Introduction
When a radiological event resulting from an accident or deliberate malicious activity is classified as a large-scale event, rapid medical and radiological triage or classification of the affected individuals is required.Reference Waselenko, MacVittie and Blakely 1 –Reference Wei and Dewji 3 In this context, biological dosimetry is a tool that can estimate the absorbed dose in individuals suspected of exposure to ionizing radiation, 4 typically applied through cytogenetic techniques to detect chromosome damage, including dicentric chromosome analysis (DCA) and cytokinesis-block micronucleus assay (CBMN). Detailed information on these assays can be found in the International Atomic Energy Agency technical document, 4 and in several ISO standards. 5 , 6 The results of biological dosimetry provide the medical team with reliable data to determine the possible clinical evolution of the affected individuals and deliver the necessary medical care.Reference Swartz, Williams and Flood 7 , Reference Swartz, Flood and Singh 8
In a radiological incident of this type, the potential number of people who may need to undergo this type of analysis could easily exceed the capacity of one or several laboratories. In this regard, biological dosimetry networks have been recognized as an important emergency response strategy. In recent years, various international networks for biological dosimetry have been established in different world regions,Reference Kulka, Wojcik and Di Giorgio 9 such as Running the European Network of Biological and Retrospective Physical Dosimetry, RENEB,Reference Kulka, Ainsbury and Atkinson 10 the Asian network for radiation dosimetry, ARADOS,Reference Kurihara, Ha and Qinjian 11 the Latin America Biological Dosimetry Network, LBDNet,Reference García, Di Giorgio and Radl 12 and the network from Canada and the United States of America, the North American BD network.Reference Wilkins, Beaton-Green and Lachapelle 13 Besides these regional networks, global networks have been set up by the WHO (BioDoseNet), the IAEA (within RANET), EURADOS, and the Global Health Security Initiative (GHSI).Reference Kulka, Wojcik and Di Giorgio 9 In addition to these networks, countries with several laboratories capable of analyzing dose biomarkers have also established their own networks to provide a response to a radiological emergency at a national scale.Reference Yoshida, Hayata and Tateno 14 –Reference Lee, Yang and Yoon 18
Although these networks constitute a robust scientific framework for the development and implementation of procedures aimed at improving radiological emergency response, one of the major challenges remains the provision of a harmonized and high-quality output. Addressing this challenge requires continuous validation, which is achieved through systematic interlaboratory comparisons (ILCs).Reference Di Giorgio, Barquinero and Vallerga 19 –Reference Bucher, Endesfelder and Pojtinger 26 The feasibility of simulating ex vivo scenarios relies on the fact that shortly after an in vivo exposure, cytogenetic assays are performed on mature lymphocytes circulating in peripheral blood. Because these cells continuously recirculate and mix throughout the vascular system, an ex vivo blending of irradiated and non-irradiated blood accurately replicates the heterogeneous lymphocyte populations observed in real-world radiological accidents and validated experimental models.Reference Barquinero, Barrios and Caballín 27 – 31
Several Spanish laboratories participate in some of the above-mentioned international biological dosimetry networks; however, other laboratories with demonstrated capability to perform biological dosimetry analyses are not currently affiliated with any network. In this context, the Spanish Nuclear Safety Council (CSN) launched a project to harmonize laboratories and establish a national biological dosimetry protocol. This study presents the results of two ILCs conducted over the last 3 years by 6 laboratories from: the Hospital Universitari i Politècnic La Fe (Valencia, Spain), the Universidad de Murcia, the Universidad de Sevilla, the Universitat Autònoma de Barcelona, the Hospital General Universitario Gregorio Marañón (Madrid, Spain), and the Centro Oncológico de Galicia (Figure 1).
Map showing the geographical locations of the participant laboratories.
Figure 1. Long description
A map of Spain with regional borders outlined in black. Six institutional logos are positioned around the map with blue arrows pointing to their respective geographical locations.
* In the Northwest, an arrow points from the logo of FUNDACIÓN Centro Oncológico de Galicia to the Galicia region.
* In the North, an arrow points from the I I G-M logo to the Madrid region in the center-north.
* In the Northeast, an arrow points from the U A B Universitat Autònoma de Barcelona logo to the Catalonia region.
* In the East, an arrow points from the Instituto de Investigación Sanitaria La Fe logo to the Valencia region.
