The Refinement of Uncertainty/Safety Factors in Risk Assessment by the Incorporation of Data on Toxicokinetic Variability in Humans
http://www.100md.com
《毒物学科学杂志》
Division of Developmental Origins of Health and Disease, Institute of Human Nutrition, Clinical Pharmacology Group, School of Medicine, University of Southampton, Biomedical Sciences Building, Bassett Crescent East, Southampton, SO16 7PX, UK
ABSTRACT
The derivation of safe levels of exposure in humans for compounds that are assumed to cause threshold toxicity has relied on the application of a 100-fold uncertainty factor to a measure for the threshold, such as the no observed adverse effect level (NOAEL) or the benchmark dose (BMD). This 100-fold safety factor consists of the product of two 10-fold factors allowing for human variability and interspecies differences. The International Programme on Chemical Safety has suggested the subdivision of these 10-fold factors to allow for variability in toxicokinetics and toxicodynamics. This subdivision allows the replacement of the default uncertainty factors with a chemical-specific adjustment factor (CSAF) when suitable data are available. This short review describes potential options to refine safety factors used in risk assessment, with particular emphasis on pathway-related uncertainty factors associated with variability in kinetics. These pathway-related factors were derived from a database that quantified interspecies differences and human variability in phase I metabolism, phase II metabolism, and renal excretion. This approach allows metabolism and pharmacokinetic data in healthy adults and subgroups of the population to be incorporated in the risk-assessment process and constitutes an intermediate approach between simple default factors and chemical-specific adjustment factors.
Key Words: safety factors; human variability; interspecies differences; metabolism; phase I; phase II; toxicokinetics; uncertainty factors; risk assessment.
TRADITIONAL APPROACHES TO RISK ASSESSMENT: THRESHOLD AND NON-THRESHOLD EFFECTS
The multidisciplinary framework of risk assessment has provided a basis for identifying adverse health and environmental outcomes related to chemical exposure and for the hazard characterization of both threshold and non-threshold effects. Quantitative risk assessment using the dose–response relationship usually derived from experimental animal data combined with low-dose extrapolation has been used to quantify the risk for human health associated with estimated levels of exposure or, alternatively, to estimate the exposure that would result in a particular level of human risk for non-threshold effects (such as cancer arising from a genotoxic mechanism). In contrast, for threshold effects, safety assurance characterizes a dose or level of exposure below which no deleterious effect is measurable in animal or human studies and applies an uncertainty factor to estimate an exposure associated with negligible health risk. Both approaches can be divided into four steps: hazard identification, hazard characterization (which includes dose–response assessment), exposure assessment, and risk characterization (Barlow et al., 2002; Dybing et al., 2002; Edler et al., 2002; Kroes et al., 2002; Renwick et al., 2003; Smith, 2002). The scientific community is now moving toward the harmonization of both approaches (Di Marco et al., 1998). Central to this harmonization process is the distinction between threshold and non-threshold toxicity, which arises from the theoretical possibility that a single molecule (or an extremely low level of exposure) of a direct-acting genotoxic compound could bind irreversibly to DNA, cause a mutation, and increase the probability that the target cell(s) would become malignant (Slob, 1999; WHO, 1999). The absence of a measurable response in animal studies cannot be taken as proof of the presence of a threshold in the dose–response relationship, because neither threshold nor non-threshold effects would show a quantifiable or statistically significant response at very low levels of exposure (Slob and Pieters, 1998; Slob, 1999; Vermeire et al., 1999). Whichever approach is used, there is an increasing need to incorporate biologically and mechanistically based data, such as that derived by physiologically based pharmacokinetic (PB-PK) modeling, quantitatively, into the risk-assessment outcome.
Safe Levels of Exposure and Markers or Surrogates for the Threshold Intake
The derivation of levels of human exposure "without appreciable health risk" for thresholded toxicants has relied on dividing a surrogate for the threshold, such as the no observed adverse effect level (NOAEL) or the benchmark dose (BMD), by an uncertainty/safety factor of a 100-fold. The resulting level of intake is expressed in milligrams per kilogram of body weight per day (Crump, 1984; WHO, 1987). The nomenclature varies around the world and includes the minimal risk level (MRL), the reference dose (RfD), or reference concentration (RfC) in the USA, the acceptable daily intake (ADI) and the provisional tolerable weekly intake (PTWI) in Europe, the estimated-concentration-of-no-concern (ECNC) in the Netherlands, and the ADI, PTWI, and tolerable daily intake (TDI) by the WHO (Dourson et al., 1996; Truhaut, 1991; WHO, 1987, 1994). The NOAEL is usually derived from chronic or sub-chronic animal studies using the most sensitive relevant species (mouse, rat, rabbit, or dog) and a dose–response relationship that includes levels producing statistically significant adverse effects (WHO, 1987). The BMD concept was proposed as an alternative to the NOAEL by Crump (1984) and was defined as a lower statistical confidence limit (e.g., 95th centile) for the dose, corresponding to a predefined low level of increase in adverse effects over the background. The BMD can be calculated from continuous or quantal data and, unlike the NOAEL it takes into account the whole of the dose–response curve (Crump, 1984). A recent probabilistic extension of the benchmark dose concept has been proposed to generate a critical effect dose (CED) from a combination of experimental dose–response data and Monte Carlo analysis. A no-adverse-effect-level (NAEL) in animals can then be calculated from the uncertainty distribution (Slob, 1999; Vermeire et al., 1999).
The Uncertainty/Safety Factor Approach
The uncertainty/safety factor approach was introduced in the United States in the mid-1950s by the Food and Drug Administration (FDA), to define legislative guidelines for food additives and environmental contaminants. The original investigators, Lehman and Fitzhugh (1954), proposed the 100-fold default factor, which was applied to the dietary concentration. They reasoned that it allowed for several areas of uncertainty (Dourson et al., 1996; Lehman and Fitzhugh, 1954; Vermeire et al., 1999):
Interspecies variability allowing extrapolation from animal to man
Human variability taking into account sensitive individuals of the population
Possible synergistic effects arising between the many food additives and contaminants to which humans are exposed.
This scheme was adopted by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) and by the Joint FAO/WHO Expert on Pesticides Residues (JMPR) in 1961 to define the Acceptable Daily Intake (ADI) as "the daily intake of chemical which, during the entire life time, appears to be without appreciable risk on the basis of all known facts at the time" (Truhaut, 1991).
The 100-fold uncertainty/safety factor can be considered to represent the product of two separate 10-fold factors that allow for interspecies differences and human variability, because a 100-fold factor is usually applied to the NOAEL (in mg/kg body weight) from an animal study and a 10-fold factor to the NOAEL (in mg/kg body weight) from a human study (WHO, 1987). These 10-fold factors have to allow for both toxicokinetic and toxicodynamic differences, and they were subdivided to take into account the aspects separately (Renwick, 1993). Values of 100.6 (4) and a 100.4 (2.5), respectively, were proposed for species differences, and equal values of 100.5 (3.16) for human variability (WHO, 1994, 1999). The aim of this subdivision of the 10-fold factors was to allow the incorporation of suitable chemical-specific data for one particular aspect of uncertainty (e.g., toxicokinetic differences between animals and humans for a compound under assessment) to replace the relevant part of the overall default uncertainty factor. In recent years a number of reviews have been produced using therapeutic drug databases and analyzing human variability for both the kinetic and the dynamic aspects (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a 2004b, 2005; Ginsberg et al., 2002; Hattis et al., 1999; Naumann et al., 2001; Renwick and Lazarus, 1998; Silverman et al., 1999). Renwick and Lazarus (1998) evaluated data for therapeutic drugs metabolized and eliminated through a wide variety of pathways, and they concluded that the general 3.16 kinetic default factor could not cover human variability for polymorphic metabolism (such as CYP2D6) or the differences between healthy adults and neonates. The authors proposed a more flexible framework to replace default values with categorical factors such as pathway-related uncertainty factors. These could be used when detailed chemical-specific toxicokinetic or toxicodynamic data were not available, but there were data on the pathway of elimination or on the mode of action, and they constitute an intermediate option between default uncertainty factors and chemical-specific adjustment factors (CSAFs) (WHO, 2001) (Fig. 1). Development of a pathway-related factor for human variability requires the selection of a value from the population distribution that covers a particular proportion of the population (e.g., the 95th or 99th centiles). The percentile selected for any risk assessment may depend on the particular end point and the regulatory agency.
