic origin. For all these reasons, radiation (or some oth- er environmental agent) could be having an important effect on human well-being and yet this could go unnoticed. Even if the increase in mutation rate is large, the consequences are likely to be so heterogeneous in their nature, so diluted by space and time, and so obscured by similar conditions from other causes as to make it impossible to associate them with their cause. Only if all the affected persons in future generations could somehow be identified and brought together at one time and place could the total impact of the mutations be apparent. One of the simplest categories of mutational damage includes those diseases and abnormali- ties that are caused by a single dominant muta- tion. The most recent compilation of McKusick (1) lists 415 such conditions with an additional 528 that are less well established. The collec- tive incidence is very roughly one percent of persons born. Some examples are polydactyly (extra fingers and toes), achondroplasia (short- limbed dwarfism), Huntington,s chorea (pro- gressive involuntary movements and mental deterioration), one type of muscular dystrophy, several kinds of anemia, and retinoblastoma (an eye cancer). A mutation of this sort pro- duces its effect in the first generation after its occurrence. In contrast, recessive mutations, which re- quire that the gene be present in duplicate in order to produce the trait, may not be ex- pressed for many generations. The trait will appear only when two mutant genes are inher- ited, one from each of the two parents. Such individually mild effects may be collectively the cause of considerable human misery. However, this may not occur for a very long time. Indeed, the gene may be lost purely by chance in the Mendelian lottery, although this is balanced on the average by those mutants that increase in number by the same process. More important, 46

there is good reason to think from animal exper- iments and from fragmentary human evidence that mutant genes are often lost from the popu-. lation because of mild dominant effects on via- bility and fertility when the gene is heterozy- gous. Thus, there is a good chance that the gene will be eliminated from the population before it ever encounters another like itself. McKusick lists 365 recessive diseases, plus 418 that are less certain. Some examples are phenylketonuria (or PKU, a form of mental deficiency), Tay Sach,s disease (blindness and death in the first few years of life), sickle cell anemia and cystic fibrosis. These are fairly common and well known, but most recessive conditions listed in the book are very rare. Recessive mutations located on the X chro- mosome are characterized by being expressed almost exclusively in males. Well known exam- ples are hemophilia (failure of blood clotting), color blindness, and a severe form of muscular dystrophy. McKusick lists 86 well established and 64 probable conditions of this sort. Be- cause the gene can be expressed in a single dose in males, which have only a single X chromo- some, X-chromosome-linked recessive muta- tions are somewhat like dominant mutations on other chromosomes in that they are expressed soon after occurrence instead of being spread out over an extended time span. Some of these dominant and recessive genes cause traits that we regard as normal, such as hair and eye color and blood groups. Others are not normal, but are so mild as to cause little concern. The great majority, however, cause diseases ranging from relatively mild to severe or even lethal. Most are so rare that they are known only to specialists. But, collectively, they are numerous enough that more than one percent of all children born will have a simply inherited disease causing an appreciable handi- cap. Another type of easily classified genetic damage is due to chromosome aberrations. Errors in chromosome distribution can lead to an individual whose cells contain too many or too few chromosomes. The well known disease mongolism is caused by an extra representa- tive of a specific chromosome (number 21). Most of the time, however, having too many or too few chromosomes leads to embryonic death; sometimes this is detected as a miscarriage, more often the death is so early as not to be detected at all. This kind of chromosome error is not thought to be strongly influenced by ra- diation, particularly at low doses. Another source of chromosome imbalance is chromosome breakage. This is less frequent than the type of distribution error mentioned above among spontaneous instances of severe human anomalies. But ionizing radiation is much more effective at breaking chromosomes than in causing errors in chromosome distribu- tion. The broken chromosomes may then reat- tach in various ways leading to rearranged gene orders, or they may be lost. The most fre- quently observed effect is a translocation—the exchange of parts between two (or more) chro- mosomes. Such a rearrangement is not harmful as long as both rearranged chromosomes, which among them have a normal gene con- tent, are present. However, children of a person with such a "balanced" translocation often receive only one of the two rearranged chromo- somes and their cells are therefore genetically unbalanced. The nature and the extent of the abnormality depends on the particular chromo- some regions that are deficient or duplicated, and on the magnitude of the imbalance. The harm ranges from rather mild to very severe, even lethal. Typically, chromosome imbalance —if it does not produce embryonic death—leads to physical abnormalities, usually accompa- nied by mental deficiency. What is most severe in one sense may not be the most tragic from the standpoint of human welfare. A chromosome aberration that causes early embryonic death may cause very little trauma, whereas the "milder" effect that per- mits the embryo to develop into a viable infant that is malformed and mentally retarded may be far more traumatic by any realistic measure of human suffering, both of the child and of his family. Among translocations that are found among normal humans, the most common by far are Robertsonian translocations. These are fusions of two chromosomes, each having a spindle attachment at the end of the chromo- some, to produce a single chromosome having the spindle attachment in the center. Such translocations have a population frequency of about 8 per 10,000. Usually the children are normal, since they inherit either the translo- cated pair or a pair of normal chromosomes. 47

However, they occasionally give rise to unbal- anced gametes leading to embryonic death or to congenital anomalies. Radiation does not ap- pear to be a major cause of these. Radiation- induced translocations are overwhelmingly of the reciprocal exchange type described in the paragraph above. In addition to those abnormalities and dis- eases that are caused by mutation of a single gene or by chromosome breakage, there are other diseases to which gene variation un- doubtedly contributes but where the inherit- ance is more complex. There is abundant evi- dence that there are inherited predispositions for many common conditions—for example, diabetes, schizophrenia, cancer, and mental retardation. It is hard to assess the magnitude of the ge- netic component and it is even harder to assess what we want to know in the context of this report—the extent to which the disease inci- dence depends on the mutation rate. But, be- cause this may well be the major way in which an increased mutation rate would exert a harm- ful effect, we shall use it as one basis for as- sessing radiation risks. There is an additional class of mutation whose importance we don,t know how to assess —those whose effects are so mild that they are not detected individually. As mentioned before, it is known in Drosophila that the most fre- quent of all mutations belong to a group that causes effects so mild that they can only be de- tected statistically in experiments involving large numbers. For example, a mutation might cause a one-percent reduction in the probabili- ty of surviving from the egg to the adult stage. Such a mutation is clearly impossible to detect in man, and very few mouse experiments are of a size to reveal it. We don,t know what the oth- er manifestations of such a mutant would be. (We cannot ask a Drosophila if it has a head- ache.) Perhaps the human counterparts of these mutations, in addition to causing a slight reduction in life expectancy, are responsible for greater susceptibility to disease, impaired physical or mental vigor, or a slight malforma- tion of some organ. We cannot ignore such mild mutations as unimportant, because (1) if Drosophila is any indication, they are by far the most frequent class of mutations; and (2) being mild, with less effect on viability and fertility, they are more likely to be transmitted to future generations and continue to have their effect over a longer time, thereby affecting more persons. Thus, their impact is multiplied by the number of generations through which they persist; and taken over the whole period, and in conjunction with other mutants, their effect may be far from negligible (See Explanatory Note 8). Despite a concern for this effect, we shall not attempt to estimate it quantitatively, for rea- sons to be discussed below. It is worth noting again, however, that in Drosophila the evi- dence is now good that this class of mutation is relatively less frequent among radiation in- duced mutations than among spontaneous mutations. The contrast between genetic and somatic concerns is striking. The low-dose somatic ef- fects that are most feared are cancer and leu- kemia. The evidence that high radiation doses have these effects is unequivocal. The evidence for low doses is less clear. For genetic effects of radiation, we have no direct evidence of human effects, even at high doses. Nevertheless, the animal evidence is so overwhelming that we have no doubt that humans are affected in much the same way. In contrast to somatic effects, where the concern is concentrated mainly on malignant disease, the genetic ef- fects are on all kinds of conditions—for the spectrum of radiation-caused genetic disease is almost as wide as the spectrum from all other causes. IV. Could an Increased Mutation Rate Possi- bly be Beneficial? So far, we have been assuming that any in- crease in the mutation rate would be harmful to future generations. On the other hand, the theory of evolution assumes that mutations are the raw material on which evolutionary progress depends. The question is sometimes raised as to whether an increased mutation rate might be a good thing, increasing our evo- lutionary potential in a time of rapidly chang- ing environment. There are several answers. One is that we don,t know what the optimum mutation rate for evolution is; probably there is no universal and simple answer. But, in any case, evolutionary theorists believe that in sexually reproducing 48

species the rate of evolution is hardly ever, if ever, limited by the mutation rate. The differ- ence between rapidly evolving and slowly evolving species is far more easily explained on the basis of such factors as availability of eco- logical opportunity and stability of the envi- ronment than by differences in mutation rate. A second reason is purely empirical. In every species studied by geneticists, the overwhelm- ing majority of mutations that have effects large enough to be readily observed are delete- rious. The most conspicuous examples of benefi- cial mutations have been those that are discov- ered only in a drastically altered environment, such as DDT-resistant mutants in insects (which are beneficial from the insect,s point of view). There are a large number of mutants whose effects are very slight and which are or- dinarily not observed and studied; many of these have been revealed recently by sensitive chemical techniques. Among such mutants, there may be some whose effects on any body function are so slight as to produce no effect at all on the individual,s well being. However, the existence of such possibly neutral mutations does not alter the general conclusion that, whenever there is an appreciable effect on the organism, the effect is almost always harmful. Natural selection preserves the rare beneficial mutants while eliminating the great majority of misfits. We believe that a genetically diverse popula- tion is more to be desired than a uniform one, and this might be regarded as an argument for a high mutation rate. But the amount of genet- ic variability existing in the population is far greater than that which arises by mutation in a single generation. Furthermore, in some poly- morphisms such as blood groups, hemoglobins, and serum proteins the entire variability may have arisen from a few mutant genes. If human mutation were to stop entirely, we should prob- ably not notice any effect at all for many gener- ations, except for some reduction in the inci- dence of severe dominant aberrations, among which are some of the most distressing diseas- es. The mutant genes now in the population arose in the past and have been pre-tested to some extent, the worst ones having been elimi- nated by natural selection. What we are saying is that there is ample genetic variability in the population for any evolutionary progress that is likely to occur in the foreseeable future. In- deed, some geneticists argue that for a long time to come the closer we can come to a muta- tion rate of zero, the better off we will be. Whether this is correct or not (and in any case lowering the spontaneous mutation rate is not now possible) the Subcommittee is convinced that any increase in the mutation rate will be harmful to future generations. V. Sources of Genetically Significant Radia- tion Exposure The sources of population exposure are treated in detail in Chapter III of this report. The main features are repeated in Table 1 for convenience of reference, together with the current Radiation Protection Guides for aver- age population exposure. The genetically significant dose (GSD) is an attempt to estimate the exposure that is rele- vant to mutation production. The gonad dose at each age and sex is weighted by the expected number of future children for a person of that age and sex. This is the procedure used in esti- mating the GSD from medical and dental radia- tion. For natural radiation, the GSD is as- sumed to be the same as the Gonad dose, since exposure is uniformly distributed over all ages. It is somewhat less than the whole body radia- tion because of shielding of the gonad by other body tissues. For fallout, occupational expo- sure, and nuclear power we have not attempted to convert these into gonad doses, since the amounts are so small. The occupational expo- sure is obtained by considering the total radia- tion received by those occupationally exposed, and treating this total as if it were uniformly distributed over the whole population. The genetically significant dose is probably con- siderably less than the 0.8 mrem in the table because of the age distribution of those who are occupationally exposed. The estimates of exposure from nuclear pow- er developments in the year 2000 are based on an assumed increase from 6000 megawatts in 1970 to 800,000 in 2000, along with correspond- ing increases in mining and fuel reprocessing. As mentioned in Chapter III, there is the fur- ther assumption that the radiation level at the site boundary is 5 mrem per year per reactor. It is important to emphasize, however, that no 489-797 O - 72 - 5 49

