This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Bartalena, L. Articles by Marcocci, C. Search for Related Content PubMed PubMed Citation Articles by Bartalena, L. Articles by Marcocci, C.

Management of Graves’ Ophthalmopathy: Reality and Perspectives¹

Dipartimento di Endocrinologia e Metabolismo, Ortopedia e Traumatologia, Medicina del Lavoro, University of Pisa, 56124 Pisa, Italy

	Abstract

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
Graves’ ophthalmopathy is an debilitating disease impairing the quality of life of affected individuals. Despite recent progress in the understanding of its pathogenesis, treatment is often not satisfactory. In mild cases, local therapeutic measures (artificial tears and ointments, sunglasses, nocturnal taping of the eyes, prisms) can control symptoms and signs. In severe forms of the disease (3–5%), aggressive measures are required. If the disease is active, high-dose glucocorticoids and/or orbital radiotherapy, or orbital decompression represent the mainstay of treatment. If the disease is severe but inactive, orbital decompression is preferred. Novel treatments such as somatostatin analogs or intravenous immunoglobulins are under evaluation. Rehabilitative (extraocular muscle or eyelid) surgery is often needed after treatment and inactivation of eye disease. Correction of both hyper- and hypothyroidism is crucial for the ophthalmopathy. Antithyroid drugs and thyroidectomy do not influence the course of the ophthalmopathy, whereas radioiodine treatment may cause the progression of preexisting ophthalmopathy, especially in smokers. The exacerbation, however, is prevented by glucocorticoids. In addition, thyroid ablation may prove beneficial for the ophthalmopathy in view of the pathogenetic model relating eye disease to autoimmune reactions directed against antigens shared by the thyroid and the orbit.

I. Introduction

II. Pathogenesis

III. Management of Graves’ Ophthalmopathy: General Principles

A. Whom to treat?

B. How to treat?

C. How to assess the effects of treatment?

IV. Management of Nonsevere Graves’ Ophthalmopathy

V. Management of Severe Graves’ Ophthalmopathy

A. Established treatments

B. Nonestablished treatments

C. Novel treatments under investigation

D. Miscellaneous treatments

E. Rehabilitative surgery

VI. Summary of Assessment and Treatment of Graves’ Ophthalmopathy

VII. Treatment of Hyperthyroidism and the Course of Graves’ Ophthalmopathy

A. Antithyroid drug treatment

B. Radioiodine therapy

C. Thyroidectomy

D. Total thyroid ablation

VIII. Treatment of Graves’ Hyperthyroidism in Patients with Ophthalmopathy

A. Nonsevere (or absent) ophthalmopathy

B. Severe ophthalmopathy

IX. Prevention of Graves’ Ophthalmopathy

A. Natural history of the disease

B. Cigarette smoking

C. Other risk factors

X. Therapeutic Perspectives

A. Anticytokine therapy

B. Somatostatin analogs

C. Colchicine

XI. Concluding Remarks

	I. Introduction

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
GRAVES’ OPHTHALMOPATHY (GO), the most frequent extrathyroidal manifestation of Graves’ disease (1), remains a pathogenetic enigma and a therapeutic dilemma. In its severe expression it is a disfiguring and invalidating disease that profoundly influences and impairs the quality of life of affected individuals (2). Despite recent progress in the understanding of its pathogenesis (3, 4, 5) and in its management (6), major controversies still exist in both areas. Even the denomination of this disorder is controversial, and the terms "thyroid eye disease" or "thyroid-associated ophthalmopathy" are both often used. This is because, although mostly associated with Graves’ hyperthyroidism, ophthalmopathy may less frequently occur also in patients with hypothyroid Hashimoto’s thyroiditis or in euthyroid subjects with no current or past evidence of thyroid hyper- or hypofunction (so-called euthyroid Graves’ disease) (1, 7).

Because excellent reviews have recently been published on the pathogenesis of ophthalmopathy (1, 3, 4, 5), only the most recent contributions in this field will be presented, while the discussion will be mostly focused on the management of the disease. Emphasis will be given to the coordinated treatment of the frequently associated hyperthyroidism, and future therapeutic perspectives will be considered.

	II. Pathogenesis

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
It is widely accepted that GO is an autoimmune disorder (1), but its pathogenesis is still incompletely understood (see Refs. 3, 4, 5 for review). Support for the concept that GO is of autoimmune origin comes from the associated histopathological changes. There is an increased volume of the extraocular muscles and orbital connective and adipose tissues (1); the extraocular muscles are edematous due to increased production of the hydrophilic glycosaminoglycans (GAGs) in the orbital tissue (3); a marked infiltration of immunocompetent cells (predominantly T lymphocytes and macrophages; to a lesser extent, B lymphocytes) is detectable (8, 9). Infiltrating T cells are mostly CD4+ (10, 11).

According to a leading pathogenetic hypothesis (1), autoreactive T lymphocytes recognizing an antigen shared by the thyroid and the orbit infiltrate the orbital tissue and the perimysium of extraocular muscles; this process may be facilitated by either circulating or locally produced adhesion molecules (5), the expression of which may be induced by cytokines (12) and be related to the activity of the disease (13). After infiltration of the orbit by T cells, the shared antigen could then be recognized by a T cell receptor on CD4+ T lymphocytes: the finding of a biased usage of the T cell receptor variable gene (14, 15) supports the concept of an antigen-specific immune reaction (9). After antigen recognition, CD4+ T lymphocytes could secrete cytokines that amplify the immune reaction by either activating CD8+ T lymphocytes or autoantibody-producing B cells (16). Phenotypic analysis of T cell clones from the orbital tissue of GO patients has revealed a predominance of T cells with a Th1 profile (interleukin-2, interferon-, tumor necrosis factor-) (17, 18), but also a Th2 profile of cytokine production (interleukin-4, interleukin-5, interleukin-10) has been reported (18, 19, 20). These differences might be related to different stages or activity levels of eye disease (21, 22), but they might also reflect differences in experimental methods employed in different studies.

Cytokines induce expression of major histocompatibility complex class II molecules (23) and heat-shock protein-72 (HSP-72), which are important for antigen recognition (5), and of intercellular adhesion molecule-1 (24), which is important for T cell recruitment. In addition, cytokines stimulate fibroblasts to synthesize and secrete GAGs (25, 26), which attract fluid into the retroorbital space, thus contributing to the development of periorbital swelling, proptosis, and extraocular muscle swelling (1). The expansion of the orbital content is also related to cytokine-induced proliferation of fibroblasts (27). Orbital fibroblasts may contribute to perpetuate the ongoing immune reaction in the orbit by protecting infiltrating T cells from apoptosis (28). Orbital fibroblasts include a subpopulation of cells (preadipocytes), which, under particular hormonal stimulation, differentiate into adipocytes (29) and may contribute to the increased volume of retroorbital adipose tissue.

