Registration Dossier

Toxicological information

Exposure related observations in humans: other data

Currently viewing:

Administrative data

Endpoint:
exposure-related observations in humans: other data
Type of information:
migrated information: read-across from supporting substance (structural analogue or surrogate)
Adequacy of study:
key study
Reliability:
2 (reliable with restrictions)
Rationale for reliability incl. deficiencies:
other: see 'Remark'
Remarks:
Acceptable well documented publication which meets basic scientific principles Read-across is justified on the following basis: The family of zinc borates that include Zinc Borate 500, Zinc Borate 2335 and Zinc Borate 415 (also known as Zinc Borate 411). Zinc borate 500 is anhydrous Zinc Borate 2335 and Zinc Borate 415 has different zinc to boron ratio. Zinc borate 2335 (in common with other zinc borates such as Zinc borate 415 and 500) breaks down to Zinc Hydroxide (via Zinc oxide) and Boric Acid, therefore the family of zinc borates shares the same toxicological properties. Zinc borates are sparingly soluble salts. Hydrolysis under high dilution conditions leads to zinc hydroxide via zinc oxide and boric acid formation. Zinc hydroxide and zinc oxide solubility is low under neutral and basic conditions. This leads to a situation where zinc borate hydrolyses to zinc hydroxide, zinc oxide and boric acid at neutral pH quicker than it solubilises. Therefore, it can be assumed that at physiological conditions and neutral and lower pH zinc borate will be hydrolysed to boric acid, zinc oxide and zinc hydroxide. Hydrolysis and the rate of hydrolysis depend on the initial loading and time. At a loading of 5% (5g/100ml) zinc borate hydrolysis equilibrium may take 1-2 months, while at 1 g/l hydrolysis is complete after 5 days. At 50 mg/l hydrolysis and solubility is complete (Schubert et al., 2003). At pH 4 hydrolysis is complete. Zinc Borate 2335 breaks down as follows: 2ZnO • 3B2O3 •3.5H2O + 3.5H2O + 4H+ ↔ 6H3BO3 + 2Zn2+ 2Zn2+ + 4OH- ↔ 2Zn(OH)2 ____________________________________________________________ Overall equation 2ZnO • 3B2O3 •3.5H2O + 7.5H2O ↔ 2Zn(OH)2 + 6H3BO3 The relative zinc oxide and boric oxide % are as follows: Zinc borate 2335:zinc oxide = 37.45% (30.09% Zn) B2O3 = 48.05% (14.94% B) Water 14.5% Zinc borate 415: zinc oxide = 78.79%; (63.31% Zn) B2O3 = 16.85% (5.23% B) Water 4.36% Zinc borate, anhydrous: Zinc oxide = 45 % B2O3= 55% (17.1 % B)

Data source

Reference
Reference Type:
publication
Title:
Derivation of an occupational exposure limit for inorganic borates using a weight of evidence approach
Author:
Maier, A., Vincent, M., Hack, E., Nance, P. and Ball, W.
Year:
2014
Bibliographic source:
Regulatory Toxicology and Pharmacology 68 (2014) 424-437

Materials and methods

Endpoint addressed:
other: Derivation of occupational exposure limit
Test guideline
Qualifier:
no guideline followed
GLP compliance:
no
Remarks:
not applicable (it is a review)

Test material

Reference
Name:
Unnamed
Type:
Constituent

Results and discussion

Results:
The data-derived intraspecies factor of 1.5 and is reported for workers for toxicokinetic variability. The US EPA (2004) divided the POD (10.3 mg B/kg-day) by a composite UF of 66 for deriving their Reference Dose (RfD). This composite factor was calculated by multiplying the subfactors of 3.3 for interspecies differences in toxicokinetics (based on data for boron clearance rates in rats versus humans), a default value of 3.2 for interspecies differences in toxicodynamics, a value of 2.0 for variability in human toxicokinetics (based on data on human variability in glomerular filtration rate), and the default factor of 3.2 to account for variability in human toxicodynamics

Any other information on results incl. tables

The relevant human effects and toxicology data were analyzed to array candidate endpoints that might ultimately serve as the most sensitive basis for OEL derivation. Existing risk assessment documents allowed for a focused evaluation of two potential systemic effects of interest (reproductive or developmental toxicity), and portal of entry effects (sensory irritation of the upper respiratory tract).