Table 1. Sources of genetically significant radiation. Estimated average amounts taken from tables in Chapter III. mrem/year Source Whole body Genetically significant exposure Natural radiation exposure Cosmic radiation Radionuclides in the body External gamma radiation 44 18 40 Total 102~ 90 Man-made radiation Medical and dental Fallout Occupational exposure Nuclear power (1970) Nuclear power (2000) 73 4 0.8 0.003 <1 30-60 Radiation Protection Guide for man-made radiation (medical excluded to the general population (for reference) 170 allowance has been made for failure to meet the expected levels of performance, nor for acci- dents or sabotage. In the United States the genetically signifi- cant exposure from diagnostic radiation in 1964 was estimated to be 55 mrem per year. By 1970, the exposure was reduced to an estimated 36 mrem per year, of which about 2/3 is in males and 1/3 in females. The genetically significant exposure from therapeutic radiation is much less, being about 5 mrem per year. That from dental radiation is still less and may be con- sidered negligible in comparison with diagnos- tic radiation. We note that, despite recent reductions, medi- cal radiation continues to be the largest man- made source of gonadal radiation. We believe that it is feasible to lower this exposure still more by such things as improved diagnostic equipment, image amplification, attention to gonadal shielding, rigorous adherence to oper- ational standards, elimination of all medically unwarranted exposure, and better training of personnel. It also appears to be technologically feasible to develop nuclear power, at least for the near future, with a genetic exposure that is a very small fraction of the natural background, and less than one percent of the present radiation protection guides. Table 1 represents only average values for the population. There may be considerable var- iation about these averages. From the stand- point of estimating the overall genetic impact of radiation on future generations, the popula- tion average is the most important figure. 50

VI. Risks versus Benefits It is not a part of the Subcommittee,s as- signed task to balance benefits against risks. Nonetheless, we should like to make some gen- eral remarks. It is clear at the outset that an assessment of risk is useful for arriving at a rational decision only if it is compared to the benefits, or to the difficulty of reducing the risk. It goes without saying that if there were no benefits there • would be no excuse for taking any avoidable risk whatsoever. Furthermore, if the risk can be decreased at an acceptable increase in cost, this should be done even though the benefit may heavily outweigh the risk. The risk estimate is useful for answering three related questions: (1) Do economic and social benefits associat- ed with the radiation outweigh the genetic cost of the radiation? (2) Do alternatives to the use of radiation also involve a risk, and if so is this risk greater or less than that of radiation? (3) Are the costs of exposure reduction too great a price to pay for the reduced genetic risk? We are fully aware that both the costs and benefits are difficult to measure with any preci- sion and often they are expressed in units that are incommensurable. Yet even crude and un- certain estimates can often lead to a more ra- tional policy than would be possible if no such assessments were available. It is with this phi- losophy that we proceed to discuss risk esti- mates. VII. Methods of Estimating the Genetic Risks from Radiation The task of the Committee is to: "(1) Review the scientific bases for the evaluation of risks at low levels of radiation exposures; (2) select the scientific basis it recommends the Environ- mental Protection Agency (EPA) to use; (3) make such estimates of risks as it deems scientifically appropriate; and (4) delineate the interpretations that can be attributed to the estimates." From the earlier sections it is clear that, al- though we are beginning to get some informa- tion from direct human studies, we must still rely mainly on experimental animals—in par- ticular the mouse—for quantitative data. We recommend the following general principles for risk estimation: 1. Use relevant data from all sources, but emphasize human data when feasible. In gener- al, when data of comparable accuracy exist, place greater emphasis on organisms closest to man. 2. Use data from the lowest doses and dose- rates for which reliable data exist, as being more relevant to the usual conditions of human exposure. 3. Use simple linear interpolation between the lowest reliable dose data and the sponta- neous or zero dose rate. In order to get any kind of precision from experiments of manageable size, it is necessary to use dosages much higher than are expected for the human population. Some mathematical assumption is necessary and the linear model, if not always correct, is likely to err on the safe side, (see Explanatory Note 10) 4. If cell stages differ in sensitivity, weight the data in accordance with the duration of the stage. 5. If the sexes differ in sensitivity, use the unweighted average of data for the two sexes. The Subcommittee has considered various ways of estimating the genetic risk. Some, which were considered and rejected, are dis- cussed later. The four ways that we have used are related to the kinds of diseases discussed in Section III. They are: 1. The risk relative to natural background radiation. 2. The risks for specific genetic conditions. 3. The risk for severe malformation and dis- ease. 4. The risk in terms of over-all ill health. Our view is that there are so many facets to the problem that several ways of estimating the risk are more useful than any single one in arriving at the best policy decisions. VIII. Risk Estimates We now present risk estimates made in the four ways listed above: A. The Risk Relative to that from Natural Background Radiation This is not really a risk estimate at all, but may nevertheless be useful as a policy guide. 51

As mentioned earlier, the natural level of radiation averages about 100 mrem per year. This varies considerably from one region to another, depending especially on the kinds of minerals present in the earth and on the alti- tude. A person who lives in a stone house may get more radiation than one who lives in a wooden house, because of the greater radioac- tivity of some rocks, such as granite. Likewise, a person who lives at a high altitude receives more radiation from cosmic rays. Exposure to man-made radiation near the level of back- ground radiation will produce additional ef- fects of a magnitude comparable to what man has experienced from this source throughout his entire history. Furthermore, since man- made radiations are not qualitatively different from natural radiation, they will not produce novel effects. These are particularly firm con- clusions because they do not require any quan- titative genetic information. Another way of stating this is to note that the annual difference in natural radiation be- tween a location in Louisiana and one in Colo- rado might be 100 mrem or more. Even a person who knows this probably doesn,t take this dif- ference into account in deciding to change his residence. We can regard man-made radiation levels of this magnitude as comparable to other risks that are often accepted. The idea of using the background radiation level as a yardstick for setting standards is not new. The BEAR Committee had this in mind as one consideration in formulating its recommen- dations. It was specifically used by the Ad Hoc Committee of the National Committee on Ra- diation Protection and Measurements (NCRP) in its recommendations for limitations on the somatic radiation dose for the general popula- tion (2). The Committee recommended that "the population permissible dose for man-made ra- diation, excluding medical and dental sources, should not be larger than that due to natural background radiation, without a careful exami- nation of the reasons for, and the expected ben- efits to society from, the larger dose." To summarize: Our first recommendation is that the natural background radiation be used as a standard for comparison. If the genetical- ly significant exposure is kept well below this amount, we are assured that the additional consequences will neither differ in kind from those which ve have experienced throughout human history ii ^r exceed them in quantity. B. Risk Estimates for Specific Genetic Con- ditions It seems most meaningful to compare the current incidence of specific genetic conditions to the expected increase from radiation. To do this, we first consider a convenient yardstick, the relative mutation risk per rem. This is the fraction by which the mutation rate would be increased by one rem of radiation. The recipro- cal of this is the mutation rate-doubling dose, the dose required to produce as many muta- tions as occur naturally. We estimate these quantitites as follows: In the absence of human data on radiation effects,, we use data from the mouse. A chronic radia- tion dose to mouse spermatogonia produces about 0.5 x 10-7 recessive mutations per rem per gene. The reproductive cells of the female, for most of their lifetime, are very much less mutable than those in the male, even from acute irradiation. Furthermore, the germ cell stages in the female that have a high mutation- al sensitivity to acute irradiation, namely, the mature oocytes, give a very low mutation rate with chronic irradiation. Therefore, we^take the average of the two sexes as 0.25 x 10~7 (See Explanatory Note 11). This may be too high, since the gene loci on which these studies were made were to some extent preselected for muta- bility; they would not be included in the study if they had not mutated at least once in the strains studied. Another reason for thinking that this may be too high is that in mice the rate of induction of dominant visible mutations is lower than that for recessives by at least an order of magnitude. Dominant mutations con- stitute a substantial part of the human genetic risk. For the spontaneous rate, human data are available. The rates are variable. Many genes have mutation rates in the vicinity of 10-5 per gene per generation, but there is reason to be- lieve that these are a nonrandom sample on the high side. We suspect that the true average rate is an order of magnitude less (See Explan- atory Note 12). If the spontaneous value is tak- en as 0.5 x 10-6 this is 20 times the induced rate per rem in the mouse (.25 x 10-"). If the average spontaneous rate is 0.5 x 10-5, the ratio is 200. So we estimate that the increase in relative mutation risk per rem is between 1/200 and 1/20, or that the doubling dose is between 20 52

and 200 rem. If we consider the mutation rate for dominant visible mutations—estimated with considerable uncertainty as about 2 x 10-9 per rem—the doubling dose is 100 rem or more un- less the spontaneous rate is less than 2 x 10-7- The extensive studies in Hiroshima and Na- gasaki permit a direct approach to the dou- bling dose based entirely on human data (see Explanatory Note 6). The death rates in the children of irradiated and control parents did not differ significantly. If the true death rate were one percent higher in the children of irra- diated parents, the probability of the observed results in a population as large as this would be less than 0.10. Assume that these parents re- ceived a total average dose of 100 rem each. It was estimated that 0.5 percent of liveborn die prior to age 8 as a result of a dominant muta- tion or chromosome aberration that occurred in the previous generation (41). So we can say that 100 rem, at the most, produced an effect equal to twice the dominant spontaneous rate; the amount required to equal the spontaneous rate for dominant deleterious effects (the "dou- bling dose") is therefore at least 50 rem. Since this represents acute exposure, whereas we are here concerned with chronic exposure, we esti- mate that the doubling dose is at least 3 times as large, or at least 150 rem. The 100 rem aver- age dose for the Japanese parents may be too high; if this were only 50 rem then the doubling dose becomes 75 rem. There are many uncer- tainties in these calculations, but they do offer strong evidence against the doubling dose for mutation being as low as 20 rem of chronic ir- radiation. The lowest possible value for the doubling dose is about 3 rem, for this is the amount of radiation received from natural sources in 30 years. Such a doubling dose would imply that all spontaneous mutations are caused by natu- ral radiation. Although this possibility cannot be completely ruled out, the evidence is strongly against it. In addition to the estimates given above, there is abundant evidence in experi- mental organisms for causes of mutation other than radiation. It seems unlikely in the extreme that man differs from all other species in being insensitive to all other causes of spontaneous mutation. We shall then use a range of 20 to 200 rem for the doubling dose. For reasons given, we doubt that it is below 20 except possibly for some spe- cial categories that do not make a large contri- bution to the total effect. If the true value is greater than 200, then any harm would arise only from being too cautious. As mentioned earlier, the original BEAR Committee used 5 to 150 for its limits; our suggested limits are somewhat higher, reflecting the new informa- tion mentioned before. Calculations based on these assumptions are given in Table 2, which gives the estimated in- crease to be expected among one million live- born individuals whose ancestors had received 0.17 mrem per year (or 5 rem per 30-year re- productive generation). If this amount of expo- sure is continued until an equilibrium is reached, the number of individuals affected with autosomal dominant traits is expected to increase to 10,250-12,500 per million live births, in contrast to the estimated present in- cidence of 10,000. The calculations are ex- plained in Note 13. The incidence figures are based mainly on an extensive survey in North- ern Ireland (3). We have accepted the judgment of the United Nations Report in determining which traits to include. For autosomal domi- nant traits, there is good reason to think that the equilibrium frequency is proportional to the mutation rate. The assumption is almost as good for X-chromosome-linked recessive traits. The population incidence in Ireland was only 4/5 the incidence among newborn. Therefore, we have assumed that expression in the first generation after radiation is one-fifth the equi- librium value; this is roughly equivalent to assuming that the average mutant persists in the population for 5 generations. The contribu- tion of recessive genes, we believe, is negligible in comparison. For one thing, known recessive diseases are somewhat less common than the dominant. But much more important, the inci- dence of recessive genes depends strongly on the way selection acts on the heterozygous car- riers. A small difference in this can outweigh a large mutational difference. This is especially true if the mutant is favored in the heterozy- gous state. The incidence is not likely to in- crease in proportion to the mutation rate. Fi- nally, whatever the equilibrium is, it is at- tained very slowly, so any effect of an in- creased mutation rate on the incidence reces- sive traits would be spread over hundreds of generations. 53

