Clinical and histologic hallmarks
Graves’ disease is a common disorder with an incidence in women of 1/1000 population/year. In addition to hyperthyroidism, 25-50% of individuals with Graves’ disease develop clinical involvement of the eyes (1). While some patients with GO experience only mild ocular discomfort, 3-5% suffer from intense pain and inflammation with double vision or even loss of vision.
The clinical symptoms and signs of GO can be explained mechanically by the increase in tissue volume evident within the bony orbit. The expanded orbital tissues cause forward displacement of the globe and impairment of venous and lymphatic outflow from the orbit. These changes, combined with the local production of cytokines and other mediators of inflammation, result in proptosis, periorbital edema, conjunctival erythema and chemosis (Figure 1).

Computerized tomographic scans show that the majority of patients with GO have enlargement of both the orbital fat and the extraocular muscles, while others appear to have only adipose tissue or extraocular muscle involvement. The extraocular muscles cells themselves are intact in early, active disease, suggesting that they are not themselves the targets of autoimmune attack. Rather, the enlargement of the extraocular muscle bodies results from an accumulation of hydrophilic mucopolysaccharides, including especially hyaluronan, within the perimysial connective tissues (2). In later stage disease, the resolving inflammatory process within the muscles may leave them fibrotic and misaligned.
The increase in the volume of the adipose/connective tissues within the orbit appears to contribute more significantly to the overall expanded orbital tissue volume than does the extraocular muscle enlargement. Computerized tomographic studies show that proptosis measurements in these patients are most closely correlated with the volume of the fat compartment (3). This expanded adipose tissue volume appears to result both from hyaluronan accumulation with attendant edema, and from the emergence of a population of newly differentiated fat cells within these tissues (4).
Thyroid (“pretibial”) dermopathy is typically a nodular thickening of the skin on the anterior lower legs. This condition is evident in approximately 15% of Graves’ patients with severe GO. The histologic changes in the subdermal connective tissues in thyroid dermopathy are similar to those within the GO orbit, with lymphocytic infiltration and hyaluronan accumulation (5), but without an increase in adipose tissue volume.

Heterogeneity in the fibroblast phenotype
The characteristic histologic changes within the tissues outlined above suggest that the orbital fibroblast constitutes the target cell in GO. Rather than being a homogeneous population of cells, fibroblasts exhibit phenotypic heterogeneity even within a single tissue (6). Some cells within the orbital fibroblast population are capable of producing hyaluronan and inflammatory prostanoids, while others (termed "preadipocyte fibroblasts" or "preadipocytes") have the capacity to differentiate into mature adipocytes (5). Connective tissues investing the extraocular muscles contain the former, while preadipocytes are found primarily in the orbital connective/adipose tissue depot. These phenotypic differences between fibroblasts within the orbit may help to explain why some patients with GO have predominant eye muscle disease, while enlargement of the adipose tissues is the major disease feature in others (8,9).
Fibroblasts also possess a wide array of tissue-specific phenotypes (10). Early studies of orbital fibroblasts focused on cytokines, their effects on orbital fibroblasts biology, and phenotypic differences between fibroblasts from the orbit and skin (1,6). For instance, orbital fibroblasts treated with IFN-? or leukoregulin synthesize high levels of hyaluronan, while dermal fibroblasts similarly treated produce only small amounts (11,12). More recent studies have centered on the particular sensitivity of orbital fibroblasts to induction of CD40 expression following IFN-? treatment. This receptor is bound by the CD154 receptor on activated T lymphocytes.CD40/CD154 ligation results in the production by fibroblasts of several mediators of inflammation, including IL-1, IL-6, and IL-8, and in the synthesis of high levels of hyaluronan (13).
Preadipocyte fibroblasts also show regional differences in the expression of adipocyte-specific genes (14,15), and vary in their adipogenic potential; PPAR-? agonists enhance differentiation of preadipocyte fibroblasts from subcutaneous sites, while those from omental sites are refractory to these agents (16). The study of such depot-specific differences in fibroblast phenotype may help to explain why patients with GO have expanded orbital adipose tissues, without evidence of involvement of other adipose tissue depots, and why the lower legs are more commonly affected than are other skin regions.
Unique anatomical features of the orbit and lower extremities appear to play a role in their prominent clinical involvement in Graves' disease (17,18). The unyielding confines of the bony orbit predisposes to compression of low-pressure lymphatic and venous channels, increasing retroocular pressure and periorbital edema. Similarly, prolonged standing contributes to compromise of these channels in the lower extremities, likely contributing to the dependent edema seen in thyroid dermopathy. Moreover, individual anatomic 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 dermopathy.

