DEVELOPMENT OF A MINIATURIZED VERSION OF DICENTRIC CHROMOSOME ASSAY TOOL FOR RADIOLOGICAL TRIAGE

Abstract

Use of ionizing radiation (IR) in various industrial, medical and other applications can potentially increase the risk of medical, occupational or accidental human exposure. Additionally, in the event of a radiological or nuclear (R/N) incident, several tens of hundreds and thousands of people are likely to be exposed to IR. IR causes serious health effects including mortality from acute radiation syndrome and therefore it is imperative to determine the absorbed radiation dose, which will enable physicians in making an appropriate clinical ‘life-saving’ decision. The ‘Dicentric Chromosome Assay (DCA)’ is the gold standard for estimating the absorbed radiation dose but its performance is time consuming and laborious. Further, timely evaluation of dicentric chromosomes (DCs) for dose estimation in a large number of samples provides a bottleneck because of a limited number of trained personnel and a prolonged time for manual analysis. To circumvent some of these technical issues, we developed and optimized a miniaturized high throughput version of DCA (mini-DCA) in a 96-microtube matrix with bar-coded 1.4 ml tubes to enable the processing of a large number of samples. To increase the speed of DC analysis for radiation dose estimation, a semi-automated scoring was optimized using the Metafer DCScore algorithm. The accuracy of mini-DCA in dose estimation was verified and validated though comparison with conventional DCA performed in 15 ml conical tubes. The mini-DCA considerably reduced the sample processing time by a factor of 4 when compared to the conventional DCA. Further, the radiation doses estimated by mini-DCA using the triage mode of scoring (50 cells or 30 DCs) were similar to that of conventional DCA using 300–500 cells. The mini-DCA coupled with semi-automated DC scoring not only reduced the sample processing and analysis times by a factor of 4 but also enabled the processing of a large number of samples at once. Our mini-DCA method, once automated for high throughput robotic platforms, will be an effective radiological triage tool for mass casualty incidents.

INTRODUCTION

Human exposure to ionizing radiation (IR) has become inevitable owing to its wide spread use in industrial, medical, research and other applications. Additionally, exposure can also result during any radiological or nuclear incident, which may involve several tens of hundreds and thousands of people. IR induced health effects may vary from mild to severe depending on radiation dose and mortality can result if severely exposed people (>3–5 Gy) do not get appropriate medical treatment in a timely manner. IR also has a stochastic risk for cancer resulting in increased deaths. Accurate personalized dose assessment is therefore critical for guiding the clinical decision making process. The dicentric chromosome assay (DCA) has been effectively used for radiation dose estimation since Bender and Gooch first described it in 1962⁽¹⁾. They utilized this technique to assess radiation dose in the exposed individuals after the Recuplex criticality accident in Hanford, WA, USA. Since then, DCA has been the method of choice for radiation dose estimation and its utility as a biodosimeter was proven many times in estimating the absorbed radiation doses in humans exposed during several large-scale radiological and nuclear incidents/accidents including Chernobyl, Goiania and Tokaimura, to name a few. DCA is considered to be the ‘gold standard’ for dose assessment owing to a number of unique features: (1) an extremely low base line frequency of 1–2 dicentric chromosomes (DCs) per 1000 metaphase cells, (2) base line frequency is not largely influenced by life style (cigarette smoking, alcohol and caffeine intake), (3) fairly specific to IR with a clear radiation dose dependency and (4) fairly long radiation dose dependent persistence of DCs (up to 4 weeks after exposure but can be extended to ~6 months using a correction factor for the half-life of T-lymphocytes) after IR exposure. One of the distinct advantages of DCA over other ‘omics’ biodosimeters (transcriptomics, proteomics and metabolomics) is the extremely low base line frequency with very little dependence on inter-individual variation.

