Toxicological Risks of Selected Flame-Retardant Chemicals (2000)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

EXECUTIVE SUMMARY 1 Executive Summary IGNITION of upholstered furniture by small open flames from matches, cigarette lighters, and candles is one of the leading causes of residential-fire deaths in the United States. On average, each year since 1990, about 90 deaths (primarily of children), 440 injuries, and property losses amounting to 50 million dollars have resulted from such fires. The U.S. Consumer Product Safety Commission (CPSC), an independent federal regulatory agency, is required to protect the public from unreasonable risks of injury and death associated with consumer products. In 1993, the National Association of State Fire Marshals petitioned CPSC to issue a performance-based flammability standard for residential upholstered furniture to reduce the risk of residential fires. If such a standard were promulgated, most residential upholstery fabric would be treated with flame-retardant (FR) chemicals. It is estimated that in the United States, FRs would be applied to as much as 600 million square yards of upholstery fabrics each year. Over the lifetime of the furniture, many consumers might be exposed to FR-treated fabric. Because some FRs are known to have toxic properties and there is a potential for exposure of millions of people to these chemicals, some have expressed concern about the use of these chemicals until they are shown to be safe. THE CHARGE TO THE SUBCOMMITTEE In its fiscal year 1999 appropriations report for CPSC, Congress requested an independent study by the National Academy of Sciences' National Research 

EXECUTIVE SUMMARY 2 Council (NRC) concerning health risks posed by exposure to FR chemicals that are likely to be used in residential upholstered furniture to meet a flammability standard that CPSC is considering. CPSC, with the help of the industry, identified the following 16 chemicals (or chemical classes) that became the focus of this NRC study: (1) hexabromocyclododecane, (2) decabromodiphenyl oxide, (3) alumina trihydrate, (4) magnesium hydroxide, (5) zinc borate, (6) calcium and zinc molybdates, (7) antimony trioxide, (8) antimony pentoxide and sodium antimonate, (9) ammonium polyphosphates, (10) phosphonic acid, (3-{[hydroxymethyl]amino}-3- oxopropyl)-dimethyl ester, (11) organic phosphonates, (12) tris (monochloropropyl) phosphate, (13) tris (1, 3- dichloropropyl-2) phosphate, (14) aromatic phosphate plasticisers, (15) tetrakis (hydroxymethyl) hydronium salts, and (16) chlorinated paraffins. The NRC assigned the project to the Committee on Toxicology (COT) of the Board on Environmental Studies and Toxicology. COT convened the Subcommittee on Flame-Retardant Chemicals, which prepared this report. Subcommittee members were chosen for their recognized expertise in toxicology, epidemiology, pharmacology, chemistry, exposure assessment, risk assessment, and biostatistics. The subcommittee was charged to review the toxicological and exposure data on the above 16 FR chemicals to assess potential health risks to consumers and the general population resulting from potential exposure to these chemicals in residential furniture. The subcommittee was also asked to identify data gaps and make recommendations for future research. The subcommittee was not charged or able, with the available data and resources, to evaluate adverse effects from occupational exposures, or the potential ecological effects that might result from the disposal of household furniture. The subcommittee also was not charged to compare FRs for efficacy, or to evaluate the cost, technology, exposure-standard achievability, or the benefits of using FRs on upholstered furniture to reduce fire risk, or to consider the toxicity of combustion products of FR materials from fires involving treated upholstery fabric. Those issues were also beyond the scope of the subcommittee's charge, expertise, and resources. THE SUBCOMMITTEE'S APPROACH To help the subcommittee gather information, the NRC commissioned papers by expert consultants on various FR chemicals, reviewed the toxicity assessments by the CPSC staff on 16 chemicals, and held public meetings at which representatives of CPSC, the U.S. Environmental Protection Agency, Congress, the National Association of State Fire Marshals, and industrial groups made presentations on issues of critical importance to the subcommit 