DEVELOPMENT OF PATHWAY-RELATED UNCERTAINTY FACTORS: THE METABOLISM AND PHARMACOKINETIC DATABASE
Pathway-related uncertainty factors have been developed using a meta-analysis of metabolism and pharmacokinetic data for probe substrates of phase I (CYP1A2, CYP2A6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, ADH, hydrolysis) and phase II (glucuronidation, glycine conjugation, N-acetylation (NAT), sulfate conjugation) metabolism and renal excretion (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a, 2004b; Walton et al., 2001a, 2001b, 2004). Both phase I and phase II metabolic pathways are known to be polymorphic; however, the in vivo functional consequences of many genetic polymorphisms have not yet been characterized, and in vivo pharmacokinetic data are not yet available. Therefore many elimination routes have been analyzed as monomorphic pathways (CYP1A2, CYP2A6, CYP2E1, CYP3A4, alcohol dehydrogenase, hydrolysis, glucuronidation, sulfate, or glycine conjugation), but the analyses will need to be reassessed if future pharmacokinetic data define a phenotypic difference. A number of polymorphic pathways were analyzed: CYP2C9, CYP2C19, CYP2D6 (phase I) and (N-acetyltranferases, NAT (phase II), for which the human data included non–phenotyped subjects (NP) for all CYP isoforms, phenotyped subjects (extensive and poor metabolisers—EMs and PMs for CYP2D6 and CYP2C19), and fast and slow acetylators (FAs/SAs for NAT).
Probe substrates were included in the analyses provided they were completely absorbed orally, and >60% of an oral dose was metabolized or eliminated via a single pathway of interest (see Dorne et al., 2001a). Meta-analyses of pharmacokinetic parameters were performed for all probe substrates/pathways and subgroups of the human population using markers of chronic exposure (oral clearances and AUCs) and acute exposure (Cmax) (data not shown). The pathway-related uncertainty factors were derived for each metabolic route to cover the 95th, 97.5th, and 99th centiles (which are presented here) of each subgroup of the human population (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a, 2004b; for reviews see, Dorne, 2004; Dorne et al., 2005). For subgroups of the human population, differences in internal dose compared to healthy adults were included in the calculation. These were based on the difference in geometric means combined with the subgroup specific variability to define the magnitude of the difference between the geometric mean for healthy adults and the value for different centiles of the subgroup.
Pathway-Related Uncertainty Factors Allowing for Interspecies Differences
Pathway-related factors for interspecies differences in kinetics could only be derived for some pathways, such as CYP1A2 (Walton et al., 2001a), glucuronidation (Walton et al., 2001b), and renal excretion (Walton et al., 2004), because for most compounds the pathway of elimination in humans did not predict the pathway in animals (Walton et al., 2001c). In consequence, there would be considerable uncertainty in using a species-specific pathway-related factor, unless there were detailed data for both humans and the animal species, in which case a compound-specific value would be derived.
Pathway-Related Uncertainty Factors for Human Variability in Healthy Adults
Pathway-related uncertainty factors were calculated for monomorphic pathways to cover the 99th centile of the healthy adult population for probe substrates of phase I (CYP1A2, CYP2A6, CYP2E1, CYP3A4, ADH, and hydrolyis), phase II (glucuronidation, glycine, and sulfate conjugation) metabolism and renal excretion. Values were all below the kinetic default uncertainty factor (3.16) with a range comprised between 1.6 and 2.2 for all pathways and up to 2.8 for CYP3A4 metabolism. The biological basis for the higher CYP3A4 variability (for the oral route) is related to the intestinal and hepatic expression of CYP3A4, possible CYP3A4 polymorphism and competition between CYP3A4 substrates and P-glycoprotein in the gastrointestinal tract (Dorne et al., 2003a).
In contrast, uncertainty factors for the 99th centile of PM and SA subjects of polymorphic pathways exceeded the 3.16 kinetic default factor and were 52 (CYP2C19), 26 (CYP2D6), and 5.2 (NAT) (assuming the parent compound was the proximate toxicant) (Dorne et al., 2002, 2003b). These values are based on oral in vivo data for compounds showing a range of clearance values; low clearance compounds would show large phenotypic differences because the difference in enzyme activity would affect both systemic clearance and bioavailability; for compounds showing high, blood-flow limited clearance, the phenotypic difference would derive largely from a difference in bioavailability (Gentry et al., 2002; Haber et al., 2002). The lack of data for the CYP2C9 polymorphism did not allow the inclusion of PMs such as individuals carrying CYP2C93/3, and for NP subjects, uncertainty factors were below the 3.16 kinetic default factor (2.2) (Dorne et al., 2004b). However, in vivo data for CYP2C9 PMs are emerging in the literature, and preliminary results show that they would require a higher degree of protection compared to CYP2C9 EMs. The variability observed within CYP2D6 EMs and the difference in internal dose between EMs and PMs depend on the presence of multiple copies of the gene and range from no copies in "strict" PMs and up to 13 copies in ultrarapid metabolizers (Dalen et al., 1998, 1999). This also applies to CYP2C19 metabolism (Wedlund, 2000) and N-acetylation (NAT2 gene), for which 8 and 14 alleles, respectively, have been found in Caucasians (Lin et al., 1993).
The uncertainty factors derived for each polymorphic pathway are "worst-case" values because they have been developed for major probe substrates for that pathway (60–100% of an oral dose in EMs or FAs). Data for minor CYP2D6 and CYP2C19 substrates (10–60% metabolism) have shown exponential relationships between the extent of metabolism in EMs/PMs and the pathway-related uncertainty factors (Dorne et al., 2002, 2003b). Overall, these relationships demonstrate that the general default toxicokinetic uncertainty factor (3.16) would cover PMs for substrates that were metabolized up to 30% by either enzyme (Dorne et al., 2002; 2003b). Such data for N-acetyltransferases and CYP2C9 substrates are not currently available.
Ethnic Differences, the Elderly, Children and Neonates
The uncertainty factors derived for each subgroup assume that subjects with lower ability to eliminate the compound will show higher susceptibility—i.e., that the parent compound is the active toxicant. Pathway-related uncertainty factors resulting from analyses of ethnic differences for mainly sub-Saharan African and Asian healthy adults raised the same concerns as those in general healthy adults, because the variability in CYP2D6, CYP2C19, or NAT would not be covered by the general kinetic default factor. An exponential relationship between the extent of CYP2C19 metabolism of different substrates in healthy Asian EMs/PMs and the corresponding pathway-related uncertainty factor has also been documented (Dorne et al., 2003b).
Both hepatic and renal function are reduced in the elderly as a result of normal aging processes (Durnas et al., 1990; Le Couteur and McLean, 1998), but pathway-related uncertainty factors derived for the elderly subpopulation were below the general 3.16 kinetic default factor, except for renal excretion (4.3) and CYP3A4 metabolism (4.6). Uncertainty factors (99th centile) for non-phenotyped elderly subjects were above the corresponding values in healthy adults and exceeded the general default values for CYP2D6 (8.4), CYP2C19 (4.3), and SAs (7.6). These data suggest that elderly PM subjects would be expected to have lower clearances than healthy adult PMs, but such data are not yet available in the literature (Dorne, 2002; 2003b).
Pathway-related factors (99th centiles) for children were below the 3.16 default factor except for the polymorphic CYP2D6 and CYP2C19 isoforms (45 and 9, respectively). Uncertainty factors (99th centile) for neonates could be derived for only five elimination pathways: CYP1A2 (14), CYP3A4 and glucuronidation (12), glycine conjugation (28), and renal excretion (3.7). These values reflect the immaturity of the neonatal liver and kidney in terms of xenobiotic metabolism, which has been the subject of a number of reviews (Besunder et al., 1988; Cresteil, 1998; Gow et al., 2001; Ginsberg et al., 2002; Renwick et al., 2000). Pharmacokinetic data for polymorphic pathways in neonates were available in only two subjects for CYP2D6 and were associated with a 19-fold and 33-fold lower clearance (Ito et al., 1998). Overall, it is probable that neonates would be the most susceptible subgroup when exposed to compounds handled by CYP2D6 and CYP2C19 metabolism.
