Temporal trends in serum polybrominated diphenyl ether concentrations in the Australian population, 2002–2013

Abstract

In Australia, systematic biomonitoring of persistent organic pollutants (POPs) in pooled serum samples stratified by age and sex has occurred every two years between 2002/03 and 2012/13. Multiple regression modeling on log10-transformed serum pool concentrations of BDEs 47, 99, 100 and 153 and on the sum of these (Σ₄PBDE) was conducted to examine trends by sex and time since baseline, stratified by age group. Temporal trends were age- and congener-specific, with the largest changes per year of observation in the 0–4 year old group, with β (SE) = −0.098 (0.013) for log₁₀BDE47; −0.119 (0.012) for log₁₀BDE99; −0.084 (0.014) for log₁₀BDE100, and −0.053 (0.013) for log₁₀BDE153, all p < 0.001. Adults over age 16 showed much smaller decreasing temporal trends for BDE47 and BDE99, no significant changes in BDE100, and, for the oldest age groups, slight increases in BDE153. As a result, Σ₄PBDE concentrations were stable over the entire time period in adults older than 16. Concentrations of each BDE in pools from females aged 31–60 were significantly lower compared to males. Relative proportions of BDE47 declined, while BDE153 accounted for a greater share of Σ₄PBDE over time. Whereas previously we saw a large elevation in the youngest age groups compared to older children and adults, this is no longer the case. This may be due to a decline in infant and toddler exposures in the indoor environment as use of PBDEs in consumer products has been phased out, suggesting temporal changes in the relative sources of exposure for young children in Australia.

1. Introduction

Polybrominated diphenyl ethers (PBDEs) are a class of brominated flame retardants (BFR) added to products to reduce flammability and rate of ignition. In 2005, the importation of raw product penta- and octa-BDE into Australia ceased (NICNAS, 2007) following the inclusion of penta-BDE and octa-BDE into the annex of the Stockholm Convention (Stockholm Convention on POPs, 2010). The commercial PBDE flame retardant mixture deca-BDE is now being considered for listing on the Stockholm Convention (Stockholm Convention on Persistent Organic Pollutants, 2013).

Biomonitoring of PBDEs using human blood serum samples from Australia began in 2002/03 and found unexpectedly high concentrations compared to “legacy” persistent organic pollutants (POPs) such as dioxins and polychlorinated biphenyls (Harden et al., 2007; Toms et al., 2008). This first study included a young age group (< 16 years) with an average age of 11 years. In 2004/05, a 0–4 years age group was also included in the biomonitoring program with an average age of2.4 years. The concentration of BDE-47 in this youngest age group was four times that of the ≥16 years group (Toms et al., 2008), thus demonstrating an inverse relationship between age and concentration, which was in contrast to legacy POPs (Harden et al., 2007). Higher concentrations in children have also been reported elsewhere, including the USA (Fischer et al., 2006; Lunder et al., 2010; Sjodin et al., 2014; Wu et al., 2015), Norway (Thomsen et al., 2002), and Pakistan (Ali et al., 2013). This became a cause for concern since limited data suggest the potential for adverse health effects from PBDE exposure, and young children may be considered to be a population of potentially greater susceptibility (Darnerud, 2003; Grandjean and Landrigan, 2014). Various epidemiological studies have examined potential associations between exposure to PBDEs and diabetes; neurobehavioral and developmental disorders; reproductive health effects; alteration in thyroid function; and adverse cognitive outcomes with early life exposure (Eskenazi et al., 2013; Kim et al., 2014; Chevrier et al., 2016; Jacobson et al., 2016). The higher concentrations of PBDEs compared to other POPs in Australia coupled with the elevated concentrations in children warranted continued biomonitoring to further assess these trends.

The Australian biomonitoring program relies upon collection and analysis of pooled serum samples and allows for the assessment of age, sex and importantly, temporal trends, in POP concentrations (Toms et al., 2012, 2014). While it is expected that decreased exposure will eventually result in decreased body burden, the longevity of products treated with PBDEs combined with lengthy estimated human half-lives of 1.8 years (BDE-47), 2.9 years (BDE-99), 1.6 years (BDE-100) and6.5 years (BDE-153) (Geyer et al., 2004), makes it difficult to anticipate the speed at which concentrations in human tissue will decrease. As well, usage change over time may result in altered PBDE congener profiles in human samples.

