External Beam Radiotherapy for Thyroid Eye Disease

Figure 1 Radiation-induced clonogenic cell death.

where in the cell that, ultimately, will affect the DNA. If ionizing radiation strikes the DNA and produces a biological effect it is referred to as the direct effect. If ionizing radiation strikes elsewhere in the cell and, subsequently, the effect is transmitted to the DNA by chemical intermediaries it is called the indirect effect. The vast majority of radiation’s effects on living tissue is via the indirect effect.

Most living tissue consists of water. When ionizing radiation interacts with water it will transiently create ionized water species. These, in turn, will go through a series of rapid and complex radiochemical reactions that will result in the formation of chemical free radicals. A free radical is a chemical species with an electron in its outer atomic shell not paired with another electron with an opposite spin. Chemical free radicals are extremely reactive species and are the source of much mischief in biology. The more commonly produced free radicals from ionizing radiation are hydroxy free radicals (OH•).

If the OH• is sufficiently long lived in the cell, it will interact with the purine and pyrimidine bases of DNA and bind to them. This will produce singleand double-strand breaks in DNA. If these breaks are irreparable by the cell’s repair enzymes, one of two pathways may be taken that lead to cell death. If the injury to the DNA is sufficient to prevent the cell from successfully carrying out replication, when the cell attempts to divide it will lose its homeostatic mechanisms, swell, burst, and die. This is referred to as clonogenic death (Fig. 1). An alternative pathway leading to cell death occurs when the cell’s monitoring processes detect serious DNA injury. If this injury is not repaired, then programmed cell suicide, called apoptosis, occurs that leads to DNA fragmentation and cell death (Fig. 2). The presumed mechanism of action of ionizing radiation in Graves’- associated orbital disease is radiation-induced death of inflammatory cells, including lymphocytes, which inhibits the inflammatory attack on the periocular tissues and reduces pain and proptosis. The presumed reason for radiation’s failure to work successfully, in

Figure 2 Apoptosis induced by radiation.

some cases, is that the inflammatory process is so longstanding, or so advanced, that fibrosis has occurred that cannot be reversed by killing acute inflammatory cells.

II.GENERATION OF EXTERNAL BEAM RADIATION THERAPY FOR THYROID EYE DISEASE

There are two commonly used types of machines for the generation of external beam radiation for the treatment of Graves’-associated orbital disease. A cobalt machine consists of a block of radioactive cobalt 60 housed in a lead box. The box has an aperture through which the radiation can escape. When the cobalt is moved to the ‘‘on’’ position it is pushed by a mechanical device over the aperture and a high-energy beam of radiation escapes. When it is pulled back into the ‘‘off ’’ position, it is removed from the aperture and the radiation is contained by the surrounding lead box.

A high-energy linear accelerator, in contrast, consists of an electron gun that fires electrons down an accelerator tube. This tube imparts additional energy to the electrons. The electrons are slammed into a tungsten target. As they slow down in the target, energy is given off. This energy takes the form of a high-energy x-ray beam that may be shaped and modulated for the purpose of administering external beam radiation therapy. In modern medical practice, most treatment of Graves’-associated orbital disease is performed with a linear accelerator.

III. RADIOTHERAPY TECHNIQUE

When utilizing radiotherapy in the treatment of Graves’ disease, the target volume typically includes the entire content of the bony orbit in order to treat all of the tissues poten-

Figure 4 Isodose plan: lines represent percentage dose received to enclosed volume.

alloy, is used to fabricate customized blocking to shield out the structures inferior and superior to the bony orbit such as the brain and sinus, which are not a target for the radiotherapy. The anterior field edge is typically set clinically at the lateral canthus, to place the beam edge posterior to the lens of the eye. This keeps the dose of radiotherapy to the ipsilateral lens quite low. Radiation beams, however, diverge. Thus, the field set behind the ipsilateral lens will diverge into the contralateral lens unless specialized techniques are utilized to avoid this. Two techniques are generally used to shield the lens: either angling fields back or utilizing half-beam blocking techniques. By angling back, the divergence is taken out of the beam in the direction of the lens by rotating the beam posteriorly, typically between 3 and 5 degrees. Alternatively, half-beam blocking techniques literally block out half of the radiotherapy field, so that the nondivergent center of the beam is located at the lateral canthus (Fig. 3). With this technique the amount of radiation reaching the lens is only that transmitted through the half beam block: about 3% of the dose (3) (Figs. 4, 5).