Reproductive toxicity

Toxicology studies indicate a potential concern for effects on the male reproductive tract. Testicular effects reported in dogs and rats by Weir and Fisher (1972) have been supported by ad¬verse reproductive findings from numerous other studies in laboratory animals (see US EPA, 2004 for a comprehensive review). These effects, however, have not been observed in humans. Multiple modern epidemiological studies on boron-exposed cohorts have examined the relationship between boron exposures, both inhaled and ingested, and changes in reproductive outcomes. These studies did not identify any increased risk of adverse semen parameters or other reproductive effects (Duydu et al., 2011; Liu et al., 2006; Robbins et al., 2010; Sayli, 1998, 2001, 2003; Scialli et al., 2010; Whorton et al., 1994), despite the presence of higher blood boron levels in boron workers, as compared to controls (Duydu et al., 2011; Robbins et al., 2011; Xing et al., 2008). Other potential reproductive effects have been identified in epidemiology studies of boron exposed populations, but they are not necessarily indicative of or correlated with other adverse or clinically observable reproductive effects and are limited in their application for risk assessment. A decrease in the Y:X sperm ratio was identified in male boron workers (Robbins et al., 2008). However, an earlier study did not identify a significant increase in the percentage of female offspring born to male boron workers (Whorton et al., 1994). The lack of concordance in these two findings clouds the relevance of this finding for risk assessment. Chang et al. (2006) reported a significant delay in pregnancy, an increase in induced abortion, and a decreased number of live births among the families of Chinese boron workers, but this effect is likely related to lifestyle factors rather than boron exposure; after adjusting their models for the covariates of age, education, soybean, alcohol, and tobacco usage, and race, the authors found no significant relationship between these effects and boron exposure. Overall, the epidemiology data indicate that there is little risk for adverse reproductive effects at the concentrations to which boron workers are exposed for two reasons: (1) the majority of studies showed no observable effect for reproductive parameters and (2) the few studies that did report effects were not consistent with the body of evidence and/or the effects reported were ascribed to confounding exposures, two important elements of causality analysis in epidemiology studies (Hill, 1965). Additionally, the doses associated with testicular effects in animal toxicology studies are higher than those associated with developmental toxicity. Since the UFs that would be applied to this endpoint do not differ significantly from those for the developmental endpoint, the OEL for reproductive toxicity would not be lower than for developmental toxicity. For this reason, an OEL for this endpoint is not further explored.