The aim of this study is to update the monitoring of PBDEs in human blood serum from samples of the Australian population from previous reports to include pools collected in 2010/11 and 2012/13. The time period of monitoring covers a period of great interest, with the ban on these chemicals (in 2005) going into effect early in the monitoring period. This allows an assessment of the effectiveness of the intervention to eliminate use of these chemicals and in turn decrease body burden. In addition, the data allow an investigation of whether or not congener profiles differ with changes to product usage.

2. Materials and methods

2.1. Sample collection

Sample collection for our human biomonitoring program relies upon collection of pooled samples of de-identified surplus sera from community pathology laboratories stratified by age and sex, as has been reported previously (Heffernan et al., 2014, (Toms et al., 2008, 2009a, b, 2012). In brief and as per our protocol, samples for the two new collection periods (2010/11 and 2012/13) were obtained in collaboration with Sullivan Nicolaides Pathology (SNP) from de-identified surplus pathology samples collected in south-east Queensland, Australia. Samples were stratified by age and sex. Pooled samples were placed into 100 mL solvent rinsed glass bottles. Age groups were as follows: 0–4; 5–15; 16–30, 31–45, 46–60, and ≥60 years. For each pool, 100 individual samples were combined using up to 1 mL of each of the 100 samples. For younger age groups, where less volume was available, < 1 mL was used with consistent volumes for all donors into a specific pool, e.g. 0.5 mL was taken from each sample. Two pools were made for each age and sex group for each time period in the 2010/11 and 2012/13 collection periods. As previously described, greater numbers of pools were collected for selected age groups at some earlier time periods.

When this methodology was first conceived, an innovative technique to assess if bias occurred using pathology samples was undertaken where two pools of pathology samples “sick” were compared to age, sex and postcode matched insurance samples “healthy” and analysed for dioxins. The normalised diference was 10% which could be explained by normal variations in analytical reproducibility (Harden et al. 2004). Therefore, there should not be bias related to the sample collection.

For each of 2010/11 and 2012/13, there were a total of 4800 samples assembled into 24 pools. In total, there are PBDE data from 249 pools created from 17,752 individual samples from 2002/03, 2004/05, 2006/07, 2008/09, 2010/11 and 2012/13. In 2002/03, separate pools were not collected for ages 0–4; rather, the youngest age group was < 16 years, with limited numbers of donor samples from children in the youngest age group. Beginning in 2004/05, pools were collected specifically for the 0 to 4 year age group. It was not possible to determine if any one donor contributed to more than one collection period due to the use of de-identified samples. We maintain ethics approval for this study through The University of Queensland Medical Research Ethics Committee and Queensland University of Technology Ethics Committee. The analysis of pooled samples by investigators at the U.S. Centers for Disease Control and Prevention (CDC) was determined not to constitute engagement in human subject research.

2.2. Chemical analysis

The PBDE measurements were made at the U.S. Centers for Disease Control and Prevention (CDC) in Atlanta and the methodologies have been described previously (Sjodin et al. 2004; Jones et al. 2012). Briefly, a set of samples was defined as 24 unknown samples with three analytical blanks and three quality assurance/quality control (QA/QC) samples. Each set was processed using a semi-automated sample preparation method. Human sera (2 g) were weighed into test tubes and fortified with internal standards (¹³C-labeled) using a 215 Liquid Handler (Gilson Inc., Middleton, WI). Formic acid and water were added to denature proteins and dilute the samples on the liquid handler. The target analytes were extracted into dichloromethane using the solid phase extraction (SPE) workstation (Rapid Trace®, Zymark, Hopkinton, MA) for the 2010/11 pools and by automated liquid liquid extraction (LLE) for the 2012/13 pools. Clean up was performed on a two layered column. The top layer comprised activated silica and the bottom layer comprised silica gel/sulfuric acid (2:1 by weight). The top layer retained polar lipids such as cholesterol, while the bottom layer degraded the remaining lipids to produce an extract free of biogenic material. Samples were evaporated to 1 mL and transferred to the gas chromatograph vials, which were previously spiked with recovery standards. Samples were further evaporated to 10 μL and analysed by gas chromatography high resolution mass spectrometry. A DFS (ThermoFinnigan, Bremen, Germany) instrument was used for the analysis. The chromatographic separations were carried out on an 1310 gas chromatograph (ThermoFinnigan, Bremen, Germany) fitted with a DB5HT capillary column (15 m, 0.25 mm inner diameter, and 0.10 μm thickness). The following congeners were targeted for analysis: BDEs-17, −28, −47, −66, −85, −99, −100, −153, −154 and −183. The results are expressed as ng/g lipid. Three of the 249 pools in this analysis had non-detected concentration for one of the four main congeners. These non-detected pool concentrations were imputed using half the limit of detection (LOD), which was dependant on sample size and blanks. A summary metric (Σ₄PBDE) was calculated for each pool as the sum of BDEs-47, −99, −100 and −153.

