The attempt to determine the nature and extent of the genetic effects of the atomic bombs detonated over Hiroshima and Nagasaki is at this writing by any one of several criteria far and away the most extensive exercise in genetic epidemiology ever undertaken. As the study has unfolded, it mirrors much of the amazing history of human (and general) genetics since World War II. This volume is designed to bring together a set of publications which will not only summarize the various investigations undertaken during the course of this program but also reflect the evolution of the program as it attempted to incorporate the new genetic technologies as they emerged.
The effort by geneticists to predict the genetic consequences for humans of exposure to ionizing radiation, of which this study is a principal component, has certainly been one of the several most serious (and controversial) social responsibilities they have faced in the past half century. But not only is this subject a major issue in its own right, it also serves as a prototype for the type of genetic epidemiology necessary to evaluate the impact on our genetic selves of other human perturbations of the environment in which our species now functions. It is a highly technical issue, ill served by overly simplistic approaches and hasty generalizations.
Between 1946 and 1974, the investigations reported herein were carried out as part of the activities of the Atomic Bomb Casualty Commission (ABCC), a program jointly operated by the U.S. National Academy of Sciences and the Japanese Institute of Health. In 1975, the ABCC was replaced by an independent binational foundation designated the Radiation Effects Research Foundation (RERF), with joint oversight by the U.S. National Academy of Sciences and the Japanese Ministry of Health and Welfare. Throughout the entire period, funding for the American contribution to the effort has been derived from the Atomic Energy Commission and its successor agencies, the Energy Research and
Development Administration and the Department of Energy; funding for the Japanese contribution has come from the Japanese Ministry of Health and Welfare. We, the Japanese and American authors of the papers which are reproduced in this volume, are grateful to the successive directors and the staff of the ABCC amd RERF, whose staunch support has so greatly facilitated the conduct of these studies.
The early events in the history of the genetics program are recorded in Chapter 2. In 1946, knowledge both of the doses of radiation sustained by survivors and the sensitivity of the mammalian genome to radiation was far inferior to the present situation. Nevertheless, as the preliminary data on post-atomic bomb Hiroshima and Nagasaki emerged, it became very likely, given the number of survivors in the two cities and their probable gonadal doses, as well as the indicators one would be forced to pursue in any study, that even a very major effort would not yield a statistically significant difference between the children of survivors receiving increased radiation at the time of the bombings (ATB) and the children of suitable controls. Furthermore, it was clear that a valid study would be expensive and of considerable duration (but at the time, no one anticipated a program that may well extend into the twenty-first century). On the other hand, the exposure of so large a group of normal human beings to radiation, some up to the limits of human tolerance, was so singular in human history that there really was no question in the collective mind of the Committee convened by the National Academy of Sciences to consider the issues, but that a genetic study should be undertaken. Nor were there any doubts in the minds of our Japanese colleagues (see pp. 30–31 of Chapter 3).
There were other unique aspects to this nascent operation. The primary focus of the first round of studies was the occurrence of congenital defects in newborn infants. This is a sensitive subject to parents all over the civilized world, invoking the folklore stigma of some occult wrong doing. This was indeed true in Japan. The study, which would thus have been sensitive in any civilized country, clearly had to be as considerate as possible of Japanese mores and culture. There was from the outset a dynamic tension, between the need to respect the parents who would be involved and their cultural conditioning, and the need to determine pregnancy outcomes accurately, for the sake of these same worried survivors and a humanity just beginning to cope with the implications of the atomic era. In this regard, the Japanese survivors were, immediately after the war, being inundated by grossly exaggerated predictions concerning the fate of their children. Indeed, in those early years no less a figure in genetics than J.B.S.Haldane could suggest, in 1955, that the doubling dose of radiation for humans could be as low as 3 roentgens. In retrospect, we would like to believe that whatever offense our visiting medical teams occasionally rendered by intrusions on families in moments of sorrow is now more than offset by the dispersal of the ugly clouds which for a period were a serious impediment to the marriage of their children. By this statement we do not deny the genetic implications of these exposures but do strongly urge, as will become apparent, that they were far less than unbridled popular belief.
