A Practical Approach, Second Edition=Ronald D. Ho.pdf

DEVELOPMENTAL AND REPRODUCTIVE TOXICITY RISK ASSESSMENT 857Law. 111,112 And even though the approach was not intended for exposures of long durations and/orlow concentrations, it has come to be more generally applied and is used in extrapolating toinhalation exposures that are not covered by the experimental data. The reliability of this approachin estimating the effects of a particular exposure at various concentrations and durations is relativelyundefined. However, recent reviews of this area have begun to question its general applicabilityover wide ranges of concentration and duration. 113,114The development of new and more appropriate models should define the relationship betweenexposure concentration and duration. Initial application of such models will require broad assumptionsabout the mode of action. Ultimately, more refined models should permit interpolation of datato untested c × t combinations, reducing the uncertainty in assessing the data and increasingconfidence in the prediction of potential human risk. The duration of exposure, even for agentswith a short half-life, can influence the developmental effects that are produced, as demonstratedby recent studies on all-trans-retinoic acid 115 and ethylene oxide. 116 The effects of exposure tohyperthermia during a brief period of development (GD 10 in the rat) have also been shown to bea function of the level and duration of temperature elevation. 117 Thus, the default adjustment ofinhalation developmental toxicity studies for duration of exposure will likely be more protectivewith adjustment from shorter to longer durations of exposure. 116,117Derivation of the human equivalent concentration (HEC) for inhalation exposures is intendedto account for pharmacokinetic differences between humans and animals, while the potentialpharmacodynamic differences are usually accounted for by a portion of the interspecies uncertaintyfactor (typically 10 0.5 ). The application of dosimetric adjustment factors (DAFs) to derive the HECis discussed in detail in the EPA’s guidance for RfC derivation. 118Currently, there is no similar procedure available to account for pharmacokinetic differencesin deriving the human equivalent dose (HED), and the basis for deriving the dose metric for oralexposure differs between cancer and noncancer risk assessment. In the case of noncancer riskassessment, oral dose is expressed on the basis of body weight 1 (e.g., milligrams per kilogram perday), while for cancer risk assessment, dose is expressed on the basis of body weight to the 0.75power (e.g., milligrams per kilogram 0.75 per day). Harmonization of these approaches is needed formore consistent dose-response assessment.b. Benchmark Dose ModelingThe benchmark dose modeling (BMD) approach has several advantages over the NOAEL approach.For example, dose-response modeling with derivation of a BMD uses all the data in fitting a model,accounts for the variability in the data, and does not limit the BMD to one of the experimentaldoses. The BMD is defined as a dose that corresponds to a particular level of response — thebenchmark response (BMR). The BMR is defined as a particular response rate for a quantalresponse, e.g., 1%, 5%, 10%, or as a specified change from control values for a continuous response,e.g., 1 SD change from the control mean. The BMR must be selected for the derivation of a BMD,and this is not straightforward, particularly in the case of continuous data, because the decision isbased on selecting the magnitude of change from controls that is considered adverse.Studies have been conducted to evaluate several approaches to calculating BMDs for standardprenatal developmental toxicity data. 119–122 These studies applied several dose-response models,both generic and developmental toxicity–specific, to a large number of standard developmentaltoxicity studies in animals (rats, mice, rabbits) that were dosed throughout the period of majororganogenesis (or in some cases throughout pregnancy). These studies showed that such modelscan be used successfully with standard developmental toxicity data. The BMD for a 5% excessrisk above controls for binomial endpoints corresponded on average to the NOAEL. Incorporationof variables, such as intralitter correlation and litter size, appeared to enhance the fit of thedevelopmental toxicity–specific models, whereas inclusion of a threshold did not. Threshold in thiscase refers to a model-derived estimate of the point at which the model can no longer distinguish© 2006 by Taylor & Francis Group, LLC