7
Implications for Risk Assessment

This report is generally concerned with the dosimetry of atomic-bomb survivors based on calculation, measurement, and biological approaches. It is appropriate to discuss some factors that could be relevant to the use of the dosimetry in, for example, risk assessment. Even in DS86, neutrons could play a part; if neutron fluence is higher in Hiroshima, as the discrepancy implies, the part played by neutrons could be larger. This section explores that aspect of the matter in an illustrative, rather than a definitive, way. It is not aimed specifically at risk estimation, which must embrace all aspects of epidemiological and dosimetric factors in both Hiroshima and Nagasaki.

CONCERNS ABOUT DS86

The apparent discrepancies between calculations and measurements of thermal neutrons in Hiroshima have led in recent years to concern that there might be substantial biases in risk estimates derived from the atomic-bomb survivor experience. It needs to be emphasized here that that concern is largely unfounded. Although any unresolved aspect of the atomic-bomb dosimetry adds to the uncertainty in risk estimates, the neutron discrepancy has only minor implications for the assessment of risks to survivors themselves. In line with the principal aim of the studies at RERF—the elucidation of the health effects among the survivors—it is reliably known from 5 decades of epidemiological follow-up what the specific radiation exposures in Hiroshima and Nagasaki have done and what past and continuing risk they have posed for the survivors (Thompson and others 1994; Pierce and others 1996). The increased cancer rates have been thoroughly studied in their dependence on distance from the hypocenter and on shielding. Estimates of risk



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 7 Implications for Risk Assessment This report is generally concerned with the dosimetry of atomic-bomb survivors based on calculation, measurement, and biological approaches. It is appropriate to discuss some factors that could be relevant to the use of the dosimetry in, for example, risk assessment. Even in DS86, neutrons could play a part; if neutron fluence is higher in Hiroshima, as the discrepancy implies, the part played by neutrons could be larger. This section explores that aspect of the matter in an illustrative, rather than a definitive, way. It is not aimed specifically at risk estimation, which must embrace all aspects of epidemiological and dosimetric factors in both Hiroshima and Nagasaki. CONCERNS ABOUT DS86 The apparent discrepancies between calculations and measurements of thermal neutrons in Hiroshima have led in recent years to concern that there might be substantial biases in risk estimates derived from the atomic-bomb survivor experience. It needs to be emphasized here that that concern is largely unfounded. Although any unresolved aspect of the atomic-bomb dosimetry adds to the uncertainty in risk estimates, the neutron discrepancy has only minor implications for the assessment of risks to survivors themselves. In line with the principal aim of the studies at RERF—the elucidation of the health effects among the survivors—it is reliably known from 5 decades of epidemiological follow-up what the specific radiation exposures in Hiroshima and Nagasaki have done and what past and continuing risk they have posed for the survivors (Thompson and others 1994; Pierce and others 1996). The increased cancer rates have been thoroughly studied in their dependence on distance from the hypocenter and on shielding. Estimates of risk

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) and probabilities of causation are not at issue. They are largely independent of the eventual resolution of the neutron discrepancy. The unresolved neutron problem is related to a second task of the investigators at RERF. The question is not what the radiation in Hiroshima has done to the survivors, but how it has done it. The issue is how much of the observed effect can be attributed to the different radiation components—the major absorbed-dose contribution by the gamma rays and the minor absorbed dose contribution by neutrons. Attempts to settle that question are not required primarily for improving the risk assessment for the survivors. They are needed to ascertain whether and how the observations in Hiroshima can be applied to the estimation of risk in populations exposed to radiation that differs from that in Hiroshima, e.g., consisting of a different mixture of gamma rays and neutrons or in being free of neutrons. As already said, such conclusions would be important to the people of the entire world. Therefore, although this report’s evaluation of the discrepancies between measurements and DS86 calculations should not, and does not, depend on how the resolution of the discrepancies might affect estimates of radiogenic risk, it is appropriate to consider in a preliminary way whether modifying the dosimetry system to reduce the discrepancies will greatly affect risk estimates. To be relevant to risk estimation, discrepancies between calculations and measurements of neutron fluence in DS86 must occur in a dose range where health effects due to the radiation exposure have been ascertained, and the neutron doses in this range must be large enough to contribute substantially to the observed effects. There is critical interest in the neutron fluences at distances from the hypocenter between 1000 m and 1500 m, which correspond in Hiroshima to mean organ doses of 0.2–2.0 Gy (Roesch 1987). The fluence values at 2000 m and those at less than about 1000 m are informative inasmuch as they can help to substantiate the values in the region of interest, and the doses below 0.2 Gy could become increasingly important since direct examination of the risk from doses below this level has been considered by Pierce and others (1996) and Pierce and Preston (2000). On the basis of the earlier dosimetry system, T65D, it had been surmised (Rossi and Kellerer 1974; Rossi and Mays 1978) that neutrons were responsible for a substantial fraction of the late health effects observed in Hiroshima. Support for that idea declined when DS86 specified considerably lower neutron doses in Hiroshima. It was then concluded that the neutrons are, even in Hiroshima, a minor potential contributor to the observed health effects and their role, although uncertain, is not critical for risk estimation. In later analyses, the neutrons were therefore accounted for crudely by applying a weighting factor of 10 to their absorbed-dose contribution. The sum of the gamma-ray absorbed dose and the weighted neutron dose was termed weighted dose and was expressed in sieverts. That approach seemed to confirm the relative unimportance of the neutrons in DS86 for risk evaluation. However, a more quantitative assessment of the actual doses is required to appreciate the situation. As we shall see in the illustration given here, a