EXECUTIVE SUMMARY 3 tée's task. Some of the speakers also provided technical documents on FR chemicals. In characterizing the human health risks from dermal, oral, and inhalation exposures to each of the 16 FR chemicals, the subcommittee generally followed the risk-assessment paradigm first articulated by the NRC in its 1983 report Risk Assessment in the Federal Government: Managing the Process. The subcommittee evaluated data relating to key elements of the risk paradigm, as discussed below. Hazard Identification In the hazard-identification phase of each assessment, determinations were made as to whether causal relationships exist between the dose of an FR chemical and an adverse health effect. To identify adverse effects associated with an FR chemical, the subcommittee reviewed human (epidemiological studies, clinical observations, and case reports) and laboratory animal data on neurotoxicity, immunotoxicity, reproductive and developmental toxicity, organ toxicity, dermal and pulmonary toxicity, carcinogenicity, and other local and systemic effects. In vitro data were also reviewed to determine the potential for genotoxicity as well as other toxic effects and to understand the mechanisms of toxic action. Toxicokinetic studies were reviewed to understand the absorption, distribution, metabolism, and excretion of the FR chemicals. Dose-Response Assessment In the dose-response-assessment phase, the subcommittee reviewed the relationships between increases in the dose of an FR chemical and changes in the magnitude of the incidence or severity of toxic effects. For all types of toxic effects other than cancer, the procedure used to quantify the dose-response relationships involved estimating the highest dose at which no adverse effects were observed. This exposure level is called the no- observed-adverse-effect level (NOAEL). In contrast, the lowest-observed-adverse-effect level (LOAEL) is the lowest dose at which a statistically or biologically significant increase in an adverse effect was reported. The NOAEL is the highest exposure level below the LOAEL at which no statistically or biologically significant increase was observed in the frequency or magnitude of an adverse effect when compared with a control group. The NOAEL or LOAEL (A LOAEL was used only when a NOAEL could not be identified) was then divided by a composite of one or more uncertainty factors (UFs) to obtain a reference dose (RfD), which 

EXECUTIVE SUMMARY 4 is an estimate of lifetime daily dose that is believed to have a reasonable certainty of no harm; for inhalation exposures, similarly defined reference concentrations (RfCs) were calculated or mathematically scaled from the RfDs. The UFs were applied to account for interspecies and intraspecies variation, differences in exposure durations and routes, use of a LOAEL when a NOAEL was not identified, adequacy of experiments, and data quality. For some types of toxic effects, notably most cancers, the subcommittee conservatively assumed that no threshold for a dose-response relationship exists or that, if one does exist, it is very low and cannot be reliably identified. Therefore, the subcommittee's risk-estimation procedure for carcinogens was different from that for noncarcinogens. In the risk-estimation procedure for carcinogens, the relationship between the incidence of cancer and the dose of a chemical reported in an epidemiological study or an experimental animal study was extrapolated linearly to much lower doses at which humans might be exposed. This procedure overestimated conservatively the excess lifetime risk of cancer resulting from lifetime exposure to a chemical at a particular dose rate. This procedure does not provide a “safe” dose with an estimated risk of zero (except at zero dose), although at sufficiently low doses, the estimated risk becomes very low and is regarded to have no public-health significance. The relationship between average lifetime daily dose and tumor incidence was fitted to a mathematical model to predict the incidence at low doses. Several such models are widely used. The subcommittee used the linearized multistage no-threshold model because it provides a conservative risk estimate (i.e., it tends to overestimate, rather than underestimate the risk). The subcommittee applied a statistical confidence-limit procedure to this risk estimate to generate an upper bound (i.e., cancer potency factor) on cancer risk. Although the actual risk cannot be determined, the subcommittee concludes that the actual risk will not exceed the upper bound. The actual risk is also highly likely to be lower than the upper bound, and it might be zero. Exposure Assessment Exposure assessment is the third phase in the risk-assessment process. Because exposures to FRs in treated residential furniture fabrics have not been studied, there are no quantitative measurements of exposures under relevant exposure conditions. The subcommittee assumed that human exposure to FR-treated fabric in homes can occur potentially via skin contact, ingestion (specifically for infants or children who might suck or chew on fabric), inhalation of particles generated during abrasion of surface fibers, and inhalation of vapors 

