Organophosphate Ester Flame Retardants: Are They a Regrettable Substitution for Polybrominated Diphenyl Ethers?

Abstract

As the use of polybrominated diphenyl ethers (PBDEs), and the entire class of organohalogen flame retardants, is declining, the use of organophosphate esters flame retardants (OPFRs) is increasing. In this paper, we ask whether OPFRs are a better choice than PBDEs.

To address this question, we compared OPFRs with PBDEs for a wide range of properties. OPFRs exposure is ubiquitous in people and in outdoor and indoor environments, and are now often found at higher levels compared to PBDE peak exposure levels. Furthermore, data from toxicity testing, epidemiological studies, and risk assessments all suggest that there are health concerns at current exposure levels for both halogenated and non-halogenated OPFRs.

Obtaining the scientific evidence needed for regulation of OPFRs can take many years. Given the large number of OPFRs in use, manufacturers can move towards healthier and safer products by developing innovative ways to reduce fire hazard for electronics enclosures, upholstered furniture, building materials and other consumer products without adding flame retardant chemicals.

TOC Art

INTRODUCTION

Flame retardants are added to consumer products and building materials to reduce fire hazard. Their use is driven by flammability standards, usually based on small-scale fire testing, which may not accurately predict real life fire behavior. ^{1, 2} Beginning in the 1970s, polybrominated diphenyl ethers (PBDEs) were added to consumer products, including furniture, children’s products, and electronics. After extensive research showed that PBDEs were persistent, bioaccumulative, and toxic, in 2004 the European Commission and California banned the use of Penta- and OctaBDE, two commercial mixtures primarily used in North America. Also, in 2004, the US Environmental Protection Agency (US EPA) negotiated a phase-out of new production of these two PBDE commercial mixtures with US manufacturers. Subsequently in 2009, the US EPA negotiated the phase out of DecaBDE production, the PBDE with the highest production volume. Penta- and OctaBDE were added to the Stockholm Convention in 2009, prompting more than 150 signatories to legislate their phase-out (see Figure 1 and Table S1 for more details about regulatory timeline). DecaBDE was added to the Stockholm Convention in 2017 and similarly phased out of use in most countries.

Figure 1.

Timeline of major regulatory milestones for PBDEs and OPFRs. Policy enactment dates are listed; implementation/ compliance dates are often later (see Table S1 for detailed information).

Unfortunately, old furniture, electronics, vehicles, and other products containing PBDEs continue to be used and reused in spite of the phase outs and bans that prevent their use in newly manufactured products. The resultant long term exposure to PBDEs, especially in low income communities, is another reason to avoid alternative flame retardants that could be similarly harmful.

As the use of PBDEs and the class of organohalogen flame retardants (flame retardants containing carbon and halogen elements, most often bromine or chlorine) is declining due to regulatory action and manufacturers’ voluntary actions, the use of organophosphate esters flame retardants (OPFRs) is increasing. For example, DecaBDE production in the US dropped from 50 million pounds (23,000 tonnes) in 2012 to less than 25 thousand pounds (11 tonnes) in 2015.³ During the same time frame, the US production volume of various OPFRs has remained constant or increased. ³

OPFRs are organic esters of phosphoric acid containing either alkyl chains or aryl groups, and they may be halogenated or non-halogenated (see Figure S1). In addition to their use as flame retardants, OPFRs are used as plasticizers in consumer products and construction materials. ^{4, 5} OPFRs are also oxidation products of phosphites, which are commonly-used antioxidants in plastic products. ^{6, 7} Commercial flame-retardant formulations that contain mainly OPFRs have replaced Penta-BDE in residential furniture. ^{8, 9} OPFRs have also been used as substitutes for Octa- and DecaBDE in electronics, with resorcinol bis(diphenylphosphate) (RDP or PBDPP) and triphenyl phosphate (TPHP) measured in televisions at mg/g levels. ^10–12 A list of the OPFRs discussed in this paper is reported in Table 1.

Table 1.

List of major OPFRs cited in the text along with CAS numbers (adapted from Zhang et al. 2016¹⁷ and Stubbings et al., 2019¹⁰⁶)

