A review on the analytical procedures of halogenated flame retardants by gas chromatography coupled with single quadrupole mass spectrometry and their levels in human samples

Abstract

Halogenated flame retardants (HFRs) market is continuously evolving and have moved from the extensive use of polybrominated diphenyl ether (PBDE) to more recent introduced mixtures such as Firemaster 550, Firemaster 680, DP-25, DP-35, and DP-515. These substitutes are mainly composed of non-PBDEs HFRs such as 2-ethyl-hexyl tetrabromobenzoate (TBB), bis(2-ethylhexyl) tetrabromophthalate (TBPH), 1,2-bis-(2,4,6-tribromophenoxy) ethane (BTBPE) and decabromodiphenyl ethane (DBDPE). Other HFRs commonly being monitored include Dechlorane Plus (DP), Dechlorane 602 (Dec602), Dechlorane 603 (Dec603), Dechlorane 604 (Dec604), 5,6-dibromo-1,10,11,12,13,13-hexachloro- 11-tricyclo[8.2.1.02,9]tridecane (HCDBCO) and 4,5,6,7-tetrabromo-1,1,3-trimethyl-3-(2,3,4,5-tetrabromophenyl)-2,3-dihydro-1H-indene (OBTMPI). This review aims at highlighting the advances in the past decade (2010-2020) on both the analytical procedures of HFRs in human bio-specimens using gas chromatography coupled with single quadrupole mass spectrometry and synthesizing the information on the levels of these HFRs in human samples. Human specimen included in this review are blood, milk, stool/meconium, hair and nail. The review summarizes the analytical methods, including extraction and clean-up techniques, used for measuring HFRs in biological samples, which are largely adopted from those for analysing PBDEs. In addition, new challenges in the analysis to include both PBDEs and a wide range of other HFRs are also discussed in this review. Review of the levels of HFRs in human samples shows that PBDEs are still the most predominant HFRs in many cases, followed by DP. However, emerging HFRs are also being detected in human despite of the fact that both their detection frequencies and levels are lower than PBDEs and DP. It is clearly demonstrated in this review that people working in the industry or living close to the industrial areas have higher HFR levels in their bodies.

Capsule:

Review of latest advances in analytical methods and synthesis of concentration levels of HFRs in human samples

Graphical Abstract

1. Introduction

Production and use of flame retardants (FRs) has vastly expanded in the last several decades. FRs can be found in a wide range of materials and products ranging from home furniture and construction materials to electronic products (Fayiga and Ipinmoroti, 2017). Halogenated flame retardants (HFRs) have a significant share of the total market of FRs due to their efficiency in curbing fires, large spectrum of use and low cost (Wabeeke 2002). Extensive use of HFRs contributes to their ubiquitous presence in the environment and in biota (Fayiga and Ipinmoroti, 2017; Zhang et al., 2016), which causes concerns for possible human exposure to these compounds.

Polybrominated diphenyl ethers (PBDEs) is a group of the mostly produced and widely used HFRs. They were developed in the 1970’s and commercialised in the form of three mixtures: Penta-BDE, Octa-BDE and Deca-BDE, each containing diverse degrees of bromine-substituted congeners (Guardia et al., 2006). PBDEs have been gradually phased out due to their adverse effects on human health and the environment (Xiong et al., 2019). Penta-BDE and Octa-BDE were banned in 2004 in USA and in 2006 in Europe following the 2004 regulations of the Stockholm Convention on persistent organic compounds (POPs). Only Deca-BDE remains in limited use in these two jurisdictions (U.S. Environmental Protection Agency, 2009). Mirex, a chlorinated chemical sold as flame retardant under the trade name Dechlorane, was banned in North America in the 1970’s.

As a result of strict regulation on these HFRs, new alternative HFRs have been introduced into the market including, 2-ethyl-hexyl tetrabromobenzoate (TBB) and bis(2-ethylhexyl) tetrabromophthalate (TBPH) in Firemaster 550 that is designed to replace Penta-BDE (Stapleton et al., 2008), and 1,2-bis-(2,4,6-tribromophenoxy) ethane (BTBPE) in Firemaster 680 that is designed to replace Octa-BDE (Zhou et al., 2014a). Replacement for Mirex includes 1,2,3,4,6,7,8,9,10,10,11,11-Dodecachloro-1,4,4a,5a,6,9,9a,9b-octahydro-1,4:6,9-dimethanodibenzofuran (Dec602), 1,2,3,4,5,6,7,8,12,12,13,13-Dodecachloro-1,4,4a,5,8,8a,9,9a,10,10a-decahydro-1,4:5,8:9,10-trimethanoanthracene (Dec603), 1,2,3,4,7,7-hexachloro-5-(2,3,5,6-tetrabromophenyl)bicyclo[2.2.1]hept-2-ene (Dec604) and 1,2,3,4,7,8,9,10,13,13,14,14-dodecachloro-1,4,4a,5,6,6a,7,10,10a,11,12,12a-dodecahydro-1,4:7,10-dimethanodibenzo[a,e]cyclooctene (Dechlorane Plus) (Shen et al., 2011; Zhou et al., 2014b). Dechlorane Plus is marketed as DP-25, DP-35, and DP-515 mixtures (Occidental Chemical Corporation, 2007) and contains syn- and anti- isomers (s-DP and t-DP) with a proportion of t-DP in the commercial mixture around 60-80% (Xian et al., 2011). Finally, 5,6-Dibromo-1,10,11,12,13,13-hexachloro- 11-tricyclo[8.2.1.02,9]tridecane (HCDBCO) was introduced in 1975 as a flame retardants containing both chloride and bromide but its usage remains largely unknown (Zhou et al., 2014b). Other halogenated flame retardants include Pentabromoethyl benzene (PBEB), hexabromobenzene (HBB), pentabromotolune (PBT), bromophenols (BPs), bromotoluenes (BTs), bromoanilines (BAs) and halogenated organophospohorus (OPFR) were also reported (Cristale et al., 2012) but won’t be the focus of this review.

PBDEs are suspected to be associated with changes in thyroid hormone levels (namely triidothryne, thyroxine and thyroid stimulating hormone) due to their potential role in the thyroid hormone homeostasis (Guo et al., 2018a; Huang et al., 2014). PBDEs may also affect physical and mental neurodevelopment in children (Herbstman et al., 2010) and their behaviour regulation and emotional impulse control (Vuong et al., 2017). More recently, association between PBDE in mothers’ breast milk and gut bacteria composition and metabolites in their newborn infants (Iszatt et al., 2019), and association of prenatal exposure to PBDEs with long-term gut microbiome structure in children (Laue et al., 2019) were also reported.

Toxicological studies for many emerging HFRs are still under development. Some studies have suggested that effects of some emerging HFRs may be similar to PBDEs including disruption of thyroid hormone. A positive correlation between TBPH and BTBPE concentration in household dust and total triiodothyronine level in occupants was reported (Johnson et al., 2013). Decabromodiphenyl ethane (DBDPE) levels in blood were negatively correlated to total triiodothyronine in humans (Zhao et al., 2021).

HFRs are highly lipophilic in general. Human exposure to HFRs occurs mostly through ingestion of foods and dusts (Ni et al., 2012; Bramwell et al., 2017) and unabsorbed HFRs are largely excreted through faeces. Bio-specimens used in human biomonitoring of HFRs include blood, breast milk, stool, hair and nails. Meconium from new-borns is also used. Such monitoring is essential for the adequate estimation of internal dose or body burden of HFRs in humans. It is critical to have analytical methods allowing for the efficient detection and quantification of both PBDEs and other HFRs.

Both gas chromatography (GC) and liquid chromatography (LC) have been employed for the separation of HFRs, depending on the chemicals to be measured. PBDEs are, for example, measured using GC methods (Abdallah 2014). GC is also used for HFRs reviewed in this paper such as TBB, TBPH, BTBPE, DBDPE and Dechlorane-like compounds including DP, Dec602, Dec603, Dec604 and HCDBCO. LC on the other hand is used for compounds whose isometry may be lost due to high GC temperature like 1,2,5,6,9,10-hexabromocyclododécane (HBCD) (Webster et al., 2009a) and for compounds with low volatility such as 4,4′-(propane-2,2-diyl)bis(2,6-dibromophenol) (TBBPA), which need to be derivatized if analysed using GC (Abdallah 2014).

Mass spectrometry (MS) is the most widely used detection technique for HFRs. GC is commonly coupled with single quadrupole mass spectrometry and electron capture negative ionization (ECNI) is the most commonly ionization mode found in routine analysis (Lin et al., 2015; Zhou et al., 2014a, 2014b). In recent years ECNI selectivity issues have pushed research into alternatives such as tandem mass spectrometry (MS/MS) especially for dechlorane like compounds (Wang et al., 2018; Neugebauer et al., 2017; Sales et al., 2017; Portoles et al., 2015; Cristale and Lacorte, 2013). The primary focus of this review will be on single quadrupole mass spectrometry.

2. Scope of the review

This review covers literature published between 2010 and 2020 containing studies that have a sample size of 10 or more. HFRs included in this review are shown in Scheme 1, along with their structures. Only GC method-based coupled with single quadrupole mass spectrometry studies are included in this review. Human bio-specimens of interest in this review are (1) blood, (2) milk, (3) stool/meconium, (4) hair and (5) nail. Review on measurement methods includes both sample preparation and instrumental conditions. With increased number of HFRs being added to the existing methods that were developed for measuring PBDEs, several analytical challenges, notably co-elution of legacy and emerging HFRs (Abdallah 2014), loss of some thermally labile HFRs at high temperature and during sample extraction/clean-up, have emerged. These challenges will be discussed in this review as well. During revision of this article, a review paper on the methodologies (both sample preparation and instrument analysis) for halogenated organic contaminants in general was published (Ayala-Cabrera et al., 2021). The newly published review paper is a good resource for the additional information about the methods that can be applied to the measurements of halogenated FRs.

