INTRODUCTION

In a radiological mass-casualty accident, fast and reliable triage identification of individuals exposed to ≥2 Gy acute whole-body equivalent dose of radiation from the worried well is essential for immediate medical treatment, as recommended by the U.S. Department of Health and Human Services Radiation Emergency Medical Management (1). Multiple biomarkers assessed in human peripheral blood such as γH2AX (2, 3), proteins (4, 5), gene expression (6–8) and miRNA (9) have been evaluated for triage as these biomarkers bypass the need for peripheral blood lymphocyte culture to evaluate DNA damage. However, standardized radiological triage guidelines by the International Organization for Standardization (ISO) are currently only available for cytogenetic endpoints scored in cultured peripheral blood lymphocytes. Both dicentric chromosomes and micronuclei (MN) can be used for cytogenetic triage, as seen in ISO 21243 (10) and ISO 17099 (11), respectively.

For triage assessment, 50 metaphases/30 dicentrics are scored for dicentric chromosome assay (DCA) (10), while 200 binucleated cells (BNCs) are scored for cytokinesis-block micronucleus (CBMN) assay (11, 12). In contrast, conventional dose assessment requires at least 500 metaphases/100 dicentrics to be scored for DCA (13), and 1,000 BNCs to be scored for CBMN assay (11) for reliable dose estimation. In manual dicentric scoring for one individual by experienced scorers, conventional triage assessment can take up to 150 min (10) or 30 min with the QUICKScan method (14). However, MN scoring is much quicker than dicentric scoring as the criteria for MN scoring is much simpler than dicentrics and requires no prior knowledge of chromosome aberrations and karyotypes (15). Despite the much longer scoring time, DCA is usually preferred over CBMN assay for dose estimation as DCA is the international “gold standard” for cytogenetic biodosimetry due to dicentrics being highly radiation specific with a low background frequency (13, 16). In contrast, background MN frequency is affected by a variety of factors, including age, sex and lifestyle (17). Furthermore, cell culture for DCA can be completed in 48 h, much faster than the conventional CBMN assay culture of 72 h.

To increase the feasibility of CBMN assay over DCA in radiological triage, previous efforts were made to reduce CBMN assay culture time to 70 h (18), 64 h (18, 19), 60 h (20), 54 h (18, 20, 21), 48 h (20, 22) and 36 h (20). Similarly, technological advancements for high-throughput assessment of MN were also developed for faster scoring and multiple samples. Semi-automated and automated MN scoring systems were developed for Giemsa (23, 24), DAPI (25–27), DAPI/Fast Green (28) and PI (24)-stained cells on microscope slides. For direct imaging with fixed cell solutions, imaging flow cytometry coupled with Rapid Automated Biodosimetry Technology (RABiT) was also developed to handle multiple low volume blood samples (29). The RABiT system can also be used for direct cell culture and fixation, cell imaging and automated MN scoring of cells from multiple donors on glass-bottomed microplates (21). However, these reagents and equipment used are highly specialized, relatively expensive and may not be commonly found in all cytogenetic laboratories. As such, we assessed the feasibility of the low-cost method of manual MN scoring with light microscopy on Giemsa-stained slides in 48 h CBMN cultures for triage in this study.

Furthermore, triage MN scoring is often performed only in whole blood (WB) cultures (12, 18, 19, 21–27, 29). Despite ISO recommending both cultures of WB and peripheral blood mononuclear cells (PBMCs) isolated from WB for CBMN assay as the target cells analyzed are assumed to be phytohemagglutinin (PHA)-stimulated T lymphocytes, plasma and other cellular components present in WB but absent in PBMCs could influence DNA damage induction and repair. As reported in our previous studies (30, 31) and other studies (32, 33), differences in CBMN parameters were seen between WB and PBMC cultures. To the best of our knowledge, this study is the first to compare triage MN scoring for WB and PBMC cultures. In addition, as PBMC separation from WB is preferred for γH2AX (3), a multi-parametric approach for triage assessment with γH2AX and CBMN assay could also be performed with PBMCs directly isolated with density centrifugation.

