Dosimetry: From Bomb, to Kerma, to Internal Organ-Dose
This chapter is divided into five parts:
Sources of Ionizing Radiation from an Atomic Bomb, p.1 The Relationship of "Kerma" with Internal Organ-Dose, p.1 Comparison of T65DR and DS86 Dosimetries, p.3 Uncertainties about Doses from Weapon-Yield, Fallout, and Activation Products, p.3 Choice of RBE Values for the Neutron Component, p.5
No quantification of cancer-risk from low-dose ionizing radiation can begin, of course, without an estimate of the mean dose received by the internal organs from which the fatal cancers arise. What is needed is an estimate which can be reasonably used for all internal organs.
We show exactly how such estimates are obtained in Chapter 9 for the T65DR cohorts (1950-1982). Then in Chapter 10, we show exactly how they are obtained in the DS86 dosimetry for the same cohorts -- for the same people.
The foundation for those two chapters is laid by this chapter.1. Sources of Ionizing Radiation from an Atomic Bomb
Kerr and colleagues provide a handy summary of the sources of neutrons and gamma rays from an atomic explosion (Kerr87b). From their Table 1, we have made the listing below.
- Prompt Neutrons from fission (weapon still intact and still able to sustain the fission process).
Time Emitted after detonation: < one microsecond.
- Delayed Neutrons emitted by the fission-products (after explosion and burn-up of the weapon).
Time Emitted after detonation: < one minute.
- Prompt Gamma Rays from the fission process itself.
Time Emitted after detonation: < one microsecond.
- Secondary-Origin Gamma Rays resulting from the interaction of neutrons with the weapon itself, with air, or with ground.
Time Emitted after detonation: From < one microsecond (for gammas resulting from neutron interaction with the weapon) to 0.2 seconds (from neutron interaction with air or ground).
- Activation Gamma Rays, produced by neutron-activation of ground or other materials.
Time Emitted after detonation: Initial, 0.2 seconds to 1 minute. Residual, 1 minute to years.
- Delayed Gamma Rays from fission-products.
Time Emitted after detonation: Initial, 0.2 seconds to 1 minute. Residual, 1 minute to years.
Kerr points out, "Because the fission products are contained in the fireball formed after the weapon explodes and because the rapidly-rising fireball reaches an altitude of about two miles (3000 m) by the end of one minute, irradiation by delayed neutrons and gamma rays soon ceases at ground level" (Kerr87b).
At Hiroshima and Nagasaki, the extremely early gamma rays and neutrons, transported from the explosion-region outward, dominate the doses received by survivors who were located at various distances from the hypocenter of the bomb. Thus, the exposure of survivors is properly regarded as acute exposure.2. The Relationship of Kerma and Internal Organ-Dose
RERF Technical Reports use terms whose meaning is fairly constant: Kerma, free-in-air kerma, environmental transmission-factor, shielded kerma, body-transmission/ organ-absorption factor, and internal organ-dose.
Kerma is the acronym for "kinetic energy released in material" (Elle87, p.6). For a discussion of "kerma," we can rely on William C. Roesch, editor of RERF's Volume 1 about the DS86 dosimetry (Roes87). He states the following, in the Editor's Note on an unnumbered page before its table of contents (italics are in the original):
"The object of the reassessment program is to determine the absorbed dose (or simply the dose) in certain organs of the people exposed to the bombs. For practicality, determination of a quantity often approximately equal to the dose, the kerma, is usually made instead. The concept of absorbed dose deals with the energy imparted by ionizing radiation to a medium per unit of mass. `Energy imparted' means the difference in energy of the particles and quanta entering and leaving a small test volume. For the particles and quanta encountered in A-bomb dosimetry, the energy difference for photons and neutrons equals the energy they give to the charged particles they produce by interactions in the volume. This difference, per unit mass, is the quantity called the kerma."
There is more to it than that -- much more -- and for an advanced discussion of the exact relationship of kerma to dose, Roesch refers to an earlier work (Roes68).
However, his Editor's Note provides explicit assurance that, in most of the organs of interest in A-bomb dosimetry, "the kerma gives a sufficiently accurate approximation to the dose" (Roes87).
