Division of Endocrinology, Metabolism, and Nutrition, Mayo Clinic, Rochester, Minnesota 55905
Address all correspondence and requests for reprints to: Dr. Rebecca Bahn, Division of Endocrinology, Metabolism, and Nutrition, Mayo Clinic, Rochester, Minnesota 55905. E-mail: bahn.rebecca{at}mayo.edu.
Hyperthyroidism in Graves disease is due to the binding of stimulatory autoantibodies to the TSH receptor (TSHr) on thyroid follicular cells. The stimulation of this G protein-coupled receptor by autoantibodies leads to excessive and uncontrolled production of thyroid hormone. Our understanding of Graves disease has increased remarkably after the cloning in 1989 of TSHr (1, 2, 3). This achievement led to insights regarding TSHr biology, including the unique two-subunit structure of this receptor that may render it especially prone to autoimmune attack (4). In contrast, the pathophysiology of Graves ophthalmopathy (GO) and thyroid-associated pretibial dermopathy (PTD) is less well understood. However, studies by several groups have begun to unravel the many complex factors contributing to development of these ocular and dermal manifestations of Graves disease.
The progression of GO and PTD from initiation to subclinical disease to fully developed ocular and dermal manifestations appears not to be a linear process. It seems rather to be a positive feedback cycle composed of mechanical, immunological, and cellular processes (Fig. 1). In this review each of the major components of this disease cycle will be examined with an eye toward understanding the limitations of current therapy and identifying targets for future therapy.
|
Many of the clinical symptoms and signs of GO can be explained on a mechanical basis by the increase in volume of intraorbital tissues characteristic of the disease (5). Although most individuals with GO have evidence of both extraocular muscle and orbital adipose tissue enlargement, some exhibit a predominance of either muscle or fat expansion (6). Individuals younger than 40 yr of age are considerably more likely to exhibit orbital fat expansion in the absence of muscle enlargement, whereas patients over 70 yr are more prone to severe, fusiform muscle enlargement without significant changes in orbital adipose tissue volume (7). Proptosis, the forward displacement of the globe, stems from this increase in orbital tissue volume within the unyielding confines of the bony orbit. Extraocular muscle dysfunction is caused, early in the disease, by swelling of the muscle bodies. Enlargement of the muscles at the apex of the orbit may lead to compressive optic neuropathy and visual loss. In later stages the extraocular muscles may become fibrotic and atrophic as a result of chronic inflammation and compression of the muscle fibers. Chemosis and periorbital edema appear to be caused primarily by decreased venous and lymphatic drainage from the orbit secondary to compression of these channels. Similarly, patients with severe PTD have compromise of low pressure lymphatics in the lower extremities (8), which probably contributes to the dependent edema seen in this condition.
Histological examination of orbital adipose and extraocular muscle tissues in GO reveals an excess of complex carbohydrates called glycosaminoglycans (GAG), in which hyaluronan predominates (9). This large, hydrophilic, polyanionic compound is composed of repeating disaccharides. One molecule of hyaluronan occupies a volume of 330,000 x 10-19 ml, or 75,000 times greater than an equivalent weight of collagen. Gross enlargement of the muscle bodies in GO is caused primarily by accumulation of GAG and edema within the connective tissues investing and separating the muscle fibers. A similar accumulation of GAG is apparent within the fatty connective tissues of the orbit. The overall increase in orbital fat volume appears to be attributable to both hydrated hyaluron and de novo adipogenesis in these tissues.
The expansion of fat and muscle tissues within the bony confines of the orbit leads to increased intraorbital pressure. This results in compression of orbital contents with traumatic injury to these soft tissues. The pressure may be relieved by forward protrusion of the globe. In this sense, proptosis could be considered to represent a natural orbital decompression. The degree of decompression is limited by the forward mobility of the globe, which is tethered in place by the rectus muscles and canthal tendons of the lid. Indeed, GO patients with minimal proptosis, but especially large extraocular muscles or orbital fat volume, are at particular risk for the development of compressive optic neuropathy.
