Validation of a High-Throughput Dicentric Chromosome Assay Using Complex Radiation Exposures

Abstract

Validation of biodosimetry assays is routinely performed using primarily orthovoltage irradiators at a conventional dose rate of approximately 1 Gy/min. However, incidental/accidental exposures caused by nuclear weapons can be more complex. The aim of this work was to simulate the DNA damage effects mimicking those caused by the detonation of a several kilotons improvised nuclear device (IND). For this, we modeled complex exposures to: 1. a mixed (photons + IND-neutrons) field and 2. different dose rates that may come from the blast, nuclear fallout, or ground deposition of radionuclides (ground shine). Additionally, we assessed whether myeloid cytokines affect the precision of radiation dose estimation by modulating the frequency of dicentric chromosomes. To mimic different exposure scenarios, several irradiation systems were used. In a mixed field study, human blood samples were exposed to a photon field enriched with neutrons (ranging from 10% to 37%) from a source that mimics Hiroshima’s A-bomb’s energy spectrum (0.2–9 MeV). Using statistical analysis, we assessed whether photons and neutrons act in an additive or synergistic way to form dicentrics. For the dose rates study, human blood was exposed to photons or electrons at dose rates ranging from low (where the dose was spread over 32 h) to extremely high (where the dose was delivered in a fraction of a microsecond). Potential effects of cytokine treatment on biodosimetry dose predictions were analyzed in irradiated blood subjected to Neupogen or Neulasta for 24 or 48 h at the concentration recommended to forestall manifestation of an acute radiation syndrome in bomb survivors. All measurements were performed using a robotic station, the Rapid Automated Biodosimetry Tool II, programmed to culture lymphocytes and score dicentrics in multiwell plates (the RABiT-II DCA). In agreement with classical concepts of radiation biology, the RABiT-II DCA calibration curves suggested that the frequency of dicentrics depends on the type of radiation and is modulated by changes in the dose rate. The resulting dose-response curves suggested an intermediate dicentric yields and additive effects of photons and IND-neutrons in the mixed field. At ultra-high dose rate (600 Gy/s), affected lymphocytes exhibited significantly fewer dicentrics (P < 0.004, t test). In contrast, we did not find the dose-response modification effects of radiomitigators on the yields of dicentrics (Bonferroni corrected P >. 0.006, ANOVA test). This result suggests no bias in the dose predictions should be expected after emergency cytokine treatment initiated up to 48 h prior to blood collection for dicentric analysis.

INTRODUCTION

Ionizing radiation deposits molecular-bond-breaking energy, which can damage DNA and lead to the formation of aberrant chromosomes with two (dicentrics) or more centromeres. Dicentrics possess a great mechanistic obstacle for cell division: pulling two linked centromeres towards opposite poles stretches damaged chromosomes and forms chromatin bridges (1). Normally, it causes a delayed death of a cell that carries unrepaired dicentric. However, a line of evidence suggests that unresolved dicentric may also trigger the cascade of mutagenic events, namely the bridge breakage and refusion cycles and chromothripsis, that can contribute to genome instability and initiation of tumor (2–5). Over the past decades, researchers concerned with radiological protection acknowledged that the presence of dicentric chromosomes accurately indicates preexisting exposures (6, 7). A method of dicentric scoring in lymphocytes [dicentric chromosome assay (DCA)] has been extensively used worldwide to identify and quantify exposures to γ, X or neutron radiations (8).

The DCA is very precise but labor-intensive analysis that requires time and skill to estimate the radiation dose. This will likely pose a bottleneck when hundreds of thousands of exposed people require biodosimetry assessment, for instance, after the detonation of a large improvised nuclear device (IND) within a U.S. city. A QuickScan method, introduced by Flegal et al., (9) can significantly reduce the time for dicentric analysis. However, even a large cytogenetic network may become overwhelmed to assign degrees of exposure to such considerable numbers of people (10). To facilitate high-throughput sample processing, the Center for Radiological Research (CRR) has recently developed a miniaturized and fully automated version of the dicentric chromosome assay (the RABiT-II DCA) (11) wherein both processing and analysis of the blood samples are significantly accelerated. For exposed individuals who are at risk for developing clinically significant acute radiation syndrome (ARS), a rapid estimation of the dose can be lifesaving.

Manifestation and severity of ARS can be affected by many factors, including radiation quality, dose, dose rate, and the type of radiation field. To date, dicentric formation by uniform (pure photon or pure neutron) fields and low- to conventional dose rates (e.g., Gy/min or less) is well established by other researchers (6, 12–16). However, the surface detonation of a large nuclear weapon will likely cause a release of mixed (photons + IND-neutrons) radiation fields at a broader range of the dose rates (17). The dose rate of fallout irradiation will be lower than that for prompt irradiation (delivered within microseconds after the burst) and will gradually decrease over hours or days (during fallout and ground shine). To minimize the errors in dose prediction, it is essential to determine whether the RABiT-II system is sensitive enough to detect dicentrics induced by different radiation types and dose rates.

Recently, novel irradiation platforms (18, 19) were built at the Radiological Research Accelerator Facility (RARAF). These facilities can generate a wide range of dose rates (from Gy/s to Gy/day) and mixed (photons + IND-neutrons) IND-like exposures, which are relevant for validating the high-throughput dicentric assay (20).

Some exposure scenarios may possess additional challenges for biodosimetry dose assessment. For example, in a severe radiological event, the U.S. Government suggested to dispense the granulocyte colony-stimulating factors (G-CSF) to forestall the symptoms of ARS and severe myelosuppression (21). However, it is not well documented whether a pre-dosimetry use of these radiomitigators may change the magnitude of the biological effects (yield of dicentrics) and interfere with dose prediction.

Currently, there are four Federal Drug Administration-approved ARS agents: Neupogen^® (filgrastim, approved 2015), Neulasta^® (pegfilgrastim, approved 2015), Leukine^® (sargramostim, approved 2018), and Nplate^® (romiplostim, approved 2021) to increase neutrophil (first 3 agents) or platelet (Nplate) production (22). Ideally, exposed individuals who qualify for this treatment (e.g., through a yes/no point-of-care test) should initiate it within 24 h postirradiation, even before determining the actual dose using more elaborate techniques, such as DCA.

G-CSF regulates the production of white blood cells, mainly granulocytes, to protect an exposed organism from infections (23). Granulocyte activation occurs via binding the G-CSFs molecules to the G-CSFR receptors on the cell surface (24). It has been documented that T-lymphocytes also have a G-CSFR receptor, and the G-CSF has been shown to alter activity and gene expression profile in T-lymphocytes (25–27). While G-CSF affects multiple pathways on T-lymphocytes, including cell growth and maintenance, it remains to be determined as to whether the G-CSF directly affects the DNA repair pathways resulting in higher-than-expected yields of dicentrics in T-lymphocytes and, hence, imprecise interpretation of the dose.

Here, we reproduced several possible exposure scenarios that can presumably affect survivors in the perimeter of a nuclear detonation of an IND device. These were: (a) pure vs. mixed radiation fields (photons + IND-neutrons) with a neutron energy spectrum that mimic Hiroshima A-bomb (0.2–9 MeV), and (b) the range of dose rates covering prompt (up to 3 × 10⁵ Gy/s) to protracted (0.4 cGy/min) components. Beyond, we addressed whether the use of G-CSF (Neupogen) or pegylated G-CSF (Neulasta) has any impact on the yield of dicentrics, thereby affecting the precision of dose determination.

MATERIALS AND METHODS

Unless otherwise noted, all reagents and plasticware used in this study were purchased from Thermo Fisher Scientific^™ (Waltham, MA).