IV. RETROSPECTIVE TRIAL RESULTS

The first reported use of radiotherapy for Graves’ ophthalmopathy was in 1936 by Henry Thomas, Jr., and Alan Woods. ‘‘X-ray treatment is being given to the orbits with the hope

Figure 5 Dose volume histogram per lens: 100% of lens receives approximately 3% of prescribed dose.

of reducing swelling. While some improvement has been noted, it is still too soon to make any definite statement on the value of radiation in this patient’’ (4).

Multiple reports have retrospectively analyzed the efficacy of radiotherapy for Graves’ ophthalmopathy. While these reports are useful in looking at responsiveness to therapy, the usual cautions involved in retrospective analyses apply. Patients are selected, and often less favorable patients are referred for radiation after failing alternative treatment regimens. The results are also further muddied by the concurrent utilization of steroids. Nevertheless, lessons are to be learned from retrospective analyses. Certainly these series provide the longest-term data on possible radiation complications.

Peterson et al. reported on 311 patients treated at Stanford between 1968 and 1988

(1). Patients were organized into groups according to the era of treatment and the total radiation dose. Two groups, the pre-1979 and post-1983 patients, received 2000 cGy in 10 fractions of 200 cGy/fraction. Patients treated between 1979 and 1983 received 3000 cGy in 15 fractions of 200 cGy/fraction. Patients were reviewed 2–4 weeks postradiotherapy and thereafter as needed. Minimum follow up was 12 months. Signs and symptoms were scored with respect to five parameters: soft tissue, proptosis, eye muscle impairment,

corneal involvement, and degree of sight loss. In response evaluation, 80% had soft tissue responses and 58% of these were complete responses. Over 75% of patients with corneal manifestations had a significant response with therapy. Abnormalities of extraocular muscle motility and proptosis also exhibited improvement: in 61% and 51%, respectively. A wide range of response to visual acuity defects was reported. Seventy-six percent of patients successfully discontinued steroid use following their radiotherapy, typically within several months of completing treatment. There was no outcome improvement in the patient group receiving higher doses. Acute side effects during treatment occurred in 10% of patients and consisted of self-limiting soft tissue inflammation. No long-term complications were observed. With follow-up to 21 years, no radiation-induced tumors have been detected.

A second large series from France of 199 patients treated between 1977 and 1996 reported similar results (5). Twenty-six percent demonstrated good or excellent overall responses to therapy with 48% partial responses. The authors reported significantly improved results for patients treated with early or moderately advanced presentation and improved results for patients treated no later than 7 months after the beginning of ophthalmopathy. In follow up, four patients have required surgical treatment for bilateral cataract. No retinopathy or tumor induction have been reported with median follow up for 100 patients to 86 months.

A number of smaller, retrospective series are available in the literature. A study by Palmer et al. of 29 patients indicated overall improvement in signs and symptoms in 48%

(6). Soft tissue changes were most responsive and relieved in 78%, with proptosis reduced in 52%. Eye muscle motility, however, was only improved in 24%. Twenty-eight percent of these patients had previous orbital decompression that may have contributed to this result. A series by Marcocci et al. reported similar improvements in proptosis and ophthalmoplegia to those of the Stanford series, but systemic corticosteroids were utilized as a fundamental part of therapy (7).

Although radiotherapy complications are generally rare, some studies have reported cataracts, retinopathy, and optic atrophy. Kinyoun et al. reported four cases of radiation retinopathy and optic atrophy after orbital irradiation (8). In retrospect, however, these cases were found to have major dosimetric errors delivering over 3500 cGy in ten 350 cGy fractions instead of 2000 cGy in 2 weeks (9). It is well recognized that larger fraction size is related to increased risk of late radiation complications. Some authors have suggested that diabetics may have an increased risk of retinopathy.

Tumor induction by therapeutic radiation is a rare but serious complication. There is no known threshold, or safe dose, below which this complication is not believed to occur, but the probability is increased with higher doses. Snijders-Keilholz et al. have published a calculation of risk of tumor induction by orbital radiotherapy for Graves’ ophthalmopathy. They calculate a risk of 0.0064 (or 6 : 1000 persons) for fatal radiationinduced cancers or 1.2% (10). Since radiation-induced malignancies have a latency period of decades, for elderly persons the risk of orbital irradiation is minimal. The authors suggest, however, that treatment be reserved for older patients due to the theoretical increase in malignancy induction that the young may survive to realize.