Developmental toxicity

U.S. EPA derived its reference dose (RfD) (2004) based on the 5% lower bound on the benchmark dose estimate (BMDL05) of 10.3 mg B/kg-day for decreased fetal weight in rats (Allen et al., 1996), using data from several developmental toxicity studies (Heindel et al., 1992; Price et al., 1994, 1996). Other recent risk assessments (ATSDR, 2010; WHO, 2009) identified the same POD, and no new epidemiology or toxicology studies were identified for the current analysis that support lower POD estimates. The critical effect level of 10.3 mg B/kg-day was determined to be an appropriate POD for OEL development after adjustment to an inhalable concentration equivalent. The US EPA (2004) divided the POD (10.3 mg B/kg-day) by a composite UF of 66 for deriving their Reference Dose (RfD). This composite factor was calculated by multiplying the subfactors of 3.3 for interspecies differences in toxicokinetics (based on data for boron clearance rates in rats versus humans), a default value of 3.2 for interspecies differences in toxicodynamics, a value of 2.0 for variability in human toxicokinetics (based on data on human variability in glomerular filtration rate), and the default factor of 3.2 to account for variability in human toxicodynamics. Applica¬tion of the composite UF of 66 resulted in a RfD of 0.16 mg B/kg- day. The World Health Organization (WHO, 2009) chose the same POD as was used in the U.S. EPA assessment, but derived a different composite UF. Using international defaults in the CSAF methodol¬ogy (which vary slightly from the U.S. EPA methods), WHO (2009) calculated a composite UF of 60. For the current assessment, a composite UF of 40 was derived based on the IPCS (2005) approach for CSAFs and additional considerations appropriate for risk assessment in an occupational setting (see Table 2). The composite UF of 40 resulting from our analysis reflects the same CSAF considerations as applied in the US EPA (2004) and WHO (2009) assessments, but with further refinement of the toxicokinetic subfactor for human variability in sensitivity. The toxicokinetic adjustment factor for human variabil¬ity was calculated in the US EPA (2004) assessment from data on the variability in glomerular filtration rate (GFR) during pregnancy; GFR was identified as the primary determinant of boron clearance rates. The US EPA (2004) modified the sigma method (Dourson et al., 1998) to calculate the lower bound of risk at 3 standard deviations (SD) instead of 2 with the goal of ensuring adequate coverage of preeclamptic women (the sensitive subpopulation), resulting in a recommended intraspecies (i.e., human variability) toxicokinetic adjustment factor of 2. The estimated lower bounds for acceptable risk at both 2 and 3 SDs, using the Sigma Method, are shown in Table 3. The intraspecies factor for toxicokinetic var¬iability can be further reduced to a data-derived value of 1.5 for occupational assessment based on variability in GFR in populations after excluding preeclamptic women (Dunlop, 1981; Krutzen et al., 1992; Sturgiss et al., 1996). The results from these three studies were averaged to increase the sample size and therefore better re¬flect the overall population distribution, including median re¬sponse and variability. The decision to average the results from these three studies also reflects that none of the available studies were significantly more robust than the others. Because the OEL is intended to protect working populations, it is not appropriate to include preeclamptic women, since they would not likely be represented in the work place. Of the total workforce, pregnant women represent a relatively small percentage of total workers. Women with preeclampsia represent an even smaller subset of this population; overall incidence of preeclampsia is estimated at roughly 3%, or less, of total pregnancies (Thornton et al., 2013; WHO, 2005). Women diagnosed with mild preeclampsia are given outpatient, and sometimes inpatient, treatment including blood pressure measurements, laboratory monitoring, physician visits twice weekly and, generally, bed rest, although this is no longer recommended as routine management of hypertension in pregnancy by the American College of Obstetricians and Gynecologists (ACOG, 2013). Women with preeclampsia are, therefore, unlikely to be working during this period of sensitivity. Although it is possible that a pregnant woman with preeclampsia could be found in an occupational setting, specifically if the woman had not been receiving prenatal care, the percentage of the working population, under such circumstances, would be very small. Additionally, working populations are more homogenous than general populations, so increasing the lower bound to 3 SDs instead of 2 is unnecessary for an occupational assessment, especially when preeclamptic women are excluded. Moreover, the GFR values for preeclamptic women are approximately 2 SD below those of healthy women (Krutzen et al., 1992), indicating that using the sigma-method with 2 SD, at most, is adequate. Application of the principles of the IPCS MOA framework showed that the evidence is not sufficient to reasonably exclude the human relevance of the observed developmental effects in animals. Thus, developmental toxicity is considered the most appropriate potential systemic endpoint as the basis for the OEL. The best POD, based on currently available data, would be 10.3 mg of boron/kg-day (66 mg B/m3) with a UF of 40, resulting in a best OEL estimate of 1.6 mg B/m3 for developmental effects.

