Cells exposed to fast neutrons often exhibit a non-Poisson distribution of chromosome aberrations due to the high ionization density of the secondary reaction products. However, it is unknown whether lymphocytes exposed to californium-252 (252Cf) spectrum neutrons, of mean energy 2.1 MeV, demonstrate this same dispersion effect at low doses. Furthermore, there is no consensus regarding the relative biological effectiveness (RBE) of 252Cf neutrons. Dicentric and ring chromosome formations were assessed in human peripheral blood lymphocytes irradiated at doses of 12–135 mGy. The number of aberrations observed were tested for adherence to a Poisson distribution and the maximum low-dose relative biological effectiveness (RBEM) was also assessed. When 252Cf-irradiated lymphocytes were examined along with previously published cesium-137 (137Cs) data, RBEM values of 15.0 ± 2.2 and 25.7 ± 3.8 were found for the neutron-plus-photon and neutron-only dose components, respectively. Four of the five dose points were found to exhibit the expected, or close to the expected non-Poisson over-dispersion of aberrations. Thus, even at low doses of 252Cf fast neutrons, when sufficient lymphocyte nuclei are scored, chromosome aberration clustering can be observed.
Californium-252 has a mean neutron energy of 2.1 MeV (1 ) and is routinely utilized in the petroleum industry and in nuclear power production (2 ). As such, understanding the biological effects and hazards of low-dose 252Cf neutron exposures is important for worker protection. Neutrons are very useful for examining biological mechanisms dependent on ionizing density or linear energy transfer (LET) due to the wide range of energies that result in variable relative biological effectiveness (RBE). Our group is currently engaged in quantifying a suite of cellular end points at multiple neutron energies, including thermal neutrons (3 ) and, in the current work, fast neutrons.
Dicentric and ring chromosome analysis in human peripheral blood lymphocytes is currently considered the gold-standard biological dosimetry method, with studies establishing direct correlations between dicentric and (much rarer) ring chromosome induction and absorbed dose after low- and high-LET irradiations (4 ).
RBE indicates the ability of a particular radiation type to produce a certain biological effect, as compared to a reference radiation, typically low-LET gamma or X rays. RBE is an experimentally determined unit-less quantity that influences radiation weighting factors (denoted by wR), theoretically determined values put forth by the International Commission on Radiological Protection (ICRP) to account for the varying biological effects induced by different radiations. However, unlike wR, which takes into account all possible biological consequences of a particular radiation, RBE values vary with dose, dose rate, biological end point and cell type (5 ) and consequently will not necessarily equal the wR values. The RBE for dicentric and ring chromosome induction in peripheral blood lymphocytes after neutron exposure has previously been examined at many different incident energies and is reported to peak around 0.385 MeV (RBEM of 94.4 ± 38.9) (6 ). In two published studies, RBE was evaluated for human peripheral blood lymphocytes after 252Cf exposures, as outlined in Table 1. Lloyd et al. reported values between 6 and 27 for neutron-only exposures, depending on the dose at which RBE was calculated (7 ). Tanaka et al. noted dose-specific RBE values from 2.3 to 7.7 (for neutron and gamma components of 252Cf exposure) and 3.3 to 12.0 (for only the neutron component of the exposure), depending on dose rate (8 ). To compare, ICRP Publication 103 indicates that the wR of 252Cf is 16.85 (9 ). This value was obtained using the coarse 252Cf neutron energy spectrum (10 ) and the piecewise mathematical equation for wR stated in (9 ), where the spectrum-averaged wR value is equal to the sum of the wR for each energy bin, evaluated at the mid-point energy of the bin, that in turn is weighted by the neutron fluence rate of the bin normalized to the energy-integrated neutron fluence rate.
It is generally accepted that non-Poisson over-dispersion of radiation-induced DNA aberration clustering is a result of either: 1. a partial-body exposure; or 2. a high-LET radiation exposure (4, 11 ), like the 252Cf exposures examined here. In addition to the differing RBE values, neither Lloyd et al. (7 ) nor Tanaka et al. (8 ) provide a description of the dicentric distribution data necessary to determine whether 252Cf exposures produce the predicted high-LET over-dispersion of the numerical binning of dicentric chromosomes (4 ). Here we present our methods, results and comments for the 252Cf RBE and the associated DNA aberration dispersion.