THE THREE PHASES OF THE STUDY
Phase I (1948–1954)
During this period, the program, in conjunction with the operation of a national rationing system, registered almost all pregnant women in Hiroshima and Nagasaki at the
completion of the fifth lunar month of pregnancy and then collected data following the birth of the child on viability, presence of congenital defect, neonatal death, and on sex of child and birthweight. Given the fact that during this period most births were attended by a midwife and occurred at home or in small facilities maintained by the midwives, collecting accurate data on the occurrence of congenital malformation with reference to all births in the two cities required a major effort directed at ensuring the examination of the child by a physician as soon as possible after birth. With respect to children who were stillborn or died during the neonatal period in Hiroshima, some 56% came to autopsy. Altogether, some 76,626 pregnancies were registered in Hiroshima and Nagasaki during this period; an additional 8,391 pregnancies were registered in Kure, which during 1948– 1950 served as a “control” city. Approximately 30% of the surviving children in Hiroshima and Nagasaki were reexamined at age 8–10 months at the ABCC, the reexamination including physical measurements.
In view of recurrent conjectures concerning the possibility that critical data on the genetic effects of the bombs were missed because of the “delayed” initiation of the program (cf. Roberts, 1990), it seems worthwhile to review the time course of the early events rather carefully. The first children conceived after the bombings would (aside from premature terminations) have been born in May 1946. Planning for the genetic studies began in November, 1946; the registration of pregnancies in this program (at the completion of the fifth lunar month) was initiated in Hiroshima in February of 1948 and in Nagasaki in July of the same year, the midpoint of these two dates being approximately May 1948. The first pregnancies to be registered represented conceptions occurring in December, 1947. The “window” during which there was no direct coverage of newborn infants was approximately 18 months. However, the liveborn infants during this 18-month period were, on the basis of birth registrations, subsequently incorporated into the studies on F1 mortality (Chapters 6, 7, and 12) and malignancies among the F1 (Chapter 11), and also drawn upon for the cytogenetic (Chapter 8) and biochemical (Chapter 9) studies. The “window” is thus rather specific for congenital defect.
A preliminary analysis of the accumulated data in 1953 revealed no significant difference between children according to the exposure category of their parents. The experience between 1948 and 1953 with the numbers of children born to parents thought to have received significant amounts of radiation at the time of the bombings made it possible to predict probable future trends with respect to the number of offspring of such parents. There was a precipitous drop in the Japanese birth rate in the early 1950s, due to the introduction of liberal abortion laws designed to help Japan cope with the rapidly expanding population created by the repatriation of Japanese from China, Korea, Formosa, and Manchuria following World War II. This fact, plus the attrition inevitable in a study of this nature, suggested that with respect to congenital malformation and stillbirth, the additional numbers to be obtained in the future were very unlikely to alter the small differences between exposure subgroups from statistically non-significant to significant.
At our request, in July of 1953, a Committee was convened by the National Academy of Sciences to examine both the analysis and the projection. The Committee concurred with the analysis, and voted to recommend termination of the clinical aspects of the program in early 1954. It was agreed, however, that the registration of births and the study of the sex-ratio and survival of the offspring should be continued, not only because of the intrinsic interest of these indicators but also with the thought that future genetic developments might bring new approaches to the cohorts which were being identified.
We then undertook the definitive analysis of these data which constitutes Chapter 3.
The results of the preliminary analysis were confirmed and extended. Further details concerning the pattern of congenital malformation in Japanese, as revealed by the at-birth and nine-months examinations, and the autopsy program in Hiroshima, are to be found in Neel (1958). A particular focus of the monograph which constitutes Chapter 3 was the exclusionary power of the data, i.e., the magnitude of the effect which could exist and go undetected, at various probability levels. Although on the basis of a burgeoning experimental literature it could not be doubted that the survivors of the bombings had sustained some genetic damage, our philosophy during this period was that it would be inappropriate to speculate on the magnitude of that damage in the absence of statistical significance. This decision was fortified by the relatively poor understanding at that time of the genetic (mutational) component in the various indicators under scrutiny.
Because marriages between cousins were relatively common among registrants in the program (6% in Hiroshima, 8% in Nagasaki), and because the individuals engaging in cousin marriages might not be randomly distributed in the two cities with reference to the hypocenter (a possibility which would introduce a source of bias into the data), we felt it best to exclude children resulting from such marriages from the analysis described in Chapter 3. With the main analysis completed, however, we then undertook a separate analysis of the much smaller sample of children born to cousins (Chapter 4), which analysis, within the limits of sample size, failed to suggest that these children exhibited any greater sensitivity to the genetic effects of radiation than the children of unrelated parents. These children of consanguinity, plus suitable controls, later became the subjects of a major study (Schull and Neel, 1965).