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) more precise detailing of the neutron and gamma doses reveals a potentially greater role for neutrons, at a total dose of about 1 Gy, than previously envisioned. THE NEUTRON/GAMMA-RAY DOSE RATIO The diagrams in Figure 7–1 represent the neutron/gamma-ray dose ratio, (the ratio of average neutron absorbed dose to gamma-ray absorbed dose) plotted against the total absorbed dose in Hiroshima. The lower diagram refers to the bone marrow, the upper diagram to the colon. The solid lines show the relations according to DS86. The ratio for Nagasaki runs parallel to that for Hiroshima but is lower by a factor of 3. The dotted lines FIGURE 7–1 Ratio of neutron absorbed dose to gamma-ray absorbed dose in Hiroshima versus total dose. Solid curves correspond to the current dosimetry system, DS86; dotted curves to neutron doses increased in line with thermal-neutron activation data (Straume and others 1992); broken curves to intermediate adjustment that might be consistent with preliminary 63Ni measurements (Chapter 3).

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) give the relations originally proposed by Straume and others (1992) to account for the discrepancies between thermal-neutron activation measurements and the DS86 computations. In view of the DS86 computations and the present assessment of the activation data discussed in Chapter 3, including preliminary 63Ni data, this relation can now essentially be discounted. The broken lines represent a tentative, smaller modification that may not be inconsistent with the current evaluation of activation. In analyses of mortality from or incidence of all solid cancers combined, it has been usual to refer to the colon dose, that is the dose to the deepest, most highly shielded organ. For the gamma rays, the choice is not critical; organ doses for the solid-tumor sites (averaged in terms of the ICRP tissue-weighting factors) are only about 5–10% higher than the colon dose. For neutrons, the reference to the colon is unsatisfactory because averaging over all organs at risk results in a neutron absorbed dose that is roughly 1.9 times the neutron absorbed dose to the colon. The analysis that refers the neutron dose to the colon is thus biased toward a low predicted neutron contribution to observed health effects. The neutron dose to an average organ at risk for solid cancer is close to that in the bone marrow, and later considerations will therefore use data on bone marrow as an approximation that is also adequate with regard to all solid cancers combined. EFFECT OF THE NEUTRON CONTRIBUTION AS INFERRED FROM RBE Figure 7–2 shows a nonparametric representation of the excess relative risk (ERR) for solid-cancer mortality versus absorbed dose to the bone marrow in Hiroshima (adapted from Chomentowski and others 2000). At low doses, statistical imprecision makes it difficult to give a reliable value of the ERR; at doses close to 2 Gy, one recognizes some bending over of the curve that complicates any extra-polation to low doses. It is therefore reasonable to consider the total effect and the fractional effect due to neutrons at an intermediate total dose, which is chosen here to be 1 Gy. With the neutron/gamma dose ratio 0.0075 for the colon at 1 Gy total dose (see Figure 7–1) the weighted dose at 1 Gy is 1.07 Gy (derived from 0.9925+(10× 0.0075)), therefore the effect contributed by neutrons is about 7% (0.075/1.07). This low value confirms the common judgment, although it depends on two assumptions; first, the choice of 10 for the relative biological effectiveness (RBE) of neutrons compared to gamma rays, which some consider low, and second, the reference site (the colon), since it underestimates the dose contributions of the neutrons. Although the relatively high dose of 1 Gy might seem to be in line with fairly low values of the neutron RBE, it must be recognized that in DS86, at a 1 Gy total dose to the bone marrow in Hiroshima, the neutron dose is only about 15 mGy. That is in the lower range of neutron doses at which excess tumor incidence has

