Occupational Exposure to Carbon Nanotubes and Nanofibers

cell surface area). However, the working lifetime8-hr TWA concentrations, BMC(L)s, based on theestimated retained lung doses are higher than thosebased on the estimated deposited lung dose. This isbecause the retained dose estimates (which assumesome particle clearance from workers’ lungs duringthe 45 years of exposure), require a higher inhaledairborne concentration to reach the estimated human-equivalentBMD(L) lung doses.The estimated deposited lung dose of CNT (assumingno clearance) may overestimate the actual CNTlung dose, given that the short-term kinetic datahave shown some CNT clearance in rats and mice[Muller et al. 2005; Deng et al. 2007; Elgrabli et al.2008b; Mercer et al. 2009; Pauluhn 2010a, b]. Onthe other hand, the estimated retained lung dose ofCNT, based on models for poorly soluble sphericalparticles, may underestimate the retained CNTlung burden, given that overloading of rat lungclearance has been observed at lower mass doses ofMWCNT (Baytubes) than for other poorly solubleparticles [Pauluhn 2010a,b]. Thus, although thereis uncertainty in the deposition and retention ofCNT in the animal and human lungs, the depositedand retained lung dose estimates reported in thisrisk assessment may represent reasonable upperand lower bounds of the actual lung doses.A.4.4 Critical EffectLevel EstimatesThe response endpoints in these animal studies ofCNT are all relatively early-stage effects. Althoughthese effects were persistent or progressive afterthe end of exposure in some studies, there was noinformation on whether these responses were associatedwith adverse functional effects. More advanced-stageresponses (grade 2 or higher severityon histopathology examination) were also evaluated,and as expected, these responses resulted inlower risk estimates (Table A–6). It is expected thatexposure limits derived from these early responsedata would be more protective than those based onfrank adverse effects. On the other hand, becauseof the lack of chronic studies, there is considerableuncertainty about the potential chronic adversehealth endpoints.The excess risk estimates at the lower LOQ (1 µg/m3) are considerably lower than those at the upperLOQ (7 µg/m3) of NIOSH Method 5040, for eitherminimal (TableA–7) or slight/mild (Table A–8) lungeffects based on the rat subchronic inhalation data.The range in the estimates in Table A–7 and A–8reflects the low precision in the animal data and theuncertainty about CNT retention in the lungs. Thereis also uncertainty about the relationship betweenthe lung dose and response, including whether thereis a threshold. For example, for slight/mild lungeffects (Table A–8), the actual risk could be as low aszero or as high as 16% at the REL of 1 µg/m3.NIOSH utilized BMD modeling methods to estimatethe critical effect level (i.e., the dose associatedwith the critical effect or benchmark response) inorder to provide a standardized method for riskestimation across studies. In contrast, the NOAELbasedapproaches do not estimate risk, but may assumesafe exposure or zero risk below the derivedOEL. BMD modeling also uses all of the doseresponsedata, rather than only a single dose for aNOAEL or LOAEL, and takes appropriate statisticalaccount of sample size, unlike NOAEL-basedapproaches. However, the BMD modeling optionsfor some of these CNT data were limited becauseof sparse data, and the dose groups with 100%response (observed in the subchronic inhalationstudies) contribute little information to the BMDestimation. A common challenge in risk assessmentis defining a biologically relevant response forcontinuous endpoints, which was also encounteredin this risk assessment. A standard practice of usinga statistical definition of the benchmark responsewas used for the continuous BMD estimation inthe absence of data on the functional significanceof the early-stage pulmonary inflammation and fibroticresponses (Section A.2.3.2).For CNT, as with other chemicals, there is uncertaintyin whether a NOAEL or a BMDL from ashort-term or subchronic study in animals wouldalso be observed in a chronic study. For example,in the Pauluhn [2010a] study, 0.1 mg/m3 was theNIOSH CIB 65 • Carbon Nanotubes and Nanofibers123