Review on age-specific exposure to organophosphate esters: Multiple exposure pathways and microenvironments

Abstract Organophosphate esters (OPEs) widely exist in the environment, raising increasing concerns about their potential health risks. This comprehensive review surveyed the occurrence of OPEs over the last decade in indoor microenvironments (i.e. residence, in-vehicle, office, and school/daycare center), outdoors, foodstuffs, drinking water, and human-related specimen (i.e. breast milk and urine) with a view to unraveling age-specific exposure to OPEs. Multiple exposure pathways including inhalation, dermal absorption, dust ingestion, and dietary ingestion were considered to prioritize their relative importance. The results showed that dietary ingestion was the main contributor, followed by dust ingestion, regardless of age. A healthy diet with less contaminated food can effectively reduce OPE intake. OPE concentrations in air and dust followed the sequence of in-vehicle and office > school/daycare center > residence > outdoors. Compared to other indoor settings, exposure in schools/daycare centers and offices contributed to a greater OPE intake in non-adults and adults, respectively. The estimated daily intake of OPEs followed the sequence of infants > toddlers > children > teenagers > adults > elderlies. Overall, OPEs posed low health risks to all age cohorts, but infants were vulnerable and subject to the highest risk, largely attributed to breast milk ingestion. This review highlights the need for more toxicity and bioaccessibility studies on OPE mixtures and metabolites to further refine the health risk assessment of OPEs. Graphical abstract

Some PBDEs have been phased out due to their toxic effects, such as endocrine-disrupting, neurotoxic and carcinogenic effects (USEPA, 2017;Wu et al., 2020). However, OPEs can also potentially induce similar toxicities, for example, endocrine-disrupting, neurotoxicity, hepatoxicity, and immunotoxicity, causing developmental and reproductive impairment in mammals (Al-Salem et al., 2020;Canbaz et al., 2017;Hu et al., 2017;Li et al., 2017;Yan & Hales, 2021). In particular, the adverse effects induced by metabolites of several OPEs may be comparable to or even stronger than those of OPEs Zhang, Yu, et al., 2020). The levels of OPEs found in the air and dust were generally much higher than those of PBDEs, which also holds true for the bioaccessibility of dust ingestion (de la Torre et al., 2020;Fang & Stapleton, 2014;Saini et al., 2019;Shoeib et al., 2019;Yadav et al., 2019). These findings imply that OPEs, as replacements of PBDEs, may not be safe.
The majority of people spend most of their time indoors, for example, Americans spend approximately 90% of their time indoors on average. This emphasizes the importance of indoor environment quality (USEPA, 2021). Concurrently, OPEs are present more abundantly in indoor air and dust than in outdoor settings Khairy & Lohmann, 2019;Wong et al., 2018). Exposure to OPEs in indoor settings via inhalation, dermal absorption, and dust ingestion might contribute significantly to human body burdens of OPEs, particularly for infants and toddlers who are susceptible to OPE intake via mouthing behavior Xue et al., 2007). Besides, OPEs occur widely in a variety of food categories and therefore dietary ingestion should be of great concern (Poma et al., 2017;Poma et al., 2018;Wang & Kannan, 2018;Zhao et al., 2019). A good example is the exposure of infants to OPEs via breastfeeding Kim et al., 2014;Ma et al., 2019). Since exposure patterns are age-specific, it is necessary to characterize age-specific exposure with consideration of multiple exposure pathways and microenvironments. But previous studies have mainly focused on dietary exposure, without considering other exposure pathways via microenvironments (Gbadamosi et al., 2021;Li, Zhao, et al., 2019).
In view of the above-mentioned issues, the present review aims to (1) examine and prioritize the contributions of different pathways and indoor microenvironments for age-specific exposure to OPEs; (2) overview the exposure levels of OPEs in general population and their health risks; and (3) provide valuable insights into human health risk assessment and feasible strategies to minimize the daily intake of OPEs for different age cohorts. Publications between January 1, 2010 and March 1, 2021 were retrieved for data compilation and analysis, and suggestions or orientations are provided to aid in future studies.

