Better Science IV. TESTING ALTERNATIVES Animal-based (in vivo) testing is characterized by massive suffering and questionable scientific value. Although safety testing does not involve the largest number of animals utilized in the United States specifically or the world in general, the numbers are still very high. In vivo testing protocols often involve severe levels of pain and distress and account for the majority of animals listed in USDA annual statistics as experiencing pain without anesthesia or analgesics. By its nature, animal-based toxicity testing is deliberately designed to cause injury, pain and/or death to some or all of the animals involved. Dr. Gerhard Zbinden, one of the world's leading toxicologists, once described a standard in vivo bioassay test as little more than "a ritual mass execution of animals." Efforts to Refine the severity of the animal's experience and Reduce the numbers involved are in progress, but have had limited success. Any advances in these activities may be overshadowed by recent calls for massive, new animal-intensive testing programs of dubious necessity. A combination of activities will be needed to replace the use of in vivo toxicity tests with more humane methods. Initially, we need realistic information on human exposure to individual chemicals to determine actual risk, as well as a comprehensive database on past human exposure experience. Little of this type of information is currently available and may not yet exist due in large part to companies and government agencies not sharing in-house data. As new and existing in vitro methods become more widely used and accepted for regulatory purposes, the importance of multi-step, tier-testing strategies will be recognized. Such approaches allow testing for multiple endpoints that more accurately reflect the mechanistic processes involved in toxic exposures. Concomitantly, there will be a shift away from tests (such as the Draize Eye Irritation which involved placing various test substances in one eye of a rabbit and subjectively recording any damage that resulted) that have more to do with symptoms than actual toxic responses. High Throughput Screening (HTS), (now widely used in-house by pharmaceutical companies in research, drug development, candidate screening of potential compounds), followed by Medium Throughput Screening (MTS) methods currently under development, will examine the absorption, disposition, metabolism and excretion of test compounds. Eventually the expanding fields of toxicogenomics and proteonomics may provide quick, accurate "toxicity on a chip" and eliminate all remaining in vivo and much of in vitro techniques. At present the process of hazard assessment can begin with computer (in silico) approaches in association with in vitro cell culture models. This combination either exists, is being developed or is under validation trials for: eye irritation skin irritation phototoxic potential (UV radiation) nephrotoxicity (kidney) reproductive toxicity skin penetration chronic toxicity blood-brain barrier gastrointestinal barrier Tier-testing strategies are either under developed or available for: physiochemical properties acute toxicity skin corrosivity skin sensitization carcinogenicity/genotoxicity xenobiotic metabolism neurotoxicity endocrine disruption What follows is a brief review of some key areas of toxicity testing and the current status of associated alternatives development and use in each. Acute Toxicity Essentially all of the basic in vivo toxicity tests measure some aspect of acute exposure with organ- or system-specific endpoints. There are, however, more general acute poisoning studies designed to provide information on substance concentrations necessary to produce death or severe injury. Routes of exposure may vary between oral, skin and inhalation. Such in vivo tests, whether combined with detailed histopathological (microscopic damage to cells and tissues) examination or not, represent little more than graduated, mass poisoning of surrogate species (e.g., mouse, rat, dog) already known to respond differently than humans. The most egregious example of such a useless test was the classical LD50 (Lethal Dose = 50% of animals die). Developed in the early 20th century to standardize the production batches of digitalis (a concern that can now be addressed using non-animal alternatives), it historically acquired a level of toxic significance for which it was never intended and was entirely unsuited. After more than two decades of criticism and documentation of the failures of the LD50, its use as a standard, worldwide test has finally ended. This long-overdue response was, however, delayed for several years due to the refusal of one national regulatory agency (the United States Environmental Protection Agency) to abandon support for its use. Development and validation programs to find in vitro replacements to measure general acute toxic exposure are underway in Europe and the United States. QSAR (computer) models are widely used by industry to either prioritize in vivo or identify the need for further in vitro tests. Since acute systemic toxicity (single exposure) results from cytotoxicity-associated (i.e., cellular damage) responses, measuring cytotoxicity in cell cultures should be predictive of the more general in vivo response. There are large existing databases of cytotoxicity information