EXECUTIVE SUMMARY 5 off-gassing from treated fabric. In estimating exposures to FRs, the subcommittee evaluated three exposure scenarios that involved different exposure routes: dermal, oral, and inhalation. For dermal exposure, the scenario was that of an adult sitting on FR-treated fabric of a couch for a substantial fraction (25%) of the time, with potential exposure over 25% of the upper torso area; clothing was conservatively assumed to provide no barrier to exposure. For ingestion, the scenario was of an infant or a child repeatedly sucking on FR-treated fabric of a chair or couch. For inhalation, the scenario was of a person spending time in a closed (but ventilated) room containing FR- treated upholstered furniture that shed FRs as small (respirable) particles, or from which FR chemicals evaporate. In all of those exposure scenarios, the subcommittee intentionally overestimated exposures by using extremely conservative assumptions. Those assumptions are discussed in detail in Chapter 3. Risk Characterization In the final phase of the risk-assessment process, the subcommittee integrated data and analyses from the other three phases (hazard identification, dose-response assessment, and exposure assessment) to determine the likelihood that individuals might experience adverse effects from the FR chemical under anticipated conditions of exposure. To characterize the health risk from exposure to a noncarcinogenic chemical, a hazard-index approach was used to judge whether a particular exposure would be likely to present a noncancer toxicological risk. A hazard index was calculated for each chemical by dividing the estimated human dose by the RfD or RfC. In the absence of adequate human carcinogenicity data for any FR chemical, the subcommittee's approach for estimating cancer risks from exposure to carcinogenic FRs involved the extrapolation of observations of cancer at relatively high doses in laboratory animals to much lower doses anticipated for humans in residential settings. The upper limit on the cancer potency factor extrapolated from animal experiments was multiplied by the estimated lifetime average dose rate to estimate an upper limit on lifetime cancer risk. The subcommittee did not recommend an acceptable cancer risk level for carcinogenic FR chemicals, because that is a regulatory policy question, not a scientific matter. Some regulatory agencies consider as acceptable excess lifetime cancer risks ranging from 1 in 10,000 (1×10−4) to 1 in 1 million (1×10−6) exposed people. 

EXECUTIVE SUMMARY 6 SUMMARY OF TOXICOLOGICAL RISK ASSESSMENTS The subcommittee's toxicological risk assessments for each of the 16 FR chemicals are presented in Chapters 4–19 and summarized in Table ES-1. Table ES-1 shows the critical toxicity end points used for derivation of RfDs, RfCs, or cancer potency factors, estimated worst-case human exposure levels, hazards indices for noncancer effects, and upper limits on lifetime excess cancer-risk estimates for carcinogenic chemicals. In the absence of adequate toxicity data to derive dermal RfDs, the subcommittee characterized potential risks from dermal exposures by using oral RfDs to calculate the hazard indices. In addition, sufficient data to derive inhalation RfCs were available for only two FRs. Inhalation RfCs for other FRs were extrapolated from oral RfDs by using typical body weights and breathing rates. Table ES-1 shows that for most of the 16 candidate FRs, the hazard indices for noncarcinogenic effects are less than 1 for all three routes of exposure. FRs with hazard indices of less than 1 are not likely to pose noncancer health risks even at the worst-case exposure levels. FR chemicals with hazard indices greater than 1 might possibly pose noncancer health risks. However, the subcommittee does not necessarily expect adverse effects at hazard indices slightly greater than 1, given the highly conservative assumptions it used to estimate risks. Carcinogenic risk assessments performed on the FRs that were found to be or likely to be carcinogenic indicate that some of the estimated excess cancer risks may be greater than 1×10−6. However, the subcommittee believes that actual carcinogenic risk is likely to be much lower because of the extremely conservative (high) exposure estimates. Several of the 16 chemicals were actually chemical classes rather than single compounds. In some of these cases, one chemical of the class was selected as a surrogate: tetrakis (hydroxymethyl) hydronium chloride for the tetrakis (hydroxymethyl) hydronium salts and their compounds; dimethyl hydrogen phosphite for organic phosphonates; and tricresyl phosphate for aromatic phosphate plasticizers. Surrogates were selected on the basis of representativeness of the class, availability of data, and most potent chemical in the class. Conclusions about the class are based on the properties of the surrogate. The risk from other members of the class might be different from the risk from the surrogate. The subcommittee's use of several UFs in the derivation of RfDs or RfCs and the intentional overestimation of exposure levels reflects a precautionary approach to the protection of public health. Such an approach is commonly practiced, but the subcommittee is aware that there are potential shortcomings in taking such an approach. Overestimating risks from FRs might result in a net adverse effect on public health if the uses of FRs that could reduce the risks of death and injury from fires were avoided because of minor toxicological risks estimated through such conservative assumptions. 