Applications of organophosphate esters
The OPEs are commonly used as additives, such as plasticizers and/or flame retardants, in polymers and as extreme-pressure and anti-wear additives in engine oils, lubricants, and hydraulic fluids. Their end-use applications cover construction materials, electronics, electricals, automotive materials, textiles, adhesives, and coatings. Particularly, OPEs have been widely utilized in the production of polyurethane (PU), polyvinyl chloride (PVC), engineering thermoplastics, and acrylonitrile butadiene styrene (ABS) (Lucintel, 2016). Individual OPEs in commonly used polymers can be elucidated in product descriptions by main suppliers worldwide (Table 1). Basically, aryl OPEs are used as flame retardants, whereas alkyl OPEs are applied as plasticizers for plastics and synthetic rubber, because of superior thermal stabilization and plasticizing properties for aryl and alkyl OPEs, respectively (Green, 1992).
TPhP and novel OPEs, for example, BPADP, CDP, RDP, and IPTPP, have been extensively employed as plasticizers and/or flame retardants in polymeric materials (Table 1), while TPhP is somewhat mixed with other OPEs for wider applications. The enclosures of computers and other electronic devices usually consist of polycarbonate/acrylonitrile butadiene styrene (PC/ABS) blends, which are often flame-retarded with TPhP. Thus, TPhP has been abundantly detected in computer monitors, boards, keyboards, LCD TV panels and covers, and AC adaptors (Kajiwara et al., 2011;Saito et al., 2007). Apart from electronic and electrical devices, TPhP has also been applied in automotive parts, for example, lacquers, varnishes, hydraulic fluids, and lubricant oils. Marklund et al. (2005) demonstrated that TPhP occurred in vehicle lubricants and waste oils as well as in aircraft turbo, engine, and accessory oils. Besides that, TnBP and TMPP are also frequently used as extreme-pressure and anti-wear additives in hydraulic and lubricant oils for preventing surface damage (Marklund et al., 2005). As a plasticizer and solvent, TnBP is utilized in cellulose esters, lacquers, plastic, and vinyl resins. Besides, TnBP is also an antifoam agent and  has wide applications in manufacturing and packaging of latex paints and floor polishes (ICL, 2021). TMP and TEP are used as plasticizers and flame retardants for rigid PU (RPU) foam, unsaturated polyester (UPE) resins, and anti-coloring agents for polyethylene (PE). TEHP with cold resistance could be used as a plasticizer for synthetic rubber and PVC. TCIPP and TDCIPP are widely used in RPU, flexible PU (FPU) foam, and epoxy and phenolic resins. Rigid PU foam is mostly employed in construction, piping, and packaging for thermal insulation, whereas FPU is primarily used as a cushioning material in furniture, bedding, carpet underlay, automobiles, and packaging (Hill, 2003). TCIPP and TDCIPP showed high detection frequencies in furniture foam, accounting for 0.5-2.2% and 1-5% of the foam weights, respectively (Stapleton et al., 2009). In addition to flame retardation, TCIPP possesses antistatic and softening properties as well as resistance to low temperature and moisture, and therefore is widely applied in laminate roofing and spray formulations. For example, PVC wallpapers have been found to contain TCIPP (Ni et al., 2007). Owing to its resistance to flame, low temperature, and ultraviolet light, TCEP is primarily employed in cellulose-nitrate-and cellulose-acetate-based materials, fireproof paint, and plastics, and can also be used in PU, polyester, and acrylic resin. As TCEP is suspected to be carcinogenic and toxic to mammalian reproduction, it is subject to risk evaluation proposed by the European Chemical Agency (ECHA) and United States Environmental Protection Agency through the Toxic Substances Control Act (ECHA, 2020;USEPA, 2019). This in turn enforces certain restrictions on the production and application of TCEP, especially in childcare supplies, residential upholstered furniture, and textile.
TBOEP has wide applications in a variety of materials besides PVC and FPU. For instance, TBOEP can be used as an antifoaming agent in ore beneficiations, or a leveling agent in waxes, floor polishes, and paper coatings, or a softener in chlorinated rubber and nitrile formulations (e.g. seals, gaskets, hoses, and shoe soles) (ICL, 2021). EHDPP can be added to flexible PVC as a plasticizer for low smoke, which also serves a flame retardant in thermoplastic PU, nitrile rubber, and cellulose acetate (LANXESS, 2021).
Novel OPEs, including BPADP, CDP, RDP, IPTPP, TXP, TMPrP, and TBPP, may become increasingly prevalent owing to their wide applicability, especially in electronic and electrical segments using PC/ABS and polyphenylene ether/high-impact polystyrene blends (PPE/HIPS). Generally, most conventional alkyl, aryl, and chlorinated OPEs can serve as plasticizers and/or flame retardants for PVC (Table 1) but are seldom applied in the production of polypropylene (PP). PE with small amounts of OPEs and PP materials are recommended for food packaging and drinking water storage. Use of less PVC materials indoors, such as non-PVC flooring and wallpapers, may effectively reduce exposure levels of OPEs. However, some proprietary OPErelated flame retardants have not been thoroughly evaluated yet, so that their environmental occurrence and health impacts remain unknown.

Non-dietary exposure sources: Indoor microenvironments and outdoor settings
The general population spends approximately 90% of the time indoors, and therefore is subject to intensive exposure to indoor air and dust (USEPA, 2021). The occurrence of OPEs in air and dust is therefore the main focus for non-dietary exposure. As OPEs are additives in consumer products without chemical bonding, they can be readily released into air and dust via volatilization and abrasion (Cao et al., 2014;Reemtsma et al., 2008). The OPEs with high vapor pressure are likely to emit to air via volatilization, and then sorb to the particulate phase. OPEs can also occur in particles, especially in large ones, via abrasion (Cao et al., 2014). Therefore, OPEs can migrate from consumer products to particles through three pathways: (1) sorb to particles from the gaseous phase upon volatilization; (2) transfer from abrasion-proof materials to fine particles through abrasion; and (3) transfer from easy-to-wear materials to relatively large particles through abrasion (Cao et al., 2014). Many studies have demonstrated that OPEs are not evenly distributed in different sizes of atmospheric particles and/or dust (Cao et al., 2014;Cao, Zhao, et al., 2019;Li, Qiu, et al., 2019;Yang et al., 2014;Zhou & Puttmann, 2019). In indoor air, for example, in office, TPrP, TnBP, TCEP, and TCIPP showed similar unimodal distribution patterns peaking in coarse particles (4.7 À 5.8 lm), whereas TDCIPP, TMPP, TBOEP, and TEHP predominantly occurred in fine particles (< 2.5 lm) (Yang et al., 2014). In outdoor air, OPEs were slightly different in each size fraction, but were similar in coarse and fine particles in urban areas (Luo et al., 2016) and during a haze episode (Cao, Zhao, et al., 2019).
The distribution of OPEs in dust is generally variable with particle sizes, and tends to cluster toward fine fractions (Cao et al., 2014;He, 2019;Li, Qiu, et al., 2019;Zhou & Puttmann, 2019). For example, high abundances of OPEs occurred in indoor dust particle sized approximately 900, 100, and 10 lm (Cao et al., 2014). The concentrations of OPEs were the highest in particle sizes < 43 lm, and somewhat increased with decreasing particle size in indoor dust in offices, cars, and public microenvironments . The OPE concentrations also increased with decreasing particle size in industrial road dust, but exhibited an uneven distribution pattern in main road dust peaking in size fractions of 90 À 150 and < 75 lm (He, 2019). Apparently OPEs can be derived from a variety of sources. The size distribution pattern of dust-associated OPEs is generally affected by the physicochemical properties of OPEs [e.g. vapor pressure and n-octanol À water partition coefficient (K ow )] and dust (e.g. total organic carbon (TOC) content and surface area), use pattern of OPEs, and migration pathways (Cao et al., 2014;. As thus, particle size distribution may assist in source diagnostics. The concentrations of 12 conventional OPEs, including TEP, TPrP, TiBP, TnBP, TCEP, TCIPP, TDCIPP, TPhP, TMPP, EHDPP, TEHP, and TBOEP, in air and dust globally in the recent decade are summarized (Tables S4 and S5). The sampling setting are categorized as residence (including home and dormitory), in-vehicle, school/daycare center, office, and outdoors. The mean levels of OPEs were higher in offices and vehicles but were the lowest in residences, either in indoor air or dust ( Figure 1). This pattern can be attributed to (1) lower ventilation in offices and vehicles; (2) higher temperatures in vehicles favor for emissions of OPEs, especially TCIPP, from electronic equipment, foam, and insulating materials; and (3) greater emission potential of OPEs, which is affected by the amounts of related insulation and building materials, furniture, and electronic devices, as well as their service life and operation time (Kemmlein et al., 2003;Liagkouridis et al., 2017;Liang et al., 2019;Sugeng et al., 2018;Wang et al., 2019). In addition, the OPE levels were much higher in indoor microenvironments than in outdoor settings, particularly in dust. This suggests that human exposure to indoor OPEs should be of great concern and additional efforts are needed in characterizing OPE levels in different indoor microenvironments. Among individual OPEs, TCIPP was the most abundant component in the air of offices (84.6%), residences (66.4%), and vehicles (52.3%). This pattern was probably resulted from widespread applications of TCIPP in PVC, polystyrene, and foam materials, such as PVC flooring, wallpaper, upholstered furniture/mattresses, and insulating materials. By comparison, TBOEP was the most dominant component of atmospheric OPEs in school/daycare centers (52.0%) and outdoors (36.7%), indicating different prevailing emission sources in these settings. Regarding indoor dust, TBOEP was also predominant in dust of offices (87.7%), school/daycare centers (85.6%), and residences (36.2%). In general, TBOEP had much higher abundances in dust than in air, because it would emit from products into air at low rates due to its low vapor pressure and migrate to dust largely through abrasion (Cao, Lv, et al., 2019;Saito et al., 2007). Consumer products and building materials abundantly occur in residences, schools/daycare centers, and offices and are readily subject to wear and tear, resulting in widespread occurrence of TBOEP in dust. The predominance of TDCIPP in vehicle dust was probably resulted from its widespread usage in PU foam with low cost and excellent stability for car manufacture. Overall, TCIPP and TBOEP are abundant in air and dust of indoor microenvironments and their health effects on humans merit more attention, such as their combined toxicities with matrices in air or dust.