Performance of DCA, however, is time consuming and laborious with an estimated turnaround time of 72–96 h. Further, timely evaluation of DCs for dose estimation in a large number of individuals provides a bottleneck due to a restricted number of trained personnel and a prolonged analysis time. To make DCA suitable for radiological triage scenarios, efforts have been taken by several laboratories to automate some of the procedural steps in DCA: chromosome preparation⁽²⁾, automated scoring^(3–9) and sample tracking⁽¹⁰⁾. Automated platforms are also commercially available (HANABI, ADS Biotec, USA) which can perform the steps of cell harvesting and metaphase chromosome preparation with minimal human intervention. However, currently available systems can process only 64 samples at a given time and the steps leading to metaphase chromosome slide preparations take ~5–6 h. In order to effectively use DCA for radiation mass casualty incidents, improvements are necessary for large-scale sample processing and analysis so that timely assessment of individualized dose is accomplished. The conventional DCA utilizes 15 ml conical tubes and processing a large number of these tubes is cumbersome, costly and time consuming as large volumes of reagents are added manually after each centrifugation step during sample fixation. To enable high throughput sample processing, we developed a miniaturized DCA (mini-DCA) technique that utilizes only 50–100 μl of whole blood samples collected by a finger stick. Samples are cultured in a 96-microtube matrix with bar-coded 1.4 ml tubes. Since all the chromosome preparatory steps are carried out using multichannel pipettes, the overall sample processing time is reduced by a factor of more than 4 when compared to the conventional DCA. To streamline the high throughput sample processing with rapid image analysis, we also optimized the semi-automated Metafer DCScore algorithm. We found that a quick manual verification of false positives or negatives considerably reduced the analysis time by a factor of 5 when compared to complete manual scoring. The mini-DCA was verified and validated through comparison with conventional DCA performed in 15 ml conical tubes. The radiation doses estimated by mini-DCA using the triage mode of scoring (50 cells or 30 DCs) were similar (±0.1–0.3 Gy) to doses estimated by conventional DCA using 300–500 cells. Our results suggest that the mini-DCA coupled with automated DC scoring can be an effective radiological triage tool for radiation mass casualty incidents.

MATERIALS AND METHODS

Sample collection and irradiation

Blood samples (~10 ml) were collected from three healthy human donors (35–55 years of age) and the blood collection was performed with the consent of donors in strict compliance with the Institutional Review Board (IRB) protocol (ORAU 000 349). Aliquots (1 ml) of samples were irradiated with different doses of X-rays (0, 0.25, 0.5, 1, 2, 3, 4 and 5 Gy) using the X-ray irradiator facility of the University of Tennessee Knoxville (UTK) at a dose rate of 2 Gy/min.

Conventional DCA

Conventional DCA was performed in 15 ml conical tubes (Falcon). For conventional DCA, 0.5 ml of mock and irradiated samples were mixed with 9.5 ml complete lymphocyte growth medium (PBMAX, GIBCO) containing fetal bovine serum, Phytohaemagglutinin, growth factors and antibiotics. To promote an optimal growth, an additional 1 ml of PHA was added for every 100 ml of PBMAX for both conventional DCA and mini-DCA. Bromodeoxyuridine (10 μM) was added to the cultures to identify the first division metaphases. The cultures were incubated at 37°C in a 5% CO₂ incubator for 48 h. Colcemid (0.1 μg/ml) was added for the last 4 h and the cultures were harvested using a standard procedure. Cells were treated with a hypotonic solution (0.56% KCl) for 18 min at 37°C and fixed in three changes of acetic acid:methanol (1:3) mixture. An aliquot of fixed cell suspension (30–40 μl) was placed at the center of acid cleaned slides. The slides were stained using 5% Giemsa for metaphase image capture and analysis.

