Keywords

FormalPara Learning Objectives

  • Mechanistic explanation of eye changes

  • Orbital fibroblasts as target cells

  • Excessive glycosaminoglycan production

  • Swelling of extraocular muscles and orbital fat

  • Autoimmune thyroid disease

  • Role of TSH receptor antibodies

  • Interaction with IGF-1 receptor

  • Role of genetic and environmental factors

Introduction

The etiology of Graves’ orbitopathy (GO) is still incompletely understood, but many aspects of the pathogenesis have been unraveled over the last decades. Family members of Graves’ patients have an increased risk to develop the same disease, and smoking is a strong risk factor both to provoke GO and to develop severe forms of GO. The mechanistic explanation of the orbital reshaping in GO is now well-understood, whereas significant progress has been made in our understanding of immunological and molecular changes leading to GO.

Mechanistic Explanation of Eye Changes in Graves’ Orbitopathy

The hallmark of Graves’ orbitopathy (GO) is swelling of extraocular muscles and orbital fat, evident from orbital imaging with CT-scans or MRI. Swollen retrobulbar tissues are due to local inflammation and production of excessive amounts of hydrophilic glysoaminoglycans (particularly hyaluronan) causing edema. When the extraocular muscles are affected, it leads to dysfunction due to a failure of relaxation. It limits movement into the field of the ipsilateral antagonist, which, if asymmetrical, gives rise to double vision [1]. The human orbit is a tight space, completely surrounded by bone except anteriorly. Here, instead of bone, there is a fascial sheet across the orbital opening called the anterior orbital septum. Because of the bony surroundings, swollen retrobulbar tissues have no other outlet than pushing the globe forwards causing exophthalmos [2].

The development of exophthalmos has even been termed “nature’s own decompression”. A tight anterior orbital septum might preclude proptosis and nature’s own decompression, resulting in higher retrobulbar pressures and optic neuropathy. This mechanism is supported by measuring retrobulbar pressure with an intraorbitally positioned micropressure transducer. In GO patients with sight loss due to dysthyroid optic neuropathy (DON), retrobulbar pressure ranged from 17 to 40 mmHg (mean 29 mmHg) falling to 9–12 mmHg upon decompression; in GO patients with marked exophthalmos but without DON, retrobulbar pressure was 9–11 mmHg, with no reduction upon surgical decompression [3]. Pressure on the optic nerve leads to colour impairment, altered pupil responses and loss of vision. In contrast, patients with equivalent intraorbital soft tissue swelling but with a lax anterior orbital septum will “self-decompress” to develop exophthalmos with no or limited rise in retrobulbar pressure. Thus, patients with muscle restriction but without exophthalmos are at risk for DON [1]. Inflammation in the eyelids causes visible edema, erythema and festoons. Upper eyelid retraction is multifactorial, due to a combination of increased sympathetic stimulation of Müller’s muscle, contraction of the levator muscle due to its direct involvement, and scarring between the lacrimal gland fascia and levator which specifically gives rise to lateral flare [1]. Corneal signs are secondary phenomena: a wide palpebral aperture leads to increased tear evaporation, which combined with poor blinking causes superficial punctate erosions and symptoms of surface irritation [1].

