Evidence for higher-than-predicted neutron exposures in Hiroshima raises some vexing questions.

by Tore Straume, Lawrence Livermore National Laboratory, University of California, Livermore, California

This article was originally published in RERF Update 4(4):3-4, 1992-93.

The Dosimetry System 1986 (DS86) has recently been used in influential reports by the United Nations (UNSCEAR 1988), the US National Academy of Sciences (BEIR V), and the International Commission on Radiological Protection (ICRP Publication 60) to estimate radiation-induced cancer risk and to make recommendations concerning appropriate radiation safety standards for workers and the public. However, a detailed comparison of neutron activation measured in mineral and steel samples from Hiroshima, with the activation levels calculated for these samples using the DS86 neutron fluence and spectrum, revealed a large discrepancy between measurements and calculations (T Straume et al, Health Phys 63:421-6, 1992). The discrepancy indicated that there were actually some 2 to 10 times more neutrons at relevant distances in Hiroshima than estimated by DS86. This finding has important implications for the estimation of radiation-induced cancer risk in humans.

Neutron activation measurements

The activation measurements available at the most relevant distances of 1-2 km in Hiroshima are presently limited to nuclides produced by thermal neutrons, which do not themselves contribute much to neutron dose. However, the measured thermal-neutron fluences in the samples are expected to be proportional to the free-in-air fast-neutron fluence at these distances and thus be proportional to neutron dose. This is because nearly all of the thermal neutrons that activated the samples actually originated from fast neutrons that were slowed down in the local environment of the sample (ie, in the ground and in buildings).

The possible effect of this discrepancy on the dosimetry for atomic bomb survivors could be that DS86 neutron doses in Hiroshima would have to be increased by as much as a factor of 2 at 1 km and a factor of 10 at 1.5 km (the so-called DS86+ dose).

Risk implications

If the neutron dose is underestimated in DS86 in proportion to the measured-to-calculated ratios for thermal-neutron activation (Straume et al, loc cit), the Hiroshima neutron dose at 1.5 km would be a factor of ~10 higher than now believed. This would result in neutrons contributing a larger fraction (and gamma rays contributing a smaller fraction) of the radiation-induced risk in Hiroshima than is predicted from DS86. Although neutrons in the DS86 system contributed less than a few percent of the dose in Hiroshima (WC Roesch, ed, US-Japan Joint Reassessment of Atomic Bomb Radiation Dosimetry in Hiroshima and Nagasaki. DS86: Final Report, Vol 1, 1987), they have a disproportionately large impact on radiation risk estimates because of their high relative biological effectiveness, which increases with decreasing dose (RL Dobson et al, Radiat Res128:143-9, 1991).

To illustrate the potential impact of this neutron discrepancy on risk estimates inferred from the Hiroshima data, the fraction of radiation-induced risk from neutrons is estimated using chromosome aberrations measured in blood lymphocytes of Hiroshima survivors (D Preston et al, RERF Technical Report 7-88, Table 4b). This is a particularly appropriate surrogate endpoint for such an evaluation because of the availability of a chromosome aberration dose-response curve (Dobson et al, loc cit) for Hiroshima-like neutrons. (Experimentally derived dose-response curves for Hiroshima-like neutrons are not available for other biological endpoints.)

The following relation was used to calculate the fraction (F_n) of aberrations produced by neutrons in Hiroshima: F_n = Y_n/Y_t = a_nD_n/Y_t. The aberration frequency for Hiroshima neutrons was estimated as the product of the slope (a_n) of the dose-response curve for Hiroshima-like neutrons (from Dobson et al, 1991) and the neutron dose (D_n) to bone marrow in Hiroshima survivors. (D_n values were provided by Akio A Awa of the Radiation Effects Research Foundation.) The total aberration frequency (Y_t) for Hiroshima survivors was obtained from D. Preston et al (loc cit) as a function of total (gamma + neutron) DS86 marrow dose (D_t). The Hiroshima aberration data (Y_t) are reported in the form “% cells with aberrations.” In contrast, the a_n value by Dobson et al was reported in the form “aberrations per cell per Gy.” To provide the numerator in the above equation with the same units as the denominator, Y_n was converted from “aberrations per cell” to “% cells with aberrations.” This transformation was made possible by the determination that the aberrations in Dobson et al were Poisson distributed among cells.

Essentially all aberrations in blood lymphocytes of Hiroshima survivors several decades after exposure are reciprocal translocations and pericentric inversions (Preston et al, loc cit). Although the a_n coefficient employed here for Hiroshima-like neutrons was obtained in vitro for dicentrics and rings, it has been shown that reciprocal translocations and pericentric inversions are induced with the same frequency as dicentrics and rings (JN Lucas et al, Int J Radiat Biol 61:830-5, 1992; LG Littlefield et al, in Medical Management of Radiation Accidents, CRC Press Inc, Boca Raton, Fla, USA, pp 109-26, 1990). Also, considerable data now show that the reciprocal translocation frequency in blood lymphocytes of whole-body-exposed individuals is fully stable with time postexposure (T Straume et al, Health Phys 62:122-30, 1992; J Lucas et al, Cytogenet Cell Genet 60:259-60, 1992; KE Buckton et al, in Mutagen-induced Chromosome Damage in Man, University Press, London, pp 142-50, 1978). Therefore, the cytogenetic effectiveness of Hiroshima-like neutrons measured in vitro at first cell division postexposure can be used to estimate the aberration frequency induced by neutrons in Hiroshima survivors.