available in the biomedical literature. The Multicenter Evaluation of In Vitro Cytotoxicity (MEIC) program, which NEAVS as well as other national and international animal and scientific organizations funded, identified 69 different methods with potential applications for acute toxicity prediction and a subset of these that has a general predictive ability of 84% for humans. In contrast a standard rodent test might have only 65% accuracy. In some cases these in vitro methods also identify human toxic mechanisms that can only be seen in the alternatives. This may explain why such non-animal methods are so widely used for non-regulatory, in-house purposes. A battery of MEIC-identified tests are currently undergoing validation as in vitro replacements for the now discredited LD50 test. Furthermore, MEIC has initiated a new series of research efforts (EDIT - Evaluation-guided Development of New In Vitro Tests) to expand the earlier work with an emphasis on the use of human cells and improved use of toxicokinetic data. The German government has published a comprehensive Registry of Cytotoxicity (RC) that collects a large volume of relevant data from in vitro cells, comparisons with in vivo LD50 values and the use of linear regression analysis. This database can be used to predict some acute toxicity and reduce the number of animals and compounds utilized for test purposes. Ultimately, acute toxicity testing will be based on a tiered-testing strategy utilizing QSARs, a battery of basal cytotoxicity tests (e.g., MEIC), assessments of biotransformations and cell-specific toxicity protocols. The testing of any compound would end, if a positive result is identified. Using animals as surrogate tasters to identify human poisons will end. Chronic Toxicity One traditional criticism of in vitro replacement alternatives was their inability to mimic or reproduce the consequences of long-term, chronic human exposure to toxic substances. This is no longer the case. Essentially all of the endpoints measured in animal-based tests can be transferred to appropriate in vitro systems. This is particularly applicable as the mechanistic bases for such endpoints become identified and characterized. In most cases, however, such tests measure the consequences of acute, not chronic exposure. As cell culture technology has evolved, it is now possible to maintain in vitro systems for longer periods of time - weeks or months. It is equally apparent that it is not necessary to maintain such cultures for years, as is done with some typical chronic in vivo tests. Long-term cell and tissue culture techniques now allow the study in vitro of the effects of chronic, repeated exposure to toxic substances, as well as the recovery from such exposure. To be a valid in vitro replacement, such long-term cultures must: retain their differentiated functions, as is the case for in vivo models; maintain stable, reproducible conditions; involve perfusion systems with continuous replacement of nutrient media, removal of waste products and allow for periodic exposure to toxic test substances; be based on serum-free cultures of human cell lines, which have unlimited lifespans and are more directly relevant to human exposures; include the possibility of co-cultures of multiple cell types to more closely match organotypic conditions. Pilot studies with systems such as Technomouse, Integra and EpiFlow have all demonstrated the feasibility of long-term cell cultures for chronic exposure studies. One test for a low-dose neurotoxin utilizing epithelial cells grown in a hollow-fiber perfusion system gave results that "mirrored" those found in traditional in vivo studies. Other ongoing programs are exploring the use of genetically engineered cells and applications of different co-culture and three-dimensional in vitro systems to address the issue of identifying problems associated with chronic toxicity. Although the MEIC Project was designed to measure the effects of acute exposure, some of the test components appear relevant to chronic situations. MEIC laboratories were able to maintain cultures for up to six weeks. As these long-term culture techniques are refined, validated and come online as in vitro replacements for traditional animal-based methods, they should not be viewed as high-throughput systems. Such cultures will still require longer periods of time. Their use, however, should only be necessary if other in vitro acute exposure tests suggest problems may exist. Computers and Toxicology Although the ability of computers to replace in vivo experiments and test is occasionally overstated, toxicity testing is ideally suited to the application of such in silico approaches. What computers do best is compare and contrast large quantities of quantitative and qualitative data. Safety testing is based on the production and use of such information. There are three basic types of computer applications in safety testing: SAR -- Structure Activity Relationships - examine known associations between chemical structures and biological activity with similar data for new substances. QSAR - Quantitative Structure Activity Relationships - creates multivariable mathematical relationships between chemical structure, physiochemical properties and biological activity. Expert Systems are any formal approach (with or without the use of computers) that allows the rational prediction of