Conclusion

Chapter 7: Conclusion

At the same time, however, the scientific community's experience with radionuclides in humans was limited, and this approach to medical investigation was new. Although the available data about human risk were encouraging and the biological susceptibility of children to the effects of radiation was not appreciated, we are left with the lingering question of whether investigators and agency officials were sufficiently cautious as they began their work with children. This is a difficult judgment to make at any point in the development of a field of human research; it is particularly difficult to make at forty or fifty years' remove. Investigators and officials had to make decisions under conditions of considerable uncertainty; this is commonplace in science and in medicine. Although the biological susceptibility of children was not then known, investigators and officials held the view that children should be accorded extra protection in the conduct of human research, and they made what they thought were appropriate adjustments when using children as subjects. If human research never proceeded in the face of uncertainty, there would be no such experiments. How little uncertainty is acceptable in research involving children is a question that remains unresolved. Today, we continue to debate what constitutes minimal risk to children, in radiation and in other areas of research. The regulations governing research on children offer little in the way of guidance, either with respect to conditions of uncertainty about risk or when risks are known.

As best as we can determine, in eleven of the twenty-one experiments we reviewed, the risks were in a range that would today likely be considered as more than minimal, and thus as unacceptable in nontherapeutic research with children according to current federal regulations. It is possible, however, that four of the eleven might be considered acceptable by the "minor increase over minimal risk" standard.[98] In these four experiments, the average risk estimates were between one and two per thousand, the studies were directed at the subjects' medical conditions, and they may well have had the potential to obtain information of "vital importance."

Physical risk to subjects is not the only ethically relevant consideration in evaluating human experiments. With the exception of the studies at Fernald, we know almost nothing about whether or how parental authorization for the remaining nineteen experiments we reviewed was obtained. And with the exception of the Fernald studies and the experiment at Wrentham, we know very little about the children who were selected to be the subjects of this research. Therefore, we cannot comment on the general ethics of these other experiments.

The experiments at Fernald and at the Wrentham School unfairly burdened children who were already disadvantaged, children whose interests were less well protected than those children living with their parents or children who were socially privileged. At the Fernald School, where more is known, there was some attempt to solicit the permission of parents, but the information provided was incomplete and misleading. The investigators successfully secured the cooperation of the children with offers of extra milk and an occasional outing--incentives that would not likely have induced children who were less starved for attention to willingly submit to repeated blood tests.

One researcher speaking almost thirty-five years ago set out the fundamental moral issue with particular frankness and clarity:

. . . we are talking here about first and second class citizens. This is a concept none of our consciences will allow us to live with. . . . The thing we must all avoid is two types of citizenry.[99]

It might have been common for researchers to take advantage of the convenience of experimenting on institutionalized children, but the Committee does not believe that convenience offsets the moral problems associated with employing these vulnerable children as research subjects--now or decades ago.

The Vanderbilt Study

The radioiron portion of the nutrition study, directed by Dr. Paul Hahn, was designed to study iron absorption during pregnancy.[c] The women, who were anywhere from less than ten weeks to more than thirty-five weeks pregnant, were administered a single oral dose of radioactive iron, Fe-59, during their second prenatal visit, before receiving their routine dose of therapeutic iron.[d] On their third prenatal visit, blood was drawn and tests performed to determine the percentage of iron absorbed by the mother. The infants' blood was then examined at birth to determine the percentage of radioiron absorbed by the fetus. The doses to the women were estimated in the study article, using crude dose-estimation methods available at the time, to be from 200,000 to 1,000,000 countable counts per minute.[e] Although the investigators did not estimate doses to the fetuses in the original study, Dr. Hahn later estimated fetal doses to be between 5 and 15 rad. This estimate, however, has been questioned.[f]

There is at least some indication that the women neither gave their consent nor were aware they were participating in an experiment. Vanderbilt study subjects, expressing bitterness at the way they believed they had been treated, testified at an Advisory Committee meeting that the proffered drink, called a "cocktail" by the investigators, was offered with no mention of its contents. "I remember taking a cocktail," one woman said simply. "I don't remember what it was, and I was not told what it was."[g] Although it is not clear what, if anything, the subjects were told, information about the Vanderbilt experiment was available to the general public. In late 1946 news reports appeared in the Nashville press.[h]

The actual risk to the fetuses in the Vanderbilt experiment has long been a matter of study. In 1963-1964, a group of researchers at Vanderbilt found no significant differences in malignancy rates between the exposed and nonexposed mothers.[i] However, they did identify a higher number of malignancies among the exposed offspring (four cases in the exposed group: acute lymphatic leukemia, synovial sarcoma, lymphosarcoma, and primary liver carcinoma, which was discounted as a rare, familial form of cancer). No cases were found in a control group of similar size, and approximately 0.65 cases would have been expected on Tennessee state rates, compared to which the three observed cases is a marginally significant excess. This led the researchers to conclude that the data suggested a causal relationship between the prenatal exposure to Fe-59 and the cancer. The investigators also concluded that Dr. Hahn's estimate of fetal exposure was an underestimation of the fetal-absorbed dose.

