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.
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 34 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. excess incidence of transformations in the entire cell population that must be considered. In fact, one cannot do otherwise. In view of the long latent period between initiation and full expression of cancers and heritable diseases, it is not possible to link with certainty any specific cancer or genetic defect with any specific exposure, either before or after the development of a tumor or a genetic defect. Although the allegation that the specific exposure and the specified cancer are causally related implies that a single cell must have been stochastically (accidentally) hit, harmed, and transformed at the time of the exposure, there is no way to show that such a microaccident involving the specific radiation did in fact take place. The fact that approximately one out of five deaths per year in the United States is due to cancer of unknown origin and that about twice that number of persons per year develop cancers of unknown origin indicates how precarious such an allegation would be. If there is no way of knowing even whether a given cancer patient is in fact a casualty of the implied microaccident, then clearly it is not possible to evaluate the harm to any single cell and the chance of its transforming. A causal connection will probably remain impossible to establish and may thus represent one of the true trans-scientific problems, as so aptly termed by Alvin Weinberg (in this volume), because it is not amenable to resolution by scientific means. Since there is no causally relevant harm to evaluate either before or after a single-cell response, the special training and experience of the physician are of little or no value beyond certifying the presence or absence of a cancer or genetic defect and its type. In particular, it is not possible to establish cause and effect by the usual medical means of harm assessment. Therefore, the question of causation can be addressed only on a probabilistic basis. PROBABILITY OF CAUSATION IN CANCER CASES The probabilistic approach termed the probability of causation (Bond, 1981a; National Institutes of Health, 1984), although suggested some time ago (Bond, 1959; Oftedal et al., 1968), has only recently received wide attention. The recent increase in interest is due to the growing number of situations in which a causal relation between a cancer in either an exposed individual or a group has been claimed (Schaffer, 1984). Because the probability of causation (PC) method and its advantages and possible drawbacks have been described in detail elsewhere (Bond, 1982b; Cox, 1984; National Institutes of Health, 1984; National Research Council, 1984; Jablon, 1985), only the relatively simple principles involved will be presented here. Because risk coefficients are required for the PC method, these, and the risk of cancer for a given dose, are discussed first. This is followed by a
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 35 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. presentation of the PC method. In particular, this section discusses the circumstances under which the PC method provides a true probability that the single agent, radiation, was causally related to the observed effect, and the circumstances under which the possibility of multiple causation and thus of shared responsibility must be considered. The risk coefficient RC, derived from a population exposed to known amounts of radiation, is the expected excess incidence of a specified cancer per unit dose of radiation. The numerical value of RC is also termed the "risk" of cancer per unit dose for the "statistically average" individual in that or a similar population. An RC can be expressed either as a function of time, usually in years (for example, excess cases per 106 persons per year per rad) or as the total excess cases over a lifetime (for example, excess cases per 106 persons per rad). Table 2 contains examples of absolute risk coefficients for several cancers. The actual values for absolute risk are rough average values selected from the extensive data published by the National Institutes of Health (1984, Table VI-1- A, p. 76). The NIH document also provides values for relative risk. The total excess incidence, or average risk, for any given dose is simply the RC multiplied by the given dose. However, the use of human studies to derive RCs for "low-level exposure" involves extrapolation to the low-exposure region where statistical limitations preclude direct observation of an excess incidence, if it exists. Such extrapolations for low-LET radiations are conservative in the sense that values for the probability of causation so derived are upper limits. Thus, the consequent error in the PC is in the direction of favoring the individual plaintiff (see Whipple, in this volume). The risk coefficients and the risk from any given dose are prospective in that they permit the estimation, for a given exposed population or the average individual in that population, of the expected excess incidence and thus the risk. They do not address directly the retrospective situation in which a TABLE 2 Absolute Risk Coefficient for Selected Cancers: Approximate Values for Young to Middle-Aged Adults Type of Cancer Risk Coefficient (cases per 106 exposed per year, per rad) Leukemia (excluding chronic 1 lymphocytic) Esophageal 0.1 Stomach 0.4 Thyroid 1 Colon 0.25 SOURCE: National Institutes of Health (1984, Table VI-1-A, p. 76).