Dietary exposure sources: Foodstuffs/breast milk/drinking water
OPEs are frequently found in food items, with levels and compositions varying widely with food categories Poma et al., 2017;Poma et al., 2018;Wang & Kannan, 2018;Zhao et al., 2019). The OPE concentrations ranged from 1.1 to 9.6 ng g À1 wet weight in food items from a local market in eastern China, and were the highest in grains (Ding et al., 2018). Wang and Kannan (2018) found that median OPE concentrations in meats and fish were higher than those in cooking oil, grains, and dairy products from a local market in Albany, New York, USA. As grains and meats/fish are the staple foods for Chinese and Americans, respectively, dietary ingestion of these staple foods should be of great concern.
Fats/oils from a Belgian local market and a Swedish food market had the highest mean OPE concentrations among all foodstuffs and might be a significant potential source for dietary exposure to OPEs for Europeans (Poma et al., 2017;Poma et al., 2018). In addition, EHDPP was the most abundant component among OPEs in most foodstuffs, except vegetables collected from Sweden, consistent with what was observed in Chinese foodstuffs (Poma et al., 2017;Zhao et al., 2019). However, the OPE profiles were considerably variable with food matrices and regions, apparently ascribed to different sample types and bioaccumulation factors of OPEs. Cross contamination from food processing and packing procedures may also contribute a fair amount of OPEs to foodstuffs (Poma et al., 2017;Poma et al., 2018;Wang & Kannan, 2018;Zhao et al., 2019). Processed foodstuffs have significantly higher OPE concentrations than non-processed foodstuffs and consequently contribute more to consumers' dietary intake (Poma et al., 2017;Zhao et al., 2019). Wang and Kannan (2018) found that OPEs were detected in food packages with a median concentration of 132 ng g À1 . However, data are limited and insufficient for understanding how OPEs migrate from food packages to foodstuffs.
Foodstuffs are generally divided into the categories of meats, fish (including shellfish), vegetables, fruits, dairy products, grains, eggs, and fats/oils (USEPA, 2011) (Table S6). Overall, mean concentrations of 12 OPEs followed the sequence of fats/oils > meats > grains > vegetables > fish (from the markets) > dairy products > eggs > fruits (Figure 2a). Residues of OPEs in raw materials of fats/oils and cross contamination during food processing may have significantly contributed to higher OPE levels in fats/oils than in other food categories (Li, Zhao, et al., 2020;Poma et al., 2017;Poma et al., 2018). Leaching from food packages may have also been another contributor, since OPEs are widely added to plastic packages and lipophilic OPEs tend to adhere to fats/oils (K€ uhn et al., 2020;Li, Zhao, et al., 2020;Wang & Kannan, 2018). Therefore, packaging materials with fewer OPE additives (e.g. PP and PE) are recommended for fats/oils storage.
Apart from fats/oils, meats and grains had the highest OPE levels and thus merit utmost attention as they are the main food items in daily life. It should be noted that dietary patterns differ among age groups; for example, infants often have a special breast milk diet. The OPE concentrations in breast milk were 22 ng g À1 lipid weight (l.w.; median value) in Japan, 73.5 ng g À1 l.w. (mean value) in Spain, 3.61 ng mL À1 (mean value) in the United States, and 10.6 ng mL À1 (median value) in China (Beser et al., 2019;Chen et al., 2021;Kim et al., 2014;Ma et al., 2019). The varying OPE concentrations in breast milk may be closely related to nursing mothers' exposure levels, which can be affected due to the use of products with flame retardants and mothers' personal characteristics Kim et al., 2014).
Drinking water is another important exposure source of OPEs. Because OPEs are omnipresent in water sources, and they can be introduced to potable water through purification processes Figure 2. (a) Box-whisker plot (box represents 25th, 50th, and 75th percentiles, whiskers represent 10th and 90th percentiles) showing the concentrations of total OPEs in different food categories (ng g À1 wet weight), breast milk, and drinking water (ng mL À1 ). Dotted line represents the mean value. Dark and white dots represent minimum and maximum values, respectively. (b) Average amounts of the daily intake of OPEs (ng kg bw À1 day À1 ) by dietary exposure for different age cohorts, estimated with data (n ¼ 74) in Table S6. Significant differences between age cohorts were discerned for dietary intake (Mann-Whitney U test, p < 0.001). (Ding et al., 2015;Lee et al., 2016;Li, He, et al., 2019). Ding et al. (2016) found that tap water was the most OPE-contaminated water source among different types of drinking water in eastern China with a median concentration of 192 ng L À1 , followed by direct drinking, barreled, well, and bottled water, with median concentrations of 59.2, 27.6, 4.50, and 3.99 ng L À1 , respectively. Li et al. (2014) also demonstrated that tap water was contaminated by a greater amount of OPEs than bottled water, which can be partially attributed to the prevalent use of PVC pipes containing residual OPEs . Although barreled and bottled water appeared to have lower content of OPEs than tap water, potential leaching of OPEs from plastic containers should also raise concern. Barreled and bottled water is commonly stored in PC and polyethylene terephthalate (PET) containers, respectively (Ding et al., 2015), which contain a range of OPEs (Table 1). Prolonged or high-temperature storage can increase the amounts of OPEs leached from the plastic containers to barreled and bottled water (Hou et al., 2021;Xiao et al., 2020). Therefore, short-time and room-temperature storage is recommended to reduce OPEs leaching from plastic containers to barreled and bottled water.