Free-in-Air Kerma :
In physics, it is common to estimate quantities (such as dose) for some imaginary, "ideal" circumstances. The term "free-in-air kerma" is one such term. Roesch, still in the Editor's Note, defines the term as follows:
"For example, in this report, it is common to determine the kerma that would be produced in tissue by the radiations at a point in air. One such condition is used so often that its kerma is given a special name: the kerma-in-tissue at a point in air over bare ground (i.e. no person present and not in or near a building) is called the free-in-air (FIA) kerma or the free-field kerma."
The ability to compute the dose to a gram of flesh suspended in air without a body present, at a given distance from a given explosion, is important, but still it is just an early step on the way to computing something which counts -- say, the average dose to the intestines of a whole, irradiated person who was shielded by a wooden house, or by a concrete factory, or by a hill.
Organ-Dose in Relation to Kerma and Transmission-Factors :
In TR-9-87, (p.6), Preston and Pierce devote a paragraph to terms which we shall quote verbatim (their underlining is shown in bold by us):
"The tissue kerma in air at specific locations unadjusted for the effect of shielding by structures or terrain will be called the free-in-air (FIA) kerma. The tissue kerma in air at the survivor location after adjustment for the effects of shielding by structures or terrain will be called kerma. Organ dose will be used to refer to the mean absorbed dose for specific organs. The terms total kerma and total organ dose will be used to mean the sum of gamma-ray and neutron kerma or organ dose, respectively. In analyses which make use of assumed or estimated RBE values the level of radiation is expressed in terms of dose equivalent. FIA kerma, kerma, and organ dose is [sic] given in SI units, gray (Gy) or milligray (mGy), while dose equivalent is expressed in sieverts (Sv) or millisieverts (mSv). The ratio of kerma to FIA kerma will be called the environmental transmission factor while the ratio of organ dose to kerma will be referred to as the body transmission/organ absorption factor."
In this book, we shall abbreviate the last term as "body transmission-factor."
Shielded Kerma :
Elsewhere in RERF reports, what is defined above as "kerma" is sometimes explicitly called "shielded kerma."
Absorbed Dose :
In some other RERF reports -- for instance, in discussion of doses from fallout in the DS86 book, Volume 1 -- the term "absorbed dose" is often used. The meaning there is explicitly an average internal organ-dose (Roes87, p.224).
Summary -- Two Modifiers for Kerma, to Obtain Organ-Dose :
When RERF reports provide values for "shielded kerma" or just "kerma" (which is the same), the kerma values need two kinds of modification in order to obtain the corresponding internal organ-doses.
One adjustment is for attenuation of both neutron and gamma dose by the body (the appropriate body transmission-factors). The other adjustment is only for the neutron component of the dose; it is the application of the factor which converts doses from high-LET neutrons to their dose-equivalents in low-LET units, according to the higher presumed Relative Biological Effectiveness (RBE) of neutrons.3. Comparison of T65DR and DS86 Dosimetries
Neutron FIA Kerma at Hiroshima :
"Hiroshima DS86 FIA neutron kerma estimates are about 10 % of the T65D estimates at all ground ranges" (TR-9-87, p.9; at page 7, TR-9-87 says that it means T65DR whenever it says T65D).
Neutron FIA Kerma at Nagasaki :
In Nagasaki, neutrons were unimportant even in the T65DR dosimetry. In the DS86 dosimetry, there is "a reduction of about 30 % in the neutron kerma" (TR-9-87, p.9).
Gamma FIA Kerma at Hiroshima :
"The DS86 FIA gamma kerma estimates in Hiroshima are larger than the corresponding T65D estimates. This difference increases with distance. Thus at a distance of 700 m from the hypocenter the ratio of the DS86 FIA gamma kerma estimate to the T65D value is 1.3 while at 2,000 m the ratio is 3.7" (TR-9-87, p.9).
Gamma FIA Kerma at Nagasaki :
The DS86 system introduces "little change" in the FIA gamma kerma for this city (TR-9-87, p.9).
Environmental Transmission-Factors :
According to TR-9-87 (p.9-10), the factors for neutrons changed very little in DS86 from their T65DR values, but "the average DS86 environmental transmission factor for gamma radiation is about 50 % of that for T65D in Hiroshima and about 60 % in Nagasaki" and this change "has a dramatic impact on gamma kerma estimates."
It should be noted that "the current DS86 system" appears to cope with shielding by wooden houses, but not with shielding by concrete factories or by all types of terrain (TR-9-87, p.9,10,12). Factories and terrain account for the omission of about one thousand of the Nagasaki survivors, in Dose-Groups 4 and 5, from the current DS86 dosimetry (TR-9-87, p.12). Such omissions were shown in detail in our Table 5-B.
