Autoimmunity and Risk Assessment Michael 1. Luster, Petia P. Simeonova, Randy Gallucci, and Joanna Matheson Toxicology and Molecular Biology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Morgantown, West Virginia USA Among the issues dealing with identifying potential adverse immunologic effects (i.e., suppression, hypersensitivity, or autoimmunity) associated with xenobiotic exposure, general agreement exists among the regulatory and pharmaceutical communities that predictive tests for autoimmunity are in most need of development in order to improve risk assessment. The estimation of risk (i.e., the probability of a deleterious effect resulting from exposure) involves both the qualitative evaluation of whether a hazard exists and the quantitative evaluation for determining an acceptable level of exposure in humans. Unless adequate human data are available, which is uncommon, this is based on animal studies. Although animal models exist to study autoimmune processes, these models do not readily lend themselves to interpretation in the risk assessment process due, for the most part, to the complexity of autoimmune disease(s), as they are multifactorial and exhibit genetic heterogeneity in humans. To improve the risk assessment process, researchers must develop and validate animal models that not only incorporate mechanistic information into the assessment process but also allow for consideration of potent genetic, physiologic, and environmental influences. Key words: autoimmunity tests, immunotoxicology evaluation, immunotoxicology methods, risk assessment, xenobiotic-induced autoimmunity. - Environ Health Perspect 1 07(suppl 5):679-680 (1999). http.//ehpnetl.niehs.nih.gov/docs/1999/suppl-5/679-680luster/abstract.html Overview A study recently conducted by the Environmental Defense Fund (1) indicates that although the number of chemicals tested for immunotoxicity is somewhat less than that for other organ systems, such as the reproduc- tive, developmental, or nervous systems, almost 20% of chemicals present at significant levels in the environment have been examined at some level for immunologic effects. A review of the literature, however, indicated the majority of those chemicals studied were examined for their ability to cause either hypersensitivity or immunosuppression, and only a relatively few were tested for their potential to produce autoimmunity (2). These observations appear at odds with results from a recent survey of the pharmaceutical industry (3). This survey, in which 12 compa- nies responded, suggested that hematologic or dermatologic problems consistent with auto- immunity (or systemic allergy) were the most common preclinical or postlaunch immuno- logic observations. Furthermore, the survey respondents indicated the most immediate need in the area of immunotoxicology evalua- tion was the development of predictive tests for autoimmunity. Epidemiologic data sup- port the survey results; approximately 5% of the U.S. population suffers from some form of autoimmune disease (4), and 5-20% of patients receiving certain drugs such as pro- cainamide or hydralazine develop drug-induced autoimmune diseases (5). The question then arises, Why is there a lack of validated autoimmune models suitable for drug or chemical screening? Certainly a large number of experimental models are used successfully to study organ-specific and sys- temic autoimmunity (Table 1). Furthermore, although many questions still remain on the etiology and biology of autoimmune diseases, scientists have provided a good road map of disease development, allowing for successful application of novel biotech therapies such as cytokine antagonists. The primary reason for the lack of validated screening assays probably stems from the complexity of the disease. First, autoimmune disease is not one disease but a group of over 30 diseases affecting dif- ferent organs, often through different mecha- nisms. Unless a common early process is identified, a single test would be unlikely to provide an adequate degree of concordance to be useful for predictive risk assessment. Second, although most diseases are governed by genetics, the degree of genetic and epige- netic influences in autoimmune diseases is such that it could drastically alter the out- come of a test. For example, the relative risk for developing autoimmunity from gold salt increases 32-fold in individuals who possess the HL-A DR3 allele (7). Experimental studies of mercury-induced autoimmunity in the Brown-Norway rat and B.10 mouse sug- gest the same genetic influences apply in ani- mal models (7). Third, development of autoimmune disease may occur through one of several mechanisms, thus increasing the need to conduct multiple tests in risk assess- ment. For example, at the cellular level, autoimmune disease can occur from either aberrant B-cell or T-cell responses. Thus, if a specific autoimmune disease results from a defect in T cells, assays