We can get some support for the numerical values in Table 2 by a different calculation. This again uses mouse data for the induced rate, since there are no suitable human data. But the rest of the calculation involves no as- sumption about either the doubling dose or the normal incidence. McKusick,s tabulation of known Mendelian traits in man lists 415 with dominant inheritance well established and 528 more where the evidence is less complete, a to- tal of very roughly 1000. Assuming the muta- tion rate in the mouse (0.25 x 10-? per rem) and 1000 mutable loci (103), we compute for one mil- lion births (106) with a parental exposure of 5 rem a number of new mutants equal to 0.25 x l0-7 x 103 x 106 x 5 x 2, or 250. The final factor of 2 comes from the fact that both parents are irradiated. This value is in the range of first generation values in Table 2 for autosomal dominants. It should be emphasized that this includes only known conditions, not those to be discovered in the future. The mouse mutation rates that serve as the basis for these estimates are all for recessive mutations. The rates for dominant mutations are known much less reliably, but the evidence is that they may be an order of magnitude low- er (4). If this is correct, the values in Table 2 are all 10 times as high as they should be. We prefer, however, to err on the safe side, so we have used the recessive rates. In addition to the mutation traits summa- rized in Table 2, there is another class of genetic change—that is, the damage caused by chromo- some aberrations, both structural and numeri- cal. There has been great progress in this area recently. Estimates of cytogenetic effects are given in Table 3. The background for the calculations is given in Explanatory Notes 14 and 15. The current incidence data are based on sur- veys of newborn children and on studies of aborted fetuses. The estimated effects from radiation are based solely on information de- rived from mice. We have assumed, as with oth- er genetic effects, that at low doses the effect is proportional to the dose. The major contribution to postnatal anoma- lies is aneuploidy, of which the best known example is mongolism. Since reproduction in this group is virtually zero, the entire incidence must be caused by new occurrences of the chro- mosome error. The evidence is that this kind of event is not very sensitive to radiation. Some Drosophila data suggest a threshold, but there is good evidence that at least some of the effect has linear relationship to dose (See Explanato- ry Note 16). There have been reports that irra- diation of women before conception causes aneuploidy (mongolism), but other and larger studies have failed to confirm this (See Explan- atory Note 17). The human data provide no suitable basis for a quantitative assessment and the Drosophila data are based on high dos- es and there is room for considerable doubt about its human relevance. We, therefore, use Table 2. Estimated effects of radiation for specific genetic damage. The range of estimates is based on doubling doses of 20 and 200 rem. The values given are the expected numbers per million live births. Number Effect of 5 rem Current incidence that are per generation per million live new First Equili- births mutants generation brium Autosomal dominant traits 10,000 2,000 50-500 250-2,500 X-chromosome-linked traits 400 65 0-15 10-100 Recessive traits 1,500 ? very few very slow increase 54

Table 3. Estimates of cytogenetic effects from 5 rem per generation. Values are based on a population of one million live births. Unbalanced rearrangements are based on male radiation only. Current incidence Effect of 5 rem per generation First generation Equilibrium Congenital anomalies Unbalanced rearrangements Aneuploidy Recognized abortions Aneuploidy and polyploidy XO Unbalanced rearrangements 1,000 4,000 35,000 9,000 11,000 60 5i 56 15 360 75 5 56 16 450 iSee footnote 1. mouse data as the basis for the estimated aneu- ploidy and translocation induction.1 It is probable, as Table 3 indicates, that ra- diation could make some contribution to spon- taneous abortions. Although this is certainly not negligible as a source of human distress, it is of much less concern than congenital anoma- lies of the live-born. There is probably a much larger class of genetic damage that results in failure of the egg to implant or in post-implan- tation death—that is, too early to be detected. Since this does not have an appreciable effect on human well-being, these mutations have not been included in the calculations. C. Risk Relative to the Current Incidence of Serious Disabilities Using 0.005 to 0.05 as the range of values for the relative mutation risk for one rem (or a doubling dose of 20-200 rem) the continuous exposure of 5 rem per 30 years (or about 170 iWe are aware that a new study in progress in Hiroshima and Nagasaki is finding increases in sex chromosome triso- my in the progeny of the irradiated population. We are as yet unaware whether this will require revision in our risk estimates, which must await analysis of the data. mrem per year) would eventually cause an increase of from 2.5 to 25 percent in the burden of mutation-caused disease. An unknown, but probably large fraction of human disease is genetically related—at least in the sense that susceptibility depends partly on genetic factors. A detailed survey in North- ern Ireland (3) led to the estimate that just over 25 percent of hospital beds and 6 to 8 percent of physicians, time are used by persons with he- reditary disease. But we cannot conclude from this that the incidence of such diseases will rise in direct proportion to the mutation rate. The relation between mutation rate and the inci- dence of complexly-inherited disease is hardly known at all and we shall have to make arbi- trary assumptions. Our disease classification follows that of the United Nations Committee. About one percent of children born have a disease for which there is evidence of dominant or X-linked inheritance, as we have indicated before. Undoubtedly, some of these are inherit- ed in a more complex way, but these are proba- bly balanced by dominant diseases that are not recognized as such. So we shall take one-per- cent as the figure in this first category. The in- cidence of these is essentially proportional to the mutation rate. Recessive diseases, the second category, are less frequent and their incidence is only very indirectly related to the mutation rate. Their 55

total incidence is less than 0.5 percent. Serious diseases caused by chromosome aberrations also have an incidence of about one-half per- cent, but these are not thought to be greatly increased by low level radiation (See the pre- vious section). Diseases of more complex etiology fall main- ly into two groups. One group, our third catego- ry, with an overall incidence of abut 2.5 percent of births, consists of malformations. About 1.5 percent are recognizable at birth, the remain- ing 1 percent developing later. The fourth cate- gory contains a mixture of constitutional and degenerative diseases. This figure is taken to be 1.5 percent, but is quite arbitrary, depend- ing on what diseases are included. Anemia, dia- betes, schizophrenia and epilepsy, for example, are included. Heart disease, ulcer, and cancer have not been included, although there is known to be a genetic component in each. We then have a total of about 6 percent of children born who have diseases in these cate- gories. The first category, dominant or X-chro- mosome-linked diseases, is assumed to increase in proportion to the mutation rate, which means that the eventual incidence will be in- creased by the dose times the increase in rela- tive risk per rem which we have taken to be between .005 and .05. The first generation effect we again take as 20 percent of this. The second category, chromosomal and reces- sive diseases, will have very little increase re- lative to dominant diseases. The chromosomal effects are not very much increased by low level radiation. Those diseases caused by recessive genes will eventually increase but the amount is uncertain. Furthermore, it would require scores of generations in the future for the full effect to be manifest. The last two categories present a more diffi- cult problem. The extent to which the incidence of these diseases depends on mutation is not known. We shall define the "mutational compo- nent" of a disease as the proportion of its inci- dence that is directly proportional to the muta- tion rate. It is not likely to be more than half for diseases considered in these categories. Some would estimate it as low as 5 percent. The values in Table 4 are based on these limits. To estimate the first generation effects in the last three categories of diseases, we note that the magnitude of the individual gene effects presumably causing these diseases is likely to be less than that of the single gene dominant mutations; therefore, we would expect a smaller fraction of the total impact to be in the first generation. We shall arbitrarily use half the 20 percent value used for dominant diseases, or 10 percent as the fraction of the equilibrium value that is expressed in the first generation, al- though the true value may well be less. The Subcommittee is aware that this classifi- cation of diseases, based on the U. N. Reports, which in turn were based on surveys in North- ern Ireland, is not very detailed and may not be entirely relevant for the United States. It would be possible to get a more detailed specification of the kinds of diseases and their relative incidences. Our reason for not being more specific in this regard is that the other uncertainties are so great. For most diseases, even if we knew the incidence with great accu- racy, we could not specify what the mutational component of this is. Until we have quantita- tive information about radiation-induced mu- tation in man and of the role of mutation in maintaining disease incidence, we think it is pointless to further refine our discussion of specific diseases. D. The Risk in Terms of Il1 Health There is danger that the previous sections, by concentrating only on fairly well defined genetically-associated diseases, have dealt with only the exposed part of the iceberg. What about the rest of human illness? It, too, has some degree of genetic determination. The most tangible measure of total genetic damage, in terms that are meaningful and im- portant to human welfare, is probably poor physical and mental health. Although we can- not measure the personal distress that this causes, we can measure morbidity in economic units such as days lost from work or medical expenses. It seems likely that the mutational compo- nent of this unspecified remainder is less than for the categories of Table 4, where the in- crease that would eventually result from a doubling of the mutation rate was taken to be roughly in the range of 1/4 to 1/2. We assume that the quantitative average is effectively mimicked by a model in which a certain fraction of the incidence has simple linear relationship 56

Table 4. Estimated effect of 5 rem per generation on a population of one million. This includes conditions for which there is some evidence of a genetic component. This table includes values from tables 2 and 3. Disease classification Current incidence Effect of 5 rem per generation First generation Equilibrium Dominant diseases Chromosomal and recessive diseases Congenital anomalies Anomalies expressed later Constitutional and degenerative diseases 10,000 10,000 15,000 10,000 15,000 50-500 Relatively slight 5-500 250-2500 Very slow increase 50-5000 Total 60,000 60-1000 300-7500 to dose and the rest is not assignable to radia- tion at all. This fraction was taken as 1/4 to 1/2. For unspecified ill-health, it is probably less, for dominant genes are thought to play a lesser role, and we shall assume a lower value of 1/5. It may well be less, but few would argue that it is much higher, so we take this as a pru- dent assumption. Using this value and again taking 20 rem as the lower limit of the mutation rate doubling dose, an exposure of 5 rem per generation would increase the equilibrium ill-health inci- dence by 5/20 x 1/5 or 5 percent of the present value. With 200 rem as the doubling dose, this would be 0.5 percent. If desired, this can be converted into eco- nomic terms. For one way, see Note 18. IX. Discussion A major concern of the Subcommittee is the possible existence of a class of radiation-in- duced genetic damage that has been left out of the estimates. By relying so heavily on experi- mental data in the mouse we may have over- looked important effects that are not readily detected in mice, or the mouse may not be a proper laboratory model for the study of man. Another source of concern is whether the low mutational response to radiation in the female mouse is applicable to the human female. The concern arises over the fact that, for most of their lifetime, the reproductive cells in the fe- male mouse are highly sensitive to cell killing, whereas those in the human are not. This raises the question of whether the low mutational sensitivity of these mouse cells can be assumed to apply to the human cells. However, there is, in general, in the mouse no clear correlation, ei- their negative or positive, between mutational sensitivity and cell killing of the type involved here, in which death occurs before cell division. Furthermore, the germ cells in the female mouse also go through a period when, like the human cells, they are very resistant to killing, and here too the mutational sensitivity, al- though high with acute irradiation, is very low with chronic irradiation—much lower than that in the male. Thus, although we have some concern about applicability of mouse female data to the human female, there are, so far, no data from any of the various germ cell stages studied in the female mouse that would indicate anything but a very low mutational response, relative to that in the male, under the usual conditions (low dose and low dose rate) of hu- man radiation exposure. If our estimates of risk are too high, the only dangers are those which might derive from ex- cessive caution, such as regulations which 57