Gamma Shielded Kerma at Hiroshima :
"In Hiroshima the average DS86 gamma kerma estimate is less than the T65D estimate for ground ranges of less than about 1,300 m and greater than the T65D estimate at larger ground ranges" (TR-9-87, p.10).
Gamma Shielded Kerma at Nagasaki :
In Nagasaki, at all distances, the DS86 gamma kerma is lower than it was in the T65D dosimetry (TR-9-87, p.10).
Neutron Shielded Kerma in DS86 :
"The ratio of neutron to total kerma increases smoothly with kerma from less than 1 % to about 6.5 % in Hiroshima and from less than 1 % to about 1.5 % in Nagasaki" (TR-9-87, p.10).
Body Transmission-Factors, Gamma Dose :
Preston and Pierce choose the dose to the large intestine (colon) as "a representative dose" for internal organs in general (TR-9-87, p.3). They report that "For intestinal doses, the DS86 body transmission/organ absorption factor is 80 % larger than the T65D factor" (TR-9-87, p.13). We agree with Preston and Pierce that this change in body transmission-factors for gamma dose is "a large effect"; it almost doubles the organ-dose from any particular shielded gamma kerma.
A comparison of old and new body transmission-factors for neutrons as well as gammas is provided in the next chapter by our Table 9-A.
The Net Effect on Organ-Doses :
It has been noted by others (TR-9-87, p.3; Fry87, p.845) that several aspects of the dosimetry-revision tend to cancel each other out -- to offset each other. For instance, in Hiroshima, organ-doses from neutrons decrease in DS86, but organ-doses from gamma rays increase -- especially in the low-dose classes.
The net effect of the new dosimetry (current version), on organ-doses in each RERF Dose-Group, can be seen from the next two chapters by comparing Table 9-C with Table 10-E.4. Uncertainties about Doses from Weapon-Yield,
Fallout, and Activation Products
Anyone who reads Volume 1 of RERF's DS86 book (Roes87) will realize that the dosimetry of the A-bomb survivors will always be approximate -- which is typical of just about every human study in the field of ionizing radiation.
The DS86 dosimetry leaves many, many questions unsettled, including the yield of the Hiroshima bomb -- which remains uncertain with a range of 12-18 kilotons (Elle87, p.8-9). Also unresolved is the question of doses from "residual radioactivity" (neutron-activation products and fallout).
Activation Products :
RERF's DS86 book points out that the creation of activation products, resulting from neutron-capture by substances in or above the ground, decreases as distance from the hypocenter increases. Near the hypocenter at Hiroshima, activation products produced an upper limit for absorbed dose of about 50 rads (50 centi-gray, or 0.5 Gray), and near the hypocenter at Nagasaki, about 18 to 24 rads (Chris87, p.21). "The cumulative exposure would be about one-third as large after a day and only a few percent after a week" (Oka87, p.223).
According to Sztanyik (Sz78), "Shortly after the detonation, thousands of people entered the affected areas in both cities for rescue work, in search of their relatives and to assist in removal of ruins." This means that an LSS survivor who was up to 10,000 meters from groundzero ATB (a presumably unexposed survivor, for instance), and who then tried to help within 1000 meters of the hypocenter, could have received an unquantifiable or poorly quantified dose of radiation.
Fission-Product Fallout :
In both Hiroshima and Nagasaki, there were regions of fallout at quite a distance from the hypocenter. In Nagasaki, the Nishiyama area received fallout about one hour after the bombing. "The upper limits on absorbed dose from gamma rays for persons continuously in the fallout area at Nagasaki ranged from about 12 to 24 rad" (Chris87, p.21). An exposure of at least one-fifth the maximum extended over some 1,000 hectares (Chris87, p.21). For Hiroshima, the estimated absorbed doses from fallout may have ranged from about 0.6 to 2 rads for those with continuous residence in the Koi-Takasu region (Chris87, p.21).
External doses in the fallout areas were first measured "some weeks or months" after the bombing. Approximate initial levels could be back-calculated "providing storms had not washed away a large portion of the activity" (Chris87, p.20).