Scheme 1:

Structures of halogenated flame retardants included in this review. BDE: brominated diphenyl ether; OBTMPI: 4,5,6,7-tetrabromo-1,1,3-trimethyl-3-(2,3,4,5-tetrabromophenyl)-2,3-dihydro-1H-indene; TBB: 2-ethyl-hexyl tetrabromobenzoate; TBPH: bis(2-ethylhexyl) tetrabromophthalate; DBDPE: decabromodiphenyl ethane; BTBPE: 1,2-bis-(2,4,6-tribromophenoxy) ethane; HCDBCO: 5,6-dibromo-1,10,11,12,13,13-hexachloro- 11-tricyclo[8.2.1.02,9]tridecane; Dec602: 1,2,3,4,6,7,8,9,10,10,11,11-dodecachloro-1,4,4a,5a,6,9,9a,9b-octahydro-1,4:6,9-Dimethanodibenzofuran; Dec603: 1,2,3,4,5,6,7,8,12,12,13,13-dodecachloro-1,4,4a,5,8,8a,9,9a,10,10a-decahydro-1,4:5,8:9,10-trimethanoanthracene; Dec604: 1,2,3,4,7,7-Hexachloro-5-(tetrabromophenyl)bicyclo[2.2.1]hept-2-ene; Dechlorane Plus: 1,2,3,4,7,8,9,10,13,13,14,14-dodecachloro-1,4,4a,5,6,6a,7,10,10a,11,12,12a-dodecahydro-1,4:7,10-dimethanodibenzo[a,e]cyclooctene

Concentrations levels of HFRs, especially the emerging HFRs, in humans in the last decade (2010-2020) are also summarized in this review. Concentrations of PBDEs in blood in a global perspective up to 2014 (Fromme et al., 2016a) and in China up to 2018 (Jiang et al., 2019) have been reviewed elsewhere. These two review papers focused on several commonly reported PBDEs congeners such as BDE47, BDE99, BDE100, BDE153 and BDE209. This review therefore focuses only on the new information on PBDEs from Mexico (Eskenazi et al., 2011; Ochoa-Martinez et al., 2016; Orta-Garcia et al., 2018), USA (Eskenazi et al., 2012) and Europe (Brasseur et al., 2014; Knudsen et al., 2017) that were not included in the above-mentioned two reviews. DP data up to 2010 have also been reviewed (Xian et al., 2011). New information about DP has since become available and is included in this review, along with other Dechlorane-like compounds such as Dec602, Dec603, Dec604 and HCDBCO, which have not been reviewed so far. In addition, environmental occurrence, fate, and toxicity of several novel brominated FRs have been reviewed by Xiong et al. (2019). Some of the studies mentioned in Xiong et al. (2019) are included in this review in more detail. Concentrations levels are given in ng/g lipid weight (lw) for blood and milk samples and in ng/g dry weight (dw) for hair and nail samples in this review, unless otherwise stated. However, both ng/g lw and ng/g dw are reported for stool/meconium samples.

3. Measurement methods for halogenated flame retardants in biological samples

3.1. Sample preparation

a. Extraction of HFRs from blood

Both serum and plasma have been used for biomonitoring of HFRs. Initial protein denaturation is generally accomplished by using formic acid, while hydrochloric acid mixed with 2-propanol is also used (Herbstman et al., 2010; Hovander et al., 2000; Ben et al., 2013; He et al., 2012). In general, mixture of hydrochloric acid and propanol is used prior to liquid-liquid extraction (LLE) while formic acid is used prior or to solid phase extraction, but no study was found comparing the performance of the two treatments.

SPE is the most frequently used technique to extract HFRs from blood. Commercially available SPE cartridge of Oasis HLB is widely used (Cequier et al., 2015; Butt et al., 2016). Other studies also reported using commercial SPE cartridges of Isolute ENV+ (Karlsson et al., 2007) and Isolute C18 (Brasseur et al., 2014). For PBDEs, a better recovery was observed when using Oasis HLB compared to Isolute ENV+ as the affinity of the latter might be too strong for PBDEs (Covaci and Voorpoels 2005). After the analytes are loaded on the cartridge, it is critical to remove the water in the cartridge by either washing the cartridge with a polar solvent such as methanol followed by vacuum drying the cartridge (Cequier et al., 2015) or by only vacuum drying the cartridge (Butt et al., 2016; Kim et al., 2016), under nitrogen gas to prevent sucking in lab air during drying. Subsequent elution of HFRs from the cartridge is done using a non-polar or slightly polar solvent (e.g., hexane, dichloromethane (DCM)) or a mixture of them (Cequier et al., 2015).

Liquid/liquid extraction (LLE) is also commonly used for the extraction of HFRs from serum. A wide range of solvent or solvent mixture have been used in LLE. Solvents used in LLE, include hexane/methyl tetrabutyl ether (MTBE) (Lu et al., 2017; He et al., 2012), methanol (Bjermo et al., 2017), acetonitrile (Drage et al., 2019a), DCM:ethyl acetate (Butt et al., 2016). Emulsion however, can form in LLE, making the phase separation difficult and tedious (Cequier et al., 2013).

Both SPE and LLE extraction methods result in similar recoveries for PBDE congeners, except for BDE209, which has a better recovery in SPE (Cequier et al., 2013; Lu et al., 2017). SPE have also been reported to separate HFRs from matrix interferent more efficiently (Miao et al., 2015) than LLE does. Additionally, SPE procedure can be automated while LLE is mostly handled manually. Lipid content is an essential information for the normalization of concentrations of HFRs in blood, milk and stool/meconium samples. A setback of the SPE and QuEChERS (described below) approach is the incompatibility with lipids determination and requires another aliquot of the same samples for the determination of lipid content in the sample (Cechova et al., 2017a).

QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Save) has been explored as an alternative method to extract HFRs from blood (Gao et al., 2016; Wang et al., 2018). The QuEChERS approach relies on the use of an organic solvent mixed with a salt to extract the analytes from the matrix into the organic phase. The organic phase is then purified to remove residual lipid by dispersive SPE with an appropriate sorbent. Two sorbents of the dispersive SPE, namely C18 and PSA (Primary Secondary Amine), have been tested. C18 showed a better reproducibility and larger capacity to remove lipids than PSA. The studies showed promising results for extracting eight PBDEs including BDE209) and five emerging HFRs, namely pentabromotoluene (PBT), pentabromoethylbenzene (PBEB), hexabromobenzene (HBB), 1,2-bis(2,4,6-tribromophenoxy)-ethane (BTBPE), and decabromodiphenyl ethane (DBDPE) (Gao et al., 2016). . QuEChERS has shown a better extraction efficiency for these HFRs when using acetonitrile, n-hexane and n-hexane/acetone solvents. Ethyl acetate and n-hexane/MTBE are less effective in extracting above-mentioned HFRs. However, large variability in DBDPE measurement using QuEChERS extraction method was reported with a relative standard deviation of 60% and more (Gao et al., 2016). Efficiency and suitability of QuEChERS for DP and DP-related compounds have not been assessed yet. QuEChERS is less tedious than SPE and LLE, thereby limiting the risk of contamination by air, solvent or sorbent contact.

b. Extraction of HFRs from breast milk

LLE is the most common method for extracting HFRs from milk. Solvents used in LLE are similar to those for blood extraction such as DCM:Hexane (Siddique et al., 2012), cyclohexane:acetone (Müller et al., 2016) and hexane:MTBE (Ben et al., 2013; Chen et al., 2015a). Contrary to blood, denaturation of milk samples using acids is not widely reported. It is used only in rare instances when blood and milk share the same extraction procedure (Ben et al., 2013).

LLE and SPE are methods that use a significant amount of glassware, liners, and cartridges that can all be sources of contamination (Siddique et al., 2016). To minimize contamination and to fully automate the extraction procedures, new extraction methods such as Accelerated Solvent Extraction (ASE) (Cechova et al., 2017b) are used. To use the ASE for liquid samples, human milk was reported to be freeze-dried and grounded prior to insertion in the ASE cell (Cechova et al., 2017b) and serum was pipetted directly on Hydromatrix and packed with precleaned cellulose disk in the ASE cell (Siddique et al., 2016). ASE shows little to no contamination (Siddique et al., 2016) but was reported to being responsible for a loss of BDE209 through adsorption in the system tubing (Webster et al., 2009b).

c. Extraction of HFRs from stool and meconium

There are only very limited reports on the methods for extracting HFRs from stool and meconium samples. HFRs in these samples can be extracted using Soxhlet devices with a solvent mixture of hexane and DCM (Jeong et al., 2016). However, ASE provides a better extraction efficiency for HFRs than Soxhlet (Chen et al., 2015a; English et al., 2017). LLE using hexane and sulfuric acid with the help of centrifuging has been used to extract and clean up polychlorinated biphenyls (PCBs) from stool (Ortega-Garcia et al., 2006). Given the similar lipophilic nature of both PCBs and HFRs, LLE could be an alternative to Soxhlet extraction for the extraction of HFRs. Microwave Assisted Extraction (MAE) is common for solid biological samples (adipose tissue, liver) as it is quicker and less tedious than Soxhlet. However, application of LLE and MAE in extracting HFRs from stools or meconium has not been reported yet.

d. Extraction of HFRs from hair

Since levels of contaminants found in hair could be a result of combined internal uptake and external absorption from the environment, various washing methods using water (Kurcharska et al., 2015; Qiao et al., 2018; Poon et al., 2015; Carnevale et al., 2014; Chen et al., 2015b), hexane:DCM (Kurcharska et al., 2015), hexane (Poon et al., 2015), 10% sodium dodecyl sulphate (SDS) (Poon et al., 2015), methanol (Kurcharska et al., 2015), shampoo (Kurcharska et al., 2015) or acetone (Kurcharska et al., 2015) have been tested to remove possible external contaminations. However, none of these procedures seems to completely and exclusively remove the externally absorbed HFRs (Kurcharska et al., 2015). As a result, studied reporting levels of HRFs in hair tend to not employ any washing procedure (Liu et al., 2016; Chen et al., 2019a).