OPFRs	Full Name	CAS	Molecular Formula	Molecular weight (g/mol)	LogKOWa	LogKOAa	TSCA a,b	DSLc
TEP	Triethyl phosphate	78–40-0	C6H15O4P	182.16	0.9	5.5	Yes (Imported) (450–4,500 tonnes)	Yes (DSL)
TNBP	Tri-n-butyl phosphate	126–73-8	C12H27O4P	266.32	3.8	7.7	Yes (450–4,500 tonnes)	Yes (DSL)
TCEP	Tris(2-chloroethyl) phosphate	115–96-8	C6H12Cl3O4P	285.48	1.6	7.6	Yes (Imported) (11–45 tonnes)	Yes (DSL)
TCIPP+	Tris(2-chloro-1-methylethyl) phosphate	13674–84-5	C9H18Cl3O4P				Yes (23,000–45,000 tonnes)	Yes (DSL)
TPP	Tripentyl phosphate	2528–38-3	C15H33O4P	308.40			No	No
TDMPP	Tris(3,5-dimethyl phenyl) phosphate	9006–37-5	C24H27O4P	410.44	8	13.5
TDCIPP	Tris(1,3-dichloro-2-propyl) phosphate	13674–87-8	C9H15Cl6O4P	430.90	3.7	10.6		Yes (DSL)
TPHP	Triphenyl phosphate	115–86-6	C18H15O4P	326.28	4.7	10.5	Yes (450–4,500 tonnes)	Yes (DSL)
EHDPP	2-Ethylhexyl diphenyl phosphate	1241–94-7	C20H27O4P	362.40	6.3	11.3	Yes (Imported) (450–4,500 tonnes)	Yes (DSL)
TBOEP	Tris(2-butoxyethyl) phosphate	78–51-3,	C18H39O7P	398.47	3.0	12.3
DOPP	Dioctyl phenylphosphonate	1754–47-8	C22H39O3P	382.52
TmCP	Tri-m-cresyl phosphate	563–04-2	C21H21O4P	368.36	6.3	12.0
TpCP	Tri-p-cresyl phosphate (Tri-p-tolyl phosphate)	78–32-0	C21H21O4P	368.37	6.3	12.0
ToCP	Tri-o-cresyl phosphate (Tri-o-tolyl phosphate)	78–30-8	C21H21O4P				Yes	Yes (DSL)
TPPP	Tris(2-isopropyl phenyl) phosphate	64532–95-2	C27H33O4P	452.52	9.1	14.0
TDBPP	Tris (2,3-dibromopropyl) phosphate	126–72-7	C9H15Br6O4P	697.61	4.2	14.1	Yes	No
TTBPP	Tris(4-tert-butylphenyl) phosphate	78–33-1	C30H39O4P	494.62	10.4	15.0	Yes (Withheld)	Yes (DSL)
IPP	Isopropylated phenyl phosphate	68937–41-7 (mix of isomers)	C21H18O4P -C27H30O4P	368.28– 452.28	NR	NR	Yes
IDDP	Isodecyl diphenyl phosphate	29761–21-5	C22H31O4P	390.45	7.3	12.0

Withheld = production volumes withheld; values with parenthesis are estimated production volumes for 2015; DSL = Canadian domestic substances list; NDSL = Canadian non domestic substances list;

^aToxic Substances Control Act Chemical Substance Inventory (TSCA Inventory), Available at: https://www.epa.gov/tsca-inventory/how-access-tsca-inventory (Accessed May 2018).

^bU.S. EPA; CDR database, 2016, Available at: https://www.epa.gov/chemical-data-reporting (Accessed May 2018).

Extensive scientific research now suggests that the entire class of organohalogen flame retardants may have hazardous properties and some authoritative bodies are now addressing this problem with a chemical class approach (See Table S1). ¹³ In 2017, the US Consumer Product Safety Commission (CPSC) accepted a petition to ban furniture, children’s products, electronic enclosures and mattresses containing any member of the class of organohalogen flame retardants. In May 2019, at the request of the CPSC, the National Academies of Sciences (NAS) released a report titled “Scoping Report for Conducting a Hazard Assessment of Organohalogen Flame Retardants as a Class”.¹⁴ This report states that “the number of chemicals in use today demands a new approach to risk assessment, and the class approach is a scientifically viable option.” The authors of the report concluded that “the best approach is to define subclasses as broadly as is feasible for the analysis. Although the challenges to a class approach might appear daunting, the alternative ­ individual assessments of hundreds of chemicals ­ is unrealistic. The report supports the idea that the class approach can prevent the problem of “regrettable substitution” of a replacement lacking adequate toxicity information for a phased out known hazardous substance. Regrettable substitution occurs because of the difficulty of changing industrial processes and a lack of toxicological information, causing manufacturers to replace a phased-out chemical with a “drop in” substitute chemical that has a similar structure, function and potential for harm. The NAS report also highlights the point that the cumulative exposures and risk are ignored when chemicals are assessed individually. In 2018, the European Commission (EU) proposed to prohibit the entire class of organohalogen chemicals in electronic display enclosures and stands. ¹⁵ --effective on April 1, 2021. The manufacturers of televisions and other electronics will need to find an alternative solution to meet flammability codes, and it is likely that industry will look to non-halogenated OPFRs.

In this paper, we ask whether OPFRs as a class have a reduced potential for harm compared to PBDEs, or if they are an example of regrettable substitution. To address this question, we compared OPFRs and PBDEs in regard to their environmental fate, measured levels indoors, exposure levels among the general population, and evidence of adverse health effects. Our comparison came from reviewing the literature, although the review was not comprehensive. We conclude by discussing current policy changes and opportunities to innovate for less hazardous materials and products with reduced potential for harm from flame retardants.