OPTIONS FOR REFINEMENTS OF SAFETY FACTORS
Quantitative data describing interspecies differences and human variability in toxicokinetics and toxicodynamics are essential if general default uncertainty factors are to be replaced with chemical-specific data, but such data are available only very rarely. This discussion has highlighted the recent development of pathway-related uncertainty factors, allowing for interspecies and human variability in kinetics, with particular emphasis on human variability, because the uncertainty factors for species differences would not be generally applicable to most pathways. This is mostly due to qualitative and quantitative interspecies differences in metabolism, questioning the relevance to humans (Walton et al., 2001a, 2001b, 2001c, 2004).
Pathway-related uncertainty factors provide an intermediate approach between the general kinetic default factor (3.16) and chemical-specific adjustment factors (CSAF) for thresholded toxicants, giving more flexible options to risk assessors depending on the level of knowledge available for a particular compound. They could replace the default kinetic factor for a particular compound under evaluation if its metabolic fate is known in humans. Identification of the specific enzyme isoform responsible for the metabolism of a xenobiotic is becoming a routine procedure using liver microsomes, heterologously expressed enzymes, and/or enzyme inhibitors (Venkatakrishnan et al., 2001). In vivo metabolism data may also be available in healthy adults using tracer doses of the compound. Overall, pathway-related uncertainty factors were found to be below the general kinetic default factor of 3.16 for most monomorphic pathways in healthy adults and most subgroups of the population, with the exception CYP3A4 and renal excretion in the elderly, and all pathways analyzed in neonates (CYP1A2, CYP3A4, glucuronidation, glycine conjugation, and renal excretion). Polymorphic pathways (CYP2D6, CYP2C19, and NATs) showed high variability and are of concern because the general kinetic default value would not be adequately protective for any subgroup of the population. An important aspect of genetic polymorphism is to identify the toxicological consequences of metabolism because subjects with impaired metabolism (PMs/SAs) and fast metabolism (EMs and FAs) could be the susceptible subgroup if the toxic species were the parent compound or the metabolite, respectively.
A crucial issue for the risk assessment of subgroups of the population is whether the uncertainty factor (whether a CSAF or a pathway-related factor) should cover a percentile of the subgroup or of the combined total population. This concept applies particularly to polymorphic pathways, because the 99th centile of a PM subgroup that was only 5% of the total population would be the 99.95th centile of the total population. In contrast, although neonates may be only a small percentage of the total population at any point in time, all individuals have been part of that subgroup at some time in their lives! An additional problem is that the frequencies of polymorphism differ between different ethnic groups of the human population. The frequency of CYP2D6 PMs in Caucasians is much higher than that in Asian populations (8% versus 1%), whereas the opposite applies to the CYP2C19 and NAT pathways (2.5% versus 15% and 40–70% versus 10–20%) (Wedlund, 2000). Given this ethnic diversity, different uncertainty factors could be developed if the value were chosen to cover a particular percentage of the total population.
Finally, neonates show the lowest enzyme activity of the whole human population, although reliable data are not available for polymorphic metabolism in this subgroup. The low activity could translate into a greater susceptibility if the parent compound were the active toxicant, but less susceptibility if the compound underwent metabolic activation. In the absence of data on the activity of the relevant pathway(s) of elimination in neonates and the consequences of metabolism (i.e., detoxication or activation), an extra uncertainty factor higher than that in adults for polymorphic metabolism (CYP2D6, CYP2C19, NAT) may be an option to be considered by risk assessors and risk managers (Dorne et al., 2005).
A limitation to the widespread application of pathway-related factors in risk assessment is that most contaminants are eliminated via multiple metabolic and excretion routes. Recently, a meta-analysis of human variability in kinetics for such compounds was carried out, and Monte Carlo modeled predictions of relevant uncertainty factors based on metabolism data and pathway-related variability were validated using published chemical-specific data for the modeled compounds (Dorne and Renwick, 2003). Probabilistic approaches describing both uncertainty and variability represent a potential future improvement by the replacement of uncertainty factors that are derived as single point estimates with a distribution that provides information about the extent of uncertainty in the risk assessment (Dorne and Renwick, 2003; Edler et al., 2002; Slob and Pieters, 1998; Swartout et al., 1998; WHO, 2001).
CONCLUSIONS
The "OMIC" sciences (genomics, proteomics, metabolomics, metabonomics) have opened up considerable opportunities to refine uncertainty/safety factors with chemical-specific data based on our mechanistic understanding at different levels (populations, individuals, cells, and molecular level). Toxicology and risk assessment have benefited enormously by our growing understanding of interspecies differences and inter-individual variability in toxicokinetics and toxicodynamics. These fundamental aspects provide the basis for the refinement of the traditional safety factor approach. This paper has described the development of pathway-related uncertainty factors that provide risk assessors a new intermediate option between default uncertainty factors and chemical-specific adjustment factors (CSAFs). Pathway-related factors have been developed for the kinetic variability, and further work would be required to investigate human variability in toxicodynamics to develop uncertainty factors based on mechanistic data or mode of action, although in reality these type of data are not available or scant (Hashemi, Walton, and Renwick, unpublished report). A pooled analysis of pharmacodynamic variability for different drug effects has been performed (Renwick and Lazarus, 1998), and inter-individual differences were similar to the variability in kinetics. Meta-analyses of a number of different pharmacodynamic end points for both the human variability and interspecies aspects would provide useful tools to replace default safety factors for the dynamic aspect and to design new probabilistic models (Edler et al., 2002; Slob and Pieters, 1998, Swartout et al., 1998; Vermeire et al., 1999; WHO, 2001).
REFERENCES
Barlow, S. M., Greig, J. B., Bridges, J. W., Carere, A., Carpy, A. J., Galli, C. L., Kleiner, J., Knudsen, I., Koeter, H. B., Levy, L. S., et al. (2002). Hazard identification by methods of animal-based toxicology. Food Chem. Toxicol. 40, 145–191.
Besunder, J. B., Reed, M. D., and Blumer, J. L. (1988). Principles of drug biodisposition in the neonate. A critical evaluation of the pharmacokinetic-pharmacodynamic interface (Part I). Clin. Pharmacokinet. 14, 189–216.
Cresteil, T. (1998). Onset of xenobiotic metabolism in children: toxicological implications. Food Addit. Contam. 15(Suppl), 45–51.
Crump, K. S. (1984). A new method for determining allowable daily intakes. Fundam. Appl. Toxicol. 4, 854–871.
Dalen, P., Dahl, M. L., Ruiz, M. L., Nordin, J., and Bertilsson, L. (1998). 10-Hydroxylation of nortriptyline in white persons with 0, 1, 2, 3, and 13 functional CYP2D6 genes. Clin. Pharmacol. Ther. 63, 444–452.
Dalen, P., Dahl, M. L., Eichelbaum, M., Bertilsson, L., and Wilkinson, G. R. (1999). Disposition of debrisoquine in Caucasians with different CYP2D6-genotypes including those with multiple genes. Pharmacogenetics 9, 697–706.
Di Marco, P. N., Priestly, B. G., and Buckett, K. J. (1998). Carcinogen risk assessment. Can we harmonise Toxicol. Lett. 102–103, 241–246.
Dorne, J. L. C. M. (2004). Impact of inter-individual differences in drug metabolism and pharmacokinetics on safety evaluation. Fundam. Clin. Pharmacol. 18, 609–620.
Dorne, J. L., Walton, K., and Renwick, A. G. (2001a). Human variability in glucuronidation in relation to uncertainty factors for risk assessment. Food Chem. Toxicol. 39, 1153–1173.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2001b). Uncertainty factors for chemical risk assessment. Human variability in the pharmacokinetics of CYP1A2 probe substrates. Food Chem. Toxicol. 39, 681–696.
Dorne, J. L., Walton, K., Slob, W., and Renwick, A. G. (2002). Human variability in polymorphic CYP2D6 metabolism: Is the kinetic default uncertainty factor adequate Food Chem. Toxicol. 40, 1633–1656.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2003a). Human variability in CYP3A4 metabolism and CYP3A4-related uncertainty factors for risk assessment. Food Chem. Toxicol. 41, 201–224.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2003b). Polymorphic CYP2C19 and N-acetylation: Human variability in kinetics and pathway-related uncertainty factors. Food Chem. Toxicol. 41, 225–245.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2004a). Human variability in the renal elimination of foreign compounds and renal excretion-related uncertainty factors for risk assessment. Food Chem. Toxicol. 42, 275–298.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2004b). Human variability for metabolic pathways with limited data (CYP2A6, CYP2C9, CYP2E1, ADH, esterases, glycine and sulphate conjugation). Food Chem. Toxicol. 42, 397–421.