MATERIALS AND METHODS
Blood was drawn by venipuncture from a healthy male blood donor (aged between 25 and 30 years old) who routinely donates anonymously for other radiobiology work at Canadian Nuclear Laboratories (CNL; Chalk River, Canada) (12, 13 ). Enrollment of a single donor is quite common in prominent neutron studies (6, 7, 14 ), and it is well documented that this individual's blood cells respond normally to low-LET radiation exposures (12, 13 ). After venipuncture, 1.5-ml aliquots of anticoagulated whole blood were transferred into 15-ml polypropylene test tubes with 1-mm-thick walls, and then transferred to the irradiation facility. All samples were maintained at room temperature prior to, and during the irradiations to minimize the effects of concurrent DNA repair (4 ).
252Cf spontaneous fission source irradiations were completed in the Health Physics Neutron Generator facility at CNL. Polypropylene test tubes containing fresh human blood were suspended around the 252Cf source in the middle of the facility. As shown in Fig. 1, blood samples were positioned 10 cm away from the 252Cf pellet (center-to-center distance) and at a height of approximately 103 cm from the facility floor. A REM-500 tissue-equivalent proportional counter, with an internal 244Cm source (15 ), was placed alongside the blood tubes to verify a constant dose rate during blood irradiation.
Analytical calculations and Monte Carlo radiation transport simulations were carried out using the Particle and Heavy Ion Transport code System (PHITS) version 2.64 (16 ) to quantify the absorbed dose rate delivered to an individual blood volume contained inside the tube holder by neutrons and photons emitted from 252Cf spontaneous fission, evaluated at the approximate mid-point date of the irradiation campaign. In the PHITS simulations, the T – Deposit tally was used to quantify the total kinetic energy deposited, per source particle, by secondary charged particles in the blood volume and the [ T – Track ] tally was used to quantify the energy-integrated incident particle fluence, per source particle, through the blood volume.
In the PHITS simulation environment, the blood volume, tube holder and radiation source were surrounded by a sphere of 100-cm radius and filled with air (air also filled the region within the tube holder above the blood volume and below the tube cap). In addition, the simulation utilized the following isotopic composition (by weight fraction) for blood, polypropylene and air (17 ): blood (1.06 g cm–3) consists of 1H (0.101866), natC (0.100020), 14N, (0.029640) and 16O (0.759414); polypropylene (0.90 g cm–3) consists of 1H (0.143711) and natC (0.856289); and air (0.001205 g cm–3) consists of natC (0.000124), 14N (0.755268), 16O (0.231781) and natAr (0.012827). Dose calculations pertaining to 252Cf neutrons utilized the neutron energy spectrum listed in (18 ) and dose calculations pertaining to 252Cf photons utilized the photon energy spectrum and photon emission data listed in (10 ). Using 252Cf neutron emission data described in (19 ) and 252Cf photon emission data described in (10 ), the neutron emission rate at the time of irradiations was 1.29 × 108 ± 0.06 × 108 neutrons s–1 [the cumulative uncertainty on the neutron emission rate of the source is 5% (19 )], the direct neutron fluence rate and direct photon fluence rate incident on the blood volume at the time of irradiation was 1.03 × 105 ± 0.05 × 105 neutrons cm–2 s–1 and 5.81 × 105 ± 0.41 × 105 photons cm–2 s–1, respectively, and the corresponding mass of 252Cf was 54.97 ± 3.88 µg. With respect to neutron dose delivery, the neutron energy spectrum emitted from 252Cf is generally fast and neutrons will typically undergo elastic scatter interactions in blood. Using tabulated energy-dependent microscopic elastic scatter cross-section data for 1H, 12C, 16O and 14N from the ENDF/B-VIII.0 library (20 ), the elastic scatter reaction rate with each element in blood was calculated for each neutron energy bin and the kinetic energy deposited in blood by each such interaction is equal to one half of the sum of the minimum and maximum kinetic energy that can be given to the recoil nucleus in question. Using this approach, the overall absorbed dose rate and absorbed dose delivered to a bare blood volume per unit energy-integrated neutron fluence is calculated to be 10.85 ± 0.54 mGy h–1 and 29.25 pGy cm2 neutron–1, respectively.