In summary, the retrospective literature supports the use of therapeutic radiation in treatment of Graves’ ophthalmopathy. The literature cautions us, however, as to the dangers of improper treatment and careful employment of radiotherapy to minimize risks of complications.

V.EVALUATION OF PROSPECTIVE RANDOMIZED TRIALS

Prummel et al. from the University of Amsterdam conducted a randomized prospective, double-blind clinical trial to compare the value of prednisone to external beam radiation therapy (11). They enrolled patients, ages 20–70 years, who were euthyroid for at least 2 months, who had not received previous treatment for ophthalmopathy other than eye drops. The study was confined to individuals with moderately severe ophthalmopathy defined by moderate to marked soft tissue involvement, proptosis 23 mm, extraocular muscle involvement, and visual loss. Patients randomized to prednisone received 60 mg orally per day for 2 weeks, 40 mg orally per day for 2 weeks, 30 mg orally per day for 4 weeks, and, thereafter, the dosage was tapered by 2.5 mg per week in addition to sham radiation therapy. Patients randomized to receive external beam radiation therapy received 2 Gy per fraction, 1 fraction per day, for 10 fractions in 2 weeks plus placebo pills to mimic the prednisone.

A response was defined by a decrease in signs or clinical symptoms from baseline values. Treatment failure was defined as an increase in symptoms. Twenty-eight patients were assigned to receive prednisone and sham radiation and 28 received radiotherapy and placebo. Therapeutic outcome after 24 weeks, as determined by change in signs or symptoms, was similar in each treatment group. Of the 28 patients who received prednisone, 14 responded (50%). Thirteen of the 28 patients receiving radiotherapy responded (46%); 36% in the radiotherapy group showed no change, as did 40% in the prednisone group. When each treatment group is considered as a whole, improvement was seen in total and subjective eye scores, which is attributable to improvement in the responders. There was no difference in degree of improvement between the two treatment groups, but the total eye score improved more rapidly in the prednisone-treated patients.

Side effects seemed more marked in the patients treated with prednisone. Mean body weight in these patients increased from 71 kg to 73 kg at 24 weeks (p 0.002). Hypertension, severe cirrhosis, hirsutism, behavioral change, and cushingoid face were more common in the prednisone-treated patients.

The study suggested that the efficacy of external beam radiotherapy and of oral prednisone in initial treatment of patients with moderately severe Graves’ ophthalmopathy was similar but, because radiotherapy was better tolerated, it might be preferred.

In contrast, a recent study by Mourits et al., also from the Netherlands, found the case for radiation therapy less persuasive (12). This study included patients with moderately severe Graves’ ophthalmopathy based on the presence of lid retraction, proptosis, impaired motility, an increase in intraocular pressure, along with enlarged extraocular muscles and increased intraorbital fat on a coronal computed tomogram (CT). A patient was judged to have moderately severe disease if he or she had, in their worse eye, motility impairment causing diplopia, proptosis 23 mm, moderate or severe eyelid swelling, or a combination of these. Patients included in this trial otherwise were similar to the prior randomized study: age 25–75 years, no treatment for orbital disease except drops, euthyroid for 3 months, and no patients with diabetes mellitus.

All the patients received either 2 Gy per fraction of external beam radiotherapy in 10 fractions over 2 weeks or sham irradiation. Patients were examined 1 day before, and at 4, 12, and 24 weeks after radiation therapy.

The definition of treatment outcome included major and minor criteria. Major criteria were improvement in diplopia grade and improvement in eye movements in any direction of 8 degrees. Minor criteria were variations of 2 mm or more in lid aperture and

reduction in eyelid swelling. The authors defined the response as ‘‘successful’’ if the patient improved in one or more major criteria or in the two minor criteria.

