Epidemiologic studies of environmental agents and systemic autoimmune diseases.

Systemic lupus erythematosus and systemic scleroderma are autoimmune diseases thought to have an exogenous trigger. This review summarizes relevant case-control and cohort studies that investigated exogenous sex hormones, silica, silicone, solvents, pesticides, mercuric chloride, and hair dyes as putative risk factors for the development of these diseases. These studies indicate that estrogen replacement therapy in postmenopausal women increases the risk of developing lupus, scleroderma, and Raynaud disease, although the increase in risk is relatively modest. Oral contraceptives may also play a role in disease susceptibility in lupus but not apparently in scleroderma. Environmental endocrine modulators, in the form of pesticides, may represent another opportunity for estrogenlike effects to occur, but there is scant evidence that these agents play a role in human systemic autoimmune disease. Although exposure to silica dust increases the risk of scleroderma in men occupied in the industry, this does not explain most male scleroderma cases. When this exposure was investigated among women, no significant risk was found. Additionally, silicone in implanted devices as well as occupational exposure to silicone-containing compounds did not pose an increased risk among women for scleroderma. The role of solvent exposure has been investigated as a risk factor for scleroderma with mixed findings. One study suggested a potential role in male patients or in those individuals with Scl-70 antibody positivity either male or female. Two other studies were unable to corroborate this finding. Mercuric chloride causes antifibrillarin antibodies and immune complex glomerulonephritis in susceptible mouse strains. Antifibrillarin antibodies, but not glomerulonephritis, occur in a subset of scleroderma patients and preliminary evidence suggests that mercury levels may be higher in this group of individuals. Hair products have been studied as possibly raising the risk of developing lupus, since such products contain an aromatic amine similar to a compound known to cause drug-induced lupus. A 1986 study suggested a positive association, but two subsequent studies did not support this association.

Systemic lupus erythematosus (SLE, lupus) and systemic sclerosis (sderoderma) are prototypical systemic autoimmune diseases affecting multiple organ systems and characterized by the presence of autoantibodies. For the purpose of this article, the term lupus will be used to refer to the systemic form of the disease and the term discoid lupus will refer to the isolated skin disease. Similarly, the term scleroderma will be used to refer to systemic sclerosis, not to morphea or localized scleroderma that is confined to the skin. These diseases are considered autoimmune because a)tissue damage is mediated by immune cells and immune cell products; b) antibodies that recognize elements of host tissue are produced on a chronic basis; and c) there is no identified external stimulus against which an immune response can be considered appropriately directed. It has long been postulated that such an external stimulus exists and provides the trigger or initiating event. Persistence of disease, and perhaps its severity as well, is dependent on a complex interplay between host tissue and immune responses presumably based on genetic features. Additionally, both lupus and scleroderma have a female preponderance, with an overall female-to-male ratio of approximately 4:1 (1,2), suggesting that sex hormones or pregnancy-related events influence disease susceptibility.
Thus, both genetic factors and sex hormones are thought to play important roles in disease risk and these features account for the occasional familial clustering and the strong female preponderance. However, these two factors together make up only a small portion of the risk, suggesting that some environmental or acquired exposure(s) is necessary for disease expression.
Interest in the possibility that environmental exposures are important in the development of scleroderma comes from compelling evidence that several pseudo-sderodermatous conditions are triggered by the ingestion of, or exposure to, certain chemicals. For example, ingestion of contaminated rape seed oil resulted in the toxic oil syndrome in Spain in 1981 (3), and a contaminated tryptophan/protein powder nutritional supplement was associated with the eosiniophilia myalgia syndrome in the United States in the late 1980s (4). Additionally, vinyl chloride disease occurred in heavily exposed workers in that industry until the syndrome was recognized and changes made to reduce the level of exposure (5). All these conditions, however, differ in many clinical features from idiopathic scieroderma, hence the term pseudoscleroderma or sderodermalike conditions, and provide limited insight into the pathogenesis of systemic sclerosis.
Similarly in the case of lupus, a syndrome of drug-induced lupus is well known and associated with several widely used medications. In the drug-induced form of the disease, signs and symptoms are similar to, if not identical with, the idiopathic disease, but the drug-induced form is usually milder and symptoms remit on drug cessation.
The central issue for this review is what role, if any, environmental exposure play in idiopathic disease expression. Because of the limited value of case reports and case series, this review will focus on case-control and cohort studies. Such studies include investigations of exogenous sex hormones, silica, silicone, solvents, mercuric chloride, and hair dyes as potential risk factors for the development of these diseases.