might permit greater hazards from other sources to replace the overestimated radiation hazard thus avoided. A more serious error would be to underestimate the effect. It is im- possible to prove absolutely that the doubling dose is not as low as the background radiation level, namely about 3 rem. But all the informa- tion from other organisms and the few data that exist for man (mainly from cell cultures and from humans that have been irradiated, especially the Japanese populations) suggest that this value is too low and that our lower limit of 20 rem for chronic radiation is very unlikely to be too high. We have assumed that data from low doses and low dose-rates in mice are appropriate for the kinds of radiation to which most of the human population will be exposed. The major exceptions are therapeutic radiation, which constitutes very little of the genetically signif- icant dose, and accidents, about which we are in no position to make any specific assumptions. Perhaps the major reservation that we have about our estimates is their failure to take adequately into account mutations that have very mild effects. As mentioned earlier, this is the most frequent class of mutations in Droso- phila and because they persist longer in the population than those with more drastic ef- fects, each mutant gene affects a correspond- ingly larger number of persons. The empirical experiments on mice argue that such genetic mutations are not making any substantial impact on mouse populations for up to 45 gen- erations of continuous radiation—far longer than we are able to consider in any meaningful way for the human population. Yet there is the possibility that one simply does not see in mice effects that would cause appreciable distress in humans. One way to approach this problem is to use a method that was urged by the late H. J. Muller, who discovered that radiation has genetic ef- fects. As mentioned in the introduction, this was one approach used by the original BEAR Genetics Committee. Since each mutant gene must eventually be eliminated from the popula- tion, one can simply measure the total number of mutations produced in a generation; this number is then the number of eventual gene extinctions or "genetic deaths" if the popula- tion size remains stable. (If the population grows, the number of gene extinctions increas- es in proportion.) There is no information given by this calculation about the time distribution of these extinctions. But it must be said that this is one approach that at least attempts to measure the total impact of mutation, integrat- ed over all future generations affected by these mutations. Despite the relative simplicity of this calcu- lation, we have not recommended it as a basis for estimation of the genetic risk. The main reason is the impossibility of equating statisti- cal gene extinctions to any meaningful measure of human misery. A "genetic death" may be the death of an embryo so early that no one ever knows about it, or it may simply be the failure to reproduce. On the other hand, it may be a lingering, painful death in early adult life that causes great distress to the person and his en- tire family. Also it is not known that mutant genes are always eliminated independently, which the calculation assumes. Furthermore, the calculation depends on the gene number, which for man can only be guessed. Finally, to equate genetic deaths to actual human death, as some have done, gives a quite erroneous pic- ture of the impact of mutation on the popula- tion. We remind all who may use our estimates as a basis for policy decisions that these estimates are an attempt to take into account only known tangible effects of radiation, and that there may well be intangible effects in addition whose cumulative impact may be appreciable, al- though not novel. X. Summary and Conclusions We have reviewed the recommendations and risk estimates of the 1956 National Academy of Sciences Committee on the Biological Effects of Atomic Radiation (BEAR) and believe that, if anything, the risks estimated at that time were on the high side. The main reasons for this are the discoveries that radiation at low dose-rates is considerably less effective than the same dose at a faster rate and that the female mouse is for much of her lifetime very resistant to radiation-induced mutation. Another reason is the failure of mice whose ancestors have been irradiated heavily for many generations (45 in

one case) to show measurable effects on viabili- ty or fertility. We recommend that calculations for low dos- es be made by assuming that the relationship between the lowest accurate measurements and zero induced effect at zero dose is linear. The assumption is plausible for mutations and chromosome breaks; for other effects, such as non-disjunction, which may depend mainly on other mechanisms, it is a conservative proce- dure in which any error is likely to be on the safe side. We also recommend that, when esti- mates are made from experimental animals, these be based on chronic or fractionated doses as being more relevant than large, acute doses to the typical conditions of radiation exposure to the human population. We take the risk of chronic radiation at low doses, relative to the spontaneous mutation rate, to be 0.005 to 0.05 per rem. This relative risk is equivalent to saying that the amount of radiation required to produce as many muta- tions as occur spontaneously in a single gener- ation (the doubling dose) is between 20 and 200 rem. The information on the radiation-induced effect comes almost entirely from mouse data. The Subcommittee recommends four bases for assessment of the genetic risk. They are arranged in order of the confidence that we have in them. The first is very firm, the second less so, the third still less, and the fourth little more than an informed guess. (1) The risk relative to the natural back- ground radiation. If the genetically significant exposure is kept well below this amount, we are assured that the additional consequences will be less in quantity and no different in kind from what we have experienced throughout human history. This base, although not quantitative, has the great merit that it is not necessary to make any quantitative assumptions about human radiation genetics. (2) The risk of specific genetic conditions. Using the relative risk (or doubling dose) given above, an estimate of the increase in diseases caused by dominant and X-chromosome-linked recessive mutations can be made for the gener- ation following radiation and for the equilibri- um increase under continuous radiation. Esti- mates of cytogenetic effects can be made direct- ly from mouse data. Numerical values are giv- en in Tables 2 and 3. (3) The risk relative to the current incidence of serious disabilities. Diseases caused by dom- inant and by X-chromosome-linked recessive mutations will eventually increase in propor- tion to the mutation rate increase. For congeni- tal anomalies and constitutional diseases, we suggest that the mutational component (or the fraction of the incidence that is proportional to the mutation rate) is between 5 and 50 percent. Numerical values based on these assumptions are given in Table 4. (4) The risk in terms of overall ill health. The contribution of the mutational component to ill health is arbitrarily taken as 20 percent. With this and a doubling dose between 20 and 200 rem a dose of 5 rem per generation would even- tually lead to an increase of between 0.5 and 5.0 percent in all illness. It is clear that these estimates are subject to great uncertainty. The ranges of plausible values are broad, and there is no assurance that the true values are within these ranges. We are well aware that future information will necessitate revisions. The estimates are pre- sented, not as accurate scientific information (as scientists we would prefer to defer judg- ment until the information is solid), but as rea- sonable values based on current knowledge which, crude and uncertain as they are, may serve as a better guide to rational uses of ra- diation than no estimates at all. In cost-benefit calculations, the discrepancy between cost and benefit may be so great that even-such crude and uncertain estimates may be very useful. Whether a risk is acceptable also depends on how avoidable it is. If the ge- netic risk is easily reduced, it is unacceptable even if the cost-benefit ratio is low. It seems clear that the genetically significant radiation exposure from fallout, from nuclear power developments, and from occupational exposure (treated as a part of the over-all pop- ulation average) is now very small relative to that from natural radiation. There is no reason to think that the dose commitment for the de- velopment of nuclear power in the next few decades should be more than about a millirem annually. The 1956 report and the guides that grew out of it were the result of an effort to balance genetic risks against the needs of so- ciety. It now appears that these needs can be met with very much less than the 170 mrem per 59

year of the current Radiation Protection Guides. Accordingly, the 170 mrem seems to provide an unnecessarily large cushion. Likewise, we believe that the currently much higher level of radiation from medical sources (mainly diagnostic) should be examined in view of the same concept. If it can be reduced fur- ther without impairing essential medical serv- ices, then the present level is unnecessarily high. APPENDICES TO CHAPTER V Explanatory Notes Note 1. UNSCE AR Reports The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) has issued a series of reports which collectively constitute a wealth of information on this sub- ject (5-8). In general, throughout this report, we shall not further document conclusions that are in the UNSCEAR reports, but instead will simply refer to these reports. The bibliographies therein are very extensive and the reader is referred to them for more detailed information. We should like to take this opportunity to thank the United Nations Secretariat for supplying us with written material and for in- dividual consultation. Note 2. History of Radiation Standards The British report specifically stated: "Those responsible for authorizing the devel- opment and use of sources of ionizing radiation should be advised that the upper limit, which future knowledge may set to the total dose of extra radiation which may be received by the population as a whole, is not likely to be more than twice the dose which is already received from the natural background; the recommend- ed figure may indeed be appreciably less than this" (9). In January 1957, the NCRP recommended that the population dose "shall not exceed 14 million man-rems per million of population over the period from conception up to age 30 and one- third that amount in each decade thereafter." This was based on the exposure practices and data of that period and the contributions of the individual sources were estimated in man-rems per million population per 30 years as: Natural radiation 4,000,000 Medical irradiation 5,000,000 Occupational exposure 150,000 Radiation in plant environs 450,000 Fallout 200,000 Total 9,800,000 Balance 4,200,000 The radiation exposures included medical, natural, and fallout radiation and all other man-made sources and allowed a cushion of over 4 million man-rems for future needs. In April 1958, the concept of population dose for man-made radiation, exclusive of medical exposure, was made more specific in the state- ment: "The radiation. . .shall be such that it is improbable that any individual will receive a dose of more than 0.5 rem in any 1 year from external radiation." It was also recommended, as in 1957, that the average body burden of radionuclides not exceed 1/10 that for radiation workers. In September 1958, the International Com- mission on Radiological Protection (ICRP) suggested that "the genetic dose to the whole population from all sources, additional to the natural background, should not exceed 5 rems plus the lowest practicable contribution from medical exposure." Because the genetic dose is calculated for a 30-year period, this would amount to an average of 170 mrem per year. The same value of 170 mrem per year had been arrived at by a different route based on the 0.5 rems per year recommended by the NCRP for an individual in the general popula- tion. It was reasoned that to hold the dose to the individual to that level, the average level 60

for a population group would have to be approx- imately 1/3 of the maximum amount, or again 170 mrem per year. Based on the published rec- ommendations of the NCRP and ICRP, the pop- ulation average of 170 mrem was adopted by the Federal Radiation Council in 1960. The history of radiation protection stand- ards has recently been reviewed (1O). Other references are by NCRP (11, 12), the FRC (13), and NCRP (14). Note 3. The Number of Genes Actually, this calculation does not assume that the number of genes is known, but rather it depends on the ratio of the overall mutation rate to that for a single locus. The ratio of the total lethal rate to that for a single locus was multiplied by 2 to 3 to allow for mutations with less than lethal effects. This led to an estimated ratio of about 104, subject to considerable uncertainty both as to accuracy of measure- ment and reliability of assumptions. The con- clusion was reinforced by the fact that the number of bands in the salivary gland chromo- somes in Drosophila is about 5000. There is recent evidence (15-21) that the number of genes (complementation units) in Drosophila is indeed equal to the number of salivary chromo- some bands, which would be 5000 per gamete, or 10,000 in the diploid cell. The human number is probably larger, but there is no comparably reliable way to estimate it. We shall not use the gene number in any of our risk estimates. Note 4. Effect of Cell Type, Sex, and Rate of De- livery of Radiation There is evidence from many organisms and many systems. This is reviewed in detail and presented in summarizing tables in UNSCEAR papers. We shall present only a short summary and refer the reader to the UNSCEAR reports for details and references to the original litera- ture. Some of the best documentation comes from mouse studies and, because it is likely to be more relevant for human risk estimation than data from insects, plants, or cell cultures, we shall present only mouse data here. The follow- ing summary, provided by W. L Russell, illus- trates the range of sensitivity of different stages in the two sexes. The data are all for single locus recessive mutations induced by acute X-radiation. Spermatogonia Spermatogonia of newborn Mature oocytes Immature oocytes in adults Immature oocytes in newborn 1.7 x 10-7/locus/rem 1.3 1.8-5.4 0 1.0 Data are not given for spermatozoa and spermatids. Although the rates for these cells are considerably higher than for spermatogon- ia, we have not given them because these stages occupy such a short part of the total pre-reproductive period in man. Likewise, the immature oocyte stage in female mice where mutation production is very low (not signifi- cantly different from the spontaneous controls) is a stage of long duration relative to the much" more sensitive mature oocytes, or immature oocytes in newborn. In the female, we are less confident of the comparability of mouse and human because of unexplained differences in the cell-killing rate (22,23). The dose-rate dependence can also be illus- trated from mouse data. For spermatogonia, the rate of production of mutations is about 1/3 as great with low dose-rates of X-rays as with a high dose-rate. There seems to be a comparable reduction when the total dose is small, even if given at a rapid rate. For mature oocytes in the female, the rate of mutation production when the dose is administered at a very slow rate is only about 1/20 the value for the same dose given quickly (24-29). Note 5. Empirical Studies of Mouse Populations There are several recent reviews of this sub- ject (30-36). Although the simplest approach to assessing radiation risks would seem to be direct observa- tion of harmful changes in offspring and later descendants of irradiated mammals, such stud- ies are generally believed to reveal only part of the total genetic damage. Recessive lethal changes in particular tend to escape detection unless special stocks and special breeding sys- tems are employed, and the same may be said of 61