Dorne, J. L., Walton, K., and Renwick, A. G. (2005). Human variability in xenobiotic metabolism and pathway-related uncertainty factors for chemical risk assessment: A review. Food Chem. Toxicol. 43, 203–416.
Dorne, J. L. C. M., and Renwick, A. G., 2003. Prediction of human variability using kinetic data and Monte Carlo modelling for the derivation of pathway-related uncertainty factors for compounds handled by multiple pathways. Toxicol. Lett. 144, S195.
Dourson, M. L., Felter, S. P., and Robinson, D. (1996). Evolution of science-based uncertainty factors in noncancer risk assessment. Regul. Toxicol. Pharmacol. 24, 108–120.
Durnas, C., Loi, C. M., and Cusack, B. J. (1990). Hepatic drug metabolism and aging. Clin. Pharmacokinet. 19, 359–389.
Dybing, E., Doe, J., Groten, J., Kleiner, J., O'Brien, J., Renwick, A. G., Schlatter, J., Steinberg, P., Tritscher, A., Walker, R. et al. (2002). Hazard characterisation of chemicals in food and diet. Dose response, mechanisms and extrapolation issues. Food Chem. Toxicol. 40, 237–282.
Edler, L., Poirier, K., Dourson, M., Kleiner, J., Mileson, B., Nordmann, H., Renwick, A., Slob, W., Walton, K., and Wurtzen, G. (2002). Mathematical modelling and quantitative methods. Food Chem. Toxicol. 40, 283–326.
Gentry, P. R., Hack, E., Haber, L. T., Maier, A., and Clewell, H. J. (2002). An approach for the consideration of genetic polymorphism data in chemical risk assessment: Examples with warfarin and parathion. Toxicol. Sci. 70, 120–139.
Ginsberg, G., Hattis, D., Sonawane, B., Russ, A., Banati, P., Kozlak, M., Smolenski, S., and Goble, R. (2002). Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature. Toxicol. Sci. 66, 185–200.
Gow, P. J., Ghabrial, H., Smallwood, R. A., Morgan, D. J., and Ching, M. S. (2001). Neonatal hepatic drug elimination. Pharmacol. Toxicol. 88, 3–15.
Haber, L. T., Maier, A., Gentry, P. R., Clewell, H. J., and Dourson, M. L. (2002). Genetic polymorphisms in assessing inter-individual variability in delivered dose. Reg. Toxicol. Pharmacol. 35, 177–197.
Hattis, D., Banati, P., Goble, R., and Burmaster, D. E. (1999). Human interindividual variability in parameters related to health risks. Risk Anal. 19, 711–726.
Ito, S., Gow, R., Verjee, Z., Giesbrecht, E., Dodo, H., Freedom, R., Tonn, G. R., Axelson, J. E., Zalzstein, E., Rosenberg, H. C., andKoren, G. et al. (1998). Intravenous and oral propafenone for treatment of tachycardia in infants and children: Pharmacokinetics and clinical response. J. Clin. Pharmacol. 38, 496–501.
Kroes, R., Muller, D., Lambe, J., Lowik, M. R., van Klaveren, J., Kleiner, J., Massey, R., Mayer, S., Urieta, I., Verger, P. et al. (2002). Assessment of intake from the diet. Food Chem. Toxicol. 40, 283–326.
Le Couteur, D. G., and McLean, A. J. (1998). The aging liver. Drug clearance and an oxygen diffusion barrier hypothesis. Clin. Pharmacokinet. 34, 359–373.
Lehman, A. J., and Fitzhugh, O. G. (1954). 100-fold margin of safety. Assoc. Food Drug Off. U.S.Q. Bull. 18, 33–35.
Lin, H. J., Han, C. Y., Lin, B. K., and Hardy, S. (1993). Slow acetylator mutations in the human polymorphic N-acetyltransferase gene in 786 Asians, blacks, Hispanics, and whites: Application to metabolic epidemiology. Am. J. Hum. Genet. 52, 827–834.
Naumann, B. D., Silverman, K. C., Dixit, R., Faria, E. C., and Sargent, E. V. (2001). Case studies of categorical data-derived adjustment factors. Hum. Ecol. Risk Assess. 7, 61–105.
Renwick, A. G. (1993). Data-derived safety factors for the evaluation of food additives and environmental contaminants. Food Addit. Contam. 10, 275–305.
Renwick, A. G., and Lazarus, N. R. (1998). Human variability and noncancer risk assessment—An analysis of the default uncertainty factor. Regul. Toxicol. Pharmacol. 27, 3–20.
Renwick, A. G., Barlow, S. M., Hertz-Picciotto, I., Boobis, A. R., Dybing, E., Edler, L., Eisenbrand, G., Greig, J. B., Kleiner, J., Lambe, J. et al. (2003). Risk characterisation of chemicals in food and diet. Food Chem. Toxicol. 41, 1211–1271.
Renwick, A. G., Dorne, J. L., and Walton, K. (2000). An analysis of the need for an additional uncertainty factor for infants and children. Regul. Toxicol. Pharmacol. 31, 286–296.
Silverman, K. C., Naumann, B. D., Holder, D. J., Dixit, R., Faria, E. C., Sargent, E. V., and Gallo, M. A. (1999). Establishing data-derived uncertainty factors from published pharmaceutical clinical trial data. Hum. Ecol. Risk Assess. 5, 1059–1090.
Slob, W., and Pieters, M. N. (1998). A probabilistic approach for deriving acceptable human intake limits and human health risks from toxicological studies: General framework. Risk Anal. 18, 787–798.
Slob, W. (1999). Thresholds in toxicology and risk assessment. Int. J. Toxicol. 18, 259–268.
Smith, M., 2002. Food Safety in Europe (FOSIE): Risk assessment of chemicals in food and diet: Overall introduction. Food Chem. Toxicol. 40, 141–144.
Swartout, J. C., Price, P. S., Dourson, M. L., Carlson-Lynch, H. L., and Keenan, R. E. (1998). A probabilistic framework for the reference dose (probabilistic RfD). Risk Anal. 18, 271–282.
Truhaut, R., 1991. The concept of the acceptable daily intake: An historical review. Food Addit. Contam. 8, 151–162.
Venkatakrishnan, K., von Moltke, L. L., and Greenblatt, D. J. (2001). Human drug metabolism and the cytochromes P450: Application and relevance of in vitro models. J. Clin. Pharmacol. 41, 1149–1179.
Vermeire, T., Stevenson, H., Peiters, M. N., Rennen, M., Slob, W., and Hakkert, B. C. (1999). Assessment factors for human health risk assessment: A discussion paper. Crit. Rev. Toxicol. 29, 439–490.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001a). Uncertainty factors for chemical risk assessment: Interspecies differences in the in vivo pharmacokinetics and metabolism of human CYP1A2 substrates. Food Chem. Toxicol. 39, 667–680.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001b). Uncertainty factors for chemical risk assessment: Interspecies differences in glucuronidation. Food Chem. Toxicol. 39, 1175–1190.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001c). Categorical default factors for interspecies differences in the major routes of xenobiotic elimination. Hum. Ecol. Risk Assess. 7, 181–201.
Walton, K., Dorne, J. L., and Renwick, A. G. (2004). Species-specific uncertainty factors for compounds eliminated principally by renal excretion in humans. Food Chem. Toxicol. 42, 261–274.
Wedlund, P. J., 2000. The CYP2C19 enzyme polymorphism. Pharmacology 61, 174–183.
WHO (1987). Principles for the safety assessment of food additives and contaminants in food. Environmenal Health Criteria. 70, 174 pp., International Programme on Chemical Safety, World Health Organisation, Geneva.
WHO (1994). International Programme on Chemical Safety: Assessing human health risks of chemicals: Derivation of Guidance values for health-based exposure limits. Environmenal Health Criteria, 170, 73 pp., International Programme on Chemical Safety, World Health Organisation, Geneva.
WHO (1999). International Programme on Chemical Safety: Assessing human health risks of chemicals: Principles for the assessment of risk to human health from exposure to chemicals. Environmental Health Criteria 210, World Health Organisation, Geneva.