Mechanical factors and trauma to soft tissues may likewise be involved in the pathogenesis of PTD. Rapoport and colleagues (10) reported the case of a Graves patient with severe, elephantiasic PTD who suffered worsening coincident with increased periods of prolonged standing. Excision of a part of the lesion with subsequent grafting of skin from the patients uninvolved thigh resulted in recurrence of the disease at the original site as well as development of dermopathy at the donor site. The inferior limit of the PTD was just above the margin of this patients shoes, with relative sparing of the feet. The researchers hypothesized that dependent edema with slower return of lymphatic fluid to the circulation might reduce the clearance and prolong the half-life of disease-related cytokines or chemokines within the affected tissues. This pooling of immune mediators would be expected to increase their local disease-producing effects.
Unique anatomical features of the orbit and lower extremities appear to contribute to their prominent involvement in the peripheral manifestations Graves disease. The unyielding confines of the bony orbit predisposes to compression of orbital low pressure lymphatic and venous channels. Similarly, prolonged standing with dependent edema contributes to the compromise of channels in the lower extremities. Moreover, perhaps individual anatomical variability, such as the shape of the orbits or variations in venous or lymphatic vessels, may place some individuals with Graves disease at special risk for the development of severe GO or PTD.
Trauma itself may act as a stimulus for the initiation of an autoimmune reaction. Indeed, it has long been known that GO patients who experience significant trauma to the skin in any region, whether accidental or due to surgery, may develop the characteristic dermal changes of PTD at that site (9). Recent studies suggest that antigen-presenting cells may be nonspecifically activated after injury to a nearby cell (11). The insult may be traumatic, delivered by a pathogen or toxin, or result from a defect in the nearby cells programming. The injured cell sends activating signals to local antigen-presenting cells, which, in turn, take up antigens and up-regulate costimulatory molecules. This may lead to activation of T cells and initiation of an antigen-specific response. Thus, it appears that tissue injury may initiate an autoimmune response in the absence of foreignness. In this case, the autoimmune process is a result, rather than a cause, of the disease or condition.
Orbital decompression surgery is a very efficacious, albeit invasive, treatment for severe GO. By removing a portion of the bony wall, this surgery increases the orbital volume available to contain the enlarged tissues. As a result, the orbital contents are returned to near-original positions, and intraorbital pressure is reduced. This allows for improved venous drainage with decreased local edema and alleviation of soft tissue trauma. In a study by Garrity et al. (12), in which 428 consecutive patients underwent transantral orbital decompression surgery at the Mayo Clinic, proptosis was reduced by a mean of 4.7 mm, and visual acuity, papilledema, visual scotomas, and exposure keratitis improved in almost 90% of the patients. The surgery was performed for treatment of optic neuropathy in 51% of patients and for relief of severe orbital congestion with inflammation in 27%. Periorbital swelling and conjunctival erythema were alleviated rapidly by the surgery, underscoring the important role played by mechanical venous and lymphatic obstruction in development of the clinical disease. In contrast, the excision of skin affected by PTD is deleterious, probably augmenting the tissue trauma associated with the condition.
Immunological contributions
A diffuse infiltration of lymphocytes, with occasional nests of lymphoid aggregates, is seen within the orbital adipose tissues of patients with GO. A similar, if more sparse, cellular infiltrate is present in the interstitial tissues of the extraocular muscles. The majority of the cells are T lymphocytes, with occasional B lymphocytes seen. Both helper/inducer (CD4+) and suppressor/cytotoxic (CD8+) T lymphocytes are present, with a slight predominance of the latter (13).
To better understand immune mechanisms operative within the orbit in GO, several groups of investigators have profiled the cytokines secreted by tissue-infiltrating T cells. Two groups reported that the majority of retroocular T cell clones produce cytokines involved in cell-mediated T helper cell type 1 (Th1)-type immune processes, namely IL-2, interferon- (IFN
), and TNF
, but not IL-4 or IL-5 (14, 15). A third group detected mRNA encoding a T helper cell type 2 (Th2)-dominant profile (IL-4, IL-5, and IL-10) characteristic of humoral responses (16), whereas another group of investigators identified clones secreting cytokines characteristic of both subtypes (17). These and other related studies suggest that T cells representing both subtypes are represented in the retroocular infiltrates in GO. It appears that Th1 cells may predominate in early disease, whereas Th2 cells may become predominant late in the course of the disease (18).
