Sex and dose rate effects in automated cytogenetics

Abstract

Testing and validation of biodosimetry assays is routinely performed using conventional dose rate irradiation platforms, at a dose rate of approximately 1 Gy/min. In contrast, the exposures from an improvised nuclear device will be delivered over a large range of dose rates with a prompt irradiation component, delivered in less than 1 μs, and a protracted component delivered over hours and days. We present preliminary data from a large demographic study we have undertaken for investigation of age, sex and dose rate effects on dicentric and micronucleus yields. Our data demonstrate reduced dicentric and micronucleus yields at very high dose rates. Additionally, we have seen small differences between males and females, with males having slightly fewer micronuclei and slightly more dicentrics than females, at high doses.

Introduction

Testing and validation of biodosimetry assays is routinely performed using conventional dose rate irradiation platforms, primarily orthovoltage irradiators, at a dose rate of approximately 1 Gy/min. In contrast, the exposures from an improvised nuclear device (IND) will be delivered over a large range of dose rates with a prompt irradiation component, delivered in less than 1 μs, and a protracted component delivered over hours and days. For planning purposes, the US government assumes an IND to be a 10kT gun-type enriched Uranium device, detonated at ground level^{(1, 2)}. The radiations generated by such a device have been discussed elsewhere⁽³⁾.

It has long been known that dose rate effects are a strong modulator of radiation response: low dose rate exposures exhibit a significant partial damage repair (see⁽⁴⁾ Chapter 5), resulting in lower yields of, for example, dicentrics. Conversely high dose rate irradiations may overwhelm the repair processes resulting in higher yields. Extremely high dose rates, on the other hand, may exhibit saturation effects at high doses possibly due to oxygen depletion, effectively resulting in lower-than-expected late effects⁽⁵⁾.

We have initiated a large demographic study to investigate age, sex and dose rate effects on dicentric and micronucleus yields in vitro, after radiation exposure. Blood was collected from 260 healthy donors (130 male, 130 female; out of a projected 800), of all age groups (from 3 to 75 y old). Blood was aliquoted and irradiated to 3 or 8 Gy at a variety of dose rates and analysed using our RABiT-II micronucleus⁽⁶⁾ and dicentric⁽⁷⁾ assays, with about one-third of the donors only analysed using the dicentric assay. This manuscript presents analysis of sex and dose rate effects seen in the donors accrued to date.

Methods

This study was approved by Columbia University’s Institutional Review Board (IRB) protocol IRB-AAAF2671. Blood from adult donors was collected at Columbia University into sodium heparin vacutainer tubes and stored at room temperature overnight to simulate shipping. Blood from pediatric donors was collected by Jean Brown Clinical Research (IRB ID#8933; Salt Lake City, UT) into sodium heparin vacutainer® tubes and shipped overnight in a temperature-controlled shipper (22°C Credo Cube, Fisher Scientific, Pittsburgh, PA). Blood was then aliquoted into Matrix Storage Tubes (Fisher Scientific, Pittsburgh, PA) and diluted with RPMI1640 (1:4 ratio, Fisher Scientific). The diluted blood samples were exposed to different radiation doses at a variety of dose rates.

 

Irradiation

Fallout

The temporal dependence of dose rate exposure from external fallout is roughly described by the so-called 7–10 rule of thumb⁽⁸⁾, i.e. that for every 7-fold increase in time post detonation, there is about a 10-fold decrease in dose rate (resulting in the dose rate decreasing with time, t, as approximately t^-1.2). Low dose rate irradiations mimicking fallout are performed using a combination of our modified X-RAD 320 delivering dose rates of 1 Gy/min to 6 Gy/d⁽⁹⁾ and the VADER (Variable Dose rate External ¹³⁷Cs irradiator⁽¹⁰⁾) delivering a dose rate of 0.1–1 Gy/d (Figure 1).

Figure 1.

Simulated 7–10 fallout dose profile.

1-mL diluted blood aliquots were placed in a home-built plastic incubator⁽⁹⁾, placed in the X-RAD 320 and irradiated (at 37°C) at a series of decreasing dose rates (Table 1). The X-RAD was operated at 320 kVp with a custom Thoreaus filter (1.25-mm Sn, 0.25-mm Cu + 1.5-mm Al; HVL 4-mm Cu). Dosimetry was performed prior to irradiating each batch of samples using a calibrated 10 × 6–6 ion chamber (Radcal, Monrovia, CA). Samples were then transferred to the VADER and irradiated at 37°C using a continuously decreasing dose rate. The VADER was calibrated annually, using the same ion chamber⁽¹⁰⁾.

