Shannon S. Joseph
Shannon S. Joseph, M.D., M.Sc. is an oculoplastic and orbital surgeon. She completed her medical school and residency training at the Johns Hopkins University School of Medicine, followed by an oculoplastic and orbital surgery fellowship accredited by the American Society of Ophthalmic Plastic and Reconstructive Surgery at the University of Michigan.
Neil R. Miller
Dr. Neil Miller is currently Professor of Ophthalmology, Neurology, and Neurosurgery at the Johns Hopkins Medical Institutions and also the Frank B. Walsh Professor of Neuro-Ophthalmology. Dr. Miller was President of the North American Neuro-Ophthalmology Society from 2000 to 2002 and Chairman of its Executive Board from 2002 to 2004. He has been past President of the International Neuro-Ophthalmology Society on three occasions: in 1982, 1992, and 2008. He is also an emeritus member of the International Orbital Society. In 2009, he was given the Lifetime Achievement Award by the American Academy of Ophthalmology (AAO). He has authored or co-authored 500 articles, 90 chapters, and 13 books in the fields of neuro-ophthalmology and orbital disease. Many of Dr. Miller’s previous fellows and residents hold faculty positions at major institutions throughout the United States and around the world.
Dysthyroid optic neuropathy (DON) is one of the most serious complications of thyroid eye disease (TED). It occurs in 4–8.6 % of TED patients and can lead to permanent vision loss [1–3]. DON is defined as the impairment of optic nerve function in TED caused by the mechanical compression or distortion of the optic nerve . This chapter will focus on the management options for DON, but we will first review its disease mechanism, clinical presentation, and diagnostic process.
Mechanism of Disease
TED is an autoimmune phenomenon that involves both antigen-dependent and antigen-independent immune systems (Fig. 28.1).
(a, b) Pathogenesis of thyroid eye disease
The fundamental change that underlies most clinical manifestations of TED, including DON, is enlargement or expansion of orbital soft tissues [4–7]. The pathogenic processes leading to this soft tissue enlargement remain under investigation; however, current theories point to the orbital fibroblasts (OF) as the principal effector cells of this process [4, 5, 8]. The orbit contains two subpopulations of OFs: Thy1+ and Thy1- cells, each of which contributes to orbital soft tissue enlargement in TED in a unique manner. Thy1+ OFs overexpress the cell surface marker Thy1 (CD90) in TED. Upon stimulation by pro-inflammatory cytokines, such as interleukin-1β, these OFs proliferate and upregulate specific hyaluronan synthases, resulting in increased hyaluronan secretion [5, 9–12]. Hyaluronan is a hydrophilic glycosaminoglycan (GAG) that accumulates within the orbit and draws water into the extraocular muscles, orbital fat, and connective tissues, leading to swelling [9, 13–15]. Thy1- OFs, which do not express Thy1, are pre-adipocytes that can differentiate into mature adipocytes, leading to expansion of orbital fat . In addition to the OFs, lymphocytes and macrophages contribute to orbital soft tissue swelling by infiltrating the tissue and generating inflammatory edema [17, 18]. Together, these mechanisms lead to the orbital soft tissue enlargement seen in TED.
As soft tissue volume increases within the rigid bony orbital walls, intraorbital pressure also increases, and mechanical compression occurs [18, 19]. When this compression affects the optic nerve, usually in the orbital apex, it can cause DON . In some patients, the rise in orbital pressure leads to proptosis that, in turn, may serve as spontaneous orbital decompression, thus decreasing the risk for DON . Proptosis typically occurs in patients with eyelid laxity and/or minimal orbital soft tissue fibrosis . Conversely, patients with less eyelid laxity and more orbital fibrosis are less likely to have proptosis and more likely to have greater intraorbital pressure, resulting in an increased risk for DON . Indeed, several studies have demonstrated that patients with DON typically do not have marked proptosis [1, 3, 20, 22].
As noted above, the location at which optic nerve mechanical compression occurs in the TED orbit is widely accepted to be the orbital apex. This theory is well supported by evidence from both radiographic imaging studies and histopathological studies [1, 6, 18, 20, 23–29]. Trokel et al. were the first to demonstrate using computed tomographic (CT) scanning crowding of the enlarged extraocular muscles at the orbital apex in patients with DON . This significant correlation between apical crowding on CT and DON has since been confirmed in numerous studies [1, 23, 26–29] (Fig. 28.2a, b).