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) FIGURE 7–2 Nonparametric representation of the excess relative risk for solid tumor mortality in Hiroshima as a function of absorbed dose to the bone marrow (adapted from Chomentowski and others 2000). The gray shaded band indicates the standard error. The two lower curves show the effect contribution by fast neutrons that is inferred in terms of the relative biological effectiveness, R1, of neutrons against a gamma-ray dose of 1 Gy. The effect of the neutrons is proportional to the neutron dose, but due to increasing neutron/ gamma dose ratio it increases more than linearly with total absorbed dose. been determined in animal studies. Likewise, 1 Gy is typically near the lowest gamma-ray dose (0.5 Gy is about a minimum) at which an excess of solid tumors can be adequately measured in animal experiments. At 15 mGy (i.e., about 1 Gy gamma dose) the observed values of RBE tend to be fairly large. Results of animal studies vary, but 20 appears to be a lower value—inferred from life-shortening in mice (Carnes and others 1989; Covelli and others 1989), which has been used as a proxy for tumor incidence—and 50, consistently seen in a large series of experiments on tumor induction in rats (Lafuma and others 1989; Wolf and others 2000), appears to be a reasonable high value. The lower RBE value of 20 implies that 50 mGy of neutrons has the same effect, E1, as 1 Gy of gamma rays. Because of the linear dose dependence for neutrons, the 15 mGy of neutrons will contribute 0.3 E1 to the observed effect Eobs= (0.985+0.3) E1=1.285 E1 at 1 Gy total absorbed dose. Thus, the neutrons contribute 23% of the ERR observed at a 1 Gy total absorbed dose in Hiroshima, and

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) 77% is due to gamma rays. Likewise, one obtains with the upper RBE value of 50 against 1 Gy of gamma rays a neutron-effect contribution of 0.75 E1, of a total effect of Eobs=0.985+0.75 E1=1.735 E1; that is, the neutrons contribute 43% and the gamma rays contribute 57% of the effect. The major point is that regardless of where in the interval 20–50 the true RBE lies, there is a substantial contribution of neutrons (as indicated by DS86) to the effects at about a 1 Gy total absorbed dose in Hiroshima. COMPARISON OF NEUTRON EFFECTS IN HIROSHIMA AND NAGASAKI If the same calculations are made for Nagasaki with only one-third as great a neutron contribution as that in Hiroshima—5 mGy instead of 15 mGy at a 1 Gy total absorbed dose—then the calculations can be summarized as in the following table (Table 7–1). The results in Table 7–1 show that from a consideration of reasonable RBEs for neutrons, the effects to be expected for the same total absorbed dose of 1 Gy should be 18–40% higher in Hiroshima than in Nagasaki only because of the greater number of neutrons in Hiroshima, as calculated in DS86. Obviously, as noted below, increasing the neutron contribution in Hiroshima because of thermal-neutron (and fast-neutron) activation measurements would increase the ratio of effects at Hiroshima versus Nagasaki further. In fact, it is well known that intercity differences, seem to exist (Pierce and others 1996), with ERR estimates for Hiroshima 1.5–2.0 times as great as those for Nagasaki and excess absolute risk estimates 1.2–1.5 times as great as those for Nagasaki, on the basis of cancer-mortality data from 1950–1990. The complexity of the issues involved in determining intercity differences, especially a bias that might have led to the large ratio of ERR estimates, are described in some detail in Pierce and others (1996) and Pierce and Preston (2000). None of those observations leads to clearly significant differences. Although the intercity difference is evidently in the direction of greater neutron effects in TABLE 7–1 Neutron/Gamma Dose Calculations for Hiroshima and Nagasaki Total Dose Gamma-Ray Dose Neutron Dose Neutron RBE Total Weighted Dose Hiroshima/Nagasaki Effect Ratio for RBE Hiroshima 1 Gy 0.985 Gy 0.015 Gy 10 1.135 Gy 1.09 1 Gy 0.985 Gy 0.015 Gy 20 1.285 Gy 1.18 1 Gy 0.985 Gy 0.015 Gy 50 1.735 Gy 1.40 Nagasaki 1 Gy 0.995 Gy 0.005 Gy 10 1.045 Gy   1 Gy 0.995 Gy 0.005 Gy 20 1.095 Gy   1 Gy 0.995 Gy 0.005 Gy 50 1.245 Gy  