It should be noted that pools from all collection periods were analysed at the U.S. CDC with the exception of the 2002/03 and 2004/05 pools which were analysed at Eurofins, Germany (Toms et al., 2008).

2.3. Statistical analysis

Mean and standard error of the mean of pool concentrations by analyte, age group, and sampling cycle were calculated. For each sample collection period, trends in log₁₀-transformed pool concentration as a function of average age of pool contributors were evaluated using linear regression. Trends in log₁₀-transformed pool concentration over years since baseline collection and by sex were examined using multiple regression (STATA IC 12.1, Stata Corp., College Station, TX). The multiple regression was stratified by age group in order to detect potential differences in time trends by age group. Contribution of each of the four major congeners (BDEs 47, 99, 100, and 153) to the sum of those four congeners was examined at different time points to evaluate potential changes in congener profiles over time.

3. Results

3.1. PBDEs in human blood serum

Human serum samples collected during 2010/11 and 2012/13 pooled according to age and sex were analysed for BDEs-17, −28, −47, −66, −85, −99, −100, −153, −154 and −183. BDEs-17, −28, −66, −85, −154 and −183 were detected in < 10% of the pools and are not discussed further. BDEs: 47, 99, 100 and 153 and Σ₄PBDE are the focus of the analyses and results presented here. Similar to previous collection cycles (Toms et al., 2008, 2012), PBDEs were detected in all pooled human blood serum samples collected in 2010/11 and 2012/13 (Table 1). The arithmetic means and standard error of the means of each congener at each time period for each age group are reported in Table 1. Because each pool consists of equal serum volumes from 100 individuals, the measured concentration in that pool represents the arithmetic mean of the concentrations in individuals contributing to that pool, and means across different pools for the same age group are equivalent to the arithmetic mean concentration of all the individuals contributing to all of those pools.

Table 1.

Average pool concentrations (ng/g lipid) and number of pools by sample collection period, age and congener. Trend by age was assessed using linear regression of log₁₀-transformed concentrations as a function of average age of individuals contributing to each pool, with p values < 0.05 indicated with bold type.

