Radiation accidents involving high doses are relatively rare. Accidents involving partial-body irradiations with severe complications are more frequent than those with uniform whole-body irradiations. For risk-based stratification, in radiation mass casualties, an approach of estimating equivalent whole-body doses is likely to be used. Therefore, it is necessary to test whether estimation of equivalent whole-body doses under partial-body exposure conditions is valid. We used an in vitro simulated partial-body irradiation model by mixing irradiated and un-irradiated lymphocytes in various proportions for validation. Similar in vitro simulated mixed culture model was previously used in cytogenetic studies (Lloyd et al. 2000
, Lloyd et al. 1987
, Blakely et al. 1995
Our laboratory’s approach for using DCA for rapid risk-based stratification of radiation exposed population is proposed in . Accordingly, following receipt of blood samples from a radiation accident, DCA is performed. Initial screening involves confirmation of irradiation above 2-Gy by analysis of only 20 metaphases instead of the typical 500–1000 metaphases scored during routine analysis (IAEA 2001
; ISO 2004
). Presence of four dicentrics in 20 metaphase spreads (0.2 dicentrics per cell) indicates a potential dose of approximately 2-Gy with a lower confidence limit (LCL) of 0.85 Gy. For cases with confirmed ≥2-Gy dose, analysis is increased to 50 metaphases. Metaphases are evaluated for indications of homogeneous or inhomogeneous distribution of dicentrics (Lloyd et al. 2000
; Prasanna et al. 2003
); partial-body exposures are indicated by variation from the expected dose-dependent distribution of number of dicentric per cell. In cases of uniform whole-body exposures, samples are categorized based on a risk-based stratification table (simplified table is shown in for clarity) into “not life-threatening (green)”, “potentially life-threatening (yellow)”, and “significantly life-threatening (red)” cohorts, as indicated (Lloyd et al. 2000
, Prasanna et al. 2003
) in the embedded table (). Assignment of victims to risk-based stratified cohorts will enable prioritization of samples for medical care and to establish sample scoring priority if additional (500 metaphase spreads) cytogenetic analysis is required. Construction of the calibration curve shown in is based on an analysis of 1000 metaphase spreads. This calibration curve for our laboratory previously published (Wilkins et al. 2008
), was fitted by a linear quadratic model, described by the equation, y = c+αD+βD2
using the CABAS software (Deperas et al. 2007). The coefficients of this fitted curve are α = 0.0293±0.0078, β = 0.0369 ±0.0036, and c = 0.0032 ±0.0016, with standard deviations as indicated. Distribution of dicentrics in cell population at all uniform doses conforms to a Poisson distribution. For non-uniform exposure, the distribution of dicentrics will be over-dispersed. For these cases, the triage dose prediction model will be used to predict the equivalent whole-body dose ().
Figure 1 Illustration of rapid risk-based stratification of radiation-exposed population by the dicentric assay after whole-body and partial-body exposures. Following receipt of blood samples from a radiation accident, DCA is performed. For confirmation of irradiation, (more ...)
and show equivalent whole-body doses estimated after analysis of 50 metaphases under in vitro simulated partial-body irradiation conditions, for 3-Gy and 5-Gy doses, along with LCL and upper confidence limit (UCL) based on exact Poisson error on yield, distribution of dicentrics, dispersion coefficients (variance/mean), u values, and nature of distribution.
Table 1 Equivalent whole-body dose estimation after simulated in vitro partial-body exposure to a 3-Gy dose. Different proportions of 3-Gy irradiated blood were mixed with unirradiated blood at various fractions. Radiation doses were estimated to the fractions (more ...)
Table 2 Equivalent whole-body dose estimation after simulated in vitro partial-body exposure to a 5-Gy dose. Different proportions of 5-Gy irradiated blood were mixed with unirradiated blood at various fractions. Radiation doses were estimated to the fractions (more ...)
Expected physical doses, derived from the percent irradiated volume, fell within the 95% LCL and upper confidence limits (UCL) of estimated equivalent whole-body doses for all fractions at both radiation doses.. However, in one case (5-Gy, 50% irradiated fraction) the underestimation was quite large (0.95-Gy estimated dose versus 2.5-Gy expected dose, LCL 0.25-Gy and UCL 2.61-Gy) and placed the victim in the ‘wrong’ category (“not life-threatening” instead of “potentially life-threatening”) using the triage dose prediction model. For one additional case, 3-Gy, 50% irradiated sample, although the estimated whole-body equivalent dose was within the 95% LCL and UCL of expected dose, the value was borderline between “not-life threatening” and “potentially life threatening” categories. Although our results are preliminary, we propose that in cases of non-uniform exposures, estimation of equivalent whole-body doses may be considered for rapid risk-based stratification. As our results are preliminary, further studies involving larger number of samples are warranted, before integration of this partial-body dose assessment approach for treatment.
Our observations on the nature of dicentric distributions at 3-Gy and 5-Gy irradiated fractions, show a Poisson distribution at the lower dose fractions and non-Poisson distribution at the higher dose fractions, as revealed by the dispersion co-efficients and u
values. It is evident that at a relatively lower dose of 3-Gy, analysis 50 metaphases is inadequate to precisely identify partial-body exposures. However, precise identification of partial-body exposures is dependent on, in addition to number of metaphases analyzed, at least three other factors: (i) calibration curve, (ii) fraction of the body irradiated, and (iii) dose to the irradiated fraction. From our 5-Gy data (), it is evident that fractions of 25–75% are identifiable as partial-body irradiations, for the number of metaphases analyzed, based on non-Poisson distribution. Therefore, it is imperative under partial-body exposure conditions, not only to estimate equivalent whole-body doses, but also consider the nature of dicentric distributions for clinical management, as previously suggested by Lloyd (1997)
. Early identification of partial-body exposures is essential in order to determine whether cytokine therapy is beneficial.
Estimates of whole-body equivalent doses using the triage dose prediction model was accurate in all cases. Only in one case the true dose was seriously underestimated, placing the victim in the “wrong category”. Lloyd et al. (2000)
has previously shown that for simulated partial-body irradiation, scoring of 50 metaphases provided satisfactory agreement between the true and the estimated dose and the percentage of irradiated fraction, at 3-Gy and above. Our preliminary results involved relatively a small number of samples and need to be further confirmed using a larger sample size, but they appear to support the notion that detection of partial-body irradiation, particularly at higher doses, is possible with only 50 metaphases. However, accurate estimation of the fraction of irradiated body could be satisfactorily determined only for large fraction (≥25%) and doses >3-Gy, using the Qdr or the Dolphin methods. The triage dose prediction model is an initial step towards rapidly assigning victims to risk-based categories. Additional testing and assessment of probability for true-positive and true-negative cases are required.