TABLE ES-1 Summary of Health Risk Assessments of 16 Flame-Retardant Chemicals Critical Dermal Oral Estimated Worst-case Human Hazard Indexa for Upper Limits on Lifetime Excess Toxicity End Exposure Levels Non-Cancer Effects Cancer-Risk Estimateb Flame-Retardant Point for RfD Inhalation RfD Cancer Dermal Inhalationc Oral Dermald Inhalatione Oral Dermal Inhalation Oral Chemical Derivation of (mg/ RfC (mg/ (mg/ Potency (mg/ (µg/m3) (mg/ RfD or RfC kg-d) m3) kg- Factor kg-d) kg- d) d) Hexabromocyclododecane Oral: liver N/C N/C 0.2 N/A 1.3× 10 0.48 2.6 6.7× 10 6.8×10−4 0.13 N/A N/A N/A toxicity −6 (particles) ×10 −6 (particles) EXECUTIVE SUMMARY 3.4 −2 5.0×10−3 (vapors) (vapors) Decabromodiphenyl oxide Oral: liver N/C N/C 4.0 9.0×10−4 1.3× 10 0.48 2.6× 3.3× 10 3.4×10−5 6.5 1.2× 10 1.2×10−7 6.7× toxicity per mg/kg- −9 (particles) 10−2 −10 (particles) × 10 −12 (particles) 10−7 d (oral) 0.38 2.7×10−5 −3 9.7×10−8 2.6×10−7 (vapors) (vapors) (vapors) per µg/m3 (inhalation) Alumina trihydrate Oral: N/C N/C 1.5 N/A 5.9× 10 0.71 1.6× 3.9× 10 1.4×10−4 1.0 N/A N/A N/A developmental −2 (particles) 10−3 −2f × 10 toxicity N/C −3 (vapors) Magnesium hydroxide Oral: derived N/C N/C 12.0 N/A 1.7× 10 0.38 2.1× 1.4× 10 9.1×10−6 1.7 N/A N/A N/A from −3 (particles) 10−2 −4 × 10 tolerable N/C −3 upper limit (vapors) for Mg- induced diarrhea in humans Zinc borate Oral: N/C N/C 0.6 N/A 6.3× 10 0.19 1.7× 1.0× 10 9.1×10−5 2.8 N/A N/A N/A developmental −3 (particles) 10−4 −2 × 10 toxicity N/C −4 (vapors) 7 

