The suitability of micronuclei as markers of relative biological effect.

Abstract

Micronucleus (MN) formation is routinely used as a biodosimeter for radiation exposures and has historically been used as a measure of DNA damage in cells. Strongly correlating with dose, MN are also suggested to indicate radiation quality, differentiating between particle and photon irradiation. The "gold standard" for measuring MN formation is Fenech's cytokinesis-block micronucleus (CBMN) cytome assay, which uses the cytokinesis blocking agent cytochalasin-B. Here, we present a comprehensive analysis of the literature investigating MN induction trends in vitro, collating 193 publications, with 2476 data points. Data were collected from original studies that used the CBMN assay to quantify MN in response to ionizing radiation in vitro. Overall, the meta-analysis showed that individual studies mostly have a linear increase of MN with dose [85% of MN per cell (MNPC) datasets and 89% of percentage containing MN (PCMN) datasets had an R2 greater than 0.90]. However, there is high variation between studies, resulting in a low R2 when data are combined (0.47 for MNPC datasets and 0.60 for PCMN datasets). Particle type, species, cell type, and cytochalasin-B concentration were suggested to influence MN frequency. However, variation in the data meant that the effects could not be strongly correlated with the experimental parameters investigated. There is less variation between studies when comparing the PCMN rather than the number of MNPC. Deviation from CBMN protocol specified timings did not have a large effect on MN induction. However, further analysis showed less variation between studies following Fenech's protocol closely, which provided more reliable results. By limiting the cell type and species as well as only selecting studies following the Fenech protocol, R2 was increased to 0.64 for both measures. We therefore determine that due to variation between studies, MN are currently a poor predictor of radiation-induced DNA damage and make recommendations for futures studies assessing MN to improve consistency between datasets.