Sampling period	Na	Mean (SE), ng/g lipid
		BDE47	BDE99	BDE100	BDE153	Σ4PBDE
2002/03
Ages 0–4		NC	NC	NC	NC	NC
Ages 0–15	12	11.01 (0.67)	4.88 (0.22)	2.91 (0.12)	4.24 (0.24)	23.04 (1.02)
Ages 16–30	16	6.14 (0.53)	2.51 (0.21)	1.54 (0.16)	2.61 (0.21)	12.8 (0.96)
Ages 31–45	20	4.83 (0.37)	2.21 (0.18)	1.11 (0.08)	2.03 (0.13)	10.19 (0.68)
Ages 46–60	12	3.91 (0.31)	1.33 (0.15)	0.94 (0.07)	1.74 (0.16)	7.91 (0.64)
Ages > 60	12	3.82 (0.78)	1.39 (0.26)	0.89 (0.14)	1.58 (0.13)	7.65 (1.27)
p for trend:		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001
2004/05
Ages 0–4	4	24 (3.7)	8.5 (0.99)	6.85 (1.02)	6.3 (0.94)	45.65 (6.57)
Ages 5–15	4	7.55 (1.01)	2.68 (0.38)	2.3 (0.29)	5.25 (1.04)	17.78 (2.29)
Ages 16–30	4	4.48 (0.31)	1.55 (0.13)	1.25 (0.12)	2.45 (0.5)	9.73 (0.94)
Ages 31 −45	4	3.73 (0.35)	1.35 (0.22)	1.09 (0.08)	2.13 (0.18)	8.29 (0.76)
Ages 46–60	4	3.73 (0.37)	1.15 (0.1)	1.1 (0.15)	2.03 (0.25)	8 (0.8)
Ages > 60	4	2.85 (0.49)	0.85 (0.14)	0.73 (0.07)	1.5 (0.2)	5.93 (0.89)
p for trend:		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001
2006/07
Ages 0–4	32	19.21 (1.61)	7.51 (0.65)	5.03 (0.44)	4.35 (0.35)	36.1 (2.91)
Ages 5–15	32	10.29 (0.94)	3.53 (0.37)	3.24 (0.34)	6.04 (0.42)	23.09 (1.9)
Ages 16–30	4	19.65 (9.19)	5.78 (2.98)	4.53 (1.89)	6.53 (2.16)	36.48 (15.02)
Ages 31 −45	4	4.78 (0.35)	1.45 (0.2)	1.2 (0.09)	2.45 (0.21)	9.88 (0.8)
Ages 46–60	4	5.13 (1.06)	1.28 (0.25)	1.2 (0.04)	1.88 (0.3)	9.48 (1.23)
Ages > 60	4	3.93 (0.9)	1.4 (0.28)	1.05 (0.18)	2.33 (0.38)	8.7 (1.58)
p for trend:		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001
2008/09
Ages 0–4	4	11.65 (0.75)	5.3 (0.81)	3.53 (0.17)	3.5 (0.19)	23.98 (1.4)
Ages 5–15	4	5.93 (0.5)	2.33 (0.14)	1.93 (0.21)	5.1 (0.4)	15.28 (1.07)
Ages 16–30	4	4.73 (0.4)	1.78 (0.25)	1.48 (0.24)	2.9 (0.3)	10.88 (1.04)
Ages 31 −45	4	4.63 (0.62)	1.58 (0.29)	1.2 (0.16)	2.88 (0.39)	10.28 (1.27)
Ages 46–60	4	3.5 (0.23)	1.35 (0.03)	1.13 (0.03)	2.45 (0.21)	8.43 (0.39)
Ages > 60	4	3.38 (0.3)	1.3 (0.06)	0.95 (0.1)	2.1 (0.22)	7.73 (0.61)
p for trend:		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001
2010/11
Ages 0–4	4	6.33 (0.36)	2.3 (0.12)	1.9 (0.11)	2.88 (0.48)	13.4 (0.99)
Ages 5–15	4	5.75 (0.55)	1.9 (0.17)	1.9 (0.14)	4.78 (0.53)	14.33 (1.33)
Ages 16–30	4	3.75 (0.25)	1.15 (0.06)	1.08 (0.03)	2.88 (0.23)	8.85 (0.53)
Ages 31 −45	4	3.38 (0.23)	1 (0.04)	1.05 (0.06)	2.85 (0.28)	8.28 (0.53)
Ages 46–60	4	2.78 (0.31)	0.8 (0.07)	0.88 (0.11)	2.68 (0.31)	7.13 (0.67)
Ages > 60	4	3.25 (0.6)	0.83 (0.09)	0.88 (0.13)	2.35 (0.19)	7.3 (0.9)
p for trend:		< 0.001	< 0.001	< 0.001	0.019	< 0.001
2012/13
Ages 0–4	4	4.35 (0.29)	1.03 (0.05)	1.43 (0.1)	1.88 (0.09)	8.68 (0.45)
Ages 5–15	4	4.35 (0.29)	1.1 (0.07)	1.38 (0.14)	3.2 (0.21)	10.03 (0.64)
Ages 16–30	4	4.48 (0.52)	1.18 (0.15)	1.35 (0.26)	3.03 (0.23)	10.03 (0.94)
Ages 31 −45	4	3.6 (0.71)	1.15 (0.32)	0.73 (0.08)	2.63 (0.48)	8.1 (1.55)
Ages 46–60	4	3.08 (0.37)	0.8 (0.07)	0.75 (0.03)	3.13 (0.58)	7.75 (0.97)
Ages > 60	4	3.15 (0.79)	0.88 (0.25)	0.63 (0.13)	2.35 (0.36)	7 (1.51)
p for trend:		0.005	0.058	< 0.001	0.909	0.020

^aN: number of pools; NC: not collected. For this collection period, samples from ages 0 to 15 were composited. Please note for 2002/03 there was only one young age group 0–15 years compared to two for all other collection periods (0–5 and 5–15 years).

