Occupational Exposure to Carbon Nanotubes and Nanofibers

used as the effect levels in evaluations of alternativemethods to derive OEL estimates. A quantitativecomparison of possible critical effect levelsis shown in Table A–12. The BMDL estimates aregenerally similar to the NOAEL estimates (within afactor of approximately 1 to 4), which suggests thatthe BMDL estimates may be reasonable despite thesparse data in the low dose region of the subchronicinhalation studies (Figure A–1).A statistical analysis was performed to compare theNOAEL and BMD estimates (in this example, theBMD is an exposure concentration, or BMC). Themaximum likelihood estimate of the excess risk (ofa minimal or higher grade of alveolar septal thickening)at 0.1 mg/m3 is 0.10 (i.e., 10%), based on theBMD model fitted to the dose-response data in thePauluhn [2010a] study (Table A–12). Yet, 0.1 mg/m3was identified as a NOAEL based on zero adverseresponse being observed [Pauluhn 2010a]. In orderto assess the precision of the estimate of the excessrisk associated with this NOAEL, the likelihood ofthe data in the NOAEL and control groups was reparameterizedin terms of the respective sum anddifference of the expected response proportions;and an upper confidence limit for the differencewas assessed by inverting its likelihood ratio teststatistic. When a nominal confidence coefficient of95% for a two-sided interval was applied, a valueof 0.17 (i.e., 17%) was obtained for the UCL of thedifference. Hence, the results supporting the use of0.1 mg/m3 as a NOAEL are also statistically consistentwith the results from the BMD model sincethe MLE of excess risk based on the model is lessthan the UCL.In a standard risk assessment approach, BMDLestimates may be considered equivalent to a NO-AEL for use as a POD in risk assessment [US EPA1994]. Once an effect level is selected in a given animalstudy, it is extrapolated to a human-equivalenteffect level (e.g., as 8-hr TWA concentration),or human-equivalent concentration (HEC). ThisHEC_POD (human-equivalent point-of-departure)is the POD for either extrapolating to a lower (acceptable)risk level or applying uncertainty factors inthe derivation of an OEL. These steps are discussedfurther in Section A.6.3.A.6.3 Alternative OELEstimation MethodsAs mentioned in the previous section, a standardrisk assessment method using animal data typicallyinvolves first identifying a critical effect levelin animals (e.g., NOAEL or BMDL), which is thePOD animal. A HEC_POD is estimated by extrapolatingthe animal dose to humans by accounting for thebiological and physical factors that influence thelung dose across species † . Lung dosimetry modelscan account for these interspecies differencesand provide equivalent dose estimates in animalsand humans given the exposure concentration andduration, the breathing rates and patterns, andthe physical properties of the aerosol. A simplifiedstandard approach in lieu of a lung dosimetrymodel to apply a total dosimetric adjustment factorto the animal effect level (Section A.6.3.1). It isuseful to evaluate both approaches given that thelung dosimetry models have not been specificallyvalidated for respirable CNT.A.6.3.1 Illustration of Human-Equivalent ConcentrationEstimationThe human equivalent concentration HEC) to aPOD animal(e.g., NOAEL) in an animal study can becalculated as:Equation A–8:HEC_POD = POD animal/ DAFwhere DAF is the dosimetric adjustment factor,andEquation A–9:DAF = (VE H/VE R) × (DF H/DF A) × (RT H/RT A) × (NF A/NF H)where VE is the ventilation rate (e.g., as total volumeof air inhaled per exposure day, m3/d) in†HEC_POD is then divided by appropriate uncertaintyfactors (UFs) to account for variability and uncertaintyin its estimation (Section A.6.3.3).NIOSH CIB 65 • Carbon Nanotubes and Nanofibers131