Blood Samples

The study was approved by Columbia University’s Institutional Review Board (IRB) protocol #AAAR0643. After the informed consent, peripheral blood was drawn by venipuncture from apparently healthy volunteers with no recent history of exposure to ionizing radiation or clastogenic agents. A total of 25 people participated in this study (15 females and 10 males aged 27 to 63 years old). Freshly drawn blood was heparinized [vacutainers with sodium heparin anticoagulant (Becton Dickinson, NJ)], aliquoted into 2D-barcoded tubes (Matrix Storage Tubes), and exposed to different radiations, doses, and dose rates. Irradiations were performed on several irradiation platforms located at RARAF, CRR, and Columbia University Irving Medical Center (CUIMC), Department of Radiation Oncology. The control samples were sham irradiated, receiving a 0 Gy dose. Experiments were carried out in replicates to verify the reproducibility of the experimental results.

Irradiation Setup for a MIXed (Photons + IND-Neutrons) Fields Study

The energy spectrum of research reactors differs from the energy spectrum of an IND device. The Columbia IND-spectrum Neutron Facility (CINF) generates a neutron field with a broad energy spectrum (0.2–9 MeV) that mimics the spectrum produced by the Hiroshima A-bomb accompanied by a concomitant γ-ray component of about 20% (18, 28, 29). To add the photon energies, we used a filtered 250-kVp Coronado Orthovoltage X-ray machine (Pittsburgh, PA) located on-site. The blood samples were irradiated as described elsewhere (30) at various proportions of neutrons to photons with an increasing percentage of neutrons from 10% to 37% (to cover the range of neutron contributions in the survivable zone after the detonation of a 10-kiloton IND device). The dose rate for the neutron exposures was 2–3 Gy/h, and the dose rate for the X-ray exposures was 1.2 Gy/min (using 250-kVp X rays; 15 mA; 0.5 mm Cu + 1 mm Al filter; HVL 2 mm Cu). As controls, pure X rays (with a maximum dose of 3 Gy) or pure neutrons [with a maximum dose of 1.2 Gy accompanied by 0.24 Gy of extra γ rays but without added X rays (83% neutrons)] were used. The calibration of the CINF, dose, and dose rate monitoring were performed using a custom A-150 muscle tissue-equivalent (TE) gas ionization chamber. Photon dosimetry for the concomitant photons was performed with a compensated Geiger-Mueller dosimeter. Dosimetry for orthovoltage X rays was performed prior to irradiation using a victoreen R-chamber (18).

Irradiation Setup for a Dose Rate Study

Low- and conventional dose rates using low-energy X rays.

The blood samples were placed inside of XRAD 320 (Precision X-Ray, North Branford, CT) on the platform located at 41.5 cm from the X-ray target. The XRAD was set for 320-kVp, 12.5 mA with a custom Thoreaus filter (1.25 mm Sn, 0.25 mm Cu + 1.5 mm Al; HVL 4 mm Cu) at a conventional dose rate (1 Gy/min) irradiation. The separation between the blood samples was less than 2 cm. The dose and dose rate were determined using a Radcal 10×6–6 (Moravia, CA) cylindrical ion chamber and verified using a built-in parallel plate transmission chamber integrated into the filter assembly. Dose uniformity was evaluated using EBT3 radiochromic film (Ashland Advanced Materials, Bridgewater, NJ). Calibration of the XRAD 320’s built-in transmission chamber was performed using an N30013 ion chamber (PTW Farmer, Freiburg, Germany). For lower dose rates, the irradiation setup was similar to the above, except that we used two other settings to reduce the dose rate: 0.1 mA, source to sample distance (SSD) 75 cm (for 0.25 Gy/h), or 2.75 mA, SSD 75 cm (for 4 Gy/h), both using the same voltage and filtering as above. To maintain optimal culture conditions (37°C, 5% CO₂) and minimize scattered radiation when irradiation times were prolonged, the blood samples were pre-mixed with RPMI-1640/10% FBS (1:1 ratio) and placed in a custom incubator made mainly of plastic (31).

Conventional and high dose rates using high-energy X rays.

A Varian TrueBeam LINear ACcelerator (LINAC) was used. A 3-cm-thick solid water phantom slab inserted with three blood samples was placed as close as possible to the gantry surface. The gantry angle was rotated to 90 degrees, and the phantom height on the treatment table was adjusted to align the beam’s central axis to the center of the blood samples. A 10 × 10 cm field size, flattening filter free (FFF) 10-MV X-ray beam of Varian TrueBeam, at 2400 MU/min, was used for high-dose-rate irradiation. The blood samples were 1.5 cm from the phantom surface, and 1 cm phantom was added for maximum dose rate (d_max is 2.3 cm). The dose profile of the FFF beam is peaked at the middle. The separation between the blood samples was less than 2 cm, and the dose difference between the blood sample in the middle (highest dose) and the adjacent blood sample was about 4%. The distance from the X-ray target to the center of blood samples is 59 cm. The dose and dose rate were measured with an ADCL-calibrated ion chamber in a solid water phantom at the same position. The highest dose rate at the center of blood samples was 1.1 Gy/s. Dosimetry for the Varian TrueBeam was performed on the day of irradiation using an N30013 ion chamber (PTW). For the conventional dose rate, the setup was similar to the above, and the irradiation distance was 100 cm from the target of X ray to the center of the blood samples (Supplementary Fig. S1; https://doi.org/10.1667/RADE-22-00007.1.S1). A 10 × 10 cm field size, 10-MV X-ray beam of Varian TrueBeam, at 100 MU/min, was used for 1 Gy/min dose rate irradiation. The blood samples were 1.5 cm from the phantom surface, and 1 cm phantom was added so that the center of samples was at d_max. The dose and dose rate were measured with an ADCL-calibrated ion chamber in a solid water phantom at the same position. The dose rate at the center of blood samples was 1 Gy/min, and the dose and dose rate to all blood samples in this setup were uniform within 1%.

Conventional, high-, intermediate- and ultra-high dose rates using electrons.

The FLASH irradiator [a custom-modified decommissioned Varian Clinac 2100 (Varian medical Systems, Palo Alto, CA)] was used to deliver electrons at intermediate and ultra-high dose rates in two modes: FLASH (9-MeV electron mode) or superFLASH (6-MV photon mode with the target retracted) (32). In FLASH mode, blood samples were irradiated using 9-MeV electrons at average dose rates of 1 Gy/min (SSD 171 cm, 1–2 electron pulses per second), 1 Gy/s [60 cm beyond the isocenter (SSD 171 cm, 180 pulses per second)], 90 Gy/s (at the top of the Y jaws), and 600 Gy/s (at the position of the field illumination mirror, 3 Gy: SSD 20 cm, 2 pulses with 5.5 ms gap; 8 Gy: SSD 24 cm 3 pulses with 5.5 ms gaps). Pulse duration for the 9 MeV electrons is approximately 0.1 μs with a repetition rate of 180 Hz; for the 6 MeV electrons about 40% of the energy is delivered in 1 μs with the rest in the following 4 μs (33). In superFLASH mode (6-MV photon mode with the conversion target removed), the samples were irradiated at a dose rate of 300 MGy/s (single pulse). The dosimetry was performed on the day of the irradiation using a NIST-traceable Advanced Markus Ion Chamber, irradiated side by side with the samples (for dose rates of 1 Gy/min and 1 Gy/s) and using EBT3 Film (Ashland, Wilmington, DE) for higher dose rates. More details are available elsewhere (33).