Table 3
Calculation of lower bound estimates for acceptable risk using the Sigma Method from Dourson et al. (1998) and information on glomerular filtration rates (GFR). Estimates indicate that a toxicokinetic (TK) uncertainty factor (UF) of 1.5 is appropriate based on the traditional sigma method approach based on 2 SD.
Study Mean GFR (SD) (mL/min) GFR at 2 SD below mean GFR at 3 SD below mean TK UF estimate based on Sigma-Method Value for 2 SD TK UF estimate based on Sigma-Method Value for 3 SD
Dunlop (1981) 150.5 (17.6) 115.3 97.7 1.31 1.54
Sturgiss et al. (1996) 138.9 (26.1) 86.7 60.6 1.60 2.29
Krutzen et al. (1992) 195 (32) 131 99 1.49 1.97
Krutzen et al. (1992) (preeclamptic population) 128 (33.9) 60.2 26.3 2.13 4.87
Average (excluding preeclamptic population) 161.5 111.0 85.8 1.47 1.93

Sensory irritation

At extremely high aerosol concentrations in animal lethality studies, some respiratory effects do occur (e.g., nasal inflammation). However, none of the inorganic borates are highly toxic in acute lethality studies, with LC50 values for several different inorganic borates reported above 2000 mg/m3, the highest concentration tested (reviewed in Hubbard (1998)). No effects on organ pathology (including of the respiratory tract) were noted and no signs of pneumoconiosis were evident at any concentration. The absence of both significant respiratory tract histopathology and acute lethality in animal studies is consistent with the epidemiology literature on the respiratory tract effects of inorganic borates (Garabrant et al., 1985; Wegman et al., 1994) and supports sensory irritation as the only significant respiratory tract response of interest for OEL development.