* In the Southeast, an arrow points from the UNIVERSIDAD DE MURCIA 1272 logo to the Murcia region.
* In the Southwest, an arrow points from the UNIVERSIDAD DE SEVILLA logo to the Andalusia region.
An inset box in the bottom right corner shows the Canary Islands.
Materials And Methods
Blood Collection and Irradiation of the Samples
The 2 ILCs were organized by the biological dosimetry laboratory of the Hospital Universitari i Politècnic La Fe. For this purpose, blood samples from the same healthy donor were obtained by venipuncture and collected into heparinized tubes. Gel heparin was used as the anticoagulant in the first ILC, and liquid heparin in the second ILC. The extractions for both ILCs were performed separately in time, one for each ILC. Before blood collection, informed consent was obtained from the donor. The study followed institutional, national, and international ethical guidelines and was approved by the Animal and Human Experimentation Ethics Committee from the Universitat Autònoma de Barcelona (ref. CEEA-6587).
All samples were irradiated ex vivo (Figure S1) in a methacrylate tank with water at 37°C using the TrueBeam HDR accelerator located at the Radiation Oncology and Medical Physics Service of the Hospital Universitari i Politècnic La Fe. Irradiations were performed using a 6 MV photon beam (6X) with a dose rate of 600 MU/min. The irradiation field at isocenter was 8 × 16 cm² in order to have adequate scatter conditions, and the samples were positioned at a depth of 9 cm. The source-to-surface distance (SSD) was set to 100 cm minus the depth, resulting in an effective SSD of 91 cm. Exposures were delivered using parallel-opposed fields, with the gantry positioned at 90° and 270°, corresponding to left-inferior (LI) and left-dorsal (LD) orientations, to allow homogeneous irradiation of the whole blood sample. The irradiations were administered at the 100% isodose level, and the doses were calculated in terms of water kerma.
For the first ILC, two irradiations were performed at doses of 2.9 Gy (sample A) and 1.4 Gy (sample B), respectively. For the second ILC, a single irradiation of 2.63 Gy was delivered, and 2 types of samples were prepared: a partial-body exposure (sample A) and a whole-body exposure (sample B). The partial-body sample was obtained by mixing 50% irradiated blood with 50% non-irradiated blood. The doses chosen for the two ILCs were close to 2 Gy because, in a radiological emergency, this value is a critical limit used to separate low exposures from severe radiation damage. After irradiation and following IAEA recommendations, the blood samples were kept at 37°C for 2 h to allow the cellular repair mechanisms to take place.
Shipment to Participants and Cytogenetic Assays
For each ILC, two heparinized tubes containing 3 mL of blood were labelled as Sample A and Sample B, and packaged in compliance with UN 3373 regulation for the transport of biological substances. To simulate realistic triage and transport network conditions, the blind samples were dispatched at ambient temperature without the inclusion of temperature loggers. Shipment was managed from the Hospital Universitari i Politècnic La Fe by a transport company certified for the shipment of biological material. The participating laboratories received, for each ILC, the two tubes within 24 h.
In the first ILC, those laboratories with knowledge in DCA were requested to report dose estimates after analyzing 20, 50, and 500 cells per sample. In the latter case, 100 dicentrics were also accepted. The choice of these specific cell counts follows established international triage and routine criteria. Under large-scale emergency triage conditions, truncated scoring of 20 cells serves as an ultra-rapid binary indicator to confirm significant exposure, whereas scoring 50 cells improves precision for early medical categorization.Reference Lloyd, Edwards and Moquet 32 –Reference Wilkins, Romm and Oestreicher 34 Conversely, a full scoring of 500 cells or 100 dicentrics represents the routine standard necessary to minimize statistical uncertainty. 4 The participating laboratory with expertise in CBMN assay was asked to report the estimated dose after the analysis of 500 binucleated cells as a compromise between speed and accuracy. In the second ILC, the laboratories that used the DCA in the previous comparison reported dose estimates using both triage scoring (50 cells) and full scoring modes (500 cells or 100 dicentrics), and the laboratory using the CBMN assay reported the estimated dose after the analysis of 500 binucleated cells.