Irradiation Setup and Radiomitigators for Cytokine Treatment Study

To mimic administering the drugs after irradiation, we exposed blood samples to γ rays (¹³⁷Cs) at a dose rate of 0.665 Gy/min. Irradiation setup for Gammacell-40 (Atomic Energy of Canada Ltd., Chalk River, Canada) was described previously (11). Calibration and dose rate monitoring were performed routinely by the chief medical physicist at Columbia University Irving Medical Center’s Department of Radiation Oncology using thermoluminescent dosimeters. After irradiation, the samples were immediately mixed with RPMI-1640 (1:1 ratio) containing Neupogen (Amgen, LOT# PIM004110471, 600 μg/ml) or Neulasta (Amgen, LOT# PIM00403124, 10 mg/ml). The final concentration for each drug was established through conversion from [gram/mass of the donor] into [μg/mL]. The recommended doses for victims of a radiation accident are (34, 35): Neupogen: 10 μg/kg/day; Neulasta, two doses, 6 mg each (,45 kg). In NHP studies Neupogen is also 10 μg/kg/day, while Neulasta: 300 μg/kg (36). On average, participants of our study were approximately 65 kg, and a total volume of blood of about 5 L. Hence, the concentrations were converted as follows: Neupogen, 10 μg/kg for 65 kg adult = 650 μg in 5 L blood = 0.1 μg/mL [final]; Neulasta, ~8.7 mg for 65 kg adult (65 kg * 6 mg Neulasta/45 kg) in 5 L blood = 1.7 μg/mL [final]. To simulate a delay between administering the drug and biodosimetry screening, the samples were kept with Neupogen or Neulasta in an incubator for 24 or 48 h before being processed by the RABiT-II DCA.

Automated Dicentric Assay in Multiwell Plates

The RABiT-II DCA was performed robotically at Columbia Genome Center High-Throughput Screening Facility on a High Throughput Screening platform (Perkin Elmer cell::explorer) as described previously (37). In brief, 30 μl aliquots of irradiated blood were cultured in plastic 96-well plates for 44 h and subsequently treated with 0.1 μg/ml colcemid for additional 8 h (a total lymphocyte culture time of 52 h). This treatment time does not let the majority of cells reach the second division (38). We estimated that in the RABiT-II DCA, approximately 93.8% and 6.2% metaphases in the first and second divisions should be expected, respectively (11). At the end of the culture time, the cells were swollen in a hypotonic solution and fixed with 3:1 methanol:acetic acid. After this, fixed cells were lysed in 1:1 water:acetic acid and solutions containing individual chromosomes were transferred into glass-bottom 96-well plates and stained using a modified 3-h PNA FISH hybridization protocol that does not require high temperatures to give access of the centromere/telomere probes to DNA. Subsequently, plates were imaged using a BioTek Cytation Cell Imaging Multi-Mode Reader (with 20× objective) and analyzed using custom software, FluorQuantDic v.4. The software classifies the chromosomes as monocentric or dicentric based on the number of detected centromere and telomere spots. The classifier parameters used in this study are given in Supplementary Table S1 (https://doi.org/10.1667/RADE-22-00007.1.S2).

Statistical Analysis

The yields of dicentrics (Y) detected by the FluorQuantDic v.4 software were plotted against the dose using the linear (Y = αD) or the linear-quadratic (Y = αD + βD²) mathematical functions, where D is a dose and α and β are the parameters. The standard errors for the dose-response curves were calculated using the Dose Estimate v.5.3 software (39). The uncertainties for α/β quotients were estimated by propagating the relative errors in quadrature, where SE represents standard error:

$$\alpha /{\beta }_{SE}=\sqrt{{(\frac{{\alpha }_{SE}}{\alpha })}^2+{(\frac{{\beta }_{SE}}{\beta })}^2}\ast \frac{\alpha }{\beta }.$$

The relative biological effectiveness (RBE) values of the different radiations and dose rates were extracted from the dose-response curves at a DNA damage level equivalent to 2 Gy of the monoenergetic γ rays (the reference radiation recommended by ICRP and NCRP) using the following formula: RBE [2 Gy] = reference radiation/test radiation.

The mitotic indexes (MI) in the dose rate study were calculated using the formula: MI = number of cells with visible chromosomes / total number of lymphocytes *100%.

Significance in the dose rate study was assessed using Students t-test. Potential effects of the radiomitigators were accessed using a one-way Analysis of variance (ANOVA) test integrated into GraphPad Prism v. 9.3.1 software.

In experiments with mixed fields, the ratios of neutron vs. photon components were more complex. Thus, we modeled the yield of dicentric/monocentric chromosomes (variable y) as a function of neutron and photon dose components using the standard linear-quadratic (LQ) formalism. We allowed the α and β parameters to be different for neutrons and photons. Interaction terms between photon and neutron doses were considered to evaluate possible deviations from additivity in the dicentric yields induced by different radiations. The analyses were performed in R 4.2.0 software. To estimate LQ model parameters and evaluate potential variability in the dicentric yield between different blood donors, we performed multiple linear regression (using the lm command in R) with dicentric yield Y as the dependent variable, and photon dose, photon dose², neutron dose, neutron dose², and donor (a categorical variable) as independent variables. Inverse variance weighting was used in the regression, with the variance for each data point estimated based on the total number of scored chromosomes (monocentrics + dicentrics). We evaluated regressions with all possible combinations of independent (predictor) variables and their binary interactions using multimodel inference (MMI) implemented by the glmulti R package. Importance scores were estimated for each predictor variable and interaction of variables based on the Akaike information criterion with sample size correction (AICc). Those predictor variables with the highest importance scores were retained for further analysis. Assessments for possible multi-collinarity of predictors were performed using variance inflation factor (VIF) calculations. Using the retained set of predictor variables, we performed a more detailed mixed-effects regression (using the lme4 R package), where random effects by donor were included along with fixed effects. This approach was intended to account for inter-donor variability in model parameters.

RESULTS

Effects of Neutron and Mixed (Photons + IND-Neutrons) Radiation Fields

The yields of dicentrics detected by the RABiT-II system in human blood cultures exposed to pure neutron, photon, and electron radiations at a conventional dose rate of 1 Gy/min are shown in Fig. 1A. The RABiT-II chromosome scoring results used to generate these curves are listed in Table 1. The RABiT-II DCA data for the ¹³⁷Cs γ-ray curve were determined previously and can be found elsewhere (37).

FIG. 1.

Linear energy transfer (LET) effects of pure neutrons, electrons, or photons measured by the high-throughput dicentric assay: Panel A: The RABiT-II DCA dose-response curves for dicentric chromosome aberrations produced by five radiation qualities: IND-spectrum neutrons (LET of ~70 keV/μm), photons (320-kVp X rays, LET of 0.4 keV/μm; 10-MV X rays, LET of 0.2 keV/μm; ¹³⁷Cs γ rays, LET of 0.91 keV/μm), and electrons (LET of 0.2 keV/μm). Each value represents an average of the dicentrics per total number of chromosomes scored automatically by the chromosome identification software (Table 1). Error bars indicate standard errors (n = 3). Panel B: RBE as a function of LET for human lymphocytes exposed to various radiation types: a comparison between RBE values extracted by the RABiT-II DCA and the classical DCA (6, 12–14, 16). All RBE values were calculated at a damage level equivalent to 2 Gy of γ rays. The black-dashed line shows the general trend of the data.

TABLE 1.