A generalized weight of evidence approach was used to integrate the results of several lines of evidence from the toxicology and human studies with the aim of characterizing the intensity of the sensory irritant response induced by exposure to inorganic borates and the concentration-response characteristics of this response. The lines of evidence considered, in order from most to least weight, were: (1) controlled human exposure studies; (2) occupational epidemiology studies; (3) inhalation-based sensory irritation studies in rodents; (4) standard hazard-based irritation assays by non-inhalation routes. The analysis of each of these lines of evidence and the resulting contributions of these data to developing a potency range for borate-induced sensory irritation for setting an OEL is provided below. The results of the BMC modeling for nasal sensation response are summarized in Table 4. With the exception of the linear model, the resulting BMCLs are very similar. The quadratic model is the preferred model since it has the highest goodness of fit p-value and the lowest AIC value (both parameters indicating the best fit), while the absolute value of the chi-square residuals (i.e., a measure of local model performance in the range of the BMR) are similar across the models. The concentration-response data and quadratic model fit are shown in Fig. 2. The resulting BMCL of 9 mg/m3 (1.4 mg B/m3) is consistent with the observations made by Cain et al. (2004) that the perceived magnitude of response was significantly above background at 10 mg/m3 sodium borate and higher. The maximum likelihood estimate (MLE) of the BMC is 13 mg/m3 (2.0 mg B/m3). The results of the BMC modeling for the throat sensory response are summarized in Table 5. The linear model has the highest good¬ness of fit P-value, although the quadratic and Hill models perform nearly as well. For the power model, the power parameter is re¬stricted to be greater than or equal to 1 to prevent an infinite slope in the concentration-response curve at the control dose, causing this model to be reduced to a linear model; resulting the same fit statistics. The absolute values of the residuals do not strongly favor one model over another because at 20 mg/m3, they appear to be smaller (i.e., better) for the linear model, but if one considers the residuals at 5 mg/m3, the Hill and quadratic models most closely match the data (i.e., have the smallest residuals). The AIC for the linear model is significantly better than the other models - reflecting superior overall fit and a smaller number of equation parameters. Thus, all of the models give roughly the same quality of fit to the experimental data, with the linear model somewhat better. The best estimates for the BMC are similar across the models; however, the BMCLs are nearly fivefold different from the lowest to highest estimate. There is sufficient model uncertainty to avoid picking a single model; therefore the results were averaged across the mod¬els. If the three results (excluding the power model result, since that reduced to the same mathematical formulation as the linear model) are averaged, a BMCL of ~8 mg/m3 is obtained (or 1.2 mg B/m3). This value is very close to the BMCL for nasal responses. Selecting the model averaging approach reflects that none of the BMC models used in this assessment are biologically based and to be preferred a priori, rather they are statistical in nature only and are not intended to infer a biological basis for selecting one curve form over another (U.S. EPA, 2012). Occupational epidemiology studies report that dust from inorganic borates can cause symptoms associated with ocular and respiratory tract sensory irritation. Several key studies are summarized in Table 6. The most commonly reported symptoms among exposed workers include dryness of the mouth, nose or throat, eye irritation, dry and/or productive coughs, and sore throat (Eisen et al., 1991; Garabrant et al., 1984, 1985; Wegman et al., 1994). Garabrant et al. (1984, 1985) also reported nose bleed, but this symptom was not identified in any other study. It is noteworthy that the workers in these studies were employed in a facility located in a desert environment, and such symptoms observed at higher rates than controls (including office workers) may reflect in part the general environmental conditions at this location. A lowest observed adverse effect level (LOAEL) for effects more severe than eye irritation was reported at an average boron oxide and boric acid particulate total level of 4.1 mg/m3 measured as total dust using a 37 mm filter method cassette (Garabrant et al., 1984, 1985). The equivalent inhalable mass would be approximately 10 mg/m3 based on demonstration that the 37-mm total dust sampler equipment under-samples suspended particles by factors ranging from 1.2 to 4.0 compared to the IOM sampler (Shen et al., 1991; Culver et al., 1994; Tsai et al., 1995; Werner et al., 1996; Katchen et al., 1998; Teikari et al., 2003; Vincent, 2007). The adjustment for inhalable mass is applicable to these reported upper respiratory tract effects since the IOM sampler would predict deposition of inorganic borate particles in the nose and throat (U.S. EPA, 1994). The dust particles associated with borate mining and processing typically have mass median aerodynamic diameters of 10-15 am and in this environment the IOM sampler collects between 2 and 3 