Besides participating in T and B cell responses, some Th1 and Th2 cytokines contribute to disease propagation by inducing classical immunomodulatory proteins in orbital and pretibial fibroblasts (19). Other participating inflammatory mediators and chemokines are elaborated by resident macrophages and fibroblasts. These factors, including IL-1, IL-6, IL-8, IL-16, TGFß, RANTES, and prostaglandin E2 (PGE2), trigger T cell migration across activated microvascular endothelium or participate directly in local inflammation (20, 21).
GO is clinically apparent in 2550% of individuals with Graves hyperthyroidism. However, characteristic changes in orbital soft tissue anatomy are detectable using sensitive imaging techniques in the vast majority of hyperthyroid patients (22). Conversely, although 10% of patients with GO are euthyroid, the majority of these patients have laboratory evidence of thyroid autoimmune disease, including antibodies directed against thyroid antigens such as TSHr or thyroid peroxidase (23). Similarly, although only 4% of individuals with Graves hyperthyroidism have clinically apparent PTD, evidence of subclinical involvement of the pretibial skin is demonstrable by dermal ultrasonography in 76% of patients (24). Whether clinical GO or hyperthyroidism occurs first, the other manifestation occurs in 85% of patients within 18 months (25). Likewise, although PTD is rarely diagnosed in patients without GO, it is likely that these cases represent remote, undiagnosed, or subtle ocular disease (26). Finally, although Graves disease is a multigenic condition that develops as a result of many interactions between relatively weak susceptibility genes and environmental triggers (27), no unique susceptibility genes have been identified in Graves patients with clinical GO or PTD (28).
These clinical and genetic observations support the concept that GO, PTD, and Graves hyperthyroidism are manifestations of the same autoimmune disease. It follows, therefore, that all components of the clinical triad may stem from an immune reaction directed against the same or a similar autoantigen. As the target in Graves hyperthyroidism is TSHr in thyroid follicular cells, this protein is a logical candidate to be similarly targeted in the orbit and skin. Such cross-reactivity might be a T cell function involving recognition of antigenic TSHr peptide fragments, processed by antigen-presenting cells within the thyroid, orbit, and skin. Alternately, the cross-reactivity could be primarily at the B cell level, with cross-reactive antibodies recognizing intact cell surface antigen expressed in these tissues. However, in contrast to the well documented occurrence of neonatal thyrotoxicosis due to transplacental transfer of TSHr, the lack of any convincing reports of neonatal GO argues against a direct role for autoantibodies in the pathogenesis of this disease.
A prerequisite for involvement of TSHr as an antigen in GO and PTD is that it be present in the orbit and skin. Studies aimed at identifying TSHr in extrathyroidal tissues have been performed by several laboratories using reverse transcriptase PCR (29, 30, 31), ribonuclease protection assays for semiquantitative detection of this low abundance mRNA (32), Northern blotting (33), or immunohistochemistry (10, 31). The results of these studies were in general agreement and showed detectable TSHr mRNA and protein in GO orbital adipose/connective tissue specimens and derivative cultures. Low abundance TSHr was also apparent in normal orbital fatty connective tissue samples and cultures. However, relative TSHr mRNA levels were shown in one study to be greater in orbital adipose tissues from GO patients than in similar tissue samples derived from normal individuals (32). Similarly, TSHr protein may be more abundant in PTD skin than in normal pretibial skin (10). Other studies have shown low abundance mRNA transcripts or protein in thymus, kidney, pituitary, and other organs not clinically involved in Graves disease (34, 35).
Both normal and GO orbital fibroblasts increase the expression of functional TSHr, as measured in a TSH-dependent cAMP assay, when cultured under conditions shown to induce adipocyte differentiation (36). Treatment of these cultures with IL-6 enhances TSHr expression (37). This cytokine, produced by thyrocytes and orbital fibroblasts (38), is elevated in the sera of patients with Graves disease (39) and may serve as an important mediator of TSHr expression in this setting. Other cytokines, including IFN, TNF
, and TGFß, inhibit TSHr expression in orbital fibroblasts (40), whereas purified immunoglobulins from GO patients with elevated thyroid-stimulating immunoglobulin titers seem to have no such effect.