In this study, we shortened the culture period of CBMN assay from 72 h to 48 h and evaluated various parameters with manual scoring of Giemsa-stained cells as a low-cost alternative. In the first part, cell proliferation indicators of nuclear division index (NDI) and percentage of BNC in all cells (%BNC), conventional and triage MN frequency (MN in 500/1,000 BNCs, MN in 100/200 BNCs) and time taken for triage MN scoring were compared in 0, 2 and 4 Gy WB and PBMC cultures from 3 donors in three conditions, varying in culture period and time of cytochalasin B (Cyt-B) addition [48 h culture (24 h at Cyt-B), 72 h culture (24 h at Cyt-B), 72 h culture (44 h at Cyt-B)]. In the second part, dose-response curves (DRCs) using radiation-induced MN/BNC from another 3 donors were constructed for WB and PBMCs in the shortened [48 h culture (24 h at Cyt-B)] and conventional [72 h culture (44 h at Cyt-B)] CBMN assay. Dose estimation was performed using radiation-induced MN frequencies after triage and conventional scoring from the first part of the study.

MATERIALS AND METHODS

Analysis of CBMN Endpoints (NDI, %BNC, Conventional and Triage MN Scoring)

CBMN endpoint analysis was manually performed by a single experienced scorer. Cells were scored with Olympus CX31 and CX33 (Olympus Co., Tokyo, Japan) at 400× magnification for NDI, %BNC and MN frequency. For NDI and %BNC, a minimum of 125 cells with intact cytoplasmic and nuclear membrane were scored per spot for a total of 500 cells. Equations for NDI (35) and %BNC are provided below. M₁, M₂, M₃, and M₄ indicate the number of cells with one, two, three or four daughter nuclei, respectively, and N is the total number of cells analyzed.

BNC identification and MN scoring were performed in accordance with the standardized criteria compiled by the Human MicroNucleus project (15). In conventional MN scoring for 0 and 2 Gy samples, a minimum of 250 BNCs was scored per spot per slide, for a total of 1,000 BNCs. In conventional MN scoring for 4 Gy samples, a minimum of 125 BNCs was scored per spot per slide, for a total of 500 BNCs. In triage MN scoring for 0 and 2 Gy samples, a minimum of 200 BNCs was scored per spot. In triage MN scoring for 4 Gy samples, a minimum of 100 BNCs was scored per spot. Detailed information on conventional and triage MN scores and time can be found in Supplementary  Tables S1 (rare-199-04-05_s01.docx),  S2 (rare-199-04-05_s02.docx) and  S3 (rare-199-04-05_s03.docx) ( https://doi.org/10.1667/RADE-00191.1.S1;  https://doi.org/10.1667/RADE-00191.1.S2;  https://doi.org/10.1667/RADE-00191.1.S3, respectively).

RESULTS

Cell-Cycle Progression Parameters in 48 h (24 h at Cyt-B), 72 h (24 h at Cyt-B) and 72 h (44 h at Cyt-B)

NDI and %BNC were first compared in the three culture conditions (Fig. 1A). NDI and %BNC were significantly lower in 48 h than 72 h cultures for 0, 2 and 4 Gy cells. In 72 h cultures, length of Cyt-B addition did not greatly affect cell-cycle progression as mostly similar NDI and %BNC were seen for 24 h at Cyt-B and 44 h at Cyt-B. Likewise, as seen in representative images at 0 Gy (Fig. 1B), 48 h cultures showed many mono- and binucleated cells while 72 h cultures showed many multi-nucleated cells.

FIG. 1

Panel A: NDI and percentage of BNC (%BNC) in all culture conditions for 0, 2 and 4 Gy WB and PBMCs. The 500 cells with intact cytoplasmic and nuclear membrane were scored. Average values from duplicate cultures are shown. Cross bars represent Mean ± SD. One-way ANOVA with Tukey's post hoc test was performed. *P < 0.05, **P < 0.01, ***P < 0.0001. Panel B: Representative images of 0 Gy 48 and 72 h WB and PBMC cultures under a light microscope. Scale bars represent 50 µm.

Conventional MN Frequency and Observed Triage MN Distribution in 48 h (24 h at Cyt-B), 72 h (24 h at Cyt-B) and 72 h (44 h at Cyt-B)

In Fig. 2, conventional MN frequencies in both WB and PBMCs were compared in the three culture conditions. Similar, but not significantly different, MN frequencies between WB and PBMCs were seen in 0, 2 and 4 Gy cells for all conditions except for 4 Gy 48 h (24 h at Cyt-B). Variations of MN frequencies in PBMC cultures were wider than WB cultures.