With respect to photon dose delivery, the energy-dependent mass-energy absorption coefficient data for blood (21 ) was utilized to calculate the kerma delivered to a bare blood volume per energy-integrated photon fluence, determined to be 6.32 pGy cm2 photon–1. Using the previously stated photon fluence rate incident on the bare blood volume, the kerma rate to the bare blood volume is 13.22 ± 0.93 mGy h–1.
Test tubes were removed from the irradiation facility at predetermined times, as detailed in Supplementary Table S1 (https://doi.org/10.1667/RR15528.1.S2). A total of five dose points of 12, 30, 60, 90 and 135 mGy were irradiated over two separate days. The lower four dose points were irradiated on the first irradiation day, and the highest dose point was irradiated on the second irradiation day.
Immediately after 252Cf irradiations, whole blood cultures were set up according to IAEA recommendations (4 ). Briefly, 1 ml of whole blood was cultured in 9 ml of Roswell Park Memorial Institute RPMI-1640 medium (Thermo Fisher Scientific™ Inc., Waltham, MA), with 15% fetal bovine serum (Millipore Sigma, Burlington, MA), 100 units ml–1 penicillin and 100 µg ml–1 streptomycin (Millipore Sigma, Burlington, MA), 20 µM of bromodeoxyuridine (BD Biosciences, San Jose, CA), and 1% phytohemagglutinin (Millipore Sigma). Cultures were incubated in a humidified environment for 48 h at 37°C with 5% carbon dioxide in air. To induce cell cycle arrest in metaphase, 0.1 µg ml–1 of colcemid (Thermo Fisher Scientific) was added for the final 4 h of incubation. Cells were further incubated in 0.075 M potassium chloride hypotonic solution (Millipore Sigma), followed by three washes with Carnoy's fixative containing three parts methanol to one part acetic acid. Slide making was completed using the Hanabi-HS Metaphase Spreader (Transition Technologies Inc., Toronto, Canada). Fluorescence-plus-Giemsa staining was completed by immersing slides in 20 µg ml–1 bisbenzimide H 33258 (Millipore Sigma), then layered with 0.6 M sodium phosphate (pH 9.0) (Millipore Sigma) and exposed to a 365 nm ultraviolet light. Slides were washed three times in ultra-pure water, and stained for 10 min in 10% Giemsa Stain (Thermo Fisher Scientific) in Gurr buffer (Thermo Fisher Scientific Inc.). Cover slips were mounted using permount mounting media (Thermo Fisher Scientific).
Slides were coded and metaphase spreads were imaged using the CytoVision® microscope system (Leica Biosystems, Buffalo Grove, IL). Complete metaphase spreads were scored manually according to the dicentric chromosome assay (DCA) criteria laid out in the IAEA Cytogenetic Dosimetry publication (4 ). Briefly, centromeres were counted to ensure metaphase spread completeness. If 46 centromeres were found, metaphase spreads were then examined for the presence of dicentric chromosomes, centric ring chromosomes and acentric fragments. Both dicentric and centric ring chromosomes were included in the aberrations tally, while acentric rings, minutes and terminal deletions were recorded as acentric fragments, following the IAEA Cytogenetic Dosimetry guidelines (4 ).
Data analysis was also carried out as recommended by the IAEA (4 ). The results of the DCA were tested for compliance with the Poisson distribution by calculating the dispersion index and u-test statistic for all doses. The goodness of fit of the dose-response curves was evaluated using the Chi-square test (χ2), and the significance of the equation coefficients was evaluated using the z test. P values less than 0.05 were considered statistically significant. The Dose Estimate software package (version 5.2) was used to ensure proper curve fitting (22 ). Errors were reported as either standard deviation (SD), the square root of the number of observations or as standard error (SE).