ENVIRONMENTAL BEHAVIOR

The idea that OPFRs are less harmful than PBDEs was largely based on the presumption that OPFRs are less environmentally persistent and hence have a lower potential for widespread environmental distribution and exposure. However, mounting evidence calls this presumption into question.

PBDEs have been classified as Persistent Organic Pollutants (POPs) under the Stockholm Convention due to their persistence, ability to undergo long-range transport, bioaccumulation potential, and toxicity to both humans and wildlife. ¹⁶ PBDEs have been documented in air, water, and terrestrial and aquatic biota at a global scale (see Tables S1-S4). They are nonpolar compounds with generally low volatility and vapor pressures (see Figure S2). PBDEs can degrade, albeit slowly, in environmental matrices through photodegradation and microbial debromination, which can create more bioavailable and toxic congeners.

OPFRs are expected to be less persistent in the environment than PBDEs.¹⁷ Data on their persistence is scant and the physical-chemical properties of OPFRs are difficult to measure or estimate, which makes prediction of their environmental behavior more uncertain than that of PBDEs. Although their higher vapor pressures lead to the expectation of higher air concentrations than that of PBDEs (see Figure S2), they are also expected to have shorter half-lives in air, and thus reduced atmospheric long range transport potential.¹⁸ Compared with PBDEs, OPFRs and especially chlorinated OPFRs are more soluble and can persist in water, which gives them the ability to undergo long range transport via waterborne routes.^{19, 20} Thus, rather than being POPs, as is the case with PBDEs, chlorinated OPFRs appear to be Persistent Mobile Organic Compounds or PMOCs, which is equally concerning.^{18, 21}

Although OPFRs were not expected to accumulate in the environment based on their physical-chemical properties, multiple measurements show that many OPFRs have achieved orders of magnitude higher concentrations in air and water in numerous environments ranging from urban areas to remote Arctic and Antarctic locations than PBDEs when they were at peak use (see Figure 2 and Tables S2-S3). OPFRs are readily scavenged from air by precipitation and then transported to surface waters because of their higher solubility and low tendency for sequestration in soil and other carbon-rich matrices compared to PBDEs. This efficient scavenging of OPFRs by rain, coupled with high emission rates, results in some urban surface water concentrations being similar to those from treated final wastewater treatment plant effluent (see Table S1). ¹⁸ In Great Lakes water, for example, OPFRs concentrations were in the range of 10–100 ng/L while ΣBDE (defined as the sum of total measured BDE congeners, which can vary from study to study) were significantly lower, with a range of 0.05–0.25 ng/L (see Table S2 for specific data). Despite their high aqueous solubility, OPFRs also accumulate in sediment by virtue of high emissions and ability for transport to aquatic systems. For example, ∑₁₄OPFR were ~0.5–50 ng/g dw in the Great Lakes sampled in 2010–2013,²² comparable to concentrations of 0.5 to 6.7 and <4 to >240 ng/g dw for Σ₉BDE and BDE-209, respectively measured at their peak usage. ²³

Figure 2.

Bar chart representing selected median concentrations of total OPFRs and total PBDEs in outdoor air (pg/m³, A), water (ng/L, B), and indoor dust (ng/g, C). Each bar represents data from a different study (see Tables S2–5 for details about the studies included and for a more comprehensive list of locations).

Finally, numerous measurements confirm not only the presence but the relative abundance of OPFRs in remote locations, which cannot be explained by local releases. For example, total OPFRs have reached a median concentration of 237 pg/m³ in Canadian Arctic air ^{19, 24–26} while ΣBDE air concentrations, at the time of peak use, were orders of magnitude lower. OPFRs have also accumulated in Arctic sediments at concentrations 10–100 times greater than those of PBDEs, which, as noted above, is unexpected since OPFRs are less likely to deposit because of their high water solubility. ²⁷

Overall, environmental measurements clearly show that OPFRs demonstrate long range transport and accumulation in the environment that rival PBDEs, despite expectations to the contrary. Further, whereas concerns with PBDEs were due to them being POPs, OPFRs fit into a higher class of concern, that of PMOCs.

INDOOR BEHAVIOR AND HUMAN EXPOSURE

Concentrations of PBDEs and OPFRs in indoor air, house dust, and on hand wipes provide critical information on the exposure potential since inhalation, hand-to-mouth contact, and dermal absorption are all important routes of human exposure to flame retardants. Below we provide evidence of relatively high exposure to OPFRs compared to PBDEs, originating from elevated levels in indoor air, house dust, and food.