recessive detrimental changes and mutations associated with small dominant effects. Never- theless, induced hereditary changes leading to skeletal anomalies (37), loss of learning ability, and changes in such quantitative characteris- tics as body weight (38, 39), have been detected by this method. Where the irradiations have been repeated over many generations, such mammalian stud- ies have posed a curious problem. If, as is gen- erally believed, most induced mutations have slight deleterious effects in the heterozygous state, the continued accumulation of such change without apparent eliminations through deaths and failures to reproduce would be ex- pected to cause eventually some obvious and substantial effects on the members of the popu- lation. This has not yet happened in any of the large-scale studies. Results obtained by Spalding and his co- workers are of special relevance in that the exposures, in this case 200 rems per generation to the male line, were continued over a total of 45 generations. It was reasoned that, if muta- tions with individually small effects do, in fact, occur with much greater frequency than muta- tions with major effects, and can accumulate to constitute a damaging genetic load, the pre- sumed effects would eventually be reflected in measurable alterations of the growth and death rates. The experiment was carried out with a highly inbred strain of mice to minimize initial chance differences in the irradiated and unirradiated lines. There were no significant differences between the irradiated and control strains in growth rate or in mortality; the life- time survival curves are almost identical in the two groups. Other such studies of mammals have shown changes in growth rates, but not in any consistent direction. As summarized by Green (30), these negative results-may be due "to the non-existence of in- duced mutations having only moderate individ- ual effects on heterozygotes, to the failure to find the right indicator trait, or to the relative- ly small sizes of the experiments so far con- ducted and their relative lack of power for dis- criminating small genetic differences in the presence of large amounts of non-genetic vari- ability." Note 6. Hiroshima-I"<iagasaki Studies and Sex- ratio as a Measure of Genetic Effect The studies of children whose parents were irradiated in the Japanese bombings have been reviewed several times (40-42). None of the measures of health and survival showed signif- icant differences, nor did physical measure- ments. Special attention has been given to the sex- ratio (43). Animal experiments have shown sex-ratio shifts in the direction expected if re- cessive X-chromosome-linked deleterious or lethal mutations were being produced (44). The only results relative to sex-ratio in the Ja- panese studies that approached statistical sig- nificance are changes in the sex-ratio, where the early results are barely significant in the expected direction; but later studies have not confirmed this. There are in the literature eight other studies (43), mostly showing results in the expected direction; that is, a reduced pro- portion of males when the mother is irradiated. Reports that irradiation of females prenatally affects the sex-ratio of their children (45) are not confirmed in the much larger and statisti- cally better controlled Japanese study (46). However, the sex-ratio although easily ob- tained from data extensive enough for consid- erable statistical precision, is notoriously sub- ject to fluctuations for genetically irrelevant reasons in both human and experimental ani- mal populations and the Subcommittee does not believe that sex-ratio is a suitable measure for assessment of the human risk. It is perhaps worth noting that, if the human female is like the mouse in being very resistant to radiation effects, the sex-ratio in the grandchildren of irradiated males may be a more revealing mea- sure of recessive X-chromosomal effects than in the children of irradiated mothers. There is another reason to be suspicious of sex-ratio data as an indicator of genetic dam- age. In the female, one of the two X-chromo- somes is inactivated in each cell. This could mean that a lethal mutation which, if it were on an autosome would not be expressed because of recessivity, might be expressed if it were on the X-chromosome since it would exert its effect in half the cells. Hence, for this reason as well as the one given in (he paragraph above, the abs- ence of a sex-ratio change in the children of irradiated parents -nay not mean anything. 62

Much more important, we think, is the absence of significant effects on physical measurements or on health and survival. Note 7. Effects of Transmutation Of the radioactive isotopes absorbed by the body, only three (H3, C^, and P32) are incorpo- rated into DNA where transmutation effects could possibly induce mutations in addition to those induced by the emitted radiation. H* be- comes helium, C14 becomes nitrogen, and P32 becomes sulphur. Committee 24 - Radioactive Nucleic Acids and Precursors - of the National Council on Radiation Protection and Measurements has considered the relative effects of transmuta- tion and radiation from tritium and carboni-i and concluded that the effect of radiation greatly outweighs that of transmutation. Transmutation of tritium in DNA thymidine produces mainly single strand breaks. Under normal growth conditions, these breaks are repaired with great efficiency. It has been found that tritium decaying in the five position in the pyrimidine ring in DNA leads to an in- creased yield of mutations in both microor- ganisms (47) and Drosophila (48) indicating that transmutations can, indeed, produce mu- tations. Nonetheless, tritium substitution for hydrogen at this position of the pyrimidine ring is extremely rare since the position binds only about 0.04% of the total nuclear hydro- gen. All the experimental evidence overwhelm- ingly indicates that the effects of intranuclear tritium are produced by beta radiation. Carbon^ can also cause effects from chemi- cal transmutation to nitrogen and these effects should be added to the radiation effects from the beta particle. Nonetheless, when there are many carbon -14 decays per nucleus the radia- tion effects would again far outweigh the con- sequences of transmutation. This situation is deemed to be similar to that which occurs with tritium. Although the transmutation of P:5a incorpo- rated into DNA can lead to strand breakage and, thus, possibly to mutations, P32 is not preferentially located in DNA which makes the transmutation effect less important. Some ex- perimental results with Drosophila (49) indi- cate that here there might be a slight effect of transmutation in the induction of mutations. This effect was found to be far less efficient than the irradiation. Under certain conditions (50), the transmutation effect was not even no- ticed. Thus, experimental results in Drosophila support the view that the contribution from transmutation is small compared to direct ra- diation effects. The general conclusion is that for these nu- clides that are incorporated into nucleic acids the beta radiations far outweigh any contribu- tion from transmutation effects and that it is, therefore, justified to consider the main effect to come from the radiation emitted when the isotope disintegrates. This is also true a for- tiori for those isotopes not incorporated into nucleic acids. Note 8. The High Frequency and Heterozygous Expression of Minor Mutations It has been known for many years that minor deleterious mutations in Drosophila are more numerous than those that produce a lethal or near-lethal effect. The first accurate quantita- tive assessment of the mutation rate of such minor genes was by Mukai (51), who used the device of letting mutations accumulate on a chromosome that was protected from the effect of natural selection by being kept heterozy- gous generation after generation with careful precautions to minimize natural selection. From the mean and variance of the decline in viability when such chromosomes were later made homozygous, he inferred that the muta- tion rate is at least 15 times the lethal muta- tion rate. These results have recently been confirmed in three independent experiments (52). Further confirming evidence comes from microorganisms showing that mutations re- sulting from substituting one amino acid for another (missense mutations) are very much underrepresented relative to chain-terminat- ing (nonsense) mutations among conditional lethals (53, 54). Presumably, the former are producing effects too small to be detected by the system employed. Although these mutants are found in very high frequency in natural populations of Dro- sophila, they are not as frequent as they would be if they were completely recessive. This 63

means that they must be eliminated from the population through heterozygous effects (55, 52). The high frequency of these mutants and their degree of heterozygous expression is such that they should have appreciable effects on the viability or fertility of the population. An increased mutation rate would, therefore, be expected to cause a general, non-specific reduc- tion in the fitness of the individuals in the popu- lation through the production of such mutants. A mitigating factor is that these individually minor mutants are less frequent, relative to severe mutants, among radiation-induced than among spontaneous Jtiutations (56). Radiation is known to produce genetic changes at all lev- els—single base replacements, insertions and deletions of nucleotides, changes involving several bases, and on up to gross chromosome rearrangements (57). However, the ratio of deletions and chromosome rearrangements to single base effects is likely to be much higher for radiation-induced than for spontaneous changes. Note 9. The Kinetics of Mutation and Chromo- some Breakage by Radiation The genetic material is DNA which contains information in the sequence of its four nucleo- tides. Each sequence of 3 nucleotides (triplet) codes for an amino acid in a protein. A gene is composed of many hundreds or more of nucleo- tides in a specific sequence. Not all DNA codes for proteins; probably the great majority has other functions, largely unknown. The DNA itself is organized into larger linear nucleopro- tein structures, the chromosomes, found in the nucleus of the cell. Any change of a nucleotide such that a given triplet will now code for a different amino acid constitutes a mutation. Other changes in cod- ing also can have mutagenic consequences. For instance, the addition, or deletion, of a nucleo- tide from DNA will shift the reading sequence of the code, since it is read 3 nucleotides at a time sequentially. Such frame shift mutants will change whole sequences of amino acids in the protein up to the point where a reverse shift can put the reading back into proper reg- ister. Thus, even a change, deletion, or addition of a single nucleotide in DNA can be a muta- tion. In addition, a larger class of mutational events arises from the breakage of the chromo- some itself with subsequent deletion or rear- rangement of the broken pieces. These changes are often large enough to be seen if the chromo- somes are examined under the microscope. Their size distribution, however, forms a con- tinuum from the very small deletion of a single nucleotide to the loss of a whole chromosome. At the bottom of the range, it is impossible to define just where a deletion should be con- sidered a point mutation in the gene rather than a chromosome breakage type of mutation. For most of the chromosome rearrangements considered in this context, with low LET irra- diation, the frequency of induced rearrange- ments is proportional to the dose over the dose range of interest. At higher doses, more com- plex kinetics are observed (58). Note 10. The Linearity, No-threshold Assump- tion As outlined in the Note 9, there is strong ev- idence that, for single locus mutations in Dro- sophila, the dose-response relationship is linear down to the lowest doses that have been ade- quately tested. There is no evidence for any threshold. If there is none, then the curve, when extrapolated to lower doses, should intersect the zero-dose ordinate at a value equal to the spontaneous rate. The observations are com- patible with this, but the statistical error is too large for this expectation to be tested with any rigor. As mentioned in the previous note, another reason to expect a linear relationship is that for very low doses there is very little opportu- nity for ionizations from independent ion tracks to occur in the same cell locality. Any effect following exponential kinetics with an exponent larger than one is bound to disappear at sufficiently low doses. For phenomena involving breaking and re- joining chromosomes, such as reciprocal translocations, it is usual (as stated in the pre- vious note) to have a power curve at moderate doses. However, the data for translocation production in Drosophila oocytes are best fit- ted by a curve with both a linear and a quad- ratic component. At low doses, the quadratic component becomes negligible. This is readily understood; even if the translocation requires two or more ionizations, these are much more 64