If the above pathogenetic mechanisms are correct, two major questions arise: 1) Which is the antigen shared by the thyroid and the orbit?; and 2) Which is the orbital cell type targeted by T cells?

The TSH-receptor (TSH-R), the autoantigen involved in Graves’ hyperthyroidism, is probably a shared antigen (30, 31, 32). This concept is supported by several lines of evidence. TSH-R transcripts have been demonstrated in the orbital tissues by RT-PCR (33, 34, 35, 36, 37), but this technique has a drawback: its sensitivity allows the amplification of virtually any gene due to illegitimate transcription (38). In addition, TSH-R variants have also been detected in orbital tissue (39, 40, 41, 42). However, the presence of TSH-R-like immunoreactivity has been shown in orbital and pretibial fibroblasts using antibodies directed to the TSH-R extracellular domain (43, 44, 45). More recently, in a sample of orbital fat from a GO patient, the major TSH-R transcripts (4.6, 1.7, 1.3 kb) were demonstrated by Northern blot analysis (46). Using in situ hybridization with a digoxigenin-labeled antisense oligonucleotide probe specific for the extracellular domain of the TSH-R, Spitzweg et al. (47) demonstrated specific perinuclear and cytoplasmic TSH-R gene expression in orbital fibroblasts from GO patients and, to a lesser extent, from normal subjects. Intact and variant TSH-R mRNA transcripts were demonstrated by Bahn et al. (48) by liquid hybridization analysis in orbital adipose/connective tissue specimens from GO patients. Ludgate and co-workers (49) reported that, using monoclonal antibodies to the TSH-R produced by genetic immunization (50), immunostaining was obtained in fibroblast-like elongated cells and in adjacent clusters of adipocytes in orbit bioptic samples of GO patients, but not in tissue specimens from pseudotumor or in extraocular muscle samples. Recently, an increased expression of the TSH-R was reported in orbital preadipocytes after differentiation into adipocytes, with a relatively greater TSH-R gene expression in GO than in normal orbital tissue specimens (51).

After overexpression of the extracellular domain of the TSH-R as fusion protein in bacteria (52), a low level of IgG binding was detectable by Western blotting in sera of 3 of 11 GO patients who had negative tests for circulating TSH-R autoantibodies (49). In addition, IgA binding to a degraded fragment of the TSH-R fusion protein was observed in 6 of 11 TSH-R autoantibody-negative GO patients (49). The role of IgA class autoantibodies was previously underscored by Arnold et al. (53), who reported IgA binding to normal orbit and skin fibroblasts.

After xenografting retroorbital tissues from GO patients into severe combined immunodeficient mice, the TSH-R antibody was detected in 7 of 9 xenografted mice (54). When syngeneic TSH-R-primed splenocytes were transferred to BALB/c or NOD mice, destructive thyroiditis with a Th1 cytokine profile occurred in NOD mice, while a Th2 response developed in BALB/c mice together with the appearance of TSH-R antibodies (55). As reported by Ludgate et al. (49), eye changes similar to those found in GO [lymphocyte and mast cell infiltration, an increase in adipose tissue, periodic acid Schiff-positive (PAS+) edema] were observed in 17 of 25 BALB/c mice, but not in NOD mice.

A genomic point mutation in codon 52 of the extracellular domain of the TSH-R, leading to a proline-for-threonine substitution, was found in 2 of 22 GO patients and in no normal subjects (56). A higher prevalence of this polymorphism was found among GO patients with other extrathyroidal manifestations of Graves’ disease (acropachy, pretibial myxedema) suggesting that it might predispose to more severe immune reactions (57). However, the role of this polymorphism in the pathogenesis of GO appears questionable in view of other studies (58, 59, 60).

Another possible explanation is that the orbital antigen cross-reacting with a thyroid antigen might be located on eye muscle cells. A 64-kDa antigen shared by the thyroid and the orbit was reported by Salvi et al. (61), but the role and specificity of this antigen have been questioned because of its expression in other tissues (62, 63). Wall et al. (64) reported that under nondenaturing conditions a 64-kDa protein, expressed in eye muscle cells but not in skeletal muscle, reacted with serum antibodies present in 67% of GO patients but not in patients with Hashimoto’s thyroiditis or in controls. Recently, the 64-kDa protein was partially sequenced and identified as the flavoprotein subunit of mitochondrial succinate dehydrogenase, with a corrected molecular mass of 67 kDa (65). Autoantibodies reactive with purified succinate dehydrogenase were detectable in 67% of patients with active GO, 30% of patients with stable eye disease, 30% of Graves’ patients without clinically apparent ophthalmopathy, and 7% of normal subjects (65). It has been claimed that the appearance of these antibodies in the circulation might predict the subsequent development of GO (66).

Other eye muscle autoantigens possibly involved in GO include 1) a 63-kDa calcium-binding protein, called calsequestrin, expressed in extraocular muscle and skeletal muscle, but not in the thyroid (67); 2) a 53-kDa protein, identified as another calcium-binding glycoprotein, sarcalumenin, expressed in extraocular muscle and skeletal muscle, but not in the thyroid (68); 3) a different 63- to 64-kDa protein, called 1D protein, cloned from a thyroid cDNA expression library (62), which is expressed in extraocular muscles, skeletal muscles, thyroid, testis, and other tissues (69, 70); and 4) a novel eye muscle protein, called G2s, with an estimated molecular mass of about 220 kDa, expressed in extraocular muscles, skeletal muscles, and thyroid (68). A higher prevalence of circulating antibodies directed against two porcine eye muscle antigens (64-kDa and 95-kDa) has been found by immunoblotting in GO patients compared with those without eye involvement or to normal controls (71). On the other hand, serum antibodies reacting with several extraocular muscle antigens have been detected in patients with nonspecific orbital inflammation (72).

A role for eye muscle cells in GO has also been suggested by the coexpression of human leukocyte antigen (HLA)-DR and heat shock protein-70 (HSP-70) in eye muscle cells from GO patients (73). Molnar et al. (74) reported the presence of IgA antibodies reacting with eye muscle fibers, with no difference, however, in their prevalence between Graves’ patients with or without ophthalmopathy.