The widespread use of PBDEs since the late 1970s resulted in near-ubiquitous human exposure, with ΣBDE in breastmilk and serum levels (both lipid adjusted) peaking in the early to mid-2000s. ^{28, 29} However, the increasing use of OPFRs following the phase-out of PBDEs has also led to increasing human exposure, and urinary OPFR biomarker levels have been steadily climbing since the early 2000s ³⁰. Interestingly, OPFRs have been found in dust at substantial levels for at least two decades. For instance, ΣOPFR levels are similar to ΣPBDE levels in house dust standard reference material (SRM 2585) (~5,000 ng/g), which was prepared from hundreds of vacuum cleaner bags collected in several US States in the mid-1990s. ^{9, 31–33} This comparison suggests that exposure to OPFRs was similar to PBDEs in the 1990s, but now appears to have increased since the phase-out of PBDEs.

The growing prevalence of OPFRs and decreasing amounts of decaBDE in indoor house dust correlates with their changing production levels in the United States. Common OPFRs such as Tri-cresyl phosphate (TCP), Tri-n-butyl phosphate (TNBP), and 2-Ethylhexyl diphenyl phosphate (EDHPP) each have had production volumes of up to 10 million pounds (4500 tonnes) yearly since 2012, ³⁴ compared to 10–50 million pounds (4500–23,000 tonnes) production volume of decaBDE in 2012, dropping to less than 25,000 pounds (11 tonnes) in 2015.³⁵

Although the percentage by weight application of PBDEs and OPFRs to polyurethane foam is roughly identical (~3–7%), OPFRs are also heavily used in electronics. As a result, OPFRs have been detected at much higher levels in indoor air (Table S4). ^{8, 36} While this is partly accounted for by OPFRs being used as both flame retardants and plasticizers, it is also likely reflective of the higher vapor pressure of OPFRs compared to PBDEs, leading to increased off-gassing of OPFRs from treated products into indoor air (Fig. S2). Studies from North America and Europe conducted in the early to mid-2000s, when PBDE use and exposure was at its height, reported average ΣBDE indoor air concentrations in the range of 100–600 pg/m³. ^37–39 By contrast, recent studies report OPFRs in the ng/m³ range, at least an order of magnitude higher than for PBDEs (Table S4). Much like PBDEs, OPFR air concentrations have been found to fluctuate seasonally. They also vary depending on microenvironment, with cars and offices often having higher concentrations compared to living spaces, reflecting high usage and emissions. ^39–41

Because OPFRs have higher vapor pressures compared to PBDEs, it might be assumed that OPFR dust concentrations would be lower than PBDE dust concentrations. However, despite regional differences in flame retardant dust concentrations, recently reported ΣOPFR geometric mean and median dust levels are either equivalent to (if considering the measured sum of PBDE congeners in commercial mixtures) or higher than ΣPBDE geometric mean and median dust levels from the early to mid-2000s (Fig 2). Reported ΣOPFR geometric mean and median dust concentrations range from low μg/g to low mg/g (Table S5). ^{3, 16, 33, 41–46} Several recent studies have estimated the daily exposure to OPFRs and PBDEs (e.g. ng/kg/day) via levels measured in indoor dust and the results demonstrate that exposure to OPFRs is higher.^{47, 48}

Although data are limited for PBDEs, OPFRs have been found on hand wipes at higher levels than PBDEs, suggesting that the magnitude of exposure via hand-to-mouth and dermal transfer pathways is potentially greater for OPFRs compared to PBDEs (Table S3). As for PBDEs, OPFR exposure may also result from dietary or diet associated intake. ^{49, 50} The use of EHDPP, TPHP and TNBP in food packaging material is approved by the US Food and Drug Administration, and their migration from plasticized film wrappers into food has been documented.⁵¹ The presence of OPFRs has also been reported in butter, bread bags, fish, and drinking water. ^52–54 Likewise, a recent study detected Tris(2-chloroisopropyl) phosphate (TCIPP) and TNBP in greater than 70% of 87 food samples and 5 tap water samples collected in Australia.⁵⁵ Based on the levels found in food, the estimated daily dietary intake was 4.1 ng/kg bw for Tris(2-chloroethyl) phosphate (TCEP), 25 ng/kg bw for TCIPP, and 6.7 ng/kg bw for TNBP. In Sweden, Poma et al. detected EHDPP in composite food samples from multiple food categories, and estimated a daily intake of ΣOPFRs (TCEP, TPHP, EHDPP, Tris(1,3-dichloro-2-propyl) phosphate or TDCIPP, and TCIPP) in the range of 6–49 ng/kg bw. ⁵⁶ The total dietary intake of Ʃ₇OPFRs in fish from the Philippines was 22 ng/kg bw/day.⁵⁷ In an US market basket study from 1988, the dietary intake for TPHP was calculated as 0.3–4.4 ng/kg/bw/day.⁵⁸ For comparison, the estimate of daily intake of ∑₁₃BDE congeners for US adults was 0.9 ng/kg/bw and 0.7 ng/kg/bw for ∑₁₈BDE congeners for Canadian adults during the 2000s. ^{59, 60}