The RABiT-II DCA Frequencies of Normal and Dicentric Chromosomes for Different Radiation Qualities and Dose Rates Used to Generate the Curves shown in Figs. 1A, 3A and B

Radiation	Dose-rate	Dose (Gy)	Chr	DC	Y ± SE	Number of donors/gender ratio
Neutrons (IND-spectrum)	0.05 Gy/min	0	4278	64	0.015 ± 0.002	4 (2 males + 2 females)
		0.6	4685	105	0.022 ± 0.002
		1	2762	92	0.033 ± 0.003
		1.2	2220	77	0.035 ± 0.004
		1.4	2957	97	0.033 ± 0.003
		1.6	2939	106	0.036 ± 0.004
Photons (320-kVp X rays)	1 Gy/min	0	15907	140	0.009 ± 0.001	6 (3 males + 3 females)
		1	14069	145	0.010 ± 0.001
		2	10641	137	0.013 ± 0.001
		4	9743	356	0.037 ± 0.002
		8	6758	586	0.087 ± 0.004
	4 Gy/h	0	5658	63	0.011 ± 0.001	4 (2 males + 2 females)
		2	2219	38	0.017 ± 0.003
		4	1340	30	0.022 ± 0.004
		8	537	26	0.048 ± 0.009
	0.25 Gy/h	0	5971	52	0.009 ± 0.001	4 (2 males + 2 females)
		2	9594	140	0.015 ± 0.001
		4	3776	79	0.021 ± 0.002
		8	2067	95	0.046 ± 0.005
Photons (10-MV X rays)	1 Gy/min	0	25335	233	0.009 ± 0.001	4 (2 males + 2 females)
		1	21185	232	0.011 ± 0.001
		2	17142	255	0.015 ± 0.001
		4	13951	309	0.022 ± 0.001
		8	11687	671	0.057 ± 0.002
	1 Gy/s	0	16278	188	0.012 ± 0.001	6 (3 males + 3 females)
		1	13237	164	0.012 ± 0.001
		2	12184	165	0.014 ± 0.001
		4	9109	280	0.031 ± 0.002
		8	7560	533	0.071 ± 0.003
Electrons (9-MeV or 6-MeV)	Control	0	12662	163	0.013 ± 0.001	8 (2 males + ± females)
	300000 Gy/s	3	7337	140	0.019 ± 0.002
	600 Gy/s	3	9135	193	0.021 ± 0.002
		8	4576	234	0.051 ± 0.003
	90 Gy/s	3	7994	188	0.024 ± 0.002
		8	3535	236	0.067 ± 0.004
	1 Gy/s	3	8898	210	0.024 ± 0.002
		8	3701	249	0.067 ± 0.004
	1 Gy/min	3	8099	151	0.019 ± 0.002
		8	3165	226	0.071 ± 0.005

Note. The data was pooled for 18 healthy volunteers of different ages, genders, and races (7 males and 11 females) and three replicates.

Abbreviations. Chr: chromosomes; DC: dicentric chromosomes; Y ± SE: dicentric yield ± standard error.

The RBE values extracted from the dose-response curves (Table 2) indicated that, at a dose rate of 1 Gy/min, the potential of highly-filtered low-energy 320-kVp X rays to induce dicentrics was higher than those of high energy 10-MV X rays. The IND-spectrum neutrons induced dicentrics more efficiently than photons or electrons and exhibited an RBE value of around 6.6 compared to the reference radiation. Overall, the results of the RABiT-II DCA were in good agreement with RBE values extracted by other researchers from the classical DCA curves (Fig. 1B and Table 2, column 6).

TABLE 2.

The RBE Values for Three Radiation Qualities (IND-Neutrons, Photons and Electrons) Extracted from the Dose-Response Curves Obtained Using the Classical DCA or the RABiT-II DCA (Figs. 1A, 3A and B)

	Radiation
Study		Type	Mean energy	Dose rate	Dose range	RBE*
Lloyd DC, et al. (6, 13, 14)	Neutrons,	fission spectrum	0.7 MeV	0.5 Gy/min	0.5–3 Gy	5.7
		D-Be cyclotron	7.6 MeV	0.3 Gy/min	0.27–3.24 Gy	4.0
		D-T generator	14.7 MeV	0.3 Gy/min	0.05–3.03 Gy	2.7
	Photons,	250-kVp X rays	not given	1 Gy/min	0.05–8 Gy	1.2
				0.2 Gy/h	0.05–8 Gy	0.9
		60 Co γ rays	1.25 MeV	0.5 Gy/min	0.25–8 Gy	1.0
				~0.2 Gy/h	0.25–8 Gy	0.7
Prasanna P, et al. (12)	Neutrons,	fission spectrum	0.71 MeV	0.25 Gy/min	0.75–2.5 Gy	2.7
	Photons,	250-kVp X rays	0.083 MeV	1 Gy/min	0.25–3.5 Gy	1.3
		60 Co γ rays	1.25 MeV	1 Gy/min	0.25–5 Gy	1.0
Karthik K, et al. (16)	Photons,	225-kVp X rays	not given	0.66 Gy/min	0.1–5 Gy	1.1
		6-MV X rays	not given	3 Gy/min	0.1–5 Gy	1.5
		60 Co γ rays	1.25 MeV	1 Gy/min	0.1–5 Gy	1.0
This study	Neutrons,	IND-spectrum	1 MeV	0.05 Gy/min	0.6–1.6 Gy	6.6
	Electrons,	9 MeV	9.10 MeV†	1 Gy/min	3–8 Gy	1.2
				1 Gy/s	3–8 Gy
				90 Gy/s	3–8 Gy
				600 Gy/s	3–8 Gy
	Photons,	320-kVp X rays	0.16 MeV§	1 Gy/min	1–8 Gy	1.3
				4 Gy/h	1–8 Gy
				0.25 Gy/h	1–8 Gy
		10-MV X rays	3 MeV‡	1 Gy/min	1–8 Gy	0.9
				1 Gy/s	1–8 Gy
		137 Cs γ rays	0.662 MeV	1 Gy/min	2–10 Gy	1.0

^*In each study, the RBE calculated at 2 Gy γ-rays equivalent.

^†From table 1 of ref. (90).

^§Calculated using SpekCalc (91).

^‡From ref. (92).

Next, we assessed the effects of the mixed (photons + IND-neutrons) field with a set of neutron fractions presumably expected in the perimeter of an IND detonation. Selected neutron percentages (10%, 19%, 37%) were set to mimic realistic neutron fractions in an urban environment [cf. fig. 5.13 in (40)]. The results of the automated scoring indicated that dicentric yields in the mixed field tended to be intermediate between those from pure photon and 83% neutron exposures (Table 3 and Fig. 3A). However, contrary to expectations, there was no apparent correlation between dicentric yields and the proportion of the neutron component in the mixed field (colored dashed lines, Supplementary Fig. S2A; https://doi.org/10.1667/RADE-22-00007.1.S3).

TABLE 3.

The RABiT-II Frequencies of Normal and Dicentric Chromosomes in Mixed (Photons + IND-Neutrons) vs. Pure Radiation Fields Shown in Fig. 2

Neutron (%) in a mixed field *	Contribution of each component (Gy)
	Neutron	X ay	Extra γ ray*	Total	Chr	DC	Y actual ± SE	Y predicted
Control	0	0	0	0	24266	320	0.013 ± 0.002	0.012
0% Neutrons (X rays)	0	1	0	1	18913	320	0.017 ± 0.001	0.014
	0	2	0	2	8437	200	0.024 ± 0.002	0.021
	0	3	0	3	16452	501	0.030 ± 0.003	0.032
10% Neutrons	0.1	0.92	0.02	1.04	23686	482	0.020 ± 0.001	0.015
	0.2	1.84	0.04	2.08	13538	439	0.032 ± 0.002	0.022
	0.3	2.76	0.06	3.12	16818	766	0.046 ± 0.002	0.033
19% Neutrons	0.2	0.84	0.04	1.08	14344	221	0.015 ± 0.001	0.016
	0.4	1.68	0.08	2.16	13926	408	0.029 ± 0.002	0.024
	0.6	2.52	0.12	3.24	15163	532	0.035 ± 0.003	0.035
37% Neutrons	0.4	0.68	0.08	1.16	14812	334	0.023 ± 0.001	0.019
	0.8	1.36	0.16	2.32	18675	509	0.027 ± 0.002	0.027
	1.2	2.04	0.24	3.48	19027	854	0.045 ± 0.002	0.038
83% Neutrons (Max.)	0.2	0.04	0.04	0.28	23200	377	0.016 ± 0.001	0.015
	0.4	0.08	0.08	0.56	27897	528	0.019 ± 0.001	0.018
	0.8	0.16	0.16	1.12	23323	535	0.023 ± 0.001	0.023
	1.2	0.24	0.24	1.68	20138	632	0.031 ± 0.002	0.029

Note. The data was pooled for three healthy volunteers of different ages, genders, and races (1 male and 2 females) and four replicates.