Most of the eye muscle antigens are intracellular, ubiquitous, and probably devoid of the disease specificity expected in an organ-specific autoimmune disorder (32). On the other hand, the G2 s protein is a cell-membrane protein and might effectively be involved in primary immune recognition. Antibodies to the other eye muscle antigens might then represent a secondary phenomenon after eye muscle damage and antigen exposure. Recently, evidence of Fas-mediated apoptosis was provided in extraocular muscle tissue from GO patients (75), but this is likely to represent a late event in the course of eye disease, preceding fibrotic changes in eye muscles.

To summarize, the orbital cell target of the autoimmune response in GO remains to be defined, but fibroblasts and adipocytes are more likely to be primarily involved; myocytes might be the object of secondary phenomena concurring with the perpetuation of the autoimmune reaction, although at the current status of knowledge, a primary role also of eye muscle cells cannot be ruled out. The nature of the putative antigen(s) shared by the thyroid and the orbit remains elusive, but many elements support the idea that TSH-R may represent the culprit antigen. The role of numerous eye muscle antigens remains to be clarified. It is conceivable that most of these eye muscle antigens, localized intracellularly, may be expressed only after eye muscle damage. Antibodies directed against these antigens might, therefore, represent not the primary event, but a secondary response and contribute to maintaining rather than triggering the ongoing autoimmune reactions in the orbit.

	III. Management of Graves’ Ophthalmopathy: General Principles

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
The majority of Graves’ patients have a mild and nonprogressive ocular involvement that does not require any specific or aggressive treatment, also because nonsevere GO often tends to improve spontaneously. In a study of 101 patients attending a thyroid-eye clinic over a 5-yr period, 59 patients were judged not to require specific treatment for ophthalmopathy and were included in a study on the natural history of the disease: ophthalmopathy spontaneously regressed with time in two thirds of cases, while it deteriorated only in 15% of cases (76) (Fig. 1). Bartley et al. (77) found that among the 120 subjects of an incidence cohort study, 89 (74%) required either no treatment or only supportive measures (Fig. 2). Thus, the two basic questions that must be addressed when evaluating a GO patient are whether he/she needs a specific treatment for eye disease and, in this case, which kind of treatment is indicated.

View larger version (14K): [in this window] [in a new window]   Figure 1. Natural history of GO. SI, Substantial improvement; MI, minor improvement; NC, no change; W, worsening. [Derived from P. Perros et al.: Clin Endocrinol (Oxf) 42:45–50, 1995 (76 ).]

View larger version (13K): [in this window] [in a new window]   Figure 2. Treatment of GO in an incidence cohort of 120 patients. NO, No treatment; SG, systemic glucocorticoids; OR, orbital radiotherapy; OD, orbital decompression; RS, rehabilitative (eye muscle and eyelid) surgery. [Derived from G. B. Bartley et al.: Am J Ophthalmol 121:200–206, 1996 (77 ).]

Definition of severity of GO is somehow arbitrary (Table 1). Undoubtedly, optic neuropathy, which can be subclinical and heralded only by changes in the visual evoked potentials (79), depicts per se a situation that can be sight threatening, especially if it is associated with an evident reduction of visual acuity. Marked proptosis may cause secondary exposure keratitis and lead to corneal ulceration or perforation. Thus, the presence of a substantial reduction of visual acuity attributable to optic neuropathy, or the presence of marked degrees of proptosis should be sufficient to define the ophthalmopathy as severe. In this regard, it may be relevant to evaluate variations in the proptosis, which may indicate a progression of the disease. Extraocular muscle dysfunction does not represent a danger for vision, but the resulting diplopia markedly influences daily activities and is responsible for major discomfort for affected individuals, especially if it is constant, i.e., present in all positions of gaze. Accordingly, extraocular muscle impairment, when causing diplopia in primary and reading gaze positions, should also be considered as an indicator of the severity of the disease. Soft tissue involvement, either inflammatory or congestive, is in most cases more striking than dangerous, although it disturbs and creates discomfort for the patient. For this reason, except for the rare cases with extremely severe periorbital swelling, conjunctival hyperemia and chemosis, soft tissue involvement should not be sufficient to define the disease as severe. Soft tissue manifestations are, however, relevant to assess the activity of the disease (see below) and for the perception of the disease by the patient. In addition, soft tissue involvement is rarely isolated in severe eye disease and, under most circumstances, is associated with some of the other expressions of the disease.

A different concept refers to the activity of the disease. The natural history of GO is not completely understood, but it seems that the ophthalmopathy undergoes an initial, active phase of progressive exacerbation, followed by a subsequent partial regression and a static, inactive phase in which the residual manifestations of the disease (e.g., proptosis, strabismus due to fibrotic changes of the extraocular muscles) are unlikely to show any further substantial change (16). If this model is correct, it is evident that the activity of the ophthalmopathy is neither synonymous nor coincident with the severity of the disease. In other words, an individual patient may have severe ocular manifestations, but the disease may have run its course (Fig. 3). To assess the activity of the ophthalmopathy, Mourits and co-workers (80) proposed a clinical activity score (CAS), which in its original formulation included 10 different items (Table 2), mainly, but not solely, reflecting inflammatory changes: giving one point to each manifestation, a score is obtained, with a range from 0 (no activity) to 10 (highest activity). This group recently evaluated the usefulness of this score in predicting the outcome of either radiotherapy or oral prednisone treatment in patients with moderately severe GO (81). They found a significantly higher CAS in the 22 responders than in the 21 nonresponders: using a cut-off point of 4, CAS could accurately predict the outcome of therapy, since 12 of 15 patients with CAS > 4 (80%) had a favorable outcome of treatment (81). It should, however, be pointed out that 10 of 22 responders (45%) had a CAS of 4 or less (81). Thus, it would appear that, while a high CAS is usually predictive of a good response to treatment, a low CAS does not necessarily rule out a possible favorable outcome of therapy. A slightly modified CAS, which does not include some of the items originally proposed by Mourits et al. (80), was indicated by an ad hoc committee of the four Thyroid Societies as a tool to record ocular changes over time after treatment of ophthalmopathy (82) (Table 2). Gorman (83) recently stated that the activity of the ophthalmopathy is indeed difficult to define, and some of the items included in the calculation of CAS, such as periorbital swelling, caruncle edema, and chemosis, may well reflect congestion rather than inflammation.