For the case where DS86 is assumed to underestimate the neutron dose, the “true” neutron dose (D_n+) was estimated from the relation D_n+ = D_nR, where R is the measured-to-calculated neutron-activation ratio from the fitted curve in T. Straume et al (Health Phys 63:421-6, 1992). Also in that case, the aberration frequency (Y_n+) estimated for Hiroshima neutrons is the product of a_n and D_n+. Again, this frequency was converted to percentage of cells with aberrations to provide compatibility with the aberration data available for Hiroshima survivors. Values for percentage of cells with aberrations, doses, and the other relevant parameters for Hiroshima are listed in the Table.

Chromosome aberration data and other parameter values for Hiroshima.a

Slant range (m) Dn (Gy) Dg (Gy) Y1 (％ cells) Yn (％ cells) R Dn+ (Gy) Yn+ (％ cells) 1000    0.056   2.34    18.4    4.81    1.9   0.106     8.87 1100    0.031   1.70    15.3    2.70    2.1   0.065     5.56 1150    0.019   1.22      9.2    1.67    2.5   0.048     4.10 1200    0.009   0.74      4.9    0.79    3.6   0.032     2.82 1350    0.002   0.305      1.6    0.177    6.8   0.014     1.20 1700    0.00023   0.055      0.26    0.020  15.0   0.0035     0.305b

aEffective slant-range values were estimated from the DS86 free-in-air kerma values that correspond to the bone marrow doses listed above and the dose-vs-range relationships for Hiroshima (WC Roesch, ed, US-Japan Joint Reassessment, Vol 1, 1987). Values for Yt are observed percentage of cells with aberrrations minus background (ie, the group that received 0-0.005 Gy) from D. Preston et al (RERF TR 7-88, Table 4b). Yn and Yn+ are percentage of cells with aberrations for DS86 and DS86+ neutron doses, estimated as described in the text. Neutron doses (Dn) to marrow were provided by Akio Awa (RERF) and gamma-ray doses (Dg) to marrow were obtained by subtracting the neutron doses from the total doses in Table 4b of RERF TR 7-88. The values for R are the measured-to-calculated ratios for thermal-neutron activation (Straume et al, Health Phys 63:421-6, 1992) at the indicated distances.

bThis predicted value of 0.305% is not significantly larger than the observed value of 0.26% plus or minus 1.68%.

The risk implication of the neutron discrepancy is illustrated in the Figure. White bars are based on the assumption that DS86 neutron doses in Hiroshima are indeed correct and do not require upward revision. In contrast, the black bars are based on the assumption that DS86 neutron doses are too low in Hiroshima and must be increased with neutrons of all energies in proportion to the measured-to-calculated thermal-neutron ratios (Straume et al, loc cit). The Figure shows that if the DS86 neutron dosimetry is correct, then neutrons contributed only a small fraction of the total radiation-induced health risk in Hiroshima. However, if Hiroshima neutron doses are underestimated in DS86 by the amounts observed for thermal-neutron activation (Straume et al, loc cit), then most of the radiation-induced risk in Hiroshima would actually be due to neutrons, not gamma rays, as currently believed. Since the DS86 gamma-ray doses for Hiroshima are approximately correct (there was relatively good agreement between gamma-ray doses measured in Hiroshima and those calculated using DS86), it can be inferred from the Figure that gamma rays would contribute about 50% of the risk at 1000 m and essentially none of the risk at 1700 m. Also, the sharply increasing contribution to risk from neutrons with increasing distance (or with decreasing dose) could impact the shapes of dose-response curves obtained from the Hiroshima data; ie, the increasing neutron dose component with distance would tend to make the curves more linear at low doses than expected from DS86.

Figure. The fraction of risk induced by neutrons in Hiroshima if the Dosimetry System 1986 is correct (DS86, white bars) or if DS86 neutron doses are increased according to the thermal-neutron activation measurements (DS86+, black bars). The fractional risk estimates are based on chromosome aberration data for Hiroshima survivors (Preston et al, RERF TR 7-88) and the effectiveness of Hiroshima-like neutrons measured in vitro (Dobson et al, Radiat Res 128:143-9, 1991). The fraction of risk by gamma rays is indicated on the right axis and is simply 1 minus the risk from neutrons.

Acknowledgment

This work was performed under the auspices of the US Department of Energy by the Lawrence Livermore National Laboratory under contract number W-7405-Eng-48 with support from the Defense Nuclear Agency, Project Number 92-281.