^*Neutron spectrum is accompanied by a concomitant photon component of about 20%.

Abbreviations. Chr: chromosomes; DC: dicentric chromosomes; Y ± SE: dicentric yield ± standard error.

FIG. 3.

Dose rate effects measured by the high-throughput dicentric assay: Panel A: Dose-response curves for dicentric aberration yields induced by photon exposures. Dashed lines with black markers correspond to high-energy X rays (10 MV); solid lines with white markers correspond to highly filtered low-energy X rays (320 kVp). Panel B: Dose-response curves for dicentric aberration yields induced by 9-MeV electrons. Each value represents an average of the dicentrics per total number of chromosomes scored automatically by the chromosome identification software (Table 1) and plotted against the dose. Error bars indicate standard errors (n = 3). Panel C: The effects of the dose rate on dicentric yields as a function of absorbed energy per unit of time (Gy/s). Dicentric yields caused by 320-kVp X rays are shown by straight lines with black markers, 10-MV X rays by dotted lines with gray markers, and 9-MeV electrons by dashed lines with white markers. For all radiations, 3 Gy data is shown in triangle-shaped markers while 8 Gy data – is in circle-shaped markers. Each data point indicates the mean value (a total of 10 and 8 donors in photon or electron study, respectively) ± standard error (n = 3).

To determine whether this result was influenced external factors such as biological variability and small sample size, first, we assessed whether the actual yields of dicentrics in our experiments (Table 3, column 8) deviate from those predicted by statistical modeling. For this, we extracted LQ coefficients from weighted dicentric yields (Y) using the fixed effects regression model analysis: predicted baseline dicentric yield in unirradiated cells, c = 1.18 × 10⁻² ± 8.0 × 10⁻⁴, P value = <2 × 10⁻¹⁶; predicted β for photons, β_p = 2.23 × 10⁻³ ± 2.26 × 10⁻⁴ Gy⁻², P value = <2 × 10⁻¹⁶; α for neutrons, α_n = 1.43 × 10⁻² ± 1.52 × 10⁻³ Gy⁻¹, P value = <2 × 10⁻¹⁶. We used these parameter values to calculate predicted (fitted) dicentric yields for the different experimental conditions, and compare them with corresponding observed values.

Multimodel inference (MMI) on weighted linear regressions with all possible combinations of predictor variables and their binary interactions suggested that the most important variables for describing this dataset were D_p² (photon dose²) and D_n (the neutron dose). In contrast, D_p (photon dose) and D_n² (neutron dose²) were less important and did not reach statistical significance. This was in agreement with the classical radiation biology concept: sparsely ionizing radiation (photons) tends to cause linear quadratic dose responses, whereas densely ionizing radiation (neutrons) produces linear dose responses. This is a consequence of the ionization patterns formed by the different irradiation modalities: photons form sparse damage, and a single photon is unlikely to result in breakage of adjacent chromosomes, forming a dicentric. On the other hand, neutrons, by generating recoil ions, form dense tracks; therefore, adjacent chromosomes are more likely to break due to the action of a single neutron. Thus, multiple independent photons are needed to form dicentrics, while single neutrons are sufficient to reach the same effect (41).

Since only D_p² and D_n (but not D_p and D_n²) were important, the predicted dicentric yields were calculated by the following equation:

$$Y=c+{\beta }_p{D}_p^2+{\alpha }_n{D}_n$$

Next, using calculated predicted yield values (Table 3, column 9), we plotted actual vs. predicted dicentric yields for each type of field reproduced in this study [Supplementary Fig. S2B–F (https://doi.org/10.1667/RADE-22-00007.1.S3), red markers (actual yield) vs. blue trend lines (predicted yield)]. This simple 3-parameter model provided a decent fit to the data, with R² = 0.477, although the model assumes that Y yields should increase proportionally with the fraction of neutrons in the field, whereas the observed data did not always conform to this expectation.

Plotted graphs revealed that 10% neutron dataset had deviations (Supplementary Fig. S2C; https://doi.org/10.1667/RADE-22-00007.1.S3) from the expected range of dicentrics, while 19% and 37% neutrons experimental data fell more closely to the predicted area. Both controls (0% and 83% neutrons) were fit well within the predicted range (Supplementary Fig. S2B and F). Hence, we assume that actual dicentric yields in the 10% neutron dataset were higher due to external experimental factors. The pooled summary for actual yields (white or black symbols) and its deviations from predicted yields (trend lines) is summarized in Figure 3B.

Next, to assess whether intermediate dicentric yields in mixed field datasets were caused by additive or synergistic interactions between photons and IND-neutrons in mixed fields, we performed a more rigorous statistical analysis using a mixed effects regression model. The photon dose-response was dominated by the quadratic (β) component, whereas the linear (α) component was not significantly different from zero and had a best-fit value of 2.43 × 10⁻³ ± 1.76 ± 10⁻³ Gy⁻¹, P value 0.169.

Mixed-effects regression versions with random intercepts and slopes by donor for either the photon or neutron dose responses turned out to be too complex for the analyzed data set, resulting in singularities during fitting. A simpler variant with random intercepts only, which assumes that inter-donor variability affects only the baseline dicentric yield but does not apply to radiation responses, was able to fit the data. The best-fit parameter values for the mixed effects regression model 6 their standard errors were: c = 1.30 × 10⁻² ± 1.84×10⁻³, p-value = 0.011; random effects standard deviation (i.e. inter-donor variability) = 2.84 × 10⁻³; β_p = 2.23 × 10⁻³ ± 2.17 × 10⁻⁴ Gy⁻², P value = <2 × 10⁻¹⁶; α_n = 1.41 × 10⁻² ± 1.46 × 10⁻³ Gy⁻¹, P value = <2 × 10⁻¹⁶. These parameters are quite similar to those for the fixed effects model above, suggesting that inter-donor variability in the intercept values was not very important.

Overall, the statistical analysis suggests that the dicentric yield from neutron radiation was dominated by the linear α component, whereas the yield from photon radiation was dominated by the quadratic β component. Hence, interactions between photons and IND-neutrons were rather additive but not synergistic (no significant multiplicative interactions).

Dose Rate Effects

Next, we simulated a range of the dose rates (from several Gy per ls to only a few Gy per several days). The results for photons or electrons delivered at low (Gy/h), high (Gy/s), intermediate or ultra-high dose rates are shown in Fig. 3A and B, and Table 1. The α and β coefficients and the α/β quotients for each dose rate are given in Table 4.

TABLE 4.

Values of the Coefficients α and β in the Equation Y = αD + βD² and α/β Quotients Which Gives the Dose-Response Curves for Photons or Electrons Delivered at a Range of the Dose Rates Shown in Fig. 3A and B

Radiation	Dose rate	α ± SE × 10−3	β ± SE × 10−3	α/β ± SE
320-kVp X rays	0.25 Gy/h	0.8 ± 0.9	0.5 ± 0.2	1.6 ± 1.9
	4 Gy/h	0.9 ± 1.6	0.7 ± 0.3	1.3 ± 2.4
	1 Gy/min	1.1 ± 0.9	1.1 ± 0.1	1.0 ± 0.8
10-MV X rays	1 Gy/min	0.9 ± 0.6	0.6 ± 0.1	1.5 ± 1.0
	1 Gy/s	1.0 ± 0.8	0.8 ± 0.1	1.3 ± 1.0
9-MeV electrons	1 Gy/min	1.3 ± 1.0	1.1 ± 0.2	1.2 ± 0.9
	1 Gy/s	1.6 ± 1.0	0.6 ± 0.2	2.7 ± 1.9
	90 Gy/s	1.6 ± 1.1	0.6 ± 0.2	2.7 ± 2.0
	600 Gy/s	1.5 ± 1.0	0.4 ± 0.1	3.8 ± 2.7