likely to be part of the same ion-cluster than from independent ion tracks. So we expect the linear component to predominate greatly at doses that commonly apply to the human popu- lation. In the mouse, two opposite types of depar- ture from linearity have been found for acute irradiation of spermatogonia. One of these has been explained by differential cell killing, and the other by repair of premutational damage. The first departure consists of an upward convexity of the dose-effect curve at high dos- es: an x-ray dose of 1000 R actually produced fewer mutations than did a dose of 600 R (59, 60). Russell,s hypothesis to account for this result is that in the heterogeneous population of spermatogonial cells some cells are more sensitive to both killing and mutation. Thus, at high doses, the sensitive cells are destroyed, leaving only those cell types that produce few- er mutations. If this effect were to extend down to lower dose levels, then the mutation rate at these levels would be higher than predicted from a linear interpolation between 600 R and 0 R. However, at 300 R, no significant depar- ture from linearity was observed. Recent work by Oakberg (61) indicates that the true stem cells in the mouse testis are not as easily killed by radiation as are the rest of the spermato- gonia, and that differential killing among these stem cells is not, in fact, likely to have any humping effect on the dose curve in the range below 500 R. Furthermore, mutation-rate stud- ies in the low dose range indicate that if there is any tendency toward such a humping it is more than counterbalanced by the opposite departure from linearity, to be described below. In Drosophila, on the other hand, Oftedal (62) has reported that spermatogonial mutations produced in the 0-300 R interval indicate a rela- tively higher mutagenicity of the lower doses in the range. He accounted for this in terms of the hypothesis invoked by Russell to explain the similar effect found in the 600-1000 R interval for the mouse. There are doubts about the sta- tistical significance of Oftedal,s results, and large-scale studies done by Abrahamson (63) in an attempt to check them have failed to reveal any evidence for a significantly increased mu- tation rate per R at doses of 20 and 100 R rela- tive to 500 R. Abrahamson,s results at face value point in the opposite direction, but are not significantly different from linearity. The second type of departure from linearity observed in the mouse consists of an upward concavity of the dose-effect curve at low doses (26). This non-linear relation for mutations that seem to be mainly the result of single- track ionization events (26, 64-66) is explained on the hypothesis that there is repair of muta- tional or premutational damage, but that the repair process is either damaged or saturated at high doses and high dose rates. This hypoth- esis, which was originally derived from the dis- covery of a dose-rate effect in mouse spermato- gonia and oocytes (67, 68), predicts that repair could operate even at high dose rates, provided that the total dose were small or given in small fractions at intervals long enough for the re- pair process to recover. As shown above, this prediction was met for small total doses. It has also proved true for fractionation. The finding of a dose-rate effect for mutation induction in mouse spermatogonia and oocytes raised anew the question of whether there might be a threshold dose or dose rate below which all mutational damage would be repaired. Explora- tion of a range of dose rates provides no evi- dence of a threshold dose rate for mutation in- duction in mouse spermatogonia (26, 69, 70). Mutation frequency drops as the dose rate is lowered from 90 R/min through 9 R/min to 0.8 R/min; but below that level, to 0.009 R/min and even 0.001 R/min, there is no further reduction in mutation frequency. Therefore, we shall make the prudent assumptions that there is no threshold dose rate in the male and that the dose response at low dose rates is linear. The female mouse, in contrast to the male, shows no levelling-off or plateau in mutation frequencies as the dose rate is lowered (26, 69, 70). At the lowest dose rate tested, 0.009 R/min, the mutation frequencies, even from high dos- es, are not significantly higher than in con- trols. Note 11. The Reason for Using Chronic Radia- tion to the Mouse Male as the Basis of Calcula- tions Mature oocytes in the mouse are relatively susceptible to radiation effects. The rate of production of point mutations is about 5 x 10-7 per locus per rem with acute radiation. Howev- er, there is a reduction to about 1/20 of this amount for chronic radiation. The stages prior 489-797 O - 7J - 6 65

to the mature oocyte are very resistant to mu- tation; hardly any mutations are produced. In the mouse the duration of the mature oocyte is about 7 weeks. It is reasonable to assume that in humans the stage of sensitivity is short rela- tive to the total pre-reproductive life cycle, as it is in the mouse, but there is no direct evidence for this. Likewise, mature spermatozoa and sperma- tids are more susceptible to radiation-induced mutation than spermatogonia, and there is no reduction in susceptibility with low dose-rate. Again, however, the time that a particular co- hort of germ cells is in the mature sperm stage is a small fraction of the whole period between conception and reproduction (See Note 4). Human exposure to radiation, insofar as this is genetically significant, is almost always in very low doses. Of the total contribution from medical radiation by far the greatest contribu- tion is from diagnostic rather than therapeutic radiation. Therapeutic radiation involves large doses at high rates, but most persons re- ceiving such radiation are past the age of re- production, or for other reasons are not likely to reproduce, or the irradiated region does not include the gonads. Diagnostic radiation is in small doses, although the rate may be fairly high. However, the total dosage is usually so small as to be more comparable to the chronic or fractionated-dosage mouse experiments. Radiation from sources related to nuclear ener- gy is small in amount and given at a slow rate. For these reasons, the mouse data on chronic irradiation and fractionated doses are more relevant for estimation of the human risk than experiments with high doses and high dose- rates. Note 12. The Average Human Mutation Rate A recent discussion of human mutation rates and of the reasons for thinking that most mea- sured rates are higher than the true average is given by Cavalli and Bodmer (71). Mutation rates in the literature, which average around 2 x 10-5 are clearly a selected sample. When a study was made of all X-linked mutants found in a complete population survey, the mean mutation rate for mutants found in the survey was about 0.4 x 10-5. However, there is reason to think that the true average may be still less because traits which are recognized as being caused by mutant genes are more likely to be recognized if the mutant has been studied be- fore; hence, there may still be a bias in favor of those with higher rates. Cavalli and Bodmer suggest that this may lower the average by another factor of 10. We shall assume that the true average lies between these values and, therefore, we take 0.5 x 10-5 and 0.5 x 10-6 as reasonable limits for the average mutation rate for human recessive genes. Note 13. The Calculations for Table 2 If the doubling dose is 200 rem, then 5 rem per generation will lead to an equilibrium increase of 5/200. For autosomal dominant traits, the present incidence is 10,000 per million; 5/200 x 10,000 = 250, the lower limit for the equilibrium value in Table 1. If the doubling dose is 20 rem, or 1/10 as much, then the equilibrium value will be 10 times as high, or 2,500. Of this amount, 20 percent is expected in the first generation, and the value would slowly rise to the equilibrium numbers if the radiation were continued at this rate generation after generation. Similar considerations apply to the X-linked calculations. The recessive X-linked genes list- ed in the 1958 United Nations report (5), as well as the autosomal recessive genes, caused a greater reduction in viability than the domi- nants. The data suggest that the average fit- ness is roughly one-half normal. At equilibri- um, the incidence of affected individuals, which will nearly all be males,- is approximately three times the mutation rate. (The equilibrium gene frequency is 3u/s, where u is the mutation rate and s in this case is 1/2. So the proportion af- fected among males is 6u, or among both sexes, since only males are significantly affected, is 3u.) The incidence of persons affected by a new mutation is one-half the mutation rate (in this case, the mutation rate in females, since the affected males get their mutant gene from their mother). So we would expect the number of per- sons affected by new mutants to be about 1/6 of the equilibrium number, less than this if the female mutation rate in humans is less than the male, as it is in mice. One-sixth of 400 is about 65, so the number of new mutants i-s estimated to be not greater than this. For a 20 rem dou- bling dose, the equilibrium value for 5 rem each generation will be (5/20) x 400 or 100. This is the upper limit estimate; the lower limit may be 66

close to zero if the female is relatively insensi- tive to radiation effects. Note 14. Chromosome Rearrangements in the Mouse Translocations can be detected in the mouse by the semisterility of heterozygotes or by the cytological observation of special prepara- tions of spermatocytes, looking for multiva- lents at diakinesis or metaphase (72). In the latter method, spermatocytes of irradiated males are examined after appropriate delays to detect translocations induced earlier in sper- matogonia. This method is very much more efficient in scoring than are genetic tests based on partial sterility, and produces a transloca- tion count following spermatogonial irradia- tion about twice that obtained by the semister- ility method (73). No assumption other than selective elimination of some translocation- bearing cells can account for the discrepancy in the two methods of screening. We shall use semisterility data in the mouse as the basis for our calculations, since this method provides a more accurate measure of the number of viable zygotes produced. Recent data, corrected for the control rate of semisterility, give 3.4 x 10-5/ gamete/R as the rate of induction (74). There are a number of studies showing essential line- arity at lower doses (63). Recent work has re- vealed a dose-rate effect with X-rays as well as gamma radiation, the reduction in incidence of translocations being greater than 2-fold at the lower dose rates (75, 76). The situation in the female is more compli- cated. Irradiation of the mouse oocyte rather frequently leads to the recovery of semisterile daughters but not of semisterile sons. Appar- ently not all reciprocal translocations are re- covered. We shall use the face value data from semisterility following oocyte radiation, which gives about 3 x 10-5. However, there are no data that we know of for earlier stages. This is a serious gap in knowledge, for we don,t know whether this will be like the situation in muta- tion production, where irradiation of earlier stages has virtually no effect. Translocations make up the great bulk of the chromosome rearrangements (the reason for this is that there is a much greater chance that the induced breaks will be in different chromo- somes than in the same chromosome). Inver- sions can be screened for anaphase bridges and fragments at anaphase I following earlier ra- diation, but it is not yet clear how accurate this is as a basis for assessment of the radiation risk. The consequences, however, should either be semisterility or the selective elimination of aneuploid gametes, hence estimates of semi- sterility in the following note will include risks from inversions. Note 15. Calculations for Table 3 The numerical values in Table 3 come from a number of sources. The data for current inci- dence are from population surveys of newborn (for a summary see Ref. 77) and studies of aborted fetuses (78-80). The estimates of radia- tion effects all depend on mouse data. With chronic radiation the rate of X-chromo- some loss when female mice are treated in the oocyte stage is about 6 x 10-6 per gamete (81). The rate from treated spermatogonia was not significantly different from the control rate. Assuming the same effectiveness for the hu- man, the fraction of XO among all zygotes would be half the above fraction, since half the deficient eggs are fertilized by Y-bearing sperm and die as very early embryos. The effect of 5 rem would then be 5x6x10-6x1/2 = 15xl0-e. This is the source of the number 15 for XO among recognized abortions in Table 3. In humans only about one in 40 of this type sur- vived to birth, so the frequency of induced XO types among live-born infants would be less than one per million. The other monosomic types die so early that they are not detected as abortions and, therefore, make a negligible contribution to human distress. There are no comparable mouse data for trisomy. The frequency of radiation-induced trisomy for the X-chromosome of Drosophila is about 1/4 that of the corresponding monosomy. Taking all this at face value for human chro- mosomes, the frequency of trisomy for the 22 human autosomes with 5 rem exposure would be 5 x 6 x 10-6 x 22 x 1/4 = 165 x 10-6. Carr (78) estimates that about one-third sur- vive long enough to produce recognized abor- tions, so the frequency of abortions from this 67