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) Hiroshima, it is impossible with the current information to ascribe differences in effects to the neutrons in DS86 alone and therefore to determine neutron RBEs from them. It is useful to note, however, that a neutron RBE of 50 would explain, with DS86 unmodified, the reported intercity difference. Taking the “speculation” in this illustration a step further, we can also note that the risk of all solid cancer and leukemia combined, 12% Sv−1 (UNSCEAR 2000), has been derived from the atomic-bomb survivor data with an assumed RBE of 10 and with reference to the colon dose. That corresponds to the assumption of an RBE of 5 with reference to the average organ dose and with an average 11 mGy of neutrons at a 1 Gy total dose in the combined sample; the effect at 1 Gy was taken to be Eobs=[0.989+5(0.011)] E1=1.044 E1. Using an RBE of 50, one would have Eobs=[0.989+50(0.011)] E1=1.539 E1. The effect, E1, of 1 Gy of gamma rays—that is the risk estimate—is thus reduced by the factor 1.044/1.539, that is, it is 8.1% Sv−1 (this will be applicable directly to acute gamma rays, although in ICRP-NCRP procedures it would be divided by a DDREF of 2 to become 4.1 % for the risk at low dose rates, slightly less than the accepted nominal value of 5.0% Sv−1 but well within the range of uncertainty for this estimate). IMPLICATIONS OF THE NEUTRON DISCREPANCY The potential implications of the neutron discrepancy—the postulate that thermal activation implies a substantially larger fast neutron component in Hiroshima than indicated by DS86 but the same fast neutron component in Nagasaki as indicated by DS86—can first be exemplified by a consideration of the large modification of neutron doses that was thought to be in line with the trend of the thermal-neutron measurements during the last few years. With the full modification of Straume and others 1992 (see dotted line in Figure 7–1) the neutron/gamma ray dose ratio at a 1 Gy total dose would be 0.055. In analogy to the above calculations, that would imply a neutron-effect contribution of 54% at a 1 Gy total dose if the RBE were 20 and a contribution of 74% if it were 50. Although the full modification of Straume and others now can probably be ruled out because of our evaluation of the recent fast-neutron measurements (see Chapter 3) with 63Ni, the example shows why the resolution of the discrepancy is important to appreciate how the risk must be apportioned between neutrons and gamma rays. The activation measurements that have already been performed provide general confirmation of DS86, at distances around 1000 m, corresponding to total doses close to 2 Gy. At 1 Gy an increase of the neutron dose—perhaps from 15 mGy to 20 mGy—is still possible, but a larger increase seems unlikely. However it is clear that any additions to the neutron component at Hiroshima will suggest a higher effect contribution by the neutrons. At smaller doses, a more substantial increase in the neutron/gamma ray dose ratio, possibly of 3–5 at 0.2 Gy, cannot be excluded and might indeed be the final outcome of fast-neutron measurements. That could have some impact on the risk

OCR for page 87
Status of the Dosimetry for the Radiation Effects Research Foundation (DS86) coefficient primarily for gamma rays (Kellerer and Walsh 2001), although uncertainties will inevitably be very large. There is some conflict between the general experience, in radiobiological studies, of upward curvature in the dose-effect relations and the finding that the dose dependence for solid tumors in Hiroshima and Nagasaki is seemingly linear. More curvature is, of course, indicative of a lower risk coefficient for gamma rays. Because more neutrons at low doses in Hiroshima would explain at least part of the seeming linearity of the overall dose dependence, it would also be consistent with a somewhat lower gamma-ray coefficient. In any case both the gamma-ray risk estimates and the neutron-risk estimates will depend, but probably not critically, on the successful resolution of the neutron discrepancy. In conclusion, it is probably already clear from the preliminary 63Ni measurements that the neutron discrepancy is smaller than at first thought and possibly within the range of uncertainties in the contribution of neutrons in Hiroshima and Nagasaki given by DS86. It is clear, however, that according to the illustration given here the effects of these neutrons, even at the DS86 level, are not negligible and, when allowed for, tend to lower the gamma-ray risk estimates slightly.