Figure 3C summarizes the effects of the dose rate on the dicentric yields. At a conventional dose rate (1 Gy/min), photons (black and gray markers) and electrons (white markers) exhibited different potentials to induce dicentrics. With decreasing the dose rate, the yields of dicentrics decreased. A significant correlation was observed at the following transition points: from 1 Gy/min to 4 Gy/h and lower (low-energy X rays) and from 1 Gy/s to 1 Gy/min (high-energy X rays). For electrons, no significant changes in dicentric yields were observed from 1 Gy/min to 90 Gy/s: the dicentric yields were approximately similar. However, at the dose rate of 600 Gy/s, we observed a significant decrease in dicentric yields (P value = 0.004, t-test, 8 donors, n = 3). It should be noted that “600 Gy/s” is an average dose rate, as the dose was delivered in number of pulses (as specified in the Methods, the actual dose was delivered as two sub-μs pulses of 1.5 Gy or three sub-μs pulses of 2.7 Gy with a 5.5 ms gap between pulses). Similar results were obtained in our laboratory using the classical version of the DCA (32) and in a large demographic study (800 individuals) ongoing at the moment (33). Notably, the dose rate effects were more clearly pronounced at 8 Gy than at 3 Gy.

Preliminary, we assumed that this decrease of dicentrics at the ultra-high dose-rate samples was caused by mitotic delay. However, no differences in mitotic indexes were observed between conventional and higher dose rate samples (Supplementary Table S2; https://doi.org/10.1667/RADE-22-00007.1.S4). This result excludes the possibility that the ultra-high dose rate induced a cell cycle arrest, resulting in lower dicentric yields.

Effects of Radiomitigators

Figure 4 shows the yields of dicentrics detected in blood samples (Table 5) treated with Neupogen or Neulasta for 24 or 48 h after a photon exposure with ¹³⁷Cs γ rays.

FIG. 4.

Effects of radiomitigators measured by the high-throughput dicentric assay. Dicentric yields from human lymphocytes precultured for 24 or 48 h with 0.1 μg/mL Neupogen (NEP, green lines) or 1.7 μg/mL Neulasta (NEU, orange lines) after exposure to ¹³⁷Cs γ rays (Table 5). Twenty-four-hour data is shown by squared markers, 48-h data – by triangle markers. Each data point indicates the mean value (4 donors) ± standard error (n = 3).

TABLE 5.

Frequencies of Normal and Dicentric Chromosomes in Lymphocytes Pre-Cultured with Neupogen or Neulasta for 24 or 48 h Postirradiation Using ¹³⁷Cs γ Rays at a Dose Rate of 0.665 Gy/min Shown in Fig. 4

Time	Dose (Gy)	CTRL			NPO			NEU
		Chr	DC	Y ± SE	Chr	DC	Y ± SE	Chr	DC	Y ± SE
24 h	0	11096	120	0.011 ± 0.001	14462	136	0.009 ± 0.001	15183	161	0.011 ± 0.001
	2	8126	114	0.014 ± 0.001	9870	128	0.013 ± 0.001	8069	123	0.015 ± 0.001
	4	4985	124	0.025 ± 0.002	6918	172	0.025 ± 0.002	4431	127	0.029 ± 0.003
	8	1625	97	0.060 ± 0.006	1734	113	0.065 ± 0.006	1120	64	0.057 ± 0.007
48 h	0	7645	74	0.010 ± 0.001	16572	172	0.010 ± 0.001	17559	185	0.011 ± 0.001
	2	4766	51	0.011 ± 0.001	8200	108	0.013 ± 0.001	7176	100	0.014 ± 0.001
	4	2910	63	0.022 ± 0.003	4627	122	0.026 ± 0.002	3309	84	0.025 ± 0.003
	8	880	58	0.066 ± 0.009	1080	62	0.057 ± 0.007	633	40	0.063 ± 0.010

Notes. The data pooled for 4 healthy volunteers of different ages, genders, and races 2 males (31-year-old Asian and 63-year-old white) and 2 females (38-year-old and 46-year-old African Americans) and three replicates.

Abbreviations. Chr: chromosomes; DC: dicentric chromosomes; Y ±SE: dicentric yield ± standard error; CTRL: controls, NPO: Neupogen; NEU: Neulasta.

To answer the question of whether the mean dicentric yields are different for control vs. treated samples, we used a one-way ANOVA test for each dose and each time because dicentrics yield will a priori depend on the dose. Since for each dose (Gy) and time (h), the mean Y values of 3 groups (control, Neupogen, and Neulasta) were compared, a total of 8 tests were performed (4 doses * 2 time points). Thus, the P value threshold was corrected to 0.05/8 = 0.006 (number of tests, 4 doses * 2 time points) as a simple Bonferroni correction for multiple testing.

Results of the statistical analysis revealed no significant differences (Table 6; P > 0.006) between control and test samples in the experimental conditions we reproduced, suggesting no effect of these radiation mitigators on the dicentric yields in human lymphocytes following ex vivo irradiation and pre-incubation with Neupogen or Neulasta for 24 or 48 h. In addition, because for in vitro drug testing it is advised to perform the concentration–response testing with relatively high concentrations (42, 43), we also tested several dilutions, including 100 fold higher concentrations than recommended (10 μg/mL Neupogen or 170 μg/mL Neulasta). However, we did not observe the difference in dicentric yields between control and test samples as well (data not shown).

TABLE 6.

Results of a One-Way ANOVA Analysis.

One-way ANOVA summary	CTRL vs. NPO vs. NEU
	24 h				48 h				24 h vs. 48 h
	0 Gy	2 Gy	4 Gy	8 Gy	0 Gy	2 Gy	4 Gy	8 Gy	0 Gy	2 Gy	4 Gy	8 Gy
df	11	11	11	11	11	11	11	11	23	23	23	23
F	0.206	0.203	0.724	0.012	0.236	1.050	2.100	1.028	0.186	0.389	1.572	0.658
P value	0.818	0.820	0.511	0.988	0.795	0.389	0.178	0.396	0.964	0.850	0.218	0.660
R squared	0.044	0.043	0.139	0.003	0.050	0.189	0.318	0.186	0.049	0.098	0.304	0.154
P value summary	ns	ns	ns	ns	ns	ns	ns	ns	ns	ns	ns	ns
Significant difference?*	No	No	No	No	No	No	No	No	No	No	No	No

Abbreviations. df: degree of freedom; CTRL: controls, NPO: Neupogen; NEU: Neulasta.

^*P < 0.006 after the Bonferroni correction.

DISCUSSION

In terms of biological effects, conventional exposures used to reconstruct dose-response calibration curves may not adequately represent all types of realistic exposures. For example, amount of chromosome damage in cells, exposed at a conventional dose rate of 1 Gy/min to pure (photons or neutrons) field, may not be the same as the damage caused by a nuclear weapon. This is because an IND device may include a mixture of radiation components (photons + neutrons) released in different time scales from prompt (blast) to protracted (fallout or ground shine) dose rates. The resulting concern is whether the biodosimetry assay in question can distinguish variations in the composition of the radiation field and dose rates. To address this question, we simulated exposures resembling (to a certain degree) the effects of realistic nuclear events. This was done with the aim of validating the accuracy of the automated version of the dicentric assay (the RABiT-II DCA), developed to screen many people at a time in the event of a radiation emergency.

Mixed (Photons + IND-Neutrons) Field Effects

The effects of neutrons in A-bomb survivors and neutron RBE in mixed radiation fields produced in Hiroshima and Nagasaki is a matter of long-term discussion (44–48). Indeed, when a nuclear detonation occurs at higher altitudes (airburst of military weapon), neutrons are attenuated faster than photons in air, and neutron doses are higher than photon doses out to ~600 m [fig. 5.7 in (40)]. However, in an urban environment (ground burst, more likely for an IND detonation scenario), photons will be attenuated more strongly than neutrons by construction materials (primarily high Z steel and concrete). In this case, the resulting neutron contribution will be higher (about 1/3 of the total dose at 800 m) [fig. 5.7 in (40)].