Dicentric chromosome assay
Each laboratory cultured lymphocytes from the blood they received following their own protocols and the IAEA 4 and ISO standard 5 recommendations. Whole-blood lymphocytes were cultured in appropriate culture medium at 37ºC with phytohemagglutinin (PHA) for mitotic stimulation. Metaphase arrest was induced using colcemid, which was added at 24 h by one laboratory (L1) and at 46-47 h by the remaining laboratories. After 48-50 h of culture, the cells underwent hypotonic treatment followed by fixation and 2-3 washing steps (Table 1). Metaphase spreads were then prepared on microscope slides for cytogenetic analysis. To avoid scoring second-division metaphases, all laboratories, except L1, stained the slides using the fluorescence-plus-Giemsa technique. Two laboratories used a semiautomated method to capture the metaphases that were then analyzed manually, and the remaining laboratories performed the cytogenetic analysis directly under the microscope. One laboratory (L3) applied automatic culture and spread systems.
Cell culture specifications by laboratory
Table 1. Long description
The table consists of six columns: Laboratory code, Assay, Treatment, Addition at, Harvest at, and Hypotonic treatment.
* L 1: dicentrics assay; Treatment: Colcemid 0.05 m L [10 micrograms/m L in P B S]; Addition at 24 h; Harvest at 50 h; Hypotonic treatment: K Cl 0.075 M for 20 min at 37 degrees Celsius.
* L 2: dicentrics assay; Treatment: Colcemid 0.05 m L [10 micrograms/m L in H B S S]; Addition at 47 h; Harvest at 49.5 h; Hypotonic treatment: K Cl 0.075 M for 20 min at 37 degrees Celsius.
* L 3: dicentrics assay; Treatment: Colcemid 0.15 m L [10 micrograms/m L in P B S]; Addition at 46 h; Harvest at 48 h; Hypotonic treatment: K Cl 0.075 M for 30 min at 37 degrees Celsius.
* L 4: dicentrics assay; Treatment: Colcemid 0.02 m L [10 micrograms/m L in P B S]; Addition at 46 h; Harvest at 48 h; Hypotonic treatment: K Cl 0.075 M for 10 min at 37 degrees Celsius.
* L 5: dicentrics assay; Treatment: Colcemid 0.05 m L [10 micrograms/m L in P B S]; Addition at 46 h; Harvest at 48 h; Hypotonic treatment: K Cl 0.075 M for 8 min at 37 degrees Celsius.
* L 6: micronuclei assay; Treatment: Cytochalasin-B [final culture concentration: 3 micrograms/m L]; Addition at 44 h; Harvest at 72 h; Hypotonic treatment: K Cl 0.075 M at 2 degrees Celsius.
Cytokinesis-block micronucleus assay
Whole-blood lymphocytes were cultured in appropriate culture medium at 37ºC. Cytochalasin B was added after 44 h. Then, after 72 h of culture (44 h + 28 h), the cells underwent hypotonic treatment followed by fixation and 2-3 washing steps (Table 1). Slides were then prepared for manual cytogenetic analysis using the microscope.
Dose-Effect Curves
To estimate the absorbed doses of the blinded samples, the participant laboratories used their own calibration curve, except for L2, which applied the reference calibration curve provided as an example in the IAEA manual. 4 According to the assay, L6 employed a calibration curve derived from the CBMN assay, while all other laboratories applied calibration curves derived from the DCA. Identification of the participating laboratories, the assays performed, and their corresponding calibration curves are indicated in Table 2.
Calibration curves by laboratory
Table 2. Long description
The table consists of nine columns: Laboratory code, Assay, Source, C, S E (C), alpha, S E (alpha), beta, and S E (beta).
* Row 1: L 1, dicentrics, 60 super Co, 0.0013, 0.0005, 0.0210, 0.0052, 0.0630, 0.0040.
* Row 2: L 2, dicentrics, 60 super Co, 0.0013, 0.0005, 0.0210, 0.0052, 0.0630, 0.0040.
* Row 3: L 3, dicentrics, 60 super Co, 0.0007, 0.0007, 0.0413, 0.0074, 0.0444, 0.0042.
* Row 4: L 4, dicentrics, 60 super Co, 0.0006, 0.0004, 0.0101, 0.0051, 0.0720, 0.0043.
* Row 5: L 5, dicentrics, 60 super Co, 0.0009, 0.0005, 0.0170, 0.0047, 0.0533, 0.0033.
* Row 6: L 6, micronuclei, 137 super Cs, 0.0148, 0.0021, 0.0305, 0.0013, em-dash, em-dash.