Owing to differences in metabolism between the two chemical classes, the comparison of PBDE and OPFR internal dose levels is challenging: PBDEs are bioaccumulative and have long half-lives (weeks to years) in the human body while OPFRs are rapidly metabolized with relatively short half-lives (hours to days). ^{61, 62} PBDEs have been measured in human serum as biomarkers of exposure, while diester metabolites have been measured in human urine as indicators of OPFR exposure. Serum PBDE levels have been documented in the picomolar to low nanomolar range (on a wet weight basis), with BDE-47 generally detected most frequently and at the highest concentration of all BDE congeners (although, it should be noted that BDE-209 and higher molecular weight congeners were often not included in past serum analyses). ^63–65 Urinary OPFR metabolite levels have been reported in the low to mid nanomolar range, with diphenyl phosphate (DPHP, metabolite of multiple OPFRs), 1-hydroxy-2-propyl bis(1-chloro-2-propyl) phosphate (BCIPHIP; metabolite of TCIPP), and bis(1,3-dichloro-2-propyl) phosphate (BDCIPP; metabolite of TDCIPP) frequently detected at high levels compared to other urinary metabolites, ^{66, 67} although regional differences exist.⁶⁸ While it is difficult to compare serum PBDE levels to urinary OPFR metabolite levels, detection frequencies among biospecimens are similar. BDE 47 was detected in 97% of serum samples (n = 2,062) tested in the 2003–2004 National Health and Nutrition Examination Survey (NHANES), and recent urine analyses have detected BDCIPP, BCIPHIPP, and DPHP in more than 95% of tested samples. ^{40, 55, 63, 68–71} From these data, it is evident that OPFR exposure is pervasive in the human population and that vulnerable sub-populations like infants and children may be more highly exposed. ^{66, 71–74}

Continuing OPFR biomonitoring efforts will help define trends regarding human exposure to OPFRs. To summarize, OPFRs and their metabolites are detectable at relatively high levels and frequency in indoor air, dust, food items, and human biospecimens. Given that OPFR exposures may still be increasing, it is clear that the current potential for indoor exposure to OPFRs is substantially greater than it was for PBDEs in the early to mid-2000s, when PBDE use was at its height.

TOXICITY AND HEALTH EFFECTS

Another important aspect of this comparison between PBDEs and OPFRs is understanding how the toxicity profiles of these classes of compounds compare with each other. As a class, 97 OPFR flame retardants were shown to have potential toxicity based on the Quick Chemical Assessment Tool (QCAT). ⁷⁵ The US EPA categorized some of the alternative flame retardants as high priority compounds that critically need more toxicological studies or for which regulatory measures could be envisaged. ⁷⁵

Although the use of OPFRs is on the rise, their toxicological hazard has not yet been well-characterized and indeed, a more complete understanding of the toxicity of PBDEs has emerged only after their regulation. This section summarizes recent toxicity data and describes a linkage between toxicity data noted in in vitro and in vivo studies with human exposure, ultimately identifying data gaps that need to be addressed. In addition to OPFRs and PBDEs, we also compared the toxicity profile of tetrabromobisphenol A (TBBPA) since it is the most highly produced brominated flame retardants globally.⁷⁶ Its production has remained fairly stable over the years, likely a reflection of its primary use as a reactive flame-retardant (as opposed to PBDEs additive use). The EU noted “no health effects of concern” for TBBPA for adults or infants based on margins of safety.⁷⁷ However, evidence in the literature suggests effects on the reproductive and nervous system development including brain and thyroid function.^{76, 78}

OPFRs, PBDEs and TBBPA appear to have comparable developmental and neurodevelopmental toxicity potential in a variety of in vitro assays that represent processes critical to neurodevelopment such as neuronal proliferation, neurite outgrowth, synaptogenesis and network formation. ^{75, 79} They have also been shown to affect reproduction, development, and motor activity in a multitude of alternative animal models such as zebrafish, Caenorhabditis elegans, and Planaria. ^79–87 Furthermore, similar to PBDEs, OPFR and TBBPA exposure appears to elicit behavioral alterations that persist into adulthood, long after cessation of developmental exposure. ^{88, 89}

To relate toxicity data from in vitro and alternative animal models, as well as traditional in vivo animal studies to human exposure, we used a high throughput toxicokinetic model (HTTK)^{90, 91} to convert the data, thereby allowing comparisons across the various exposure scenarios (Figure 3). The main purpose of this analysis is two-fold. The first is to compare in vitro and in vivo PODs (points of departure) to inform readers on how novel rapid strategies relate to traditional rodent studies in their toxicological outcomes, so that they may be used as complementary approaches. The second is to compare the minimum risk levels (MRLs- i.e., PODs values including safety factors, which are equivalent to RfDs or reference doses) to measured human exposures.

Figure 3.