From the standpoint of energy absorption, the levels of DNA damage should be proportional to the dose. Hence, if the resulting dose was the same, there is no reason to anticipate drastic differences in the DNA damage between a pure photon field compared to a mixed field which consists of photons and some neutron components (49). At the same time, neutrons deposit energy via a different mechanism (recoil protons) and form dense ionization tracks rather than the more diffuse ionizations caused by Compton- and photo-electrons. This may result in a more complex structure of the breaks (50), which is harder to repair because they comprise several lesions within one or two helical turns of the DNA (51).

Compared to experiments with a mixed field consisting of photons and α particles (52–55), there is limited cytogenetic data (dicentrics as a biological endpoint) addressing the matter of possible interactions of photons with neutrons in inducing DNA damage. To our knowledge, there were two prior cytogenetic studies aimed to gain experience in simulating criticality incidents with mixtures of gamma and neutron radiations. Purrott et al. showed that the ratio of the dicentric yields from gamma and neutrons varies considerably over the possible range of the gamma to neutron proportions (with rising neutron components of 25%, 50%, 55%, 80%, and 90%) expected in criticality exposures (56). In contrast, Voisin et al. (57) suggested no effects of a neutron component (50% and 83%) in a mixed field on resulting dicentric yields.

We speculate that the reason for inconsistency between our conclusions and that of Voisin et al. (57) may be due to the neutron ratios tested: we did not test the induction of dicentrics by the neutron percentages above 37%, as these are not likely to be seen in regions with overall survivable radiation dose (40) after a nuclear detonation [Voisin et al. data (57) is, however, relevant to criticality accidents]. In contrast, our RABiT-II DCA data is consistent with Purrott et al. (56) and indicate some degree of interaction between photons and neutrons in inducing dicentrics in human lymphocytes. Our statistical analysis suggests that the nature of this interaction is rather additive than synergistic (no significant multiplicative interactions), although we were not able to establish a simple correlation between dicentric yields and the proportion of the neutrons in the mixed field due to higher dicentric yields in 10% field likely caused by some external reasons.

Our earlier biodosimetry studies with this type of mixed radiation field (X rays + IND-neutrons + incident γ component) also showed that even a small percentage of added neutrons can modify biological effects compared to the same dose of pure photons. This was demonstrated for various biological endpoints, including radiation-responsive genes, ribosome biogenesis, mRNA translation, the expression of important hematopoietic cytokines, and the formation of micronuclei (30, 58–61). Furthermore, in vivo studies in mice and non-human primate (baboon) models showed that equivalent doses of pure γ rays and mixed-field radiations (photons + neutrons) do not produce equivalent biological effects, and ARS occurs at lower doses of mixed-field radiation (62, 63). The biological effects were rising with an increase in the total neutron fraction (58, 62). Compared to a pure γ field, a decline in lymphocyte counts and delayed decline of neutrophils and platelets were observed (63). Hence, the results for early (gene and protein expression) and late (cell death) biological endpoints were consistent with the observations of dicentrics made in this work. Apparently, the mixed (photons + IND-neutrons) exposures may appear to have a greater suppressive effect on T-lymphocytes and other immune cells than the same dose of pure photons, suggesting a more profound impact on the hematopoietic system and manifestation of ARS.

Dose Rate Effects

In the IND scenario, the dose rate effects need to be studied at both very high- and very low-dose rates. On the high end, photon exposures from the blast of a nuclear detonation occur in a fraction of a microsecond (17), regardless of the distance from the detonation, as all photons travel at the same velocity, and there is no temporal dispersion of the radiation. Thus, individuals in close proximity to the blast will likely receive high doses at high-dose rates, while those further away will receive lower doses but still at high-dose rates. At a survivable distance (~1 km), it is expected that individuals will receive few-Gy doses at dose rates on the order of Gy/μs, where there is the potential for radiation to alter the cellular chemistry and modulate the amount of damage formed (64). This does not hold for neutrons, where “chromatic aberrations” will significantly increase the irradiation time.

On the other hand, when dose rate is lowered, and radiation is delivered over several hours, the time for repair increases (41). In agreement with this, our data suggest that when radiation exposure spreads over several h, a time comparable with DSB repair times (65), aberrations of two-track origin (β values, Table 4) can be modified by repair mechanisms that have time to operate (66). Comparing the α and β coefficients, the main differences in the observed yields were due to the differences in the β terms (aberrations produced by separate tracks). This was consistent with Lloyd et al. (14), which demonstrated that only the β coefficient is dependent on the dose rate and decreases as the dose rate decreases. Interestingly, the β coefficient also decreased at higher dose rates. The α/β quotients for photons did not differ much, while for electrons, they seemed to increase at higher dose rates. However, based on the estimated errors, these differences were not significant.

Moving towards higher dose rates (1 Gy/min to 90 Gy/s), we observed saturation effects in the number of dicentrics induced by electrons. However, when 8 Gy was delivered at a very high dose rate, the number of dicentrics significantly (P= 0.004, t-test) decreased (Fig. 3C). These results were reproducible in our laboratory for an ongoing 800-person demographic study and a blind classical dicentric test made side-by-side manually (32, 33). As we observed no mitotic index changes in 600 Gy/s samples, we assumed that a lower amount of dicentrics in these samples may be related to the so-called “FLASH effect” (64) rather than cell cycle arrest.

Several physico-chemical mechanisms were proposed to cause the “FLASH effect” at ultra-high dose rates. These include a local depletion of oxygen by radiation, resulting in less peroxyl radical damage on DNA (67–70), and hence, less double-strand breaks that can initiate formation of dicentrics. To establish this, Small et al. (71) performed a series of experiments with plasmids to clarify the link between ultra-high dose rates and DNA damage at a nanoscale level. These experiments showed no statistically significant variation in DSB induction between very high- and conventional dose rates. However, generally, Small et al. (71) did not exclude that the plasmid model lacks many key features that may lead to the FLASH mechanism, e.g., oxygen metabolism. In addition, in the cellular environment, the radical scavenging capacity is orders of magnitude higher; hence, the radical chemistry is completely different. Indeed, in experiments with living cells, researchers documented less DNA damage in cells due to enhanced radiation recovery and reduced incidence of radiation-induced senescence (72). In hamster cells, ultra-high dose rates caused fewer dicentric chromosomes (73, 74). These effects were equally applicable to the stem cells (75). Therefore, our observations are consistent with other researchers and support the assumption given by Small et al. (71) that the differences in DNA damage yields can be expected in more complex and metabolically active systems (isolated plasmid DNA vs. living cell).

Theoretical modeling suggests that the FLASH effect takes place at relative oxygen pressures below 5% (76). Indeed, at 1.6% oxygen, the FLASH-sparing effect started from 5–10 Gy at 600 or 800 Gy/s (77, 78). In addition, using alkaline comet assay, Cooper et al. (79) demonstrated that the FLASH dose rates reduce DNA damage in peripheral blood lymphocytes at oxygen levels below 3%. Thus, in this work, elevated radioresistance due to low oxygen consumed by ultra-high dose rate seems to be a very plausible explanation for the reduction in dicentrics observed in samples exposed to 8 Gy at 600 Gy/s. However, the exact cause cannot be specified as we did not measure the actual oxygen levels in our experiments. At this point, we can assume that the initial oxygen level in the tubes resembled normoxic conditions (20%) as the irradiations were performed in equilibrium with the air, and quiescent blood cells were not metabolizing at the moment of exposure. Moreover, we assume that the FLASH dose rate (600 Gy/s) may not significantly deplete oxygen in the tubes, as this can be seen only when the oxygen level was initially below a certain threshold (80).