Exogenous Sex Hormones and the Risk of Developing Systemic Lupus
Because both lupus and scleroderma have a strong female preponderance, several studies have examined the role of female hormones in disease susceptibility. Table 1 summarizes the principal findings of these studies.
Sanchez-Guerrero et al. (6) investigated the association between postmenopausal estrogen replacement therapy and the risk of developing systemic lupus using data from the Nurses' Health Study cohort, an established cohort of women begun in 1976. These authors reported an age-adjusted relative risk (RR) for developing lupus in this group of 2.1 (95% confidence interval [CI], 1.1-4.0) for ever users of estrogen and a relative risk of 2.5 (95% CI, 2.5-5.0) for current users. Additionally, risk increased with duration of use. This analysis was based on the finding of 40 cases of definite SLE [meeting the  (8), who examined the association of postmenopausal estrogen use with the development of both systemic lupus and discoid lupus, a form of lupus confined to the skin and sometimes seen as a precursor of, or concomitant condition with, the systemic form. These investigators conducted a case-control analysis based on the General Practice Research Database of the United Kingdom and the resultant series consisted of 75 cases (42 cases of SLE and 34 cases of discoid lupus) and 295 controls. The overall adjusted RR for current users versus past and nonusers was not elevated over that of controls. However, analysis of long-term users, as defined by 25 or more months of use, showed an adjusted RR estimate of 2.8 (95% CI, 1.3-5.8) for the development of SLE and discoid lupus combined. The RR in this study was adjusted for body mass index, hysterectomy, oophorectomy, and smoking status. Therapy with estrogen alone (unopposed estrogen) for greater than 24 months had an adjusted RR of 5.3 (95% CI, 1.5-18.6), whereas the combination of estrogen plus progestogen had an insignificantly elevated RR of 2.0 (95% CI, 0.8-5.0). Participants in this study were almost exclusively white; these results should be applied with caution to a nonwhite population.
These studies found no significant association of smoking status, body mass index (6,7), hysterectomy, and oophorectomy (7) with the outcome of SLE or discoid lupus.
In terms of oral contraceptive (OC) use and the risk of developing SLE, Sanchez-Guerrero et al. (9) reported a study analogous to the hormone replacement study discussed above, using the Nurses' Health Study cohort. These authors found 99 cases of definite SLE and 88 cases of more stringently defined SLE (cases meeting the American College of Rheumatology criteria plus having a disease diagnosis by a rheumatologist) among 121,645 women in the cohort. They reported an RR of definite SLE of 1.4 (95% CI, 0.9-2.1) for past users of OCs. Using the more stringent definition of SLE, the RR was 1.9 (95% CI, 1.1-3.3). The RR estimates were adjusted for age and postmenopausal hormone use. No significant increase in risk was seen with increased duration of OC use and no significant trend was observed according to the time since first use. Because the cohort was 95% white, effects of race could not be studied and these findings may not apply to other ethnic groups.
A smaller case-control study reported by Strom et al. (10) of 195 SLE patients and 143 friend controls failed to find an association between SLE and either any use or recent use of oral contraceptives (odds ratio [OR] = 0.6; 95% CI, 0.2-1.4).
These two studies taken together suggest that the increased risk of developing SLE with oral contraceptive use is relatively small and the effect weak.