Critical Dermal Oral Estimated Worst-case Human Hazard Indexa for Non- Upper Limits on Lifetime Excess Cancer- Toxicity End Exposure Levels Cancer Effects Risk Estimateb Flame- Point for RfD Inhalation RfD Cancer Dermal Inhalationc Oral Dermald Inhalatione Oral Dermal Inhalation Oral Retardant Derivation (mg/ RfC (mg/ (mg/ Potency (mg/ (µg/m3) (mg/ Chemical of RfD or kg-d) m3) kg-d) Factor kg-d) kg-d) RfC Calcium and Oral: N/C 2.0× 10−3 6.0× 2.6×10−5 6.3× 10 0.19 1.7× 10 9.5×10−2 0.28 N/A 5.0×10−6 N/A zinc molybdates increased 10−4 per µg/m3 −3 (particles) 10−4 uric acid (inhalationf N/C EXECUTIVE SUMMARY levels; ) (vapors) Inhalation: degeneration of respiratory epithelium Antimony Oral: liver N/C 2.0× 10−4 0.2 7.1×10−4 2.0× 10 0.24 5.2× 0.1 1.2 2.6 N/A 1.7×10−4 N/A trioxide toxicity; per µg/m3 −2 (particles) 10−4 × 10 (particles) Inhalation: (inhalationf N/C −3 noncancer ) (vapors) pulmonary toxicity; lung tumors Antimony Inadequate N/C N/C N/C N/A 2.0× 10 0.24 5.2× —g —g —g N/A N/A N/A pentoxide and data for any −2 (particles) 10−4 sodium route N/C antimonate (vapors) Ammonium Oral: N/C N/C 300 N/A 2.2 0.71 5.9× 7.3× 10 6.8×10−7 2.0 N/A N/A N/A polyphosphates calcification (particles) 10−2 −3 × 10 of the kidney N/C −4 (vapors) 8 

Phosphonic acid Inadequate N/C N/C N/C N/A 2.8× 0.35 7.5× __ h __ h __ h N/A N/A N/A data for any 10−2 (particles) 10−4 route UE (vapors) Organic Oral: lung N/C N/C 0.12 5.4×10−3 per 2.2 0.72 5.9× 18.3 1.7×10−3 0.49 6.1 1.1×10−6 9.1× phosphonates hyperplasia mg/kg-d (particles) 10−2 (particles) ×10−2 (particles) 10−6 (dimethyl hydrogen and alveolar/ (oral) UE (vapors) 6.6×10−4 phosphite) bronchiolar 1.5×10−6 per (vapors) adenomas or µg/m3 carcinomas (inhalationf) EXECUTIVE SUMMARY observed Tris Inadequate N/C N/C N/C N/A 1.5 0.48 4.0× —g —g —g N/A N/A N/A (monochloropropyl) data for any (particles) 10−2 phosphates route UE (vapors) Tris (1, 3- Oral: N/C N/C 5.0× 6.0×10−2 per 2.6× 0.48 4.0× 0.52 2.7×10−2 8.0 1.6 8.2×10−6 6.6× dichloropropyl-2) testicular 10−3 mg/kg-d 10−3 (particles) 10−2 (particles) ×10−4 (particles) 10−5 phosphate atrophy and (oral) UE (vapors) UE (vapors) seminal 1.7×10−5 per vesicle µg/m3 effects (inhalationf) Aromatic phosphate Oral: liver N/C N/C 7.0× N/A 3.0× 0.48 4.0× 4.3× 1.9×10−3 0.57 N/A N/A N/A plasticizers (tricresyl lesions and 10−2 10−3 (particles) 10−2 10−2 (particles) phosphate) adrenal 417 (vapors) 1.7 (vapors) gland toxicity Tetrakis Oral: liver N/C N/C 3.0× N/A __i 0.43 9.4× N/C 4.1×10−2 0.31 N/A N/A N/A hydroxymethyl toxicity 10−3 (particles) 10−4 (particles) phosphonium UE (vapors) chloride (THPC) 9 