It is important to mention that in experiments with ultra-high dose rates and normal cells may not respond equally to the presence of oxygen. For example, in normal air, fewer 53BP1 foci were observed in normal fibroblasts after FLASH irradiation but not in the cancer cell line (72). Because in some cells the FLASH effect can be observed even at normal oxygen levels, researchers suggest that the FLASH mechanism may only in part be driven by depletion of oxygen (77). This assumption is also supported by the state-of-art radiation chemistry models, which do not consider that oxygen depletion and transient hypoxia is the only main mechanism for the FLASH effects (81). Thus, it remains to be determined what caused the reduction in dicentrics we observed in 600 Gy/s samples: whether it was indeed because the local oxygen was at some point consumed by radiation delivered at an ultra-high dose rate or due to other external factors (e.g., an intrinsic susceptibility of lymphocytes to FLASH or the individual characteristics of the electron beam and FLASH pulse delivery in our facility (82)) or more complex molecular mechanism.

Yields of Dicentrics Depend on the Energy of X Rays

There has been a push to move away from radioisotope-based irradiators toward orthovoltage- and Linac-based irradiation platforms (83). However, varying the hardness of the radiation impacts ionization density and, indirectly, DSB and, therefore, dicentric yields. Based on microdosimetric considerations, we would expect the megavoltage X rays to have less “clustered” lesions and, therefore, lower dicentric yields than orthovoltage X rays. Indeed, in our experimental system, at higher doses delivered at a conventional dose rate, 320-kVp X rays induced more dicentrics than 10-MV X rays. This conclusion was in good agreement with Schmid et al. (84), who showed increased dicentric yields in human–hamster-hybrid cells exposed to X rays with lower energies.

A similar prior experiment (a comparison between low-energy vs. high-energy X rays using a classical DCA) was conducted by Karthik et al. (16). Although the conclusions of that study were different from those made here [specifically, high-energy X rays induced more dicentrics than low-energy X rays (Table 2, row 3, column 6, RBE6-MV X rays = 1.5 vs. RBE225-kVp X rays = 1.1)]. We assume that this could be a result of the dose rate differences in the Karthik et al. study (16) (Table 2, row 3, column 4: 3 Gy/min 6-MV X rays vs. 0.66 Gy/min 225-kVp X rays). Indeed, our data suggests that dicentric yields were sensitive to the dose rate changes (Fig. 3A and B). In contrast, in our experiments, both radiations were delivered at the same dose rate of 1 Gy/min. Thus, the discrepancies between conclusions made in this study and Karthik et al. (16) may be due to the different dose rates (with a higher dose rate for 6-MV X rays) used in the latter study.

Besides dicentrics, similar conclusions were obtained using the other biological endpoints. For instance, Yachi et al., (85) showed that, at the same absorbed dose, the diagnostic 60–250-kVp X rays induce double-strand breaks more efficiently (more γ-H2AX foci) than therapeutic 6-MV X rays (85). Consistently, using a colony survival assay, Okamoto et al., (86) reported a higher RBE value for 200-kVp X rays compared to 6-MV X rays (86). Thus, in agreement with other researchers, our data showed that the DNA damage tends to be higher as the mean X-ray energy is lower (85). This is primarily due to changes in microscopic energy deposition patterns. Low-energy electrons, such as those generated by orthovoltage X rays, have a short range, and thus most of the photon energy is deposited locally (the range of a 100 keV electron is about 140 μm in water); at higher energies, the range of the secondary electrons increases drastically (the range of a 1 MeV electron is about 4 mm in water) (87) reducing the clustering of the ionization events and hence DSB yields.

Applied to external beam therapy, the X rays with higher energies may be safer for cancer patients because it may reduce the risk of secondary cancers in irradiated healthy tissue. Of course, chromosome aberrations may not be the sole events leading to cancer development; yet, parallelism exists due to similar underlying molecular mechanisms (46).

G-CSF Treatment does not Interfere with the Automated Dicentric Analysis

In an effort to prevent infections and restore bone marrow functions, medical countermeasures can be administered to the victims of an IND explosion (34). Optimally, these drugs should be administered within 24 h postirradiation in-field triage using a yes/no point of care assay (88). In this case, validation of the actual dose for these individuals will happen later, after initiation of the G-CSF treatment. Results published earlier suggest that G-CSF acts as a strong immune regulator and can directly modulate immune responses from T cells via its receptor on the T-lymphocyte surface (26). Hence, because the dicentric analysis is generally performed in T cells, there is a possibility that ongoing G-CSF treatment can affect the results of this test leading to an incorrect interpretation of the dose. We assessed whether 24 and 48 h incubation of irradiated lymphocytes with G-CSF and pegylated G-CSF can interfere with the dose prediction by the RABiT-II DCA. Our measurements and statistical analysis indicated no significant differences between control and test samples [P > 0.006 (Bonferroni corrected), one-way ANOVA, Table 6] treated with cytokines up to 48 h postirradiation (Table 5 and Fig. 4). This result suggests that in the in vitro conditions we reproduced, the cytokine treatment did not affect the biodosimetry prediction of the dose using the high-throughput dicentric test. Earlier at the CRR, the effects of G-CSF on the micronucleus assay were tested, and no effects on micronucleus yields in ex-vivo or in vivo models were found as well. Therefore, this result and conclusion were consistent for both cytogenetic endpoints.

In conclusion, the data obtained in this work demonstrates the capacity of the RABiT-II DCA, a high-throughput dicentric test, to differentiate between various radiation qualities and dose rates. The trend of relationships between RBE values and LET of radiation was consistent with common wisdom [e.g., Fig. 1B in this work and fig. 2 in ref. (89)]. Overall, our experimental data suggest the suitability of the RABiT-II DCA for an emergency biodosimetry assessment aimed to identify individuals requiring medical attention. In a future perspective, we are planning to complement this assay with a machine learning approach proposed by Shuryak et al. (61). This will allow reconstruction of the partial body exposures.

Supplementary Material

Supplementary file 1

Supplementary Fig. S1. 1 Gy/min irradiation setup at a Varian TrueBeam LINear ACcelerator (LINAC).

Supplementary file 2

Supplementary Table S2. The RABiT-II DCA mitotic indexes for different electron dose rates. The data pooled for 4 healthy volunteers (1 male and 3 females).

Supplementary file 3

Supplementary Fig. S2. Dicentric yields in the mixed field study: A. Dicentric dose-response curves for mixed (10%, 19%, or 35% neutrons) and pure (0% or 83% neutrons) radiation fields as a function of the total dose. B-F. Actual vs. Predicted dicentric yield dose reconstruction.

Supplementary file 4

Supplementary Table S1. The classifier parameters for automated chromosome identification using a custom software, FluorQuantDic v.4.

FIG. 2.

Mixed vs. pure radiation field effects measured by the high-throughput dicentric assay: Panel A: A bubble chart shows relationships between three variables: the neutron dose (y-axis), the photon dose (x-axis), and dicentric yields (Table 3, column 8) induced by pure or mixed (photons + IND-neutrons) fields as a function of photon and neutron mixed doses (the third dimension, encoded by the radius of the circles, n = 4). The yield percentages (%) are encoded by the color matrix depicted in the chart on the right. Black-dashed arrows show each percentage of neutrons (%). Panel B: Dicentric dose-response curves for mixed (10%, 19%, or 35% neutrons) and pure (0% or 83% neutrons) radiation fields as a function of total dose. Actual dicentric yields (Table 3, column 8) represent an average of dicentrics per normal chromosomes scored by the software and are indicated by white (mixed field) or black (pure field) symbols. Error bars indicate standard errors. Predicted dicentric yields (Table 9, column 9) are estimated using best-fit parameters for the fixed effects regression model and are shown by colored dashed (mixed fields), black straight (pure photon field), or black dotted (IND-neutron field with parasitic photons) trendlines.