Miniaturized DCA (mini-DCA)

For mini-DCA, a 96-microtube matrix (Thermo Scientific) containing bar-coded 1.4 ml tubes was used. Multichannel pipettes were used for all the liquid handling steps involved in culture set up, cell harvesting and fixation. For each mock or irradiated sample, triplicate cultures were set up. Cultures were set up in mini tubes by placing 100 μl of whole blood sample and 900 μl of complete growth medium. After mixing thoroughly, the cultures were incubated at 37°C in a 5% CO₂ incubator for 48 h. At the completion of 44 h, 10 μl of colcemid (0.1 μg/ml) was added for the last 4 h and the cultures were harvested and fixed as described before. Briefly, the micro-plates containing the 1.4 ml tubes were spun at 1200 rpm for 2 min and the supernatant (~900 μl) was carefully aspirated from the tubes using a multichannel pipette. For hypotonic treatment, 900 μl of freshly made 0.56% KCl was added to each tube and the tubes were incubated for 18 min at 37°C. After hypotonic treatment, the micro-plates were spun and the cells were subjected to fixation using three changes of acetic acid:methanol (1:3) mixture. The fixed cell suspension (25–30 μl) was dropped onto acid cleaned slides and the slides were subjected to fluorescence plus Giemsa (FPG) staining technique. Briefly, slides covered with a few drops of Hoechst 33258 were exposed to short wavelength ultraviolet light for 30 min followed by washing in Sorenson Buffer (pH 6.8) and incubation at 60°C for 20 min in 2XSSC. After staining with Giemsa (5% in a buffered solution), the slides were air-dried and subsequently mounted with coverslips using DPX.

Image capture and analysis

Metaphase chromosomes were detected using ×10 objective lens using the metaphase finder algorithm of Metafer and the images were captured with an immersion ×63 oil lens objective. The captured metaphase images were subjected to DCScore algorithm of Metafer for the detection of DC frequencies. A quick manual correction, if needed, was performed for removing the false positives and negatives before estimating the absorbed radiation dose using the Chromosome ABerration cAlculation Software (CABAS). A standard Co-60 calibration curve generated in our laboratory was used to compare the physical dose with the biodose estimated by both mini-DCA and conventional DCA. The fit coefficients for the X-rays calibration curve are C = 3.6 × 10⁻³, α = 6.0E × 10⁻² and β = 4.3 × 10⁻².

RESULTS

Comparison of quality and performance of mini-DCA with conventional DCA by Giemsa staining method

The mini-DCA was optimized using a microtube matrix containing 96 bar-coded tubes of 1.4 ml capacity. All the liquid handling steps were carried out using the multichannel pipettes (Figure 1). We estimated that each mini culture performed in a 1.4 ml tube containing 100 μl of whole blood sample yielded an average of 100–200 metaphase cells/slide, which was sufficient for a triage mode of scoring. Typically, 2–3 slides were prepared from each tube yielding ~400–600 metaphases per culture. To verify and validate the performance of mini-DCA, conventional DCA was performed on the same blood samples. Initially, blood samples collected from three healthy donors were irradiated with different doses of X-rays (0, 1, 3 and 5 Gy) and aliquots of samples were processed for both conventional (15 ml conical tubes) and mini-DCA (1.4 ml tubes in a 96-microtube matrix) methods. The main purpose for developing mini-DCA is to increase high throughput sample processing for radiological triage and therefore a triage mode of scoring (50 cells or 30 dicentrics) was employed for dose evaluation by mini-DCA while a total of 300–500 metaphase cells were analyzed for dose assessment by conventional DCA. The samples processed by conventional DCA and mini-DCA were evaluated by both manual and semi-automated scoring respectively using Metafer. Since semi-automated DC scoring is rapid and often preferred for triage scenarios, the accuracy of semi-automated scoring was compared with manual scoring. Representative pictures of metaphase chromosomes prepared from 3 Gy of X-rays irradiated lymphocytes by both conventional DCA and mini-DCA are shown in Figure 2. Heavily damaged cells were chosen to demonstrate the accuracy of automated detection of DCs by DCScore algorithm. Red rectangle boxes in both metaphase cells indicate the DCs that were detected by automated Metafer DCScore algorithm. The results of comparative analysis of conventional and mini-DCA are shown in Table 1 and Figure 3. In all the samples (with the exception of a single dose point of 3 Gy in donor 1), the biodose estimated by analysis of either 50 metaphases or 30 DCs in the mini-DCA was similar (±0.1–0.3 Gy) to the biodose estimated by the conventional analysis of 300–500 cells. Further, radiation doses estimated by both mini-DCA and conventional DCA were also very similar to the physical dose.