toxicity of test substances. QSARs are currently widely used for non-regulatory purposes in the study of skin corrosion, skin irritation, eye irritation, the blood-brain barrier, acute toxicity, metabolic endpoints (to name a few), and as the initial step in tiered testing strategies. Pharmaceutical companies routinely used QSARs to design new chemical agents and drugs, with applications to toxicology only recently adopted (although not yet formally validated for toxicity endpoints). SARs utilize computers to identify portions of molecules that are known to be associated with specific (in this case -- toxic) biological properties. Both QSARs and SARs have the advantage over in vivo of being based on detailed knowledge of chemical structure (determined by physical analysis); easily transferred to computer automation; and extremely quick response times. There are several current limitations on the utility of such computational approaches. These include: the absence of good quality toxicological information despite decades of animal tests and human exposure data; overly simplistic models of complex toxicological mechanisms; and extrapolation beyond the area covered by the available data. These restraints are being addressed and removed as: there is more reliance on mechanisms of toxicity than simple chemical structure; increased use of human cells and tissues; better characterization of cellular receptor structure and properties; improvements in computer databases; and incorporation of information from the human genome project and toxicogenomics. Despite their limitations and continued development, there are already a number of high-powered programs available and in widespread use for non-regulatory purposes. These include: TOPKAT (Toxicity Prediction by Computer-Assisted Technology), CASE (Computer Automated Structure Evaluation), COMPACT (Computerized Optimized Parametric Analysis of Chemical Toxicity), DEREK (Deductive Estimation of Risk from Existing Knowledge), Hazard Expert, ONcologic and Meteor. Several of these systems have, for example, overall accuracies of 60 to 90% for some standard toxicity endpoints (e.g., rodent carcinogens). Ocular Toxicity The classic Draize Eye Irritancy Test was characterized more than two decades ago as extremely inhumane, of questionable relevance to human or animal exposure to harmful substances, and scientifically flawed with high degrees of inter- and intra-laboratory variability for the same test materials. Computer simulations documented a lower correlation between repeat tests of the same chemicals than would be tolerated for any in vitro test used for regulatory purposes. The Draize is simply not consistently reproducible and thus cannot be reliably used to predict human risks. Although criticized on both ethical (pain and distress to the animals involved) and scientific (multiple, significant anatomical and physiological differences between the eyes of rabbits and humans), there still remains a widespread misconception among some toxicologists and regulatory officials and agencies, that the Draize provides a valid measure of eye irritation potential. The poor quality of Draize data also contributes to difficulties in replacing it. Due to the seriously compromised nature of existing in vivo data, potentially valid alternatives have technically failed validation efforts when compared to bad Draize data. Reanalysis of previously conducted validation studies for in vitro Draize replacements, unbiased by the poor quality animal data, suggest these replacements are adequate to identify potentially hazardous materials, especially if utilized as a tier-testing strategy. A number of in vitro replacements are widely used in-house by industry to eliminate nearly all requirements for the classical Draize test. They are also accepted on a case-by-case basis by several national regulatory agencies other than the United States. In vitro replacements for the Draize Eye Irritancy Test include one or more of the following, done in combination with each other and several QSAR computer analysis programs: use of chorioallantoic membrane of a fertilized hen's egg (does not damage the embryo) (HET-CAM test) is the only one to include responses of the circulatory system. fluorescein leakage (utilizing MDCK dog kidney monolayer cell cultures and measuring dye leakage due to damaged inter-cellular connections. neutral red uptake (NRU) and neutral red release (NRR), which are indicators of cytotoxicity (cell injury/death) and are excellent predictors of responses to severe irritants. agarose diffusion, which measures cell death in cultures exposed to test materials. EpiOcular, which is a sophisticated reconstituted human corneal epithelial cell culture that provides data on cell viability, release of indicators for inflammation and changes in membrane permeability. A comprehensive German study of these methods proved that the HET-CAM and NRU tests could identify severe eye irritants and eliminate them from further testing protocols. There remains a need to create new in vitro methods to be added to the existing battery of replacement alternatives to the Draize. These should be based on specific cellular and molecular endpoints associated with positive eye irritation responses. At its simplest level, any substance producing a positive skin irritancy response should be labeled as an eye irritant