Table 1.

X-RAD 320 parameters for modeling the initial component of fallout irradiation.

Step	SSD (cm)	I (mA)	Dose rate (Gy/min)	Time (min)	Dose (Gy)
3 Gy
1	75	12.5	0.33	3	1
2	75	3.82	0.09	6	0.55
3	75	0.95	.025	19	0.46
4	75	0.48	.013	13	.17
5	75	0.1	.004	80	.32
VADER			∝t-1.2	46 h	0.5
8 Gy
1	40	12.5	1	3	3
2	75	7	0.18	8	1.5
3	75	2.3	0.06	13	0.75
4	75	0.9	0.024	20	0.5
5	75	0.45	0.013	60	0.78
6	75	0.1	0.004	140	0.554
VADER			∝t-1.2	43 h 44 m	0.92

During these 48-h-long irradiations, the rest of the aliquots were irradiated and then kept in incubator at 37°C, 5% CO₂.

Ultra high dose rate

High dose rate irradiations were performed using the FLASH irradiator at the Radiological Research Accelerator Facility⁽¹¹⁾, a modified Clinac (Varian medical Systems, Palo Alto, CA) that can deliver electron dose rates between 1 Gy/min and 600 Gy/s.

Aliquots were irradiated using 9-MeV electrons at a dose rate of 1 Gy/min (SSD 171 cm, 1–2 electron pulses per second), 1 Gy/s (SSD 171 cm, 180 pulses per second), ‘600 Gy/sec’ (3 Gy: SSD 20 cm, 2 pulses with a 5.5-ms gap; 8 Gy: SSD 24 cm, three pulses with 5.5-ms gaps) or as a single pulse of 6-MeV electrons. The 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.

Dosimetry was performed using an 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 film dosimetry for higher dose rates. More details are available elsewhere⁽¹¹⁾.

Sample analysis

All samples were kept at 37°C until the ‘fallout’ irradiations were complete. We therefore did not see reason to hold them an additional two hours at 37°C as recommended by IAEA⁽¹²⁾. Upon the completion of irradiations, equal aliquots of the blood (150 μL) were transferred from tubes into the different 96 multiwell plates (for micronucleus and dicentric assays), centrifuged for 2 min at 250 g to replace RPMI1640 with PB-max (Fisher Scientific), and processed using RABiT-II micronucleus⁽⁶⁾ and dicentric⁽⁷⁾ protocols:

CBMN

After 44 h of incubation, a total of 30 μL of Cytochalasin-B (Sigma-Aldrich, St Louis, MO) were added to the cultures at a final concentration of 6 μg/mL. Cells then were cultured for an additional 10 h. The next steps were performed at room temperature. After incubation, the cells were swollen in a hypotonic solution (0.075 M potassium chloride), fixed and washed four times with methanol: acetic acid (3: 1). The fixed cells were transferred to glass-bottom 96-well Matriplates (Brooks Life Sciences, MA) filled with 0.3 mL of methanol: acetic acid (10: 1) and centrifuged at 250 g/2 min. After aspiration of the liquid from wells and complete evaporation of the fixative, 0.2 mL of PBS containing 0.15 μg/mL DAPI (4′, 6-diamidino-2-phenylindole; Thermo Fisher Scientific) was added to the cells for DNA staining.

Dicentric assay

After 44 h of incubation, a total of 30 μL of Colcemide (Gibco, bought through Fisher Scientific) were added to the cultures at a final concentration 0.1 μg/ml. Cells then were cultured for an additional 8 h. At the end of the culture time, the cells were swollen in a hypotonic solution and fixed with 3:1 methanol: acetic acid. The fixed cells were burst in water: acetic acid (1:1 ratio) to extract chromosomes from mitotic nuclei (‘chromosome soup’)⁽⁷⁾. The chromosome solutions were transferred into glass-bottom 96-well plates (Fisher Scientific) and then individual chromosomes were stained using a modified PNA FISH hybridisation protocol that does not require high temperatures and prolonged staining times⁽⁷⁾. The stained chromosomes were counterstained with DAPI.