FIG. 2

Conventional MN frequencies of WB and PBMCs cultured in 48/72 h and Cyt-B added at 24/44 h. Average values from duplicate cultures are shown. Cross bars represent Mean ± SD. Unpaired two-tailed t test was performed. **P < 0.01.

Next, Fig. 3 shows the observed triage MN distributions in 2 and 4 Gy WB and PBMC cultures. Triage MN distributions were adjusted to 200 BNCs for 2 Gy and 100 BNCs for 4 Gy samples. As seen from the line graphs, MN distribution patterns were distinct between 2 and 4 Gy cultures regardless of culture type and condition.

FIG. 3

Observed triage MN distributions for 2 and 4 Gy WB and PBMCs in three culture conditions. As shown by the line graph, MN distribution patterns were distinct between 2 and 4 Gy samples regardless of culture type and conditions. MN distributions were adjusted to 200 BNCs for 2 Gy and 100 BNCs for 4 Gy. Bar chart represents the mean while error bars show SD of three donors.

DRCs of Shortened 48 h (24 h at Cyt-B) and Conventional 72 h (44 h at Cyt-B) CBMN Assay

As culture period and length of Cyt-B treatment had some effect on cell-cycle progression and MN frequency, we focused on subsequent comparisons between the shortened 48 h culture (24 h at Cyt-B) and the more frequently used conventional 72 h culture (44 h at Cyt-B). For both WB and PBMCs, DRCs were constructed with pooled radiation-induced MN/BNC to account for the different age and gender of the three donors (Fig. 4). DRC coefficients and SE, and their F test P values were shown in Table 1.

FIG. 4

Pooled DRCs of shortened 48 h (24 h at Cyt-B) and conventional 72 h (44 h at Cyt-B) WB and PBMC cultures with radiation-induced MN/BNC.

TABLE 1

Coefficients and Statistical Results of Pooled CBMN DRCs

Despite differences in DRCs seen in each donor ( Supplementary Fig. S1 (rare-199-04-05_s06.docx),  Table S6 (rare-199-04-05_s06.docx);  https://doi.org/10.1667/RADE-22-00191.1.S6), pooled DRCs were very similar for WB and PBMC cultures of 48 h, while DRCs differed between WB and PBMC cultures of 72 h.

Time Taken for Triage MN Scoring

In Table 2, the average time taken for triage MN scoring for 0, 2 and 4 Gy was shown. For 48 h cultures of both WB and PBMCs, scoring time was within 8 min for 0 Gy, and 20 min for 2 and 4 Gy. For 72 h cultures of both WB and PBMCs, scoring time was within 4 min for 0 Gy, 9 min for 2 Gy and 12 min for 4 Gy.

TABLE 2

Time Taken for Triage MN Scoring in Three Culture Conditions

Dose Estimations Compared with the Number of BNCs Scored and Culture Period

In Fig. 5, both triage and conventional dose estimations were compared based on the number of BNCs scored (i.e., triage MN scoring in 100/200 BNCs vs. conventional MN scoring in 500/1,000 BNCs) and culture period (i.e., shortened 48 h vs. conventional 72 h culture). For triage dose estimates, error bars depicted the minimum and maximum estimated doses if dose estimations were reported based on the first 100/200 BNCs for triage in a realistic emergency triage situation, while the mean was the dose estimated from the averaged MN scores. The number of BNCs scored for MN did not affect dose estimation as doses were within ±0.2 Gy. Estimated doses from averaged MN scores were mostly within ±0.5 Gy of the actual doses and were generally higher in 48 than 72 h cultures.

FIG. 5

Estimated doses from triage and conventional MN scoring of actual doses of 2 and 4 Gy. Bar chart height shows the dose averaged from eight 15 µl spots for triage MN score and dose averaged from duplicate cultures for conventional MN score. Error bar shows the minimum and maximum estimated doses. *Dose estimates with extrapolated MN frequency due to insufficient cells scored. Panel A: 48 h (24 h at Cyt-B) WB and PBMC cultures. Panel B: 72 h (44 h at Cyt-B) WB and PBMC cultures.