Inhalation intake
The inhalation intake of OPEs from bulk air can be estimated using Equation (1) of Table S1. Particle size greatly influences the deposition efficiencies of aerosols in the human respiratory tract and subsequently the amounts of particle-bound OPEs deposited in the head airways, tracheobronchial region, and alveolar region (Hinds, 1999). Therefore, inhalation intake of OPEs may be underestimated or overestimated if particle size or bioaccessibility is not considered. Wannomai et al. (2021) suggested that the bioaccessibility values of inhaled TCEP, TCIPP, and TDCIPP in indoor dust were 5.4-19%, 4.6-35%, and 3.2-22%, respectively, at sizes <250 lm. However, such inhalation bioaccessibility may differ from that of atmospheric particles. Zeng et al. (2019) found that TCEP, TCIPP, and TDCIPP in atmospheric fine particulate matter (PM 2.5 ) had higher inhalation bioaccessibility values (55-97%) than TPhP (26%), TMPP (2.3%), and EHDPP (1.2%). Data on the inhalation bioaccessibility of atmospheric OPEs remain scarce.
Future studies should consider two factors in assessing inhalation bioaccessibility. The first factor is the components and particle sizes of aerosols, because aerosols are usually source-specific and their components (e.g. TOC and metal ions) may affect the kinetics of OPEs in bioaccessibility assays. More importantly, particle sizes investigated should cover the majority of aerosols deposited in the respiratory tract (Kastury et al., 2017;Zeng et al., 2019). The second factor is the compositions and pH of artificial lung fluids. Since Gamble's solution and artificial lysosomal fluid have different pH values, OPEs could somewhat be degraded in artificial lung fluids . The stability of OPEs in artificial fluids should be considered, particularly hydrolysis of OPEs triggered by pH. Aryl OPEs are known to degrade rapidly in artificial lung fluids and also are more readily subject to hydrolysis with increasing alkaline pH, as compared to other OPEs (Su et al., 2016;Zeng et al., 2019). Clearly, hydrolysis of OPEs at different pH values combined with solution types needs to be clarified, and artificial lung fluids should be optimized to truly reflect the inhalation bioaccessibility of OPEs in human lungs.

Assessment via environmental media
Previous studies have indicated that dermal absorption is an important pathway for OPE intake and is probably more significant than inhalation (Cao, Zhao, et al., 2019;Tao et al., 2019;Yadav et al., 2019). Abdallah et al. (2016) obtained dermal absorption efficiencies of TCEP, TCIPP, and TDCIPP at 28%, 25%, and 13%, respectively, using an ex vivo skin model, and at 16%, 11%, and 9%, respectively, using the EPISKIN TM model. Frederiksen et al. (2018) also revealed that TCEP and TCIPP could rapidly penetrate the skin in a Franz diffusion cell system, whereas TDCIPP and TPhP had lower transdermal permeability coefficients, i.e., the percutaneous penetration rates decreased in the sequence of TCEP > TCIPP ! TBOEP > TiBP ! TnBP > TDCIPP > TPhP > TMPP. These studies indicated that the transdermal permeability coefficients of OPEs were negatively correlated with log K ow (Abdallah et al., 2016;Frederiksen et al., 2018). However, dermal absorption efficiencies varying among studies imply that robust methods should be developed and validated to assess the transdermal delivery of OPEs. Besides, the transdermal permeability coefficients of gaseous OPEs have rarely been reported. A model developed by Nazaroff (2012, 2014) indicated that the transdermal permeability coefficients of gaseous alkyl OPEs except TBOEP were 0.09-0.64 m h À1 , much lower than those of gaseous chlorinated OPEs (1.5-5.4 m h À1 ) Thus, dermal absorption of gaseous chlorinated OPEs in indoor air is of more concern (Hu et al., 2019).
Dermal intake of OPEs from dust can be estimated by Equation (2) of Table S1. Pawar et al. (2017) showed that the dermal bioaccessible fractions of TCEP, TCIPP, and TDCIPP via dust were 10%, 17%, and 19%, respectively, using an in vitro physiologically based extraction test (PBET). However, most studies have focused mainly on dermal absorption of OPEs via dust, but less on those via water, gas, and atmospheric particulate matter (Ali et al., 2018;Cao, Lv, et al., 2019;Yadav et al., 2019). This knowledge gap needs to be filled to fully understand dermal absorption of OPEs.