View larger version (14K): [in this window] [in a new window]   Figure 3. Hypothetical relationship between disease activity and severity in the natural history of GO. [Reproduced with permission from W. M. Wiersinga and M. F. Prummel: J Clin Endocrinol Metab 80:345–347, 1995 (133 ). © The Endocrine Society.]

A prolongation of T2 relaxation time at MRI has been found in GO patients with active eye disease, and the response to immunosuppressive therapy has been associated with a decrease in this parameter (88). However, the MRI signal recognizes fluid accumulation and does not necessarily reflect inflammation.

After the initial report by Postema et al. (89) concerning the orbital uptake of [¹¹¹In]octreotide in GO patients, some groups have applied this receptor-mediated scintigraphy (octreoscan) to the evaluation of the disease. While some (89, 90) but not all studies (91, 92, 93) found a higher uptake of the tracer in patients with more severe forms of ophthalmopathy, all studies indicated a relationship between the octreoscan positivity and the activity of the disease: the orbital octreotide accumulation was higher in patients with active ophthalmopathy than in those with inactive eye disease (90, 91), implying also that GO patients with positive octreoscans might have a successful outcome of medical therapy. In this regard, a positive predictive value of 90–92% has been reported (94, 95). In addition, successful management of ophthalmopathy has been associated with a decrease in orbital octreotide uptake (90, 93). However, as recently pointed out by Wiersinga et al. (96), data on the accuracy and precision of this expensive and rather nonspecific technique are too limited to propose it as a mandatory indicator of GO activity or as a tool to identify patients prone to respond to immunomodulatory treatment.

Thus, the reliability of the proposed indicators of activity of GO (CAS, internal eye muscle reflectivity at ultrasound, GAG determination in urine or plasma, T2 relaxation time at MRI, positive octreoscan) must still be demonstrated with certainty. Gorman (83) stated that it may be preferable to define measurable attributes (lid fissure width, range of extraocular muscle motion, diplopia fields, volume of extraocular muscle, and connective tissue) rather than to speculate on the activity of the ophthalmopathy. However, identification of active forms of the disease, especially in cases with moderately severe ophthalmopathy, may be crucial for the prediction of the subsequent response to treatment and, therefore, for deciding whether an individual patient needs to be treated medically or surgically. Since a good correlation between octreoscan results and T2 relaxation time at MRI (90) or CAS (89, 93) has been reported, it is possible that a combination of the different parameters discussed above may provide a better definition and identification of patients with active GO, but this remains to be established.

To summarize, severity and activity of GO are not synonymous, but both are important in deciding whether a given patient requires treatment for his/her ophthalmopathy and which type of treatment is indicated. Duration of eye disease per se is less relevant for the therapeutic decision, although it is reasonable to assume that long-lasting (>2 yr) eye disease has limited features of activity. If the ophthalmopathy is nonsevere, no aggressive medical or surgical treatment is required, even though the disease shows some signs of activity. If the patient has severe ocular involvement, assessment of the degree of activity is important: patients with active eye disease are likely to respond to medical treatment (especially glucocorticoids and/or orbital radiotherapy), whereas such a treatment is unlikely to be of benefit in patients with inactive GO, who are then candidates for surgical treatment (orbital decompression or rehabilitative surgery) (see Section III.B).

B. How to treat?
Even though the pathogenetic mechanisms of GO are still incompletely understood, the resulting changes occurring in the orbit, i.e., the swelling of retrobulbar fibroadipose tissue and extraocular muscles and the dysfunction of extraocular muscles, can readily explain the clinical expression of eye disease (4). Proptosis, which can be considered a sort of "nature’s decompression" (83), causes stare and, together with lid retraction, exposure keratitis responsible for foreign body/gritty sensation, pain, lacrimation, and photophobia. Extraocular muscle dysfunction causes restriction of eye movements with diplopia and blurring of vision. The increased orbital content may lead to optic nerve compression, with impaired color vision and decreased visual acuity. Inflammation, but also the venous engorgement suggested by the increased size of the superior ophthalmic vein (97, 98), coincide with periorbital swelling.

If this is the mechanical basis of GO, treatment should be aimed either at reducing the volume of the orbital content or at increasing the available space in the orbit. The former ("medical decompression") utilizes drugs (e.g., glucocorticoids) or treatments (e.g., orbital radiotherapy) that may reduce the ongoing inflammation by nonspecific actions or by intervening in the putative causes of the disease. Surgical decompression is not intended to act on the etiology of ophthalmopathy, but only on the mechanical effects of eye disease. Supporters of medical decompression underscore the possibility that medical treatment may avoid surgery or reduce the activity of inflammation so as to improve the outcome of subsequent surgery (99). Supporters of surgical decompression emphasize the immediate effectiveness of surgery and the not-infrequent failure or partial effectiveness of conservative approaches. The choice between medical and surgical decompression, in addition to the assessment of GO activity, ultimately depends on several considerations. These include the local availability of experienced orbit surgeons or skillful radiotherapists, the existence of contraindications to glucocorticoid treatment, and the lack of a prompt response of sight-threatening manifestations, such as optic neuropathy, to medical treatment. It should be pointed out that selection of surgical decompression does not exclude the subsequent need for glucocorticoids or orbital radiotherapy to eliminate the disease. On the other hand, selection of medical decompression does not preclude the subsequent utilization of surgical decompression if functional and/or rehabilitative results are not satisfactory.

In a recent survey of members of the European Thyroid Association, the majority of European thyroidologists selected glucocorticoid treatment for the index case, and 23% preferred orbital radiotherapy, but orbital decompression was not widely indicated as the first-line treatment (100). Surgical treatment is preferred by prestigious institutions, such as the Mayo Clinic, with a wide experience in this approach. As pointed out in a recent commentary, the wide divergence in the therapeutic approach to GO reflects also the current lack of knowledge as to the best practice (101). It is conceivable that the development of guidelines for the treatment of the eye disease by an ad hoc international committee might make the approach to the management of this disease more uniform and standardized, and improve the outcome of therapy, either medical or surgical. However, more numerous controlled and randomized studies on a large number of patients are necessary before such consensus guidelines can be developed.

C. How to assess the effects of treatment?
A great deal of controversy on the results of the treatment of GO reported in the different series depends on several factors. Many available studies have been retrospective and might have introduced biases in the selection of patients, in particular because of the frequent lack of control groups. Control groups of patients not receiving a given treatment are essential to rule out the possibility that observed ocular changes are related to the natural history of the disease. The enrollment of patients with different degrees of severity or activity of the ophthalmopathy might also have contributed to conflicting results.