Statistical Analysis
The whole-body estimated doses were reported with a corresponding 95% confidence interval (CI) using Merkel’s method, while uncertainties associated with the simulated partial-body exposure were calculated by the Delta method. In the latter case, the partial-body dose was estimated using the Dolphin’s method, and to calculate the initial fraction of irradiated cells, a survival coefficient (D0) of 2.7 was assumed.
Laboratory performance was assessed using Z-ScoresReference Di Giorgio, Barquinero and Vallerga 19 , 35 , 36 as in equation 1:
Where
$ {x}_i $
is the estimated dose;
$ {x}_{ref} $
is the reference dose value;
$ {s}^{\ast } $
is the robust standard deviation and
$ {u}_{ref} $
is the standard uncertainty of
$ {x}_{ref} $
. The robust standard deviation was determined using Algorithm A and the uncertainty of the assigned reference dose was considered negligible. In accordance with the conventional interpretation of ISO 13528:2022,
35 the Z-Score allows the classification of the participants’ results as satisfactory
$ \left(\left|z\right|<2\right) $
, questionable
$ \left(2<\left|z\right|<3\right) $
or unsatisfactory
$ \left(\left|z\right|>3\right) $
.
In addition, when 20 and 50 cell counts were evaluated, an acceptance limit of
$ \pm $
0.5 Gy was checked as recommended for triage dosimetry,Reference Lloyd, Edwards and Moquet
32 whereas the 500 cell counts were evaluated by their 95% confidence intervals.
All statistical procedures were carried out using the fitting modules, the dose estimation modules, and the new module for ILCs from the software Biodose Tools v.3.7.1.Reference Hernández, Endesfelder and Einbeck 37 , Reference Francès-Abellán, Endesfelder and Hernández 38
Results
Dose Estimation of Blinded Samples
First ILC
The laboratories identified as L2, L5, and L6 did not obtain enough material during the culture procedure to report counts of 500 cells for both samples A and B. In consequence, L2 and L5 reported the dose estimates based on counts of 20 and 50 cells only, and L6 could not perform the CBMN assay established here in 500 cells. Thus, in the first ILC, because half of the laboratories could not report the results for 500 cells, the Z-Score was not applied to results based on 500-cell scoring.
For sample A (2.9 Gy), all laboratories overestimated the administered dose (Figure 2); however, in all cases the reported values were classified within the correct dose range (high exposure). When 20 cells were scored (L1-L5), L1 and L3 reported dose estimates outside the ±0.5 Gy acceptance limit, while L2, L4, and L5 remained inside this limit. Increasing the number of scored cells to 50 (L1-L5) improved the results, with only laboratory L3 reporting a dose estimate marginally outside the acceptance limit. The laboratories which could evaluate 500 cells (L1, L3, and L4) obtained results where the 95% confidence intervals (CI) included the reference dose. Z-Score values (Figure 4) indicated questionable performance for L1 when scoring 20 cells, and for L3 when scoring both 20 and 50 cells.
Plots of the different dose estimations of the first ILC with 20, 50, and 500 analyzed cells by laboratory. Dose ranges are defined as: Low exposures (<1 Gy) in green, medium exposures (1-2 Gy) in yellow, and high exposures (>2 Gy) in red. a-c: Sample A simulates a whole-body irradiation of 2.9 Gy. d-f: Sample B simulates a whole-body irradiation of 1.4 Gy.
Figure 2. Long description
A multi-panel figure with two rows and three columns. The top row shows Sample A (2.9 G y) and the bottom row shows Sample B (1.4 G y). Columns represent 20, 50, and 500 analyzed cells respectively. Each plot has an X-axis labeled Laboratories (L 1, L 2, L 3, L 4, L 5) and a Y-axis labeled Dose (G y). Backgrounds are color-coded: green for low exposure (less than 1 G y), yellow for medium (1 to 2 G y), and red for high (greater than 2 G y).
* Panel a (Sample A, 20 cells): Five data points with large error bars centered around 3 G y. A red dashed reference line sits at 2.9 G y.
* Panel b (Sample A, 50 cells): Five data points with smaller error bars than panel a, more closely aligned with the 2.9 G y reference line.
* Panel c (Sample A, 500 cells): Three data points (L 1, L 3, L 4) with the smallest error bars, tightly clustered around the 2.9 G y line.
* Panel d (Sample B, 20 cells): Five data points with large error bars mostly in the green and yellow zones. A red dashed reference line sits at 1.4 G y.