designed to detect a defect in B cells would provide a false nega- tive. At the molecular level, depending upon the xenobiotic, an autoimmune disease can develop from the expression (unmasking) of cryptic determinants, altered immunoregula- tion such as defects in the expression of inter- leukin 4 or interferon-y, or defects in the establishment of tolerance, which often involve missed deletion or activation of autoreactive T cells. Finally, when using ani- mal models, there is some uncertainty regard- ing what actually constitutes autoimmunity. This is reflected in humans by a lack of well- defined diagnostic tests for identifying autoimmune disease and is discussed in detail in other articles in this monograph. Despite these challenges, attempts to develop predictive screening assays for detecting xenobiotic-induced autoimmunity have proceeded in several institutions includ- ing Virginia Commonwealth University (Richmond, Virginia) under the National Toxicology Program, the University of Utrecht (Utrecht, The Netherlands) funded by the Dutch Organization for Scientific Research, and the University of Dusseldorf (Dusseldorf, Germany). Models proposed to evaluate the potential of xenobiotics to induce autoimmunity are * Popliteal lymph node assay with reporter antigens * Increased titers of autoantibodies * Examination of immunoglobulin complexes/deposits (immunohistology) * Spontaneous animal models. These models are clearly different from those described in Table 1, as the latter are designed to investigate the mechanisms of a specific autoimmune disease. Of the four assays listed, immunohistology, in which immunoglobulin deposits or complexes are evaluated either on suspect organs or in the periphery, has been evaluated the least, whereas the popliteal lymph node assay with reporter antigens has been studied the most (8). Details This article is based on a presentation at the Workshop on Linking Environmental Agents and Autoimmune Diseases held 1-3 September 1998 in Research Triangle Park, North Carolina. Address correspondence to M.l. Luster, Toxicology and Molecular Biology Branch, Health Effects Laboratory Division, NIOSH, 1095 Willowdale Rd., Morgantown, WV 26505. Telephone: (304) 285-6060. Fax: (304) 285-5708. E-mail: myl6@cdc.gov Received 1 1 January 1999; accepted 19 April 1999. Environmental Health Perspectives * Vol 107, Supplement 5 * October 1999 679 LUSTER ET AL. Table 1. Examples of experimental models used to study autoimmune diseases.a Organ-specific autoimmunity Induced by immunization (EAE, AA) Spontaneous mice (NOD,b transgenics) Toxicant-induced (streptozotocin, Cd) Systemic autoimmunity Allogeneic reactions Neonatal thymectomy Spontaneous mice (New Zealand mixed)c Abbreviations: AA, autoimmune arthritis; EAE, experimental autoimmune encephalitis; NOD, nonobese diabetic. 'Adapted from Pelletier et al. 16). bDevelop immune diabetes. CProne to develop systemic lupus erythematosus. of these methods and the results so far obtained have been discussed by Pieters and Albers (9). However, none of these models have under- gone a vigorous validation process to provide sufficient confidence for use in risk assessment. Such a validation process would involve, at the minimum, establishing an acceptable level of concordance and an inter/intralaboratory vali- dation exercise. The former would involve determining the assay's sensitivity and speci- ficity using agents known to induce auto- immune disease in humans. The inter/ intralaboratory validation component is some- what less defined but would examine assay reproducibility, feasibility, accuracy, and in cer- tain instances, cost effectiveness. Data used in risk assessment are derived primarily from animal toxicology studies. When available, findings from epidemiologic and controlled clinical exposure studies take precedent. Results obtained from in vitro studies, structure-activity relationships, (SARs), or mechanistic investigations are normally used only in a supporting role. Mechanistic studies are particularly helpful, however, as without them all defaults in the risk assessment process are assumed valid (e.g., threshold, animal vs human sensitivity, interindividual variability). Epidemiologic data represent the "gold stan- dard" in risk assessment. Epidemiologic studies involving autoimmune diseases would be com- plicated by the likelihood that the disease inci- dence in the exposed population would be relatively low because of the multifactorial and poly-genetic nature of the disease. This would be further confounded by difficulties in clinical diagnosis because of a lack of defined end points available. Some of these problems would be circumvented in controlled clinical studies. However, such studies may be limited with Table 2. Structure-activity relationships of potential interest. Activity Chemical example Estrogenic Diethylstilbestrol Thymolytic Cyclosporin A, cyclophosphamide Formation of protein adducts