Sex Hormones and the Risk of Scieroderma
Beebe et al. (11) studied the association between reproductive history, OC history, and estrogen replacement use in 472 female scleroderma patients and 2,227 female controls. These authors found that OC use did not influence the risk of developing scleroderma (age and year of birth adjusted odds ratio [aOR] = 0.94 ;95% CI, 0.74-1.22). Additionally, earlier age of menarche (aOR =1.00; 95% CI, 0.79-1.22) or ever being pregnant (aOR = 0.86; 95% CI, 0.64-1.15) were not associated with scleroderma.
However, estrogen replacement therapy was associated with a small but significant increased risk of disease development with an aOR of 1.40 (95% CI, 1.10-1.77). Average age at first use was not different between cases and controls, suggesting that younger age at menopause for the cases was not responsible for this result.
Raynaud phenomenon is a reversible vasospastic condition that causes transient ischemia of the digits on cold exposure and sometimes also with emotional stress. It is most often seen as a primary condition in the absence of a connective tissue disease and affects approximately 5% of the adult American population. The autoimmune disease with which it is most frequently associated is scleroderma, in which Raynaud phenomenon affects 95% of cases. Primary Raynaud disease is seen more frequently in women than in men. Fraenkel et al. (12) studied the association of estrogen replacement therapy and Raynaud phenomenon in 497 postmenopausal women in the Framingham Offspring Study. The prevalence of Raynaud phenomenon was 8.4% in women not taking hormone replacement therapy, 19.1% in those receiving estrogen alone, and 9.8% in those receiving estrogen plus progesterone. The aOR (adjusted for age, body mass index, use of alcohol, cigarettes, and 0-adrenoreceptor antagonists) was 2.5 (95% CI, 1.2-5.3) with estrogen alone and 0.9 (95% CI, 0.3-2.6) with estrogen plus progesterone.
In summary, estrogen replacement therapy in postmenopausal women may increase the risk of developing lupus, scleroderma, and Raynaud disease, although the increase in risk is relatively modest. Evidence to date indicates that the addition of a progestogen to estrogen may serve to ameliorate this risk.
OCs may play a role in disease susceptibility in lupus, but evidence to date suggests this is not the case for sderoderma. It should be mentioned that these effects refer to the use of medications before disease onset and do not necessarily imply that the drugs would increase the severity ofestablished disease.
The discussion above has concentrated on the various medical uses for exogenous estrogens as they affect the risk of autoimmune disease. There are potential environmental endocrine modulators in the form of pesticides that could also influence the risk for these diseases. Certain pesticides are known to be endocrine disrupters including polychlorinated biphenyls (PCBs), DDT metabolites, and 2,3,7,8-tetrachlorodibenzo~p-dioxin (TCDD). PCBs exert antiestrogen activity by inhibiting estrogenmediated signal transduction (13,14). In an analogous fashion, the longlasting metabolite of DDT (p,p-DDE) can bind to the androgen receptor and disrupt hormonal signaling (13). TCDD binds to the aryl hydrocarbon receptor, which also Environmental Health Perspectives * Vol 107, Supplement 5 * October 1999 plays a role in hormonal signaling (15). Chronic and acute exposures of mice to TCDD have suppressed total hemolytic complement and third component of complement levels, impairing host defenses to bacterial pathogens (16).
At face value, this type of immunosuppression might be considered beneficial for autoimmune disease. However, complement is needed for efficient clearing of immune complexes from the circulation and individuals with particular complement deficiencies are at a higher risk for developing lupus. Although much work has been done on the teratogenicity and immunosuppressive effects of these pesticides, little work has been done to study potential associations between these compounds and the risk for human auto-immune disease.
There is a case report of scleroderma occurring following exposure to an herbicide combination of bromobutyl methyl uracil, dichlorophenyl dimethylurea, and aminotriazole (17). Aminotriazole can cause a contact dermatitis (18), but none of these compounds have been previously associated with autoimmune disease. A case-control study of 472 female scleroderma cases and 2,227 controls found an increased risk of scleroderma in those exposed (per self-report) to herbicides and pesticides (aOR, 2.19; 95% CI, 1.16-4.15) but did not distinguish between these two exposures (19).