(a, b) Axial (a) and coronal (b) CT images show Apical crowding in a patient with thyroid eye disease and bilateral optic neuropathy
Nevertheless, what established the clinicopathological correlation between apical crowding and DON were the histopathological findings of exenterated orbital contents from a patient with TED . In this study, Hufnagel et al. used immunohistochemical stains to confirm the presence of partial optic nerve atrophy in their specimen. More importantly, they observed that the axon loss was most pronounced in the sections of the optic nerve from the orbital apex . They noted no evidence of inflammatory infiltrate of the optic nerve . Taken together, current evidence suggests that DON is caused by mechanical compression or distortion of the optic nerve at the orbital apex due to enlargement of orbital soft tissue. DON thus is a form of compressive optic neuropathy, and the goal of management for DON should be to decrease orbital soft tissue enlargement and/or increase the space in which the orbital soft tissue reside, thereby relieving the intraorbital pressure and compression, especially at the orbital apex.
Clinical Presentation and Diagnosis
Compressive optic neuropathies typically present with decreased visual acuity, dyschromatopsia, visual field defects, a relative afferent pupillary defect (RAPD) (unless the condition is bilateral), and either normal-appearing optic disks or optic disk swelling . The onset of DON is highly variable [1, 32–34]; it can be acute or insidious , progressing over days, weeks, or even months. Other ocular comorbidities, such as exposure keratopathy, glaucoma, and cataract, may be present as confounding factors [1, 32, 33], sometimes resulting in delayed diagnosis (Fig. 28.3a, b).
(a) Corneal ulceration causing decreased vision in the left eye of a patient who also had a left optic neuropathy. (b) Same patient from (a). Non-contrast CT scan, coronal view. Note crowding in the left orbital apex but not in the right
To identify the most diagnostically useful clinical features of DON, the European Group on Graves’ Orbitopathy (EUGOGO) analyzed the clinical presentation of 94 eyes in 47 patients in a multicenter study . Fifty-five of these eyes had definite DON. Of these 55 eyes, 80 % had reduced visual acuity compared with 32 % in those without DON, 77 % had dyschromatopsia compared with 7 % in those without DON, 45 % had an RAPD compared with none in those without DON, and 56 % had optic disk swelling compared with 5 % in those without DON . Even though an RAPD and optic disk swelling were not very sensitive findings, they were highly specific, rendering them diagnostically valuable signs. Collectively, current evidence suggests that visual acuity, color vision, pupillary reaction to light, and optic disk appearance are arguably the four most important element of the clinical exam in diagnosing DON [3, 36, 37]. Other clinical features such as extraocular movement abnormalities, proptosis, and the clinical activity score (CAS, a score based on classic subjective features of inflammation such as swelling, redness, and pain)  have lower sensitivity and specificity in diagnosing DON. The EUGOGO survey found that extraocular movement restriction was slightly more common in eyes with DON than eyes without DON (71 % versus 52 % for upgaze, 18 % versus 0 % for downgaze, 33 % versus 5 % for abduction, 15 % versus 15 % for adduction) , but there was no significant difference in the amount of proptosis between eyes with DON and eyes without (62 % versus 63 %) . As previously mentioned in this chapter, numerous studies have shown that proptosis does not correlate with the presence or severity of DON [3, 20, 22, 23, 36, 39]. The CAS for eyes with DON was slightly higher, but 39 % of patients with definite DON had a CAS less than 3 . This finding is consistent with the previously reported observation that even though DON is, by definition, indicative of severe TED, patients with DON often lack or at least have relatively mild soft tissue stigmata of TED [1, 2, 20, 36, 40]. Resistance to retropulsion on clinical exam has been suggested to correlate with elevated intraorbital pressure and, therefore, DON [21, 41]; however, the sensitivity and specificity of this finding have not been assessed. In summary, although assessment of ocular motility and alignment, amount of proptosis, CAS, and degree of resistance to retropulsion are important elements of the clinical exam when assessing patients with known or suspected TED, they may not be the most sensitive or specific features for DON.