and not be further tested. Skin Toxicity There are four basic aspects of skin toxicity that are routinely tested: corrosivity, irritation, sensitization and absorption. Skin corrosivity can be easily measured using in vitro systems such as EPISKIN (a three-dimensional human skin model) and EpiDerm (a reconstructed human skin model) - both of which measure cell viability as an endpoint and have been accepted for regulatory purposes in the European Union and by the OECD. In addition, a non-cellular test, Corrositex, is approved in the EU and the United States, but only for acids, bases and their derivatives. Corrositex utilizes a specialized collagen matrix membrane. There is no longer any justification to do animal testing for this endpoint, especially if the in vitro methods are combined with a tiered approach involving such physical parameters as pH. Skin irritation potential is currently measured using the classic Draize skin test, the lesser-known cousin of its ocular counterpart. Several promising in vitro tests are currently subject to validation programs. These include the EPISKIN and EpiDerm systems, as well as Prediskin (a human skin culture derived from plastic surgery discards) and a variety of sophisticated QSAR models - one of which had a sensitivity of 85% and specificity of 92%. Animal-based tests for this same endpoint should be eliminated in the very near future with an in vitro replacement used immediately for prescreens and priority setting. Skin sensitization assays are designed to identify a substance's potential to produce contact dermatitis. Increased responses associated with this endpoint have helped to create new or reapply existing replacement alternatives. Several computer techniques (DEREK, TOPKAT and CASE) include this endpoint. L'Oreal is developing a human reconstituted epidermis, multi-cell culture model that includes such unique components as melanocytes, keratinocytes and Langerhans cells. MatTek Corporation is actively working on an in vitro replacement for the Murine Local Lymph Node Assay (LLNA), a reduction alternative currently accepted in the EU, United States and by the OECD. Complete replacement of in vivo skin sensitization tests is a realistic short-term expectation. Percutaneous absorption can be measured using any one of the currently available in vitro reconstituted human skin equivalents since studies show that such methods provide data predictive of human and animal exposure to test substances. In vitro replacements for this endpoint have already been accepted by the OECD. Genotoxicity / Mutagenicity (changes to genes) and Carcinogenicity (cancer-causing) It is now widely accepted by regulatory officials and toxicologists that screening for mutagenic potential can be done via in vitro methods such as the Bacterial Reverse Mutation Test; In Vitro Cell Line Mutation Test, or the In Vitro Chromosomal Aberation Test. There are also several SAR, QSAR and Expert Systems available for this endpoint. The only potential problem with these assays is the existence of mechanisms that produce non-genotoxic carcinogenesis. New in vitro tests based on Syrian Hamster Embryo (SHE) cell lines may address this concern. However, some toxicologists question the significance of such carcinogens to human risk assessment since their activity profiles are often only identified in mice. Toxicokinetics / Biokinetics -- ADME Actual systemic toxicity depends on several variables -- external dose; rate of exposure; absorption, distribution, metabolism and excretion (ADME); and the intrinsic characteristics of the test material. All of these can be identified and modeled using computer and in vitro approaches. Studies focusing on ADME are now human-based, mechanistic protocols that provide both predictive and computerized models. Although classical in vivo toxicity tests are based on dose/response relationships for entire animals, a more realistic approach might focus on concentration/response curves at the actual toxic target within the recipient's body. In vitro methods are especially useful for such studies on the biological activity and mechanisms of toxic response of chemicals. Programs such as MEIC have provided evidence of the value and utility of this approach. Perhaps most significantly, the creation of toxicokinetic-derived QSAR programs will allow toxic exposure from one test or set of tests to be used to predict the response for other types of tests. This would eliminate the need for the latter and replace the animals used with simple abstinence. As a specific example, the Environmental Protection Agency (EPA) announced on July 14, 2003, that they had conducted a review to determine if chemical companies could use physiologically-based pharmacokinetic computer models to extrapolate data from previous oral toxicity studies to predict potential hazardous consequences of inhalation exposure to the same substance. The EPA endorsed this approach to both reduce the number of costly tests required and to eliminate some current uses of animals. As mechanistic data, in vitro methods, computer simulation and DNA-chip technology continues to improve and intertwine, the justifications offered to defend continued in vivo testing requirements will become more and more tenuous. Pyrogen Testing This test is designed to identify potential bacterial contamination of injectable products (originally), implants, medical devices, dialysis machines, cellular therapies, recombinant proteins and IV products. Injectable drugs have been around for more than 100 years. Sixty years ago the rabbit pyrogen test (involving injection of test materials to check for reactions to contamination) was developed and subsequently millions of rabbits died. Twenty-five years ago the LAL (Limulus amebocyte lysate) alternative was developed based on the coagulation response of horseshoe crab blood when exposed to bacterial toxins. In theory blood is collected from the crabs, who are then released. In practice, poor technique and carelessness lead to a high percentage of crab fatalities. To avoid killing the crabs and limitations of the LAL test, as well as the need for replacing the 400,000 rabbits still used worldwide, researchers in Europe developed a new pyrogen test based on human isolated blood cells, cell lines and whole blood incubation to detect the presence of fever reaction products - a direct measure of pyrogen contamination that would affect human patients. This approach can also identify both immunostimulants and immunosuppressants. Phototoxicity Although it took seven years to complete the validation/approval/adoption process, there now is an in vitro replacement alternative to identify phototoxic (i.e., drugs and chemicals become toxic when human recipients are exposed to sunlight) potential. The 3T3 Neutral Red Uptake Phototoxicity Test (3T3 NRU PT) utilizes a mouse-derived cell line which measures the degree of cellular damage (cytotoxicity) of the cultures and toxicants when tested in the presence and absence of non-cytotoxic exposure to UVA light. Additional validation tests are currently being conducted on other in vitro methods (e.g., EpiDerm PT) to provide additional phototoxicity alternatives. These in vitro methods are accepted by both the OECD and the European Union testing authorities. There is no further justification for the continued use of in vivo tests for this toxic endpoint. Embryotoxicity / Teratogenicity There are currently more than a dozen in vitro methods representing various aspects of the reproductive process. The use of immortalized mammalian cell lines, especially embryonic (not derived from therapeutic abortions) stem cells are being used to create in vitro assays for teratogenicity that are directly predictive of human toxic risks. Using rodents for such studies is especially inappropriate due to the major physiological, biochemical and structural differences between human and rodent placentas. The Embryonic Stem Cell Test (EST) has been validated by the European Center for the Validation of Alternative Methods (ECVAM) and accepted in the European Union for the identification of embryotoxicants. Of the currently available alternatives, it is the only one suitable for high throughput screening and avoids killing large numbers of pregnant animals. It also identifies three unique endpoints representing the principal reproductive toxicological mechanisms. Endocrine Disruptors This represents a newly hypothesized class of potential toxic effects on human and wildlife reproductive systems for which there were no existing animal-based tests. Although there is evidence that humans may be unaffected by endocrine disruptors, there is also evidence of negative impacts on other species (especially wildlife). Current proposals for in vivo-based testing protocols share a set of serious problems including: lack of reproducibility insufficient or no validations inability to apply standard validation requirements questionable relevance of the data For these reasons multiple in vitro screens and QSAR computer models are being developed based on mechanistic endpoints that can only be examined using such alternatives. Unlike most existing in vivo tests, because this area of toxic concern is entirely new, it may be possible to create an alternatives-focused, tier-testing strategy that will "do it right" the first time. Metabolic Toxicity Some chemicals and drugs are essentially nontoxic but become hazardous once ingested and metabolized by the body. For this reason, information from in vitro systems utilizing human cell lines, genetically engineered human cells and subcellular components as well as several computer-based systems (METEOR, Hazard Expert, Metabol Expert, COMPACT) are being utilized to detect metabolism-mediated toxicity. Because of the enormous species differences in metabolic parameters (especially between humans and rodents -- the animals most frequently used for such tests), it is critical that such studies utilize human-based in vitro techniques and human data for computer simulation. This is one area of toxicology for which animal models are widely acknowledged by toxicologists to be inappropriate. Work is currently underway to create a simple microchip that will provide all of the necessary metabolic information simultaneously and in a human-specific context. Nephro (Kidney) Toxicity For many years primary cultures of kidney cells have been powerful tools to study renal function and toxicity. A number of in vitro toxicity endpoints are currently being investigated with an emphasis on using immortalized renal epithelial cell lines (e.g., MDCK cells originally derived from dogs). In order to reproduce some of the structural and cellular complexity of the kidneys, new perfusion culture