Silica and Silicone
Occupational exposure to silica in the form of particulate silica dust has been associated with both the development of lupus (20) and the development of scleroderma [for review see (21)]. Such occupations include mining, sandblasting, foundry work, and grinding of certain materials. Because these occupations usually involve male workers, almost all reported cases have been male. Burns et al. (22) investigated the association between such silica exposure in women and the development of scleroderma in a case-control study involving 274 female scleroderma cases and 1,184 female controls in Michigan. There were 12 cases exposed compared to 36 controls, with an adjusted OR of 1.50 (95% CI, 0.76-2.93). Thus, although the numbers of exposed individuals are small, these data do not support an important role for silica exposure in scleroderma disease development in women.
Silman and Jones (23) found no cases of silica exposure (by self-report or by expert review as probable) among 56 male scleroderma cases in the United Kingdom.
Several epidemiologic studies have failed to find a significant association between silicone breast implants and the development of scleroderma (22,(24)(25)(26)(27). One of these studies (22) also evaluated the risk of any implanted medical device, including pacemakers, joint prostheses, and indwelling catheters/shunts, many of which also include silicone products as part of the device. No association with scleroderma was found for these devices. Additional sources of silicone exposure through occupations and hobbies were examined by these investigators, but, again, no association was demonstrated. Silica and silicone exposure is discussed in greater detail in another artide in this monograph (28).
However, when these associations were tested in case-control studies, variable results were obtained. Nietert et al. (47) reported a study involving 178 scleroderma patients and 200 controls from whom occupational histories were obtained. To assess solvent exposure, exposure scores were computed using job-exposure matrices based on a standard code. Maximum intensity of exposure and cumulative exposure were calculated. A broad category of any solvent was used as well as four specific solvents: trichloroethylene (TCE), trichloroethane (TCA), carbon tetrachloride (CCl4), and benzene. Among men (n = 37), those with scleroderma were more likely than controls to have a high cumulative intensity score (OR = 2.9; 95% CI, 1.1-7.6) for any solvent exposure. Men were also more likely to have a high maximum intensity score for TCE (but not other solvent) exposure (OR 3.3; 95% CI, 1.0-10.3). This effect was not seen for the women (n = 141). However, if the analysis was confined to those male and female cases who had the Scl-70 autoantibody (a marker for more severe disease), significant solvent-disease associations were observed.
Somewhat at variance with these results is the report by Lacey et al. (48) of a casecontrol study of 472 female scleroderma patients and 2,227 controls in Ohio and Michigan, which reported a significantly increased risk of scleroderma in those with self-reported occupational activities of professional cleaning or maintenance (aOR =2.18; 95% CI, 1. 48-3.22). These activities are more likely to provide exposure to waterbased solutions, than organic solvents. These investigators did not find a significant association with dry cleaning (aOR = 1.46; 95% CI, 0.91-2.34) or with TCE exposure (aOR = 2.29; 95%, CI 0.92-5.71). Paint thinners and removers were significantly associated with disease development (aOR = 1.82; 95% CI, 1.32-2.51). These last solvents are mixtures of hydrocarbons of varying molecular weights but with properties distinct from the chlorinated hydrocarbon solvents TCE, TCA, or CC14.
Silman and Jones (23) reported an occupational analysis of 56 men with scleroderma in the United Kingdom compared to 97 controls (56 age-and geographic areamatched patients without connective tissue disease from general practitioners and 41 friend controls). Job histories and detailed descriptions and information regarding the specific agents vinyl chloride, polyvinyl chloride, silica, organic solvents, epoxy resins, and formaldehyde were obtained. Analysis was performed based on self-report as well as expert review. No significantly elevated ORs France 'This table is limited to reports of systemic sclerosis, not localized scleroderma and not sclerodermalike disorders. bVarious aromatic organic solvents. cSolvents included trichloroethane, xylene, trimethylbenzene, and naphthalene. dlerbicide contained bromacil or bromobutyl methyl uracil, diuron or dichlorophenyl dimethylurea, and aminotriazole. 'Not well characterized. were calculated for any category of exposure. Exposure to organic solvents was stratified on duration of use but no significant trend of increasing OR was found compared with either control group, with most ORs centering around unity. Table 3 is a summary of these results.
A study by Kilburn and Warshaw (49) reported an increase in the prevalence of some SLE symptoms as well as an increase in antinudear bodies in a group of individuals in Tucson, Arizona, exposed to water contaminated with TCE, TCA, inorganic chromium, and other chemicals. This report is difficult to interpret because of concerns regarding exposure misclassification and case and control selection.
The role that solvents may play in disease initiation is unclear. Data from animal studies demonstrate that some solvents and/or their metabolites can bind to nucleic acids and proteins (50,51) and can reduce both the humoral and cell-mediated response (52)(53)(54). Additionally, exposure to TCE and one of its metabolites (dichloroacetyl chloride) induced autoantibody formation in genetically susceptible autoimmune-prone MRL +/+ mice (55). Byers et al. (56) studied a human population exposed to TCE in the water supply and found elevated levels ofT lymphocytes as well as elevated levels of CD4+ and CD8+ cells, suggesting that this agent or its metabolites could alter human immunity. This study did not address the issue of autoimmune disease in this population. The role of such solvents in inducing or accelerating scleroderma or lupus remains unclear, as noted in the discussion above of the case-control studies.

Mercuric Chloride
Mercuric chloride is an immunomodulating agent that causes immune complex glomerulonephritis and autoantibody production in susceptible mouse strains (57,58). The autoantibody produced recognizes a nudeolar protein, fibrillarin (59). Evidence from Arnette et al. (60) suggests that mercury modifies fibrillarin leading to autoantigenicity, which then induces a persistent self-reaction to the native or unmodified fibrillarin. Antifibrillarin antibodies are found in only 8% of scleroderma patients, and, although relatively infrequent in occurrence, are associated with more severe disease.
Urinary mercury excretion has been reported to be higher in sderoderma patients who are positive for antifibrillarin antibodies than in antifibrillarin-negative patients or normal controls (61). These levels, however, were still in the normal or unexposed range, as would be expected in individuals who showed no signs or symptoms of mercury toxicity. Also, neither scleroderma patients in general nor antifibrillarin antibody-positive scleroderma patients develop immune complex glomerulonephritis.