WHO (2001). International Programme on Chemical Safety: Guidance document for the use of chemical-specific adjustment factors (CSAFs) for interspecies differences and human variability in dose-concentration response assessment. 76 pp., World Health Organisation, Geneva. http://www.ipcsharmonize.org/CSAFsummary.htm.(J. L. C. M. Dorne and A. )
ABSTRACT
The derivation of safe levels of exposure in humans for compounds that are assumed to cause threshold toxicity has relied on the application of a 100-fold uncertainty factor to a measure for the threshold, such as the no observed adverse effect level (NOAEL) or the benchmark dose (BMD). This 100-fold safety factor consists of the product of two 10-fold factors allowing for human variability and interspecies differences. The International Programme on Chemical Safety has suggested the subdivision of these 10-fold factors to allow for variability in toxicokinetics and toxicodynamics. This subdivision allows the replacement of the default uncertainty factors with a chemical-specific adjustment factor (CSAF) when suitable data are available. This short review describes potential options to refine safety factors used in risk assessment, with particular emphasis on pathway-related uncertainty factors associated with variability in kinetics. These pathway-related factors were derived from a database that quantified interspecies differences and human variability in phase I metabolism, phase II metabolism, and renal excretion. This approach allows metabolism and pharmacokinetic data in healthy adults and subgroups of the population to be incorporated in the risk-assessment process and constitutes an intermediate approach between simple default factors and chemical-specific adjustment factors.
Key Words: safety factors; human variability; interspecies differences; metabolism; phase I; phase II; toxicokinetics; uncertainty factors; risk assessment.
TRADITIONAL APPROACHES TO RISK ASSESSMENT: THRESHOLD AND NON-THRESHOLD EFFECTS
The multidisciplinary framework of risk assessment has provided a basis for identifying adverse health and environmental outcomes related to chemical exposure and for the hazard characterization of both threshold and non-threshold effects. Quantitative risk assessment using the dose–response relationship usually derived from experimental animal data combined with low-dose extrapolation has been used to quantify the risk for human health associated with estimated levels of exposure or, alternatively, to estimate the exposure that would result in a particular level of human risk for non-threshold effects (such as cancer arising from a genotoxic mechanism). In contrast, for threshold effects, safety assurance characterizes a dose or level of exposure below which no deleterious effect is measurable in animal or human studies and applies an uncertainty factor to estimate an exposure associated with negligible health risk. Both approaches can be divided into four steps: hazard identification, hazard characterization (which includes dose–response assessment), exposure assessment, and risk characterization (Barlow et al., 2002; Dybing et al., 2002; Edler et al., 2002; Kroes et al., 2002; Renwick et al., 2003; Smith, 2002). The scientific community is now moving toward the harmonization of both approaches (Di Marco et al., 1998). Central to this harmonization process is the distinction between threshold and non-threshold toxicity, which arises from the theoretical possibility that a single molecule (or an extremely low level of exposure) of a direct-acting genotoxic compound could bind irreversibly to DNA, cause a mutation, and increase the probability that the target cell(s) would become malignant (Slob, 1999; WHO, 1999). The absence of a measurable response in animal studies cannot be taken as proof of the presence of a threshold in the dose–response relationship, because neither threshold nor non-threshold effects would show a quantifiable or statistically significant response at very low levels of exposure (Slob and Pieters, 1998; Slob, 1999; Vermeire et al., 1999). Whichever approach is used, there is an increasing need to incorporate biologically and mechanistically based data, such as that derived by physiologically based pharmacokinetic (PB-PK) modeling, quantitatively, into the risk-assessment outcome.
Safe Levels of Exposure and Markers or Surrogates for the Threshold Intake
The derivation of levels of human exposure "without appreciable health risk" for thresholded toxicants has relied on dividing a surrogate for the threshold, such as the no observed adverse effect level (NOAEL) or the benchmark dose (BMD), by an uncertainty/safety factor of a 100-fold. The resulting level of intake is expressed in milligrams per kilogram of body weight per day (Crump, 1984; WHO, 1987). The nomenclature varies around the world and includes the minimal risk level (MRL), the reference dose (RfD), or reference concentration (RfC) in the USA, the acceptable daily intake (ADI) and the provisional tolerable weekly intake (PTWI) in Europe, the estimated-concentration-of-no-concern (ECNC) in the Netherlands, and the ADI, PTWI, and tolerable daily intake (TDI) by the WHO (Dourson et al., 1996; Truhaut, 1991; WHO, 1987, 1994). The NOAEL is usually derived from chronic or sub-chronic animal studies using the most sensitive relevant species (mouse, rat, rabbit, or dog) and a dose–response relationship that includes levels producing statistically significant adverse effects (WHO, 1987). The BMD concept was proposed as an alternative to the NOAEL by Crump (1984) and was defined as a lower statistical confidence limit (e.g., 95th centile) for the dose, corresponding to a predefined low level of increase in adverse effects over the background. The BMD can be calculated from continuous or quantal data and, unlike the NOAEL it takes into account the whole of the dose–response curve (Crump, 1984). A recent probabilistic extension of the benchmark dose concept has been proposed to generate a critical effect dose (CED) from a combination of experimental dose–response data and Monte Carlo analysis. A no-adverse-effect-level (NAEL) in animals can then be calculated from the uncertainty distribution (Slob, 1999; Vermeire et al., 1999).
The Uncertainty/Safety Factor Approach
The uncertainty/safety factor approach was introduced in the United States in the mid-1950s by the Food and Drug Administration (FDA), to define legislative guidelines for food additives and environmental contaminants. The original investigators, Lehman and Fitzhugh (1954), proposed the 100-fold default factor, which was applied to the dietary concentration. They reasoned that it allowed for several areas of uncertainty (Dourson et al., 1996; Lehman and Fitzhugh, 1954; Vermeire et al., 1999):
Interspecies variability allowing extrapolation from animal to man
Human variability taking into account sensitive individuals of the population
Possible synergistic effects arising between the many food additives and contaminants to which humans are exposed.
This scheme was adopted by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) and by the Joint FAO/WHO Expert on Pesticides Residues (JMPR) in 1961 to define the Acceptable Daily Intake (ADI) as "the daily intake of chemical which, during the entire life time, appears to be without appreciable risk on the basis of all known facts at the time" (Truhaut, 1991).
The 100-fold uncertainty/safety factor can be considered to represent the product of two separate 10-fold factors that allow for interspecies differences and human variability, because a 100-fold factor is usually applied to the NOAEL (in mg/kg body weight) from an animal study and a 10-fold factor to the NOAEL (in mg/kg body weight) from a human study (WHO, 1987). These 10-fold factors have to allow for both toxicokinetic and toxicodynamic differences, and they were subdivided to take into account the aspects separately (Renwick, 1993). Values of 100.6 (4) and a 100.4 (2.5), respectively, were proposed for species differences, and equal values of 100.5 (3.16) for human variability (WHO, 1994, 1999). The aim of this subdivision of the 10-fold factors was to allow the incorporation of suitable chemical-specific data for one particular aspect of uncertainty (e.g., toxicokinetic differences between animals and humans for a compound under assessment) to replace the relevant part of the overall default uncertainty factor. In recent years a number of reviews have been produced using therapeutic drug databases and analyzing human variability for both the kinetic and the dynamic aspects (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a 2004b, 2005; Ginsberg et al., 2002; Hattis et al., 1999; Naumann et al., 2001; Renwick and Lazarus, 1998; Silverman et al., 1999). Renwick and Lazarus (1998) evaluated data for therapeutic drugs metabolized and eliminated through a wide variety of pathways, and they concluded that the general 3.16 kinetic default factor could not cover human variability for polymorphic metabolism (such as CYP2D6) or the differences between healthy adults and neonates. The authors proposed a more flexible framework to replace default values with categorical factors such as pathway-related uncertainty factors. These could be used when detailed chemical-specific toxicokinetic or toxicodynamic data were not available, but there were data on the pathway of elimination or on the mode of action, and they constitute an intermediate option between default uncertainty factors and chemical-specific adjustment factors (CSAFs) (WHO, 2001) (Fig. 1). Development of a pathway-related factor for human variability requires the selection of a value from the population distribution that covers a particular proportion of the population (e.g., the 95th or 99th centiles). The percentile selected for any risk assessment may depend on the particular end point and the regulatory agency.
