Assessing risk to humans from chemical exposure by using non-animal test data

Toxicology in Vitro 19 (2005) 921–924 www.elsevier.com/locate/toxinvit

Assessing risk to humans from chemical exposure by using non-animal test data Robert D. Combes FRAME, 96-98 North Sherwood Street, Nottingham, NG1 4EE, UK Available online 22 August 2005

Abstract The use of data from non-animal toxicity methods in risk assessment has mainly been limited to hazard identiWcation and for elucidating mechanisms of toxicity. However, there is a need to extend the use of in vitro tests to hazard characterisation and risk assessment. This might be feasible by: (a) increased use of human cells of diVerent types; (b) better maintenance of diVerentiated cells in culture for long periods; (c) use of genetically-engineered cells with useful characteristics; (d) development of complex organotypic cell systems; (e) development of co-cultures of diVerent cell types; and (f) development of techniques for long term culturing, repeat dosing and assessment of recovery. Also, it will be necessary to obtain more information on the diVerences between cells in culture and in situ in tissues, and on the eVects of dosing in vitro and in vivo, to develop realistic and meaningful uncertainty factors to allow in vitro information to be used for risk assessment in its own right, and in conjunction with animal data. These issues and a suggested proposal for using in vitro data in risk assessment are discussed.  2005 Elsevier Ltd. All rights reserved. Keywords: Hazard identiWcation; Hazard characterisation; Risk assessment; Extrapolation from in vitro to in vivo; Safety factors

1. Introduction

2. The potential use of in vitro methods in risk assessment

Assessing the risks from exposure to potentially toxic substances is primarily undertaken by using quantitative information derived from animal test methods (Illing, 2001; Renwick et al., 2003). The use of data from nonanimal (in vitro) testing approaches for risk assessment has traditionally been conWned to: (a) verifying a negative result in vitro; and (b) deWning a potential for in vivo toxicity that has been demonstrated in vitro (especially in genotoxicity testing). However, a new approach to risk assessment is urgently needed to: (a) cope with an increasing demand for hazard data; and (b) reduce the use of laboratory animals. This could be greatly facilitated by making more use of in vitro data for risk assessment.

The process of assessing and dealing with risk from exposure to chemicals comprises: (a) hazard identiWcation (HI); (b) hazard characterisation (HC); (c) risk characterisation (RC), (d) exposure assessment (EA); (e) risk assessment (RA), by using safety factors to adjust no adverse eVect values (NOAELs); and (f) risk management (RM) (see Anon, 1999a,b). Most available in vitro tests are primarily suitable for HI, but not for the other stages of risk assessment. These tests include many that have been recently validated for regulatory use (Spielmann and Liebsch, 2002). For a test to be useful for HC it should: (a) be mechanisticallybased with a biologically plausible relationship between the endpoint being predicted and the phenomenon being

0887-2333/$ - see front matter  2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tiv.2005.07.001

922

R.D. Combes / Toxicology in Vitro 19 (2005) 921–924

modelled; (b) ideally have been validated against human data; (c) have a relevant endpoint that occurs in the target species of interest (usually humans); and (d) have a prediction model that is readily related to toxicity in target species. For a test to be useful for risk assessment purposes it should also: (a) model routes of exposure relevant to those experienced by the target species; (b) produce quantitative data from dose–response information relevant to target species exposures; (c) provide dose– response curves from which potency estimations can be derived; and (d) identify threshold dose levels and plateau eVects that can be extrapolated to target species exposure scenarios, and that can be used to clearly deWne NOAELs. There are many ways to improve the interpretation of in vitro data with respect to human hazard mainly by using: (a) cultured human cells; and (b) organotypic cell culture systems. Human cells possess relevant receptors, targets and metabolising enzymes. Organotypic cultures (a) are multilayered, with spatial diVerentiation; (b) have cells that can communicate in three dimensions; and (c) can comprise diVerent cell types (Clothier et al., 2000; Combes et al., 2002; Garle and Fry, 2003). The development of such models has been facilitated by tissue engineering techniques (Bhandari et al., 2001). Organotypic cultures usually also possess useful physiological properties, such as barrier function (Clothier et al., 1995; Passonen-Seppänen et al., 2001). Moreover, the diVerentiation of cells into layers, as well as the presence of more than one cell type, oVers a complex set of potential target cells for toxicity. In vitro methods are often very useful for investigating: (a) any synergistic or antagonistic eVects; (b) the contribution of any impurities to toxicity; (c) the contribution of any metabolites to toxicity, and the ability of speciWc enzymes to activate or detoxify a substance; (d) structure/activity relationships; (e) species/organ speciWcity; (f) receptor-binding aYnity (agonistic/antagonistic activity); (g) dose–responses (relative potency and the existence of thresholds); and (h) reversible toxicity.