Table 1.

Comparison of dicentric yields and dose estimation between conventional DCA (C-DCA) and mini-DCA (M-DCA).

	Phys. dose (Gy)	Cells scored	No. Dic.	Dic./Cell	Estd. dose (Gy)	LCL 95% CI (Gy)	UCL 95% CI (Gy)
C-DCA (Donor 1)	0	500	0	0	0	0	0.15
	1	500	61	0.12	1.08	0.91	1.27
	3	500	399	0.79	3.26	2.86	3.68
	5	350	656	1.87	5.20	4.80	5.61
M-DCA (Donor 1)	0	50	0	0	0	0	0.80
	1	50	9	0.18	1.38	0.83	2.03
	3	29	31	0.62	3.84	3.11	4.65
	5	17	30	1.76	5.04	4.08	6.09
C-DCA (Donor 2)	0	500	2	0.004	0.008	0.19
	1	500	69	0.14	1.17	0.99	1.36
	3	500	397	0.79	3.26	2.86	3.68
	5	377	684	1.81	5.11	4.71	5.52
M-DCA (Donor 2)	0	50	0	0	0	0	0.80
	1	50	9	0.18	1.38	0.83	2.03
	3	35	30	0.86	3.41	2.74	4.14
	5	16	30	1.9	5.20	4.21	6.29
C-DCA (Donor 3)	0	500	0	0	0	0	0.15
	1	500	74	0.15	1.22	1.05	1.41
	3	500	375	0.75	3.16	2.77	3.58
	5	320	570	1.78	5.06	4.66	5.47
M-DCA (Donor 3)	0	50	0	0	0	0	0.80
	1	50	6	0.12	1.07	0.53	1.73
	3	40	30	0.75	3.16	2.54	3.85
	5	15	30	2	5.39	4.36	6.50

Figure 1.

Development of miniaturized DCA for high throughput sample processing. (A) Lids for the 96-microtube matrix, (B) 96-microtube matrix, (C) multichannel pipette and (D) sample reservoir for carrying out liquid handling steps involving culture set up, harvesting and chromosome preparation steps. (E) Barcoded 2D matrix tube (1.4 ml capacity). Comparison of sample processing time for mini-DCA and conventional DCA is shown.

Figure 2.

Representative images of metaphase chromosomes prepared from 3 Gy X-rays irradiated human lymphocytes by both conventional DCA (C-DCA) and miniaturized DCA (M-DCA). The rectangle boxes shown in red color indicate the dicentric chromosomes that were detected by Metafer automated DCScore algorithm. Heavily damaged cells were chosen to demonstrate the efficiency of DCs detection by Metafer.

Figure 3.

Comparative analysis of radiation dose estimation by mini-DCA and conventional DCA in ex vivo irradiated samples of three donors. Note that the radiation doses estimated by miniaturized DCA (M-DCA) are highly similar to conventional DCA (C-DCA).