Extraction of HFRs in hair can be accomplished by vortexing the solution at room temperature, followed by sonication in a water bath at 38 °C using a solvent mixture of acetone:hexane (1:1) (Chen et al., 2019a). Another method uses a digestion process with nitric acid and oxygen peroxide (1:1) in a water bath at 60 °C for 2 hours, followed by extraction with hexane:DCM (4:1) (Liu et al., 2016). A softer digestion condition with only 10% nitric acid under sonication for 25 min was also reported (Kurcharska et al., 2015). Alternatively, hair samples can also be incubated for approximatively 12 to 14 hours with hydrochloric acid (4 M) and hexane:DCM (4:1) at 40 °C (Poon et al., 2015; Carnevale et al., 2014; Chen et al., 2015b). There is no mention of possible degradation of the target HFRs under hydrochloric acid treatment for an extensive time (12-14 hours) in any of the published reports. Soxhlet extraction with acetone:hexane (1:3) was also reported (Qiao et al., 2018).

e. Extraction of FRs from nails

Extraction of HFRs from nails is similar to the method for hair, either by using nitric acid and oxygen peroxide (Meng et al., 2020, Liu et al., 2016) or directly using DCM:hexane (3:1) mixture in a sonication bath (Chen et al., 2019b). Similar to hair samples, washing procedure for nail cannot exclusively remove externally absorbed HFRs and therefore is not employed in studies that report HFRs in nail.

3.2. Clean-up of extracts

The extracts usually require further purification prior to being injected into GC/MS instruments for analysis. Since the matrix is largely removed during the extraction step and the analytes are in an organic solvent or solvent mixture, a common clean-up method can be applied to extracts from the above-mentioned matrices of interest, namely blood, milk, stool/meconium, hair and nail. The purpose of the clean-up step is to remove nonpolar lipid and other components that have been co-extracted with HFRs, which can interfere with instrumental analysis.

Most purification methods involve the use of silica gel that had been acidified with sulphuric acid (Brasseur et al., 2014; Kim et al., 2016; Cequier et al., 2015; Herbstman et al., 2010; Jeong et al., 2016) because it is a quick and cost-effective method. However, lipids are destroyed in this process, so the lipid content of the samples must be then determined either prior to the purification step – if the extraction method allows it - or separately on another aliquot of the sample, using either gravimetric (Abdallah 2014; Ben et al., 2013; Sahlström et al., 2015) or colorimetric (for blood only) methods (Zhou et al., 2014b; Ben et al., 2013). Another overlooked setback of acidified silica is the loss of TBPH (Liu et al., 2016; Butt et al., 2016) during purification. Acidified diatomaceous earth has been proposed as an alternative in this case (Lu et al., 2017). The use of non-acidified silica gel for clean-up of serum samples after an SPE extraction using hexane:ethyl acetate as elution solvent was reported in one study too (Butt et al., 2016), in which the authors demonstrated that when hexane:ethyl acetate was used as a elution solvent very little matrix was eluted with analytes. Florisil cartridge is also used (Chen et al., 2015a; Liu et al., 2016; Qiao et al., 2018), especially for extracts of hair and nail samples.

Alternatively, serum extraction and purification may be accomplished in a single step using SPE cartridges topped with sulphuric acid (Gao et al., 2016). This procedure greatly simplifies the process, but DBDPE recoveries were found to be highly variable. Another simplified method for feces is to use a sulphuric acid treatment before fractionation on a non-acidified column for purification (Sahlström et al., 2015).

Use of sulfuric acid in the clean-up destroys lipids in the extract. Non-destructive clean-up methods have been employed to collect the lipids from the same aliquot of samples for determining the lipid content in the sample. These methods were applied to milk analysis and include Gel Permeation Chromatography (GPC) with a styrene divinylbenzene copolymer solid phase (Siddique et al., 2012), Dialysis and Freezing Lipid Filtration (FLP). Among these methods, dialysis using a semipermeable membrane gives the best recovery of lipids in human milk (Cechova et al., 2017a; Cechova et al., 2017b). However, the 48h dialysis is time consuming and may pose a challenge when a large number of samples need to be processed. FLP is an easy and economic method but it shows lower recovery than GPC for most of HFRs of interest (Cechova et al., 2017a).

3.3. GC/MS method for HFRs

Optimization of the HFRs analysis can be achieved through selection of the best conditions on different parts of the GC/MS instrument including sample evaporation in GC injection port, separation of analytes in the GC column, ionization and detection of analytes in the mass spectrometer (MS). These are discussed in the following subsections.

a. GC injection

Injection is most commonly carried using a split/splitless inlet, different temperatures of the split/splitless inlet covering in the range of 240°C to 300°C have been reported for measuring PBDEs and several emerging HFRs. Although an injection temperature of 325°C was recommended for the analysis of PBDEs (Björklund et al., 2004), BDE209 showed substantial degradation at such high temperature in split/splitless injection (Stapleton 2006). For this reason, measurements of HFRs that include BDE209 are usually carried out at relatively low temperature range of 240 °C to 260 °C, while measurements of PBDEs without BDE209 are done at a higher temperature range of 280 °C to 300 °C. Effects of split/splitless injection temperature on the two highly brominated HFRs, namely BDE209 and OBTMPI, were also assessed in authors’ lab in Health Canada. It indicates that the most abundant peaks were obtained at 240 °C for both BDE209 and OBTMPI. The peak height decreased with increased injection temperature, indicating a potential loss of these two HFRs at higher temperatures (Figure 1). A lower temperature (220 °C or 200 °C) on the other hand does not allow proper evaporation resulting in much lower or disappearance of peak abundance.

Figure 1:

OBTMPI (A) and BDE209 (B) peak intensity at different injection temperatures.

Discrimination of HFRs during split/splitless injection can be avoided by using a Programmed Temperature Vaporization (PTV) injection. Conventional split/splitless injections rely on high temperature to quickly evaporate the analytes. The principle of PTV is to introduce the extract solution in the liner at a low temperature, and then vent the solvent out of the injection port while the analytes are trapped in the liner. The temperature is programmed to allow efficient transfer of analytes into the column. There is an increased interest in using PTV as it is a soft way to evaporate compounds limiting discrimination and thermal degradation of thermally labile compounds such as BDE209 in the GC inlet (Miao et al., 2015). Although there is no report on PTV for OBTMPI, similar volatility of OBTMPI and BDE209 implies that PTV would be a suitable technique to improve OBTMPI analysis too. PTV however, must be optimized correctly to be effective as the venting step is often susceptible to loss of analytes. For example, it is necessary to use a low vent flow and an initial temperature below the boiling point of the solvent for the PTV to work properly (Wei et al., 2010). Venting out the solvent in the injection port also allows injecting a higher extract volume (50-100 μl) (Tollbäck et al., 2003) than in conventional split/splitless injection that limits to a few microliters.

Alternatively, cool on-column (COC) injection where the analytes are injected as a liquid directly into the column could present similar advantages as PTV. In this configuration the initial temperature of the column oven and the injection needle have to be maintain below the solvent boiling point. It allows to reduce the chance of discrimination and thermal degradation. However, to our knowledge no study has reported the use of COC injection for routine analysis of flame retardants.

b. GC Column

Thirty- meter columns are most commonly used in the separation of HFRs. A longer column does not favour the elution of HFRs. A shorter 15-meter column is used when heavily brominated FRs such as BDE209, DBDPE and OBTMPI are included in the analysis along with other HFRs (Abdallah 2014). Five -meter columns are also used when only heavily brominated FRs are targeted for analysis (Zhou et al., 2014a). It is reported that when BDE209 was analysed together with other PBDE congeners in a 30-meter column, the detection limit of BDE209 was about 10 folds higher than other congeners (Webster et al., 2009b).

Higher flows of helium as carrier gas reduce the retention time of compounds in the column, thereby improving the signal intensity for heavily brominated FRs such as BDE209. For example, peak height of BDE209 increased by 2.6 times at 5 ml/min and 6.8 times at 10 ml/min compared to regular 1 ml/min (Bending and Vetter 2013).

Stationary phase of the columns used for separating HFRs are coated with cross-linked, non-polar polymer materials such as 100% dimethylpolysiloxane (DB-1MS) or 95% dimethylpolysiloxane with 5% diphenylpolysiloxane (equivalent to DB-5MS, DB-5HT, RTX-1614). 100% Dimethylpolysiloxane stationary phase is mainly used when the analysis includes highly brominated FRs to reduce retention time of analytes in the column (Abdallah 2014). The column RTX-1614 (15m) designed for routine analysis of PBDEs is a robust alternative method that allows quick analysis of both PBDEs (Pirard and Charlier, 2018; Guo et al., 2018b) and most Dechlorane-like compounds (Guo et al., 2017). RTX-1614 has also been reported for its superior repeatability for Dechlorane-like compounds and superior response signal for brominated Dechlorane-like compounds (HCDBCO, Dec604) (Rjabova et al., 2017). However, DB-5MS still shows higher response signal of chlorinated Dechloranle-like compounds (Dec602, Dec603, Dechlorane Plus) compared to RTX-1614 or DB-5HT (Rjabova et al., 2017). Use of ZB-semivolatiles capillary column (5% phenyl-arylene-95% dimethyl-polysiloxane) have also been reported for separation of most of the compounds of interest including Dechloranes (Dechlorane Plus, Dec602, Dec603), PBDEs, TBB, TBPH, BTBPE, DBDPE (Sales et al., 2017). DB-XLB and RTX-500 however, have shown poor response for highly brominated HFRs (Stapleton 2006; Björklund et al., 2004).

c. MS ionization

Choice of ionization mode in MS is also important for the optimized analytical conditions. Electron capture negative ionization (ECNI) is a preferred choice of ionization mode over electron impact (EI) mode as the former is more sensitive for most HFRs, especially for brominated FRs (BFRs) such as PBDEs (Lin et al., 2015). For BFRs, ECNI often produces bromide ion (signals at m/z 79 and m/z 81) as the major fragment, which can be easily monitored with a single quadrupole detector or a Time of Flight (ToF) detector (Abdallah 2014). However, since signals of m/z 79 and m/z 81 can be generated from a lot of BFRs under ECNI, this ionization technique cannot be used to quantify co-eluting BFRs. A good separation of BFRs in the GC column is required in this case. However, full separation of BFRs may become increasingly challenging as the number of HFRs included in the analysis keep expanding. Monitoring of m/z 79 / m/z 81 for most BFRs in ECNI also prevents the use of ¹³C labelled standard of these BFRs to assess the recoveries since they also generate signals of m/z 79 and m/z 81. In this case, High-Resolution Mass Spectrometry (HRMS) with electronic ionization (EI) (Feo et al., 2012; Link et al., 2012; Fromme et al., 2016b) or tandem MS/MS with EI ionization (Cristale et al., 2013, Tolosa et al., 2020) would be a better choice for its ability to monitor fragments that are specific to a given BFR.