* Panel e (Sample B, 50 cells): Five data points with moderate error bars centered near the 1.4 G y reference line.
* Panel f (Sample B, 500 cells): Three data points (L 1, L 3, L 4) with small error bars closely matching the 1.4 G y reference line.
Legends on the right define Dose Ranges (high, low, medium) and Reference doses (Whole-body and 0.5 G y).
For sample B (1.4 Gy), some laboratories overestimated the dose while others underestimated it (Figure 2). However, when 20 and 50 cells were scored (L1-L5), the laboratories L2-L5 were classified in the correct dose range (medium exposure) and inside the ±0.5 Gy acceptance limit. For L1, the reported estimates for 50 cells were also inside the correct range and acceptance limit, but for 20 cells, the reported range was mistaken for a low exposure, and the acceptance limit was surpassed with an estimated dose of 0.73 Gy. When 500 cells could be scored (L1, L3, and L4), the dose estimates were slightly overestimated but remained in the correct dose range. The reference dose was clearly included in the 95% CI for L3 and L4, whereas for L1 it overlapped with the lower boundary. The laboratories L1-L5 reported a satisfactory Z-Score value for 20 and 50 cells; however, as shown in Figure 4, L1 exhibited a tendency toward questionable performance compared with the other laboratories for this sample.
Dose estimates and Z-score values for the first ILC can be consulted in the supplementary data (Tables S3, S7, and S8).
Second ILC
For the partial-body exposure (sample A), after scoring 50 cells, although L2-L4 identified cells with two or more dicentrics, needed to detect an overdispersion that would suggest partiality, only L2 obtained a u-value higher than 1.96, indicating significant overdispersion and a partial-body exposure (defined as dispersion index (DI) > 1 with a u-value >1.96). However, Figure 3a shows the estimated doses assuming partial-body exposure for L2-L4. Both L2 and L4 were classified in the correct range (high exposure), but the only dose estimate inside the ±0.5 Gy acceptance limit was L4. The result for L3 fell outside both the range and the acceptance limit. In consequence, the three estimated partial body doses (L2-L4) vary significantly with a standard deviation (σ) = 1.342.
Plots of the different dose estimations of the second ILC with 50 and 500 cell counts by laboratory. Dose ranges are defined as: Low exposures (<1 Gy) in green, medium exposures (1-2 Gy) in yellow, and high exposures (>2 Gy) in red. a-b: Sample A simulates a partial-body irradiation of 2.63 Gy. The whole-body equivalent was calculated for comparison. c-d: Sample B simulates a whole-body irradiation of 2.63 Gy.
Figure 3. Long description
A four-panel set of dot plots titled Second I L C. Each plot has Laboratories L 1 through L 6 on the x-axis and Dose in G y on the y-axis. Backgrounds are color-coded: green for low exposure (0 to 1 G y), yellow for medium (1 to 2 G y), and red for high (above 2 G y).
* Panel a (Sample A, 50 cells): Shows high variability. Partial-body estimates (open diamonds) range from approximately 1.8 to 4.5 G y, mostly in the red zone. Whole-body estimates (solid circles) are lower, between 0.8 and 1.8 G y, spanning green and yellow zones.
* Panel b (Sample A, 500 cells): Estimates are more tightly clustered. Partial-body estimates are consistently in the red zone around 2.5 to 3.0 G y. Whole-body estimates cluster in the yellow zone around 1.5 G y.
* Panel c (Sample B, 50 cells): All estimates are for whole-body irradiation. Data points cluster around the 2.63 G y reference line in the red zone, with large error bars extending down into the yellow zone.
* Panel d (Sample B, 500 cells): Whole-body estimates are more precise, closely following the 2.63 G y dotted reference line within the red zone. Laboratory L 6 uses a triangle symbol for C B M N assessment.
Reference lines include a dashed red line for partial-body (2.63 G y), a dotted line for whole-body (1.3 or 2.63 G y), and a dashed grey line at 0.5 G y.
Both L1 and L5 could not identify a partial-body exposure when counting 50 cells; thus, they reported the whole-body estimate. The whole-body estimation obtained by L5 (1.57 Gy) was within the expected range and met the ±0.5 Gy acceptance limit, whereas the estimation reported by L1 (0.63 Gy) satisfied neither the acceptance criterion nor the correct range. For comparison, Figure 3 (a and b) shows the estimated whole-body doses for all the laboratories (L1-L5), equivalent to approximately 1.3 Gy with a σ = 0.43.