3. Limitations of in vitro test systems In vitro test systems: (a) lack normal methods for absorption, distribution, metabolism and excretion (ADME); (b) lack intact immune, endocrine and nervous systems; and (c) are constrained by diYculties in obtaining certain cells and tissues and maintaining them in a fully-diVerentiated state (Kahn et al., 1993). However, many of these limitations can be alleviated by using advances in tissue culture methodology (Elton et al., 1999), and also by obtaining cells from human tissue banks (Orr et al., 2002; Combes, 2004). In risk assessment, eVects detected at the dose level applied to the test system (the external dose) must be

related to the eVects that would be caused by the dose that actually reaches the target cells (internal dose). The internal target organ dose can be predicted by undertaking toxicokinetic studies, particularly ADME. A knowledge of target organ metabolism can be used with the above information, as also can information on target organ eVects, to more closely relate the eVects of external and internal doses by developing biologically-based dose–response models, using in vivo data. Thus, it is also possible to use doses levels in vitro that reXect target organ concentrations. In addition, extrapolating between species can be facilitated by PBPK modelling (Blaauboer, 2002). It is also possible to use: (a) culture systems where regular disposal of old media and supply of fresh media occur (e.g. hollow Wbre cultures; Pfaller et al., 2001); and (b) a wide variety of metabolising systems (Combes, 1992). False negative results can arise from inherent inadequacies of the test system or to artefactual eVects, for example where a reactive, potentially toxic metabolite is quenched by an intracellular or extracellular component (e.g. by non-speciWc binding to serum protein; see Benoit et al., 1987), or where a metabolite or test material cannot enter the target indicator cells. It is, therefore important to assess the bioavailability, biokinetics and biodynamics of a substance tested in an in vitro assay (Gülden and Seibert, 1997). Cells in tissue culture systems and in situ in the body can also diVer in their behaviour and response to toxic insult due to diVerences in (a) structure and morphology, and in attachments and communications to cells in situ within the body (e.g. causing a loss of polarity (Yee and Day, 1999)) and (b) the proportions of diVerent cell types present.

4. Discussion Risk assessment is generally conducted by applying to the hazard data an overall adjustment factor of 100-fold. This factor is usually divided into two principal components of 10-fold each: (a) a factor taking account of interspecies diVerences and (b) a factor to account for inter-individual diVerences. These, in turn have each been broken down still further into two factors of 2.5 (for toxicodynamics) and 4.0 (for toxicokinetics) in the case of interspecies diVerences, and of 3.16 each for toxicodynamics and toxicokinetics for inter-individual diVerences (Anon, 1999b; Renwick, 1993). The use of these adjustment factors means that a value for toxicity (e.g. a NOAEL value) obtained in an animal test is divided by 100 so as to adopt the precautionary principle by setting a safe level of exposure at two orders of magnitude below one directly indicated by the animal data. There is no objective reason why an overall adjustment factor of 100 should be used, and the level of

R.D. Combes / Toxicology in Vitro 19 (2005) 921–924

adjustment necessary will in reality depend on numerous variables, including the test method, the species, the route of application, the test substance, the way the animals are handled, fed, housed, and the relationship between the endpoints measured and the toxicity being predicted in humans. Thus, adjustment factors are little better than arbitrary numbers. If arbitrary adjustment factors can be used to interpret animal test data for risk assessment, then in principle they could be used with in vitro data for the same purpose, particularly where such data are deemed relevant for HC or the later stages of risk assessment. A suggested rationale for applying such an adjustment to in vitro data for risk assessment purposes is presented in Fig. 1, based on the assumption that in vitro test systems are more sensitive than animal models to toxic insult (due to their properties discussed earlier), and retaining an overall adjustment factor of 100. This proposal has been made merely to stimulate discussion. In order for more accurate adjustment factors to be derived for using toxicity data from in vitro systems, additional research is required particularly with regard to the factors discussed in this paper. This will include an in-depth comparison of the cytoskeleton in cells in vivo and in vitro (Svitkina et al., 1997). The cytoskeleton, a network of protein Wbres throughout the cell that provides structural support, with important roles in dynamic processes such as signalling and intracellular

100-FOLD OVERALL UNCERTAINTY FACTOR* INTER-INDIVIDUAL DIFFERENCES 5-FOLD@

EXTRAPOLATION IN VITRO TO IN VIVO 20-FOLD+ DIFFERENCES IN METABOLISM (2.72-fold)

DIFFERENCES IN DOSE (2.72-fold)

CELLS TO WHOLE BODY (2.72-fold)

DIFFERENCES IN METABOLISM (2.24-fold)

DIFFERENCES IN KINETICS & DYNAMICS (2.24-fold)

*could be higher factor; eg 150-fold +if human cells obviating spp extrapolation; @human receptors/targets

Fig. 1. Suggested application of uncertainty factors to in vitro data.