Assessment via hand wipes
Approximately 50-800 mg of dust can adhere to toddlers' hands per day (Sugeng et al., 2017). Duggan et al. (1985) found that most dust particles adhered to schoolchildren's hands had sizes <10 lm. Hee et al. (1985) also suggested that fine particles with sizes <246 lm were more readily adhered to children's hands. This indicates that the adhesive ability of dust varies with particle size on hands and even on skin surface. It should be noted that the occurrence of OPEs in dust could be size-dependent, and the dermal absorption efficiency of OPEs might also be closely related to dust particle size (Cao et al., 2014;Lao et al., 2020;Mercier et al., 2011). Bulk dust, therefore, may not robustly indicate dermal absorption efficiency by skin contact. Moreover, dermal absorption of OPEs can occur via direct contact with synthetic products (Keller et al., 2014). Therefore, only OPE concentrations in dust may not be sufficient for evaluating human dermal absorption. Hand wipe may be an alternative tool for dermal absorption assessment. The OPE levels in hand wipes were associated with human exposure levels, for example, their association with urinary metabolites, including TPhP with DPhP and TDCIPP with BDCIPP (Hoffman, Garantziotis, et al., 2015;Tao et al., 2018). The levels of TCIPP, TDCIPP, and TPhP in children's hand wipes were also significantly associated with those in dust (Tan et al., 2018). Interestingly, Yang et al. (2019) found that most OPEs in hand wipes were significantly related to those in cell phone wipes. Since hand wipes are collected individually and are noninvasive, the results can be more reliable for different age cohorts. Hence, use of hand wipes is a feasible and alternative approach for estimating human dermal absorption. Residues of OPEs collected in hand wipes can be applied to estimate dermal absorption by Equation (3) of Table S1.
Gauze pads and isopropyl alcohol are the most common materials used for hand-wipe sample collection (Hoffman, Garantziotis, et al., 2015;Liu et al., 2018;Xu et al., 2016), and twill may be a more suitable material than gauze for wiping OPEs off hands (Beaucham et al., 2019). In addition to wiping materials, the physiochemical properties of OPEs (e.g. lipophilicity), solvents, and the number of sequential wipes may also affect the efficiency of hand wiping (Beaucham et al., 2019). Therefore, robust methods of hand-wipe collection are required to develop for refining the assessment of human dermal absorption.

Dust ingestion intake
The daily intake of OPEs through dust ingestion can be calculated by Equation (4) of Table S1. Numerous studies suggested that ingestion bioaccessibility should be considered in estimating ingestion intake of dust-associated OPEs, which is greatly affected by the physicochemical properties of target compounds and dust types, including molecular weight, water solubility, TOC content, and particle size (Fang & Stapleton, 2014;He et al., 2016;Quintana et al., 2017;Wannomai et al., 2020). These studies indicate that TCEP and TCIPP have higher ingestion bioaccessibility than other investigated OPEs. Ingestion bioaccessibility of OPEs were negatively correlated with log K ow , largely within the range of 5-11, and their bioaccessible fractions were smaller than those with log K ow < 4.8 (Fang & Stapleton, 2014;Guo et al., 2019;Wannomai et al., 2020). Combined with log K ow , TOC content can remarkably influence the partitioning of hydrophobic OPEs between dust and gastric fluid, and higher TOC content in dust can impede the release of more hydrophobic OPEs into the gastric fluid, resulting in lower ingestion bioaccessibility (He et al., 2016).  indicated that OPE ingestion bioaccessibility in the size fraction of 200-2000 lm were much greater than those in the size fraction of <43 lm. But Fang and Stapleton (2014) suggested that the OPE bioaccessibility decreased with increasing particle size. These findings indicate that the dust size distribution might be a significant factor in dust ingestion bioaccessibility of OPEs, but so far inconsistent results have been reached probably because of the source specificity of dust. Therefore, future studies should also address the size dependency of ingestion bioaccessibility.
Because conventional PBET without considering dynamic absorption may result in underestimation of the bioaccessible fraction for hydrophobic OPEs, an absorbent is recommended for PBET to maintain dynamic absorption in the gastrointestinal tract. For instance, Tenax beads were found to improve bioaccessibility of OPE, particularly for TPhP (Fang & Stapleton, 2014). However, artificial digestive fluids may not be identically representative of human digestive fluids, and one type of absorbent cannot maintain dynamic absorption for all OPEs due to a wide range of physicochemical properties for OPEs. Therefore, selecting an appropriate absorbent remains a challenge. Only focusing on mass transfer from dust to digestive fluid without considering other matrices in the gastrointestinal tract can also lead to large uncertainties. For example, Zeng et al. (2019) found that the addition of food nutrients (proteins and carbohydrates) can increase the oral bioaccessibility of OPEs in PM 2.5 , especially for TPhP, EHDPP, TMPP, and TEHP. The lack of considerations on digestive tract absorption and metabolism may underestimate the ingestion bioaccessibility of target compounds (Fang & Stapleton, 2014;He et al., 2016). Therefore, the results from in vivo and in vitro studies should be compared to each other to better understand the potential mechanisms influencing the ingestion bioaccessibility of OPEs.