In addition, inductively coupled plasma mass spectrometry (ICP-MS), which is considered to be more sensitive than EI or ECNI (Krol et al., 2012), has only recently been used for monitoring PBDEs with low bromine content (BDE28, BDE47, BDE100, BDE153 and BDE154) (Bergant et al., 2018). Atmospheric pressure chemical ionisation (APCI) are also used as a choice of ionization (Wang et al., 2018; Neugebauer et al., 2017; Sales et al., 2017; Portoles et al., 2015).

d. Ionization source temperature

Ionization source temperatures reported in the literature range from 120°C to 250°C for ECNI. For PBDEs, the optimal temperature is around 200 °C. In general, signal intensity of the bromine ion (m/z 79 and m/z 81) increases with increased source temperature for PBDEs, especially for highly brominated congeners, with the exception of BDE209 (Hites, 2008). Monitoring ions of chlorinated dechlorane compounds (Dechlorane Plus, Dec602, Dec603) and HCDBCO shall be selected accordingly to the source temperature. For t-DP, the fragment [M]⁻ remains the most abundant at lower source temperature but its intensity decreased significantly at 250°C. The most abundant fragment of s-DP switches from [M]⁻ to [M-6Cl]⁻ when source temperature changed from 150°C to 250°C. As a result, a lower temperature is recommended for DP analysis to use high mass fragments for its quantification (Torre et al., 2010). Effects of source temperature in ECNI on two dechloranes (Dec602 and Dec603) were investigated in authors’ lab in Health Canada and the results are shown in Figure 2. A lower source temperature favours the intensity of high mass fragments ([M]⁻ and [M-Cl]⁻) while a higher temperature favours the signal of low mass fragments such as m/z 235/237 cluster (corresponding to ion [C₅Cl₅]⁻). In conclusion, as a balance we recommend using a low temperature around 150°C when dechlorane compounds are included in the analysis. Otherwise, operating at a higher ionization temperature (200-250°C) gives a higher signal for bromine ion fragments.

Figure 2:

Mass spectra of Dec602 and Dec603 at a source temperature of 150°C, 200°C and 250°C. Spectra were obtained with on an Agilent 7890A GC/MS Triple Quad (Agilent Technology, California, USA) with a DB-1MS column (30 m x 0.25 mm x 0.25 μm) from J&W Scientific, California, USA. Helium was used as carrier gas at 1.2 ml/min. Pulse splitless injection of 2 μl was used for sample introduction with a temperature of the inlet of 240 °C. The MS was operated in ECNI mode using methane as collision gas. Temperature of the transfer line and Quadrupole were maintained at 320 °C and 150 °C.

3.4. Quality assurance and quality control measures

Most literature reporting measurement methods also include quality assurance and quality control (QA/QC) information. These QA/QC include linear range of calibration, blank contamination, limit of detection or limit of quantification and recovery.

Multi-point calibration was used to determine the linearity of calibration, commonly within the range of 0.1-1000 ng/ml for ECNI in selected ion monitoring mode (Fromme et al., 2015; Cequier et al., 2015). It is not unusual to find target analytes at various levels in blanks of all matrix, especially for PBDEs and Dechlorane Plus. As a result, samples are usually blank corrected by subtracting the value in the samples by the blank value (Kim et al., 2016; Cechova et al., 2017; Brasseur et al., 2014).

Limits of detection/quantification of PBDEs, Dechlorane Plus, Dec602, Dec603 and HCDBCO are similar and in the order of 0.01-0.1 ng/g lipid weight (lw, for blood and milk) or dry weight (dw, for hair and nails) but can jump to 1-10 ng/g lw or dw for Dechlorane Plus and HCDBCO (table S1, S2 and S4). Limits of detection of BDE209 in hair, nails and stool tend to be higher around 0.05-5 ng/g dw (English et al. 2017; Qiao et al., 2018; Zheng et al., 2011) but can also been found to be around 100 times higher than lower PBDEs at 10-15 ng/g dw, (Qiao et al., 2019; Jeong et al., 2016).

In comparison, limits of detection of TBB, TBPH, BTBPE, OBTMPI and DBDPE tend to be higher in general. In blood samples their limits are within 0.01-80 ng/g lw (Butt et al., 2016; Guo et al., 2018a; Qiao et al., 2018; Zhou et al., 2014a). In hair and nails, their limits are within 0.1-20 ng/g dw (Qiao et al., 2018; Qiao et al., 2019; Zheng et al., 2010; Chen et al., 2019b). In milk Cechova et al. (2017), reports a limit of detection of 0.001 ng/g lw for TBB, TBPH and BTBPE, while Zhou et al., (2014a) reports a limit of detection of 0.03, 0.15, 0.86, 0.20 and 1.7 ng/g lw for TBB, TBPH, BTBPE, OBTMPI and DBDPE.

Depending on the information provided, either limits of detection (LOD) or limits of quantification (LOQ) are given in supplementary materials Tables S1, S2, S3 and S4. Detailed information about LOD or LOQ in each cited studies can be found in the associated references in these tables.

Matrix spikes give various recoveries in blood and milk, mostly within the range 50-150% and with a standard deviation around 10-30% (Qiao et al., 2018; Zhou et al., 2014a; Tay et al., 2019; Pan et al., 2020; Tao et al., 2017; Cechova et al., 2017). However, recoveries lower than 50% were also reported notably for BTBPE (Tay et al., 2019), DBDPE (Zhou et al., 2014a), HCDBCO (Cechova et al., 2017) and BDE209 (Zhou et al., 2014a). Duplicate analysis shows a coefficient of variation mostly within 10-30% but can be higher for compounds found extensively in blanks such as BDE47 (Brasseur et al., 2014).

Recoveries in stool samples for PBDEs are found to be mostly above 50% with standard deviation reported below 25% (Chen et al., 2015a; English et al., 2017; Sahlström et al., 2015). The only study investigating emerging flame retardants in stool (Sahlström et al., 2015) reports mainly acceptable recovery (within 47-105%) except for TBB and TBPH at 39% and 21%.

In hair and nails, recoveries are mostly within 60-110% with standard deviation below 20% (Meng et al., 2020; Liu et al., 2016; Chen et al., 2019b; Qiao et al., 2019; Zhang et al., 2013).

Selected ion monitoring is a common detection mode in GC/MS analysis. At least two major ions are selected for each target chemical. Positive identification of a peak is ensured by the ratio of two selected ions. Different range of ion ratios are considered acceptable in various studies. For example, accepted ion ratios of within ±15% (Brasseur et al., 2014), ±20% (Kim et al., 2016) or ±30% (Zhou et al., 2014b) have been reported.

3.5. Challenges in the measurements of HFRs in biological samples

a. Background contamination

Potential contamination during sample preparation and instrumental analysis is a challenge in the measurements of HFRs in biological samples (Siddique et al., 2016). Blank contamination is especially difficult to control for several PBDE congeners such as BDE47, BDE99 and BDE153 as they are ubiquitously present in the environment and hence they are always present in laboratory consumables and air. Several remediation measures can be implemented to reduce the contamination. It is critical to thoroughly wash glassware with dichromic acid mixture (Miao et al., 2015; Siddique et al., 2012), limit the contact of samples with plastic materials and bake glassware and solid sorbents (silica, sodium sulfate, sand) at 400-500 °C prior to their use (Siddique et al., 2012; Butt et al., 2016). It is important to be aware that contamination also comes from commercial organic solvents and organic-free water (Siddique et al., 2016) or in-house generated milli-Q water. Drying steps on cartridges need to be performed under a gentle stream of nitrogen to avoid unnecessary contact of absorbent with laboratory air (Kim et al., 2016). For the instrumental analysis, it is important to keep GC baseline as low as possible by removing residual fat in the extract during the clean-up step and by maintaining the GC/MS system clean (Shi et al., 2013; Gao et al., 2016). Highly halogenated HFRs tend to quickly contaminate both GC injection port and ionization source resulting in reduced sensitivity over time. In addition, to minimize photo-degradation of HFRs, use of both aluminium foil to cover the samples during sample preparation and amber GC vials for GC/MS analysis is highly recommended (Butt et al., 2016).

b. Choice of surrogates and internal standards

The ideal choice for surrogates and internal standards is to use the isotope labelled chemical of the target compound. For many BFRs, as mentioned before, predominant ion in soft method of ionization such as ECNI is bromide (m/z 79 and m/z 81). If a BFR can only be monitored by its bromide signal, its isotope labelled standard cannot be used as internal standards or recovery surrogates in ECNI-MS analysis as they generate the same bromide signals of m/z 79 and m/z 81. As a result, alternative compounds must be used to substitute the labelled counterpart as a proxy. This inevitably leads to some degree of bias in calculating recoveries of measured BFRs. These substituting compounds must be chosen carefully by taking into consideration their structural similarities with the targets.