DEVELOPMENT OF PATHWAY-RELATED UNCERTAINTY FACTORS: THE METABOLISM AND PHARMACOKINETIC DATABASE
Pathway-related uncertainty factors have been developed using a meta-analysis of metabolism and pharmacokinetic data for probe substrates of phase I (CYP1A2, CYP2A6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, ADH, hydrolysis) and phase II (glucuronidation, glycine conjugation, N-acetylation (NAT), sulfate conjugation) metabolism and renal excretion (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a, 2004b; Walton et al., 2001a, 2001b, 2004). Both phase I and phase II metabolic pathways are known to be polymorphic; however, the in vivo functional consequences of many genetic polymorphisms have not yet been characterized, and in vivo pharmacokinetic data are not yet available. Therefore many elimination routes have been analyzed as monomorphic pathways (CYP1A2, CYP2A6, CYP2E1, CYP3A4, alcohol dehydrogenase, hydrolysis, glucuronidation, sulfate, or glycine conjugation), but the analyses will need to be reassessed if future pharmacokinetic data define a phenotypic difference. A number of polymorphic pathways were analyzed: CYP2C9, CYP2C19, CYP2D6 (phase I) and (N-acetyltranferases, NAT (phase II), for which the human data included non–phenotyped subjects (NP) for all CYP isoforms, phenotyped subjects (extensive and poor metabolisers—EMs and PMs for CYP2D6 and CYP2C19), and fast and slow acetylators (FAs/SAs for NAT).
Probe substrates were included in the analyses provided they were completely absorbed orally, and >60% of an oral dose was metabolized or eliminated via a single pathway of interest (see Dorne et al., 2001a). Meta-analyses of pharmacokinetic parameters were performed for all probe substrates/pathways and subgroups of the human population using markers of chronic exposure (oral clearances and AUCs) and acute exposure (Cmax) (data not shown). The pathway-related uncertainty factors were derived for each metabolic route to cover the 95th, 97.5th, and 99th centiles (which are presented here) of each subgroup of the human population (Dorne et al., 2001a, 2001b, 2002, 2003a, 2003b, 2004a, 2004b; for reviews see, Dorne, 2004; Dorne et al., 2005). For subgroups of the human population, differences in internal dose compared to healthy adults were included in the calculation. These were based on the difference in geometric means combined with the subgroup specific variability to define the magnitude of the difference between the geometric mean for healthy adults and the value for different centiles of the subgroup.
Pathway-Related Uncertainty Factors Allowing for Interspecies Differences
Pathway-related factors for interspecies differences in kinetics could only be derived for some pathways, such as CYP1A2 (Walton et al., 2001a), glucuronidation (Walton et al., 2001b), and renal excretion (Walton et al., 2004), because for most compounds the pathway of elimination in humans did not predict the pathway in animals (Walton et al., 2001c). In consequence, there would be considerable uncertainty in using a species-specific pathway-related factor, unless there were detailed data for both humans and the animal species, in which case a compound-specific value would be derived.
Pathway-Related Uncertainty Factors for Human Variability in Healthy Adults
Pathway-related uncertainty factors were calculated for monomorphic pathways to cover the 99th centile of the healthy adult population for probe substrates of phase I (CYP1A2, CYP2A6, CYP2E1, CYP3A4, ADH, and hydrolyis), phase II (glucuronidation, glycine, and sulfate conjugation) metabolism and renal excretion. Values were all below the kinetic default uncertainty factor (3.16) with a range comprised between 1.6 and 2.2 for all pathways and up to 2.8 for CYP3A4 metabolism. The biological basis for the higher CYP3A4 variability (for the oral route) is related to the intestinal and hepatic expression of CYP3A4, possible CYP3A4 polymorphism and competition between CYP3A4 substrates and P-glycoprotein in the gastrointestinal tract (Dorne et al., 2003a).
In contrast, uncertainty factors for the 99th centile of PM and SA subjects of polymorphic pathways exceeded the 3.16 kinetic default factor and were 52 (CYP2C19), 26 (CYP2D6), and 5.2 (NAT) (assuming the parent compound was the proximate toxicant) (Dorne et al., 2002, 2003b). These values are based on oral in vivo data for compounds showing a range of clearance values; low clearance compounds would show large phenotypic differences because the difference in enzyme activity would affect both systemic clearance and bioavailability; for compounds showing high, blood-flow limited clearance, the phenotypic difference would derive largely from a difference in bioavailability (Gentry et al., 2002; Haber et al., 2002). The lack of data for the CYP2C9 polymorphism did not allow the inclusion of PMs such as individuals carrying CYP2C93/3, and for NP subjects, uncertainty factors were below the 3.16 kinetic default factor (2.2) (Dorne et al., 2004b). However, in vivo data for CYP2C9 PMs are emerging in the literature, and preliminary results show that they would require a higher degree of protection compared to CYP2C9 EMs. The variability observed within CYP2D6 EMs and the difference in internal dose between EMs and PMs depend on the presence of multiple copies of the gene and range from no copies in "strict" PMs and up to 13 copies in ultrarapid metabolizers (Dalen et al., 1998, 1999). This also applies to CYP2C19 metabolism (Wedlund, 2000) and N-acetylation (NAT2 gene), for which 8 and 14 alleles, respectively, have been found in Caucasians (Lin et al., 1993).
The uncertainty factors derived for each polymorphic pathway are "worst-case" values because they have been developed for major probe substrates for that pathway (60–100% of an oral dose in EMs or FAs). Data for minor CYP2D6 and CYP2C19 substrates (10–60% metabolism) have shown exponential relationships between the extent of metabolism in EMs/PMs and the pathway-related uncertainty factors (Dorne et al., 2002, 2003b). Overall, these relationships demonstrate that the general default toxicokinetic uncertainty factor (3.16) would cover PMs for substrates that were metabolized up to 30% by either enzyme (Dorne et al., 2002; 2003b). Such data for N-acetyltransferases and CYP2C9 substrates are not currently available.
Ethnic Differences, the Elderly, Children and Neonates
The uncertainty factors derived for each subgroup assume that subjects with lower ability to eliminate the compound will show higher susceptibility—i.e., that the parent compound is the active toxicant. Pathway-related uncertainty factors resulting from analyses of ethnic differences for mainly sub-Saharan African and Asian healthy adults raised the same concerns as those in general healthy adults, because the variability in CYP2D6, CYP2C19, or NAT would not be covered by the general kinetic default factor. An exponential relationship between the extent of CYP2C19 metabolism of different substrates in healthy Asian EMs/PMs and the corresponding pathway-related uncertainty factor has also been documented (Dorne et al., 2003b).
Both hepatic and renal function are reduced in the elderly as a result of normal aging processes (Durnas et al., 1990; Le Couteur and McLean, 1998), but pathway-related uncertainty factors derived for the elderly subpopulation were below the general 3.16 kinetic default factor, except for renal excretion (4.3) and CYP3A4 metabolism (4.6). Uncertainty factors (99th centile) for non-phenotyped elderly subjects were above the corresponding values in healthy adults and exceeded the general default values for CYP2D6 (8.4), CYP2C19 (4.3), and SAs (7.6). These data suggest that elderly PM subjects would be expected to have lower clearances than healthy adult PMs, but such data are not yet available in the literature (Dorne, 2002; 2003b).
Pathway-related factors (99th centiles) for children were below the 3.16 default factor except for the polymorphic CYP2D6 and CYP2C19 isoforms (45 and 9, respectively). Uncertainty factors (99th centile) for neonates could be derived for only five elimination pathways: CYP1A2 (14), CYP3A4 and glucuronidation (12), glycine conjugation (28), and renal excretion (3.7). These values reflect the immaturity of the neonatal liver and kidney in terms of xenobiotic metabolism, which has been the subject of a number of reviews (Besunder et al., 1988; Cresteil, 1998; Gow et al., 2001; Ginsberg et al., 2002; Renwick et al., 2000). Pharmacokinetic data for polymorphic pathways in neonates were available in only two subjects for CYP2D6 and were associated with a 19-fold and 33-fold lower clearance (Ito et al., 1998). Overall, it is probable that neonates would be the most susceptible subgroup when exposed to compounds handled by CYP2D6 and CYP2C19 metabolism.
OPTIONS FOR REFINEMENTS OF SAFETY FACTORS
Quantitative data describing interspecies differences and human variability in toxicokinetics and toxicodynamics are essential if general default uncertainty factors are to be replaced with chemical-specific data, but such data are available only very rarely. This discussion has highlighted the recent development of pathway-related uncertainty factors, allowing for interspecies and human variability in kinetics, with particular emphasis on human variability, because the uncertainty factors for species differences would not be generally applicable to most pathways. This is mostly due to qualitative and quantitative interspecies differences in metabolism, questioning the relevance to humans (Walton et al., 2001a, 2001b, 2001c, 2004).