The Missing Information :
According to the chapter on residual radioactivity in RERF's DS86 Volume 1, "Many factors affecting the accuracy of the measurements are not well known 40 years after the bombs, therefore exposure estimates must be rough approximations. In general, the exposure rates were not measured soon enough to avoid some weathering and they were not repeated often enough to account for subsequent weathering or to provide a time distribution of radioactivity. The number of sites monitored was too small to develop a good estimate of detailed geographical distribution of the radioactivity. Also, in such surveys, it is difficult to avoid unrepresentative sampling and it is not known whether such a sampling bias exists. Finally, the details of calibration and measurement are not always available" (Oka87, p.206).
Handling of This Confounding Variable :
The likelihood that the "unexposed" group (Dose-Group 1) is not a group with zero dose has needed facing for a long time (Go81). The radiation community is starting to acknowledge it:
"At the present time doses due to residual activity are not calculated by the DS86 system. It is recommended that the few individuals from areas of high residual radioactivity not be included in the nonexposed cohort for epidemiological studies" (Chris87, p.21).
"The individual exposures from residual radiation may not be significant compared with the direct radiation at the time of the bomb. On the other hand, individuals with potential exposure from these sources are dubious candidates for inclusion in a cohort that was presumably not exposed" (Oka87, p.224).
"Care has to be taken to exclude those exposed in this way from the control population" (Fry87, p.847).
The words sound fine, but one must wonder why the exclusions were not made 30 years ago. It would be a very questionable practice indeed to make additional exclusions now, when 40 years of outcome are at hand.
Steps in the Wrong Direction :
The nominally "unexposed" Dose-Group 1 is not the only group affected by permanent uncertainties about activation products, fallout, and even the size of the bomb at Hiroshima -- the city which provides 60,470 of the 91,231 survivors (from Table 9-D).
The lower is the Dose-Group, however, the greater is the accuracy of dosimetry required, before a Dose-Group could conceivably be subdivided -- even prospectively. Nonetheless, RERF has recently started to subdivide Dose-Groups 2 and 3 into lower-dose and upper-dose halves (see Tables 26-N and 26-O, and Kato87). We regard this as a move in the wrong direction, scientifically.
The undivided Dose-Group 2 has a shielded kerma range of 1-9 rads, with a mean internal organ-dose of only l.5 to 1.9 rems (see Row 11 in Tables 9-C and 10-E). In other words, in both dosimetries, there is a very small difference in mean dose between Dose-Group 1 (nominally "zero" dose) and the undivided Dose-Group 2. The doses from fallout and/or from activation products encountered during rescue-work could easily be bigger than this dose-difference for thousands of people in Dose-Groups 1 and 2.
The idea that reliable dose-differences can be created within this small dose-range in the future -- by perpetual adjustment of the input -- strains belief. Under the circumstances, we see no way for scientists ever to have confidence in alleged dose-differences between Dose-Group 1 and Dose-Group 2.
Reasonable Handling of Dose-Groups 1 + 2 :
Hidden doses from residual radioactivity are a confounding variable of relatively small importance in the cohorts where a few rads more, or a few rads less, in a group's mean dose cannot have much of an impact on results. But for Dose-Groups 1 and 2, permanent and well-founded uncertainty about true doses can result in uncertainty about which cohort really received the higher dose.
The realistic solution is both scientifically solid and simple: It consists of treating Dose-Groups 1 and 2 as not provably different, and just combining them into a single, very low-dose Reference Group. In the T65DR dosimetry, this dose would be only 654 millirems (see Table 9-C, Row 11, far right); in the DS86 dosimetry, this dose would be only 861 millrems (see Table 10-E, Row 11, far right.)
Dose-Groups 1 and 2 very often need to be combined anyway for a different purpose: To reduce the small-numbers problem. The combination has frequently been used by analysts of the A-Bomb Study (Bee78; Kato82; Land84; Toku84; Waka83; also Beir80 at p.155).
The combination of Dose-Groups 1 and 2 is scientifically strong, and seems far more likely to produce believable results, under the circumstances, than subdivision. It is not clear why anyone is suddenly moving in the less credible direction, of sub-division.5. Choice of RBE Values for the Neutron-Component
It is now recognized that exposure of the A-bomb survivors by neutrons is a very small part of their total exposure. Part 3 of this chapter reported that, in Hiroshima, neutrons account for less than 1 % up to about 6.5 % of the total shielded kerma doses; in Nagasaki, the range is less than 1 % up to about 1.5 % of the total shielded kerma doses. The percentage of total organ-dose which comes from neutrons is even lower, after shielded kermas are adjusted by the body transmission-factors, because those factors are much lower for neutrons than for gammas (Table 9-A). The overwhelming part of the exposure in both cities was caused by gamma rays.