Flame retardant plasma concentrations measured or estimated from ingestion using data from house dust, breastmilk, and/or handwipe samples (colored bars and circles) are compared to the most potent in vitro concentration per chemical (black dots) and in vivo point-of-departure (POD; triangles). The filled triangles represent rat plasma concentrations based on in vivo POD values (when available), and open triangles represent the Minimum Risk Levels (MRLs). The colored bars represent the range of concentrations and the circles represent the mean, median, or maximum median (see Supporting Information for further details on these calculations).

First, plasma concentrations were simulated in the model using data from house dust, breastmilk, or hand wipes samples (Figure 3 and Table S7). Important input parameters into the model were the estimated oral exposures based on child feeding (for breastmilk), ingestion due to hand-to-mouth actions (for dust), as well as chemical-specific parameters, such as the fraction of the chemical bound to plasma, intrinsic metabolic clearance, pKa, LogP (lipophilicity), and molecular weight. A detailed description of these values and calculations can be found in Tables S7-S9. The internal exposure values calculated with the HTTK model were then compared to POD values from developmental or neurodevelopmental-associated in vitro assays and/or in vivo alternative animal (i.e. zebrafish) assays to understand how in vitro PODs relate to human exposure. A POD is defined as a dose-response point that marks the starting point for low dose extrapolation—that is, the POD is the exposure level at which an effect is seen.

The in vitro POD to human plasma concentration comparison found that the in vitro POD for TPHP, BDE-47, TDCIPP, and TBBPA lies within the range of estimated plasma concentrations from human exposure, as indicated by an overlap of the colored bars with the black filled circles (Figure 3). For other compounds such as Trimethyl phenyl phosphate (TMPP), Isodecyl diphenyl phosphate (IDDP), EHDPP, Isopropylated phenyl phosphates (IPP), Tert-butylated phenyl diphenyl phosphate (BPDP), and TCEP, the biological activity in vitro occurred at higher concentrations than those estimated from human exposure data. However, it should be noted that even though the in vitro POD appears higher for these latter compounds, there is limited exposure data for these OPFRs. This is important because, while we do not know the potential health effects of these OPFRs, they have similar patterns of in vitro activity at comparable concentrations to the phased-out flame retardant (e.g. BDE-47) and to TBBPA.

Where available (TMPP, TDCIPP, TCEP), we then compared the HTTK-modeled plasma concentrations to the POD values obtained from in vivo rat studies (open triangles) that are currently used to set the Minimum Risk Levels (MRLs, open triangles). According to ATSDR, an MRL is an estimate of the daily human exposure to a hazardous substance that is likely to be without appreciable risk of adverse health effects over a specified duration of exposure⁹² This comparison showed that for some compounds such as TDCIPP, the in vivo rodent POD lies within the range of human exposure while for others such as TCEP, toxicity in animal studies is noted at a higher concentration compared with human exposure. Differences in sensitivities between the findings in rodents to that in human-derived cell-based models, or toxicokinetic, toxicodynamic and/or exposure characteristics should be taken into account. For some of the less well studied OPFRs such as IPP, BPDP, EHDPP and IDDP, even though it may appear that there is a wide margin of exposure between current serum concentrations in humans and predicted toxicological effects, it should be noted that there is relatively sparse exposure data available. Furthermore, due to recent CPSC and EU regulations (Table S1), the use of the non-halogenated OPFRs is projected to be on the rise.

While this general strategy provides insights into those OPFRs for which concern could be greatest, it has several caveats. For example, it does not consider sensitive populations or even genetic variability within a population, which could significantly change the interpretation.⁹³ This approach is also limited by several model constraints; e.g., the in vivo POD used to set MRLs for some compounds such as BDE-47 and TBBPA could not be plotted because in silico model parameter estimates were unavailable. Additionally, as only developmental or neurodevelopmental-associated in vitro assays were used to evaluate in vitro potency of these compounds, other assays surveying a broader biological space could indicate additional biological disruption.

Nonetheless, these findings indicate that the in vitro activity for some of the OPFRs (i.e. TDCIPP, TPHP) is comparable to that of the phased-out BDEs (e.g. BDE-47) and lie within the range of human exposure (TPHP). Importantly, animal data are sparse for many of these OPFRs, but for compounds where data do exist, the in vitro activity appears to be at comparable levels to the in vivo PODs for some compounds (e.g. TDCIPP). Hence, it is imperative to consider novel strategies to integrate data to provide rapid and timely relevant information for human health protection, especially for sensitive populations, to complement time and cost intensive animal studies.

EPIDEMIOLOGICAL EVIDENCE

Epidemiological evidence points to concern for PBDEs and, more recently, for OPFRs. PBDEs are well-established neurodevelopmental toxicants (see Table 2). In a meta-analysis, Lam and coworkers concluded that there was sufficient evidence supporting an association between developmental PBDE exposure and reduced IQ. ⁹⁴ A similar conclusion was also reached by the extensive review of the literature conducted by the National Academies of Sciences, Engineering and Medicine, although their work was restricted to BDE-47. ⁹³

Table 2.