Halothane Altered immune regulation Mercury, interferon-y Myeloperoxidase substrates Procainamide Altered methylation Hydralazine respect to the size of the population and the length of treatment and could only be conducted with pharmaceuticals. Despite the challenges in developing appropriate tests for autoimmunity that can be used in risk assessment, the considerable amount of data generated by immunologists and pharmacologists pertaining to basic mech- anisms of chemical-induced autoimmune dis- eases have provided a conceptual framework that allows the establishment of potential SARs (Table 2) (6, 10). These SARs are by no means definitive. As the database increases, no doubt some will not be supported and others will be added. In all cases these relationships are supported by basic understanding of immunologic and pharmacologic processes. For example, estrogens are known to be a major factor in classic autoimmune diseases, presumably because of their ability to stimu- late certain components of the immune sys- tem (11). Laboratory studies have shown that thymolytic chemicals can induce autoimmu- nity when given neonatally by altering normal patterns of autoreactive T-cell deletion, a process that occurs in the thymus early in life (12). Chemicals that form protein adducts or damage tissue in such a way to allow expres- sion of cryptic determinants would provide novel host antigens that could be recognized by T cells. Agents that have adjuvant activity or biologicals that stimulate certain cytokines may shift the balance of T-helper 1 and T-helper 2 cells and allow exacerbation of preexisting autoimmune disease (13). Common features associated with many drugs that induce autoimmune diseases are that they serve as myeloperoxidase substrates and/or cause changes in methylation. The underlying biology for the latter associations are less clear but may involve formation of the specific antigenic epitopes responsible for the autoimmune response. With regard to the association with myeloperoxidase substrates, it has been suggested that many of the chemi- cals require metabolism in proximity to immune cells in order to be antigenic; immune cells such as monocytes contain high levels of myeloperoxidase (14). In any case, based upon the need to develop predictive screening models to iden- tify the potential of xenobiotics to induce or exacerbate autoimmunity, combined with our increased understanding of immune and pharmacologic mechanisms and the rapidly increasing array of available animal disease models, suitable tests should be forthcoming. REFERENCES AND NOTES 1. Environmental Defense Fund. Toxic Ignorance. Washington, DC:(1 997). 2. Dean JH, Cornacoff JB, Rosenthal GJ, Luster Ml. Immune sys- tem: evaluation of injury. In: Principles and Methods in Toxicology, 3rd ed (Hayes AW, ed). New York:Raven Press, 1994;1065-1090. 3. Dean JH. Immunotoxicity assessment in the pharmaceutical industry. Toxicol Lett 102/103:247-255 (1998). 4. Jacobson DL, Gange SJ, Rose NR, Graham NMH. Epidemiology and estimated population burden of selected autoimmune dis- eases in the United States. Clin Immunol Immunopathol 84:223-243 (1997). 5. Bigazzi PE. Autoimmunity caused by xenobiotics. Unpublished presentation at the 4th Summer School in Immunotoxicology, 18-20 October 1995, Aix-les-Bains, France. 6. Pelletier L, Ramanathan S, Druet P. Autoimmune models. In: Comprehensive Toxicology, Vol 5 (Lawrence D, ed). New York:Elsevier, 1997;365-380. 7. Miller FW. Genetics of environmentally-associated rheumatic diseases. In: Rheumatic Diseases and the Environment (Kaufman l, Varga J, eds). New York:Chapman & Hall 1999:33-40. 8. Albers R, Broeders A, Van der PijI A, Seinen W, Pieters R. The use of reporter antigens in the popliteal lymph node assay to assess immunomodulation by chemicals. Toxicol AppI Pharmacol 143:102-109 (1997). 9. Pieters R, Albers R. Screening tests for autoimmune-related immunotoxicity. Environ Health Perspect 107(suppl 5):673-677 (1999). 10. Griem P, Wulferink M, Sachs B, Gonzalez JB, Gleichmann E. Allergic and autoimmune reactions to xenobiotics: how do they arise? Immunol Today 19:133-141 (1998). 11. Homo-Delarche, F, Fitzpatrick, F, Christeff, N, Nunez EA, Bach, JF, Dardenne, M. Sex steroids, glucocorticoids, stress and autoimmunity. J Steroid Biochem Molec Biol 40:619-637 (1991). 12. Sakaguchi S, Sakaguchi N. Organ-specific autoimmune disease induced in mice by elimination of T cell subsets. V: Neonatal administration of cyclosporin A causes autoimmune disease. J Immunol 142:471-480(1989). 13. Chazerain P, Meyer 0, Kahn MF. Rheumatoid arthritis-like dis- ease after cu-interferon therapy. Ann Intern Med 116:427-439 (1992). 14. Greim P, Gleichmann E, Shaw CF. Chemically-induced allergy and autoimmunity: what do T cells react against? In: Comprehensive Toxicology, Vol 5 (Lawrence D, ed). New York:Elsevier, 1997;324-338. 680 Environmental Health Perspectives * Vol 107, Supplement 5 a October 1999