Imaging

The 96-well glass-bottom plates were imaged using a Cytation 1 Cell Imaging Multi-Mode Reader using a 20× objective (Biotek, Winooski, VT) and analysed using a custom software, FluorQuant v.6.1 (micronucleus assay) and FluorQuantDic v.4 (dicentric assay), written in house in Visual C++ using the OpenCV computer vision libraries (Version 3.1, www.opencv.org). For micronuclei, the software identifies binucleate and mononucleate cells and micronuclei based on size and proximity. For the dicentric assay, the software classifies the chromosomes as monocentric or dicentric based on the number of detected centromere and telomere spots⁽⁷⁾.

Statistical analyses

We imported the data into R 4.2.0 software⁽¹³⁾ to model the dose responses for micronuclei and dicentric chromosome yields, and to statistically assess the potential effects of sex and dose rate on these responses. Samples with < 20 binucleated cells (BN) or < 20 monocentric chromosomes (MC) were excluded from the analysis because these samples generated unreliable yields due to low numbers of cells or chromosomes. The number of retained samples was 1122 (out of 1349) micronucleus samples and 2052 (out of 2092) Dicentric samples. The resulting samples had on average 860 binucleated cells or 870 scored chromosomes (roughly equivalent to 20 metaphases). The data were not normally distributed (as assessed by skewness, kurtosis and Shapiro–Wilk normality test), and, consequently, we used nonparametric statistical tests and quantile regression, as described below.

Since raw micronucleus yield (per binucleated cell) was found to significantly decrease at high radiation doses, we calculated a ‘linearized’ corrected index as follows:

(1)

where Mi is the number of micronuclei in the sample, BN the number of binucleated cells, MN the number of mononucleated cells and k is an adjustable parameter. We used quantile regression (implemented by the quantreg R package) to model the dose response of the median (50th percentile) of Y_MN using a linear quadratic (LQ) function with parameters α and β, and a baseline value c, as follows, where D is the dose (in Gy):

(2)

During this procedure, we varied the parameter k in Eq. (1) so that the β term in Eq. (2) approached zero and became statistically consistent with zero. In other words, parameter k was adjusted such that the median of the resulting Y_MN index would have essentially a linear dose response.

For dicentric chromosomes (Y_DC), we used the standard linear-quadratic dose response function. The median value was fitted using quantile regression, whoich increased with dose over the studied range of doses.

We also fitted quantile regression versions for Y_MN or Y_DC that included potential sex effects. This was done by adding the binary sex variable (0 = male, 1 = female) to the model as an intercept modifier term, and also as interaction terms of sex with D and D². In this manner, we assessed potential sex effects as modifiers of baseline micronuclei or dicentric yields, and as modifiers of radiation dose response parameters.

To assess dose rate effects, we split the data into five dose rate categories: 0 = fallout; 1 = 1 Gy/min; 2 = 1 Gy/s; 3 = 600 Gy/s; 4 = single 5 μs pulse. We compared Y_MN or Y_DC values at different dose rate categories at each radiation dose, using Kruskal–Wallis tests, followed by pairwise Mann–Whitney U tests with Bonferroni correction. To assess sex effects, we used Mann–Whitney U tests to compare values for different sexes (0 = male, 1 = female) at each dose.

Results and discussion

Corrected micronucleus yields

A linear dose response for the corrected micronucleus index, Y_MN was achieved using k = 70 (Figure 2). The resulting β term in Eq. (2) was 1.6 × 10⁻⁴ ± 9.6 × 10⁻⁴ Gy⁻² (P-value of 0.87). Dropping this non-significant β term produced the following parameters (± standard errors) for a linear dependence of the median Y_MN on dose: c = 0.200 ± 0.002, α = 0.095 ± 0.002 Gy⁻¹.

Figure 2.

Fitted linear dose response for the median-corrected micronucleus index (Y_MN).

Dose rate effects

Figure 3 shows the pooled data from all donors, ignoring sex and age.

Figure 3.

(a) Micronucleus and (b) dicentric yields measured to date, using various dose rates. Data pooled across all donors Data in Supplementary Tables 1 and 2.

For Micronuclei, dose rate effects were statistically significant at both doses (Kruskal–Wallis test P-values = 0.0081, 1.4 × 10⁻⁷, respectively) but the dose rate effect magnitude was not large. Some pairwise U-tests with Bonferroni correction were also significant. At both doses, median Y_MN values were highest at 1 and 600 Gy/s and lower at the other dose rates. The decline in yields for fallout dose rates is possibly due to DNA repair of partial damage, where some types of damage (e.g. single strand breaks) are repaired faster than they can accumulate to form more complex damage clusters.