Periodically we encounter questions concerning (unspecified) pressures brought to bear on the study by the Atomic Energy Commission during these early days. Let us say very simply that we were never, either in Washington (where the pressure would be on the National Academy of Sciences) or in Japan, aware of any political pressure meant to influence the organization of the study or the way the data were analyzed. Problems there were, in the conduct of the study and the analysis of the data, but this was not one of them.
Phase II (1955-1968)
During this period, data collection was essentially limited to the sex-ratio of newborn infants in the two cities, and the survival of liveborn infants. The sex-ratio studies were thought to be important because of earlier, equivocal findings that the ratio had indeed been altered by parental exposure, whereas the continuing registration of liveborn infants (now through civil procedures rather than a pregnancy registry) ensured a proper population for study, should future events dictate a resumption of a hands-on program. In addition, an effort was initiated to determine if there was a relation between parental radiation histories and a simple battery of anthropometric measurements of Hiroshima schoolchildren obtained in 1965 in the course of an annual school examination program.
Among births occurring in Hiroshima and Nagasaki during this period, only some 15% involved parents who were thought to have incurred significant radiation exposures ATB. Since the periodic updating of the survival status of each liveborn child was time consuming, and there was a great excess of births to parents receiving no radiation ATB in the city samples, the study of survival was reorganized in 1958. Three cohorts of children were established at this time, namely, one comprised of all children born to survivors one or both of whom were “significantly” exposed (within 2,000 meters of the hypocenter ATB, the “proximally exposed”) and two age-sex-city matched control cohorts, one of children born to parents not in the city ATB, and the other to parents one or both of whom
were “distally exposed” (i.e., in either city but beyond 2,500 meters from the hypocenter ATB). Chapters 5, 6, and 7 present analyses based on these cohorts describing the findings on sex ratio and survival of liveborn infants through 1969. Chapter 6 is particularly important because it describes and compares in considerable detail these three cohorts, from which the subjects for studies yet to come were derived.
A word must be said about the assignment of radiation doses during this period. The analyses described in Chapters 3–6 are based on radiation categories established by us on the basis of distance from the hypocenter, shielding, and symptoms ATB. A potential weakness in the data was that the mother supplied the father's radiation history at the time she registered the pregnancy. It was not until the early 1960s that the Dosimetry Section of the ABCC had in hand detailed radiation histories on all significantly exposed survivors residing in these cities at the time of the 1950 Japanese national census, at which point a direct assessment of fathers' exposures became available. A codified approach to calculating doses for survivors, known as the Tentative 1965 Dosimetry System (T65D), was introduced in 1965 and was the basis for the analysis of Chapter 7. This dose schedule was replaced in 1986 by a system of assigning doses which is presumably definitive, the Dosimetry System, 1986 (DS86). It is noteworthy that the results of the analyses of phase III, employing the DS86 schedule, are in substantial agreement with the results of the analyses during phases I and II, employing other dose systems, although there was a substantial gain in statistical power in the transition from the early analyses based on group doses to the later analyses based on individual doses.
Phase III (1969–1990)
This phase of the study recognized both the emergence of important new technologies for the study of these cohorts of children and the growing insights into the genetic contribution of mutation in the parental generation to congenital defect, malignancy, and early death. The first of the studies to be initiated during this period, in 1968, dealt with the cytogenetic findings in a subset of the F1 Mortality Cohorts. Inasmuch as for practical reasons the necessary venous blood samples were not obtained until the child had attained 13 years of age, the findings are considered valid with respect to sex-chromosome aneuploidy and balanced chromosomal rearrangements, conditions in which survival is essentially normal, but not with respect to unbalanced chromosomal rearrangements or autosomal aneuploidy (Chapter 8). In 1975, after a three-year pilot study, a search was initiated for mutations resulting in electrophoretic variants of a series of 30 proteins of serum and erythrocytes, and in loss of activity in a series of 11 erythrocyte enzymes, again in a subset of the F1 Mortality Cohort sampled at age 13 or above (Chapter 9).