Within each of the first 5 sample collection periods (2002–2011), there was a statistically significant inverse trend with age for each of the congeners and for Σ₄PBDE. For the most recent sample collection period, the concentrations of BDE-100 and BDE-153 were no longer significantly inversely associated with age. In general, a marked decrease in the degree of difference between concentrations in the youngest age groups compared to adults was observed for all congeners (Table 1, Fig. 1).

Fig. 1.

Arithmetic means and standard errors (SE) for Σ₄PBDE by age group and sample collection periods. NC = not collected; no pools for ages 0 to 4 were collected in 2002/2003, although some samples from this age group may have been included in the 0 to 15 age group, which was the youngest age group collected for that collection period.

3.2. Temporal trends

The addition of 2 cycles of data (2010/11 and 2012/13) to the previously-reported data from 2002/03 to 2009/10 (Toms et al., 2009a, b; Toms et al. 2012) allows assessment of temporal trends over more than a decade of sampling (Table 2). In multiple regression, concentrations of all four major congeners as well as Σ₄PBDE were inversely associated with years since baseline regression in the youngest age group, with nearly a 10-fold decline in Σ₄PBDE over the 10 years of sampling for this age group (β = 0.092, 95% confidence interval [CI] −0.116, −0.068). The strongest declines in this age group were observed for BDE-99, a pentabromodiphenyl ether, followed by BDE-47, a tetrabromodiphenyl ether. Statistically significant declines in these two congeners were observed in most age groups, while results for BDE-100 and BDE-153 were more mixed. BDE-100, a pentabromodiphenyl ether, declined with years since baseline in all age groups, however, the decline was statistically significant only in the youngest two age groups. In contrast, for the hexabromodiphenyl ether BDE-153, trends differed by age. Although the youngest age group showed statistically significant declines over the sampling period, concentrations in the pools collected from individuals ages 31 and older showed small but statistically significant increases over this time period. For the Σ 4PBDE analysis, only the two youngest age groups showed statistically significant decreases over the time period of observation (Table 2). Older age groups showed no significant trend with time. As a result of the dramatic decline in concentrations observed in the youngest age group, concentrations in the 5–15 year age groups exceeded those in the youngest age groups for the most recent two time periods, and the previously remarkable age differences from young to old have diminished, although not disappeared entirely (Fig. 1). Results from the 2006/07 collection period show higher concentrations in two pools in the 16–30 year age groups, one male and one female. The reason for these elevated concentrations is unknown but may be related to one or some individuals contributing to the pool elevating the overall concentration within the pool.

Table 2.

Multiple regression results examining associations between log₁₀-transformed concentrations of BDEs-47, −99, −100, −153, and Σ₄BDEs with sex and years since baseline, stratified by age group. Statistically significant coefficients (p < 0.05) are bolded.