Critical Dermal Oral Estimated Worst-case Human Hazard Indexa for Non- Upper Limits on Lifetime Excess Cancer- Toxicity Exposure Levels Cancer Effects Risk Estimateb End Flame- Point for RfD Inhalation RfD Cancer Dermal Inhalationc Oral Dermald Inhalationc Oral Dermal Inhalation Oral Retardant Derivation (mg/ kg- RfC (mg/ (mg/ Potency (mg/ kg- (µg/m3) (mg/ Chemical of RfD or d) m3) kg-d) Factor d) kg-d) RfC Chlorinated Oral: Liver N/C N/C 0.3 N/A 0.59 0.28 1.6× 1.9 2.7×10−4 5.3 × N/A N/A N/A paraffins and kidney (particles) 10−2 10−2 EXECUTIVE SUMMARY toxicity N/C (vapors) aThe hazard index is calculated by dividing exposure levels by RfDs or RfCs. A hazard index of <1 indicates that the exposure at the intended levels is not likely to pose noncancer health risks; a hazard index of >1 was considered to possibly pose a concern for noncancer effects. bLifetime excess cancer risk above the background lifetime cancer incidence was calculated by multiplying the cancer-potency factor by the exposure estimate. For all routes of exposure, lifetime risk calculations were estimated from lifetime average exposures. cVapor exposure levels were calculated based on the vapor pressure measurements for unreacted starting material. In reality, most or all material becomes bound to upholstery fabric following curing. Vapor exposure levels for chemical entities formed during the curing process were not calculated. dToxicity information was not available to derive a dermal RfD; the subcommittee used oral RfDs as best estimates for internal dose from dermal exposure. eToxicity information was not available to derive an inhalation RfC; inhalation RfCs were estimated from oral RfD data using Equation 7 in Chapter 3 to estimate risk. fThe cancer-potency factor following inhalation is for exposure to particles and vapors. gThere are inadequate toxicity data from any route of exposure to derive RfDs or RfCs for these compounds. However, structurally related compounds were found to be a health concern at the worst-case exposure levels. Therefore, the subcommittee recommends that exposure measurements be made to determine the need for toxicity studies. hThere are inadequate toxicity data for phosphonic acid from any route of exposure to derive RfDs or RfCs; this chemical is likely to crosslink with fabric components, precluding dermal exposure to this FR in furniture upholstery. iNot calculated because THPC is likely to polymerize after application to the upholstery fabric. Abbreviations: N/A: not applicable because the chemical is not carcinogenic by the relevant route; N/C, not calculated because of inadequate data; UE, modeling produces unrealistic short-term exposure estimates for this material; therefore, maximum exposures are unknown. 10 

EXECUTIVE SUMMARY 11 Despite the lack of a complete database, the subcommittee concludes that the following FRs can be used on residential furniture with minimal risk, even under worst-case assumptions: • hexabromocyclododecane, • decabromodiphenyl oxide, • alumina trihydrate, • magnesium hydroxide, • zinc borate, • ammonium polyphosphates, • phosphonic acid (3-{[hydroxymethyl]amino}-3-oxopropyl)-dimethyl ester,1 • tetrakis hydroxymethyl phosphonium salts (chloride salt) On the basis of the hazard indices for noncancer effects and/or the potential for cancer, the subcommittee recommends that exposure studies be conducted on the following FRs to determine whether toxicity studies need to be conducted: • antimony trioxide, • antimony pentoxide and sodium antimonates,2 • calcium and zinc molybdates, • organic phosphonates (dimethyl hydrogen phosphite), • tris(monochloropropyl) phosphates,2 • tris(1, 3-dichloropropyl-2) phosphate, • aromatic phosphate plasticizers (tricresyl phosphate), and • chlorinated paraffins. It is possible that an individual could be exposed by all three routes: oral, dermal, and inhalation. In such cases, the hazard indices or cancer-risk estimates may be summed across the various routes of exposure. This approach is extremely conservative, because it is unlikely that an individual would be 1There are inadequate toxicity data for phosphonic acid from any route of exposure to derive RfDs or RfCs; this chemical crosslinks with fabric components, precluding human exposure to this FR in furniture upholstery. Therefore, this chemical can be used safely as an FR 2There are inadequate toxicity data from any route of exposure to derive RfDs or RfCs for these compounds. However, structurally related compounds—antimony trioxide and tris(1, 3-dichloropropyl-2) phosphate—were found to possibly be health concerns at the worst-case exposure levels. Therefore, the subcommittee recommends that exposure measurements be made to determine the need for toxicity studies. 