Hair Products
Drugs that contain aromatic amines can induce an SLE-like syndrome [for review see Price and Venables (62)]. Some hair products, including hair dyes, straighteners, and permanents, contain the aromatic amine hydrazine, some of which can be absorbed through the scalp. This observation prompted the investigation of environmental exposures to similar compounds and the risk of developing lupus. The use of hair dyes was initially reported in 1986 to increase the risk for SLE (63), but this finding has not been confirmed in two larger and more recent studies. In the 1986 study, Freni-Titulaer et al. (63) reported a positive association between hair product use and connective tissue disease in southeastern Georgia. However, in 1992 Petri and Allbritton (64)  reported their study of hair product use and the risk of development of lupus as well as the risk of having more severe disease. The patient group consisted of 218 members of the Hopkins Lupus Cohort; controls were sexmatched and age-similar relatives (n = 178) or friends (n = 186). The study reported no significant difference in exposure to hair dye, hair straightener, or hair permanent, and no significant differences in measures of SLE disease activity in patients who used these hair products after their diagnosis versus those who did not use the products.

Summary
There is a relative paucity of case-control or cohort studies investigating environmental exposures and the risk of developing the systemic autoimmune diseases lupus and scleroderma, in large part because these diseases are relatively rare. In total, the current state of the risk factor analysis suggests that no currently known exposure or combination of exposures explains an important proportion of the occurrence of either lupus or scleroderma. The cause of these chronic diseases is unknown and the situation is douded by the fact that there could be multiple triggers for these diseases.
Current evidence supports the finding that estrogen replacement therapy increases the risk (approximately 1.5to 3-fold) of developing systemic and discoid lupus, scleroderma, and Raynaud disease. The use of estrogen-progestogen combination therapy may ameliorate this risk. Oral contraceptives may double the risk for SLE, but there is lack of evidence to support this conclusion for scleroderma.
Silica exposure does not explain most male cases of scleroderma and does not appear to play a significant role in scleroderma risk in women. The role of solvent exposure, particularly in raising the risk for scleroderma in men who have a higher prevalence of such occupations, remains unclear. Hair products do not appear to raise the risk of SLE, contrary to earlier reports.

Mechanisms
Potential mechanisms of induction of autoimmunity for these diverse environmental agents can be classified into three broad categories: a) a change in the hormonal Environmental Health Perspectives * Vol 107, Supplement 5 * October 1999 M.D. MAYES Abbreviations: Cl, confidence interval; RR, relative risk. 'Based on 178 scleroderma cases (37 men and 141 women) and 200 controls 160 men and 138 women); 19 women and 10 men were Sc1-70 positive (47). hBased on 472 female scleroderma cases and 2,227 female controls 147). cBased on 56 male cases and 97 male controls [56 general practitioner controls and 41 friend controls (23)1. milieu to favor estrogenic stimulation of the immune system; b) suppression of one section of the immune system (such as complement levels), which disrupts normal immune surveillance; and c) chemical binding to a selfantigen forming a neoantigen, thus breaking tolerance by inducing immunity to the unmodified native molecule as well as to the modified antigen.
These mechanisms are currently speculative at best but are supported by animal models. It should be noted that the induction of autoantibodies does not necessarily result in the expression of autoimmune disease. Many of the autoantibodies are considered markers of disease with no known role in tissue damage. Several family studies have found that apparendy healthy family members and even spouses of lupus and scleroderma patients may have these autoantibodies but seldom express disease. This may reflect a shared environmental exposure but the lack of the specific genes required for the "correct" immune response.

Opportunities for Future Research
This review is relatively brief because the number of well-designed epidemiologic studies in this field is small. The existence of large population cohorts as well as an established patient cohort for scleroderma (including the Scleroderma Registry, Wayne State University, Division of Rheumatology, Detroit, Michigan) should facilitate this research. Other potential exposures such as herbicides, pesticides, and mercury, as well as exposures to infectious agents, need to be investigated. More detailed studies of occupational and environmental exposures are clearly warranted by preliminary data and are currently feasible given the interest of the scientific community and availability of large patient cohorts.