When DON is suspected based on clinical findings, or when the etiology is unclear, ancillary testing is necessary to ensure accurate diagnosis. Three ancillary testing modalities have been most commonly used in the diagnosis of DON: orbital imaging, perimetry, and visual evoked potentials (VEPs). We will discuss each of these tests in more detail, with a special focus on orbital CT scanning, one of the best diagnostic tests currently available for DON.
Magnetic resonance (MR) imaging, ultrasonography, and CT scanning are all commonly used to evaluate patients suspected of having TED. MR imaging provides the most detailed imaging of the orbital soft tissue anatomy and is useful in assessing if interstitial edema and evidence of active inflammation are present [42–44]. Ultrasonography is an economical and safe method of evaluating orbital soft tissue enlargement as well as changes in internal reflectivity of the extraocular muscles [42–44], but it requires an experienced ultrasonographer. In addition, a major disadvantage of both modalities is their inferiority to CT scanning in imaging bony structures [42–44]. DON occurs from mechanical compression or distortion of the optic nerve at the orbital apex. Therefore, orbital CT scanning, which provides precise imaging of the orbital bony structures, including the orbital apex, is the imaging modality of choice not only for the diagnosis of DON but also for determining its management [42–44].
Orbital CT scans to evaluate for DON are performed without intravenous contrast, using a spiral CT technique with 2-mm-thick slices in the axial plane. Coronal and sagittal images should be reconstructed . Imaging features such as extraocular muscle enlargement, orbital fat compartment enlargement, orbital fat prolapse, proptosis, lacrimal gland displacement, superior ophthalmic vein dilation, medial bowing of the lamina papyracea, and apical crowding have all been described in patients with DON [1, 6, 23–29, 45–47]. Two of these features, extraocular muscle enlargement and apical crowding, have been shown to have the most reliable predictors of DON (Fig. 28.2a, b).
Numerous studies have documented a correlation between extraocular muscle enlargement on orbital CT scanning and DON [6, 23, 24, 27, 40, 45–47]. Feldon et al. performed the first quantitative volumetric assessment demonstrating that TED patients with DON have significantly higher extraocular muscle volume compared with those without DON . To quantify extraocular muscle enlargement more reproducibly, Barrett et al. used a “muscle index” (MI): the percentage of orbital width occupied by the extraocular muscles measured in the plane at the midpoint between the orbital apex and the posterior globe . Sensitivity and specificity varied depending on the level of MI used. The best combination was found to be for MI of 60 %, which was 79 % sensitive and 72 % specific for DON .
Another CT parameter that has been shown to be an excellent predictor of DON is optic nerve crowding at the orbital apex [1, 3, 23–29, 42]. Nugent et al. proposed a subjective grading scale for apical crowding based on coronal imaging at the apex: grade 0 represents no effacement of perineural fat planes by enlarged extraocular muscles; grade 1 represents 1–24 % of effacement; grade 2 represents 25–50 % of effacement; and grade 3 represents greater than 50 % effacement . Studies have shown that grade 3 apical crowding on CT scanning has sensitivity ranging from 62 to 80 % and specificity ranging from 70 to 91 % [26, 28, 29]. With the advent of multidetector CT (MDCT) scanners, quantitative measures and precise image reformatting have become readily available. Goncalves et al. developed a more objective measure of apical crowding using MDCT: the orbital crowding index (CI) . This is a ratio between the square area measurements of the extraocular muscles and the orbital bone area, measured at three well-defined coronal planes . The best combination of sensitivity and specificity (92 % and 90 %, respectively) was found for a CI of 0.575 at 18 mm from the interzygomatic plane .
Collective evidence shows that orbital CT scanning is one of the best diagnostic tools currently available for DON. Every patient suspected of having DON based on the clinical examination, or for whom the diagnosis is unclear but DON is a possibility, should undergo this imaging technique. For patients with DON who require surgery, orbital CT scanning is also useful for preoperative planning, postoperative assessment, and, at some centers, intraoperative stereotactic navigation (see below) [43, 50].