techniques (e.g., EpiFlow) were developed that allow longer-term, simultaneous cultures of two or more cell types. Efforts are also underway to replace all of the animal-derived cell lines and cultures with their respective human counterparts and to identify consistently relevant in vitro toxic endpoints. Neuro (Brain/Nerve) Toxicity The routine use of in vitro methods for research and testing models dates back more than twenty years. Long-term cultures of neural and support cells are currently in use by industry to screen for toxic effects of pharmaceuticals, agricultural chemicals and other compounds. In addition, a large number of in vitro systems are being developed as toxicity screens and indicators of multiple toxicity endpoints. These include neuronal cell lines, genetically engineered cells and reaggregating brain cell cultures (which reproduce some of the in vivo complexity of the brain). Eventually a tiered testing (multiple levels of pass/fail testing) strategy, incorporating several of these in vitro methods should be sufficient to identify neurotoxic hazards. There is also evidence that not all potential endpoints need to be examined to adequately predict substances of concern. Toxicogenomics If toxicology is to eventually evolve from its primitive beginning in quantifying the mass poisoning of various species of animals, the final high-tech destination may be in the field of toxicogenomics and its sister disciplines of proteonomics and metabonomics - all of which integrate the interactions between human genes and toxic substances, proteins and metabolic activities respectively. The ultimate goal of toxicogenomics is a single or series of DNA chips that would provide almost immediate toxicity profiles of all test substances. Such chips can provide vast amounts of data on gene expression in response to specific conditions. A single chip can replace the information derived from 20,000 individual experiments. There is evidence that of the mind-boggling number of potential gene expression patterns in the human genome (10 30,000), the number with relevance to toxicologic responses is approximately 317. Once the 256 types of human cells are represented, there only remains another 60 sites of potentially relevant responses to toxic exposures. This is a fairly small number for existing microarray technology. Each microarray includes thousands of tiny pieces of DNA which allow simultaneous examination of overall patterns of gene expression. The goal of toxicogenomics is to determine which of these patterns are associated with each of the classical toxicity endpoints. Once identified, these arrangements could then be used to predict potential toxicity of new substances. It is already known that human genes respond to the presence of a compound and any damage associated with it. Genes also respond in characteristic patterns to changes in levels of metabolically important compounds and the internal environment of the cells in the body. Because of its use of specific gene expression information, microarray-based toxicology would involve a more mechanistic approach to hazard identification and characterization - certainly more relevant to humans than anything currently derived from historically crude animal poison experiments. A recent set of microarray experiments identified a set of twelve diagnostic points that provided 100% predictive accuracy for five different types of toxic substances. It has also been established that gene expression profiles (as on the chips) correlate with results of histopathology, clinical chemistry and known mechanisms of toxicity. Studies are currently underway to apply this technology to the fields of hepatotoxicity (liver), genotoxicity (genes) and nephrotoxicity (kidney). It is also likely that, once fully developed, toxicogenomics will provide the scientific proof that animal-based toxicity testing has little or no relevance to human risk assessment. Use of such chips will become a standard part of any future validation process for in vitro or in vivo safety tests and animal models intended for use in basic biomedical research. Consider the potential consequences of documenting entirely unrelated gene expression profiles for a human disease and its putative animal model surrogate. Microarrays, once validated and widely adopted, should change toxicology into a high-throughput, predictive discipline with unique sets of biomarkers (gene expression patterns) for toxic endpoints and classes of toxicants. This technology is also uniquely suited to interact with existing in vitro methods. For all of these reasons, pharmaceutical and chemical manufacturing companies are investing heavily in the field of toxicogenomics and creation of DNA microarray chips. This is the beginning of a new age of drug and chemical evaluation. Enthusiasm for this new, high-tech approach to toxicology may be premature since the biological relevance of gene expression patterns needs to be established and its predictive abilities validated. Some toxicologists have proposed conducting a limited number of animal toxicity tests in order to create the DNA patterns for the chips. << Back To Contents | Next Section >>

Other NEAVS Fact Sheets: Benefits of Non-Animal Tests | Xenotransplants | Animal Welfare Act | Limitations of Animal Tests | Non-Animal Product Safety Test Alternatives << Back to Better Science  | Home