times more mass per unit volume of air than the total dust sampler (Culver et al., 1994; Katchen et al., 1998). A conversion factor of 2.5 has been suggested for converting "total" personal exposure measures from industries similar to the borate mining and processing facility to equivalent inhalable aerosol exposures (Werner et al., 1996; Vincent, 2007). This is further supported by paired 37 mm closed face cassette and 25 mm IOM sampling at a borate processing facility in France (Shen et al., 1991). Symptoms related to pulmonary function or respiration were also reported, but were less common, with 5% of subjects reporting shortness of breath and chest tightness, and fewer than 2% reporting chest pain and hemoptysis (indicative of potential hemorrhage) (Garabrant et al., 1985). Pulmonary function measurements (e.g., forced expiratory volume in one second (FEV1), forced vital capacity (FVC), FEV1/FVC, peak flow rate (Vmax) and forced expiratory flow at 75% (FEF75) were unchanged (Garabrant et al., 1985; Hu et al., 1992). Hu et al. (1992) provided additional analysis on the concentration-response relationship of nasal irritation symptoms (Table 7) in the same cohort. These data indicate that, at concentrations approximating 10 mg/m3 of inspirable borate, the probability of a worker experiencing an irritation event during a 6-h exposure period (TWA-6) is low (i.e., a finding that supports the results of the BMC modeling (see Table 4) and the observation by Cain et al. (2004) that perceived response of sensory irritation will occur at approximately 10 mg/m3 sodium borate and higher. A risk of 1:1000 is often used as a benchmark for acceptable risk probability for OEL setting purposes, stemming from risk assessments supporting OSHA permissible exposure limits (PELs), which are often for irreversible and severe effects. There is no published consensus on the acceptable percentage of the population affected at an OEL for minimal sensory irritation (Paustenbach, 2001; Gaffney and Paustenbach, 2007). The sensory irritation potential of inorganic borates has also been investigated in a recent airway sensory irritation respiratory depression (RD50) study of boric acid and sodium borate conducted in male Swiss-Webster mice based on the ASTM E981-04 (2004) standard test method (Ball et al., 2012). For boric acid, a single 30-min exposure to boric acid dust aerosol at high concentrations produced a maximum decrease in respiratory rate of 24%. Clinical observations included opacity and partial closure of the eye at exposures greater than 513 mg/m3, which were attributed to the high level of dust loading. These results were replicated in a second study that tested concentrations at the limits of the aerosol generation system employed (Ball et al., 2012) a 21% decrease in respiratory rate was seen at the highest boric acid concentrations tested of 1018 mg/m3. Similarly to the results for boric acid, it was not possible to achieve an aerosol concentration high enough to result in 50% respiratory depression in mice for sodium bo¬rate based on the results in the mouse sensory irritation model. The highest concentration of sodium borate that was achievable with acceptable control of the aerosol concentration was 1704 mg/m3 with a resulting decrease in respiratory rate of 33%. Based on these results, the RD50 is greater than 1704 mg/m3 for sodium borate. Although the highest achievable concentration was below the RD50 value for sodium borate, based on the high aerosol concentrations achieved with respiratory rate depression values below 50%, it is clear that boric acid and sodium borate have low potency as sensory irritants. The ASTM standard uses a value of 0.03 x RD50 for estimation of an occupational exposure limit (supported by the analysis of Schaper, 1993). Alarie et al. (1980) has established that a value of 0.01 x RD50 as the concentration where no sensory irritation would be seen in humans. The maximum achievable concentration of 1704 mg/m3 sodium borate was below the RD50. Thus, based on the conclusions of Alarie et al. (1980), the product of 0.01 x 1704 mg/m3 (the maximum achievable concentration) is 17 mg/m3 disodium tetraborate pentahydrate (2.5 mg B/m3), a concentration at which human sensory irritation would not be expected. Based on the ASTM method the OEL would be expected to be greater than 0.03 x 1704 mg/m3 or 51 m3 (7.6 mgB/m3). That absence of significant irritant potency in the RD50 assay (Ball et al., 2012) is consistent with observations from standard irritancy assays, which show that inorganic borates are not skin irritants. Inorganic borates range from negligible to mild as eye irritants. The lack of boric acid irritancy to mucosal surfaces is not surprising in view of the use of saturated boric acid solutions in eye wash applications (ATSDR, 2010). Overall, the results from the standard irritancy tests further highlight the likely sensitivity of the subjective responses from the human volunteer studies, suggesting that derivation of an OEL from such data is likely to be health protective. Of the multiple lines of evidence regarding sensory irritation, the result of the human volunteer study by Cain et al. (2004) provides the most scientifically rigorous POD for the sensory irritation endpoint. Cain et al. (2004) concluded that perceived sensory irritation is significantly increased at concentrations approximating 10 mg/m3 sodium