DISCUSSION

Triage scoring and dose estimation with cytogenetic markers during a radiation mass-casualty should be efficient and reliable, such that individuals exposed to whole-body equivalent ≥2 Gy can be quickly distinguished from the worried well for immediate medical treatment. The current “gold standard” of DCA requires only 48 h for cell culture, but is limited by the long scoring time and prior knowledge of chromosome karyotypes for scoring. On the other hand, CBMN assay is conventionally cultured for 72 h, but scoring is much faster as it involves visually identifying MN with a “mean diameter within 1/16th to 1/3rd of the main nuclei” (15, 34).

Recent attempts in reducing reporting time for estimated doses involve automated scoring and robotics for high-throughput analysis, but they can be very costly. In our study, we showed that a shortened 48 h CBMN culture is feasible for triage assessment for both WB and PBMC cultures up to 4 Gy. For individuals with low mitogen response to PHA stimulation, more slides can be made for scoring. Acute radiation syndrome symptoms such as skin erythema, nausea, diarrhea and low blood cell counts can also be used simultaneously to identify individuals exposed to higher doses up to 4 Gy (48). Furthermore, dose estimates were within ±0.5 Gy of actual doses and triage MN scoring in 100/200 BNCs was sufficient for accurate dose estimations. Using manual triage scoring with light microscopy, 0 Gy samples can be identified within 8 min while 2 and 4 Gy samples can be scored within 20 min, in slides with appropriate cell concentrations.

The 72 h (44 h at Cyt-B) CBMN culture is the most commonly used (49) as it was shown to obtain an optimal BNC frequency for both WB and PBMC cultures (13, 34), although 68–70 h cultures were also used by multiple laboratories (50). Our results showed that the length of Cyt-B treatment did not significantly affect NDI and percentage of BNC for 72 h cultures. In addition, while our results showed PBMCs with higher NDI and percentage of BNC than WB cultures, the opposite was instead seen in Ellard and Parry's study (32). It is thus important to evaluate the optimal CBMN culture conditions in WB and PBMCs for multiple individuals in each population as differences in individual susceptibility to Cyt-B had been shown even within the same age group (51).

Moreover, in Köksal et al.'s study, reducing CBMN culture time from 72 h to 48 h led to a conclusion that 48 h was too short to obtain desirable BNC frequencies for MN scoring (52). While this is undeniably true as a much lower percentage of BNC was seen in our 48 h cells (0 Gy: 15–32%, 2 Gy: 6–20%, 4 Gy: 2–9%), a sufficient number of BNCs was still able to be obtained for MN scoring of > 10,000 BNCs in 0 Gy and >1,500 BNCs in 4 Gy for a 5 ml culture (500 µl WB) in the three donors used in DRC construction. Likewise, as shown by Rodrigues et al. (22) for 48 h cultures of 2 ml (200 µl WB), an approximate average of 1,400 and 800 BNCs scored in 0 and 4 Gy with imaging flow cytometry, respectively. Even though more time is needed to score MN in cultures showing low percentage of BNC, a shorter CBMN culture is highly applicable for triage as a reduced number of BNCs is scored to account for the lower percentage of BNC. In addition, by personal observation, it was easier to locate BNCs for MN scoring in 48 h cultures as cells were mostly mono- and binucleated.

Some differences in MN frequency between shortened 48 h culture and conventional 72 h cultures were seen in our experiments in Figs. 2 and 4, though no strong conclusions can be drawn due to a low number of donors analyzed. In Rodrigues et al.'s study, a much lower MN frequency in 48 h (24 h at Cyt-B) than 72 h (24 h at Cyt-B) WB cultures in automatically scored cells with imaging flow cytometry (22), which could be due to damaged cells dividing slower than healthy cells (53). In Köksal et al.'s study comparing 48 h (24 h at Cyt-B) and 72 h (44 h at Cyt-B) CBMN assay, a lower MN frequency was also seen in 48 than 72 h WB cultures in manually scored cells (52). This finding was also supported other studies by Lee et al. (20) and Almássy et al. (54), where an increasing MN frequency was seen in elongated culture periods. However, differences in donor population could have also contributed to MN frequency variability (55). Nevertheless, a 48 h CBMN culture period was able to easily distinguish irradiated from unirradiated samples, thus supporting the use of a shortened CBMN assay for radiation triage as results are able to be reported at a much quicker time. It also reinforces the recommendation that the same culture protocols should be used for CBMN DRC construction and dose estimation.