PBDEs congeners that were not present in commercial mixtures are used as recovery surrogates or internal standards in the analysis of PBDEs and other HFRs. For example, BDE77, BDE105, BDE139 and BDE138 have been reported as surrogates and standards for the analysis PBDE congeners with low level of bromine substitution using ECNI source (Qiao et al., 2018; Kim et al., 2016; Gao et al., 2016; Ochoa-Martinez et al., 2016). Three carbon-13 labelled PBDE congeners, ¹³C-BDE153 (Zhou et al., 2014a), ¹³C-BDE183 (Zhou et al., 2014b), and ¹³C-BDE209 (Miao et al., 2015; Gao et al., 2016) were also used for the medium and highly substituted PBDEs as these compounds can be monitored using signals other than bromide ion (m/z 79 and m/z 81). Other labelled standards used as surrogates or internal standards in the measurements of bromine-containing HFRs include either ¹³C-TBPH or ¹³C-TBB for TBB and TBPH (Butt et al., 2016) and either ¹³C-s-DP or ¹³C-t-DP for s-DP and t-DP (Siddique et al., 2012). However, use of either ¹³C-TBPH or ¹³C-TBB in ECNI means that either TBPH or TBB cannot be monitored with the bromide ion (m/z 79 and m/z 81).

c. Co-elution of HFRs on GC column

Mostly observed co-eluting peaks in analysis of HFRs are the pairs of HCDBCO and Dec602, BDE154 and BB153, BDE99 and TBB, and Dec604 and BDE183 (Allen et al., 2013), (Abdallah 2014; Bjermo et al., 2017). Among them, HCDBCO and Dec602 are monitored with different m/z signals in ECNI so it is not necessary to fully separate them. BDE154 is present but in less than 5% of the total mass of the commercial mixture (Guardia et al., 2006). BB153, on the other hand, is the most commonly used polybrominated biphenyl (PBB) and is still present in the environment and in humans due to its long half-life (29 year in the human body) (Sjödin et al., 2003). It is necessary to be aware of the co-elution of these two compounds as they both are monitored with signals of m/z 79 and m/z 81 under ECNI. Separation of BDE99 and TBB as well as BDE183 and Dec604 is required in ECNI analysis as the signals of bromide ion (m/z 79 and m/z 81) are used for quantification of these chemicals (Hites 2008).

d. Need for small final volume due to small sample quantity available for analysis

Some biological samples such as blood cannot be obtained in large quantity. Moreover, emerging HFRs are often present at low concentrations in biological samples. Due to the combination of these two factors, it is often necessary to concentrate the sample by reducing the final volume for GC/MS analysis in order to achieve appropriate detection of HFRs. Concentrating samples to a small final volume around 20-50 μl have been reported (Cechova et al., 2017a). Nonane (Neugebauer et al., 2017) or Dodecane (Kim et al., 2016) are commonly used as volume keeper for the final extract, while others use hexane (Miao et al., 2015). Another way to overcome the need for small final volume is to use PTV, which allow a larger injection volume to increase instrument detection of analytes in the samples (Tollbäck et al., 2003).

4. Concentrations of Halogenated Flame Retardants in Humans

Levels of HFRs in humans are summarized according to groups of chemicals. Levels of DP including its dechlorinated products in blood, milk, stool/meconium, hair and nail are summarized in Table S1 (see Supplementary data), while other dechloranes (Dec602, Dec603, Dec604 and HCDBCO) are in Table S2. Levels of five non-PBDE brominated FRs (TBB, TBPH, BTBPE, OBTMPI and DBDPE) are summarized in Table S3. Since levels of PBDEs in human blood and milk have already been reviewed as mentioned in section 2 (Fromme et al., 2016a; Jiang et al., 2019), only levels of PBDEs in stool, meconium, hair and nails are presented in Table S4. In the following subsections levels of these groups of HFRs in various bio-specimens (blood, milk, stool including meconium, hair and nail) are discussed.

4.1. Blood

Blood is the most commonly used matrix to assess internal burden for HFRs (Esteban and Castano 2009). The levels determined in blood are linked to observed health outcomes. For example, PBDEs levels in blood are associated with changes in thyroid hormone (Guo et al., 2018a; Huang et al., 2014), and impacts on physical and mental neurodevelopment (Herbstman et al., 2010). However, blood collection is an invasive process, which may discourage participation in human biomonitoring and epidemiological studies. It also poses challenges to collect blood from very young children.

4.1.1. Levels of Dechlorane plus in blood

Dechlorane plus (DP) exists in both syn- (s-DP) and anti- (t-DP) isomeric forms. Levels of DP in the environment and in biota including humans has been reviewed in 2011 (Xian et al., 2011). Since then, DP levels in blood were reported in several countries including: Canada (Zhou et al., 2014b), China (Chen et al., 2015b; He et al., 2013; Qiao et al., 2018), France (Brasseur et al., 2014), Germany (Fromme et al., 2016a), Norway (Cequier et al., 2015; Tay et al., 2019), South Korea (Kim et al., 2016) and USA (Liu et al., 2016) (Table S1). While the levels of DP in general populations are similar around the world at sub to low ng/g lipid weight (lw) (reported median of 0.2-2.4 ng/g lw for s-DP and reported median of 0.50-6.2 ng/g lw for t-DP), higher levels (50 ng/g for s-DP and above 100 ng/g for t-DP) can be found in industrial workers of e-waste recycling facilities in China (Yan et al., 2012; Zhang et al., 2013; Chen et al., 2015b).

In general, t-DP is in higher concentration in human blood than s-DP. The fraction of t-DP in total DP (sum of s-DP and t-DP), denoted as f_anti, is used to assess transformation of DP in the environment and its metabolization in biota (Xian, et al., 2011). Commercial mixtures of DP have a f_anti value of approximatively 0.6-0.8 (Xian, et al., 2011). The values of fanti in various environments including ambient air, sediments, indoor settled dust and various biota have been summarized (Xian, et al., 2011). In the environment, degradation of DP is a complex process that is influenced by photolytic reactions in air and microbial enzymatic degradation in sediments. Values of f_anti are higher in human blood and other matrices among urban and rural residents away from e-waste recycling plants (median of 0.72) than among e-waste recycling plants workers and residents living nearby the plants (median of 0.61), indicating that t-DP is more persistent than s-DP in humans (Figure 3).

Figure 3:

Ratio of the concentration of t-DP in total DP (sum of s-DP and t-DP) (f_anti) in (A) blood from industrial workers (compiled from 5 studies), (B) blood from urban and rural resident (13 studies), (C) hair and nails from industrial workers (7 studies), (D) hair and nails from urban and rural resident (8 studies), (E) blood, hair and nails from industrial workers (12 studies) and (F) blood, hair and nails from urban and rural resident (21 studies). The straight line represents the median value and “X” the arithmetic mean of f_anti values. Value of f_anti was calculated from the mean and median value of the s-DP and t-DP concentration given in the literature unless it was already reported by the author.

Furthermore, a few studies also reported the dechlorinated DP compounds in blood samples. Although the exact positions where the chlorine is eliminated were not determined, the number of chlorines remained in DP molecule is indicated by the suffix number following the Cl letters. As such, the two common dechlorination products have been named as Cl₁₁DP (meaning one Cl in DP is replaced by H, or DP−Cl+H) and Cl₁₀DP (DP−2Cl+2H). For example, in e-waste recycling plant workers in China, the concentration of t-Cl₁₁DP (arithmetic mean = 1.47 ng/g lw) was almost two orders of magnitude lower than the parent compound t-DP (median = 103.6 ng/g lw) in a Chinese study (sampled in 2011) (Yan et al., 2012). Similar concentration ratio was found in another Chinese study involving e-waste recycling plant workers, in which geometric mean of t-C₁₁DP of 1.9 ng/g lw and the median of t-DP of 120 ng/g lw were reported (sampling date unknown) (Chen et al., 2015b). In comparison, median concentration of t-Cl₁₁DP in Chinese maternal sera was 0.371 ng/g lw for mothers living for a long time (>20 years) in the vicinity of e-waste recycling activities and 0.155 ng/g lw for mothers residing in the vicinity e-waste activities for less than 3 years (sampled in 2010/2011) (Ben et al., 2014). In the same study, dechlorinated DPs were not detected in cord blood samples (Ben et al., 2014).

Dechlorinated DPs can be formed due to photolytic degradation (Wang et al., 2013). Biotic dechlorination of DP remains largely unknown. Studies reporting dechlorinated DP are inconclusive on whether the Cl₁₁DP and Cl₁₀DP are formed prior to entering the human body or result from human metabolization. Formation of dechlorinated DP during sample analysis due to either dirty liner or high GC injection temperature is reported too (Xian et al., 2011). Therefore, reporting of dechlorinated compounds should be treated carefully.

4.1.2. Levels of other Dechloranes in blood

Several dechloranes, including Dec602, Dec603, Dec604 and HCDBCO in blood were reported in Canada (Zhou et al., 2014b; Kim et al., 2019), France (Brasseur et al., 2014), Germany (Fromme et al., 2015), Norway (Cequier et al., 2015; Tay et al., 2019), China (Guo et al., 2018a) and South Korea (Kim et al., 2016) (Table S2). Among them, Dec602 and Dec603 had higher detection frequencies (>75%, for both compounds) showing their ubiquity in daily life. Dec604 was measured in some studies but not detected (Brasseur et al., 2014; Zhou et al., 2014b; Kim et al., 2016). This might be attributed to its recent arrival on the market (Rjabova et al., 2016).

Detection frequencies (DF) and levels of HCDBCO reported were not consistent among different countries. Highest level of HCDBCO was found with a median concentration of 10 ng/g lw (DF: 95%, sampled in 2015) in China near a petrochemical plant (Guo et al., 2018a). On the other hand, HCDBCO was not detected in a Norwegian study (DF: 0%, LOD: 126 pg/g serum, sampled in 2012). In Canada, it has been reported in two studies, both from residents in the Sherbrooke area. One of them reported low detection frequency of below 20% (Zhou et al., 2014b) (sampled in 2014/2016 in children and their parents) and the other around 66% (Kim et al., 2019) (sampled in 2008-2009 in maternal blood). However, even when the detection frequency is below 20%, the average level in child (n = 170) is still notably high (1.62 ng/g lw) showing a skewed biomonitoring data. Skewed distribution of HCDBCO in environmental and biological samples are commonly reported (Kim et al., 2019).