Other ancillary tests that have been used to facilitate the diagnosis of DON include static perimetry and VEPs. It is well established that patients with DON can present with a variety of visual field defects on perimetry. Central and paracentral scotomata tend to be the most common, but arcuate and altitudinal scotomata have also been observed [34, 36, 37, 51, 52]. The EUGOGO multicenter survey reported that 71 % of eyes with definite DON had abnormal visual fields, compared with 13 % in eyes without DON, although the method of visual field testing was not standardized in this study . Unfortunately, a myriad of ocular conditions can produce field defects, including exposure keratopathy, lenticular changes, glaucoma, other causes of optic neuropathy, retinal disorders, lid malposition, and refractive error. Many of these can be comorbidities of TED. Indeed, up to 70 % of TED patients without DON have also been reported to have reproducible visual field defects, many of which are also central or paracentral . Moreover, static perimetry is also subject to non-ophthalmologic factors such as technician errors, patient test-taking errors, and cognitive factors (usually on the part of the patient). In summary, static perimetry may be informative and should be obtained as part of the assessment of a patient suspected to have DON, but it is not very sensitive and is definitely not specific for DON.
Another test that has been used for both the diagnosis and monitoring of DON is VEPs. Patients with DON often have abnormal VEP latency and amplitude [1, 3, 54–59]. Several studies have reported on the remarkably high sensitivity of abnormal VEPs for DON [1, 3, 54–59]. For example, Neigel et al. observed that VEPs were abnormal in 94 % of 58 patients with DON versus 9.1 % in a control group . Comparatively, the sensitivity of other tests were much lower in their study, including decreased visual acuity (52.6 %), RAPD (35 %), dyschromatopsia (64 %), and visual field defect (66 %) . The most common changes in VEPs observed in DON are decreased amplitude and increased latency of both the P2 and the P100 peaks [54, 55, 57, 59]. Interestingly, after patients undergo treatment for DON with either steroids or surgical decompression, their VEPs tend to normalize, with the amplitudes improving more than the latencies [55, 57]. Nevertheless, although VEPs may be highly sensitive in facilitating the diagnosis of DON and monitoring patients after treatment, the test results are not very specific. Factors such as opaque media, uncorrected refractive error, and other causes of optic neuropathy can affect results. Furthermore, the testing equipment is not readily available, so VEPs have not been widely adopted as a part of the routine diagnostic process for DON.
In summary, timely and accurate diagnosis is imperative to the successful management of DON . Most cases of delayed diagnosis of compressive optic neuropathy are caused by failure to perform color vision or visual field testing, failure to check for an RAPD, and failure to obtain appropriate imaging studies . There currently are no standardized diagnostic guidelines for DON. Based on the currently available evidence, however, all patients with DON should have periodic complete ophthalmologic assessments, with special attention to color vision testing, visual fields, pupillary responses to light stimulation, and the appearance of the optic disks. In cases concerning for DON, orbital CT scanning should be obtained to assess for the degree of enlargement of the extraocular muscles as well as apical crowding. If VEPs are available, they may yield additional information. If visual impairment is confirmed based on these tests and no other identifiable etiology can account for the impairment, then the diagnosis of DON is likely.
Management of DON
Dysthyroid optic neuropathy, once diagnosed, mandates urgent and, sometimes, emergent, treatment. Once optic neuropathy develops, permanent visual loss may result, and by the time optic atrophy becomes apparent, visual recovery is unlikely. Hence, early and accurate diagnosis and treatment of DON may improve a patient’s visual prognosis.
Risk Factor Modification
The risk factors of DON include male gender, older age, smoking, and significant comorbidities such as diabetes mellitus (DM) . One of the first steps in the management of DON is modification of risk factors. Smoking is a key modifiable behavioral risk factor for TED and DON . Among patients with Graves’ disease, smokers are more likely to develop TED (odds ratio 7.7, 95 % CI 4.3–13.7)  than nonsmokers, and smokers with TED also tend to have more severe disease and poorer response to treatment [62–64]. Bartalena et al. reported that in 150 patients with severe TED, 93.8 % of nonsmokers responded to treatment (high-dose oral prednisone followed by orbital radiotherapy), whereas only 68.2 % of smokers responded . Similarly, Eckstein et al. observed that smoking influences the treatment response of TED in a dose-dependent manner, with delayed and decreased therapeutic response in smokers . Therefore, to optimize response to therapy and prevent further progression of disease, smoking cessation should be discussed upon diagnosis and remain an integral part of management for all patients with TED and DON.