Testing the efficiency of mini-DCA for radiation doses at ≥0.25 Gy

We next performed mini-DCA and conventional DCA on samples that were irradiated with relatively low doses of X-rays (0, 0.25, 0.5 and 1 Gy). Results of biodose estimation relative to physical dose by both methods are shown in Table 2. Metaphase cells obtained by both mini-DCA and conventional DCA were analyzed by semi-automated method using the DCScore algorithm of Metafer after removing the false positives and negatives. The biodose detected by mini-DCA was found to be 0, 0.33, 0.39 and 0.92 Gy for the physical doses of 0, 0.25, 0.5 and 1 Gy. The biodose detected by conventional DCA was 0, 0.33, 0.59 and 1.21 Gy for the corresponding physical doses of 0, 0.25, 0.5 and 1 Gy. With both methods, the estimated radiation doses deviated by 0.1–0.2 Gy from the physical doses. These variations are well within the specified limit of ±0.5 Gy by IAEA for the DCA based radiation dose estimation.

Table 2.

Comparison of dicentric yields and dose estimation between conventional DCA (C-DCA) and mini-DCA (M-DCA).

	Phys. dose (Gy)	Cells scored	No. Dic.	Dic./Cell	Estd. dose (Gy)	LCL 95% CI (Gy)	UCL 95% CI (Gy)
C-DCA (Donor 3)	0	537	0	0	0	0	0.12
	0.25	1393	35	0.025	0.33	0.24	0.44
	0.5	1221	62	0.05	0.59	0.48	0.71
	1	397	58	0.14	1.21	1.02	1.42
M-DCA (Donor 3)	0	326	0	0	0	0	0.20
	0.25	358	9	0.02	0.33	0.15	0.56
	0.5	427	13	0.03	0.39	0.22	0.60
	1	189	18	0.095	0.92	0.64	1.24

Semi-automated dicentric scoring is efficient and time saving

For the current study, we utilized the DCScore algorithm of Metafer for a rapid detection of DCs in the samples processed by mini-DCA and conventional DCA.

We found that the chromosome quality is a key determining factor for the success of automated or semi-automated scoring. When the quality of metaphases was good without any overlapping chromosomes and staining debris, the automated detection of DCs was highly accurate (>92–95% of manual scoring). We estimated that the average time taken by Metafer for the capture (8.4 s) and analysis (0.12 s) of a single metaphase cell at ×63 objective was 8.5 ± 0.5 s relative to 35–45 s for manual scoring mode (depending on the number of DCs in a metaphase cell), reducing the image capture and analysis time dramatically by a factor 4–5. Manual verification of automated scoring (semi-automated) was still found to be faster than complete manual scoring by a factor of 2. We estimated that by removing false positives (detection of overlapping chromosomes or touching chromosomes as dicentrics) and false negatives (lack of detection of some true DCs by the Metafer DCScore algorithm), the accuracy of semi-automated scoring could be substantially increased to more than 95%. During the course of analysis, we found that Metafer did not efficiently detect all the tricentric or tetracentric chromosomes that are induced at relatively high frequencies following exposure to radiation doses exceeding 3 Gy. Further improvements and refinements in the DCScore algorithm are needed for enabling a complete automation of DCs detection.

DISCUSSION

DCA is considered to be the ‘gold standard’ for absorbed radiation dose assessment in humans after accidental, incidental and occupational exposure. The conventional DCA on an average takes ~3–4 days for absorbed radiation dose estimation by any cytogenetic biodosimetry laboratory. DCA is relatively easy to perform when a few radiation overexposure cases are to be analyzed but its application to radiological triage is largely constrained by a number of factors: (i) fairly long sample processing and analysis time, (ii) restricted number of cytogenetic biodosimetry laboratories and (iii) limited number of trained radiation cytogenetic personnel. Therefore, efforts and improvements are continually made to facilitate a large-scale sample processing and to reduce the turnaround time of DCA. Using an automated scanning Metafer platform, Schunk et al.⁽¹¹⁾ demonstrated the feasibility of automated cytogenetic imaging for metaphase chromosomes, micronuclei and DNA damage in cells after the comet assay. To facilitate the rapidity of DC analysis, various laboratories utilized the semi-automated DC scoring using Metafer DCScore algorithm^{(3, 12)}. Independently, Rogan et al.⁽¹³⁾ developed a prototype software system with sufficient sensitivity and capacity to estimate radiation dose estimates in a mass casualty event. To match up with the advancements made for automated DC analysis, technical improvements are certainly needed for developing automated high throughput platforms for large-scale sample processing and chromosome preparation steps, which will be useful for any radiation mass casualty incident.