Overview of data on specific OPFRs including the CAS registry number, major uses, US production volumes for 2015, summary of biomonitoring data, major regulatory actions, and risk assessment/epidemiological data. Data from US EPA’s Chemical Data Reporting, NIH PubChem, and the Chemical Hazard Data Commons.

Chemical/ CASRN	Uses	US prod. volume (2015)	Biomonitoring	Regulatory Actions	Associations in epidemiological studies and risk assessment
TPHP/ 115–86-6	Industrial; Commercial; Consumer: foam seating and bedding, plastic and rubber products, nail polish. FDA Indirect Additives used in Food Contact Substances	1–10 million lbs	Increasing exposure; 30 ubiquitous exposure in NHANES- 92%, higher exposures in women and higher exposures in 6–11 yo children. 71	CA SCP1- Repro & Neuro tox; WA CSPA2	Adverse reproductive outcomes;107 prenatal exposures associated with decreased IQ and working memory in children;108 in combination with other OPFRs, increase in behavior problems. 96
TDCIPP/ 13674–87-8	Industrial; Commercial; Consumer: foam seating and bedding	10–50 million lbs	Increasing exposure;30 ubiquitous exposure in NHANES- 92% and higher exposures in women. 71	CA Prop 65- carcinogen; CA SCP- Carcinogen; WA CSPA	In combination with other OPFRs, prenatal exposures associated with decreased IQ and working memory in children; 108 cancer risks of concern for infants;97 cancer risks of concern for children. 99
IPP/ 68937–41-7 (mix of isomers)	Industrial; Commercial; Consumer: foam seating and bedding, plastic and rubber products.	1–10 million lbs		CA SCP- Repro, Neuro; WA CSPA
EHDPP/ 1241–94-7	Industrial; Commercial; Consumer: foam seating and bedding, plastic and rubber products FDA Indirect Additives used in Food Contact Substances	1–10 million lbs		WA CSPA; MDH3 Chemical of high concern
IDDP/ 29761–21-5	Industrial; Commercial; Consumer: foam seating and bedding, plastic and rubber products	Withheld		CA SCP- Repro, Neuro; MDH Chemical of high concern
TMPP (TCP)/ 1330–78-5	Industrial; Commercial; Consumer: plastic and rubber products, lubricants and greases, and other unspecified uses	1–10 million lbs		CA SCP- Repro, Neuro; WA CSPA
TCEP/ 115–96-8	Industrial, unspecified	25,000–100,000 lbs	Ubiquitous exposure in NHANES- 89% and higher exposures in 6–11 yo children. 71	CA Prop 65- carcinogen; EU toxic to reproduction; CA SCP carcinogen; WA CSPA	In combination with other OPFRs, increase in behavior problems; 96 risks of concern for cancer and reproductive effects for infants. 97
RDP (PBDPP)/ 57583–54-7	Industrial; Commercial; Consumer: plastic and rubber products and other unspecified uses	1–10 million lbs		CA SCP
TCIPP/ 13674–84-5	Industrial; Commercial; Consumer: foam insulation, building/ construction materials, foam seating and bedding products, electronic products	50–100 million lbs	Widespread exposure NHANES- 61%, higher exposures in 6–11 yo children, and higher exposures in women. 71	CA SCP- Carcinogen; WA CSPA	In combination with other OPFRs, increase in behavior problems; 96 paternal levels associated with decreased fertilization; 95 risks of concern for cancer and reproductive effects for infants. 97
TNBP/ 126–73-8	Industrial; Commercial; Consumer, including adhesives/ sealants and inks/ toners FDA Indirect Additives used in Food Contact Substances	1–10 million lbs	Ubiquitous exposure in NHANES- 81% and higher exposures in women. 71	EU CMR4; MDH Chemical of High Concern; WA CSPA

¹CA SCP = CA Safer Consumer Products Program Candidate Chemical List

²WA CSPA = WA Children’s Safe Products Act Chemicals of High Concern for Children

³MDH = Toxic Free Kids Act Chemicals of High Concern or Priority Chemicals

⁴EU CMR = EU Carcinogen, Mutagen or Reproductive Toxicant

Evidence from epidemiological studies is now indicating that OPFRs could also be causing adverse effects at ambient exposures. For example, Carignan et al. (2017) reported that urinary concentrations of DPHP, a metabolite of several OPFRs, as well as a standalone flame retardant, were significantly associated with adverse reproductive outcomes (e.g., failed fertilization and implantation) in 211 US women undergoing in vitro fertilization. ³⁰ Carignan et al. (2018) also found that paternal urinary concentrations of BDCIPP were associated with reduced fertilization.⁹⁵ Castorina et al. (2017) reported evidence for the developmental toxicity of OPFRs, finding that ∑OPFR metabolites measured in pregnant women, and particularly DPHP, were significantly associated with decreased IQ and working memory of their 7 year old children in the CHAMACOS birth cohort in California. ⁹⁵ Lipscomb et al. (2017) found a significant dose-dependent relationship between the exposure of 92 US children (ages 3–5 years) to ∑OPFRs (TCIPP, TCEP, TPHP) measured using silicone wristbands, and teacher ratings of less responsible behaviour and more externalizing behaviour problems. ⁹⁶