Diabetes mellitus also is an important risk factor for DON. Neigel et al. reported that 15.5 % of patients in their DON group were diabetics, compared with 1.7 % in their control group . Similarly, Kalmann et al. noted that DON developed in 33.3 % of TED patients with DM compared with 2.9 % in the TED patients without DM . TED patients with DM also seem to have a more recalcitrant course, developing more severe DON with a worse visual prognosis [1, 37, 65]. One proposed mechanism underlying this relationship is that diabetic microvascular changes in the optic nerve may sensitize it to compression . No studies have evaluated whether or not there is a dose-response relationship between the severity of DM and the progression or therapeutic response of DON; however, upon diagnosing DON in a patient with DM, it would be prudent to alert the patient’s primary care provider of this diagnosis and emphasize to the patient the importance of glycemic control.
Systemic Thyroid Status
A major systemic factor that must be considered when managing patients with TED is their systemic thyroid status. Although patients may be hyperthyroid, hypothyroid, or euthyroid when they first develop symptoms and signs of TED, TED typically presents simultaneously or within 18 months of the onset of hyperthyroidism [66, 67]. No correlation between the level of thyroid dysfunction and DON has been reported in the literature, but evidence suggests that TED patients who are persistently dysthyroid have more severe eye disease than those who are euthyroid . Therefore, the restoration and maintenance of normal thyroid function should be a key treatment goal in every patient with TED.
Three treatments are used to normalize thyroid function: radioablation using iodine-131, antithyroid medication, and total or partial thyroidectomy. Two large randomized controlled trials have shown that iodine-131 treatment is more likely than other forms of antithyroid therapy to be followed by the development or worsening of TED [68–73]. The mechanism underlying this phenomenon is likely related to the destruction of thyroid tissue by radioactive iodine therapy leading to the release of thyroid antigens that are shared between the thyroid and the orbital soft tissues, which subsequently exacerbates the autoimmune reaction within these tissues . The risk factors for developing or worsening of TED triggered by radioablation include high pretreatment serum T3 levels (more than 5 nmol), high serum thyrotropin binding inhibiting immunoglobulins (TBII) levels, and an active smoking history [69, 70]. This adverse effect of radioactive iodine therapy may be prevented or at least reduced by concomitant administration of low-dose (0.5 mg/kg/day) systemic oral steroids beginning 1 week prior to radioablation and continuing for several months following radioablation . In summary, euthyroidism should be the goal in all patients with TED. For patients who already have TED or who do not have TED but have the risk factors described above and in whom radioablation is to be performed, we recommend pre- and posttreatment with low-dose systemic corticosteroids.
Treatments for DON
Three major treatment modalities may be used to treat DON: systemic glucocorticoids, orbital decompression, and orbital radiotherapy. Some patients may require a combination of these therapies. Because DON is rare, randomized controlled trials to assess the efficacy of the various treatment modalities are difficult to conduct. Most of the current literature consists of retrospective studies and will be reviewed below.
Glucocorticoids (GCs) have been used to treat TED for more than half a century [39, 74, 75]. GCs have anti-inflammatory properties and, at higher doses, are immunosuppressive. Specifically, GCs suppress lymphocytes, hinder recruitment of monocytes and macrophages, and inhibit the release of inflammatory mediators . Not surprisingly, patients with more active orbital inflammation tend to be more responsive to GC therapy than those with long-standing or inactive disease associated with fibrotic orbital soft tissue . In addition, GCs inhibit glycosaminoglycan synthesis in fibroblasts . By reducing the inflammation and swelling of the orbital soft tissues, GCs decrease orbital soft tissue and extraocular muscle enlargement, reduce apical crowding, and relieve compression of the optic nerve.
Both local and systemic GC treatments for patients with TED with and without DON have been used. Local treatment consists of retrobulbar and subconjunctival injections of steroids . The efficacy of this treatment for DON has not been extensively studied, but given the site of the compression, the risk of tissue injury from multiple intraorbital injections, and the large doses of systemic GCs usually required to control DON, we and others do not recommend this route of administration . Instead, systemic GC administration is the route of choice for DON. The oral regimen routinely requires an initial large daily dose (1–2 mg/kg/day) with a prolonged taper over several months. As might be expected, this regimen is associated with significant systemic side effects . Over the past two decades, five randomized controlled trials (RCTs) have shown that intravenous (IV) administration of glucocorticoids is superior to oral administration in treating both DON and severe TED without DON and is associated with fewer side effects . Thus, the IV route is preferred over the oral route by most physicians .