On the other hand, per rad of dose delivered to an organ, a large body of radiobiological evidence indicates that radiations of high-LET (Linear Energy Transfer), like neutrons and alpha particles, are more potent in causing biological effects than low-LET radiations like gamma-rays and X-rays. Therefore, every analyst of the A-Bomb Study confronts the question: For human carcinogenesis, what is the relative biological effectiveness (RBE) of a rad from fission neutrons versus a rad from A-bomb gamma-rays?
A Formal Definition of RBE :
George Kerr provides one common definition of RBE, as follows:
"The RBE is defined as the absorbed dose from orthovoltage x rays, divided by the absorbed dose from another radiation needed to produce the same level of biological effect" (Kerr88, p.242).
(Orthovoltage X-rays, also called medium voltage X-rays, fall in the range of about 180-400 kilovolts (Des89, p.656).)
The definition of RBE means that, if one rad of neutron exposure produces the same level of biological effect as 10 rads of orthovoltage X-rays, then the RBE value for neutrons would be 10, relative to the X-rays. However, if two rads of gamma radiation are required to produce the same level of biological effect as one rad of orthovoltage X-rays, then the RBE of the gamma rays would be 0.5, and the RBE of neutrons relative to the gamma rays would be 20, not 10.
The biological effect which this book considers among A-bomb survivors is the increase in the cancer mortality-rate, per unit dose of acute low-LET exposure. There is no theory or body of evidence which permits any generalization that RBE for cancer-induction will be the same as RBE for some other biological endpoint, or that RBEs for humans will be comparable to RBEs for other species.
Below, we shall briefly review the types of evidence which might provide guidance on the RBE of fission-neutrons versus gamma exposure, for cancer-induction in the human.
Past Evidence -- A-Bomb Study :
By 1965, the T65D system of dosimetry was in place, with its estimates of neutron and gamma doses to the survivors of the A-bombing. Readers who examine Table 9-B will see that the shielded kerma doses (T65DR) estimated for neutrons -- relative to the corresponding gamma doses -- are very low at Nagasaki, and quite important at Hiroshima.
Among analysts of the study, it became conventional practice to multiply the shielded kerma doses from neutrons by an RBE factor of 5 or 10 for greater carcinogenic potency, compared with gamma rays. As we shall see, however, the evidence soon cast doubt on 5 to 10 as the proper RBE range for neutrons, within the T65DR dosimetry.
In 1977, McGregor and co-workers published findings on breast-cancer in Hiroshima and Nagasaki survivors. A major finding of their study was stated twice, as follows:
"The Hiroshima and Nagasaki dose-response curves were similar, which suggested approximate equivalence of neutron and gamma radiations in their carcinogenic effect on breast tissue, and were consistent with a linear model" (McGr77, p.799).
"The dose-response function was reasonably linear and was similar in the two cities. There was no evidence suggesting that gamma and neutron radiations entail different risks per rad" (McGr77, p.808).
In 1978, Mole published a review on the subject of breast-cancer induction by ionizing radiation, in which he stated the following: "The incidence of breast cancer per rad was closely similar at Hiroshima and Nagasaki, for deaths (Mole, 1975) and for diagnoses (McGregor et al., 1977), showing that the RBE of fission neutrons cannot much exceed one for induction of breast cancer in women" (Mole78, p.402).
In 1979, Land and McGregor further amplified their conclusion concerning the RBE of neutrons in breast-cancer induction by radiation: " . . . For breast cancer there was no epidemiological evidence that the effects of neutron and gamma radiation were markedly different. Therefore, no account was taken of radiation quality in the present report" (Land79, p.17). And, indeed they did not take any account of RBE; all their calculations used breast-dose in rads (total gamma plus neutron dose).
Plentiful Flags of Warning :
It is amazing that these findings did not produce great flags of warning that something was radically wrong. The findings indicated that (A) either the estimate of neutron RBE values like 10 must be seriously wrong for carcinogenesis in humans, or (B) the estimate of neutron fluences at Hiroshima must be seriously exaggerated.