Congener	Sex (female vs. male)		Years since baseline
	β (SE)	p	β (SE)	p
log10BDE47
Ages 0–4	0.009 (0.055)	0.873	−0.098 (0.013)	< 0.001
Ages 5–15	0.005 (0.043)	0.902	− 0.036 (0.008)	< 0.001
Ages 16–30	−0.005 (0.076)	0.943	−0.013 (0.01)	0.209
Ages 31–45	−0.117 (0.038)	0.004	−0.011 (0.005)	0.039
Ages 46–60	−0.1 (0.039)	0.015	−0.012 (0.005)	0.035
Ages > 60	− 0.123 (0.063)	0.060	−0.002 (0.009)	0.814
log10BDE99
Ages 0–4	0.011 (0.049)	0.820	−0.119 (0.012)	< 0.001
Ages 5–15	− 0.004 (0.045)	0.934	− 0.056 (0.008)	< 0.001
Ages 16–30	0.009 (0.07)	0.896	−0.03 (0.01)	0.003
Ages 31 –45	−0.127 (0.044)	0.006	− 0.032 (0.006)	< 0.001
Ages 46–60	−0.139 (0.056)	0.018	−0.016 (0.008)	0.047
Ages > 60	− 0.085 (0.07)	0.232	−0.013 (0.009)	0.166
log10BDE100
Ages 0–4	0 (0.058)	0.999	−0.084 (0.014)	< 0.001
Ages 5–15	0.009 (0.047)	0.844	− 0.028 (0.009)	0.003
Ages 16–30	− 0.006 (0.073)	0.935	−0.005 (0.01)	0.609
Ages 31–45	− 0.085 (0.036)	0.023	−0.008 (0.005)	0.100
Ages 46–60	− 0.069 (0.036)	0.066	−0.005 (0.005)	0.326
Ages > 60	−0.107 (0.052)	0.049	−0.005 (0.007)	0.476
log10BDE153
Ages 0–4	− 0.026 (0.053)	0.621	−0.053 (0.013)	< 0.001
Ages 5–15	−0.11 (0.038)	0.005	−0.002 (0.007)	0.789
Ages 16–30	− 0.216 (0.051)	< 0.001	0.013 (0.007)	0.063
Ages 31–45	−0.187 (0.021)	< 0.001	0.016 (0.003)	< 0.001
Ages 46–60	−0.184 (0.028)	< 0.001	0.025 (0.004)	< 0.001
Ages > 60	−0.152 (0.03)	< 0.001	0.021 (0.004)	< 0.001
log10Σ4PBDEs
Ages 0–4	0.003 (0.049)	0.953	−0.092 (0.012)	< 0.001
Ages 5–15	− 0.028 (0.039)	0.477	− 0.03 (0.007)	< 0.001
Ages 16–30	− 0.05 (0.068)	0.467	−0.008 (0.009)	0.368
Ages 31–45	−0.13 (0.031)	< 0.001	−0.007 (0.004)	0.086
Ages 46–60	−0.125 (0.03)	< 0.001	−0.001 (0.004)	0.833
Ages > 60	−0.119 (0.051)	0.028	0.002 (0.007)	0.784

3.3. Trends by sex

Concentrations of each congener and Σ₄PBDE were nearly always significantly lower in adult females (ages 31–60) than in males (Table 2). Statistically significant regression coefficients on the log_10- transformed concentrations were generally in the range of just less than −0.1 to −0.2, corresponding to approximately 20 to 50% lower concentrations in women of these age groups compared to men at the same ages (Table 2). Fig. 2 illustrates these differences for Σ₄PBDE in pools from males and females ages 31 to 45 across all time periods. In contrast, concentrations of the measured congeners in the youngest children did not differ by sex. For persons aged 5 to 15 and aged 16 to 30, concentrations did not differ by sex except for BDE-153, where again females exhibited lower concentrations than males.

Fig. 2.

Arithmetic mean (SE) Σ₄PBDE in males and females ages 31–45 across all sampling periods.

3.4. Relative contributions of key congeners

The average contribution of each of the four congeners to Σ₄PBDE was assessed across pools in the youngest age group and in pools from all age groups 16 and older, and the relative contributions were compared at the 2004/05 and 2012/13 collection time periods. BDE-47 consistently contributed the most to the sum of the four dominant BDE congeners, followed by BDE-153. Over time from the 2004/05 collection period to the most recent collection reported here, the proportion of BDE-47 decreased slightly in the youngest age group and in persons ages 16 and over. BDE-99 also decreased somewhat in both age groups. The relative proportion of BDE-153 increased in both age groups (Fig. 3).

Fig. 3.

Relative proportions of four BDEs for children ages 0 to 4 and for adults ages 16 and older at two time periods.

4. Discussion

The pooled serum analyses reported here represent more than a decade of systematic biomonitoring of serum PBDE concentrations in the Australian general population. The time period covered includes significant regulatory changes affecting the use of PBDE compounds. The Australian National Industrial Chemicals Notification and Assessment Scheme (NICNAS) found that importation of raw chemical penta- and octa-BDEs into Australia ceased in 2005, and that no manufacture of these chemicals occurred in Australia. Importation and manufacture of these two groups of BDEs were formally banned in 2007. While data on amounts of PBDEs in products imported into Australia currently or historically is not available, a study of BDEs in products in Australian homes found some detectable concentrations of octa-BDEs and deca-BDEs indicating that sources of exposure are still present (Gallen et al. 2014). Over the same time period, selected classes of PBDEs were identified under the Stockholm Convention as persistent organic pollutants, and recently the production and use of these compounds has been severely restricted. While the turnover rate of furniture and electronic and electrical goods from PBDE-treated to non PBDE-treated products is unknown, PBDEs are detected in landfill leachate from South East Queensland, Australia. Gallen et al. (2016) found BDE-47 and BDE-99 were the PBDE congeners detected in the highest concentrations in the leachate collected from four of six open landfills sampled. This indicates that PBDE-treated products are being disposed of into landfill (Gallen et al. 2016) and assumedly replaced with non PBDE-treated ones.

