Characterization of Risk to Human Health

HBCD has low acute toxicity. In the one chronic bioassay identified, dietary exposure to mice for 18 months did not result in increased incidence of any tumours. The results of a limited database indicate that HBCD does not have significant genotoxic potential in vitro or in vivo.

Short-term repeated-dose toxicity studies have identified effects upon the liver and the thyroid with adverse effect levels ranging from 29.9 to 125 mg/kg-bw per day (Chengelis 1997; van der Ven et al. 2006). The European Union had selected the LOAEL of 29.9 mg/kg bw per day as one of two critical effect levels upon which to characterize risk (EU RAR 2008).

The European Union had selected a NOAEL (10 mg/kg-bw per day) from the Ema et al. (2008) two-generation reproductive assay with Crl:CD(SD) rats for assessing risk to susceptible populations for long-term exposure (EU RAR 2008). The LOAEL was 101 mg/kg-bw per day, based upon a dose-related decrease in fertility index in the F0 generation, a significant decrease in the number of primordial follicles in the ovary and an increased incidence of animals with decreased size of thyroid follicles in the two highest dose groups in both the F0 and F1 generations.

One study identified an endpoint of potential concern for susceptible subpopulations (i.e., infants and children). Eriksson et al. (2006) exposed neonatal (day 10) male NMRI mouse pups to HBCD by gavage once, at either 0, 0.9 or 13.5 mg/kg-bw. At the age of three months, the mice were assessed for spontaneous behaviour, and learning and memory capability. The lowest dose was the LOAEL, 0.9 mg/kg-bw, based upon significantly altered spontaneous behaviour (hyperactive condition and reduced habituation). While changes in spontaneous behaviour have not been noted in other animal studies, this endpoint was taken into consideration in risk characterization.

Low adverse effect levels were noted in a recent one-generation study with rats (van der Ven et al. 2009; Lilienthal et al. 2009a). The authors stated that “the most sensitive” effects were decreased mineral density in trabecular bone in F1 females (critical effective dose, 0.18 mg/kg-bw per day), decreased concentration of apolar retinoids in liver in F1 females (critical effect dose, 5.1 mg/kg-bw per day) and increased immune response in F1 males (critical effective dose, 1.45 mg/kg-bw per day). For each of these three endpoints, a dose-response is either not clear (e.g., trabecular bone mineral content, increased immune response) or is evident only at the higher levels of exposure (apolar liver retinoids). Due to the limitations of this study, it was not considered further for risk characterization.

HBCD has been detected in human blood, cord serum, human breast milk, dust and ambient air in Canada . Data were identified for concentrations in foods in the United States . The most relevant study for human health risk characterization was determined to be the two-generation reproductive assay with rats (Ema et al. 2008).

To assess the risk from exposure of the general population of Canada to HBCD over a lifetime, a conservative NOAEL of 10 mg/kg-bw/day was selected from the Ema et al. (2008) two-generation reproductive toxicity study. Additionally, it was considered appropriate to charactize the magnitude of the margin between potential exposures to infants and children and the LOAEL of 0.9 mg/kg-bw derived from the Eriksson (2006) study.

The most highly exposed subpopulation of the general population of Canada was breast fed infants of 0 - 6 months age at 1.1 x 10^-1 µg/kg-bw per day derived from combined intakes from food, soil, dust and other environmental media (Health Canada 2008). This finding correlates well with the estimate derived by Eljarrat et al. (2009) for nursing infants in A. Corûna, northwestern Spain , of 1.75 x 10^-1 µg ΣHBCD/kg-bw per day. The exposures for formula-fed infants and non-formula-fed infants 0–6 months of age were 1.0 x 10^-2 µg ΣHBCD/kg-bw per day and 4.0 x 10^-2 µg ΣHBCD/kg-bw per day, respectively. The upper-bounding estimates of exposure for the general population of Canada , as reported in Appendix D, considers levels of HBCD in household dust and food.

Based on product scenario modelling, the highest consumer product exposure estimate was 5.6 x 10^-5 µg/kg-bw per day for an infant (0 - 6 months of age) from mouthing of flame-retarded textile or upholstered furniture.

A comparison between the critical effect level identified for the general population (10 mg/kg-bw per day) and the upper-bounding estimates of exposure for the general population (0.047 µg HBCD/kg-bw per day) results in a margin of exposure of 213 000. Additionally, the margin between upper-bounding exposures (1.1 x 10^-1 µg HBCD/kg-bw per day) for breast-fed infants and the LOEL of 0.9 mg/kg-bw per day results is 8200. These margins of exposure are considered adequate to address uncertaintities in the exposure and health effects databases. Since the intakes of the rats in the two-generation study spanned in uteroexposure, lactation and feed, the margins between the estimated intakes from human milk and the critical effect level in the animal database are considered protective of human health. This is consistent with the conclusions in the EU RAR (2008), which derived a margin of safety (MOS) of 7.0 x 10⁵ and concluded no concern for reproductive/fertility toxicity for breast-fed infants and no concern for breast-fed infants for repeated-dose toxicity with a MOS of 1.5 x 10⁶.

The margin of exposure between the estimated intake of 5.6 x 10^-5 µg/kg-bw per day by infant mouthing of textiles and the most conservative LOAEL in the animal database of 0.9 mg/kg-bw per day is 1.8 x 10⁷.

There is moderate confidence in the database of toxicity studies for HBCD. Although the only carcinogenicity bioassay was inadequately reported, the available genotoxicity studies were negative. The critical two-generation reproduction study was reported to be compliant with the OECD guideline and conducted in accordance with good laboratory practices. Furthermore, consistent effects at similar levels of exposure were observed among the studies.

Canadian environmental media data were available for several media including levels in Canadian human breast milk, outdoor air, food biota (lake trout) and dust. Uncertainty exists in the use of the Canadian maximum level of HBCD in dust of 1300 µg/kg. There is uncertainty in the use of a surrogate median level of 0.0002 µg/m³ from the United Kingdom (Abdallah et al. 2008a) and in the use of total HBCD levels in food from a market basket survey conducted in the United States (Schecter et al. 2009). In regard to the latter, the uncertainty is probably low given that similar food commodities are available in Canada , and U.S. levels would likely be representative of levels found in Canada . Worst-case exposure estimates are achieved through the inclusion of limit of detection values for non-detects. A level of HBCD in lakes in the U.K. was used as surrogate data in the absence of data on HBCD in Canadian drinking water. Uncertainties also are associated with the assumptions incorporated into the consumer product scenario model. The mouthing scenario estimates for infants and toddlers were similar to the ones conducted for the TCEP screening assessment (Canada 2009) and appear to be underestimates when compared to the EU derived sub-scenario; however, when the EU RAR value of 84 mg/m² fabric is substituted in the EU scenario, rather than the 2 mg/m² value, the estimates are comparable. An intake derived from a back calculation of blood level was similar to that of the intake from food sources, and a comparable estimate of exposure was derived by Eljarrat et al. (2009) for nursing infants. For these reasons, there is high confidence in the environmental media and consumer product derived estimates of exposure for the general population of Canada and the resulting derived margins of exposure.