The PHITS simulations revealed that the neutron absorbed dose rate, delivered to a bare blood volume and to a blood volume inside a tube holder, was 10.49 ± 0.52 mGy h–1 for both conditions. This dose rate is of the same order of magnitude (difference = 21%) as the ambient dose equivalent rate (8.47 ± 0.42 mGy h–1) obtained from the product of the average ambient dose equivalent per unit fluence for the 252Cf neutron energy spectrum, 385 pSv cm2 neutron–1 (23 ), and the neutron fluence rate incident on the blood volume, 1.03 × 105 ± 5.13 × 103 neutrons cm–2 s–1, divided by the spectrum-averaged wR value for 252Cf of 16.85. As discussed in (3 ), the ambient dose equivalent per unit fluence was estimated from the dose equivalent per unit fluence delivered by an expanded and aligned planar neutron source to a point located 10 mm into the interior of a 30-cm diameter International Commission on Radiation Units and Measurements tissue sphere. Therefore, the absorbed dose rate delivered to the blood volume in tube and the ambient dose equivalent rate being of the same magnitude indicates that for the 252Cf neutron energy spectrum, dose delivery to blood is an adequate representative of dose delivery to deep organs.
The photon absorbed dose rates, delivered to a bare blood volume and to a blood volume inside the tube holder, are 7.54 ± 0.53 mGy h–1 and 8.09 ± 0.57 mGy h–1, respectively. The increase in the absorbed dose rate in the presence of the tube holder is attributed to secondary electrons created in the tube wall, from incident photon interactions, being able to enter the blood volume and depositing their kinetic energy. The total absorbed dose rate delivered to the blood volume in the tube holder from incident neutrons and incident photons is 18.58 ± 0.77 mGy h–1.
A total of 6,100 cells were examined across six dose points bearing, in total, 348 dicentric and centric ring chromosome aberrations (Table 2). As a function of dose, the 252Cf aberration yield varied from 1 aberration in 1,100 metaphase spreads at 0 mGy, to 159 aberrations in 1,000 spreads at 135 mGy. Therefore, the background yield was 0.001 aberrations per cell, and the maximum aberration yield was 0.159 aberrations per cell. The majority of cells had no aberrations, with a small portion of cells exhibiting between one and three dicentric and/or centric ring chromosomes. Figure 2 illustrates the 252Cf dose-response relationship compared to the extrapolated 137Cs reference radiation dose-response curve. The quality of the data was tested by calculating the dispersion index (σ2/y) and the u-test statistic. The dispersion index was assessed by dividing the variance by the mean, where a dispersion index of unity indicates alignment with the Poisson distribution. Results above 1.96 indicated a non-Poisson over-dispersion at the 5% significance level (4, 24 ). Three of the five exposure data points, i.e., 30 mGy, 60 mGy and 90 mGy, demonstrated elevated dispersion and u values >1.96 indicating non-Poisson over-dispersion (Table 2). The 135 mGy dose point was nearing a significant over-dispersion, with a u value of 1.82.
The linear and linear-quadratic dose-response relationships for 252Cf dicentric and centric ring aberrations (A) were calculated using the iteratively reweighted least-squares method recommended by the IAEA for both the neutron plus photon [n + γ] and neutron [n] dose components (4, 22 ). The neutron-only aberrations were calculated using the method proposed by Lloyd et al. (7 ) by extrapolating the number of photon-induced aberrations per cell at each dose point based on the previously published in-house 137Cs dose-response curve (13 ). This value was then subtracted from the total aberrations per cell to give the neutron aberrations per cell (Table 1). Linear and linear-quadratic regression equations and their errors are given in Table 3. Where χ2 was greater than the degrees of freedom (df), the standard error was increased by (χ2/df)1/2. All fits demonstrated non-significant χ2-test P values (Table 4) that indicated the observed data points did not differ from the fitted data, confirming a good dose-response curve fit. The z test indicated the β coefficient was not significant for both linear-quadratic fits, confirming a preference for a linear dose-response curve for both the [n + γ] and [n] data (Table 4).