Recent volumetric studies of Caucasian human orbits have quantified the size of orbital contents. Bony orbital cavity volume (OV) in control males is greater than in control females (28.9 cm3 vs 24.9 cm3, p < 0.001); likewise, orbital fat volume in males is greater than in females (16.2 cm3 vs 14.0 cm3, p < 0.001), and the same is true for muscle volume (4.2 cm3 vs 3.7 cm3, p < 0.001) [4]. The sex difference disappears when applying the ratio of fat volume to orbital volume (FV/OV 0.56 ± 0.11 vs 0.56 ± 0.10) and of muscle volume to orbital volume (MV/OV 0.15 ± 0.02 vs 0.15 ± 0.02). There exists a direct correlation between age and FV/OV and a weaker indirect correlation between MV/OV and age, indicating an increase of fat volume and a decrease of muscle volume with advancing age [4]. Age-specific reference ranges of FV/OV and MV/OV ratios could thus be established [4]. These data have been applied to a series of 95 consecutive referrals with untreated GO. All patients were Caucasians and had definite GO, but the patient mix was heterogenenous ranging from mild to severe ophthalmopathy [5]. In 25% of patients, neither FV nor MV was exceeding the upper normal limit, 5% had only an increased FV, 61% had only increased MV, and 9% had both increased FV and MV. It should be realized that patients without increased fat or muscle volumes (values lower than the 97.5th percentile of the reference interval) could have volumes which increased recently to values still within the P2.5–P97.5 reference interval. Comparing patients with and without increased MV (MV/OV 0.24 vs 0.16 p < 0.00; FV/OV 0.59 vs 0.60, NS), patients with increased MV were older (52 year vs 45 year), had more proptosis (22 mm vs 20 mm), more impaired ductions (e.g. elevation 40° vs 45°), higher diplopia scores (1 vs 0), and higher serum TBII concentrations (9.7 vs 4.2 U/L). Relative to patients without increased FV, patients with increased FV (FV/OV 0.81 vs 0.57, p < 0.00; MV/OV 0.22 vs 0.20, NS) had more proptosis (24 mm vs 21 mm) and less diplopia (score 0 vs 1) [5]. Whereas these associations make sense, the differential involvement of orbital fat and extraocular muscle enlargement in GO remains incompletely understood [6]. It could exist a number of GO variants, each with a different immunopathogenesis: one with predominant muscle enlargement and another with predominant fat enlargement, and mixtures. An alternative hypothesis is that the increase in orbital fat is just a late phenomenon in the natural history of GO [6]. This view is strengthened by the finding that a long duration of GO at diagnosis is associated with a greater orbital fat volume as compared to a shorter duration (FV/OV >1 year 0.65 vs FV/OV <1 year 0.55, p = 0.004), whereas muscle volume was similar (MV/OV >1 year 0.22 vs MV/OV <1 year 0.21, NS) [6]. The hypothesis is further supported by a 4-year follow-up of patients with mild GO who did not require specific treatment: fat volume was greater at follow-up (FV/OV at baseline 0.57 vs FV/OV at 4 year 0.65, p < 0.000) whereas muscle volume was smaller at follow-up (MV/OV at baseline 0.17 vs MV/OV at 4 year 0.14, p < 0.000) [7, 8]. The swelling of orbital fat appears to be a rather late phenomenon, in line with the transition of a subset of orbital fibroblasts into mature adipocytes (adipogenesis) during the immunopathogenesis of GO.

Immunopathogenesis of Graves’ Orbitopathy

Orbital fibroblasts (OF) are considered the target cells of the autoimmune attack in GO. For retrobulbar T cells from GO patients recognize autologous OF (but not eye muscle extracts) in a MHC class I restricted manner, and proliferate in response to autologous proteins from OF (but not from orbital myoblasts). Conversely, OF proliferate in response to autologous T cells dependent on MHC class II and CD40-CD40L signaling [9]. OF are capable, when stimulated, to produce excessive amounts of glycosaminoglycans (GAG), notably hyaluronic acid (HA). GAGs are very hydrophilic compounds and thus attract much water, resulting in edematous swelling. GAGs accumulate in the endomysial space between muscle fibers. There is no increase in the number of muscle fibers and no ultrastructural damage to the muscle cells themselves in GO (except in very advanced cases when some damage may be seen). OF investing the extraocular muscle fibers and residing within the orbital connective/fatty tissue are heterogeneous cells, which can be classified based on the presence or absence of the cell surface glycoprotein CD90 also known as thymocyte antigen-1 (Thy-1) [10, 11]. OF expressing this antigen are capable of excessive HA production and are abundant in the extraocular muscles. Conversely, OF within the orbital connective/fatty tissue are Thy-1 negative and characteristically undergo adipogenesis under appropriate conditions.

Immunocompetent cells in the orbital tissues consist of CD4+ and CD8+ T cells, few B cells, monocytes, and abundant macrophages [12]. Many of these cells are activated memory cells, frequently located adjacent to blood vessels. Macrophages, but also B cells, might present the responsible autoantigen—most likely the TSH receptor- to T cells, which are then stimulated to recognize OF. Activated infiltrating T cells produce cytokines and chemokines capable of remodeling orbital tissues. The cytokine profile in the early stages of GO is predominantly derived from Th1 cells, whereas cytokines are mostly derived from Th2 cells in patients with a GO duration >2 years [13]. Data suggest GO is primarily a T-cell mediated disease, initiated by the migration of T-helper cells into the orbit. Cytokines induce expression of immunomodulatory proteins on orbital endothelial cells and fibroblasts, such as HLA, hsp-72 and several adhesion molecules, generating T-cell migration. Cytokine-activated OF synthesize chemo attractants, IL6 and RANTES, perpetuating the immune attack.