For patients with DON, IV administration of GCs is typically given as daily doses of methylprednisolone (MP) for 3–5 days, following which a decision is made as to whether to continue daily or weekly IV administration or switch the patient to oral steroids. This regimen is similar to that used for other autoimmune or inflammatory conditions, such as systemic lupus erythematosus, rheumatoid arthritis, and giant cell arteritis [83–85], and it can lead to rapid visual recovery in patients with DON [86–89]. Guy et al. reported five patients with severe DON treated with IV MP, 1 g/day, for 3 days. All five patients experienced significant improvement in visual acuity, color vision, and visual field defects immediately after the completion of therapy . A larger retrospective study by Mourits et al. examined the longer-term efficacy of IV pulse therapy for DON . Sixty-two consecutive patients with DON were treated with four courses of IV MP, 500 mg every other day, followed by either an oral prednisone taper or orbital radiotherapy. Vision, proptosis, supraduction, and CAS all improved significantly by the first day after completion of therapy. Changes in other findings of DON such as visual field defects, dyschromatopsia, presence or absence of an RAPD, disk appearance, and results of VEPs were not documented in this study; however, by the time the disease had been stable for at least 6 months, 39 % of patients retained normal vision, whereas 61 % had to undergo surgical decompression for persistent or recurrent DON .
These retrospective studies support the short-term efficacy of IV pulse therapy with GCs for DON; however, if a majority of patients treated in this manner eventually have to undergo surgical decompression, should the first-line therapy be surgery instead of IV GCs? Wakelkamp et al. conducted a small randomized controlled trial to answer this question . Nine patients were randomized to IV MP pulse therapy followed by an oral taper over 4 months. Six patients were randomized to immediate three-wall orbital decompression surgery. At 52 weeks, there was improvement in visual acuity, “total eye score,” and CAS in 56 % of the steroid group compared with 17 % in the surgery group. Moreover, although 56 % of patients in the steroid group eventually required surgical decompression or orbital radiotherapy, 83 % of the surgery group required subsequent steroid treatment, orbital radiotherapy, or both. Therefore, immediate surgery has no advantage over initial pulse therapy, at least in this study. This result is not surprising as, in contrast to GC treatment, surgery does not have a targeted effect on either the underlying immune response or the ongoing orbital inflammation in TED . In fact, surgery may generate additional stress and inflammation in the orbital soft tissue. Thus, we believe that IV GC pulse therapy should be the first line of therapy for DON unless the patient has a contraindication to steroid use.
There is currently no standardized dosing regimen for IV GC pulse therapy or for the subsequent oral prednisone taper in patients with DON. No RCTs have been conducted to determine the most efficacious regimen. There also is no clear evidence regarding the benefit of adjuvant radiotherapy for DON after GC therapy. Based on the available evidence, most would agree with a regimen of IV MP, 1 g/day, for three consecutive days, repeated in the second week if necessary. If the response is sufficient, an oral prednisone taper can be initiated at the completion of the pulse therapy. If the response is insufficient or if the patient cannot tolerate or has contraindications to steroid treatment, either orbital decompression surgery or radiation therapy should be considered [90–92].
Systemic GC therapy is well known to have a multitude of adverse effects, many of which have serious health consequences and some of which can be lethal. Four studies that specifically investigated the use of IV GC pulse therapy for DON observed the following side effects: peptic ulcer, osteoporosis, abscess formation, increased insulin requirement, weight gain, cushingoid appearance, hypertension, central retinal vein occlusion, and irritability [86–89]. Other known serious GC side effects such as gastrointestinal (GI) bleed, liver dysfunction, and adrenal insufficiency were not reported by the authors of these studies, but this may be due to the small sample size in each. Indeed, much larger studies have been conducted on IV GC treatment for moderate-to-severe TED, which typically involves weekly infusions of IV GCs with a cumulative dose ranging from 4.2 to 12 g . A meta-analysis of a total of 1,461 patients with moderate-to-severe TED treated with IV GCs reported a morbidity and mortality rate of 6.5 % and 0.6 %, respectively . The mortalities were due to acute liver failure and cerebrovascular or cardiovascular events [92–97]. All but one of these patients had received a cumulative dose of at least 8 g when the fatal adverse event occurred . Therefore, it would appear that the higher the cumulative dose of GCs, the greater the risk for a serious and potentially fatal adverse effect.