borate and higher. This was selected as the point of departure for the sensory irritation OEL. This result was confirmed in our analysis by additional benchmark dose modeling from the Cain et al. (2004); where a POD of 9 mg/m3 (or 1.4 mg B/m3) was estimated. The most sensitive effect as well as the most relevant and reliable result derives from the human volunteers under controlled conditions. Given differences in the relative level of precision inherent in the study designs comparing well-controlled volunteer studies and work-place epidemiology, the results of Cain et al. (2004) are well supported by the occupational epidemiology literature. The most scientifically rigorous POD based on animal sensory irritation data would be 1704 mg/m3 (254 mg B/m3) from the RD50 studies; not only is this potential POD magnitudes higher than those identified through the Cain et al. (2004) data, but the resulting OEL is much higher (~6-fold higher). For the sensory irritation endpoint, a composite UF value of 1 is appropriate. This reflects the conclusion that the concentration-response estimate from Cain et al. (2004) represents a lower bound estimate of response for a sensitive population. A factor of 1 to account for human variability in response is appropriate when the POD is derived from a NOAEL or NOAEL surrogate (such as the BMCL) and for when the sensitive population serves as the basis for the POD (Haber et al., 2001). This use of a factor of 1 does not infer that there is no variability in human response, rather that sensitive individuals were already accounted for in the estimation of the POD. This conclusion was supported by several considerations. First, in worker populations, the degree of variability based on toxicokinetic or toxicodynamic considerations is expected to be lower than for the general population, since the occupational population is only a subset of the general population. Second, a reduced factor for intraspecies variability is often used for sensory irritants, based on the principle that there is lower variability for direct contact effects than other systemic responses, and that only dynamic, not kinetic variability, is relevant for such effects. A third consideration reflects the nature of the study population in Cain et al. (2004) that consisted of non-smoking, young adults with no current active rhinitis or cold symptoms. The volunteers in the Cain et al. (2004) study represent a population that is at the sensitive end of the distribution for sensory irritants in a general population cohort of adults. Although interindividual variability is signiicant for chemosensory responses, analyses of controlled exposure studies suggests that younger age, presence of allergic rhinitis, and coincident odor stimuli tend to increase nasal sensitivity, while smoking tends to decrease sensitivity (Shusterman, 2002, 2007). Persons with respiratory infections might be more sensitive to some effects of respiratory toxicants, however multivariate logistic regression analysis indicates that smokers and borate workers with colds were less sensitive to irritation than nonsmokers and workers without colds (Eisen et al., 1991). The concordance in effect thresholds estimated from borate exposed workers to that derived from the studies (Cain et al., 2004, 2008) also suggests that the POD used for the OEL is suficiently representative of a worker population NOAEL for sensory irritation and the use of a small UF for this consideration. Nevertheless, the number of subjects in the clinical studies was small and an argument could be made for increasing the factor for human variability as a result. Human variability has already been addressed in the concentration response assessment by the selection of the POD estimate. The use of a minimal composite UF reflects the POD selection approach. Several decisions were made in the modeling process to assure that a sensitive response threshold was identified for the test subjects included in the study. A reduced factor is supported by the use of several health-protective assumptions in conducting the concentration-response modeling that were chosen to increase the overall margin of safety in the assessment. The effect level used as the POD is the lower bound estimate of the sensory response (i.e., BMCL10) rather than the maximum likelihood estimate (i.e., BMC) - reflecting subject response and experimental variability. They selected 15% CO2 as the cut point for deriving the OEL even though a value of 17.7% may have been the best estimate (mean response) for the onset of irritation, based on a subset of test subjects (Cain et al., 2008). Additionally, we used a 5% CO2 response to represent the control response, although a lower background response of 2.5% response may have been suitable. All of these modeling decisions resulted in a lower bound estimate of the point of departure that was meant to ensure that the data are representative of the sensitive human population. Despite these decisions, the POD was developed using a small number of subjects and is not a direct estimate of overall worker variability. To the degree that the risk assessment does not consider that POD as reflective of a sensitive population NOAEL a larger factor for human variability in response could be applied. For the reasons noted above to be concordant with current risk assessment practice, such a factor would need to be 3 or less.