As knowledge of the genetic basis of the childhood malignancies improved, it was felt worthwhile to make the incidence of these malignancies in the cohorts being followed for survival the object of a special study (Chapter 11). Finally, the early data on untoward pregnancy outcomes were reanalyzed (Chapter 10), and the data on survival were brought up to date through 1985 (Chapter 12). The DS86 dose system mentioned earlier was available for the latter three analyses but not for the first two (the cytogenetic and biochemical studies). However, in the paper (next section) which explores the implications of the findings for the genetic doubling dose, the data of Chapters 8 and 9 have also been put in the context of DS86 doses.
AN ESTIMATE OF THE GENETIC DOUBLING DOSE OF RADIATION
The complex of data assembled in the course of these studies can first of all be viewed as an empirical exercise in evaluating the tangible risks to children of parental radiation exposure. In particular, we note that within the limits of the statistical error inherent in estimations of this type, they refute many of the unbridled and ill-advised speculations appearing in the past in the public presses of the world, speculations whose impact on the morale of exposed survivors the reader can well imagine. It would in our opinion, however, have been a breach of public trust to terminate the analysis at this point, especially in light of the explosive increase in recent years in knowledge concerning the mutational component of the various indicators employed in the study. In this connection, we note that the control data collected during phase III provide major input to our knowledge of spontaneous mutation rates with respect to certain types of chromosomal and protein variants.
None of the analyses of the data sets just mentioned yielded results suggestive of a significant difference between the children of exposed and of controls. The average combined gonadal doses of the exposed parents, however, approximated what on the basis of experiments with mice have been extrapolated to be the gametic doubling dose of acute radiation for humans. As these analyses unfolded, we began to sense a possible discrepancy between these findings on the children of survivors and the expectations based on projections from the mouse paradigm which had dominated the risk assessments of various national and international committees. There might be a message in the consistent lack of any notable difference between the cohorts under study.
On the grounds that we could accept the hypothesis that ionizing radiation produced mutations in humans, and that the corpus of data from Japan was the most substantial to be available in this century, we accordingly undertook a cautious exploration of the implications of the data for the human genetic doubling dose of ionizing radiation (Chapter 13). The latter is defined as the amount of acute or chronic (including low-level intermittent) radiation that will produce the same mutational impact on a population as occurs spontaneously each generation. We prefer this metric to any absolute calculation of the genetic impact of radiation because of the perspective which a doubling dose estimate supplies. In point of fact, doubling doses undoubtedly differ according to the genetic phenomenon under consideration; the estimate which was derived was an average. Undertaking to estimate a doubling dose represents a major evolution in our thinking concerning the most appropriate way to address these data.
The application of this approach to these data has not been easy. The procedures we have employed have no precedent. In the absence of the statistical difference between study groups, which is the traditional point of departure for most scientific propositions, we could just be manipulating the noise in the system. Furthermore, to develop the argument, we must make assumptions concerning the contributions which spontaneous mutation in the parental generation made under the conditions existing in post-war Japan to the complex of congenital defect, stillbirth, and death during the neonatal period, as well as to death during infancy and childhood. We needed also to devise the most appropriate way to combine results across studies, and to decide on the appropriate dose rate factor, in extrapolating from the results of acute radiation to what the results of chronic radiation might be.
The resulting estimate of the gametic doubling dose, of approximately 2 Sv for the acute radiation exposures experienced by the survivors of the bombings, and, employing a dose rate factor of 2, of 4 Sv for chronic or intermittent, low-LET radiation, is substantially
higher than several national and international committees have projected for humans on the basis of their evaluation of the experimental murine data. Inasmuch as for technical reasons discussed in Chapter 13 the estimate could not incorporate the data on physical development, sex ratio, and balanced chromosomal rearrangements, none of which indicators were suggestive of a radiation effect, the estimate is thought to be conservative. This apparent discrepancy between the human and mouse data led to an effort to reanalyze the existing published data on murine radiation genetics, together with the addition of some previously unpublished data, the better to understand the reason(s) for the apparent difference. In fact (Chapter 14), in our hands much of the discrepancy disappeared: the best estimate of the gametic doubling dose for the levels of acute radiation employed in the experiments on mice is approximately 1.4 Sv, and for chronic or intermittent, low-LET radiation, employing a dose rate factor of 3 for the extrapolation, in the neighborhood of 4.0 Sv. (The higher dose rate factor for the murine data is necessitated by the substantially larger gonadal doses employed in the experiments with mice.) For reasons that become apparent in Chapters 13 and 14, we find it impossible to assign the usual statistical limits to these two estimates, which in fact are not based on homologous material. We accept, however, that the error to be attached to both these estimates is wide, and both estimates may be subjected to considerable revision, which, however, seems unlikely to drop either estimate below 2 Sv for chronic, low-LET radiation.