Macrophages may present antigen to T-cells (CD40L) through provision of costimulatory and proinflammatory cytokines. Activated T-cells bind to CD40+ fibroblasts, inducing proinflammatory compounds (like cytokines, COX2, PGE2) and excessive GAG production. A subset of OF may differentiate into mature adipocytes, associated with increased expression of TSH receptors [14].

Fibrocytes are bone marrow derived cells from the monocyte lineage, expressing CD45, CD34, CXCR4, collagen 1, Tg and functional TSH receptors. Circulating fibrocytes are highly abundant in GO patients, and seem to infiltrate orbital connective tissue where they might transition to CD34+ OF [15]. Teprotumumab, a human monoclonal IGF-1R blocking antibody, attenuates the actions of IGF-1 and TSH in fibrocytes; specifically, it blocks the induction of proinflammatory cytokines by TSH [16]. It raises the question about the nature of the autoantigen in Graves’ orbitopathy.

TSH Receptors Are the Major Autoantigen in Graves’ Orbitopathy

TSH receptors (TSHR) in human retrobulbar tissues were detected in 1993, in GO patients to a greater degree than in healthy persons [17]. It raised the hypothesis that TSHR autoimmunity (the hallmark of Graves’ hyperthyroidism) could also impact the orbit. Full-length functional TSH receptors are expressed on OF [18], in active stages of the disease to a greater extent than in the inactive stages, and directly related to IL-1β levels in the orbit [19]. Graves’ immunoglobulins (isolated from the serum of patients with Graves’ hyperthyroidism) as well as M22 (a monoclonal TSH receptor stimulating antibody) recognize TSH receptors on OF as evident from increased cAMP and hyaluronan production in cell cultures of differentiated human OF [9]. Another post-TSHR signaling pathway runs not via cAMP but through PI3K (phospho-inositide 3-kinase) resulting in adipogenesis [10, 20].

Clinical studies support the role of TSHR. First, serum TSHR-Ab concentrations are higher in patients with both Graves’ hyperthyroidism and GO than in patients with Graves’ hyperthyroidism without GO [21]. Second, serum TSHR-Ab are directly related to the activity and severity of GO [22]. Third, the higher serum TSHR-Ab, the higher the risk of an unfavorable course of GO [23]. Fourth, serum TSHR-Ab fall after thyroidectomy or treatment with antithyroid drugs, but increase substantially after radioactive iodine therapy. 131I therapy (but not thyroidectomy or antithyroid drugs) is associated with development or worsening of GO in about 15% of patients [24, 25]. One could argue that not all GO patients have concomitant Graves’ hyperthyroidism: about 10% of all GO patients are euthyroid or even hypothyroid. However, serum TSHR-Ab can be detected in the majority if not all of such patients. But the most convincing argument for the TSH receptor as the primary autoantigen in GO is derived from experimental animal studies. Genetic immunization of mice with the TSH receptor A-subunit plasmid have produced a fair animal model of GO [26, 27]. Despite inherent limitations to the mouse model (e.g., the lateral wall of the orbit in rodents is not made of bone but consists of a connective tissue septum), the model resembles reasonably well the human condition.

Cross-Talk Between TSH Receptors and IGF-1 Receptors

Another autoantigen in GO might be the insulin like growth factor-1 receptor (IGF-1R). IGF-1R are indeed upregulated on OF of GO patients, but serum IGF-1 and IGF-binding proteins in GO patients are normal. Graves’ immunoglobulins are able to induce hyaluronan synthesis in OF; the effect can be blocked by monoclonal IGF1-R blocking antibodies, suggesting a pathway independent of the TSHR [28]. The authors suggested Graves’ IgG contain antibodies stimulating the IGF-1R. IGF-1R antibodies were detected in 10% of GO patients and in 11% of controls [29]. IGF-1R antibodies were not related to GO disease activity or severity. Their serum concentrations were rather constant, demonstrating relatively stable expression over time. IGF-1R antibodies failed to stimulate IGF-1R autophosphorylation but instead inhibited IGF-1-induced signaling. Available data do not support the idea that stimulating IGF-1R antibodies are involved in the pathogenesis of GO [30]. Further studies provided proof that TSHR stimulating antibodies do not activate IGF-1 receptors [31]. Convincing evidence that IGF-1 receptors do not act as primary autoantigens in GO is obtained from animal studies applying genetic immunization with IGF-1Rα plasmid [27]. These mice generated high levels of IGF-1Rα antibody, but did not develop apparent pathology. Of interest is that some animals immunized with TSHR A-subunit developed low-titer IGF1-Rα antibodies shortly after immunization. It can be concluded that the TSHR and likely its A-subunit, is the primary autoantigen in GO. TSHR are “the culprit as well as the victim” not only in Graves’ hyperthyroidism, but also in Graves’ orbitopathy [32].