In 1980, while writing Radiation And Human Health (Go81), I examined all the evidence and concluded there was something radically wrong with the neutron story -- a conclusion based not only on the breast-cancer findings but also on the leukemia findings and on frequency of small head-size for those irradiated in utero. These concerns were discussed in Go81 under the listing "Neutron Issue at Hiroshima-Nagasaki" (brain-damage data at pp.730-3 and pp.736-7; breast-cancer data at p.246; leukemia data at p.380 and pp.668-9). At page 246, I stated the following explicitly:
"The paper by McGregor and co-workers showed that there is no significant difference between the findings from Hiroshima and those from Nagasaki, and that there is no evidence to suggest that an RBE value for neutrons other than 1.0 (indicating no difference in effectiveness) is needed. A higher RBE had been suggested earlier because there was a higher neutron component in the radiation at Hiroshima than at Nagasaki. Like McGregor and co-workers, we shall use an RBE of 1.0 for neutrons and we can therefore combine the Hiroshima and Nagasaki findings. In other words, we shall treat rads absorbed from neutrons just like rads absorbed from gamma rays, in terms of their cancer-producing effects. However, in correcting kerma doses in air to absorbed tissue doses, we shall use appropriate factors for gamma rays and neutrons (see chapter 6)."
I made the choice of assigning RBE = 1.0 to neutrons for all cancers and leukemia in the A-bomb experience, since that was the only rational choice at the time.
Of course I recognized that the true RBE could be much higher than 1.0, but if that were the case, then the neutron-component must have been grossly exaggerated in the T65DR dosimetry for Hiroshima. Since there was no way for me to prove that the neutron-component at Hiroshima was much too high, assigning RBE = 1.0 for neutrons was a reasonable way not to participate in an obvious error. If I had blindly accepted both RBE = 10 and what turned out to be a 10-fold overestimation of neutrons at Hiroshima, I would have used the equivalent of RBE = 100 for Hiroshima neutrons. Within a year, I learned that neutrons had indeed been overestimated at Hiroshima by about 10-fold (Lo81, Fig. 1, p.663).
Current Evidence -- A-Bomb Study :
When the DS86 dosimetry corrected the neutron errors in the A-Bomb Study, one consequence was to lessen the difference between Hiroshima and Nagasaki in their neutron-to-gamma ratios. In addition, the organ-doses from neutrons are extremely low compared with the organ-doses from gamma-rays, as noted at the beginning of this section.
Under the circumstances, it is hardly surprising that analysts are unable to learn anything from the study about the true neutron RBE for cancer-induction. The inquiry requires a level of accuracy and precision which the data cannot possibly meet.
In TR-9-87, Preston and Pierce comment: "It is well-understood that RBE cannot be usefully estimated from the cancer mortality data, because the gamma-ray and neutron exposures to individuals are very highly correlated. What little information on this [which] was available within the T65D dosimetry was largely due to the ratio of neutron to gamma-ray exposures differing substantially between cities. Since this is no longer the case in the DS86 dosimetry, even less information is now available about RBE in these data" (Pr87b, p.27).
Preston and Pierce relate that they tested constant RBE values from 1 to 50 in the DS86 system, and could not find evidence that one value was better than another (Pr87b, p.27).
Their colleagues at RERF, Shimizu and co-workers, also attempted to evaluate neutron RBE, in TR-5-88. They comment: "Since the neutron dose is very small under the DS86 system, an analysis of the dose response using the gamma and neutron doses separately or to estimate the neutron RBE is difficult" (Shi88, p.36). After making extensive efforts and producing sets of estimates, they conclude: "However, the uncertainties in these estimates are too large to permit serious consideration of these RBEs" (Shi88, p.36-37).
In order to make their own estimates of lifetime fatal cancer-risks in the DS86 dosimetry, Shimizu and co-workers ended up by using a constant neutron RBE value of 10, and going ahead with their calculations.
Elsewhere, Fry and Sinclair also acknowledge that neutrons in the A-Bomb Study are so sparse that " . . . direct estimates of neutron relative biological effectiveness may be precluded or be much more difficult" (Fry87, p.845).
Other Types of Evidence :
With respect to the correct RBE value for neutrons, there simply exists no relevant human epidemiological evidence -- a conclusion reached also by Warren Sinclair, who says "there are no human data" (Sin88, p.151).
Analysts might hope to put some upper limits on RBE values for neutron exposure, by examining RBE values for alpha exposure versus low-LET exposure. Unfortunately, the human epidemiological evidence on the proper RBE for alpha exposure is still cloudy, at best.