After scoring 500 cells, only laboratories L5 (DCA) and L6 (CBMN) did not detect the irradiation as partial, with u-values lower than 1.96. However, L5 detected an overdispersion that was nearly significant and reported the partial-body estimate. For L6, which did not find binucleated cells with two or more MN, it was not possible to evaluate overdispersion, and a whole-body estimate was reported (Figure 3b). All the reported partial-body dose estimations (L1-L5) based on counts of 500 cells included the reference dose in their 95% CI and were classified in the correct dose range. The whole-body estimation obtained by L6 (1.67 Gy) also met these criteria.
The estimates of the irradiated cell fraction (Table S6) based on counts of 50 cells deviated more from the true value of 0.5 (50% irradiated blood) than those based on counts of 500 cells (Mean Absolute Deviations were 0.37 and 0.13, respectively).
For the whole-body exposure (sample B), all six laboratories estimated the dose analyzing both 50 and 500 cells, except for L6, which directly reported the 500-cell estimate according to plan. When scoring 50 cells, the dose estimates from all laboratories were within the correct dose range, corresponding to high exposure, and approximately within the ±0.5 Gy acceptance limit, although the estimates for L1, L2, and L5 were very close to the boundary of this limit. For the 500-cell estimations, all participating laboratories estimated doses within the correct dose range, and the reference dose was always included in the 95% CI (Figure 3c,d). The analysis based on 50 cells presented a higher standard deviation (σ = 0.43) than for 500 cells (σ = 0.25).
Figure 4 presents the Z-scores obtained for the estimated doses of samples A and B. Only three laboratories were able to report dose estimates for sample A based on 50 cells; thus, Z-scores were not calculated. The reported values (sample A500 and sample B50, B500) fell inside the satisfactory range (
$ \pm $
2).
Plots of the Z-score value for samples A and B of the first ILC (a-b) and the second ILC (c-d) by laboratory. The color legend indicates the performance.
Figure 4. Long description
A four-panel scatter plot titled Z-Score. Each plot has an x-axis labeled Laboratories with discrete points L 1 through L 5 or L 6 and a y-axis labeled Z-Score ranging from minus 3 to 3. The background is divided into horizontal color bands: green for satisfactory (Z between minus 2 and 2), yellow for questionable (Z between 2 to 3 and minus 2 to minus 3), and red for unsatisfactory (Z greater than 3 or less than minus 3).
* Panel a, First I L C Sample A: Shows data for 20 cells (pink circles) and 50 cells (blue triangles). Most points are in the green zone, with L 1 and L 3 pink circles reaching the yellow questionable zone.
* Panel b, First I L C Sample B: Shows data for 20 cells and 50 cells. All points for L 1 through L 5 fall within the green satisfactory zone.
* Panel c, Second I L C Sample A: Shows data for 500 cells (blue triangles). All points for L 1 through L 5 are within the green satisfactory zone.
* Panel d, Second I L C Sample B: Shows data for 50 cells (pink circles) and 500 cells (blue triangles). All points for L 1 through L 6 are within the green satisfactory zone.
A legend on the right indicates that pink circles represent smaller cell counts (20 or 50) and blue triangles represent larger cell counts (50 or 500) depending on the I L C round.
Dose estimates and Z-Score values for the second ILC (Tables S6, S9, and S10), and the distribution of aberrations reported by each laboratory for both ILCs (Tables S1, S2, S4, and S5) can be consulted in the supplementary data.
Discussion
The number of biological dosimetry laboratories varies among countries, as does the type of institution to which they are affiliated. Most countries have only a small number of laboratories, typically one or two, with the primary laboratory most often affiliated with government institutions, emergency preparedness organizations, or radiation protection authorities. When a second laboratory exists, it usually belongs to the defense sector.Reference Wilkins, Carr and Lloyd 39 However, Spain presents a somewhat particular situation, with a relatively high number of laboratories, all of which belong to the hospital or university sector. Some of these laboratories are regular participants in international networks such as RENEB or BioDoseNet, taking part in their ILCs (L1, L3, and L4), while other laboratories have either temporarily discontinued such international collaborations or have not yet established them (L2, L5, and L6). In this context, the CSN launched a project to establish and harmonize an interactive national rapid-response framework capable of developing and supporting a National Protocol for Biological Dosimetry through complementary expertise among participating groups in Spain.
