Genotoxic mixtures and dissimilar action: concepts for prediction and assessment

Abstract

Combinations of genotoxic agents have frequently been assessed without clear assumptions regarding their expected (additive) mixture effects, often leading to claims of synergisms that might in fact be compatible with additivity. We have shown earlier that the combined effects of chemicals, which induce micronuclei (MN) in the cytokinesis-block micronucleus assay in Chinese hamster ovary-K1 cells by a similar mechanism, were additive according to the concept of concentration addition (CA). Here, we extended these studies and investigated for the first time whether valid additivity expectations can be formulated for MN-inducing chemicals that operate through a variety of mechanisms, including aneugens and clastogens (DNA cross-linkers, topoisomerase II inhibitors, minor groove binders). We expected that their effects should follow the additivity principles of independent action (IA). With two mixtures, one composed of various aneugens (colchicine, flubendazole, vinblastine sulphate, griseofulvin, paclitaxel), and another composed of aneugens and clastogens (flubendazole, doxorubicin, etoposide, melphalan and mitomycin C), we observed mixture effects that fell between the additivity predictions derived from CA and IA. We achieved better agreement between observation and prediction by grouping the chemicals into common assessment groups and using hybrid CA/IA prediction models. The combined effects of four dissimilarly acting compounds (flubendazole, paclitaxel, doxorubicin and melphalan) also fell within CA and IA. Two binary mixtures (flubendazole/paclitaxel and flubendazole/doxorubicin) showed effects in reasonable agreement with IA additivity. Our studies provide a systematic basis for the investigation of mixtures that affect endpoints of relevance to genotoxicity and show that their effects are largely additive.

Fig. 1

Induction of MN by aneugens and clastogens in the CBMN assay using CHO-K1 cells. MN induction is presented as percentage of MN positive binucleated cells. The graphs show the data for at least three independent experiments (red dots, exception: benzo[α]pyrene was tested only once); solvent controls are shown on the left (green dots as indicated). The regression curves (thick black lines) are shown with their 95 % confidence belts (dashed lines). Estimated threshold concentrations are indicated by the vertical dashed lines. Mean baseline levels of MN within the cells are depicted as horizontal lines. The grey areas show the cytotoxic concentrations determined in the MTT assay (MTT–EC₄₀) (colour figure online)

Fig. 2

Predicted and observed induction of MN by two mixtures of aneugens or aneugens and clastogens in the CBMN assay. Mixture I was composed of flubendazole, colchicine, griseofulvin, paclitaxel and vinblastine (a), and mixture II of flubendazole, doxorubicin, etoposide, melphalan and mitomycin C (c). Prediction curves were derived from CA (green curves in a, c as labelled) and IA (light blue curves in a, c as labelled), with dashed lines as the respective 95 % confidence belts. Prediction curves were also generated from a hybrid CA/IA model (dark blue lines in b and d as labelled) for mixture I (b) and mixture II (d), with compounds grouped according to strict criteria of similar and dissimilar mechanism of action. All mixtures were designed at a ratio of the estimated threshold concentrations of the individual compounds and tested as dilution series (the mixture concentrations corresponding to the sum of the individual threshold concentrations are indicated as ΣITC). Data are shown from at least three independent experiments (red dots a–d), together with their regression curves (thick black lines) and 95 % confidence belts (dashed lines b, d). Threshold concentrations (vertical dashed lines) and mean baseline levels of MN (horizontal line) were estimated by regression analysis (see Table 2 for more information). The grey areas show the cytotoxic concentrations determined in the MTT assay (MTT–EC₄₀) (colour figure online)

Fig. 3

Predicted and observed induction of MN by three mixtures of aneugens or aneugens and clastogens in the CBMN assay. Mixture III was composed of flubendazole, paclitaxel, doxorubicin and melphalan (a), mixture IV of flubendazole and paclitaxel (c), and mixture V of flubendazole and doxorubicin (c). Prediction curves were derived from CA (green curves as labelled) and IA (light blue curves as labelled), with dashed lines the respective 95 % confidence belts. Prediction curves were re-calculated by using only non-threshold regression models and are shown for mixture III (b), mixture IV (d) and mixture V (f). All mixtures were designed at a ratio of the estimated threshold concentrations of the individual compounds and tested as dilution series (the mixture concentration corresponding to the sum of the individual threshold concentrations is indicated as ΣITC). Data shown are from at least three independent experiments (red dots), together with their regression curves and 95 % confidence belts (thick black curves with dashed lines, b, d and f). Mean baseline levels of MN (horizontal line) were estimated by regression analysis (see Table 2 for more information). The grey areas show the cytotoxic concentrations determined in the MTT assay (MTT–EC₄₀) (colour figure online)