One contribution to the uncertainty comes from human evidence which suggests that radiation-induced cancer from alpha exposure may show up earlier than from low-LET exposure (Go83; Go85). If RBE is going to reflect lifetime excess cancer-risk per rad, but if the existing follow-ups are all incomplete, RBEs based on incomplete follow-ups will overestimate the carcinogenic potency per rad of the radiations which induce cancers earlier. (This situation may be analogous to the tendency to overstate the relative radiation-inducibility of leukemia per rad, just because leukemia shows up earlier than most of the solid tumors.)
Since the correct RBE values for alpha exposure are still so uncertain, they cannot provide much guidance for inferring correct RBE values for neutrons.
Then what kind of evidence can be used, for estimating the RBE of neutrons?
Although the relevance of experiments with cell-studies and with other species is always uncertain, there is no other guidance on neutrons. Radiobiological experiments have been indicating that biological dose-responses from neutrons are usually linear at low doses, and then begin to flatten.
Fry (Fry81, p.232) has stated that, for experimental animal data, " . . . results consistently show that the dose-response curve for tumor incidence after exposure to neutron radiation bends over at relatively low doses." Kerr (Kerr88, Fig.3, p.245) cites one mouse-study of female mammary adenocarcinomas where the linear dose-response from neutron irradiation appears to be shifting to supra-linearity (concavity-downward) in the region of 10-20 rads of neutron dose.
Is RBE Constant or Variable, in the A-Bomb Study?
In the absence of any human evidence contradicting the experiments referred to above, we are going to assume that dose-response for neutrons in the A-Bomb Study is linear. The neutron doses received by the A-bomb survivors are so low that they would certainly lie in the linear segment of the neutron dose-response curve.
We can ascertain their values in DS86, if we look ahead to Chapter 10. We take the shielded kerma values from Table 10-D, Rows 2 and 6, and multiply them by the body transmission-factor of 0.19 from Table 9-A, in order to obtain mean organ-doses. The results, below, show that the highest mean neutron organ-dose is about 4.369 rads. Indeed, very few people in the study received an organ-dose over one rad from neutrons.Dose- Hiroshima Nagasaki Group Neutron Neutron Organ-Dose Organ-Dose 1 0 rad 0 rad 2 0.009 rad 0 rad 3 0.058 rad 0.002 rad 4 0.270 rad 0.050 rad 5 0.620 rad 0.126 rad 6 1.330 rad 0.243 rad 7 2.165 rads 0.410 rad 8 4.369 rads 0.994 rad
With respect to the gamma dose in the A-Bomb Study, the shape of dose-response for cancer-induction is supra-linear (concave-downward), as will be shown in Chapter 14. Nonetheless, for the short dose-segment below 5 rads, we can make the simplification that dose-response is linear.
If the dose-response below 5 rads is linear for both the neutrons and the gamma-rays, and if neither has a threshold, it follows that the RBE-value of neutrons will be constant at all neutron doses which occur in the A-Bomb Study. With respect to human carcinogenesis, we have shown elsewhere in this book that no threshold exists for low-LET radiations.
Therefore, in our analyses of the A-Bomb Study, we use a constant RBE value for neutrons. The remaining question is simply: What value should we use?
The Fallacy of RBE = 100 :
As noted above, Shimizu and co-workers use the constant RBE value of 10 for neutrons with the DS86 dosimetry. We do not fault that choice. Almost any choice is arbitrary, in the absence of relevant human evidence.
But some choices can and should be ruled out, because real-world human evidence invalidates a key premise on which they rest. We refer to suggestions (for instance, in Beir80) that the RBE for neutrons versus low-LET radiations is destined to rise progressively as total dose goes down toward zero dose.
The suggestion would be valid, of course, if dose-response were linear for neutrons, and concave-upward for low-LET radiation. Under such circumstances, the neutron RBE would necessarily vary with dose-level, and would increase at lower doses. If the dose-response curve of the low-LET radiation were presumed to have a very flat region near zero dose, then the ratio of slopes (or biological effect) at equal doses of the two radiations could easily rise to 100 or more, in the zero-dose region. Indeed, if the low-LET dose-response were assigned a threshold, the neutron RBE would rise to infinitely high values at doses below the alleged threshold-dose for the low-LET radiation.