The results from the first ILC, in line with the findings reported by Prasanna et al.,Reference Prasanna, Moroni and Pellmar 40 indicate that for doses above 2 Gy (2.9 Gy in the present study), an initial screening of 20 cells rather than the conventional 500 can reliably classify samples within the correct high-dose exposure range. This corroborates that a reduced cell count can work as an efficient preliminary step for identifying individuals who may require more detailed analysis. However, at lower dose levels, such as sample B (1.4 Gy), the analysis of only 20 cells appears insufficient for accurate dose range classification. In agreement with other experiments such as Wilkins et al.,Reference Wilkins, Romm and Oestreicher 34 scoring 50 cells seems sufficient to obtain reliable whole-body dose estimates.
Findings from the second ILC highlight the complexity of partial-body exposure assessment. While whole-body dose estimation relies on a uniform damage distribution, identifying a partial-body exposure requires detecting an overdispersed distribution of the induced aberrations. Lloyd et al.Reference Lloyd, Edwards and Moquet 32 previously demonstrated that, for simulated partial-body exposures, scoring 50 cells provided satisfactory estimated doses and proportion of the irradiated fraction at doses above 3 Gy. The findings reported by Prasanna et al.Reference Prasanna, Moroni and Pellmar 40 further corroborate that such an approach is feasible at high dose levels; however, accurate estimation of the irradiated body fraction was shown to be reliable only for relatively large fractions (>25%) and doses exceeding 3 Gy when using Dolphin’s method. Adding to these observations, our results support that, at doses below 3 Gy (2.63 Gy), analyses based on only 50 cells were insufficient for the reliable detection of partial-body exposures. When 50 cells were analyzed, only 3 laboratories identified partial exposure, and the corresponding dose estimates showed substantial deviations from the expected values in 2 out of the 3 cases. In contrast, when 500 cells were scored, all laboratories specializing in dicentric analysis produced consistent dose estimates. Furthermore, the reported irradiated body fraction values (Table S6) suggest that even at 50% fractions, the estimation needs more than 50 cells to obtain reliable results in doses lower than 3 Gy.
The detection of partial-body exposure has important medical implications, as medical management and treatment strategies may differ depending on the exposure type. A clear example of this is provided by a radiological accident with industrial radiography that occurred in Chilca, Peru. 31 In this accident, the most affected individual exhibited blood cell aplasia, and, due to the heterogeneous nature of his exposure (u-value > 1.96), it was anticipated that a progressive hematological recovery would occur. In fact, animal studies have shown that even 5% bone marrow shielding results in a large increase in survival from hematopoietic acute radiation syndrome.Reference Van Bekkum and Schotman 41 Consequently, a less aggressive hematological treatment approach was administered 31. In this regard, the software used in this study, Biodosetools, automatically performs calculations for both whole-body and partial-body exposures when partial exposure is assessed, requiring no additional analytical effort. Given that most laboratories failed to detect partial exposure when scoring 50 cells, we consider that calculating both partial- and whole-body doses provides a rapid risk-based stratification. When the whole-body dose estimated with 50 cells falls within the medium or high-dose range, a good practice would be to prioritize these cases and increase the number of cells analyzed to better determine whether the exposure was partial or whole-body.
Other previous studies, such as Port et al.,Reference Port, Barquinero and Endesfelder 24 conducted an ILC in which participating laboratories applied all biological dosimetry assays available within their expertise to estimate radiation dose, and then their performance was evaluated separately for each assay. In the present study and for both ILCs, each participating laboratory performed its preferred technique (CBMN assay or DCA) to estimate the dose, and this result was compared among them. The obtained results suggest that laboratories can contribute to dose assessment in large-scale radiological incidents by applying the biological dosimetry technique with which they are most experienced or equipped. It is important to mention that the CBMN assay, while successfully assessing dose estimation for whole-body exposure and whole-body estimation of the partial-body simulation, could not reliably identify partial-body exposure under the conditions tested. In other simulated partial-body exposures, when comparing both dicentrics and MN, it has been indicated that without accident information, the MN assay could not reliably distinguish between whole- and partial-body irradiation.Reference Mendes, Mendonça and Barquinero 29 After a homogeneous exposure, and contrary to dicentrics, MN are usually overdispersed; therefore, approaches based on zero-inflated Poisson models, such as the Dolphin’s method, are probably not fully suitable. In this sense, alternative models such as zero-inflated negative binomial distribution could improve the suitability of detecting partial-body doses using the CBMN assay. Additional analyses are therefore required to determine whether this limitation is assay-specific. Notably, previous studies have reported successful partial-dose estimation using the CBMN assay when scoring more than 500 cells.Reference le Roux, Slabbert and Smit 42 –Reference Purnami, Nurhayati and Syaifudin 44
The overall dose estimation results showed good agreement with the reference dose for each sample, with approximately 69% of participants obtaining acceptable results in the first ILC and 77% in the second, comparable to the 75% reported by Pan et al.Reference Pan, Ruan and Gao 16 However, if the analysis is restricted to the Z-Score values, which could only be calculated when results were available from all participants, the proportion of acceptable estimates would increase to 80% for the first ILC and 100% for the second.