Several studies have examined correlations between the level of TSHr autoantibodies and the severity of GO. Although the results of these studies are conflicting, the most carefully designed found both thyroid-stimulating immunoglobulins and thyroid binding inhibitory immunoglobulins to be closely correlated with GO clinical activity (41). A weaker, but significant, correlation was noted as well between autoantibody levels and proptosis. Although these results support the concept that TSHr may be an important autoantigen in GO, they do not necessary indicate TSHr autoantibodies in pathogenesis. It appears from available experimental data that cellular immunity with local secretion of cytokines, rather than any direct effect of humoral immunity, plays the more important role in pathogenesis. The high titer autoantibodies measured in patients with severe GO may reflect the intensity of the TSHr-directed immune response.
Several other candidate autoantigens, including the muscle proteins G2s and D1, as well as a 64-kDa eye muscle membrane antigen have been studied as potential autoantigens in GO (42, 43). Antibodies directed against these muscle proteins are present in the sera of some GO patients, but are also detectable in normal sera. Most investigators agree that extraocular muscle antigens are probably not involved in the initiation of GO as primary autoimmune targets. Rather, it appears that these antibodies are produced secondary to the disease process, reflecting the release of sequestered cytoskeletal proteins from damaged eye muscles. Recently, immunoreactivity for intact thyroglobulin was detected in fibroadipose orbital tissue samples from GO patients (44). This finding echoes the interesting earlier hypothesis that thyroglobulin might travel from the thyroid to the orbits via cervical lymphatic channels and might function there as an autoantigen in GO (45).
Direct evidence that any candidate antigen or cell type is an autoimmune target in GO has been difficult to obtain, largely due to difficulties in obtaining affected orbital tissues and autologous lymphocytes. However, several studies have examined reactivity of peripheral or orbital T cell lines against various purified antigens or orbital tissue preparations. These T cell lines recognized autologous cultures of orbital fibroblasts in an MHC class I-restricted manner and also proliferated in response to thyroid membrane preparations and purified TSHr antigen (46, 47). No response to crude eye muscle extract, autologous peripheral blood mononuclear cells, allogeneic fibroblasts, or purified protein derivative of mycobacterium tuberculosis was noted. Similarly, low molecular weight protein fractions derived from extraocular or abdominal muscles and particular candidate antigens, including recombinant D1 and thyroglobulin, were not recognized. Although instructive, the utility of these studies is limited by their reliance on peripheral T cells, on orbital T cell lines rather than clones, or on cells and cell fractions rather than recombinant or endogenously processed proteins. However, taken together, they do support a primary role for orbital fibroblasts as target cells or for TSHr as an orbital autoantigen in GO.
Studies concerning the pathogenesis of Graves disease and GO have long been hampered by the lack of an animal model reproducing the salient clinical and biochemical characteristics of the disease. However, as a useful model was developed recently in which naive BALB/c mice were inoculated with splenocytes primed with either a TSHr fusion protein or TSHr cDNA (48). Thyroiditis, associated with elevated thyroid hormone levels and production of low titer TSHr autoantibodies, was induced in the majority of these animals. However, these antibodies did not stimulate the receptor, but, rather, inhibited TSH binding to the receptor. Thus, the model does not entirely reflect the autoimmune milieu present in Graves disease. However, examination of mouse orbits revealed histological changes characteristic of GO, including lymphocytic and mast cell infiltration, accumulation of adipose tissue, hydrophilic mucopolysaccharides and edema, dissociation of muscle fibers, and evidence of TSHr immunoreactivity. Although falling short of clearly implicating TSHr as the orbital autoantigen in GO, this model does support the concept that enhanced TSHr expression in the orbit may contribute to the development of the disease.
