PRINCIPLES OF TOXICOLOGY - Biology East Borneo

460 RISK ASSESSMENTleading to important differences in apparent dose–response relationships (i.e., those based strictly onapplied dose). One approach used to enhance extrapolation among species is PBPK modeling. UsingPBPK models, scientists are able to predict target organ doses of a chemical (or a critical metabolite,if that is important for toxicity) in test species and humans. With this information, corrections can bemade for pharmacokinetic differences among species, leading to better extrapolation of dose–responserelationships. The principal limitation of PBPK analyses is that they are very data-intensive, and PBPKmodels have been constructed and validated for only a few chemicals.Since PBPK models are rarely available, simpler approaches to extrapolating doses must be usedin most situations. One of the simplest approaches is to convey doses per unit body weight. Largeranimals (or humans) are assumed to require larger doses to produce the same toxic effect in directproportion to their body weight. This is the dose metric most commonly used when extrapolatinginformation on noncancer health effects among species. Interestingly, a different dose metric has beenused traditionally for extrapolating dose–carcinogenicity relationships. In this situation, doses haveusually been scaled according to the surface area of the animal. Since surface area for most species isa function of body weight, and can be approximated by body weight raised to the 0.67 power, thisfunction has been used to convert doses used in animal studies (e.g., rodent cancer bioassays) toequivalent human doses. In biology, empirical observations suggest that many biochemical andphysiological processes seem to scale among species according to surface area, while differences inothers seem to correspond more closely to changes in weight. The correct scaling for doses is notentirely obvious, and could conceivably be different for different chemical classes or toxicologicaleffects. Recently, there have been recommendations that scaling for carcinogen doses use a factorintermediate between body weight (or body weight raised to the power of 1) and surface area (bodyweight raised to the power of 0.67); that is, body weight raised to the power of 0.75.Does the choice of scaling factor really make a difference? To illustrate the answer, consider theextrapolation of dose information between a mouse and a human. If a dose for a noncancer effect in amouse were converted to a human dose based on surface area, rather than on body weight as iscustomary, the SHD dose would be reduced by a factor of 12–14. On the other hand, switching fromsurface area scaling to body weight scaling for carcinogenicity data would result in a 12–14-folddecrease in cancer risks estimated from the same dose–response information. The difference in use ofbody weight versus surface area for extrapolating between rats and humans is not as large—about6-fold—but still might be considered significant.18.5 RISK CHARACTERIZATIONThe purpose of the risk characterization step is to integrate information provided by the hazardidentification, dose–response assessment, and exposure assessment in order to develop risk estimates.Risk information may be conveyed in a qualitative manner, quantitative manner, or both. A qualitativeassessment may describe the hazard posed by chemicals of concern, discuss opportunities for exposure,and reach some general conclusions that the risks are likely to be high or low, but would not providenumerical estimates of risk. A quantitative risk characterization, on the other hand, includes numericalrisk values. Theoretically, there are many ways that numerical risks could be calculated depending onthe specific questions being addressed in the risk assessment. For example, risks could be expressedas an individual’s excess lifetime risk of developing a particular health effect as a result of chemicalexposure. Risks could also be expressed on a population basis (e.g., estimated number of extra casesof a disease per year attributable to chemical exposure), or as the relative risk of an exposed populationversus an unexposed population. Other ways of expressing risks, or health impacts, could be in termsof loss of life expectancy or lost days of work.As discussed in Section 18.4, the most common means of expressing cancer risk associated withchemical exposure is in the form of individual excess lifetime cancer risk. When calculated forregulatory purposes, these values are intended to represent upper-bound estimates. That is, a cancerrisk of one in one million means that an individual chosen at random from the exposed population is