cause per million live births is about 55. A small fraction of this number would survive to produce live births with congenital defects. Again, we are taking the mouse data at face value, which yields a negligible effect from ir- radiated males. To estimate the number of unbalanced rear- rangements induced by 5 rem, we rely on trans- locations leading to semisterility in the mouse. For low dose irradiation, the frequency of sem- isterility among progeny of males irradiated in the spermatogonial stages is about 1.5 x 10-5 per rem. There is some indication that the rate in humans may be higher. This comes from studies of Brewen (82) (unpublished) which show in lymphocyte cell cultures twice as many dicentrics in human cells as in mouse cells. There is good reason to believe that at low dos- es the number of translocations should be Ap- proximately equal to the number of dicentrics; for chromatid breaks this has been shown ex- perimentally (83). In the absence of better in- formation, we shall estimate the human rate by doubling that for the mouse. Only balanced translocations are detected as semisterility in the offspring. The ratio of the frequency of undetected, unbalanced translo- cations to that of balanced translocations in the offspring of irradiated individuals depends on a number of unknown factors. Under certain and possibly unrealistic assumptions (such as random segregation in a ring quadrivalent) it might be as high as 4:1. On the other hand, in a chain quadrivalent, 2:1 appears more likely. With an excess of alternate segregations, a ratio of 0:1 may be approached. However, all this applies only if the translocation took place between chromosomes. In the case of chromatid translocations, the ratio may be extremely high because the mitotic descendants of the cell in which the breakage occurred are likely to be already unbalanced. Since radiation may in- duce chromosome breaks as well as chromatid breaks, and we do not know which type is more common in the germ line, we shall use 4:1 as a reasonably conservative overall estimate of the ratio. The estimate for 5 rem to spermato- gonia is 5 x 1.5 x 10-5 x 2 x 4 = 600 x 10-6. The estimates for progeny of irradiated fe- males are still more dubious. The amount of semisterility in the progeny of irradiated fe- males is about the same as for males with acute radiation. However, the relative effectiveness of chronic radiation may be less than for sper- matogonia. Translocations induced in oocytes are chromatid exchanges; the ratio of balanced to unbalanced gametes is unknown. There are also some unexplained peculiarites in the fe- male mouse data referred to in Note 14. We are aware of no quantitative data on translocation production in female mice from radiation of stages before the mature oocyte. For single locus mutations, these stages are much less sensitive to radiation, indeed almost immune. This may be true for translocations as well, but we are not sure. For all these reasons, we hard- ly know how to begin a quantitative measure. It is not likely that the female rate is higher than the male, so we shall again be conserva- tive and simply assume the male rate. Doubling the male rate, we get 1200 unbal- anced and 300 balanced translocations in a population of 1 million exposed to 5 rem. Most of these would eventually be eliminated from the population in the form of very early em- bryonic deaths that are not detected. It has been estimated that about 30 percent would be expressed as recognized abortions. The number leading to congenital anomalies among the live- born is a much smaller fraction, probably con- siderably less than 5 percent. Taking 5 percent of 1500 as the upper limit for congenital anom- alies and 30 percent as the estimate for abor- tions gives the values 75 and 450, given as equi- librium values in Table 3. The unbalanced prod- ucts would affect the first generation, so this is 12/15 of the total. The values in Table 3 very likely are too high, for the various reasons given above. We are not completely sanguine, however. In mammalian cell cultures, including human, the production of translocations by radiation compared with the spontaneous rate suggests a very low doubling dose. It is not clear what the fate of these would be if they occurred in germ cells. Many may be eliminated through individ- ual cell deaths. Note 16. Radiation Induced Nondisjunction: Is there a Threshold? Recent work on the induction of nondisjunc- tion in the fruit fly by ionizing radiations has 68

led to two quite different interpretations. Re- ports from one laboratory (84, 85) suggest that chromosomal interchange may play a signifi- cant role in bringing about improper segrega- tions of chromosomes as a result of misalign- ments on the meiotic spindle at division I. An alternative hypothesis of a threshold dose be- low which nondisjunction cannot be induced has been proposed (86), based on the failure to find increases in sex chromosome trisomy at doses below 1200 R. However, a more recent report from the latter laboratory (87), shows that an appreciable fraction of sex chromo- some monosomics are also trisomic for the small fourth chromosome, a finding that is re- quired by the interchange model. Since nega- tive findings can scarcely provide evidence of total lack of an effect at low doses, it seems prudent to suppose that nondisjunctions may result from radiation-induced breakage and interchange, or possibly from other events for which thresholds cannot be demonstrated. Note 17. Is Maternal Radiation a Significant Cause of Human Nondisjunction? The only human observations that are rele- vant to this question pertain to autosomal tri- somy, especially trisomy 21 which results in mongolism (Down,s syndrome). One group of studies, retrospective and prospective, deals with medical radiation, mostly for diagnostic purposes, and shows a significant association between pre-conception radiation of the mother and the probability of the child being trisomic. The association was found in the retrospective studies by Uchida and Curtis (88), with 81 mothers of mongoloid children and 81 matched controls, and by Sigler et al. (89), with 216 in each group. Another study (90), failed to show even a trace of the effect, but since there were only 51 mothers of mongoloid children and 51 matched controls these data hardly carry enough weight to invalidate the other studies, especially since the prospective study by Uchi- da et al (91) again revealed a significant asso- ciation. The large body of data from.Hiroshima and Nagasaki (92) fails to show any relationship between irradiation and mongolism. The dis- crepancy between the two sets of data is highly significant under any reasonable assumption of the dosages involved in the medical radia- tion. The fact that the Japanese study involved large numbers of normal people (not patients), was done prospectively, and involved the entire population of newborns, all carefully exam- ined, gives it statistical precision and makes it less susceptible to possible extraneous causes. A possible alternative explanation of the stud- ies involving medical x-rays is that, other fac- tors being equal, women who are receiving x- rays are in poorer health than those who are not and that the poor health is the cause of nondisjunction. It is likely that high doses of radiation will cause nondisjunction in man just as it does in Drosophila and probably does in the mouse. However, the mouse data are concerned with monosomy rather than trisomy and chromo- some loss can occur through processes other than nondisjunction. The way that high doses of radiaton upset normal disjunction may be quite different from the way in which other genetic effects are produced. The radiation may act on the spindle and, in particular, may not be produced by a single ion cluster. Therefore, the argument against a threshold used for oth- er genetic effects may not apply to nondisjunc- tion. In view of the fact (reported in Note 16) that chromosome interchanges, known to be caused by radiation, can cause nondisjunction in Dro- sophila, it would be imprudent to assume an absolute threshold. Hence, our calculations are done on a linear assumption as described in Note 15. This may lead to a gross overestimate of the risk, but it is very unlikely to be an un- derestimate. Note 18. An Attempt to Measure the Economic Cost of Radiation Cost-benefit calculations may necessitate that the cost of radiation exposure be mea- sured in direct economic terms such as dollars. We give here an illustrative example of how this might be done, patterned after Lederberg (93, 94). Assume that the present cost per capita for poor health is $400 per year. (This is based on an estimated $80 billion in medical expenses in 1970 and a population of about 200 million.) $400 per year is $12,000 for a 30-year period. In 69

section VIII-D we estimated that the equilibri- um amount of illness from 5 rem of exposure per 30-year generation would be increased be- tween 0.5 and 5 percent. One rem would produce an increase of 0.1 to 1.0 percent. As fractions of $12,000, these percents are $12 and $120. Thus, one rem per generation continued until equilib- rium is reached would add an amount of illness equivalent to a cost of $12 to $120 per person per 30 years. This implies that the amount of damage done by one rem, when integrated over all future generations corresponds to a cost of $12-120 - regardless of whether equilibrium has yet been reached or not. Thus we say: The total future cost of one man-rem, in terms of health costs paid for in present dollars, is between $12 and $120. This may provide one way for putting a dol- lar value on a dose commitment of one rem that could be used in cost-benefit calculations. The cost would be distributed over many genera- tions in the future. We do not undertake to analyze the moral or economic implications of choosing a discount rate for genetic damage that is expressed far in the future, nor do we allow for changes in the purchasing power of a dollar. We do not intend to imply that all money spent for health costs is directly caused by ill-health, nor on the con- trary that this represents the total costs of poor health. Furthermore, our presentation of a suggested dollar equivalent for the illness induced by one rem does not imply that we seek to exchange dollars for lives. It is essential, however, that our society elaborate some ra- tionale for the allocation of resources in the interest of maximizing our health - a task that demands further analysis by specialists in health economics. REFERENCES (J) McKusick, V. A. 1971. Mendelian Inheritance in Man. Johns Hopkins Press, Baltimore. 3rd. Ed. (2) NCRP, 1960. Somatic radiation dose for the general population. Report of the Ad Hoc Committee of the Na- tional Committee on Radiation Protection and Measure- ments, 6 May 1959. Science 131:482-486. (3) Stevenson, A. 1959. Rad. Res. (Suppl.) 1:306-325. (4) Searle, A. G., Personal communication. (5) UNSCEAR, 1958. General Assembly. Report of the United Nations Scientific Committee on the Effects of Atomic Radiation. 1958 (A.3838: G. A. Official Records, 13th Sess. Suppl. No. 17). (6) UNSCEAR, 1962. (A/5216: G. A. Official Records, 17th Sess. Suppl. No. 16). (7) UNSCEAR, 1966. (A/8314: G. A. Official Records. 21st Sess. Suppl. No. 14). (8) UNSCEAR, 1969. (A/6314: G. A. Official Records, 24th Sess. Suppl. No. 13). (9) British Medical Research Council, 1956, 1960. The Hazards to Man of Nuclear and Allied Radiations. Lon- don. Her Majesty,s Stationery Office. (10) Taylor, Lauriston, S., Radiation Protection Stand- ards, CRC Press 1971. (11) NCRP, 1957. Maximum permissible radiation expo- sures to man (NCRP). Amer. J. Roent. 77:910-913. (12) NCRP, 1968. Maximum permissible radiation expo- sures to man (NCRP). Radiology 71:263-266. (13) FRC, 1960, 1962. Background Material for the Devel- opment of Radiation Protection Standards, Staff Re- ports of the Federal Radiation Council. (14) ICRP Publicatons No. 1(1956) and 6(1964). (15) Lifschytz, E., and R. Falk. 1968. Fine structure anal- ysis of a chromosome segment in Drosophila melanogas- ter. Analysis of x-ray-induced lethals. Mutation Res. 6:235-244. (16) Lifschytz, E., and R. Falk. 1969. Fine structure anal- ysis of a chromosome segment in Drosophila melanogas- ter Analysis of ethyl methane sulphonate-induced le- thals. Mutation Res. 8:147-155. (17) Shannon, M. P., T. C. Kaufman, and B. H. Judd. 1970. Lethality patterns of mutations in the zeste-white re- gion of Drosophila melanogaster. Genetics 64:58 (1K) Berendes, H. D. Polytene chromosome structure at the submicroscopic level. A map of region X, I-4E of Dro- sophila melanog-aster. Chromosoma (Berlin) 29:118-130, 1970. (19) Lefevre, George, Jr. Salivary chromosome bands and the frequency of crossing over in Drosophila melanogas- ter. Genetics 67:497-515.1971. (20) Kaufman, T. C., M. W. Shen, and B. H. Judd. The com- plementation map of mutations in a small region of the X-chromosome of Drosophila melanogaster. Genetics 61: Supplements 30 and 31,1969. (21) Rayle, E. E., and M. N. Green. A contribution to the genetic fine structure of the region adjacent to white in Drosophila melanogaster Genetics: 39:497-507. 1968. (22) Baker, T. G., 1969. The effects of X-irradiation on mammalian oocytes in organ culture. Biophysik 6:39-45. (23) Baker, T. G., 1971. Comparative aspects of the effects of radiation during oogenesis. Mut. Res. 11:9-22. (24) Russell, W. L., 1968. Recent studies on the genetic effects of radiation in mice. Pediatrics 41:223-230. (25) Russell, W. L., 1967. Factors that affect the radia- tion induction of mutations in the mouse. Ann. Acad. Brasil, Cienc, 39 Suppl. 66-75. (26) Russell, W. L., 1968. Repair mechanisms in radiation mutation induction in the mouse, in Recovery and Repair Mechanisms in Radiology. Brookhaven Symp. Biol. 20:179-189. (27) Russell, W. L., 1967. Recent studies on the genetic effects of radiation in mice. In Proc. I., Int,l, Symp. Bio- logical Interpretations of Dose from Accelerator-pro- duced Radiation (R. Wallace Ed.) USAEC, Div. of Tech. Info. Conf. 670305, pp. 81-87. (28) Searle, A. G., and R. J. S. Phillips. 1968 Genetic insen- sitivity of the mouse dictyate oocyte to chronic irradia- tion, in Effects of Radiation on Meiotic Systems, Interna- tional Atomic Energy Agency, Vienna, pp. 17-25. (29) Batchelor, A. L., R. J. S. Phillips, and A. G. Searle, 1969. The ineffectiveness of chronic irradiation with neutrons and gamma rays in inducing mutations in fe- male mice. British J. Radiol. 42:448-451. 70