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 36 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. specific cancer is alleged to have been caused by a specified earlier exposure. The retrospective situation is addressed by the probability of causation method. In simplest form, the approach is described by the ratio where PC is the probability of causation, and RR is the expected excess incidence of cancer due to a radiation exposure of a population (equal to the risk to the average person); RB is the baseline normal or spontaneous incidence of cancer of a given type in that population; RO is the risk from other kinds of exposure (including that due to any additional radiation exposure other than that claimed to be causative); and RC is the risk from chemical carcinogens. The expected excess incidence RR is simply the absorbed dose in rads to the organ in which the specified tumor is presumed to have originated, multiplied by the risk coefficient for that tumor type. Although the PC formulation denotes absolute risk, the relative risk can be used instead without altering the principles involved (National Institutes of Health, 1984). Uncertainties with respect to radiation exposure exist in the dose, the baseline incidence, and the risk coefficients used to calculate a PC. Significant improvement in the first two measures is possible in principle, given the required expenditure of time and effort. Although better data in the third area, risk coefficients, are becoming available as current epidemiologic studies on exposed human populations progress and as the doses are refined, there is a limit to the extent that uncertainty can be reduced considering the hoped-for decline in the number of new, excessively exposed populations. Thus, Equation 1 will yield a fraction representing the probability that a specific exposure was causative. For instance, suppose that a 50-year-old male has developed a myelocytic leukemia, and that the exposure sustained at age 30 and claimed to be causative was 2 rads of X rays. The risk coefficient is approximately 1 Ã 10-6 per person, per rad, per year (Table 2), and the baseline leukemia incidence for an individual of age 50 is approximately 10 Ã 10-6 per year (Table 3). The probability of causation, assuming RO and RC are negligible, is then Consider first the low-level exposure regions. Here the linearity of the absorbed dose-quantal response function indicates independent action by each event involving a target, and thus complete genetic or carcinogenic
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 37 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. transformation of the cell as the result of a single event. That is to say, there is no measurable contribution from multihit action (the action of smaller hits from the same carcinogen, radiation, each ineffective in itself, combining to cause a quantal cell transformation). It is therefore particularly unlikely that a hit from a different carcinogen could have played a role in the initiation of any of these radiation-associated cellular transformations. Such results effectively rule out multiple causes for the initiation of a single cancer. In other words, the evidence is strong that, for low-level exposures, radiation and other possible causes of cancer are mutually exclusive. TABLE 3 Approximate Baseline Incidence of Selected Cancers in Young to Middle- Aged Adults Type of Cancer Risk Coefficient (cases per 106 exposed, per year, per rad) Leukemia (excluding chronic 70 lymphocytic) Esophageal 40 Stomach 100 Thyroid 40 Colon 350 SOURCE: National Institutes of Health (1984, Table VII-1, pp. 103â104). This deduction is supported by the results of studies on the combined effects of radiation and chemicals at low and high exposures. The results of one such investigation are shown in Figure 4 (Bond et al., 1984a), where there is a lack of synergistic action in the linear, low-level exposure region. The PC in this region can thus be taken to be a true probability that radiation alone is causally related. When cancer arises from low-level exposure, then, one logical approach to dealing with causation and liability is to assign responsibility for the cancer to the party under whose aegis the radiation exposure was experienced, but only if the PC exceeds 50 percent. An alternative might be to award compensation incrementally between a certain minimum level of probabilityâperhaps 20 percentâand 50 percent. Several other graduated schemes could be worked out. An upper-limit amount of compensation could be included for the highest PC, together with a "sole remedy" clause limiting adjudication to this solution. In any event, this approach lends itself to an administrative solution according to an agreed-upon plan, rather than the alternative of litigating each claim separately, in either a workmen's compensation or a tort claims forum. Let us consider next exposures well above the low-level exposure region but well short of the large-exposure region characterized by radiotherapy
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 38 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. (for an example see Figure 1, where exposures in the range of 100 rads were delivered in a time much shorter than that required for repair processes). Here the increasing nonlinearity of the dose-response curve indicates cooperative action between radiation hits that are too small separately to cause a mutation or carcinogenic transformation. Because of this multihit characteristic, one such hit could conceivably be contributed by a chemical mutagen or carcinogen that may be present. Figure 4 Synergism between the effects of radiation and those of a chemical mutagen, demonstrable in Tradescantia only in the larger-dose multihit region. Note that in the low-dose single-hit region, the combined response is no more than simply additive and may be antagonistic or mutually protective. That such combined-effect action is possible is also shown in Figure 4, for the case in which the Tradescantia cells were exposed both to radiation and to the chemical mutagen ethylmethanesulfonate (EMS), an alkylating agent (Bond et al., 1984a, 1984b). Note that synergistic action is apparent in that the response to the combined action of the radiation and the chemical mutagen clearly exceeded the sum of the responses to the agents administered separately. When PC is calculated for this high-level region, one cannot exclude the probable contribution of the combined action of both agents in the shared mode. It is reasonable to infer further that there was multiple