RBEM was calculated as a ratio of the 252Cf and the 137Cs photon α coefficient values that represent the linear component of the dose-response curves, as shown in Table 3. A previously published 137Cs linear-quadratic dose response for aberrations from an earlier CNL study was used as the photon reference exposure (Table 3) (13 ). Consequently, a linear dose-response relationship for the neutron and photon components of 252Cf revealed a RBEM[n + γ] of 15.0 ± 2.6 and a RBEM[n] of 25.7 ± 3.8 for dicentric plus centric ring induction.
Dose-specific RBE values for the five 252Cf exposures were also calculated by dividing the 137Cs photon dose required to generate a given effect by the 252Cf dose that gave the same effect, as evaluated in Table 2.
In the current study, we analyzed and report here on the DNA aberration dispersion, which, to our knowledge, had not previously been reported for high-LET 252Cf exposure of ∼2.1 MeV neutrons. Dicentric chromosomes and (much rarer) ring chromosomes were both included in the analysis (4 ). Four of five dose points were found to exhibit, or were close to exhibiting, non-Poisson over-dispersion (Table 2). The statistical power is low at the lowest dose. Only the lowest exposure dose point of 12 mGy, with the higher uncertainty of 11 ± 3 aberrations in 1,000 cells, did not approach or achieve over-dispersion. It is possible that increasing the number of cells scored could eventually result in the predicted high-LET over-dispersion at all dose points. However, due to the limited sample available for this study, it was not possible to explore this option. Non-Poisson over-dispersion is characteristic of high ionization density and has been observed in other neutron studies, including those at similar fast neutron energies of 1.151 MeV (6 ), 1.6 MeV (25 ) and 4.85 MeV (6 ).
Linear and linear-quadratic regression fits were both evaluated, revealing a stronger relationship for the linear dose-response function due to the non-significant linear-quadratic β coefficient (Table 4). Both previously published 252Cf studies report linear dose-response relationships (7, 8 ). Other DCA studies in human lymphocytes with neutron energies between thermal and 1.151 MeV predominantly report linear dose-response relationships (6, 14, 26–28 ), while higher energies tend to show a mix of linear and linear-quadratic fits (6–8, 25, 26, 29–32 ).
There are currently two other published studies detailing 252Cf RBE for human lymphocytes. Lloyd et al. (7 ) described RBE [n] values ranging from 6 to 27 for 252Cf neutron doses between 118–2,894 mGy, in reference to 60Co photons at a similar dose rate (Table 1). Tanaka et al. (8 ) investigated several dose rates and found that dose-specific 252Cf RBE ranged from 2.3 to 7.7 for dicentric aberration yields at 1,000 mGy when 252Cf neutron and photon doses were considered together, and 3.3–12.0 when the neutron component of the dose was considered alone. These values were generated in reference to either 60Co or 137Cs photons at matched dose rates. For easier comparison here, the RBEM values for both prior studies were calculated using their data. For the study by Lloyd et al. (7 ), the RBEM[n + γ] was found to be 24 ± 11 and the RBEM[n] was of 33 ± 15. For Tanaka et al. (8 ), RBEM[n + γ] values ranged from 6.3 to 7.7, and the RBEM[n] values were between 9.8 to 11.9, with lower-dose-rate exposures generating a slightly higher RBEM (Table 1). In both cases, the DCA data conformed to a linear dose relationship, and the calculated RBEM values were similar to the highest reported dose-specific RBE.
For the current study, a RBEM[n + γ] of 15.0 ± 2.6 was evaluated for the combined neutron and photon components of 252Cf, and an RBEM[n] of 25.7 ± 3.8 was found for only the neutron component of 252Cf, compared to 137Cs photons. This was based on a thorough examination of 1,000–1,100 metaphase spreads per dose point (Table 2). Both the RBEM[n + γ] and RBEM[n] values fall between RBEM values of Lloyd et al. (7 ) (RBEM[n + γ] = 24 ± 11, RBEM[n] = 33 ± 15) and Tanaka et al. (8 ) (RBEM[n + γ] = 6.3 – 7.7, RBEM[n] = 9.8 – 11.9). The RBEM[n] reported here (25.7) is similar to the RBEM reported for other fast neutron energies, including 4.85 MeV mono-energetic neutrons (RBEM of 32.3 ± 13.3) (6 ), and is higher than the ICRP wR value of 16.85 for 252Cf neutrons (9, 10 ).