Functional interactions between TSHR and IGF-1R have been demonstrated in cultured OF obtained from GO patients. Simultaneous activation by TSH and IGF-1 synergistically increases hyaluronan secretion in OF [33]. Blockade of IGF-1R inhibits both hyaluronan synthesis and Akt phosphorylation induced by the monoclonal TSHR stimulating antibody M22 or IGF-1 [34]. Such studies suggest a crosstalk between TSHR and IGF-1R [35]. The TSHR is a heptahelical G-protein coupled receptor, whereas the IGF-1R is a receptor tyrosine kinase. Despite these fundamental structural differences, both receptors are phosphorylated by G-protein receptor kinases, which enables β-arrestin binding [36]. Arrestins mediate receptor internalization and also activate the mitogen-activated protein kinase (MAPK) pathway. TSHR must neighbor IGF-1R for crosstalk in GO fibroblasts to occur, and this depends on arrestin-β-1 acting as a scaffold [37]. TSH activates both G-proteins and β-arrestin, suggesting the different signals in the TSHR are propagated in differentiated intramolecular pathways [36]. A signalosome has been proposed, in which arrestin-β-1 mediates proximity and crosstalk of the TSHR and IGF-1R [38]. Combination therapy targeting simultaneously TSHR and IGF-1R might be more effective than targeting either receptor alone (the TSH receptor by either small molecule TSHR antagonists or monoclonal TSHR blocking antibodies, and the IGF-1R by monoclonal IGF-1R blocking antibodies) [36, 38].

Genes and Environment

A number of susceptibility genes for Graves’ disease has been identified, notably TSHR, HLA, CTLA4, PTPN22, CD40 and FRCL3. However, the frequency of particular polymorphisms in these susceptibility genes do not differ substantially between Graves’ hyperthyroid patients with and without Graves’ orbitopathy [39]. In the absence of clear genetic markers specific for GO, it is difficult to explain why, among Graves’ hyperthyroid patients with the same level of TSH receptor antibodies, some will develop GO whereas others will not. TSHR immunoglobulins are heterogeneous, comprising antibodies that stimulate or block the TSH receptor or antibodies that are neutral with respect to their binding to the TSHR; the various classes of TSHR-Ab might have differential effects on orbital tissues.

Alternatively, environmental factors could play a role, and in this respect, it is noteworthy that smoking greatly increases the risk for GO (odds ratio 7.7, 95% CI 4.3–13.7) [40]. The proportion of GO among all referred patients with Graves’ hyperthyroidism decreased from 57% in 1960 to 35% in 1990, and to 29% in 2010 [41, 42]. The secular decline in smoking is likely causally related to a lower prevalence of GO in the last decades [43]. GO patients referred in 2012 had less severe and less active ophthalmopathy than those referred in 2000 [44]. Orbital muscle volume is larger in GO patients who are current smokers compared to GO patients who are never smokers or ex-smokers; orbital fat volume had no relationship with smoking behaviour [45].

A new risk factor for GO might be hypercholesterolaemia [46]. Statin use (for >60 days in the past year vs <60 days or nonuse) was associated with a 40% decreased hazard of GO (adjusted HR 0.60, CI 0.37–0.93) in a population of patients with newly diagnosed Graves’ hyperthyroidism [47]. No significant association was found for the use of nonstatin cholesterol-lowering medication or cyclooxygenase 2 inhibitors and the development of GO. Subsequent studies confirmed high serum cholesterol levels as a potential risk, at least in patients with a relatively short duration of Graves’ disease [48, 49]. The pleiotropic anti-inflammatory actions of statins, which are not related to their cholesterol-lowering action, could be key in explaining the association with GO [46].