Dietary ingestion intake
Dietary ingestion via foodstuffs, breast milk, and drinking water can be estimated by Equations (5), (6), and (7), respectively (Table S1). Few studies have focused on the bioaccessibility of OPEs via food ingestion; therefore, 100% bioaccessibility is commonly assumed and applied for estimating OPE dietary intake. However, previous studies indicate that fat content and cooking method greatly impact the bioaccessibility of PBDEs, that is, low fat content and cooking processes (e.g. boiling, frying, and steaming) could result in decreased PBDEs bioaccessibility (Mi et al., 2017;Yu et al., 2010;Zhang, Tang, et al., 2020). These findings provide valuable insight into the bioaccessibility of OPEs through food ingestion. Although cross contamination from food processing may increase OPE levels in foodstuffs, cooking processes may possibly decrease the bioaccessibility of OPEs due to thermal degradation or transformation (Zhang, Tang, et al., 2020). As thus, cooking processes may be a feasible way to reduce intake of OPEs via dietary ingestion. It should be noted that the degradation or transformation products of OPEs from cooking processes may possibly pose another health risk, but their identification, bioaccessibility, and toxicity remain unclear and need investigation. The bioaccesibility of OPEs impacted by fat, protein, carbohydrate, and fiber content of foodstuffs and cooking methods also remain largely unknown and need to be further evaluated.

Estimated intake through multiple exposure pathways by internal exposure
Because the bioaccessibility and bioavailability of OPEs are difficult to evaluate, internal exposure has been widely adopted to estimate OPE body burden Fromme et al., 2014). Body fluids, such as urine and blood (e.g. plasma and serum), are generally employed to estimate body burden; particularly, urine is frequently used as a noninvasive sample to understand OPE intake Chen et al., 2018;Kim et al., 2014;Li, Dong, et al., 2019). Generally, the body burden calculated by urinary metabolites using equation (8) of Table S1 integrates all exposure pathways of OPEs.
OPEs can be biodegraded in human body by phase I and phase II metabolisms, and some of these metabolites are excreted via urine (Hou et al., 2016). As metabolites, DnBP, BCEP, BCIPP, BDCIPP, DPhP, and BBOEP are commonly paired with TnBP, TCEP, TCIPP, TDCIPP, TPhP, and TBOEP, respectively, for body burden estimation Fromme et al., 2014;Hou et al., 2016;Li, Dong, et al., 2019). The molar fraction of metabolite to parent OPE (F ue ) is a crucial parameter for this estimation, but has undergone limited investigation. Van  Other factors may have also caused the aforementioned inconsistency. For instance, a diester metabolite might not always pair with an exclusive OPE; DPhP can be derived from the degradation of TPhP, and even from EHDPP, RDP, and BPDPP (Ballesteros-Gomez et al., 2015;Heitkamp et al., 1985). Apart from metabolism, diester OPEs in human body might originate from direct intake, because diester OPEs could be introduced to environmental media or foodstuffs from industrial applications, as well as biotic and abiotic degradation of triester OPEs (Fu et al., 2017;Li, Zhao, et al., 2020;Liu et al., 2021;Su et al., 2016). Therefore, care should be exercised in using F ue , for example, direct intake of diester OPEs should be considered. Apart from diester metabolites, hydroxylated OPEs (OH-OPEs) are the potential metabolites of OPEs, but are rarely reported in human urine and seldom employed for intake estimation Fromme et al., 2014;Hou et al., 2016;Li, Yao, et al., 2020). This is probably because the metabolic levels of OH-OPEs may not be comparable to those of diester OPEs in human, or most of them could not be quantified due to the lack of commercial standards (Li, Yao, et al., 2020); they were often detected by high-resolution mass spectrometry. Use of OH-OPEs, seemingly better in estimating body burden of OPEs, should be further explored in future studies.

Hazard identification and dose-response assessment
Based on in vivo acute toxicity, the median lethal dose (LD 50 ) of individual OPE is greater than 50 mg kg À1 in oral and dermal administrations, showing no to low level of concern for this endpoint. In contrast, the median lethal concentration (LC 50 ) of a given OPE is greater than 1 mg L À1 in inhalation administration, indicating low to moderate level of concern (Table S9) (Illinois EPA, 2007). Inhalation exposure to OPEs is, therefore, of great concern. However, there are certain variations on effective doses related to different exposure routes, species, and sex (Table S9).
In vivo chronic toxicity studies on OPEs remain limited and might not be efficient enough to reveal the toxicities of OPEs for humans.
In vitro assays suggest toxicities of OPEs are non-negligible, which include developmental and reproductive toxicity, immunotoxicity, hepatotoxicity, neurotoxicity, etc.; exposure to OPEs in vitro can trigger agonistic or antagonistic effects on nuclear receptor activities or alter related gene expressions, eventually posing adverse effects on mammalian development, reproduction, and metabolism (Al-Salem et al., 2020;Canbaz et al., 2017;Hu et al., 2017;Li et al., 2017). The toxicity data of individual OPE on mammalian are still sparse, either in vivo or in vitro, especially for TMP, TPrP, TIBP, and novel OPEs (Table S9). Since the mode of action and mechanisms of absorption, distribution, and metabolism remain unclear, genomics, proteomics, and metabolomics will be conducive to understand these mechanisms and the margin of exposure between in vivo and in vitro (Morris et al., 2014;Zhang et al., 2022). Moreover, the combined toxicities of OPEs needs to be evaluated thoroughly, particularly for TPhP with other OPEs (Araki et al., 2020;Zhang et al., 2020b). For example, it has been evidenced that co-exposure to TPhP and TDCIPP can synergistically effect gene expressions and cellular activities ; TPhP combined with TCIPP is associated with rhino-conjunctivitis (Araki et al., 2020). The dose levels and ratios of OPE congeners, as well as sample matrices related to the environment, are crucial factors for combined toxicity and should be carefully assessed in future studies.