(1)

AD

DR

(2)

IVD

MI

(3)

IVD

DR

923

traYcking, is obviously a key determinant of cell behaviour. In addition, more work on the diVerences between other physiological processes, such as transporters in cells in tissue culture and in the body (Hanu et al., 2000), and on in vitro biokinetics and biodynamics is required. Gülden and Seibert (1997) have previously shown that in vitro data can be corrected to account for compound bioavailability in in vitro test systems. The Health and Safety Executive in the UK has assessed the risk of exposure to several methyl benzimadazole carbamate fungicides, in particular benomyl and carbendazim. These chemicals exhibited clear threshold dose levels for aneuploidy (chromosome loss), detected in vitro by measuring interference with microtubule activity in cultured cells. The threshold doses were converted into estimated NOAELs by mathematical modelling, and the use of the chemicals was regulated on the basis of these values (Anon, 2003). Studies on the eVects of methylmercury on neurogenesis in vivo in rats and in vitro, by using primary midbrain cell cultures (Faustman et al., 2002) found that cell cycle progression was inhibited at the equivalent of 4– 7
g Hg/g in the cells (converted from 1–2
M in cell culture). This was comparable to estimated brain levels of Hg required to exert the same eVects and developmental toxicity to the brain in vivo (Lewandowski et al., 2000). The authors concluded that ‘a carefully chosen in vitro system can provide dose–response data which are quantitatively similar to those observed in vivoƒ. such an approach also provides a framework by which in vitro data can be included in dose/response assessment.’ The above examples indicate that in vitro data can be used, in conjunction with adjustment factors, in the overall assessment of animal toxicity data for risk assessment. Thus, instead of regulatory agencies relying primarily on in vivo data (pathway (1) in Fig. 2), they should use in vitro information in a parallel way for risk assessment (pathway (2) in Fig. 2) (as originally proposed by Chamberlain, 1996), or even in its own right (pathway (3) in Fig. 2). In conclusion, the possibility of using in vitro data for risk assessment should seriously be considered, as a means of easing the likely testing and regulatory burden of legislation.

EA

RA

MI

AD = animal data; IVD = in vitro data; DR = dose response; MI = mechanistic information; EA = exposure assessment; RA = risk assessment

Fig. 2. Using in vivo and in vitro data for risk assessment.

924

R.D. Combes / Toxicology in Vitro 19 (2005) 921–924

Acknowledgement Some of the ideas in this manuscript arose from discussions between members of the FRAME Toxicity Committee Working Group on Risk Assessment, to whom the author is grateful.

References Anon, 1999a. Principles for the assessment of risks to human health from exposure to chemicals (International Programme on Chemical Safety, Environmental Health Criteria 210), Geneva, Switzerland, World Health Organisation. Anon, 1999b. Risk assessment approaches used by UK government for evaluating human health eVects of chemicals. Institute for Environment & Health, Leicester, UK, 57pp. Anon, 2003. Uncertainty factors: their use in human health risk assessment by the UK government. The Interdepartmental Group on Health Risks from Chemicals, MRC Institute for Environment and Health, Leicester, p. 68. Benoit, J., Cormier, M., Wepierre, J., 1987. EVect of proteins on the assessment of surfactant cytotoxicity by an in vitro test: possible correlations with in vivo data. Toxicology in Vitro 1, 91–96. Bhandari, R.N.B., Riccalton, L.A., Lewis, A.L., Fry, J.R., Hammond, A.H., Tendler, S.J.B., ShakesheV, K.M., 2001. Liver tissue engineering: a role for co-culture systems in modifying hepatocyte function and viability. Tissue Engineering 7, 345–357. Blaauboer, B.J., 2002. The applicability of in vitro-derived data in hazard identiWcation and characterisation of chemicals. Environmental Toxicology and Pharmacology 11, 213–225. Chamberlain, M., 1996. A risk assessment paradigm: OECD, cited in Final Report of the OECD Workshop on Harmonization of Validation and Acceptance Criteria for Alternative Toxicological Test Methods, Paris, France: OECD (slides available from Secretariat) 60pp. Clothier, R.H., Atkinson, K.A., Garle, M.J., Ward, R.K., Willshaw, A., 1995. The development and evaluation of in vitro tests by the FRAME Alternatives Laboratory. ATLA 23, 75–90. Clothier, R., Orme, A., Walker, T.L., Ward, S.L., Kruszewski, F.H., DiPasquale, L.C., Broadhead, C.L., 2000. A comparison of three cytotoxicity endpoints in the corneal HCE-T model. ATLA 28, 293–302. Combes, R.D., 1992. The in vivo relevance of in vitro genotoxicity assays incorporating enzyme activation systems. In: Gibson, G.G. (Ed.), Progress in Drug Metabolism, vol. 13. Taylor and Francis Limited, London, pp. 295–321. Combes, R.D., 2004. The use of human cells in biomedical research and testing. ATLA 32 (1), 43–49. Combes, R., Balls, M., Bansil, L., Barratt, M., Bell, D., Botham, P., Broadhcad, C., Clothier, R., George, E., Fentem, J., Jackson, M., Indans, I., Laozou, G., Navaratnam, V., Pentreath, V., Phillips, B., Stemplewski, H., Stewart, J., 2002. An assessment of progress in the use of alternatives in toxicity testing since the publication of the report of the Second FRAME Toxicity Committee (1991). ATLA 30, 365–406. Elton, R.C., Rhodes, P., Fry, J.R., 1999. Evaluation of a short-term rat proximal tubule incubation system for the detection of nephrotoxicants. ATLA 27, 433–448.