A 1969 study, funded by the AEC and conducted by one of the investigators from the 1963-1964 study, attempted to reconstruct the doses of Fe-59 to the fetuses in the original Vanderbilt study.[j] The investigators observed that the one case of leukemia might have been due to radiation damage, but that the doses in the other two cases were low; therefore, the relationship between the radiation exposure and the cancer in those cases might not have been causal. However, the researchers also noted that due to incomplete data, they could not estimate the dose absorbed by the fetus with confidence and that no definitive conclusions could be drawn from this study as to whether these exposures resulted in damage to the fetus.[k]

The Vanderbilt study raises many of the same ethical issues as the experiments reviewed in this chapter. Like these experiments, the Vanderbilt study offered no prospect of medical benefit to the pregnant women or their offspring, raising the question of the conditions under which it is acceptable to put children at risk for the benefit of others, whether before or after birth. What could the investigators reasonably have been expected to know about the risks to which they put their subjects? Did they exercise appropriate caution in exposing fetuses to radiation? What were the pregnant women told, if anything, and was their permission sought? Who were these women, and how were they positioned relative to pregnant women, generally?

The Committee did not have the resources to pursue these questions in both research in which children were the subjects and research in which children were exposed as fetuses. We did establish that the Vanderbilt study was not the only experiment during this period to expose fetuses in research that offered no prospect of medical benefit to them or their mothers. While the Committee did not conduct an exhaustive review of the scientific literature, we did find twenty-seven human radiation studies that included pregnant or nursing women as subjects between 1944 and 1974.[l] Of these studies, eight were considered therapeutic, and nineteen offered no prospect of benefit to the subject. Most of the nineteen were tracer experiments.

These studies were performed in order to examine human physiology during pregnancy or to study the uptake of radioactive substances by fetuses or nursing infants.[m] They generally addressed valid scientific questions that could not be investigated in other populations. Knowledge of fetal exposure to radioiodine, for example, was relevant to issues such as potential harm to the fetus from maternal uptake of radioiodine in diagnostic tests or to estimate the potential effects of environmental exposure to radioiodine on the human fetus. In other studies, radioactive iron was administered to better understand the physiology of maternal and fetal intake of iron during pregnancy.

m . Of the nineteen tracer experiments (funded by the government) involving pregnant or nursing women identified by the Committee, only three administered tracer doses to nursing women that offered no prospect of benefit; in at least one of the studies the infants were exposed. In one case, six nursing women were given radioiodine to determine excretion in breast milk, the infants were not given the exposed milk. In another case, two infants were intentionally exposed to the breast milk of their mothers, who were given I-131. An I-131 tracer study on the general population, incidentally included two nursing women. The report indicates that both had been nursing their children, and since there is no indication that the mothers were warned to avoid breastfeeding after the exposure, it is quite probable that the infants were exposed.

Nasopharyngeal Irradiation

An average of 150 patients a month, mostly children, were given the treatment at the Johns Hopkins clinic over a period of several years.[d] Many children received the treatment more than once as recurrent lymphoid tissue was considered an indication for treatment.

Crowe and his colleagues reported that the results following irradiation of the nasopharynx alone were not only as good as, but often better than, those following removal of tonsils and adenoids.[e] In review articles, they noted that approximately 85 percent of treated patients responded with decreased numbers of infections and/or improved hearing when treated at young ages. They also concluded that "it is effective, safe, painless, inexpensive and has proved particularly valuable for prevention of certain ear, sinus and bronchial condition in children."[f]

Although early articles by Crowe and colleagues indicate that nasopharyngeal radium treatments were accepted as standard procedure for the prevention of childhood deafness, these treatments, like most standard interventions in medicine, had not been subjected to formal scientific evaluation. A controlled study was conducted from 1948 to 1953 by Crowe and his colleagues to determine "the feasibility of irradiation of the nasopharynx as a method for controlling hearing impairment in large groups of children associated with lymphoid hyperplasia in the nasopharynx; to draw conclusions concerning the per capita cost of such an undertaking as a public health measure."[g] Crowe et al. wrote in an NIH "Notice of Research" that "the procedure of treatment is

not new, as an individual measure; this is the first adequately controlled experiment of sufficient size for accurate statistical analysis."[h]