The 137Cs reference photon curve was generated using much higher absorbed doses (500 mGy to 3,500 mGy) than in the current 252Cf study (13 ). Thus, it was necessary to extrapolate the photon data, given that it is not practical to score photon DNA aberrations that are statistically viable at the very low absorbed dose range reported here. Most of the neutron RBEM values reported in the literature, including those of Lloyd et al. (7 ), are derived using 60Co as the reference radiation source (6, 14, 26–30, 32 ). RBEM can be affected by the reference radiation source, and using 60Co as a reference radiation can result in a higher, but not statistically different, RBEM compared to 137Cs (26 ).
Dose-specific RBE[n + γ] decreased from 12 and 13 to 8 with increasing doses ranging from 12 mGy and 30 mGy to 135 mGy (Table 2). The dose-specific RBE[n] also demonstrated an overall decrease with increasing dose, with the highest RBE of 23 at 7 mGy and the lowest RBE of 13 found at 51 mGy (Table 2). This trend is consistent with previously reported studies (7, 8 ). As expected, the highest dose-specific RBE[n + γ] and RBE[n] values of 13 and 23 are similar to the RBEM of 15.0 ± 2.6 and 25.7 ± 3.8 reported previously.
There was a dose-rate discrepancy between the 252Cf irradiations reported here (18.58 ± 0.77 mGy h–1) and the 137Cs reference radiation (49,800 mGy h–1). Given the very low dose rate of the 252Cf source, a matched dose and dose-rate reference radiation curve was not practical. Instead, a previously published higher-dose-rate 137Cs dose-response curve was used for the reference radiation. Dose-rate discrepancies between neutron and reference radiations have been noted in several other DCA RBE studies (27, 28 ). Lloyd et al. (7 ) and Tanaka et al. (8 ) used matched or nearly-matched 252Cf and reference radiation dose rates. However, the dose rates used by Tanaka et al. (8 ) were 1,200 mGy h–1, 120 mGy h–1 or 12 mGy h–1, and the highest RBE of 7.7 was reported after γ-ray and neutron irradiations at the lowest dose rate of 12 mGy h–1. Similar RBE[n + γ] values of 7.6 and 7.7 were reported by Tanaka et al. (8 ) for dose rates of 120 mGy h–1 and 12 mGy h–1, respectively, the lower value being of the same order as the dose rate of 18.58 mGy h–1 used here. It is already well-established that low-LET radiation effects are susceptible to dose-rate variations and long irradiation times (4 ), and the 12 mGy h–1 137Cs reference radiation curve used by Tanaka et al. (8 ) included protracted photon irradiation times. It is possible this would have drastically affected the scope of chromosome aberrations available for assessment, resulting in an elevated RBE.
To our knowledge, this work represents the first measurements for a dicentric and ring chromosome dose response after low-dose 252Cf[n + γ] exposures (12–135 mGy), and for DNA aberration distribution data after dicentric chromosome analysis of 252Cf-irradiated whole blood. A linear dose-response relationship was found for peripheral blood lymphocytes with RBEM[n + γ] and RBEM[n] values of 15.0 ± 2.6 and 25.7 ± 3.8, respectively. Future work should allow for better dose-rate matching of neutron and photon dose response, and more robust scoring of the aberrations in the low-dose range to further evaluate DNA aberration dispersion at all dose points.
Table S1. Detailed 252Cf neutron irradiation plan.
This work was funded through Atomic Energy of Canada Limited's Federal Science and Technology Program. We gratefully acknowledge the CNL Health Centre staff for performing routine phlebotomy in support of this research project.
Editor's note. The online version of this article (DOI: https://doi.org/10.1667/RR15528.2) contains supplementary information that is available to all authorized users.