Pathway-related uncertainty factors provide an intermediate approach between the general kinetic default factor (3.16) and chemical-specific adjustment factors (CSAF) for thresholded toxicants, giving more flexible options to risk assessors depending on the level of knowledge available for a particular compound. They could replace the default kinetic factor for a particular compound under evaluation if its metabolic fate is known in humans. Identification of the specific enzyme isoform responsible for the metabolism of a xenobiotic is becoming a routine procedure using liver microsomes, heterologously expressed enzymes, and/or enzyme inhibitors (Venkatakrishnan et al., 2001). In vivo metabolism data may also be available in healthy adults using tracer doses of the compound. Overall, pathway-related uncertainty factors were found to be below the general kinetic default factor of 3.16 for most monomorphic pathways in healthy adults and most subgroups of the population, with the exception CYP3A4 and renal excretion in the elderly, and all pathways analyzed in neonates (CYP1A2, CYP3A4, glucuronidation, glycine conjugation, and renal excretion). Polymorphic pathways (CYP2D6, CYP2C19, and NATs) showed high variability and are of concern because the general kinetic default value would not be adequately protective for any subgroup of the population. An important aspect of genetic polymorphism is to identify the toxicological consequences of metabolism because subjects with impaired metabolism (PMs/SAs) and fast metabolism (EMs and FAs) could be the susceptible subgroup if the toxic species were the parent compound or the metabolite, respectively.
A crucial issue for the risk assessment of subgroups of the population is whether the uncertainty factor (whether a CSAF or a pathway-related factor) should cover a percentile of the subgroup or of the combined total population. This concept applies particularly to polymorphic pathways, because the 99th centile of a PM subgroup that was only 5% of the total population would be the 99.95th centile of the total population. In contrast, although neonates may be only a small percentage of the total population at any point in time, all individuals have been part of that subgroup at some time in their lives! An additional problem is that the frequencies of polymorphism differ between different ethnic groups of the human population. The frequency of CYP2D6 PMs in Caucasians is much higher than that in Asian populations (8% versus 1%), whereas the opposite applies to the CYP2C19 and NAT pathways (2.5% versus 15% and 40–70% versus 10–20%) (Wedlund, 2000). Given this ethnic diversity, different uncertainty factors could be developed if the value were chosen to cover a particular percentage of the total population.
Finally, neonates show the lowest enzyme activity of the whole human population, although reliable data are not available for polymorphic metabolism in this subgroup. The low activity could translate into a greater susceptibility if the parent compound were the active toxicant, but less susceptibility if the compound underwent metabolic activation. In the absence of data on the activity of the relevant pathway(s) of elimination in neonates and the consequences of metabolism (i.e., detoxication or activation), an extra uncertainty factor higher than that in adults for polymorphic metabolism (CYP2D6, CYP2C19, NAT) may be an option to be considered by risk assessors and risk managers (Dorne et al., 2005).
A limitation to the widespread application of pathway-related factors in risk assessment is that most contaminants are eliminated via multiple metabolic and excretion routes. Recently, a meta-analysis of human variability in kinetics for such compounds was carried out, and Monte Carlo modeled predictions of relevant uncertainty factors based on metabolism data and pathway-related variability were validated using published chemical-specific data for the modeled compounds (Dorne and Renwick, 2003). Probabilistic approaches describing both uncertainty and variability represent a potential future improvement by the replacement of uncertainty factors that are derived as single point estimates with a distribution that provides information about the extent of uncertainty in the risk assessment (Dorne and Renwick, 2003; Edler et al., 2002; Slob and Pieters, 1998; Swartout et al., 1998; WHO, 2001).
CONCLUSIONS
The "OMIC" sciences (genomics, proteomics, metabolomics, metabonomics) have opened up considerable opportunities to refine uncertainty/safety factors with chemical-specific data based on our mechanistic understanding at different levels (populations, individuals, cells, and molecular level). Toxicology and risk assessment have benefited enormously by our growing understanding of interspecies differences and inter-individual variability in toxicokinetics and toxicodynamics. These fundamental aspects provide the basis for the refinement of the traditional safety factor approach. This paper has described the development of pathway-related uncertainty factors that provide risk assessors a new intermediate option between default uncertainty factors and chemical-specific adjustment factors (CSAFs). Pathway-related factors have been developed for the kinetic variability, and further work would be required to investigate human variability in toxicodynamics to develop uncertainty factors based on mechanistic data or mode of action, although in reality these type of data are not available or scant (Hashemi, Walton, and Renwick, unpublished report). A pooled analysis of pharmacodynamic variability for different drug effects has been performed (Renwick and Lazarus, 1998), and inter-individual differences were similar to the variability in kinetics. Meta-analyses of a number of different pharmacodynamic end points for both the human variability and interspecies aspects would provide useful tools to replace default safety factors for the dynamic aspect and to design new probabilistic models (Edler et al., 2002; Slob and Pieters, 1998, Swartout et al., 1998; Vermeire et al., 1999; WHO, 2001).
REFERENCES
Barlow, S. M., Greig, J. B., Bridges, J. W., Carere, A., Carpy, A. J., Galli, C. L., Kleiner, J., Knudsen, I., Koeter, H. B., Levy, L. S., et al. (2002). Hazard identification by methods of animal-based toxicology. Food Chem. Toxicol. 40, 145–191.
Besunder, J. B., Reed, M. D., and Blumer, J. L. (1988). Principles of drug biodisposition in the neonate. A critical evaluation of the pharmacokinetic-pharmacodynamic interface (Part I). Clin. Pharmacokinet. 14, 189–216.
Cresteil, T. (1998). Onset of xenobiotic metabolism in children: toxicological implications. Food Addit. Contam. 15(Suppl), 45–51.
Crump, K. S. (1984). A new method for determining allowable daily intakes. Fundam. Appl. Toxicol. 4, 854–871.
Dalen, P., Dahl, M. L., Ruiz, M. L., Nordin, J., and Bertilsson, L. (1998). 10-Hydroxylation of nortriptyline in white persons with 0, 1, 2, 3, and 13 functional CYP2D6 genes. Clin. Pharmacol. Ther. 63, 444–452.
Dalen, P., Dahl, M. L., Eichelbaum, M., Bertilsson, L., and Wilkinson, G. R. (1999). Disposition of debrisoquine in Caucasians with different CYP2D6-genotypes including those with multiple genes. Pharmacogenetics 9, 697–706.
Di Marco, P. N., Priestly, B. G., and Buckett, K. J. (1998). Carcinogen risk assessment. Can we harmonise Toxicol. Lett. 102–103, 241–246.
Dorne, J. L. C. M. (2004). Impact of inter-individual differences in drug metabolism and pharmacokinetics on safety evaluation. Fundam. Clin. Pharmacol. 18, 609–620.
Dorne, J. L., Walton, K., and Renwick, A. G. (2001a). Human variability in glucuronidation in relation to uncertainty factors for risk assessment. Food Chem. Toxicol. 39, 1153–1173.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2001b). Uncertainty factors for chemical risk assessment. Human variability in the pharmacokinetics of CYP1A2 probe substrates. Food Chem. Toxicol. 39, 681–696.
Dorne, J. L., Walton, K., Slob, W., and Renwick, A. G. (2002). Human variability in polymorphic CYP2D6 metabolism: Is the kinetic default uncertainty factor adequate Food Chem. Toxicol. 40, 1633–1656.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2003a). Human variability in CYP3A4 metabolism and CYP3A4-related uncertainty factors for risk assessment. Food Chem. Toxicol. 41, 201–224.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2003b). Polymorphic CYP2C19 and N-acetylation: Human variability in kinetics and pathway-related uncertainty factors. Food Chem. Toxicol. 41, 225–245.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2004a). Human variability in the renal elimination of foreign compounds and renal excretion-related uncertainty factors for risk assessment. Food Chem. Toxicol. 42, 275–298.
Dorne, J. L. C. M., Walton, K., and Renwick, A. G. (2004b). Human variability for metabolic pathways with limited data (CYP2A6, CYP2C9, CYP2E1, ADH, esterases, glycine and sulphate conjugation). Food Chem. Toxicol. 42, 397–421.
