(in vitro), of TNF-α and IL-6 in small preliminary open-label trials has provided evidence for a pathogenic role of these cytokines in the active phase of GO, which probably precedes adipogenesis, but further studies are needed to understand the potential therapeutic implications of such an approach [51].
In addition to the traditional Th1 (cell-mediated) or Th2 (humoral) T-cell responses, recent unexpected results in animal models of autoimmunity revealed the need for T cells secreting IL-17, accordingly labelled Th17 cells. The current view is that Th1 cells initiate the autoimmune response but Th17 cells are required for tissue damage; indeed, lesions (e.g., multiple sclerosis plaques and rheumatoid synovial cells) contain high levels of IL-17 [52]. As yet little is known about the importance of these developments in GH or GO (Fig. 6) and the majority of studies assess regulatory T cells and Th17 cells in peripheral blood, rather than the target organ. In this setting, Kahaly et al. [53] reported similar levels of regulatory T cells in GO patients and controls whilst Peng et al. [54] demonstrated increased numbers of Th17 and Th22 cells in the periphery of GD patients which correlated with thyroid-stimulating antibody levels.
There is an epidemiological association between GO and TRAb:
•TRAbs are detectable in nearly 100% of patients with GO;
•TRAbs are present in the patients in whom GO is associated with autoimmune thyroiditis.
Fig. 6. Role of various T-cell subsets in the initiation and progression of autoimmune diseases. TLR, Toll-like receptor; Treg, regulatory T cells; SNPs, single-nucleotide polymorphisms; TSAb, thyroid-stimulating antibody; GO, Graves’ orbitopathy; GD, Graves’ disease.
There is a gross correlation between the presence/levels of TRAb and severity of GO [55]. In 1 study, thyrotrophin-binding inhibitory immunoglobulin and thyroid-stimulating immunoglobulin levels were closely correlated with GO clinical activity scores. Also, there is a correlation, although weak, between TRAb levels and proptosis [1]. A follow-up study showed that the levels and prevalence of TRAb were higher in patients with a more severe course of GO; in 50% of patients, TRAb appeared to be a significant risk factor for GO, independent of age and smoking [56].
There is experimental evidence that TSH-R-induced immunity is strongly associated with GO. In a preclinical GO mouse model high TRAb levels were detectable in 100% of TSH-R A-subunit immunized mice, and almost all mice developed GO-like disease. However, individual mice developed a mixture of different levels of either stimulating and/or blocking antibodies which are known to differently stimulate or dampen downstream signalling. The resulting net effect could be an explanation why individual mice developed different degrees of inflammation, myopathy or adipogenesis although a direct correlation was not shown [22, 23].
The direct responsibility of TRAb in GO is under debate. While GO is usually absent in hyperthyroid neonates with TRAb-driven hyperthyroidism, TSH-R activation may be relevant, as illustrated by the proptosis reported in infants harbouring germline-activating mutant TSH-R [57]. As mentioned above, early stages of adipogenesis are stimulated in orbital preadipocytes expressing such activating mutant receptors and a monoclonal thyroid-stimulating antibody (M22) is able to replace the adipogenic stimulus provided by insulin [58]. Furthermore, gene expression profiles of cells experiencing TSH-R activation demonstrated an increase in transcripts for HA synthases 1 and 2 when compared with the controls. Subsequent experiments revealed that HA production could also be stimulated by TSH or a monoclonal stimulating TRAb in subcutaneous preadipocytes (providing an explanation for the classic myxoedema of hypothyroidism) or normal (but not GO) orbital cells [59].
Activation of TSH-R may contribute to pathogenetic mechanisms indirectly. A recent case study described a woman whose GO, despite being resistant to steroid therapy, was significantly improved by a cyclo-oxygenase 2 inhibitor (celecoxib) [60].
A possible rationale could be:
•It has been shown that adipogenesis could be triggered by activated T cells through the production of a PPAR-γ agonist in a cyclo-oxygenase-2-dependent manner [61]
•The patient’s lymphocytes had elevated cyclo-oxygenase 2 expression
•Although not evaluated in this case, TSH-R expression by the patient’s lymphocytes is likely as demonstrated on murine dendritic cells and on a subset of CD45 RB lymph node T cells [62]
•Therefore, circulating TRAbs could have activated the T-cell subset to produce a PPAR-γ agonist which triggered adipogenesis. As a consequence, orbital TSHR expression would have increased and promoted infiltration of TSH-R-specific T and B cells. These could then elaborate a variety of cytokines (e.g., IL-1β) or autoantibodies (e.g., to IGF-1R) and stimulate HA production by fibroblasts
Finally, TSH-R activation may also be due to thyrostimulin, a new member of the glycoprotein hormone family which is expressed in many tissues, including the orbit, but is not present in the circulation [63]. Of interest, proptosis developed in mice overexpressing the thyrostimulin β2-subunit [64].
Is There a Familial Predisposition or a Specific Genetic Background for Graves’ Orbitopathy?
Among the non-modifiable risk factors for GO, male gender and age are significant, although more so for the severity of the disease than for its mere occurrence [65].
Genetic predisposition to GD is demonstrated by familial clustering and twin studies. However, as far as GO is concerned, no clear familial aggregation was observed in 114 consecutive patients with severe GO. Only 3/114 had a family history of GO, and all 3 were second-degree relatives [66].
GD appears to be inherited as a complex multigenic disorder and candidate gene studies (HLA on chromosome 6p21, CTLA4 on chromosome 2q33, LYP on chromosome 1p13 and TSHR on chromosome 14) are promising. The CTLA4 gene is associated with susceptibility to GO. The G allele at exon 1 CTLA4(50)A/G polymorphism is associated with GO (odds ratio 1.65). More importantly, G allele frequency is correlated with severity of GO. The T allele at intron 1 CTLA4(1822)C/T,