The most common serious adverse effect reported in TED patients treated with IV GCs is acute liver damage (ALD) [97, 98]. The estimated total frequency of GC-associated ALD is 1 %, with 30 % of these cases being lethal . The mechanism underlying this serious complication is unclear, but studies suggest three major risk factors: fatty liver (steatosis), autoimmune hepatitis, and subclinical viral hepatitis [92, 98]. To minimize risk of ALD, the following tests should be obtained prior to initiating GC therapy for DON: liver function tests, liver ultrasound, non-organ-specific autoantibodies that may be associated with autoimmune hepatitis (microsomal antibodies, anticentromeric antibodies, antimitochondrial antibodies, and smooth muscle antibodies) , and viral markers for hepatitis B and C .
Cardiovascular and GI risk factors also should be assessed prior to initiating systemic GC therapy for DON. These include hypokalemia, cardiac arrhythmia, uncontrolled hypertension, uncontrolled diabetes, and history of a GI bleed . Therefore, in addition to the aforementioned hepatic workup, the following tests should also be performed: complete blood count, basic metabolic panel, fasting glucose test, lipid panel, urinalysis (urine culture), and fecal occult test. If testing suggests that a patient is at high risk for developing an adverse effect from GC therapy, alternate treatment modalities should be considered. In addition, once GC therapy is initiated, all patients should be monitored closely for side effects.
Surgical Orbital Decompression
For patients with DON who do not respond to IV GC pulse therapy or for those who have contraindications to or cannot tolerate the side effects of GCs, surgical decompression should be considered . The goal of surgical decompression is to remove parts of the rigid bony orbital walls and/or orbital fat, thereby effectively expanding the available orbital space. This allows the enlarged and congested orbital soft tissue to expand into newly available space, relieving the orbital pressure and reducing the compression on the optic nerve [99, 100]. Numerous techniques to decompress the TED orbit have been developed since the early 1900s. We will discuss the major techniques that are in use today.
The orbit has several bony regions that could be subject to decompression: the anterior lateral wall, the deep lateral wall, the orbital floor, and the medial wall  (Fig. 28.4).
(a) Bony orbital anatomy. (b) Areas amenable to removal for both general orbital decompression and optic nerve decompression
The anterior lateral wall consists of the zygoma and is covered by the temporalis muscle and buccal fat. The deep lateral wall consists of the greater wing of the sphenoid and the lesser wing of the sphenoid anterior to the tip of the superior orbital fissure. The orbital floor overlies the maxillary sinus. The medial wall overlies the ethmoid sinuses. Although decompression of each of these areas has been shown to be effective to varying degrees in treating severe TED, it is the decompression of the posterior medial wall that is especially important in the treatment of DON  as the posterior medial wall is in close proximity to the orbital apex, the location at which the optic nerve compression occurs.
The medial wall can be approached either endoscopically or non-endoscopically (Fig. 28.5). Kennedy et al. first described using the intranasal endoscope to perform medial and inferior orbital wall decompression to treat DON . In this approach, a total ethmoidectomy is first performed. The middle turbinate is retained to prevent the prolapse of orbital fat over the sphenoid osteum and decreases the risk of a postoperative cerebrospinal fluid (CSF) leak. The lamina papyracea is then removed in piecemeal fashion, leaving a small piece intact in the frontal recess to prevent prolapsing fat from obstructing the frontal sinus. The exposed periorbita is incised from the face of the sphenoid anteriorly to facilitate decompression of the orbital contents into the ethmoid sinus. This approach has comparable efficacy as but fewer side effects than the previously used transantral approach involving a Caldwell-Luc antrostomy and external ethmoidectomy [102–106].