Dorne, J. L., Walton, K., and Renwick, A. G. (2005). Human variability in xenobiotic metabolism and pathway-related uncertainty factors for chemical risk assessment: A review. Food Chem. Toxicol. 43, 203–416.
Dorne, J. L. C. M., and Renwick, A. G., 2003. Prediction of human variability using kinetic data and Monte Carlo modelling for the derivation of pathway-related uncertainty factors for compounds handled by multiple pathways. Toxicol. Lett. 144, S195.
Dourson, M. L., Felter, S. P., and Robinson, D. (1996). Evolution of science-based uncertainty factors in noncancer risk assessment. Regul. Toxicol. Pharmacol. 24, 108–120.
Durnas, C., Loi, C. M., and Cusack, B. J. (1990). Hepatic drug metabolism and aging. Clin. Pharmacokinet. 19, 359–389.
Dybing, E., Doe, J., Groten, J., Kleiner, J., O'Brien, J., Renwick, A. G., Schlatter, J., Steinberg, P., Tritscher, A., Walker, R. et al. (2002). Hazard characterisation of chemicals in food and diet. Dose response, mechanisms and extrapolation issues. Food Chem. Toxicol. 40, 237–282.
Edler, L., Poirier, K., Dourson, M., Kleiner, J., Mileson, B., Nordmann, H., Renwick, A., Slob, W., Walton, K., and Wurtzen, G. (2002). Mathematical modelling and quantitative methods. Food Chem. Toxicol. 40, 283–326.
Gentry, P. R., Hack, E., Haber, L. T., Maier, A., and Clewell, H. J. (2002). An approach for the consideration of genetic polymorphism data in chemical risk assessment: Examples with warfarin and parathion. Toxicol. Sci. 70, 120–139.
Ginsberg, G., Hattis, D., Sonawane, B., Russ, A., Banati, P., Kozlak, M., Smolenski, S., and Goble, R. (2002). Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature. Toxicol. Sci. 66, 185–200.
Gow, P. J., Ghabrial, H., Smallwood, R. A., Morgan, D. J., and Ching, M. S. (2001). Neonatal hepatic drug elimination. Pharmacol. Toxicol. 88, 3–15.
Haber, L. T., Maier, A., Gentry, P. R., Clewell, H. J., and Dourson, M. L. (2002). Genetic polymorphisms in assessing inter-individual variability in delivered dose. Reg. Toxicol. Pharmacol. 35, 177–197.
Hattis, D., Banati, P., Goble, R., and Burmaster, D. E. (1999). Human interindividual variability in parameters related to health risks. Risk Anal. 19, 711–726.
Ito, S., Gow, R., Verjee, Z., Giesbrecht, E., Dodo, H., Freedom, R., Tonn, G. R., Axelson, J. E., Zalzstein, E., Rosenberg, H. C., andKoren, G. et al. (1998). Intravenous and oral propafenone for treatment of tachycardia in infants and children: Pharmacokinetics and clinical response. J. Clin. Pharmacol. 38, 496–501.
Kroes, R., Muller, D., Lambe, J., Lowik, M. R., van Klaveren, J., Kleiner, J., Massey, R., Mayer, S., Urieta, I., Verger, P. et al. (2002). Assessment of intake from the diet. Food Chem. Toxicol. 40, 283–326.
Le Couteur, D. G., and McLean, A. J. (1998). The aging liver. Drug clearance and an oxygen diffusion barrier hypothesis. Clin. Pharmacokinet. 34, 359–373.
Lehman, A. J., and Fitzhugh, O. G. (1954). 100-fold margin of safety. Assoc. Food Drug Off. U.S.Q. Bull. 18, 33–35.
Lin, H. J., Han, C. Y., Lin, B. K., and Hardy, S. (1993). Slow acetylator mutations in the human polymorphic N-acetyltransferase gene in 786 Asians, blacks, Hispanics, and whites: Application to metabolic epidemiology. Am. J. Hum. Genet. 52, 827–834.
Naumann, B. D., Silverman, K. C., Dixit, R., Faria, E. C., and Sargent, E. V. (2001). Case studies of categorical data-derived adjustment factors. Hum. Ecol. Risk Assess. 7, 61–105.
Renwick, A. G. (1993). Data-derived safety factors for the evaluation of food additives and environmental contaminants. Food Addit. Contam. 10, 275–305.
Renwick, A. G., and Lazarus, N. R. (1998). Human variability and noncancer risk assessment—An analysis of the default uncertainty factor. Regul. Toxicol. Pharmacol. 27, 3–20.
Renwick, A. G., Barlow, S. M., Hertz-Picciotto, I., Boobis, A. R., Dybing, E., Edler, L., Eisenbrand, G., Greig, J. B., Kleiner, J., Lambe, J. et al. (2003). Risk characterisation of chemicals in food and diet. Food Chem. Toxicol. 41, 1211–1271.
Renwick, A. G., Dorne, J. L., and Walton, K. (2000). An analysis of the need for an additional uncertainty factor for infants and children. Regul. Toxicol. Pharmacol. 31, 286–296.
Silverman, K. C., Naumann, B. D., Holder, D. J., Dixit, R., Faria, E. C., Sargent, E. V., and Gallo, M. A. (1999). Establishing data-derived uncertainty factors from published pharmaceutical clinical trial data. Hum. Ecol. Risk Assess. 5, 1059–1090.
Slob, W., and Pieters, M. N. (1998). A probabilistic approach for deriving acceptable human intake limits and human health risks from toxicological studies: General framework. Risk Anal. 18, 787–798.
Slob, W. (1999). Thresholds in toxicology and risk assessment. Int. J. Toxicol. 18, 259–268.
Smith, M., 2002. Food Safety in Europe (FOSIE): Risk assessment of chemicals in food and diet: Overall introduction. Food Chem. Toxicol. 40, 141–144.
Swartout, J. C., Price, P. S., Dourson, M. L., Carlson-Lynch, H. L., and Keenan, R. E. (1998). A probabilistic framework for the reference dose (probabilistic RfD). Risk Anal. 18, 271–282.
Truhaut, R., 1991. The concept of the acceptable daily intake: An historical review. Food Addit. Contam. 8, 151–162.
Venkatakrishnan, K., von Moltke, L. L., and Greenblatt, D. J. (2001). Human drug metabolism and the cytochromes P450: Application and relevance of in vitro models. J. Clin. Pharmacol. 41, 1149–1179.
Vermeire, T., Stevenson, H., Peiters, M. N., Rennen, M., Slob, W., and Hakkert, B. C. (1999). Assessment factors for human health risk assessment: A discussion paper. Crit. Rev. Toxicol. 29, 439–490.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001a). Uncertainty factors for chemical risk assessment: Interspecies differences in the in vivo pharmacokinetics and metabolism of human CYP1A2 substrates. Food Chem. Toxicol. 39, 667–680.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001b). Uncertainty factors for chemical risk assessment: Interspecies differences in glucuronidation. Food Chem. Toxicol. 39, 1175–1190.
Walton, K., Dorne, J. L., and Renwick, A. G. (2001c). Categorical default factors for interspecies differences in the major routes of xenobiotic elimination. Hum. Ecol. Risk Assess. 7, 181–201.
Walton, K., Dorne, J. L., and Renwick, A. G. (2004). Species-specific uncertainty factors for compounds eliminated principally by renal excretion in humans. Food Chem. Toxicol. 42, 261–274.
Wedlund, P. J., 2000. The CYP2C19 enzyme polymorphism. Pharmacology 61, 174–183.
WHO (1987). Principles for the safety assessment of food additives and contaminants in food. Environmenal Health Criteria. 70, 174 pp., International Programme on Chemical Safety, World Health Organisation, Geneva.
WHO (1994). International Programme on Chemical Safety: Assessing human health risks of chemicals: Derivation of Guidance values for health-based exposure limits. Environmenal Health Criteria, 170, 73 pp., International Programme on Chemical Safety, World Health Organisation, Geneva.
WHO (1999). International Programme on Chemical Safety: Assessing human health risks of chemicals: Principles for the assessment of risk to human health from exposure to chemicals. Environmental Health Criteria 210, World Health Organisation, Geneva.
WHO (2001). International Programme on Chemical Safety: Guidance document for the use of chemical-specific adjustment factors (CSAFs) for interspecies differences and human variability in dose-concentration response assessment. 76 pp., World Health Organisation, Geneva. http://www.ipcsharmonize.org/CSAFsummary.htm.(J. L. C. M. Dorne and A. )