EXECUTIVE SUMMARY 12 exposed at the upper limit for one route of exposure and even less likely that the same individual would be exposed at the upper limits for two or more routes. Because one route of exposure typically dominates the risk assessments, summing the hazard indices or cancer-risk estimates does not materially change the conclusions regarding the safety of FRs. UNCERTAINTIES ASSOCIATED WITH RISK ESTIMATES The subcommittee recognizes that there are major uncertainties associated with its toxicological risk estimates. The uncertainties stem from the inadequacy or absence of relevant toxicity and exposure data. In the absence of an adequate toxicity database, the subcommittee applied UFs to account for interspecies and intraspecies differences, route-of-exposure differences, less than lifetime exposure studies, absence of a NOAEL, and inadequate or inferior data. Depending on the information available, the subcommittee used several UFs, each ranging from 1 to 10. However, the exact magnitude of each UF that should be used is unknown, and these uncertainties can be reduced only by research to provide the needed information. The subcommittee identified no quantitative measurements of exposure to FRs under conditions approximating their use in residential furniture upholstery. In the absence of relevant exposure information, the subcommittee made extremely conservative assumptions to overestimate the levels; therefore, there is considerable uncertainty in these exposure estimates. In the absence of any relevant exposure information, the subcommittee is unable to quantify the magnitude of uncertainty associated with these exposure estimates. The subcommittee could not locate any data on differential toxicity of the FRs in various susceptible human subpopulations or in experimental animals. Therefore, it was unable to develop risk estimates for susceptible subpopulations. The uncertainty regarding their risk can only be reduced by studying the toxicity or toxicokinetics in young or aged animals or in animals with certain pre-existing disease conditions. In some cases, the actual chemical form of an FR chemical in treated fabric is different from the pure chemical because of chemical reactions (e.g., polymerization and cross-linking) with components of the fabric or reactions with FR-formulation components during the manufacturing or curing processes. If the chemical form of the applied FR changes in the treated fabric, the subcommittee's risk estimates might be inaccurate. 

EXECUTIVE SUMMARY 13 DATA GAPS AND RESEARCH RECOMMENDATIONS On the basis of its evaluation of toxicity and exposure data on FR chemicals, the subcommittee identified data gaps and research needs. The subcommittee concludes that it is unnecessary to fill all data gaps and recommends that research be done only to reduce key uncertainties for performing toxicological risk assessments. The subcommittee believes that obtaining appropriate exposure measurements for some of the FRs can provide relevant information quickly and inexpensively. If research shows that actual exposures are lower than the subcommittee's conservatively overestimated levels, there may be no need or a reduced need to perform toxicity studies that are much more expensive and time-consuming. To estimate exposure, the subcommittee recommends that research be done for some FRs to measure (1) extraction of FRs from treated fabric into saline, (2) volatilization of FRs, and (3) potential for FRs to be released from treated fabric during wear that could lead to the generation of airborne particles that contain FRs. The subcommittee recommends that the CPSC collect such information and do its own risk assessments. Once exposure data are collected, if the CPSC desires to derive RfDs or RfCs with greater confidence for those FRs that have hazard indices of greater than 1, the subcommittee recommends a tiered research approach. In this approach, the first step may include in vitro tests for genotoxicity and other effects, and short-term (e.g., 28-day) toxicity studies via the relevant routes for assessing various toxicity end points (e.g. neurotoxicity, pulmonary toxicity, and reproductive and developmental toxicity). If the results of the research show no basis for concern for toxicity at the subcommittee's worst-case estimated exposure levels, then no further research is recommended, and the FR can be used safely. If the studies indicate a basis for concern about adverse effects for anticipated human exposure conditions, the subcommittee recommends that the chemical be evaluated in a 90- day subchronic study. If the results of the subchronic study do not show a health concern, no further research is recommended. The subcommittee recommends that a chronic toxicity study be done only when the results of short-term and subchronic studies indicate a basis for concern about cancer at expected exposure levels. Because of the extremely conservative assumptions it used in deriving RfDs and RfCs and in estimating exposure levels, the subcommittee does not recommend further research for noncancer effects for those FR chemicals that have hazard indices of less than 1 (see Table ES-1). 

EXECUTIVE SUMMARY 14