Likewise, Cyt-B addition at 24 h or 44 h for conventional 72 h cultures could affect MN frequency and estimated doses. Our observation of a slightly higher MN frequency in 72 h (24 h at Cyt-B) than 72 h (44 h at Cyt-B) culture was also seen by Lee et al. (20) and Köksal et al. (52). When Cyt-B was added at 24 h in 72 h cultures, MN were likely scored in BNCs arrested after the first cell division. When Cyt-B was added at 44 h instead, MN were likely scored in BNCs arrested after the first cell division, if mitosis was delayed, and after the second cell division. As MN contains unstable aberrations such as acentric fragments or whole chromosomes (34), MN frequency could decrease with multiple cell divisions. As stated in EPR-Biodosimetry, Cyt-B should be preferably added at 24 h for “biological dosimetry to ensure only first division cells are captured” (13). As supported in Duthoo et al.'s study (53) of PHA-stimulated non-irradiated and irradiated T cell cycles, the estimated total cell division time was 16.6 h in non-irradiated cells. First-cycle divisions were completed in 48–72 h and unaffected by irradiation of up to 2 Gy after PHA stimulation. Furthermore, prolonged G2/M arrest was observed with increasing dose up to 4 Gy. The length of Cyt-B treatment is thus an important factor affecting MN frequency and hence, consistency in protocols for CBMN DRC construction and dose estimation is important.

Furthermore, to the best of our knowledge, this is the first time CBMN DRCs were constructed for the same donors comparing WB and PBMC cultures of 48 h (24 h at Cyt-B) and 72 h (44 h at Cyt-B). The 48 h cultures showed similar DRCs for WB and PBMCs, while a higher DRC was seen in 72 h cultures of WB than PBMCs. Similarly seen in Rodrigues et al.'s study, a much higher DRC was seen in 72 h than 48 h WB cultures for Cyt-B added at 24 h (22). Likewise, Sioen et al. also showed a higher DRC in 70 h (23 h at Cyt-B) WB than fresh PBMC cultures of up to 2 Gy (53). On the contrary, in Lue et al.'s study comparing 70 h (44 h at Cyt-B) 50 µl WB and 500 µl PBMC cultures, DRCs in both types of cultures were very similar up to 6 Gy (56). Further study is required to better understand the differences in DNA repair mechanics and MN induction after irradiation in WB and PBMCs. Once again, population differences could have influenced the dose-response relationships observed, and hence CBMN assays of multiple conditions should be separately evaluated in each laboratory. Despite the differences in DRCs seen, our dose estimations of 48 and 72 h WB and PBMC cultures were mostly within ±0.5 Gy of actual doses.

MN scoring in 200 BNCs was shown by McNamee et al. to reliably identify individuals exposed to ≥1 Gy gamma irradiation with manual scoring on acridine orange-stained cells (12), and thus 200 BNCs is recommended by ISO 17099 (11) for initial triage assessment. For higher doses, 200 total MN can also be scored (12). We also showed that 100/200 BNCs were sufficient for triage dose estimates as doses were within ±0.2 Gy as compared to conventional MN scoring. Furthermore, for higher doses, 100 BNCs could be scored instead of 200 BNCs. As our paper's focus was on manual MN scoring with a multiple unit mechanical tally counter, the direct data output was the number of cells with 0 to 6 MN (i.e., MN distribution). However, additional calculations were required to obtain total cells scored and total MN. To quicken triage identification in different dose categories, we have shown that observed triage MN distribution can be directly used to distinguish 2 and 4 Gy WB and PBMC cultures for both 48 and 72 h cultures.

However, depending on the 15 µl spot analyzed by the same scorer, triage MN of 48 h cultures in 8 spots scored showed differences of up to 60 MN in 200 BNCs (CV: 7–17%, ±0.8 Gy) in 2 Gy and 50 MN in 100 BNCs (CV: 5–12%, ±0.8 Gy) in 4 Gy. In an international inter-laboratory MN scoring exercise (57), the median intra-scorer CV of 2 spots for 1 and 2 Gy for 1,000 BNCs was 14 and 11%, respectively. Our previous study also showed differences of up to 60 MN in 1,000 BNCs and inter-scorer CV of 1–10% in 2 Gy (31). As manual MN scoring is highly dependent on visual interpretation, doses estimated could result in the false-negative rejection for the triage threshold dose of 2 Gy. Some donors also showed low percentage of BNC as insufficient BNCs were scored for triage analysis on 2 and 4 Gy PBMC for a 15 µl spot after 48 h CBMN culture. As such, if irradiated individuals were identified with 48 h CBMN triage, we highly recommend multiple scorers to score the same 2-spot slide(s). If the CV is within 20%, the average MN frequency can be used for more reliable dose estimation.