Table 4
Summary of benchmark concentration (BMC) modeling for nose irritation.
Model Chi-square residual (at 5 mg/m3) Chi-square residual (at 20 mg/m3) AIC P-value BMC (mg/m3) BMCL (mg/m3)
Linear 0.592 0.258 221 0.37 18 15
Quadratic 0.118 -0.67 221 0.54 13 9.3
Power -0.182 -0.383 222 0.42 14 7.1
Hill 0.167 -0.613 223 0.23 13 7.4

.

Table 5
Summary of Benchmark Concentration (BMC) Modeling for Throat Irritation.
Model Chi-square residual (at 5 mg/m3) Chi-square residual (at 20 mg/m3) A1C P-value BMC (mg/m3) BMCL (mg/m3)
Linear 0.809 -0.121 234 0.81 17 14
Quadratic 0.628 -0.475 236 0.71 14 7.9
Power 0.809 -0.121 234 0.81 17 14
Hill 0.538 -0.532 236 0.72 13 3.0

Preliminary OEL based on evaluation of animal and human studies

Based on these limited data, an OEL for irritant and systemic effects could be derived using the BMCL10 for nasal sensory irritation in humans (derived from Cain et al., 2004) or the BMDL05 for developmental effects in animals (US EPA, 2004). In comparing the effect levels across endpoints, it was determined that the effects on the male reproductive tract occurred at higher or equal doses than the doses that caused developmental effects and the studies for the developmental endpoints were more reliable for characterizing dose-response (US EPA, 2004). Therefore, in the context of setting an OEL, protection from developmental toxicity should also be adequate to protect from reproductive toxicity as well as toxicity to other systemic target organs. Each of these critical effect levels is adjusted using appropriate uncertainty factors to determine the best OEL estimate per IPCS (2005) CSAF methods. The OEL for protection against systemic effects (based on developmental toxicity as the critical effect) can be calculated as follows: 10.3 mg B/kg-day (the critical effect level for developmental effects) x 58 kg body weight for a female worker/9.1 m3 of air inhaled per 8-h work shift/40 (the UF) = 1.6 mg B/m3 (ICRP, 1975). Body weight and air intake rate default values used by other OEL-setting organizations would yield even higher OEL estimates for the systemic effect. Since the various borate compounds all form boric acid in the body, the appropriate dose metric for systemic toxicity is based on boron equivalents. The OEL for protection against sensory irritation is estimated to be 1.4 mg B/m3, which is essentially the same as that based on developmental toxicity, after consideration of uncertainties (and the degree of precision afforded by risk assessment methods). Thus, the final OEL recommendations show consistency in the prevention of sensory irritation and systemic effects. Irritation, being the more sensitive and least variable of the two endpoints, was selected as the basis for the OEL. Because of the slope of the irritation response, it is possible that less sensitive individuals will have no indications of irritation at this value. Therefore, borates do not necessarily have strong warning properties and the presence of irritation, or lack thereof, should not be used to ensure that acceptable exposure levels are maintained.

Applicant's summary and conclusion

Conclusions:
The US EPA (2004) assessment factors of 1.5 and 2.0 for variability in human toxicokinetics (based on data on human variability in glomerular filtration rate) are reported for sensitive population and workers, respectively. The US EPA (2004) used the default factor of 3.2 to account for variability in human toxicodynamics
Executive summary:

Inorganic borates are encountered i n many settings worldwide , spurring international efforts to develop exposure guidance (US EPA, 2004; WHO, 2009; ATSDR, 2010) and occupational exposure limits (OEL) (ACGIH, 2005; MAK, 2011). We derived an updated OEL to reflect new data and current international risk assessment frameworks. We assessed toxicity and epidemiology data on inorganic borates to identify relevant adverse effects. International risk assessment frameworks (IPCS, 2005, 2007) were used to evaluate endpoint candidates: reproductive toxicity , developmental toxicity , and sensory irritation. For each endpoint a preliminary OEL was derived and adjusted based on consideration of toxicokinetics, toxicodynamics and other uncertainties. Selection of the endpoint point of departures (PODs) is supported by dose-response modeling. Developmental toxicity was the most sensitive systemic effect. An OEL of 1 . 6 m g B / m 3 was estimated for this effect based on a POD of 63 mg B / m3 with an uncertainty factor (UF) of 40. Sensory irritation was considered to be the most sensitive effect for the portal of entry. A n OEL of 1.4 mg B / m3 was estimated for this effect based on the identified POD and an UF of 1. An OEL of 1.4 mg B / m3 as an 8 -h time - weighted average (TWA) is recommended.

The US EPA (2004) divided the POD (10.3 mg B/kg-day) by a composite UF of 66 for deriving their Reference Dose (RfD). This composite factor was calculated by multiplying the subfactors of 3.3 for interspecies differences in toxicokinetics (based on data for boron clearance rates in rats versus humans), a default value of 3.2 for interspecies differences in toxicodynamics, a value of 2.0 for variability in human toxicokinetics (based on data on hu¬man variability in glomerular filtration rate), and the default factor of 3.2 to account for variability in human toxicodynamics