The results of analysis of pooled sera for PBDEs presented in this study demonstrates that there have been dramatic decreases in the concentrations of four major PBDE congeners (BDEs 47, 99, 100, and 153) in serum pools collected from children ages 0 to 4 over the period from 2004/05 to the most recent collection period of 2012/13. The rate of decrease in PBDE concentrations in serum pools from young children is greater than that observed in adults. For some ages and congeners (notably, BDE-153 in adults) no decreases over this time period have been observed. BDE-153 has a relatively long estimated half-life of elimination in humans compared to lower brominated PBDEs, with an estimated half-life of 6.5 years compared to 1.8 years for BDE-47 in humans (Geyer et al. 2004). It has been suggested that human liver microsomes metabolise BDE-47 and −99 but not −153 (Lupton et al. 2009). Thus, even if use or exposure to BDE-153 in the environment has decreased, serum levels of BDE-153 might change more slowly than more metabolically and environmentally labile BDE compounds.

There are limited PBDE temporal trend data with which to make comparisons and different usage patterns across various countries means comparisons should be made with caution. Temporal trend data from two Californian studies of adults demonstrate a decrease in concentrations of BDEs-47, −99, −100 and −153 up to 2011/12 (Parry et al. 2018) but followed by a plateau and slight increase from 2011 to 2015 (Hurley et al. 2017).

Studies have suggested that human exposure to PBDEs occurs through various pathways including dust contact and ingestion, inhalation, and dietary exposure (Lorber 2008; Harrad et al. 2010; Schecter et al. 2010). There is limited data on PBDE intake via food in Australia but dietary intake of PBDEs is documented sources for adults (Domingo 2012). In addition, for young children, potential exposure pathways and sources influencing serum concentrations include placental transfer (Zhao et al. 2013), ingestion of human milk (Toms et al. 2007; Abdallah and Harrad 2014) and child-specific behaviours such as mouthing and elevated ingestion of dust or soils (Hoffman et al. 2016). Thus, for young children, exposure levels are influenced by the general environment (dust, indoor air, food, and presence in mouthed or handled objects) and also by maternally mediated exposure via placental transfer and ingestion of human milk (Toms et al. 2012; Shin 2018).

Evaluation of temporal trends of the key congeners in young children compared to trends in females of reproductive age provides some insight into the relative contribution of the maternally mediated exposures vs. environmental exposures for early life exposure. Fig. 4 shows the trends in BDE-47 and BDE-153 for children ages 0 to 4 and females in the age groups 16 to 30 and 31 to 45 combined. While the children show strong declines over the time period of monitoring, concentrations in females of reproductive age for these two congeners remained relatively constant. This suggests that the declines observed in young children are not the result of declining maternally mediated exposure via placental transfer or breast milk, but may rather reflect lower exposures over time from the environment and/or food ingestion. PBDEs were detected in all samples of house dust (n = 10) in Brisbane Australia in 2007 with the mean BDE-47 at 91 ng/g dust (Toms et al. 2009a, b). A recent study of house dust (n = 12) in Melbourne, Australia in 2016 found the mean-47 and BDE-209 concentrations to be 9 ng/g dust (McGrath et al. 2018). While comparisons must be made with caution due to small sample sizes and different geographical locations, this does indicate a 10 fold decrease for BDE-47 in household dust which may influence exposure via indoor environments. Assessment of PBDEs in baby foods in Australia in 2015 suggests a decrease in concentration from a previous time point of 2004 but again, comparisons must be made with caution due to small sample sizes (Toms et al. 2016). Children born in the most recent time periods appear to have lower environmental exposures than those born early in the monitoring period. The observed decreases in PBDE concentrations in the youngest age group of children who were born post-PBDE ban (after 2005) may reflect a reduction in exposure related to the banning and subsequent removal of these PBDEs and products made with them from the market.

Fig. 4.

Temporal trends in average pool concentrations for children ages 0 to 4 in comparison to pools from females of reproductive age (16 to 45) for A) BDE-47, and B) BDE-153. Bars show arithmetic mean and standard error by sample collection period. Average pool concentration for 2006/07 for BDE-47 for females of reproductive age omits one pool with an outlier concentration(45.3 ng/g lipid).