Of greatest importance is the manner in which the effects of treatment have been assessed in the different series. In many past studies, the assessment of treatment outcome was based mostly on variations of exophthalmometer readings, while other relevant expressions of eye disease were discounted. A substantial improvement was provided by Donaldson et al. (102), who introduced a numerical score (ophthalmopathy index, OI) based on the NOSPECS [N, No signs or symptoms; O, only signs, no symptoms; S, soft tissue involvement; P, proptosis; E, extraocular muscle involvement; C, corneal involvement; S, sight loss (due to optic nerve involvement)] classification of eye changes of Graves’ disease. In this numerical system each class of eye changes received a score from 0 to 3 according to the degree of involvement; after the scores were added, an OI was derived, ranging from 0 to 15 (102). For many years and in many studies, the OI represented the tool for the assessment of the effects of different treatments. Although it undoubtedly represented a major advancement toward a standardization of ocular evaluation, the OI was then criticized, owing to its intrinsic limitations. These included 1) its high dependence on subjective, rather than objective, evaluation; 2) the same weight given to eye manifestations of different severity and danger (e.g., soft tissue changes vs. optic neuropathy); 3) the difficulty in recording subtle changes in the different categories of eye involvement; and 4) the fact that patients do not progress from one class to another in a sequential fashion. In other studies, different numerical scores have been used, such as the total eye score, which gave different weight to the various items (103). Alternatively, clinical responses received an overall evaluation (excellent, good, fair, no responses) (102). However, these approaches did not completely solve the problem of standardization of ocular assessment and of the evaluation of treatment results.

An ad hoc committee of the four Thyroid Sister Societies (82) proposed a revised classification of eye changes of Graves’ disease, in which the criteria to evaluate the different classes of eye involvement were indicated. Most parameters received a quantitative evaluation, although there was some space left for semiquantitative (CAS) or qualitative (patient’s self-assessment) criteria (80). Most importantly, it was stated that 1) the NOSPECS classification could be maintained as a mnemonic aid for bedside evaluation of the patient; and 2) NOSPECS-derived numerical scores should no longer be used for reporting results of treatment, which should be described by specific and separate measurements derived from the revised classification of GO. This classification has not been revised since 1992 and needs to be validated. However, among other views, Gorman (83) is of the opinion that, because the proximate causes of GO are the swelling of extraocular muscles and retrobulbar fibro-adipose tissue, and the shortening and restricted contraction of extraocular muscles, assessment of the effects of treatment should basically rely on some relevant measurements, i.e., volume of extraocular muscles, volume of retrobulbar fibro-adipose tissue, proptosis, lid fissure width, range of extraocular motion on perimeter, and quantitation of diplopia fields. Other parameters, such as optic nerve function, periorbital edema, conjunctival injection, and chemosis, although useful to appraise the clinical effectiveness of treatment, should, in Gorman’s opinion, be regarded as secondary to the above proximate causes of GO (83). The validity of the CAS has also been criticized because of the difficulty in defining the activity of the disease (83) or the inclusion of symptoms or less readily assessable signs (104). These issues should represent a matter of argument for a new consensus on GO.

To summarize, recent years have witnessed an improvement in the assessment of ocular changes after treatment, owing to the introduction of more objective, quantitative measurements. This trend toward quantitation of changes should be further enhanced and encouraged, ideally as the result of a new consensus among international experts in this field. The standardization of ocular evaluation is crucial for a correct assessment of the results of treatment.

	IV. Management of Nonsevere Graves’ Ophthalmopathy

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
Most patients with Graves’ disease have mild ocular manifestations that do not require any aggressive treatment. In these cases the ophthalmologist can suggest simple local supportive measures that are usually sufficient to obtain symptomatic relief until eye disease becomes inactive (Table 4 and Ref. 6). A change in sleep position and elevation of the bed may be of some help to reduce periorbital edema; however, it is open to debate whether diuretics, used in the past, are useful in this regard (105). Photophobia can be alleviated by the use of sunglasses; the foreign body/gritty sensation related to a defective tear film is usually controlled by the use of artificial tears or ointments (6). If lagophthalmos is present, as suggested by tearing and eye irritation in the morning, taping the eyelids shut during the night is useful to prevent nocturnal corneal drying (6). Guanethidine or ß-blocking eye drops have been used for eyelid retraction (105). The long-term efficacy of guanethidine eye drops is questionable, and side effects, including ptosis, miosis, vasocongestion, and punctate keratitis, have been reported (105). ß-Blocking eye drops have also been used with variable degrees of success. Prisms may be beneficial for correction of mild diplopia if they are tolerated by the patient (106). In this regard, stick-on prisms allow a greater flexibility than ordinary prisms.

	V. Management of Severe Graves’ Ophthalmopathy

Top Abstract I. Introduction II. Pathogenesis III. Management of Graves’... IV. Management of Nonsevere... V. Management of Severe... VI. Summary of Assessment... VII. Treatment of... VIII. Treatment of Graves’... IX. Prevention of Graves’... X. Therapeutic Perspectives XI. Concluding Remarks References

 
Management of severe GO represents a difficult task that does not constantly provide favorable results. Recent years have seen the proposal of novel treatments that add to the list of established treatments (glucocorticoids, orbital radiotherapy, orbital decompression) (Table 5).

Glucocorticoids have been used in GO through different routes: oral, local (retrobulbar or subconjunctival), and, more recently, intravenous (6). Oral glucocorticoids have usually been employed at high doses (prednisone, 60–100 mg/day, or equivalent doses of other steroids) and for prolonged periods of time (several months) (6). Many studies have documented a high effectiveness of high-dose oral glucocorticoids on soft tissue changes and optic neuropathy, whereas the decrease in proptosis and the improvement in ocular motility have not always been impressive (1, 6, 108). Recurrence of active eye disease is a rather frequent problem with oral glucocorticoid therapy, not only when the drug is withdrawn, but also when its dose is tapered down (6). Interestingly, in one study the rate of recurrence was abated when cyclosporine was administered concomitantly with and after glucocorticoid therapy (111). Prummel and co-workers (112) reported that the percentage of a cohort of patients with moderately severe ophthalmopathy who responded successfully to prednisone therapy (14/28, 50%) was not significantly different from that of patients who had favorable responses to orbital radiotherapy alone (13/28, 46%). In summary, favorable effects of high-dose oral glucocorticoids are reported in slightly more than 60% of cases (range, 40–100%) (Table 6).