Our study, however, has several limitations. First, only one scorer was used in this study as many slides needed to be analyzed for two experiments and six donors. Second, we did not compare the time taken from culture to dose estimate report production between manual and automated MN scoring. Third, duplicate cultures were not performed in DRC construction. As 30 ml peripheral blood from each donor was required for culture set-up of 4 conditions, duplicate cultures were not feasible due to lack of manpower and blood sample volume restrictions according to ethical guidelines in our study. Fourth, only three donors of similar age were used for DRCs. According to the ISO standard, at least three donors in different age groups and gender should be used for CBMN DRCs (11). Fifth, this study focused on the ideal radiation scenario of a homogenous whole-body irradiation. Accidental heterogeneous high dose exposures account for more than 95% of the cases (58). As this study serves as an initial evaluation on the feasibility of manual MN scoring in 48 h cultures for triage, the above limitations will be addressed in the next study. Lastly, individual donor variations were seen in both parts of the study and could affect dose estimation reliability. It was to be expected as MN is not as radiation specific as dicentrics and many factors such as age, gender and smoking history could have contributed to the variations. Increased sensitivity, especially in the lower doses, could be achieved by combining MN scoring with centromere staining to distinguish MN containing acentric fragments or whole chromosomes (59). However, at the higher doses of 0.75–2 Gy, similar dose estimates and their confidence limits were obtained in MN frequencies with or without centromere staining (60). From the triage point-of-view, conventional MN scoring is sufficient for high dose identification, though estimated dose differences caused by donor or scorer variation prove difficult to resolve.

In conclusion, our study showed that with a low-cost method of manual MN scoring on Giemsa-stained cells and shortened CBMN assay to 48 h (24 h at Cyt-B) for both WB and PBMCs, radiation triage identification of individuals exposed to a whole-body equivalent dose of ≥2 Gy can be easily performed. In our donor population, triage MN scoring was completed in 8 min for 0 Gy and in 20 min for 2 and 4 Gy samples. Discrimination between 2 or 4 Gy exposure was feasible based on observed MN distribution patterns. Triage dose estimates were also mostly within ±0.5 Gy of actual doses. We thus highly recommend the 48 h CBMN assay for initial triage in a radiation mass-casualty accident. If more reliable dose estimates are required, dose estimation with 1,000 BNCs in CBMN assay or 500 metaphases in DCA can be additionally performed on selected individuals.

SUPPLEMENTARY MATERIALS

 Supplementary Table S1 (rare-199-04-05_s01.docx). Coefficient of variance (CV) for conventional MN frequency in duplicate cultures (Expt 1)

 Supplementary Table S2 (rare-199-04-05_s02.docx). Scoring time (min:s) in (a) 48 h (24 h at Cyt-B) WB and (b) PBMC cultures; (c) 72 h (24 h at Cyt-B) WB and (d) PBMC cultures; (e) 72 h (44 h at Cyt-B) WB and (f) PBMC cultures.

 Supplementary Table S3 (rare-199-04-05_s03.docx). Observed triage MN frequency in (a) 48 h (24 h at Cyt-B) WB and (b) PBMC cultures; (c) 72 h (24 h at Cyt-B) WB and (d) PBMC cultures; (e) 72 h (44 h at Cyt-B) WB and (f) PBMC cultures.

 Supplementary Table S4 (rare-199-04-05_s04.docx). Raw data and Poisson test results for DRC construction for (a) 48 h (24 h at Cyt-B) WB and (b) PBMC cultures; (c) 72 h (44 h at Cyt-B) WB and (d) PBMC cultures.

 Supplementary Table S5 (rare-199-04-05_s05.docx). Raw data of (a) NDI and (b) percentage of BNC in duplicate cultures (Expt 1)

 Supplementary Fig. S1 (rare-199-04-05_s06.docx). CBMN DRCs of (a) each donor and (b) each culture condition.

 Supplementary Table S6 (rare-199-04-05_s06.docx). Coefficients and statistical results of each donor's CBMN DRCs in 4 culture conditions.