In this study, we describe mini-DCA for high throughput sample processing to reduce the turnaround time for radiation dose estimation. The miniaturized DCA has several distinct advantages over conventional DCA: (i) enables the processing of a large number of samples, (ii) utilizes only 100 μl of whole blood sample for analysis, (iii) cost effective and 100 samples can be processed in place of 20 samples using the conventional DCA, (iv) sample culturing and processing times are reduced by a factor of more than 4 when compared to conventional DCA and (v) highly suitable for precise dose estimation using triage mode of scoring. The radiation doses estimated by mini-DCA (performed in a 96-microtube matrix) over a wide range of physical doses (0–5 Gy) correlated well with the dose estimates of conventional DCA. For the current triage mode of inter-comparison studies on DCA, 20–50 metaphase cells were effectively used for radiation dose estimation^{(8, 14)}. In the present study, the triage mode of scoring (50 metaphase cells or 30 DCs) employed for mini-DCA yielded dose estimates, which were grossly similar to that of conventional DCA. These observations suggest that triage-scoring mode can be effectively utilized for mini-DCA without compromising the accuracy of dose estimation.

Several studies in the past have utilized the triage mode of DC scoring for various simulating conditions of whole body and partial body exposures and for mass casualty incidents^{(3, 5, 8, 12, 15–22)}. Di Giorgio et al.⁽¹⁶⁾ initiated an inter-comparison study where the participants of 14 laboratories were asked to perform the dose estimates from the slides of samples irradiated with two doses (0.75 and 2.5 Gy) by both triage (50 or 100) and conventional (500–1000) scoring modes. Although there were some discrepancies among the laboratories, the overall estimated doses by most laboratories fell within ±0.5 Gy. In an earlier study, Romm et al.⁽²⁰⁾ tested the concept of triage scoring of DCs and found that scoring 50 metaphases in contrast to 500–1000 cells of conventional DCA under simulated condition was sufficiently sensitive to guide and diagnose acute radiation syndrome (ARS). Flegal et al.^{(17, 23)} utilized a new scoring technique termed as ‘DCA QuickScan’ and found that scoring 1000 cells or 200 DCs provided statistically significant results comparable to that of conventional DC scoring. In another inter-comparison study, Beinke et al.⁽⁸⁾ found no statistically significant differences for the dose estimates based on 20, 30, 40 or 50 cells and suggested the use of 20–50 cells without any loss of precision of radiological triage dose estimates (±0.5 Gy). Utilizing the mini-DCA in this study, we demonstrated that the triage scoring was effective even at doses below 0.5 Gy and the estimated biodoses were within a narrow margin of ±0.1–0.3 Gy with the physical doses.

An earlier study⁽²⁴⁾ described the development of a miniaturized assay for the detection of radiation-induced micronuclei. In this study, we developed and demonstrated the utility of mini-DCA for precise dose estimation using a triage mode of DC scoring. Our improved mini-DCA version can easily support high throughput processing of several hundreds of samples by reducing the turnaround time for personalized dose estimate following a radiological or nuclear mass casualty incident. Our mini-DCA method, once automated for high throughput robotic platforms, will be an effective radiological triage tool for mass casualty incidents.

FUNDING

The financial assistance received from the US Department of Energy (#DE-AC05–06OR23100 and Technology Integration Grant ORISE-17-CM 989) from NA-84 is gratefully acknowledged. REAC/TS, an organizational program of the Oak Ridge Institute for Science and Education, is operated by Oak Ridge Associated Universities for the US DOE. The content is solely the responsibility of the authors and does not reflect the official views or opinions of either the US DOE or ORAU.