Extrathyroidal TSHR expression
The close clinical relationship between Graves' hyperthyroidism and GO (18), and the finding of a correlation between thyroid-stimulating autoantibody levels and the clinical activity of GO (19), suggest that immunoreactivity against TSHR may underlie both conditions. The concept that TSHR-expressing orbital adipose tissue may be targeted in GO evolved from early studies showing TSH binding to guinea pig adipose and retro-orbital tissues, or to porcine orbital connective tissue membranes (20, 21). The expression of this receptor in human fat tissue was first suggested by studies showing regulation of lipolysis by physiologic levels of TSH in human fetal and newborn, but not adult, adipocytes (22). These results implicated TSH and its receptor in the normal regulation of thermogenesis in early post-natal life.
A prerequisite for involvement of TSHR as an autoantigen in GO is that it be expressed in affected orbital tissues. Studies aimed at identifying TSHR in the orbit tissues have been performed by several laboratories using many different approaches. Results of these studies are in general agreement, demonstrating the presence of TSHR mRNA and protein in both GO and normal orbital adipose tissues and derivative cultures (23-28). In addition, our laboratory reported that TSHR expression is up-regulated in GO orbital fat compared with normal orbital adipose tissues (29). A recent study by another group supported these findings and additionally showed a positive correlation between TSHR mRNA levels in GO orbital adipose tissues excised during decompression surgery and the patients’ clinical activity score (30). Similarly, TSHR appears to be more abundant in PTD (pretibial dermopathy) skin than in normal pretibial skin (17).
The only animal model of Graves’ disease in which ocular changes suggestive of GO have been reported to date was developed by Marian Ludgate and colleagues (31). This group transferred T cells primed in animals immunized with a TSHR fusion protein or vaccinated with TSHR cDNA. While thyroiditis and antibodies directed against the TSHR were reported in these animals, hyperthyroidism and thyroid-stimulating autoantibodies were not produced. The authors described tissue edema, dissociation of muscle fibers, minimal lymphocytic infiltration, and the presence of TSHR immunoreactivity within the orbital adipose tissues in the majority of immunized animals. However, while the histopathology appeared promising, the mice did not develop any of the characteristic clinical signs of GO. Of particular interest is a recent publication by this same group in which they brought into question the interpretation of the thyroid and ocular changes reported in their original study (32). Nevertheless, the partial success of this model suggests that transfer of TSHR-primed T cells may hold potential for the induction of ocular disease, and supports the concept that TSHR may be an important orbital autoantigen.

Relationship between adipogenesis and TSHR expression
We studied the relationship between adipogenesis and TSHR expression in cultures of orbital preadipocyte fibroblasts undergoing in vitro differentiation. Levels of mRNA encoding TSHR, as well as leptin and adiponectin (genes expressed exclusively by mature adipocytes), were found to be approximately 10-fold higher in differentiated cultures compared with control cultures. In addition, relatively greater expression levels of these genes was apparent in cultures derived from GO orbital tissues than in normal orbital cultures (33, 34).
We also examined expression of these genes in uncultured GO and normal orbital adipose tissue specimens. We found TSHR, leptin, PPAR-? and adiponectin mRNA levels to be several-fold higher in the GO than in the normal tissues (Figure 2), with significant positive correlations noted between levels of TSHR mRNA and mRNA levels of the adipocyte genes (35).



These results suggested to us that adipogenesis may be enhanced in the GO orbit, and that increased TSHR expression is a consequence of this process.

Potential role of the insulin-like growth factor receptor (IGF-1R)
Recent studies by Terry J. Smith and colleagues demonstrated that fibroblasts from patients with Graves' disease are activated by immunoglobulins (IgG) from these same donors to synthesize the RANTES and IL-16, molecules that provide signals for the infiltration of immunocompetent cells into areas of inflammation (36). This activation appears to be mediated through the IGF-1 receptor pathway as it is blocked in these cells by specific IGF-1 antibodies or transfection with a dominant-negative mutant IGF-1R (37). The IgG-stimulated activation of this receptor does not, however, appear to be restricted to fibroblasts from the orbit and pretibial skin, as fibroblasts obtained from diverse sites in these Graves' patients behaved similarly. In contrast, fibroblasts obtained from patients without known autoimmune disease, regardless of their site of origin, do not respond to these IgGs. These findings suggest that IGF-1R may be a novel second autoantigen in Graves' disease, playing an important role in lymphocyte trafficking. The relatively restricted involvement of the orbit and pretibial skin in the extrathyroidal manifestations of Graves' disease may be explained, in part, by the exquisite sensitivity of fibroblasts from these sites to stimulation by cytokines and other immune factors (38).

Summary
Histologic examination of orbital tissues in GO reveals that the characteristic changes result primarily from hyaluronan accumulation with edema, expansion of the fat compartment, and infiltration of the tissues by T lymphocytes. Studies using cells obtained from these tissues have shown that the orbital fibroblast is a resident cell uniquely capable of participating in these diverse cellular processes. These cells are particularly sensitive to stimulation by cytokines and other immune mediators, responding by increasing CD40 expression, synthesizing large quantities of hyaluronan, and secreting inflammatory cytokines. In addition, the preadipocyte subpopulation of fibroblasts is capable of differentiating into mature adipose cells that exhibit high levels of TSHR. Fibroblasts have also been shown to display IGF-1R. When bound by IgG from Graves' patients, these receptors initiate downstream signaling that results in RANTES and IL-16 production and leads to local lymphocytic infiltration. The relative site-specificity of orbital and pretibial involvement in Graves' disease may be explained both by the relative sensitivity of these fibroblasts to immune mediators, and by the unique anatomical features of these sites that appear to predispose them to compression of low-pressure lymphatic and venous channels.