These results and others previously reportedReference Wilkins, Beaton-Green and Lachapelle 13 , Reference Pan, Ruan and Gao 16 , Reference Oestreicher, Samaga and Ainsbury 20 show that blood samples could be shared among laboratories in the event of a major radiological accident to perform biological dosimetry. Specifically, this study demonstrates that the logistics of a Spanish network would be feasible. However, there are several limitations identified in these ILCs. The first limitation concerned difficulties with cell culture, which prevented three laboratories from reporting the required 500 metaphase counts during the first ILC. For comparison, in the RENEB ILC conducted in 2017,Reference Gregoire, Barquinero and Gruel 22 only 6 of the 38 participating laboratories failed to reach the required 500 metaphases, while in our first exercise it represented half of the participating laboratories (3 out of 6). However, a substantial improvement was observed during the second exercise, in which all participating laboratories successfully reached the required number of metaphases, corresponding to a 100% success rate. The second identified issue was the ability to detect partial-body exposures when analyzing only 50 cells during the second ILC, which has been discussed above. The third issue was the limited number of participating laboratories, which prevented the calculation of Z-Scores when not all laboratories were able to report a result and decreased the statistical robustness of the analysis.
In the future, although these results demonstrate the capacity of Spanish dosimetry laboratories to successfully share blood samples in the event of a major radiological accident, perform dose estimations, and classify individuals within the appropriate exposure range for triage, further ILCs will be required to improve the standardization of biodosimetry practices and to maintain a reliable national protocol.
Conclusions
In the present study, 2 ILCs were conducted performing 2 well-established techniques in biological dosimetry (DCA and CBMN). This type of approach likely better reflects a real emergency scenario, in which all laboratories with an implemented dose biomarker would need to be activated. Although the way in which different techniques are compared should be further reconsidered, the results obtained in these ILCs reinforce the importance of biological dosimetry networks as an emergency response strategy to increase analytical capacity and available resources. The results show that, for the participating laboratories, scoring 20 cells under triage conditions for whole-body exposures using the DCA was enough to classify within the correct dose range for high doses, and 50 cells was generally enough to obtain reliable dose estimates. For partial-body exposures, the findings indicate that scoring more than 50 cells is required at doses below 3 Gy (2.63 Gy) even at 50% partiality to achieve reliable dose estimation. Regarding the CBMN assay, while a good result was achieved for whole-body estimations when 500 cells were obtained, further experimental studies are necessary to better adjust its performance in partial-body exposure triage scenarios.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2026.10471.
Author contribution
AFA: Former analysis; statistical analysis; writing original draft. AMP: Former analysis; project administration. JA: Sample irradiation. MA: Former analysis; writing—review and editing. GA: Writing—review and editing. ID: Former analysis; writing—review and editing. JLFG: Former analysis; writing—review and editing. CFM: Former analysis. JAGG: Former analysis; writing—review and editing. ILB: Former analysis; writing—review and editing. AM: Sample irradiation. AMMD: Former analysis; writing—review and editing. MMD: Former analysis; writing—review and editing. AO: Former analysis; writing—review and editing. NP: Former analysis; writing—review and editing. JPC: Sample irradiation; writing—review and editing. MJPR: Former analysis; writing—review and editing. MJRG: Former analysis. ASP: Sample irradiation. JIVB: Sample irradiation. JFB: Former analysis; supervision; writing original draft.
Funding statement
This work was supported by the Consejo de Seguridad Nuclear Español under Grant PR-076-2022.
Competing interests
None.
Open access