(30) Green, E. L. 1968a. Genetic effects of radiation on mammalian populations. Ann. Rev. Genet. 2:87-120. (31) Green, E. L., 1968b. Body weight and embryonic mor- tality in an irradiated population of mice. Mutat. Res. 6:437-448. (32) Newcombe, H. B., 1971. The genetic effects of ionizing radiation. Adv. Genetics 16:329-303. (33) Newcombe, H. B., 1972. Effects of radiation on human populations. Proc. 4th Int. Cong. Human Genet, (in press). (34) Luning, K. G., 1972. Effects of irradiated mouse popu- lations. IV. Effects on productivity 7th - 18th generation. Mut. Res. (35) Newcombe, H. B., and John F. McGregor. Persistence of heritable changes in an irradiated rat population af- ter cessation of the exposures. Radiation Research 47:450-460(1971). (36) Spalding, J. F., M. R. Brooks, and G. L. Tietjen 1969. Lifetime body weights and mortality distribution in mice with 10 to 35 generations of ancestral X-ray exposure. Genetics 63:897-906. (37) Ehling, V. H., 1966. Dominant mutations affecting the skeleton in offspring of X-irradiated male mice. Genetics 54:1381-1389. (38) Newcombe, H. B., J. F. McGregor, 1964. Learning ability and physical well-being in offspring from rat populations irradiated over many generations. Genetics 40:1065-1081. (39) Newcombe, H. B., J. F. McGregor, 1965. Heritable effects of radiation on body weight in rats. Genetics 52:851-860. (40) Neel, J. V., and W. J. Schull. 1956. The effect of expo- sure to the Atomic Bombs on Pregnancy Termination in Hiroshima and Nagasaki. Natl. Acad. Sci. Pub. 461. (41) Kato, H., W. J. Schull, and J. V. Neel 1966. Survival in children of parents exposed to the atomic bomb, a cohort- type study. Amer. J. Human Genet. 18:339-373. (42) Schull, W. J., J. V. Neel, and A. Hashizume 1966. Some further observations on the sex ratio among infants born to survivors of the atomic bombings of Hiroshima and Nagasaki. Amer. J. Hum. Genet. 18:328-338. (43) Schull, W. J. 1963. Hereditary effects. Nucleonics 21 (3):54-57. (44) Havenstein, G. B., B. A. Taylor, J. C. Hansen, and A. B. Chapman 1968. Genetic effects of cumulative X-irra- diation on the secondary sex-ratio of the laboratory rat. Genetics 59:255-274. (45) Meyer, M. B., T. Merz, and E. L. Diamond 1969. Inves- tigation of the effects of prenatal x-ray exposure of human oogonia and oocytes as measured by later repro- ductive performance. Amer. J. Epidem. 89:619-635. (46) Jablon, S., andH. Kato, 1971. Sex ratio in off spring of survivors exposed prenatally to the atomic bombs in Hiroshima and Nagasaki. Amer. J. Epidem. 93:253-258. (47) Person, S. 1968. Lethal and mutagenic effects of triti- um decay produced by tritiated compounds incorporated into bacteria and bacteriophages. In Biological Effects of Transmutation and Decay of Incorporated Radioiso- topes, International Atomic Energy Agency, Vienna. pp. 29-64. (48) Kieft, P. 1968. Induction of recessive lethals by H3 uridine and H3 thymidine in Drosophila. In Biological Effects of Transmutation and Decay in Incorporated Radioisotopes, International Atomic Energy Agency. Vienna, pp. 65-78. (49) Oftedal, P., and W. D. Kaplan. 1968. Some aspects of transmutation studies in Drosophila. In Biological Ef- fects of Transmutation and Decay of Incorporated Ra- dioisotopes. International Atomic Energy Agency, Vien- na. pp. 79-90. (50) Lee, W. R., C. K. Oden, C. A. Bart, C. W. Debney, and R. F. Martin. 1966. Stability of Drosophila chromosomes to radioactive decay of incorporated phosphorus-32. Genetics 53:807-822. (51) Mukai, T. 1964. The genetic structure of natural pop- ulations: I. Spontaneous mutation rate of polygenes con- trolling viability. Genetics 40:1-19. (52) Mukai, T., S. I. Chigusa, L. E. Mettler, and J. F. Crow 1972. Mutation rate and dominance of genes affecting viability in Drosophila melanogaster. Genetics (in press). (53) Whitfield, J. J., R. G. Martin, and B. N. Ames, 1966. Classification of amino-transferase (c gene) mutants in the histidine operon. J. Mol. Biol. 21:335-355. (54) Langridge, J., and J. H. Campbell, 1969. Classification and intragenic position of mutations in beta-galactosi- dase gene of Escherichia coli. Molec. Gen. Genet. 103:339- 347., (55) Temin, R. G., H. U. Meyer, P. S. Dawson, and J. R. Crow 1969. The influence of epistasis on homozygous viability depression in Drosophila melanogaster. Genet- ics 61:497-519. (56) Friedman, L. 1964. X-ray induced sex-linked lethal and detrimental mutations and their effect on the viabili- ty of Drosophila melanogaster. Genetics 49:689-699. (57) de Serres, F. J., H. V. Malling, 1969. Identification of the genetic alterations in specific locus mutants at the molecular level. Jap. J. Genet. 44, Suppl. 1:106-113. (58) Wolff, S. 1962. The kinetics for two-break chromo- some exchanges. J. Theoret. Biol. 3:304-316. (59) Russell, W. L. Lack of linearity between mutation rate and dose for X-ray-induced mutations in mice. Ge- netics 41:658-659, 1956. (60) Russell, W. L., and Liane Brauch Russell. Radiation- Induced Genetic Damage in Mice. In: Progress in Nuclear Energy Series VI, Volume 2.—Biological Sciences, eds. J. C. Bugher, J. Coursaget, and J. F. Loutit. Pergamon Press, Ltd., London, 1959, pp. 179-168. (61) Oakberg, E. F. A new concept of spermatogonial stem cell renewal in the mouse and its relationship to genetic effects. Mutation Res. 11:1-7,1971. (62) Oftedal, P. 1968. A theoretical study of mutant yield and cell-killing after treatment of heterogeneous cell populations. Hereditas 60:177-210. (63) Abrahamson, S. 1972. Unpublished data furnished to the Subcommittee. (64) Russell, W. L. Observed mutation frequency in mice and the chain of processes affecting it. Ciba Foundation Symposium on "Mutation as Cellular Process," 1969, pp. 216-228. Eds. G. E. W. Wolstenholme and Maeve O,Connor. (J. & A. Churchill, London, 1969.) (65) Russell, Liane B. Definition of functional units in a small chromosomal segment of the mouse and its use in interpreting the nature of radiation-induced mutations. Mutation Res. 11:107-123, 1971. (66) Auerbach, C., B. J. Kilbey, 1971. Mutation in Eukar- yotes. Annual Review of Genetics 4:163-218 (ed. H. Ro- man - Academic Annual Review, Inc., Palo Alto, Calif.) (67) Russell, W.L., Liane Brauch Russell, and Elizabeth M. Kelly. Radiation dose rate and mutation frequency. Science 128:1546-1550,1958. (68) Russell, W. L. The effect of radiation dose rate and fractionation on mutation in mice. In "Repair from Ge- netic Radiation Damage," ed. F. Sobels. Pergamon Press (Oxford), 1963, pp. 205-217; 231-235. (69) Russell, W. L. Mutagenesis in the mouse and its appli- cation to the estimation of the genetic hazards of radia- tion. Proc. IVth International Congress of Radiation Research, Evian, France, June 29-July 4,1970. In press. 71

(70) Russell, W. L. The genetic effects of radiation. In "Peaceful Uses of Atomic Energy," Vol. 13, IAEA, Vien- na, 1972. pp. 487-500. (71) Cavalli-Sforza. L. L., and W. F. Bodmer 1971. The Genetics of Human Populations. W. H. Freeman and Co., San Francisco. (72) Evans, E. P., G. Breckon and C. E. Ford 1964. An air- drying method for meiotic preparations from mammali- an testes. Cytogenetics 3:289-292. (73) Ford, C. E., A. G. Searle, E. P. Evans and B. J. West, 1969. Differential transmission of translocations in- duced in spermatogonia of mice by irradiation. Cytoge- netics 8:447-470. (7-0 Luning, K. G.. and A. G. Searle 1971. Estimation of the genetic risks from ionizing radiation. Mut. Res. 12:291-304. (75) Ford, C. E., A. G. Searle, and E.- P. Evans, 1969. Dif- ferential transmission of translocations induced in sper- matogonia of mice by x-irradiation. Cytogenetics 8:447- 470. (76) Searle, A. G., E. P. Evans, and C. E. Ford, 1968. Stud- ies on the induction of translocations in mouse spermato- gonia. I. The effects of dose-rate. Mut. Res. 6:427-436. (77) Trunca Doyle, C. 1972. A controlled cytogenetic study of patients with idiopathic mental retardation syn- dromes. Ph. D. Thesis, University of Wisconsin, Madison. (78) Carr, D. H., 1970. Chromosomes and abortion. In Human Population Cytogenetics, Ed. by P. A. Jacobs et al. Edinburgh Univ. Press, pp. 201-257. (79) Carr, D. H., 1971. Chromosomes and abortion. Ad- vances in Human Genetics. 2:201-257. (80) Boue, J. E., and A. Boue, 1970. Les aberrations chro- mosomiques dans les avortements spontanes humains. La Presse Medicale 78:635-641. (S1) Russell, L. B., Unpublished data. (82) Brewen, J. G. Genetic Hazards of Ionizing Radia- tions: Radiation Cytogenetics Extrapolated from Mouse to Man. (Unpublished mms.) (8.T) Patau, Klaus, Unpublished data. (84) Parker, D. E., and J. H. Williamson 1970. Heterolo- gous interchange at meiosis in Drosophila. III. Inter- change-mediated nondisjunction. Mutation Res. 9:273- 286. (85) Parker, D. E., 1970. Coordinate nondisjunction of Y and fourth chromosomes in irradiated compound-X fe- male Drosophila. Mutation Res. 9:307-322. (86) Traut, H. 1970. Nondisjunction induced by X-rays in oocytes of Drosophila melanofraster. Mutation Res. 10:125-132. (87) Traut, H., and W. Sheid, 1971. The production of mon- osomic-trisomic individuals in Drosophila melanogaster by X-irradiation of immature oocytes. Mutation Res. 13:429-432. (88) Uchida, I. A., and E. J. Curtis 1961. A possible asso- ciation between maternal radiation and mongolism. Lancet, ii. 848-850. (89) Sigler, A. T., A. M. Lilienfeld, H. B. Cohen, and J. E. Westlake 1965. Radiation exposure in parents of child- ren with mongolism (Down-s syndrome). Bull. Johns Hopkins Hosp. 117, 374-399. (90) Carter, C. O., K. A. Evans, and A. M. Stewart 1961. Maternal radiation and Down-s syndrome (mongolism). Lancet ii, 1042. (91) Uchida. I. A., R. Holunga, and C. Lawler 1968. Mater- nal radiation and chromosomal aberrations. Lancet, ii, 1045-1049. (92) Schull, W. J., and J. V. Neel 1962. Maternal radiation and mongolism. Lancet i, 537. (M) Lederberg, J. 1971a. Squaring an infinite circle: Ra- diobiology and the value of life. Bull. Atom. Sci. Sept. 1971, pp. 43-45. (94) Lederberg, J. 1971b. Foreword, to The Mutagencity of Pesticides: Concepts and Evaluation, Ed. by S. Epstein, and M. S. Legator, MIT Press, Cambridge, Mass. 72