View larger version (43K): [in this window] [in a new window]   Figure 4. Overall results of glucocorticoid therapy administered through the oral, intravenous, or local (subconjunctival or retrobulbar) routes on GO. See Section V.A.1 for details of individual studies.

This prompted the evaluation of local (retrobulbar or subconjunctival) glucocorticoid therapy. After discrepant results of uncontrolled studies, in a prospective study we submitted 44 patients to combined treatment with orbital radiotherapy and retrobulbar glucocorticoids (14 injections of 40 mg methylprednisolone acetate at 20- to 30-day intervals): excellent or good results were observed only in 11 patients (25%) compared with 60% of favorable responses in patients receiving oral glucocorticoids (115). As illustrated in Fig. 4, the overall results of local glucocorticoid therapy appear less satisfactory than those obtained with the systemic administration of steroids. However, side effects are limited to transient ocular discomfort or pain; rare cases of conjunctival hemorrhages have been reported. Thus, local glucocorticoid therapy may be considered in patients with active ophthalmopathy and with major contraindications to the systemic administration of glucocorticoids (6).

In summary, glucocorticoids remain a fundamental therapeutic tool for GO. Glucocorticoids are particularly effective on active disease, soft tissue inflammatory changes, and optic neuropathy, but also on extraocular muscle dysfunction (if not associated with fibrotic changes). Proptosis appears to be influenced to a lesser extent by this treatment. A substantial proportion of patients (20–40%) respond only partially or do not respond at all to glucocorticoid treatment. It is conceivable that the effectiveness of treatment may be improved by properly selecting patients who are prone to have beneficial results, i.e., those with a high degree of disease activity, with ophthalmopathy of recent onset and/or with evidence of recent progression of eye involvement. If glucocorticoids are selected, the systemic route appears more effective than the local route, although it is more frequently associated with adverse effects. Intravenous administration appears to bear advantages over the oral administration in terms of effectiveness and possibly of side effects, but this remains to be proven by randomized studies. Since recurrences are not infrequent when the drug is tapered or withdrawn, glucocorticoid treatment needs to be continued for several months.

2. Orbital radiotherapy. External radiotherapy has been used for GO for almost 60 yr and still represents a mainstay in the management of the disease (116). It was initially directed to the hypothalamus and the pituitary, based on the assumption that the ophthalmopathy might be due to an exophthalmogenic factor of pituitary origin or to hypothalamic dysfunction (116). Subsequently, irradiation was correctly directed to the orbital tissue, the true target of the pathological process.

The rationale for the use of radiotherapy for GO resides both in its nonspecific antiinflammatory effect and in the high radiosensitivity of lymphocytes infiltrating the orbital space (117). Lymphocytes are generally suppressed with relatively low doses of radiation, and the helper/suppressor T lymphocyte ratio is also altered by radiotherapy (116). In addition, radiotherapy might also reduce GAG production by orbital fibroblasts (118). Whether the reported effectiveness of orbital radiotherapy in GO is related either to its nonspecific antiinflammatory action, or to specific immunosuppressive effects, or both remains to be clarified.

Many of the limitations encountered with the old orthovoltage apparatus, such as the low energy and the relevant side scatter of irradiation, were overcome by the introduction of high-energy apparatus (cobalt unit and, especially, linear accelerator), which allowed a better collimation, a limited side scatter, and low penumbra (119). Donaldson and co-workers (102) were the first to use a 4–6-megavolt linear accelerator in a group of 23 patients with severe GO, who had an OI ranging from 4 to 12. Excellent or good results were obtained in 15 patients (65%), even in those who had previously responded poorly to systemic glucocorticoid treatment; results were less favorable in patients with longstanding eye disease and more frequently satisfactory in patients with a rapid progression of eye disease (102). While all categories of ocular manifestations responded to radiotherapy, long lasting extraocular muscle involvement appeared to be least responsive (102). Beneficial effects, especially on soft tissue changes and optic neuropathy, have been reported in other subsequent studies, while the reduction in proptosis and the improvement in ocular motility, especially if longstanding, have often been less impressive (120). In a large series of 311 patients treated with different radiation doses, factors that apparently influenced the outcome of radiotherapy in a negative manner included male gender, advanced age, need for concomitant treatment of hyperthyroidism, and no history of hyperthyroidism (121).

At present, most centers utilize linear accelerators delivering 4–6 megavolts and use a 4 x 4-cm lateral field slightly angled posteriorly to avoid as much as possible irradiation to the contralateral lens. The use of higher energy sources have not proven to be particularly advantageous. The most common delivered dose is 20 grays (Gy) per eye (117); this cumulative dose is usually fractionated in 10 daily doses over a 2-week period to reduce the cataractogenic effect of irradiation (116). Nakahara et al. (122) reported that a cumulative dose of 24 Gy provided better results than a dose of 10 Gy. Recently, Kahaly et al. (123) reported that a therapeutic scheme of 1 Gy per week over a 20-week period was equally effective and possibly better tolerated than the classical 2-week scheme. The use of higher cumulative doses of radiation (30 Gy vs. 20 Gy) does not produce any increase in the effectiveness of treatment (121). Thus, at present it would appear that the dose of 20 Gy should be considered the optimal dose for orbital radiotherapy of GO.

Orbital radiotherapy is usually well tolerated. It may be associated with a transient exacerbation of inflammatory eye signs and symptoms (102), but this is unlikely to occur if glucocorticoids are concomitantly administered (124). Cataract is a possible complication of irradiation to the lens, but fractionation of the dose should maintain the radiation exposure of the lens below the threshold dose for radiation-induced cataract (116). Radiation retinopathy is an extremely rare complication of radiotheraphy (125); probably errors in dosage calculation and radiation technique account for most of the few reported cases (126), but some cases remain unexplained (127). Systemic microvascular disease due to diabetes mellitus or to previous chemotherapy may increase the risk for radiation retinopathy. Tallstedt et al. (128) reported increased retinopathy in all 3 patients who had diabetic retinopathy before radiotherapy. Thus, although there is no consensus view on this issue, it is reasonable to regard this condition as a contraindication to orbital radiotherapy for GO (117). Transient blindness attributed to nonvascular involvement of the optic nerve was reported in one patient (129).

A major concern relates to the possibility that orbital radiotherapy may be carcinogenic. Snijders-Keilholz et al. (130) calculated a theoretical risk of 1.2% for the occurrence of secondary tumors. This view is not shared by other authors (131), who estimated a theoretical risk of 0.3%. A proper answer to this question should be provided by a careful reevaluation of patients with a long (15–20 yr or more) follow-up period after irradiation. So far, no case of secondary tumor after orbital radiotherapy for GO has been reported in the literature (117). Nevertheless, it seems prudent to avoid irradiation in young (possibly <30 yr) patients.