In summary, the constitutive TSHr expression detectable at low levels in adipose and connective tissues from many anatomical sites as well as in other extrathyroidal tissues may have little physiological relevance in normal individuals. However, in the setting of Graves disease with circulating autoantibodies and T cells directed against this antigen, a more generalized, systemic autoimmune reaction may ensue. The finding that particular cytokines may affect TSHr expression in orbital fibroblasts suggests a role for these or other local or circulating immune factors in the modulation of expression within the orbit in GO. It may be that overexpression of TSHr plays a primary pathogenic role in GO and PTD. Alternately, locally enhanced expression of this protein may be secondary to the local disease process, but important nonetheless in disease progression.
The association between smoking and GO is quite strong, representing the major risk factor known for this condition. The odds ratio, relative to controls, has been reported to be as high as 20.2 for current smokers, and 8.9 for current and ex-smokers, suggesting a direct and immediate effect of smoking (49). In addition, several studies have shown that among patients with GO, smokers have more severe eye disease than nonsmokers. The mechanisms involved in this association are unknown. That smoking is linked to other autoimmune diseases, including rheumatoid arthritis and Crohns disease, suggests there may be a generalized stimulation of autoimmune processes in smokers. Direct effects of tobacco products or hypoxia on orbital fibroblast metabolism have been reported as well (50).
Immunosuppressive therapies, including oral or iv corticosteroids, cyclosporin, octreotide (in Europe), and orbital irradiation, are mainstays of the medical treatment of GO. However, only approximately 65% of patients respond to these agents, and the response is generally modest (51). The use of more specific immunosuppressive treatments, such as particular recombinant cytokines, cytokine receptors, or anticytokine monoclonal antibodies, may improve efficacy and decrease limiting side effects. However, because the immunological milieu appears to change over the course of the disease, a better understanding of disease evolution will be important. It is likely that some of these agents will be useful only in early, middle, or late stage disease or in immunologically active disease. Such an approach would also require the development of novel clinical and laboratory approaches to disease staging or the assessment of disease activity. The use of these agents in disease prevention or treatment of mild GO or PTD will come with the development of new agents with few adverse side effects.
Cellular contributions
The primary cellular function of fibroblasts appears to be synthesis of GAG and enzymes necessary for GAG remodeling and degradation. These activities are central to the elaboration, organization, and disposal of extracellular matrix. Fibroblasts were long considered to be a relatively homogeneous population of cells. However, recent studies have shown these cells to possess a wide array of tissue-specific phenotypes. These site-specific characteristics may contribute to the involvement of certain fibroblast-containing tissues in the extrathyroidal manifestations of Graves disease. For instance, hyaluronan synthesis in fibroblasts obtained from the orbit is relatively insensitive to T3 and dexamethasone, whereas its production in skin fibroblasts is markedly inhibited by physiological concentrations of these hormones (52). Similarly, orbital fibroblasts treated with IFN and leukoregulin synthesize high levels of hyaluronan, whereas levels in similarly treated dermal fibroblasts are much lower. Likewise, orbital fibroblasts produce extremely high levels of PGE2 after treatment with proinflammatory cytokines, whereas the response of dermal fibroblasts is less dramatic (53). Finally, unlike cultured fibroblasts from other anatomical sites, those obtained from the orbit do not express hyaluronidase (54). This enzyme is necessary for the degradation of hyaluronan, and its absence in orbital fibroblasts may contribute to the local accumulation of hydrated GAG in the orbit in GO.
Fibroblast heterogeneity extends even to subpopulations of these cells within a single tissue. These subsets can be differentiated on the basis of protein, carbohydrate, and ganglioside cell surface markers (55). Although some markers define subsets with functional characteristics such as sensitivity to cytokine-induced PGE2 or IL-8 expression, others appear to identify precursor cells capable of differentiating into more specialized cell types, such as chondrocytes, myofibroblasts, and adipocytes (56). Such fibroblast heterogeneity within a single tissue may contribute to differences in disease expression between individuals with GO, such as whether orbital fat expansion or extraocular muscle enlargement is the more prominent feature (6, 7).
The finding of TSH-stimulated lipolysis in rat epididymal adipose cells was the first indication that TSHr might be expressed in tissues outside the thyroid (57). The concept that TSHr-expressing orbital tissue may be targeted in GO evolved from early studies showing TSH binding to guinea pig adipose and retroorbital tissues or to porcine orbital connective tissue membranes (58, 59). The expression of this receptor in human fat tissue was suggested by studies showing regulation of lipolysis by physiological levels of TSH in human fetal and newborn, but not adult, adipocytes (60). These results implicated TSH and its receptor in the regulation of thermogenesis in early postnatal life.