This work was funded by NIH for the entire period of study. As recorded in an NIH grant application, the study involved approximately 7,000 children screened for hearing impairment.[i] Of those screened, approximately 50 percent were selected for further study based on the chosen criteria for hearing loss. Half of this study group was irradiated with radium, while the other half served as a control group. Crowe and colleagues reportedly concluded from this study (published in 1955) that the radium treatments did shrink swelling of lymphoid tissue and improve hearing.[j] This type of therapy was ultimately discontinued because of newly available antibiotics and the use of transtympanic drainage tubes, as well as awareness of the potential risks of radiation treatment.

In addition to the targeted lymphoid tissue, the brain and other tissues in the head and neck region, including the paranasal sinuses, salivary glands, thyroid, and parathyroid glands are also exposed to significant doses of radiation during the radium treatments, prompting concern that these treated individuals might have been placed at increased risk for radiation-induced cancers at these sites. Dale P. Sandler et al., in their 1982 study of the effects of nasopharyngeal irradiation on excess cancer risk for children treated at the Johns Hopkins clinic, found "a statistically significant overall excess of malignant neoplasms of the head and neck among exposed subjects," based however on only four cases in comparison with 0.57 expected.[k] This excess was accounted for mainly by three brain tumors that occurred in the irradiation subjects. One other malignant tumor, a cancer of the soft palate, was also reported. The Department of Epidemiology at the Johns Hopkins University has undertaken a further follow-up study of the Crowe et al. cohort of children irradiated there, previously studied by Sandler et al.[l] Verduijn et al., in their 1989 study of cancer mortality risk for those individuals (mostly children) treated by nasopharyngeal irradiation with radium 226 in the Netherlands, reported that "the present study has found no excess of cancer mortality at any site associated with radium exposure by the Crowe and Baylor therapy. Specifically,

the finding of Sandler et al. of an excess of head and neck cancer was not found in this study group."[m]

Among the Japanese atomic bomb survivors, no excess of brain tumors was found. However, several studies have noted an increased risk of both benign and malignant brain tumors following therapeutic doses of radiation to the head and neck region during childhood.[n] From the Committee's own limited risk analysis of these experiments, we concluded that the brain and surrounding head and neck tissues would be put at highest risk and estimated the lifetime risk at approximately 4.35 per 1,000 and an increased relative risk of 62 percent.[o]

The Hopkins nasopharyngeal study raises different ethical issues than those posed by the other experiments reviewed in this chapter, all of which offered no prospect of medical benefit to the children who served as subjects. By contrast, the nasopharyngeal irradiation experiment was designed to determine whether children at risk for hearing loss would be better off receiving radiation treatments or not receiving such treatments. A central issue here was whether it was permissible to withhold this intervention from "at risk" children. The application of radium was at this point a common, but scientifically unproven, treatment for children at risk of hearing loss; the risks of the treatment were not well characterized. If it was really unknown which was better for children--receiving radium or no intervention--then the medical interests of the children were best served by being subjects in the research because, as a consequence, they would have a 50 percent chance of receiving the better approach. The nasopharyngeal experiment thus belongs to a class of research the Committee did not investigate--therapeutic research with children.

The local gamma dose to the head and neck region was assumed to be distributed according to an inverse square law d(r) = 690/r2 rad. The Committee approximated the exposed region of the body by a sphere with radius 10 centimeters. This was felt to be a conservative assumption, because although the dose does not go to zero at the base of the neck, a 10-centimeter sphere would also extend outside the skull. Averaging this dose distribution over the exposed sphere, the average dose to the head was found to be 20.7 rad. The exposed volume is about 4189 cm3, or 29 percent of the total body, so the average whole body dose is about 6.0 rad. Multiplying this by the BEIR V risk coefficient for children exposed at age five, 1.4/1,000 person-rad, produces a lifetime risk of about 8.4/1,000. This calculation assumes that the brain and other head tissues have average radiosensitivity. BEIR V also gives absolute-risk coefficients for brain cancer ranging from 1 to 9 per million person-year-rad, with 3 being a reasonable average. Applying this figure to an average head dose of 20.7 rad, the Committee estimates a lifetime risk of about 4.35/1,000. The corresponding relative risk coefficients average about 3 percent per rad, so this dose would correspond to an excess relative risk of 62 percent.

Table 2. Summary and Risk Analysis for Studies Examined by the Advisory Committee

* Risk estimates are reported as average values for each experiment; maximum values ( ) are reported when available.