For dicentric yields, Kruskal–Wallis tests for dose rate effects reached significance at 3 Gy (P-value = 0.0027), but not at 8 Gy (P-value = 0.090). Some pairwise U-tests with Bonferroni correction were statistically significant at 3 Gy: between fallout and 1 Gy/min and between fallout and 1 Gy/s. At 3 Gy, median yields were approximately the same at dose rates of 1 Gy/s and higher, slightly lower at 1 Gy/min and paradoxically highest when the dose was delivered over 2 d. At 8 Gy, median yields were highest at dose rates of 1 Gy/s and lower, and lower at others.

The data imply that a standardised calibration curve cannot be used to reconstruct dose without also knowing the dose rate. That being said, we have had success in using an AI-based algorithm to reconstruct dose, without knowing the dose rate, based on a combination of micronucleus and dicentric yields (manuscript under review).

Sex effects

Although not obvious from inspection of the distributions (Figure 4), some differences between males and females were found. Sex effects for micronuclei were not significant (P-values > 0.05) at 0 or 3 Gy, based on Mann–Whitney U tests. Significance (P-value = 2 × 10⁻⁵) was reached at 8 Gy, where females have a higher median value than males (1.08 vs. 0.82).

Figure 4.

Distribution of (a) micronucleus yields and (b) dicentric yields across the samples segregated by sex (shown for a dose rate of 1 Gy/s, similar trends seen for other dose rates). Males are shown on the left, females on the right.

Quantile regressions which included sex-dependent terms suggested that females may have a lower intercept c and higher dose response slope α than males (Table 2). Sex effects on both slope and intercept reached statistical significance at the 25th percentile, and slope only at 50th and 75th percentiles.

Table 2.

Best-fitting parameters for quantile regressions that modeled the medians (50th percentiles) of either the corrected micronucleus index (Y_MN = c + αD + S[c’ + α’D]) or dicentric chromosomes (Y_DC = c + αD + βD² + S[c’ + α’D + β’D²]). Parameter c indicates the baseline yield, α and β indicate linear and quadratic dose response coefficients, in Gy⁻¹ and Gy⁻², respectively. The sex variable S (0 = females, 1 = males) was introduced into the models to assess potential sex effects. c’ therefore represents the modifier for parameter c for males, compared with females. α’ and β’ represent modifiers for the linear and quadratic dose response coefficients, respectively.

Parameter	Best fit	Std. err.	P-value
	Micronuclei
c	0.225	0.033	<10−6
α	0.081/Gy	0.008	<10−6
c’	−0.063	0.034	0.065
α’	0.030/Gy	0.009	6.7 × 10−4
	Dicentrics
c	0.0159	0.0011	<10−6
α	0.0018/Gy	6.0 × 10−4	0.0027
β	7.3 × 10−4/Gy2	8.0 × 10−5	<10−6
c’	−9.1 × 10−4	0.0017	0.60
α’	2.3 × 10−4/Gy	0.0010	0.82
β’	−7.0 × 10−5/Gy2	1.3 × 10−4	0.59

For dicentric yields, the assessment of sex effects by Mann–Whitney U tests did not reach statistical significance at 0, 3 or 8 Gy.

Quantile regressions suggested that sex-dependent terms did not reach statistical significance at any tested percentiles (25th or 75th in addition to the 50th). The results for the 50th percentile are shown in Table 2. Consequently, sex did not appear to significantly modify either the intercepts or the linear or quadratic dose response parameters for dicentrics.

This implies that a single calibration curve can be used to reconstruct dose irrespective of the individual’s sex.

Conclusions

We present preliminary data from a large demographic study we have initiated to investigate dose rate, sex and age effects in micronucleus and dicentric yields in human blood samples irradiated ex vivo. Dose rate effects could be detected for both endpoints, particularly at the high dose rate range, but their interpretation is so far not straightforward.

Sex effects are not detectable for dicentric chromosomes but were detected by some analyses for corrected micronucleus index. An analysis of age effects will be performed when the study is completed.

Up to date raw data (micronucleus and dicentric yields for individual samples) for this study will be uploaded to Immport (import.org)⁽¹⁴⁾ under study accession SDY2079—Sex and Dose Rate Effects in High Throughput Cytogenetics.

Funding

This work was supported by grant number U19-AI067773, and Contract Number HHSN272201600040C, both from the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIAID or NIH.