External exposure to OPEs
The global consumption of OPEs likely reached the largest growth rate around 2010, and still sharply increased in the recent decade ( Figure S1). It is necessary to fully evaluate exposure of general population to OPEs in the recent decade, which can give some directions to minimize OPEs exposure and associated health risk. The age cohorts were introduced to better elucidate the impacts of exposure pathways and microenvironments on human health, and were divided into the following groups: Infants (< 1 year), toddlers (1 to <3 years), children (3 to <12 years), teenagers (12 to <18 years), adults (young and middle-aged; 18 to < 65 years), and elderlies (!65 years). Exposure pathways, including inhalation, dermal absorption and ingestion of dust, and dietary ingestion (via foodstuffs and/or breast milk and drinking water), were considered to estimate the daily intake of OPEs, employing the methods mentioned in Text S1. Overall, the daily intake of OPEs through all exposure pathways decreased in the sequence of infants (mean value: 1320 ng kg bw À1 day À1 ) > toddlers (884 ng kg bw À1 day À1 ) > children (518 ng kg bw À1 day À1 ) > teenagers (274 ng kg bw À1 day À1 ) > adults (216 ng kg bw À1 day À1 ) > elderlies (175 ng kg bw À1 day À1 ), shown in Figure 3. Infants had the greatest daily intake of OPEs, primarily ascribed to breastfeeding (Figure 2b). Nursing mothers should be aware of nearby potential sources of OPEs, and avoid using cosmetics and beauty products during breastfeeding periods to minimize any accidental exposure (Mendelsohn et al., 2016;Tang et al., 2021).
Diet was the predominant source for daily intake of OPEs for all age cohorts, although the distribution patterns were somewhat different among the cohorts (Figure 2b). Toddlers had greater dietary daily intake of OPEs than other age cohorts via foodstuff ingestion, with dairy products accounting for 37.5% of the total intake amount. In other age cohorts, fats/oils contributed more than 20% of the total intake amount as they had high OPE concentrations (Poma et al., 2017;Poma et al., 2018). For children, OPE intake via dairy products and fats/oils was comparable to each other. For adults and elderlies, meats, grains, and vegetables were the main food categories for dietary intake of OPEs, except for fats/oils. Vegetarians (not eating meats and fish) and vegans (not eating meats, fish, eggs, and dairy products) ingested 134 and 113 ng kg bw À1 day À1 , respectively, from foodstuffs, and their OPE intakes reduced by approximately 15-30% compared to those on a regular diet. However, fish, fruits, and eggs contributed much less OPEs via ingestion, and drinking water contributed the least for all age cohorts. Clearly, a low-fat diet, particularly a vegetarian diet, may effectively reduce the daily intake of OPEs. Some indigestible additives may be deliberately added to foodstuffs to reduce OPE intake, such as chitosan and montmorillonite which can decrease OPE oral bioaccessibility (Guo et al., 2019).
Apart from dietary intake, dust ingestion is the most important pathway for OPE intake ( Figure 3). Infants, toddlers, and children take in greater amounts of OPEs than adults via dust ingestion, mostly due to their mouthing behavior and widespread occurrence of OPEs in commercial products (Hoffman, Butt, et al., 2015;Peng et al., 2020;Stapleton et al., 2011;Xue et al., 2007). The frequency of hand-to-mouth contacts usually decreases with increasing age for these three age cohorts, with an average of 6.7-28.0 times per hour (Xue et al., 2007). Since OPEs are prevalent in the indoor environment, the mouthing behavior should not be omitted in these age Figure 3. Amounts of the daily intake of OPEs (ng kg bw À1 day À1 ) for infants, toddlers, children, teenagers, adults, and elderlies, estimated with the data in Table S4 (indoor air, n ¼ 54), Table S5 (indoor dust, n ¼ 163), and Table S6 (foodstuffs and drinking water, n ¼ 74). Dietary ingestion including ingestion of foodstuffs (breast milk) and drinking water. Dermal absorption refers to the amounts of OPEs intake via dermal contact to indoor dust. White dotted line represents mean value, gray and red cross marks represent minimum and maximum values, respectively. Box represents 25th, 50th, and 75th percentiles, whiskers represent 10th and 90th percentiles, and dots represent 5th and 95th percentiles. Significant differences in daily take between exposure pathways were discerned for every age cohort (Mann-Whitney U test, p < 0.001).
cohorts. Teenagers, adults, and elderlies take in less OPEs via dust ingestion which also decreases with increasing age, since they may not have mouthing behavior; however, they may ingest dust by eating foodstuffs adhered with dust or by hand-to-mouth contact unintentionally. Compared to dust ingestion, inhalation and dermal absorption via dust seemed to be minor pathways. However, OPE intake via dermal absorption and dust ingestion was comparable to each other for teenagers. As oral bioaccessibility of OPEs was not considered, this may lead to overestimation of daily intake of OPEs. If the oral bioaccessibility was accounted, the mean daily intake of OPEs via dust ingestion decreased by almost 50% compared to the intake with 100% bioaccessibility. Nevertheless, OPE bioaccessibility remains unclear, which should be further investigated and considered to improve the estimations of daily intake and relative health risk. Infants are exposed to higher OPE levels in residences than in vehicles, while toddlers, children, and teenagers have the greatest daily intake of OPEs in schools/daycare centers, followed by that in residences and vehicles (Figure 4). Toddlers and children are subject to higher OPE exposure in schools/daycare centers, and nap mats of PU foam may be an important source (Stubbings et al., 2018). Besides, the use of consumer products as well as building and decoration materials might contribute to a considerable amount of OPEs in schools/daycare centers. Certified lowenergy preschools were verified to have lower OPE levels in indoor environments than non-certified low-energy and reference preschools (Persson et al., 2018). Therefore, OPE-free or environmentally friendly materials are highly recommended for use in schools/daycare centers to decrease OPE levels (Persson et al., 2018;Stubbings et al., 2018;Young et al., 2021). Adults and elderly individuals had the greatest daily intake of OPEs in offices, followed by residences. Compared to other exposure scenarios, in-vehicle exposure contributed the least amount of OPEs to the general population because it had the shortest exposure duration. Drivers, especially professional ones, may be exposed to greater amount of OPEs in vehicles than in residences, as they experienced longer exposure duration and greater levels of OPEs in vehicles (Figure 1). Therefore, good ventilation should be maintained in vehicles to minimize OPE exposure.