Faustman, E.M., Ponce, R.A., Ou, Y.C., Mendoza, M.A.C., Lewandowski, T., Kavanagh, T., 2002. Investigations of methylmercury-induced alterations in neurogenesis. Environmental Health Perspectives 110, 859–864. Garle, M.J., Fry, J.R., 2003. Sensory nerves, neurogenic inXammation and paid: missing components of alternative irritation strategies? a review and a potential strategy. ATLA 31, 295–316. Gülden, M., Seibert, H., 1997. InXuence of protein binding and lipophilicity on the distribution of chemical compounds in in vitro systems. Toxicology In vitro 11, 479–483. Hanu, R., McKenna, M., O’Neill, A., Resneck, W.G., Bloch, R.J., 2000. Monocarboxylic acid transporters, MCT1 and MCT2, in cortical astrocytes in vitro and in vivo. American Journal of Physiology— Cell Physiology 278, C921–C930. Illing, H.P., 2001.Toxicity and Risk—Context, Principles and Practice Taylor and Francis, Basingstoke 168pp. Kahn, C.R., Young, E., Lee, I.H., Rhim, J.S., 1993. Human corneal epithelial primary cultures and cell lines with extended life span: in vitro model for ocular studies. Investigative Ophthalmology and Visual Science 34, 3429–3441. Lewandowski, T.A., Ponce, R.A., Bartell, S.M., Faustman, E.M., 2000. Comparison of developmental toxicity of methylmercury in vitro and in vivo: potential value of in vitro-derived data for dose– response assessment. Society for Risk Analysis Annual Meeting, Arlington, VA, December (). Orr, S., Alexandre, E., Clark, B., Combes, R., Fels, L.M., Gray, N., Jnsson-Rylander, A-C., Helin, H., Koistinen, J., Oinonen, T., Richert, L., Ravid, R., Salonen, J., Teesalu, T., Thasler, W., TraVord, J., van der Valk, J., von Versen, R., Weiss, T., Womack, C., Ylikomi, T., 2002. The establishment of a network of European human research tissue banks. Cell and Tissue Banking 3 (2), 133–137. Passonen-Seppänen, S., Suhonen, T.M., Kirjavainen, M., Miettinen, M., Urtti, A., Tammi, M., Tammi, R., 2001. Formation of permeability barrier in epidermal organotypic culture for studies on drug transport. Journal of Investigative Dermatology 117, 1322– 1324. Pfaller, W., Balls, M., Clothier, R., Coecke, S., Dierickx, P., Ekwall, B., Hanley, B.A., Hartung, T., Prieto, T., Ryan, M.P., Schmuck, G., Sladowski, D., Vericat, J.-A., Wendel, A., Wolf, A., Zimmer, J., 2001. Novel advanced in vitro methods for longterm toxicity testing: The Report and Recommendations of ECVAM Workshop 45. ATLA 29, 393–426. Renwick, A.G., 1993. Data-derived safety factors for the evaluation of food additives and environmental contaminants. Food Additives Contaminants 10, 275–305. 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., Muller, D.J.G., Smith, M.R., Tritscher, A., Tuijtelaars, S., Brandt, P.A.v.d., Walker, R., Kroes, R., 2003. Risk characterisation of chemicals in food and diet. Food and Chemical Toxicology 41, 1211–1271. Spielmann, H., Liebsch, M., 2002. Validation successes: chemicals. ATLA 30 (2), 33–40. Svitkina, T.M., Verkhovsky, A.B., Borisy, G.G., 1997. Electron microscopy of the cytoskeleton on cultured cells. In: Motta, P.M. (Ed.), Recent Advances in Microscopy of Cells, Tissues and Organs Antonio DelWno Editore S.r.l., Rome, pp. 93–100. Yee, S., Day, W.W., 1999. Application of Caco-2 cells in drug discovery and development. In: Woolf, T.F. (Ed.), Handbook of Drug Metabolism. Marcel Dekker, Inc., New York, NY, pp. 507–522.