Adipocyte precursor cells, termed preadipocytes or preadipocyte fibroblasts, can be isolated from various adipose or connective tissue-containing regions of the body and stimulated in vitro to undergo adipogenesis (61). The expression of TSHr increases in parallel with adipogenesis in cultures of rat preadipocytes (62) and human preadipocytes obtained from patients with GO (63). In the latter cultures, approximately 520% of cells appear capable of responding to adipogenic stimuli. The addition of cytokines, including IFN, TNF
, and TGFß, to either rat or human fibroblast cultures has been shown to inhibit both differentiation of the preadipocyte subpopulation and expression of TSHr in these cells (64, 65).
The mechanisms involved in the apparently selective expansion of adipose tissues within the orbit in GO are not well understood. The process of adipogenesis is tightly regulated, and key molecules initiate the differentiation of preadipocytes to mature fat cells. It may be that the balance of these factors is shifted in the GO orbit toward activation of this process. Molecules known to be important in the initiation of adipogenesis include natural ligands of the peroxisome proliferator activator receptor (PPAR
) (66). The PPAR
agonist rosiglitazone is a potent stimulator of adipogenesis in orbital fibroblast cultures and of TSHr expression in these cells (67). It is possible that a similar endogenous ligand is activated within the orbit in the setting of Graves disease, leading to expanded orbital fat volume. Similarly, such a ligand would be expected to increase the local expression of TSHr and thus enhance the autoimmune reaction directed against this receptor.
Novel therapies for GO and PTD aimed at fibroblast processes might include treatment with cytokine antagonists to inhibit cytokine-stimulated biosynthesis GAG or PGE2. Another approach might be to use cyclooxygenase-2 inhibitors to target PGE2-related inflammation in the orbit and skin. In addition, the novel PPAR- antagonists currently under development might be of benefit if they were to inhibit the differentiation of preadipocytes and their expression of TSHr, thereby decreasing both orbital tissue volume and the load of this putative orbital autoantigen.
Systemic connective tissue inflammation
The orbit and pretibial skin are the extrathyroidal sites most commonly and severely affected in Graves disease. However, that the characteristic dermal lesions can develop at other sites, especially after trauma, suggests more widespread subclinical involvement. Other evidence in support of this hypothesis is that urinary GAG are elevated in Graves patients compared with controls regardless of whether severe GO or PTD is present (68). Levels of hyaluronic acid and chondroitin sulfate in these patients are substantially higher than would be expected were the orbits the only abnormal source of GAG. Rather, the elevations in urinary GAG in these patients appear to reflect widespread stimulation of fibroblast metabolic processes, as might occur in the setting of systemic connective tissue inflammation.
As discussed elsewhere in this review, extrathyroidal TSHr expression does not appear to be limited to orbital and pretibial connective tissues, but is detectable in tissues throughout the body (69). If systemic connective tissue inflammation and fibroblast activation are indeed present in Graves disease, they might be initiated by widespread T cell recognition of TSHr on fibroblasts. In addition, a recent study demonstrated that fibroblasts from the orbit and several regions of the skin can be activated by Graves IgG to express T cell chemoattractant activity (70). This activity, attributable in large part to RANTES and IL-16, could stimulate the recruitment of activated lymphocytes to regions of the skin not usually manifesting thyroid dermopathy and might contribute to widespread dermal involvement. In the background of this subclinical systemic connective tissue inflammation, the positive feedback cycle discussed in this review might lead to the production of the severe ocular and dermal manifestations of Graves disease (Fig. 2).
|
Footnotes
Abbreviations: GAG, Glycosaminoglycan; GO, Graves ophthalmopathy; IFN, interferon-
; PGE2, prostaglandin E2; PPAR
, peroxisome proliferator activator receptor
; PTD, pretibial dermopathy; Th1, T helper cell type 1; Th2, T helper cell type 2; TSHr, TSH receptor.
Received December 30, 2002.
Accepted January 30, 2003.
References