Internal exposure to OPEs
As mentioned in section 4.5, urinary metabolites of OPEs can shed light on the exposure levels of OPEs (Fromme et al., 2014;Hoffman, Garantziotis, et al., 2015;Hou et al., 2020). Urinary diester Figure 4. Amounts of the daily intake of OPEs (ng kg bw À1 day À1 ) from different indoor microenvironments for different age cohorts, estimated with OPE concentrations in indoor air and dust in residence (n ¼ 116), vehicle (n ¼ 21), school/day care (n ¼ 50), and office (n ¼ 30) (Tables S4 and S5). The rectangular bar represents the mean value and the error bar represents the standard deviation. Significant differences in daily take between microenvironments were discerned for every age cohort (Mann-Whitney U test, p < 0.001).
metabolites including DBP (sum of DnBP and DiBP), BCEP, BCIPP, BDCIPP, DPhP, and BBOEP, were employed to further evaluate P 7 OPE exposure (Table S8). The mean concentration of total urinary metabolites was higher in non-adult urine than in adult urine. The mean daily intake of P 7 OPE estimated by urinary metabolites were 2060 and 470 ng kg bw À1 day À1 for nonadults and adults, respectively. These results further indicate that children are exposed to higher OPE levels, and effective measures should be adopted to minimize their exposure levels. For example, hand washing is a feasible way to reduce the amount of OPE residues on hand surfaces, with a removal efficiency of >50%, and to decrease OPE absorption (Liu et al., 2017;Nielsen, 2010). Among adults, pregnant women and mothers had a greater daily intake of P 7 OPE than other adults, with a mean value of 729 ng kg bw À1 day À1 , probably because they have special diets for pregnancy or post-pregnancy nutrition. They should maintain balanced diets and need to refuel quite often to get enough calories and nutrients during pregnancy and breastfeeding period. They may eat more foodstuffs containing protein, fiber and healthy fat, these foodstuffs and greater amount of ingestion in turn contribute relatively higher amount to their body burdens of OPEs ( Figure 2a). Besides, pregnant women and mothers may be more vulnerable to be exposed to continuous exposure sources from indoor microenvironment or personal care products and cosmetics than other adults. The OPE levels in house dust were found to be associated with concentrations of their metabolites in urine of pregnant women, and housework such as frequent mopping could also increase the urinary metabolite concentrations during pregnancy (Kuiper et al., 2020;Percy et al., 2020). Increased OPE metabolite concentrations during pregnancy were also reported to be associated with the use of nail polish, perfumes, and cosmetics (e.g. lipstick), in which OPEs were commonly detected (Ingle et al., 2020;Tang et al., 2021;Young et al., 2018). Infants and fetuses are generally sensitive to xenobiotics. Their exposure levels may be effectively alleviated by minimizing mothers' exposure. In this regard, more attention should be paid to pregnant and nursing mothers' health and their prenatal and postpartum exposure.

Risk assessment
Risk assessment of all exposure pathways for different age cohorts was estimated based on 100% bioaccessibility, using a hazard quotient to determine the risk level, the sum of which was defined as the hazard index (HI). The mean HI values for infants, toddlers, children, teenagers, adults, and elderlies were 6.1 Â 10 À2 , 5.1 Â 10 À2 , 3.2 Â 10 À2 , 1.7 Â 10 À2 , 1.3 Â 10 À2 , and 1.1 Â 10 À2 , respectively. The HI values within the confidence interval ranked from the highest to the lowest in the sequence of infants > toddlers > children > teenagers > adults > elderlies ( Figure 5), and were all below the risk threshold (HI ¼ 1). Thus, the general population was subject to low risk of OPE exposure via all pathways in the recent decade. However, data on the reference dose of OPEs for each exposure pathway remain limited, especially for inhalation and dermal exposure. The health risks posed by OPEs may be underestimated as the reference doses of OPE metabolites are not available. The combined toxic effects of OPE mixtures also remains largely unknown, and more studies should be directed toward refining the health risk assessment of OPE exposure.

Conclusions and future perspectives
This comprehensive review on age-specific exposure to OPEs, in the context of multiple microenvironments and exposure pathways, can shed light on potential health risks posed by OPEs and provide insights to feasible suggestions to reduce the daily intake of OPEs. Dietary ingestion is the major pathway for all age cohorts exposed to OPEs, and breastfeeding contributes the most to daily intake of OPEs for infants. Apart from dietary intake, dust ingestion is the main source of OPE body burden in humans. Non-adults contain a greater amount of OPE body burdens and are under greater health risk than other age cohorts. Although the daily intake of OPEs is below the reference dose, their accumulation in human body and possible health risks in long-term exposure should not be neglected, particularly for potential adverse impacts on the growth and development of infants and toddlers. This is because the knowledge on the chronic toxicities and toxic accumulation of OPEs and their metabolites remains scarce, and little is known about their reference doses in different exposure pathways and possible synergistic or additive effects of exposure to OPE mixtures. As infants and toddlers are susceptible to an increase in the body burden of OPEs and likely place their hands or objects in their mouths, OPE-free or clean materials should be provided to infants and toddlers to reduce their daily intake of OPEs. Reference doses of OPEs for such vulnerable groups should be developed in future investigations.
As particulate OPEs are size-dependent in air and dust, and particle sizes greatly influence their transport, fate, and bioaccessibility. Future studies should pay attention to the occurrence of size-fractionated particulate OPEs in environmental media as well as their size-specific bioaccessibility via inhalation, dermal absorption, and ingestion. The bioaccessibility should be carefully considered in daily intake estimation to refine health risk assessment. Metabolic mechanism of OPEs also remains largely unknown, which may undermine the accuracy of body burden estimated by internal methods. Their metabolites should be further identified in human body through examining the metabolites of OPEs in urine, blood, or breastmilk by suspect/non-target screening using high-resolution mass spectrometry. The significant knowledge gaps of combined toxicities of OPEs in environment-related levels and ratios need to be filled in future studies, such as by employing related tests in genomics, proteomics, and metabolomics.

Disclosure statement
The authors declare no competing financial interests. Figure 5. Cumulative probability (%) of hazard index for different age cohorts exposed to OPEs via indoor inhalation, indoor dermal absorption and ingestion of dust, and dietary ingestion.