4.1.3. Levels of PBDEs in blood

PBDEs in children and in cord blood up to 2015 worldwide (Fromme et al., 2016a) and in Chinese populations up to 2018 (Jiang et al., 2019) have been reviewed, respectively. Additional information on PBDEs in human blood has since become available and it is summarized as following. 16.5 ng/g (lw, cord blood, sum of the median of BDE −47, −99, −100 and −153) in New York (sample collected in 2001) (Herbstman et al., 2010), 130.9 ng/g (lw, 2-year-old, sum of the geometric mean (GM) of BDE −47, −99, −100 and −153) and 42.8 (lw, 8-year-old, sum of the GM of BDE −47, −99, −100 and −153) in Cincinnati (2005 – 2008 for 2-year-old and 2011-2014 for 8-year-old) (Liang et al., 2019), 87.8 ng /g (lw, sum of the GM of BDE −28, −47, −85, −99, −100, −153 and −154, 7 year old born and raised in California but whose mothers were from Mexico) in California (2007 – 2008) (Eskenazi et al., 2011) were reported from USA. In comparison, PBDEs concentrations reported in Australia (2014-2015) (Drage et al., 2019b) and European countries including Germany (several sampling campaign between 2002 and 2009) (Link et al., 2012), Denmark (2011) (Knudsen et al., 2017), France (2003 – 2005) (Brasseur et al., 2014), Norway (2012) (Cequier et al., 2015) are significantly lower and the sum of the mean of BDE −47, −99, −100 and −153 are often in the range of 1-10 ng/g (lw). In Mexico the sum of geometric mean of BDE-47, 99, 100, 153, 154 in children is 29.5 ng/g (lw) in Juarez (Ochoa-Martinez et al., 2016) (2012), 29.0 ng/g (lw) in San Luis (Perez-Maldonado et al., 2017) (2015 - 2016) and 42 ng/g (lw) in the Guadalajara area (2012 - 2013) (Orta-Garcia et al., 2018). In Sherbrooke, Canada, PBDEs concentrations (median (ng/g lw) of 5.43 1.41 1.37 3.32 for BDE −47, −99, −100 and −153, respectively, in children) have been found to be slightly higher than in Europe but lower than in USA (Kim et al., 2019).

Generally, detection frequencies in blood for BDE47, BDE99 are high (>80%). BDE153 is also a major congener with various detection frequencies. For the general population a detection frequency between 60-100% is commonly reported for these three major PBDEs.

Figure 4 reconstructed the concentrations of BDE47 and BDE153 data in North America, Europe and East Asia for comparison. It can be seen in the figure that concentration of BDE47 in North America (median: 28.18 ng/g lw) is an order of magnitude higher than in Europe (1.99 ng/g lw) and East Asia (2.72 ng/g lw). Results are similar for the BDE99 and BDE100 (not shown in the figure). It is worth noting that BDE47 is by far the most abundant congener in North America, but in East Asia the ratio of BDE47 over BDE153 is close to 1 (Figure 4). The main reason for this difference is the large amount of Penta-BDE used in USA (Abbasi et al., 2019). Ratio of BDE47 over BDE153 in Europe is close to two, which is lower than in North America but higher than that in East Asia (Figure 4).

Figure 4:

Concentrations of BDE47, BDE153 in blood in North America (Canada, Mexico, the US), Europe (Belgium, Denmark, France, Germany, the Netherlands, Norway, Poland, Spain, Sweden) and East Asia (China, South Korea). The straight line represents the median value and “X” the arithmetic mean. A total of 12 studies were used to assess BDE47 in North America, 17 in Europe, 17 in East Asia and 12 for BDE153 in North America, 15 in Europe, 15 in China.

BDE209 is a thermally labile HFR that requires particular care during the instrumental analysis and so it is not always reported in studies on PBDEs level in blood (Fromme et al., 2016a). Although, BDE209 was detected in internal human samples (Roosens et al., 2010), it is less accumulative compared to lower substituted PBDEs in the body (Hakk and Letcher, 2003). As a result, its detection frequency in blood can be lower than BDE47, BDE99, BDE153. The relative content of BDE209 in PBDEs (the sum of BDE47, BDE99, BDE100, BDE153 and BDE209) in blood varied from 1-5% in USA to 10-50% in locations where BDE47 is less predominant such as in Europe.

4.1.4. Levels of other Brominated FRs

There are limited data available for other brominated FRs in blood (Table S3). Among them, DBDPE was detected at a median concentration of 39.2 ng/g lw in a cohort of university students in southern China (DF: 100%, LOQ: 5.59 ng/g lw, sampled in 2014) (Qiao et al., 2018) and at a median concentration of 180 ng/g lw (DF: 100%, Reporting limit (RL): 23-73 ng/g lw, sampled in 2015) in school age children (average age of 10) living near a petrochemical plant in southern China (Guo et al., 2018a). In comparison, it was detected in only 5.9% of blood samples with a maximum level of 123 ng/g (lw) in Canada (LOD: 3.5 ng/g lw, sampled in 2008/2009) (Zhou et al., 2014a) and was barely detected in Norway (DF: 2%, LOD: 84 pg/g serum, sampled in 2013) (Tay, et al., 2019).

BTBPE concentration ranged from <DL (3.2 ng/g lw) to 16 ng/g lw with a low detection frequency of 3.9% in Canada (LOD: 3.2 ng/g lw, sampled in 2008-2009) (Zhou et al., 2014a). Similar low detection frequency of this compound was observed in China (DF: 4.65%, LOQ: 35.2 ng/g lw, sampled in 2014) (Qiao et al., 2018) and in Norway (DF: 9%, sampled in 2012 (Cequier, et al., 2015) and DF: 0%, sampled in 2013 (Tay et al., 2019), LOD: 34 pg/g serum) as well. However, for school age children living near a petrochemical plant in China, BTBPE was detected at a much higher detection frequency of 70% with median concentration of 0.83 ng/g lw (RL: 0.48-1.5 ng/g lw, sampled in 2014) (Guo et al., 2018a).

Levels of TBB (median: 5.6 ng/g lw, DF: 97%, RL: 0.74-2.4 ng/g lw) and TBPH (median 6.6 ng/g, DF: 83%, RL: 0.48-1.5 ng/g lw) were determined in school age children living near a petrochemical complex in southern China (sampled in 2015) (Guo et al., 2018a). TBB (mean: 5.4 ng/g lw, DF: 57%, LOD: 0.38 ng/g lw) and TBPH (range: <LOD to 164 ng/g lw, DF: 16.7%, LOD: 7.3 ng/g lw) have been measured in Canada (sampled in 2008-2009) (Zhou et al., 2014a). These two compounds also have been measured in USA (sampled in 2014) at a median of 7.3 ng/g lw (DF: 92%, LOD not reported) for TBB and 40 ng/g lw (DF: 16%, LOD not reported, median calculated by taking into account only samples with detectable concentration of analyte) for TBPH (Liu et al., 2016). However, they were not detected in any of the 43 blood samples from Durham county in USA (LOD: 56 ng/g lw and 24 ng/g lw for TBB and TBPH, respectively, sampled in 2008-2010) (Butt et al., 2016) or in Norway ((LOD: 72 pg/g, serum and 14 pg/g, serum for TBB and TBPH, respectively, sampled in 2013) (Tay et al., 2019). Low detection of TBB and TBPH in some areas might be attributed to their recent introduction to the market as a replacement of penta-BDE mixture. TBPH has high hydrophobicity with a log Kow of 12 (Zhou et al., 2014a) as compared to a log Kow of 6.8, 7.3 and 7.9 for BDE47, BDE99 and BDE153 (Braekevelt et al., 2003) and so that might prevent its bioaccumulation in the biota (Liu et al., 2016). TBB and TBPH could also metabolize to 2,3,4,5-tetrabromobenzoic acid, and mono-(2-ethyhexyl) tetrabromophthalate, respectively, in humans (Roberts et al., 2012; Liu et al., 2016).

OBTMPI in blood was reported in Norway (Cequier et al., 2015; Tay et al., 2019) and Canada (Zhou et al., 2014b) with low detection frequency of 2%, (LOD: 30 pg/g serum, sampled in 2013) and 9.8% (LOD: 1.5 ng/g lw, sampled in 2008-2009), respectively.

4.1.5. Comparison of levels of HFRs in blood

Figure 5 shows levels of several HFRs including BDE47, BDE99, BDE153, t-DP, s-DP, Dec602, Dec603, TBB and BTBPE in blood worldwide. The figure provides a picture of overall concentrations of HFRs in human bloods. Among them, BDE47 is the dominant compound (median: 2.72 ng/g lw) followed by t-DP (median: 1.80 ng/g lw) and other two major PBDEs: BDE153 (Median: 1.58 ng/g lw) and BDE99 (median: 1.38 ng/g lw). Concentrations for the other HFRs are about 5-10 times lower (median: 0.44, 0.11, 0.19, 0.19 ng/g lw for Dec602, Dec603, TBB and BTBPE, respectively). However, it is important to note that the levels are the global average. For example, levels of PBDEs in European countries are significantly lower than in North America (Figure 4). As a result, the gap in concentrations between PBDEs and the rest of HFRs would be higher for North America than the figure shows.

Figure 5:

Comparison of concentrations among three major PBDE congeners, two DP isomers and four novel flame retardants in humans. The straight line represents the median value and “X” the arithmetic mean. Levels are in human blood, except a combined blood and milk data for TBB and BTBPE due to their scarce data in blood. Data on DBDPE, TBPH, Dec604, HCDBCO were too limited in number to be properly represented in the figure. Data for BDE47, BDE99 and BDE153 were taken from previous review by Fromme et al. (2016a) and Jiang et al. (2019). Additional data from Mexico (Orta-Garcia et al., 2018; Ochoa-Martinez et al., 2016; Eskenazi et al., 2011), Denmark (Knudsen et al., 2017), South Korea, USA (Vuong et al., 2017) and France (Brasseur et al., 2014) were added. s-DP, t-DP, Dec602, Dec603, TBB and BTBPE data were from Table S1, Table S2 and Table S3. Several DP data from workers of e-waste recycling plants in China (Yan et al., 2012; Zhang et al., 2013; Chen et al., 2015b) were not included as they were not representative of the exposure for general populations. For most studies, median was chosen to represent the exposure; if the median was not reported geometric mean or arithmetic mean was selected.

4.2. Milk

While blood remains the most common matrix for HFRs biomonitoring in humans, breast milk is also widely used, especially to assess the exposure of breast-fed children to HFRs (Johnson-Restrepo and Kannan 2009). Maternal milk is a preferred choice over young children’s blood to assess exposure of breast-fed children as blood from children at their early age can’t always be easily accessible for ethical reasons.