In summary, the available studies have reported, with few exceptions, overall favorable effects of orbital irradiation in about 60% of GO patients (Table 8). It should be mentioned that in a recent double-blind prospective study in which 42 patients received orbital radiotherapy only to one orbit, no significant differences were observed between the treated and the untreated orbit (132). It is possible that these negative results are related to the selection of patients, because only patients with mild-to-moderate GO were enrolled (132).

3. Orbital radiotherapy combined with glucocorticoids. Orbital radiotherapy and systemic glucocorticoids can be used for GO either alone or in combination. Prummel and co-workers (112) reported in a randomized study that orbital radiotherapy and oral glucocorticoids, used as a single therapeutic agent, had similar effectiveness on the ophthalmopathy. This led the Dutch group to propose that orbital radiotherapy be considered the treatment of choice in patients with moderately severe GO, i.e., without sight-threatening manifestations, in view of its better tolerability compared with glucocorticoid therapy (133). In two randomized, prospective studies we showed that orbital radiotherapy combined with high-dose oral glucocorticoids was more effective than either orbital radiotherapy alone (134) or oral glucocorticoids alone (124). In addition to these synergistic effects, the combined regimen exploits the prompter effects of glucocorticoids and the more sustained action of irradiation. The inclusion of glucocorticoids prevents radiation-associated transient exacerbation of ocular manifestations, while the inclusion of orbital radiotherapy probably reduces the prevalence of recurrences of eye disease, not infrequently observed when glucocorticoids are withdrawn. Thus, we suggest that in patients with severe GO, defined according to the criteria indicated in Table 1, this combined therapeutic regimen should be employed, if conservative therapy, rather than orbital decompression, is selected.

4. Orbital decompression. Orbital decompression is, with glucocorticoid therapy and orbital radiotherapy, a milestone in the treatment of GO. Its aim is to provide, through the removal of part of the bony components of the orbit, an increased space for the increased orbital content (1). Although it does not act on the pathogenetic mechanisms of the ophthalmopathy, it is very effective on proptosis and on the other ocular manifestations caused by venous congestion.

In the past, the surgical approach to the treatment of GO was limited by the risks of surgery and, therefore, indications for decompression were mostly represented by marked proptosis and by optic nerve compression, especially if no beneficial effect was obtained with glucocorticoids and/or orbital radiotherapy. In reviewing the records of 428 consecutive eye surgery patients at the Mayo Clinic, Garrity et al. (135) noted that the indications for orbital decompression were: optic neuropathy in 217 patients (51%), severe orbital inflammation in 116 (27%), proptosis in 90 (21%), and glucocorticoid side effects in 5 (1%). However, rehabilitative (cosmetic) surgery represented the indication for orbital decompression in several studies (136, 137, 138), as well as in 20% of cases in a survey of American ophthalmologists (139).

Several techniques, aimed at removing portions of one to four walls of the orbit, have been used (1). The four-wall technique is rarely used and may be considered in cases of very severe ophthalmopathy. The lateral approach is of limited effectiveness, because the removal of the lateral wall alone is usually associated with a limited decrease in proptosis (1). The superior (transfrontal) approach removes the roof of the orbit and is effective, but nowadays it is rarely used because of the risks associated with this procedure, i.e., intracerebral hemorrhage, damage to the frontal lobe, meningitis, and sensation of pulsation behind the globes (140).

The inferior (transantral) approach is still very popular. This technique has been modified to remove also the lateral wall of the orbit, although it is mainly aimed at the floor and the medial wall of the orbit (140). In a large series from the Mayo Clinic, Garrity et al. (135) reported that 402 of 453 eyes (89%) with visual acuity worse than 20/20 improved or remained the same; defects of the visual field ameliorated or regressed in 245 of 269 eyes (91%), preoperative papilledema was reduced in 99 of 105 eyes (94%), and preoperative exposure keratitis improved in 178 of 195 eyes (92%). The average decrease in proptosis was 4.7 mm and was sustained over an extended follow-up period (135). This technique has the advantage that there is no external scar, and decompression of the orbital apex, where the optic nerve is mostly compressed, is very effective (140). Complications are not infrequent. In a large series (135), sinusitis occurred in 18 patients (4%), lower eyelid entropion in 38 (9%), numb lip in 23 (5%), cerebrospinal fluid leakage in 15 patients (3.5%), and frontal lobe hematoma in 1 (0.2%). The major drawback of this procedure is the high incidence of postoperative diplopia, which may affect up to two thirds of patients with no diplopia before surgery (135). Nunery et al. (141) noted that postoperative diplopia was rare among patients who had no preoperative diplopia (1 of 25 patients, 4%), while worsening of preexisting diplopia was very frequent (22 of 36 patients, 61%). Tallstedt (140) noted an incidence of new postoperative diplopia in 32 of 63 patients (51%) operated on by the transantral approach at the Karolinska Hospital.

Removal of the floor and the medial wall can also be accomplished by an anterior approach through a transconjunctival or translid incision (140). This approach appears to be associated with a lower incidence of worsening of diplopia, because it is more difficult to remove the posterior part of the ethmoid, thus avoiding the prolapse of the posterior portion of the orbital content (140). In a review of American ophthalmologists, this technique was associated with a 6% worsening of diplopia, compared with 41% after transantral decompression (139). Postoperative worsening of diplopia was observed in 5 of 33 patients (15%) evaluated by Hutchinson and Kyle (142) after a two-wall operation using the translid approach. The translid approach is simpler and less morbid than the transantral technique, but it seems to be associated with a lower recession of the proptosis (143).