Preoperative and postoperative CT scans showing results of bilateral posterior orbital decompressions for thyroid eye disease characterized in part by bilateral optic neuropathy. Note expansion of the medial rectus muscles into the space previously occupied by the anterior and ethmoid sinuses. The patient has also had bilateral lateral and intracranial decompressions because of his severe proptosis
Non-endoscopic techniques can also effectively decompress the posterior medial wall and orbital floor. The medial wall can be approached by a transcaruncular or transcutaneous route. The orbital floor can be accessed through the inferior fornix transconjunctivally, with or without an additional lateral canthotomy (the “swinging eyelid approach) or through a subciliary incision [107–116]. Although the more minimally invasive approaches are typically preferred, in patients with significant periorbital swelling or conjunctival chemosis, these smaller incisions may not provide sufficient posterior access. In such settings, larger incisions such as a coronal incision may be more reasonable. A combined approach using both endoscopic and transconjunctival or transcutaneous incisions has also been reported to be beneficial for the treatment of DON [117, 118].
The transcaruncular approach is a technique that is particularly well studied and is now commonly performed by most oculoplastic surgeons to decompress the medial wall in the treatment of DON [109, 110] (Fig. 28.6). The incision is made between the plica and the caruncle. Blunt dissection is carried out medially to the posterior lacrimal crest. The periorbita is incised posterior to the posterior lacrimal crest and elevated to expose the medial orbital wall. The entire medial orbital wall and parts of the inferior orbital wall are then removed. The osteotomy extends superiorly to the level of the anterior and posterior ethmoidal arteries (that are identified and cauterized), anteriorly to the posterior lacrimal crest, inferiorly to the bony maxilloethmoidal strut, and laterally to the infraorbital canal . Posteriorly, the bone inferior and posterior to the posterior ethmoidal artery is removed to decompress the orbital apex. The advantages of the transcaruncular approach over the transcutaneous approaches include more rapid entry into the orbit, less manipulation of ocular adnexal structures, and better cosmetic outcome . Retrospective studies have found the transcaruncular approach to be efficacious in the treatment of DON patients who are refractory to IV GC therapy, with significant improvement in visual acuity, color vision, and visual fields [110, 120].
Schematic drawing of the transcaruncular approach to the optic nerve. For DON, a total ethmoidectomy would be performed
In addition to removal of the medial wall and orbital floor, some surgeons also decompress the lateral wall at the same time. This approach is called the “three-wall decompression” [121, 122]. Others perform a “balanced decompression,” removing parts of the medial and lateral walls but leaving the orbital floor intact . Retrospective case series have shown each of these approaches to be efficacious in treating TED, but none has been studied with respect to treatment of DON [82, 119, 121, 122]. We believe that all surgical procedures are of equal benefit as long as they result in sufficient posterior decompression. Removal of the lateral wall alone or removal of bone from the anterior medial wall or floor only usually is not beneficial. Nevertheless, a study by Choe et al.. evaluated the outcome of medial wall (transcaruncular approach) versus lateral wall decompression in patients with DON and found no significant difference in the clinical outcome measures (visual acuity, color vision, pupillary reaction, visual field) between these two approaches  although the lateral wall group achieved more proptosis reduction than the medial wall group. Although this study suggests that lateral wall decompression is as effective as medial wall decompression in treating DON, we have not found this to be the case.
Lateral orbital wall decompression was the earliest orbital decompression technique described for the treatment of TED and remains one of the most popular techniques used today. For the treatment of TED without DON, lateral wall decompression is frequently the first line of surgical approach for many surgeons. The traditional technique, used in the aforementioned Choe et al. study, involves a lateral canthal incision and removal of part of the zygoma from approximately the frontozygomatic suture to just above the zygomatic arch [119, 124, 125]. This allows the orbital contents to decompress into the region of the temporalis muscle . The medullary space of the sphenoid bone can also be removed to allow further decompression. A modified technique of lateral wall decompression, called the deep lateral wall approach, was described by Goldberg et al. in the early 1990s . In this technique, the lateral wall is removed medially up to the inferior orbital fissure, anteriorly to posterior to the orbital rim, superomedially to the superior orbital fissure, and posteriorly up to the dura of the middle cranial fossa. The lateral wall, including the marrow of the sphenoid, is removed with a high-speed surgical drill using a cutting and/or diamond burr tip . The advantages of the deep lateral wall decompression technique include posterior axial displacement of the globe rather than sideways or inferior displacement and significantly lower risk of consecutive strabismus, sinusitis, and CSF leak [101, 126]. The efficacy of the deep lateral approach in treating DON has not yet been assessed.