Furthermore, unknown sources and/or pathways may have been contributing to children’s PBDE intake as pharmaco-kinetic modeling of the PBDE biomonitoring data measured between 2002–2003 and 2010–2011 showed that currently known pathways do not sufficiently explain measured PBDE concentrations in children (Gyalpo et al. 2015).

Whereas previously we saw a large elevation in the youngest age groups compared to older children and adults, this is no longer the case, and if these trends continue, we expect that in subsequent monitoring the youngest age groups will have concentrations similar to maternal concentrations. If this pattern holds, this may be indicative of temporal changes in the relative sources of exposure for young children in Australia. This suggests that placental transfer, human milk, and general exposures in the food supply will become the major sources of exposure for infants and young children, while indoor environments will contribute relatively lower amounts than in the early 2000s when the biomonitoring began. Further investigation of PBDE sources is warranted.

Sex differences in PBDE concentrations have been reported previously, with higher PBDE concentrations in some studies (Meneses et al. 1999; Schroter-Kermani et al. 2000; Thomsen et al. 2002; Takasuga et al. 2004; Bjermo et al. 2017; Gravel et al. 2018; Kim et al. 2018) while other studies have reported either no difference or higher concentrations in females (Schecter et al. 2005; Gomara et al. 2007; Abou-Elwafa Abdallah et al. 2017; Ma et al. 2017). The observation in the current study that women ages 31 to 60 have consistently and significantly lower serum PBDE concentrations than men may result from one or more factors. Women in this age group may have borne children, and transfer of PBDEs to the foetus via placental transfer or via breastfeeding may result in lower remaining body burdens compared to men of the same age (Vizcaino et al. 2014). Interestingly, studies have not found evidence of depuration of PBDEs during the duration of breastfeeding (LaKind et al. 2009; Marchitti et al. 2017). Another factor that could influence the relative concentrations of PBDEs in serum might be differences in body composition between women and men. Since PBDEs are lipophilic, they distribute in lipid tissue throughout the body. Women have higher body fat levels compared to men. This could result in a larger volume of distribution for the same ingested dose, resulting in a lower lipid-adjusted concentration in serum (Porta et al. 2012). Finally, it is possible that women experience lower average doses or have more efficient elimination than men, although there is no evidence to support either of these potential explanations. It is not possible from the data collected in this study to determine which of these factors are responsible for the observed differences between women and men in these age groups.

Several limitations and considerations apply to the evaluation of the data presented here. The benefits and limitations of reliance on serum pools have been previously discussed (Heffernan et al. 2014). In particular, little information on inter-individual variation is available. For compounds with strong age trends, the grouping of pools by age ranges could bias results if the average ages of the individuals contributing to the pools are not relatively consistent. The mean ages for pooled samples representing a given age group were generally similar across the different collection periods. For example, mean age ranged from 2.2 to2.9 years in the 0–4 age group and 11 to 12.3 years in the 5–15 age group. This suggests that differences in the ages of individuals contributing to the serum pools did not systematically affect the mean PBDE concentrations.

5. Conclusion

Ongoing temporal monitoring has revealed the lowest concentrations of PBDEs detected in young children in Australia to date. However, temporal trends in other age groups are less consistent. In adults, significant declines over time are observed for BDE-47 and BDE-99, but no trend in BDE-100 was observed, and concentrations of BDE-153 increased. As a result, over the time period of monitoring reported here, concentrations of Σ₄PBDE in adults were stable. The dramatic differences in concentrations between young children and adults observed in the early monitoring cycles have diminished, and the most recent time periods show only modest inverse trends in concentrations with age for BDE-47 and BDE-100, and no inverse trends with age for BDE-99 and BDE-153.

This systematic monitoring based on pooled sampling allows for assessment of age, sex and temporal trends using an efficient and cost-effective collection method. While this type of sample collection and monitoring is not longitudinal in using the same participants over time, it provides a time series of cross-sectional datasets that allow evaluation of key aspects of changing exposures representative of the general population. Due to the smaller age ranges, changes in concentrations and profiles become apparent much sooner in the pools from children than from the adults. This work also demonstrates the value of including samples from children in biomonitoring programs, despite difficulties in obtaining the samples.