Limited risk assessments of OPFRs have been conducted in the last 10 years. Infants’ exposure to TCEP, TDCIPP, and TCIPP present in children’s products and residential upholstered furniture was associated with increased risk of cancer ⁹⁷ and reproductive effects for TCEP and TCIPP. ^{72, 98} This finding is further supported by a study investigating TDCIPP exposure in US infants. ⁷² Bradman et al. (2014) identified cancer risks exceeding California health-risk based guidelines for TDCIPP in early childhood education environments. ⁹⁹ Canada issued a draft risk assessment proposing that TCIPP be considered toxic based on human exposure to foam-containing upholstered furniture, but TDCIPP was not considered to pose a threat to human health at current exposures.¹⁰⁰ Authoritative government bodies have evaluated current data and listed several OPFRs as chemicals of concern or known to have specific health hazards (see Table 1 and Table S4).

Overall, data from traditional toxicity testing, new approach methods, epidemiological studies, and risk assessments all indicate health concerns for both halogenated and non-halogenated OPFRs.

POLICY APPROACHES TO OPFRS

As concerns about OPFR exposure and toxicity have emerged, regulators in the US, EU, and other jurisdictions have responded with policies that gather data, inform consumers, limit flame retardants of concern, and/or change flammability standards to reduce the need for flame retardants. Such responses have been aimed especially at vulnerable populations (e.g., children) and uses in common consumer products that result in widespread exposures (see Figure 1, Tables 2 and S1). In the US, the federal law covering most industrial, commercial, and consumer product chemicals, the 1976 Toxic Substances Control Act (TSCA), was acknowledged to be ineffective.¹⁰¹ Faced with a lack of regulation for decades, individual US states have stepped in to issue their own regulations. As shown in Figure 1, initial policies for individual OPFR chemicals focused more on data gathering (for example, the Washington Children’s Safe Product Act requiring reporting of chemicals use to the state) or labeling (for example, California Proposition 65 requiring warnings).

The past decade has seen a shift towards changing flammability standards so that flame retardants are not needed in consumer products when they do not provide a significant fire safety benefit. In 2016, TSCA was updated; although there were potential improvements in the law, it remains to be seen if public health protections from toxic chemicals will be improved. ^{102, 103} Canada, through its Chemical Management Plan, has regulated PBDEs and TCEP. ⁹⁹ OPFRs and other flame retardants in Canada are being assessed with a recommendation for declaring TCIPP potentially harmful to human health,¹⁰⁰ but these assessments remain constrained to a chemical-by-chemical basis rather than taking a more comprehensive approach to managing OPFRs and other flame retardants as a class.

The trends in Figure 1 and Table S1 indicate that policy makers are increasingly concerned with the use of hazardous chemicals in consumer products, especially those that result in exposures to children. An emerging trend is to require informed substitution (e.g. 2015 Minnesota House Bill 1100 – MN 2015) or to avoid regrettable substitution (e.g., 2016 Washington DC Carcinogenic Flame Retardant Prohibition Amendment Act¹⁰⁴), by creating health criteria for replacement chemicals.

LOOKING FORWARD

Here we have shown that, as with PBDEs in the past, OPFRs are now being used in high volumes, are sufficiently persistent to be detected globally, present health hazards, and may cause harm to humans, especially children, at current exposure levels. Given the large number of OPFRs on the market, obtaining the level of evidence a government often requires to regulate each compound would prove to be expensive and lengthy. However, manufacturers and purchasers can make informed choices now to eliminate use of potentially harmful chemicals.

As PBDEs and other organohalogen flame retardants are phased out, instead of replacing them with OPFRs with a similar potential for harm, we suggest pursuing creative “out-of-the-box” strategies such as improved product design to minimize flammability and using alternative materials that are inherently flame resistant. For example, one manufacturer reduced flame-retardants in TV cases by removing the power supply from inside the TV. An external power source was used (like in a laptop power cord), thus eliminating the need for flame-retardants in the plastic case around the TV.¹⁰⁵ Furthermore, only using flame retardants when they provide a proven benefit can reduce health hazard without impacting fire safety.

This paper has shown that the replacement of PBDEs with OPFRs is likely a regrettable substitution. The time has come for manufacturers, with the help of the scientific community, to stop moving from the use of one family of harmful chemicals to the next and to instead find innovative ways to reduce both fire hazard and the use of hazardous chemicals.