The reasoning above, however, is simply inapplicable and irrelevant here. For human carcinogenesis, there is no threshold dose, and the evidence from the A-Bomb Study has clearly shown for a long time that dose-response from gamma rays is not concave-upward (see Chapters 14 and 22).
The human epidemiological evidence on gamma dose-response overrules any suggestion that the neutron RBE might need to be raised to very high values in the A-Bomb Study. Moreover, such evidence from real, whole humans must prevail over data from the lab, where radiobiologists have generated every dose-response one could imagine or desire, in their experiments.
Describing experimental work, Fry says "Because of the marked variation in the shape of dose-response curves for low-LET radiation, RBE values vary almost infinitely" (Fry81, p.224). Indeed, in his discussion of possibilities, Fry includes the non-linear threshold model for carcinogenesis by low-LET radiation (Fry81, Fig.4, p.228) -- the model which leads to the infinite RBE value for neutrons.
But it is now 1990, and we must dismiss that model (and many others) for human carcinogenesis, because we have human data which provide a reality-based answer about low-LET dose-response and about the absence of any threshold. There is simply no basis for accepting the suggestion that neutron RBE always rises to very high values at low doses.
Current ICRP-NCRP Position on Neutron RBE :
Kerr reports that "Both the ICRP and the NCRP are now recommending essentially the same guidance with respect to the quality factor for fast neutrons: an increase by a factor of two" (Kerr88, p.242).
(Quality factor, Q, is the term used for RBE in radiation protection, while the term RBE is often reserved for radiobiological experiments. ICRP stands for International Commission on Radiological Protection, and NCRP stands for (USA) National Council on Radiation Protection and Measurements.)
The recommended two-fold increase is necessarily made in the absence of any direct human evidence on neutron-potency. Kerr acknowledges that "The Q for neutrons is based on a large, unfocused body of experimental data on RBE. Orthovoltage x rays are the usual reference radiation, but gamma rays from 137-Cesium and 60-Cobalt have also been widely used as reference radiations. The mixed use of reference radiations, acute versus fractionated exposures, and high versus low dose rates, can easily result in a factor of 2, or more, discrepancy in the measured values of RBE for the same biological end point" (Kerr88, p.243).
In circumstances like this, it is fair to say that the scientific basis for increasing RBE by a factor of two is thin, and that the basis for the customary value of 10 is also thin.
Nonetheless, there is a persuasive basis for assigning some value to RBE when one is dealing with credible dose-estimates for radiations having LET values which differ greatly. After all, Linear Energy Transfer is a measure of spatial concentration of energy-transfers, and there is no doubt that the biological impact of ionizing radiation is strongly tied to the spatial concentration of its energy-depositions.
The RBE Values Chosen for This Book :
In our analyses of the A-Bomb Study, we are using a constant RBE value of 20 for neutron organ-doses in the DS86 dosimetry. Therefore, we are using a constant RBE value of 2 in the T65DR dosimetry. The RBE of 2, combined with the mistaken T65DR estimates of neutrons, is about equivalent to RBE = 20 with the correct estimate of neutrons. By contrast, if we were to use RBE = 20 in the T65DR dosimetry, it would be equivalent to using a constant RBE value approaching 200, because the neutrons at Hiroshima were overestimated by about 10-fold, and were nearly negligible at Nagasaki.
All our analyses will use a neutron RBE of 2 for the T65DR dosimetry, with its overestimate of neutron doses unaltered, and a neutron RBE of 20 for the DS86 dosimetry, where neutron doses are supposed to be correct.
In the T65DR dosimetry, even though some of the neutron organ-doses (in rads, or centi-grays) appear to lie beyond the linear segment of the presumed neutron dose-response curve, in reality they did not lie beyond it. Therefore we again avoid participating in the dosimetry error when we presume -- in both dosimetries alike -- that neutron dose-response is linear in the study's neutron dose-range.
Because the RBE of 2, combined with the mistaken T65DR estimates of neutrons, is about equivalent to the RBE of 20 with the correct estimate of neutrons, the differences (if any) between our findings in the T65DR and in the DS86 systems cannot be blamed on a use of different values for neutron RBE. We have, in effect, used the constant RBE of 20 in both dosimetries.
If the RBE of 20 is too high for neutrons -- and it may well be too high -- it will lead to an underestimate of radiation-induced cancer-risk in this book, since a higher RBE value raises the total dose in rems (centi-sieverts) without increasing the observed cancers at all.