It is hoped this anthology will mirror to a considerable degree the increasing sophistication of the study of human genetics during this 40-year period. Had the techniques that became available in the 60s and 70s been available in 1946, the study would have followed a very different course. We will return in the final chapter to what this continuing evolution in genetic technology suggests for future studies.
Inasmuch as even today the issues are occasionally confused, a clear distinction must be drawn between these studies and studies on the children in utero at the time of the bombings. The extensively documented microcephaly, mental retardation, and other abnormalities of these children (reviewed in Miller, 1956; Schull et al., 1989) reflect the well-known direct somatic (teratogenic) effects of radiation but not maternal genetic damage.
Even as these studies were unfolding, remarkable insights into the ways the genetic material protects itself against damage were developing. At the outset of these studies, virtually nothing was known about genetic repair mechanisms, nor about the array of endogenous and exogenous mutagenic influences to which DNA is “normally” exposed. Radiation appeared then as a unique insult; now it can be seen as one of many insults for which evolution has prepared DNA. We suggest that radiation—whose adverse somatic and genetic effects are of course clear—can now be regarded as only one of many disruptive forces with which DNA must cope and that the higher doubling doses which we are now postulating for mice and humans should be viewed as encouraging evidence of the robustness of the genetic material. This statement is not meant to encourage complacency in the face of the many biological challenges facing humans and the ecosystems in which they function but rather to contribute to the growth of perspective in setting ecosystem and exposure priorities.
There has been a keenly felt responsibility associated with the conduct of these studies. It is the essence of science that one investigator's results can be duplicated by others. But the bombings were a unique situation, and, because of both the post-war circumstances and the necessary scale of any program which purported to meet the demands of the situation, only a single comprehensive study has been undertaken. Furthermore, certain aspects of the study have involved carefully negotiated access to confidential vital statistics, with
great care to preserve personal anonymity. In principle, much of the data reported herein on morbidity and mortality could be reconstructed from government vital statistics, given access to the exposure data of the RERF, and the cytogenetic and biochemical observations could be independently repeated. In the real world, this is unlikely to happen. On the other hand, there is a precedent for the RERF making its data available to qualified investigators not associated with the organization, a precedent which the Foundation is of course prepared to meet with respect to the genetic data.
Finally, we would be remiss if we failed to acknowledge the remarkable cooperation that this study has enjoyed throughout its history. No study undertaken under these post-war circumstances could expect to escape criticism, both individual and organized, and the programs of the ABCC and RERF, including this one, are not exempt. In the final analysis, however, the ultimate test of the public's attitude toward a program is its cooperation. We note that the cytogenetic and biochemical studies, which, long after the Occupation had been terminated, involved obtaining venous blood samples from individuals 13 years or older (i.e., “free agents”), enjoyed 90% cooperation from the “children” and those of their parents still living in the cities at the time they were selected for study. Those familiar with such studies will realize what a remarkable record this is.
Haldane, J.B.S. 1955. Genetical effects of radiation from products of nuclear explosions . Nature176:115.
Miller, R.W. 1956. Delayed effects occurring within the first decade after exposure of young individuals to the Hiroshima atomic bomb. Pediatrics18:1– 18.
Neel, James V. 1958. A study of major congenital defects in Japanese infants. Amer. J.Hum. Genet.10:398–445.
Roberts, L. 1990. British radiation study throws experts into tizzy. Science248:24–25.
Schull, W.J., and Neel, J.V. 1965. The Effects of Inbreeding on Japanese Children. New York: Harper and Row. pp. xii and 419.
Schull, W.J., Otake, M., and Yoshimaru, H. 1989. Radiation-related damage to the developing human brain. In: Baverstock, K.F., and Stather, J.W. (eds.), Low Dose Radiation: Biological Basesof Risk Assessment. London: Taylor and Francis, pp. 28–41.