4.2.1. Levels of Dechlorane plus in milk

DP in human milk has been reported in Canada (Zhou et al., 2014b), China (Ben et al., 2013; Pan et al., 2020) and Europe (Cechova et al., 2017b) (Table S1). A study comparing DP levels in milk in three European countries (Slovakia, the Netherlands and Norway) showed low detection frequencies (<30%) for DP with arithmetic mean values of 0.111, 0.278, 0.355 ng/g (lw) for s-DP and 0.057, 0.155, 0.055 ng/g (lw) for t-DP in Slovakia (sampled in 2011-2012), the Netherlands (sampled in 2011-2014) and Norway (sampled in 2003/2006), respectively (Cechova et al., 2017b). Data from Canada are similar to those from Europe with a detection frequency of 40% and 50% for s-DP and t-DP, respectively and a median value of 0.02 ng/g lw for the t-DP (Zhou et al., 2014b) (sampled in 2008-2009). DP levels in China were much higher. One Chinese study found a median of 1.33 ng/g and 3.32 ng/g (lw) for s-DP and t-DP, respectively, in residents living around an e-waste recycling facility and 0.50 ng/g and 1.58 ng/g (lw) for residents away from the e-waste recycling facility (sampled in 2010-2011) (Ben et al., 2013). Another Chinese study reported a median of 0.581 ng/g lw and 1.64 ng/g lw for s-DP and t-DP, respectively (sampled in 2011-2012) (Pan et al., 2020).

4.2.2. Levels of other Dechloranes in milk

The Canadian study mentioned in previous subsection (Zhou et al., 2014b) has reported Dec602 (DF: 82%, LOD: 0.03 ng/g lw), Dec603 (DF: 72%, LOD: 0.01 ng/g lw), Dec604 (DF: 0%, LOD: 0.10 ng/g lw) and HCDBCO (DF: 69%, LOD: 0.05 ng/g lw) in human milk (Table S2). HCDBCO was investigated in one European study (Cechova et al., 2017) but was not found in any of the samples from Norway (sample size: 305), the Netherlands (sample size: 116) and Slovakia (sample size: 37) with a limit of detection of 0.004 ng/g lw.

4.2.3. Levels of PBDEs in milk

Two recent reviews on PBDEs provide detailed PBDE concentrations in milk (Zhang et al., 2017; Jiang et al., 2019). These two reviews summarised the data from 2000 to 2015. The results presented in these two reviews show that, similar to levels in human blood samples, levels of PBDEs in human milk samples in Europe (median in the range of 0.4 to 6.3 ng/g, lw) and Asia (median in the range of 1.5 to 19.5 ng/g, lw) are comparable but the levels in North America are at least a order of magnitude higher (median in the range of 19.9 to 54.5 ng/g, lw). BDE47 is again the most dominant congener founds in human milk followed by the other three congeners: BDE99, BDE100 and BDE153.

4.2.4. Levels of other brominated FRs in milk

There is scarcity of data on other BFRs in human milk (Table S3). Both detection frequency and level of other BFRs are lower than those of PBDEs. TBB, TBPH and BTBPE measured in Europe (Cechova et al., 2017b) (Norway, the Netherland and Slovakia) show similar concentrations. The arithmetic means of TBB are 0.086, 0.240 and 0.037 ng/g lw in Norway (DF: 47%), the Netherlands (DF: 69%) and Slovakia (DF: 59%) respectively, the arithmetic means of TBPH are 0.613, 1.34 and 0.622 ng/g lw in Norway (DF: 29%), the Netherlands (DF: 15%) and Slovakia (DF: 27%) respectively and finally the arithmetic means of BTBPE are 0.586, 1.103 and 0.023 ng/g lw in Norway (DF: 24%), the Netherlands (DF: 26%) and Slovakia (DF: 22%) respectively. In this study LOD are 0.001, 0.023, 0.001 ng/g lw for TBB, TBPH and BTBPE.

In Canada, TBB (DF: 78.1%, LOD: 0.03 ng/g lw) and TBPH (DF: 32.4%, LOD: 0.15 ng/g lw) were detected in human milk while BTBPE (LOD: 0.86 ng/g lw) was not detected at all (sampled 2008-2009) (Zhou et al., 2014a). Detection frequency of OBTMPI (DF: 1.0%, LOD: 0.20 ng/g lw) and DBDPE (DF: 8.6%, LOD: 1.7 ng/g lw) in the same Canadian samples were very low (Zhou et al., 2014a). TBB in Canadian samples (median of 0.41 ng/g lw) is about 10 times higher than TBB values in Europe. In China, a relatively higher detection frequency (88%, LOD: 27 pg/ml milk) and median level (5.96 ng/g) was reported for DBDPE (sampled in 2014) (Chen et al., 2019c).

Studies in the United Kingdom show that detection frequencies of TBB, TBPH, BTBPE and DBDPE were 44%, 36%, 28% and 4%, respectively, in samples collected in 2010 (Tao et al., 2017), which is comparable to data found in the Netherlands, Norway and Slovakia (Cechova et al., 2017b). However, for samples collected in 2014-2015 detection frequencies of TBB, TBPH, BTBPE and DBDPE increased to 90%, 50%, 40% and 10%, respectively (Tao et al., 2017), showing an upward trend of this compounds in the United Kingdom over time.

4.3. Stool and meconium

Stool, including meconium, has not been widely used for human biomonitoring of HFRs, but it is gaining interest as a non-invasive alternative to blood for assessing internal exposure (Sahlström et al., 2015; English et al., 2017; Chen et al., 2015a; Jeong et al., 2016). It is also becoming popular due to increasing interest in understanding the effects of flame retardants on animal and human gut microbiome. Stool can be collected in large quantities. However, stool may only reflect the exposure of a snapshot in time, usually a few days and may not always represent the average daily exposure. On the other hand, meconium is useful to assess prenatal exposure on a larger timespan given its accumulation over the 2^nd and 3^rd trimesters (Jeong et al., 2016).

4.3.1. Dechlorane plus and other Dechloranes in stool

HCDBCO was detected in Sweden (Sahlström et al., 2015) with a geometrical mean of 0.3 ng/g lw (DF: 23%, LOD: 225 pg/g dw). There is no other report on DP and other dechloranes levels in stool or meconium so far.

4.3.2. Levels of PBDEs in stool samples

Adjusted to the lipid weight, concentrations of tetra-, penta- and hexa-PBDE congeners in stool are lower than concentration in blood, while concentrations of BDE209 are higher in stool than in blood (Sahlström et al., 2015). Possible reason for this may be the size of BDE209, which prevents it from crossing biological membranes to be accumulated in human tissues (Hakk and Letcher, 2003). As a result, BDE209 is excreted from body untransformed in faeces. Although PBDE levels in blood and milk in North America are higher than in other regions in the world, PBDEs have not been measured in stool samples in North America yet. PBDEs in stool samples were only measured in Australia (English et al., 2017; Chen et al., 2015a) and Sweden (Sahlström et al., 2015). BDE209 was the predominant congener in stool samples with levels at least one order of magnitude higher than the other PBDE congeners. For example, the median of BDE209 was 1.5 ng/g dw (dry weight basis), while median values of BDE47, BDE99, BDE100, BDE153 are 0.14, 0.05, 0.05 and 0.01 ng/g dw, respectively, in Queensland, Australia (sampled in 2015-2016) (English et al., 2017). In Uppsala, Sweden only BDE47 and BDE153 were detected at a median concentration of 0.41 and 0.23 ng/g lw (infants’ samples collected in 2009-2012) (Sahlström et al., 2015), while the median level of BDE209 was 18 ng/g lw.

Only one study reported PBDEs in meconium, in which BDE209 contributed 29% of the total concentration of PBDEs, while contribution of BDE47 was 44% (sampled in 2012) (Jeong et al., 2016). In comparison, BDE209 is the main PBDE congener in infant’s stool in Australia (English et al., 2017) and Sweden (Sahlström et al., 2015), accounting for 86% and 97% of the total PBDE, respectively. Relatively low concentration of BDE209 in meconium resembles the profiles of PBDEs in maternal blood.

4.3.3. Levels of other brominated FRs in stool samples

The Swedish study mentioned in 4.3.2. (Sahlström et al., 2015) is the only study that measured other BFRs in stool. In this study, TBB was not detected in any of the sample (LOD: 650 pg/g dw), whereas BTBPE and OBTMPI were detected with a low detection frequency of 13% and 23%, respectively, despite the fact that, like BDE209, BTBPE is also eliminated in faeces unmetabolized (Stapleton et al., 2009; Hakk and Letcher 2003). Their low detection frequency is most likely due to their recent introduction into commerce. The Swedish study also reported TBPH (DF: 50%, LOD: 4.5 ng/g dw) with a median of 9.5 ng/g lw and DBPDE (DF: 100%) with a median of 4.7 ng/g lw.

4.4. Hair

Hair is another non-invasive bio-specimen. It is already used routinely for drug analysis notably in forensic science (Daniel et al., 2004). Levels of HRFs in hair however, should be interpreted with caution because contaminants in hair is often a result of both external absorption and internal uptake (Kucharska et al., 2014; Carnevale et al., 2014). Differences in HFRs contents with the distance to the scalp has been also reported (Carnevale et al., 2014), adding a bias during the sampling process but it also allows to monitor flame retardants concentration over time by segmenting the hair. To be consistent, hair sampling should always be done on the similar portion of the hair (usually at 1 cm from the root) but this is not always respected (Chen et al., 2015b; Qiao et al., 2018). Overall, hair analysis is of interest to assess exposure over a longer period of time compared to blood or stool.

4.4.1. Levels of DP in hair

In the US, s-DP and t-DP have been detected at 0.17 and 0.82 ng/g dw (median of detected samples only) with a detection frequency of 38% and 14% (Liu et al., 2016). In China, the concentrations of individual DP isomers in populations living in non-industrial areas were generally below 1 ng/g dw, which are similar to those in USA (Chen et al., 2019a; Qiao et al., 2018) and (Zheng et al., 2010) (Table S1). Concentration of DP in hair from e-waste recycling area in China however is much higher, ranging from 2 ng/g dw to above 100 ng/g dw (Zheng et al., 2010). The same study shows a clear decreasing trend of DP levels in hair from e-waste recycling plant workers (arithmetic mean = 6.86 and 8.52 ng/g dw for s-DP and t-DP, respectively) to residents living near the e-waste plant (2.48 and 3.6 ng/g dw) and to residents not from e-waste recycling area (<1 ng/g dw). Similar results have been also reported in another Chinese study, showing DP levels around 5 times higher in e-waste recycling plant workers than in residents nearby the industrial sites (Zhang et al., 2013).