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 39 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. responsibility for the cancer. This interpretation could be important, especially if additional and simultaneous exposure to a chemical carcinogen or mutagen could be demonstrated. Here also, harm that is detectable in an organ at such relatively high exposures could be relevant only to assessing the severity of the acute organ damage in the individual who was dosed by an agent or agents, and not to the evaluation of possible cellular transformations that might be relevant to cancer. Only the statistical approach is adequate to deduce either a true probability of causation or an indication of the relative contribution of two or more agents. Current evidence is that, for low-level exposures, multiple causative agents for a given tumor are most unlikely. This indicates that the PC in this low-level exposure region is a true probability that a single agent (radiation) was causative, and that it does not represent an assignable share of the causation. As with considerations of cancer alleged to be caused by a single agent, medical knowledge alone does not equip one to deal quantitatively with questions of multiple causation. Medical training and experience are invaluable when dealing with the cause(s) of a quantal response in the context of demonstrable harm or competing injuries. However, such training is of little or no value in determining the causative role of a particular risk or of competing risks. It should not be inferred from the above, however, that multiple, essentially simultaneous exposures to carcinogens or mutagens occur frequently, nor that synergism is frequent. Of two mutagenic and carcinogenic compounds tested recently (Bond, 1984), one (EMS) was found to be synergistic with radiation. Even that compound showed a cooperative effect only with large exposures, and the factor was no more than about 2. A 1982 report of UNSCEAR indicates strongly that, although further experimental study of possible synergistic effects is needed, there is at present little evidence that combined effects will emerge as a serious problem. To quote directly from the report: For humans in environmental circumstances the Committee has been unable to document any clear case of synergistic interaction between radiation and other agents, which could lead to substantial modifications of the risk estimates for significant sections of the populations. Presumably this is due to the fact that most of the agents likely to act synergistically with radiation, as judged by the results of animal experiments, are not found in sufficient concentration in nature. A specific exception is the case of tobacco smoke, which raises essentially problems of industrial hygiene in some working environments [UNSCEAR, 1982, p. 762]. This evaluation implies that, for exposures to more than one carcinogen, the PC for each exposure can be evaluated separately. That is to say, RC in Equation 1 can, in most instances, be brought to the numerator in place of
CAUSALITY OF A GIVEN CANCER AFTER KNOWN RADIATION EXPOSURE 40 original typesetting files. Page breaks are true to the original; line lengths, word breaks, heading styles, and other typesetting-specific formatting, however, cannot be About this PDF file: This new digital representation of the original work has been recomposed from XML files created from the original paper book, not from the retained, and some typographic errors may have been accidentally inserted. Please use the print version of this publication as the authoritative version for attribution. RR , and the corresponding PC can be evaluated as for radiation. If there is in fact synergism between two carcinogens, then an assignment of shared responsibility must be considered. If radiation is one of the carcinogens, however, the shared mode would be considered for high-level but not for low- level exposure. An estimate of the PC is much more powerful than is the dose alone either for screening cases or determining more definitive estimates of the likelihood of causality. This follows from the dependence of the PC on additional variables such as the baseline incidence and the RC, either of which can vary substantially, both innately and with age and sex. It is clear from Equation 1 that PC is greater with large values for dose and RC and smaller with large values for the baseline incidence. As an example of the usefulness of the method with large doses of radiation delivered accidentally, consider the Marshallese exposed to as much as 175 rads of penetrating gamma radiation. Out of a total of about 75 high-dose individuals, one male exposed in early childhood developed myelocytic leukemia at age 18. The immediate, and correct, conclusion would be that one case, particularly in a small population, permits no positive deductions with respect to the leukemogenic potential of 175 rads of gamma radiation. Moreover, the product of the dose and the RC yields an annual risk of the order of 10-3, or 10-2 at most, which might also be construed as a basis for dismissing a causal relationship. However, the PC calculated from this dose and from information of the kind provided in Table 1 is about 80 percent. The figure should be larger because the RC in Table 1, estimated for older individuals and for lower doses and dose rates, should be increased by a factor of 3 or more. Thus, few would disagree with a decision to handle responsibility for the leukemia on the basis that the dose of 175 rads was in fact causative. Near the other extreme, consider the worker exposed to radiation under the controlled conditions of a nuclear power plant or laboratory. From actual recent experience (UNSCEAR, 1982), the large majority of such individuals will probably not receive more than 5 to 10 rads in a working lifetime. If such a worker should develop cancer of the colon, the probability of causation would be of the order of 0.1 percent. Should the cancer be rarer and have a larger risk coefficient, the PC would be larger (with leukemia, about 8 percent). If the threshold PC for a judgment of causality were near 50 percent, then few people would assume that the radiation was involved causally with either cancer. Clearly, in situations where the PC is in the range of 50 percent, the values used in the calculation of PC would come under closer scrutiny. A graded rather than a threshold approach to compensation should reduce the pressure to make fine distinctions.