Removal of portions of three walls (floor, medial, and lateral walls) can be accomplished by either combining the transantral (or translid) decompression with lateral decompression or by the coronal approach. In the latter, a skin muscle incision is made from ear to ear 1 cm behind the hair border; after incision of the periosteum, the subgaleal flap is turned down to the supraorbital rim, the periorbita is incised in all quadrants, and then the lateral wall, most of the ethmoid, and the medial portion of the floor are removed (138). This technique was reported to be associated with a greater reduction of proptosis and lower prevalence of postoperative diplopia compared with the transantral technique (138). It was observed that balancing the decompression and preserving the medial orbital strut between the ethmoid sinus and the orbital floor may minimize the risk of postoperative diplopia (144). In a large, retrospective study (138), the mean decrease in proptosis was 4.3 mm (range 0–10 mm), the largest reduction being observed in those patients who had the highest (>27 mm) preoperative Hertel readings. Complications were limited to 13 cases (10%), including damage of the infraorbital (n = 2) or supraorbital (n = 1) nerves, temporary unilateral hypoesthesia (n = 6), enophthalmos (n = 1), and asymmetry (n = 3) (138). Thus, it seems that the coronal decompression technique is safe and effective and bears a low risk of postoperative diplopia compared with the two-wall transantral approach. In patients with severe optic nerve involvement, it may be less advantageous than the transantral technique, because the orbital apex is less effectively decompressed by the coronal approach.

A different approach to decompress the orbital content may be represented by the removal of orbital fat through medial-upper and lateral-lower anterior orbitotomy. An average decrease in proptosis of 1.8 mm was found (range 0–6 mm), the largest average reduction (3.3 mm) being observed in patients with preoperative Hertel readings greater than 25 mm (145). Side effects were limited to temporary motility impairment of the inferior oblique muscle in two patients (145). With few exceptions (146, 147), this procedure seems to produce a rather limited decrease in proptosis. In mild to moderate cases, orbital lipectomy was associated with eyelid surgery, with good esthetic and functional improvement and no complications (148).

In summary, orbital decompression is a very effective therapeutic procedure for GO. It is beneficial for most expressions of the disease, with particular regard to proptosis and optic neuropathy, but also to congestive manifestations of the disease. The choice between medical and surgical treatment of ophthalmopathy relies, among other factors, on the availability of a skillful orbit surgeon. Increasing expertise in this field has expanded the indications for orbital decompression, which is currently carried out not only for sight-threatening conditions, but also for rehabilitative purposes. The selection of the different surgical techniques depends not only on the experience of the orbit surgeon, but also on the clinical situation of the patient. If optic nerve compression is severe, the transantral approach is probably better, because it allows a more marked decompression of the nerve at the orbital apex. The three-wall coronal approach is preferable in patients who do not have severe optic neuropathy and do not have preoperative diplopia, because the risk of new postoperative diplopia is lower than with the transantral approach. Irrespective of the surgical technique, orbital decompression does not solve the problem of preoperative diplopia, and a relevant proportion of patients will need extraocular muscle correction surgery (see below).

B. Nonestablished treatments
1. Cyclosporine. The autoimmune origin of GO prompted the attempt to use immunosuppressive drugs for this disease. The experience is, however, limited to small series of patients treated with azathioprine, cyclophosphamide, or the immunomodulatory agent, ciamexone, usually in uncontrolled trials (6).

The immunosuppressive drug that has been more thoroughly evaluated in the management of GO is cyclosporine. This drug affects both humoral and cell-mediated immune reactions, since it inhibits cytotoxic T cell activation and antigen presentation by monocytes and macrophages, but it also induces activation of T suppressor cells and inhibits production of cytokines (149). Although cyclosporine seems to be more effective on the early immune response (e.g., after organ transplantation) than on an already established immune response (e.g., in autoimmune disease), the above actions might explain the observed effectiveness in autoimmune diseases, especially if of recent onset (150).

Several reports have evaluated the effectiveness of cyclosporine administration in GO. Although the initial report showed a dramatic improvement of ocular conditions in 2 patients treated with the drug (151), these positive effects of cyclosporine were not uniformly confirmed in later studies (6). In a controlled, randomized, and prospective study, Kahaly et al. (111) compared the effects of oral prednisone with those of oral prednisone combined with cyclosporine; prednisone was stopped in both groups after 10 weeks. Inflammation regressed in both groups, but proptosis decreased more in the cyclosporine-prednisone group; likewise, diplopia ameliorated more effectively in this group (111). In addition, the favorable effects of treatment were more persistent in the cyclosporine-prednisone group, since only 1 of the 20 patients of this group had a relapse compared with 8 patients in the other group (111). Side effects attributable to cyclosporine were rather frequent in this study, including one case of Klebsiella pneumonia, four cases of hypertension, four cases of increased liver enzymes, and several minor effects, such as hirsutism, paresthesias, and swelling of the gums; however, all appeared to be reversible (111).

In the other randomized trial on the effects of cyclosporine on GO, two groups of 18 patients each were treated with either cyclosporine or prednisone (103). During the 12-week period of treatment, a response, as assessed by a decrease in the extraocular muscle enlargement, a decrease in proptosis, an improvement in visual acuity, and a decrease in total eye score, was observed in 11 patients treated with prednisone and only in 4 patients treated with cyclosporine (103). Interestingly, retreatment of nonresponders of both groups with a combination of the two drugs (using a lower dosage of prednisone) was often associated with a therapeutic response (103). In this study, cyclosporine was better tolerated than prednisone, but 6 cases of hypertension, 1 case of diverticulitis requiring drug withdrawal, and 1 case of irreversible rise in serum creatinine levels could be attributed to cyclosporine (103).

In summary, the use of cyclosporine has been reported in several studies, but only two of them (103, 111) were randomized and controlled. Thus, the favorable effects of cyclosporine reported in some uncontrolled studies must be interpreted with caution. The study by Prummel et al. (103) indicated a lower efficacy of cyclosporine compared with prednisone as a single-agent treatment, but both Prummel et al. (103) and Kahaly et al. (111) suggested that a combination of cyclosporine and prednisone may be more effective than either treatment alone. Thus, the use of cyclosporine might be maintained, in association with glucocorticoids, in patients who are resistant to glucocorticoids alone and in whom the persistent activity of the disease warrants a continuing medical intervention. Side effects of cyclosporine are not negligible; some of them can be severe, calling for caution in the use of this drug (e.g., doses lower than 7.5 mg/kg/day).

2. Plasmapheresis. The rationale for the use of plasmapheresis in the treatment of GO was represented by the assumption that this procedure might remove either immunoglobulins or immune complexes possibly involved in the pathogenesis of the disease, reproducing the beneficial effects observed in other autoimmune disorders. In addition, plasmapheresis might affect plasma viscosity and complement components.

Favorable results with this procedure were observed, in the first report, in 4 of 7 patients with rapidly progressive ophthalmopathy; the three treatment failures were attributed to the long duration of the disease with the likely associated fibrotic changes (152). Likewise, Glinoer et al. (153) reported a marked clinical improvement in 8 of 9 patients, especially regarding soft tissue changes, proptosis, intraocular pressure, and visual