Several studies also reported higher concentrations of DP in hair from female than from male (Chen et al., 2015b; Chen et al., 2019a). The reason remains unclear; it might be due to difference in metabolism between genders or because female hair tend to be longer than male hair and therefore is subjected to longer external absorption of contaminants (Chen et al., 2015b).

Figure 3 shows that f_anti in hair (0.76) is similar to f_anti in blood (0.72) for the general population, which is also close to f_anti value of commercial DP (0.60-0.80). Similar to the trend in blood samples, f_anti in hair of occupationally exposed workers is lower at a median of 0.54. It was noticed that f_anti in hair is comparable to the value in dust (Zheng et al., 2010).

Dechlorinated DPs were measured in the three studies in China (Qiao, et al., 2018), (Chen et al., 2015b; Zheng et al., 2010). t-Cl₁₁-DP was only detected in 23.5% in female hair and 7.69% in male hair from the general population (Sampled in 2014, LOD: 0.02 ng/g dw) (Qiao et al., 2018). In the same study, t-Cl₁₀DP was only barely detected (DF <4%, LOD: 0.06 ng/g dw). Similar to DP, arithmetic mean of t-Cl₁₁DP in hair decreased from 0.06 ng/g dw in e-waste recycling plant workers to 0.03 ng/g dw in resident living nearby an e-waste facility (Zheng et al., 2010). Higher median level of t-Cl₁₁DP in female hair (0.42 ng/g dw) than in male hair (0.10 ng/g dw) was also reported (Chen et al., 2015b).

4.4.2. Levels of other Dechloranes in hair

There is no data available on other dechloranes (Dec602, Dec603, Dec604 and HCDBCO) in hair.

4.4.3. Levels of PBDEs in hair

Tetra- and penta- congeners of PBDEs are not frequently detected in hair and their concentrations in China remain low. The mean value of BDE47 was 0.33-0.42 ng/g in rural/urban area and 1.33-1.97 ng/g dw in industrial area (Zheng et al., 2011). Data in France also show low detection frequency and concentration, where BDE47 was detected at < 1.00 ng/g dw and BDE99, BDE100 and BDE153 barely detected (< 25%, LOD of 1.00, 0.60, 0.94 and 0.15 ng/g dw for BDE47, BDE99, BDE100 and BDE153) (Iglesias-Gonzalez et al., 2020) (Table S4). Similar to PBDEs in blood, levels of BDE47 in USA remain high with median value of 33 ng/g (Liu et al., 2016).

BDE209 is the main congener found in hair in all reported studies with detection frequencies around 95-100% and concentrations around 5-10 ng/g dw in non-industrial areas (Liu et al., 2016; Zheng et al., 2011) and up to 41.6, 24.2 and 136 ng/g dw in industrial areas (Qiao et al., 2019; Zheng et al., 2011) (Table S4). BDE209 does not bioaccumulate in the human body (Hakk and Letcher 2003) and transfer of BDE209 from blood to hair is limited (Hakk and Letcher 2003). Therefore, it is reasonable to assume that high concentrations of PBE209 in hair are likely a result of external exposure to dust instead of its enrichment in hair from the blood stream (Liu et al., 2016; Chen et al., 2019a).

4.4.4. Levels of other brominated FRs in hair

TBB has been reported in hair in USA at a median of 85 ng/g dw and TBPH at a median of 78 ng/g dw in the same samples (Table S3) (Liu et al., 2016). An increase in BTBPE levels in hair from residents not associated with e-waste recycling area (arithmetic mean: 0.1 ng/g dw and 0.12 ng/g dw (Zheng et al., 2011)) to residents living near the e-waste recycling plants (arithmetic mean: 0.60 ng/g dw (Zheng et al., 2011)) to e-waste recycling plant workers (arithmetic mean: 1.21 ng/g dw (Zheng et al., 2011) and 3.39 ng/g dw (Qiao et al., 2019)) was well demonstrated in China. Similar trend was also observed for DBDPE (9.57 ng/g dw in residents of reference area and 154 ng/g dw in e-waste recycling plant workers) (Zheng et al., 2011; Qiao et al., 2019). OBTMPI in hair was not reported.

Again, the results of these BFRs could be influenced by the external exposure of hair, mainly in contact with dust. For example, the high concentration of TBB and TBPH in hair was thought to be a result of external exposure, since TBB is metabolised quickly and TBPH does not bioaccumulate efficiently in humans (Liu et al., 2016).

4.5. Nails

Similar to hair, nail is also a non-invasive matrix (Liu et al., 2016; Meng et al., 2020), but it is difficult to make the distinction between external and internal exposures (Chen et al., 2019b).

4.5.1. Levels of dechlorane plus in nail

DP in nail was reported only in USA and China (Table S1). The levels of s-DP and t-DP in USA has median values of 0.61 and 2.00 ng/g dw in fingernail and 0.32 and 0.78 ng/g dw in toenail (sampled in 2014), respectively, with detection frequencies all below 30% (Liu et al., 2016). The total DP in nail in China was reported at a median of 1.22 ng/g dw (sampled in 2016) (Chen et al., 2019b).

4.5.2. Levels of other dechloranes in nail

There is no data available on other dechloranes (Dec602, Dec603, Dec604 and HCDBCO) in nail.

4.5.3. Levels of PBDEs in nail

Similar to the findings in hair, BDE209 is the main congener of PBDEs in nail with its detection in almost all samples (Table S4). One Chinese study reported an arithmetic mean of 540 ng/g dw in nail in e-waste recycling plant workers, compared to about 140 ng/ng dw in general population (sampled in 2016) (Meng et al., 2020). Another Chinese study reported a median of 12.7 ng/g dw for BDE209 (sampled in 2016) (Chen et al., 2019b). Levels of other commonly measured PBDE congeners (BDE47, BDE99, BDE100 and BDE153) in these two studies are lower and less frequently detected (Table S4). Lower BDE209 levels were reported in USA with a median of 7.7 and 8.7 ng/g dw in fingernail and toenail, respectively (sampled in 2014) (Liu et al., 2016). PBDE profiles between the two countries are also different. In USA, BDE47 was more prevalent with 100% detection frequency and its level exceeds that of BDE209 (Liu et al., 2016).

4.5.4. Levels of other brominated FRs in nail

Among other BFRs, TBB was found in nails at a median of 40 ng/g dw (DF of 96%, fingernail) and 91 ng/g dw (DF of 94%, toenail) in USA (sampled in 2014) (Liu et al., 2016), and at 26.7 ng/g dw in China (sampled in 2016) (Chen et al., 2019b). TBPH follows a similar pattern with a median value of 74 ng/g dw in fingernail and 116 ng/g dw in toenail in USA (Liu, et al., 2016), and 28.1 ng/g dw in nail in China (Chen et al., 2019b). DBDPE was only reported in one study at a median of 7.47 ng/g dw (Chen et al., 2019b) (Table S3).

Conclusions

The number of halogenated flame retardants (HFRs) that are being monitored has expanded over the past decades. The addition of emerging HFRs to the legacy PBDEs in quantification poses new challenges in both sample preparation and instrumental analysis. While traditional SPE and LLE methods are still commonly used, they need to be improved. Modified procedures in both extraction solvents and choice of new sorbents have been introduced to improve performance. In addition, new techniques such as ASE and QuEChERS are gaining interest.

Despite of sharing common lipophilic properties among PBDE congeners, difference in volatility is a challenge in PBDEs analysis. Notably, although being a dominant congener among PBDEs, BDE209 is often left out in the analysis of PBDE congeners because of its low volatility and sorption onto activated surfaces. Emerging HFRs have similar issues; some of them are extremely low volatile, including OBTMPI and DBDPE. As a result, GC-MS conditions including injection port temperature in GC and ionization source temperature in MS require careful balance to accommodate large number of HFRs with a wide range of volatility.

Different lengths of GC columns must be considered for optimal separation of HFRs in different applications. In general, a 30-meter column delivers the best separation for the large number of HFRs, while a shorter column is better for separating thermo-labile or low volatile HFRs.

ECNI remains the most common choice of ionization in MS for HFRs. It is very sensitive for most of HFRs, especially BFRs such as PBDEs. Since most BFRs produce bromine fragment in MS under ECNI as the major signal to be used for quantification, ECNI cannot distinguish BFRs that co-elute from GC column. Such a lack of selectivity in the MS detector results in a high demand for better separation of these HFRs in GC columns, which becomes increasingly challenging with expanding number of HFRs to be measured in a single analysis. New ionization technology such as APCI or ICP has been recently introduced to complement ECNI.

New information on the levels of HRFs in various human bio-specimens has become available in the past decade. These data are summarized in four tables in the Supplementary data (Tables S1, S2, S3 and S4), each table deals with a particular group of HFRs. These data provide a baseline information on the presence of HRFs in humans. On the global scale, among the HFRs, PBDEs still have the highest levels in humans, followed by DP. Of particular interest is the levels of emerging HFRs that are being introduced in commerce as replacement of the legacy HFRs such as PBDEs; biomonitoring information is critical for the development of regulations on these emerging HFRs. High levels of HFRs including several emerging HFRs in people working in the e-waste recycling plants or living near e-waste facilities are well documented in China. It is worth noting that reported HFRs data are only limited to a few countries, mainly located in North America (Canada, the US, and Mexico), Europe and in East Asia (China and South Korea). There is little to no data from other countries and regions such as South America, Russia and India, as well as many developing countries that are currently major importers of e-wastes containing flame retardants.

• Halogenated flame retardants (HFRs) reported from 2010-2020

• Both PBDEs and emerging HFRs are reviewed

• Advances and challenges in analytical methods are highlighted

• Concentrations of HFRs in humans are summarized and synthesized

• Higher human exposure to HFRs is associated with industrial settings