Thirty patients were assigned to radiotherapy and 30 to placebo. All 30 patients assigned to radiotherapy completed the treatment. Of those assigned to placebo, 29 completed therapy. Treatment outcome was successful at week 18 in 60% of the irradiated patients and in 31% of the sham irradiated patients ( p 0.04). Motility improved in 82% of the patients after radiotherapy and 20% after sham irradiation ( p 0.004). Eyelid swelling was persistent in 36% of the irradiated patients and 42% of the sham irradiated patients ( p 0.99). In more quantitative assessments there was no difference between the irradiated and sham-irradiated patients. The irradiated patients showed improvement in their clinical activity score more quickly than the nonirradiated patients. A significant number of patients in both arms of the study ultimately went on to undergo surgery.

The mixed record of radiotherapy in this second randomized study lead the authors to conclude that, in patients with moderately severe Graves’ ophthalmopathy, radiotherapy should only be used for the treatment of motility impairment and not to ameliorate other signs or symptoms of the disease.

Clearly, both studies suffer from a small patient population. If radiotherapy were beneficial, it would have to be considerably better than the alternative to be demonstrable in studies with only 28 to 30 patients in each arm. The solution to the uncertainty created by these studies might be resolved by a large-scale cooperative group trial, perhaps involving several countries, to generate sufficient patients to answer the questions posed.

VI. CONCLUSION

The proposed mechanism of action of radiotherapy in the treatment of Graves’-associated orbital disease is killing of inflammatory cells and subsequent reduction in muscle swelling. Radiation beams are commonly generated by a linear accelerator. Therapy typically utilizes parallel opposed lateral photon beams to a dose of 20 Gy in 10 fractions over 2 weeks. Although retrospective studies generally support the value of external beam radiotherapy in selected patients, prospective studies offer a more mixed view. Further prospective trials are warranted to refine our understanding of the role of external beam radiotherapy.

REFERENCES

1.Peterson IA, Kriss JP, McDougall R, Donaldson SS. Prognostic factors in the radiotherapy of Graves’ ophthalmopathy. Int J Radiat Oncol Biol Phys 1990; 19:259–264.

2.Hall EJ. Radiobiology for the Radiologist. 4th ed., 1994.

3.Snow A. The use of independent collimation in the treatment of Graves’ ophthalmopathy. Br J Radiol 1999; 72:389–391.

4.Thomas HM, Woods AC. Progressive exophthalmos following thyroidectomy. Bull John Hopkins Hosp 1936; 59:99–113.

5.Beckendorf V, Maalouf T, George JL, Bey P, Leclere J, Luporsi E. Place of radiotherapy in the treatment of Graves’ orbitopathy. Int J Radiat Oncol Biol Phys 1999; 43:805–815.

6.Palmer D, Greenberg P, Cornell P, Parker RG. Radiation therapy for Graves’ ophthalmopathy: a retrospective analysis. Int J Radiat Oncol Biol Phys 1987; 13:1815–1820.

7.Marcocci C, et al. Orbital cobalt irradiation combined with retrobulbar or systemic corticosteroids for Graves’ ophthalmopathy: a comparative study. Clin Endocrinol 1987; 27:33–42.

8.Kinyoun JL, Kalina RE, Brower SA, Mills RP, Johnson RH. Radiation retinopathy after orbital irradiation for Graves’ ophthalmopathy. Arch Ophthalmol 1984; 102:1473–1476.

9.Parker RG, Withers HR. Radiation retinopathy—letter to the editor. JAMA 1988; 259:43.

10.Snijders-Keilholz A, De Keizer RJ, Goslings BM, Van Dam EW, Jansen JT, Broerse JJ. Probable risk of tumour induction after retro-orbital irradiation for Graves’ ophthalmopathy. Radiother Oncol 1996; 38:69–71.

11.Prummel MF, Mourtis MP, Blank L, Berghout A, Koornneef L, Wiersinga WM. Randomized double-blind trial of prednisone versus radiotherapy in Graves’ ophthalmopathy. Lancet 1993; 342:949–954.

12.Mourits MP, van Kempen-Harteveld ML, Garcia MGB, Koppeschear HPF, Tick L, Terwee CB. Radiotherapy for Graves’ orbitopathy: randomized placebo-controlled study. Lancet 2000; 355:1505–1509.

Figure 1 Patient with exophthalmos, who was treated with eyelid lengthening. Camouflage surgery, repositioning the eyelids over a prominent globe, is suboptimal from an esthetic and functional standpoint. ( 2001, Regents of the University of California.)

likely to request surgery. We now better appreciate the phenomena of compressive orbitopathy: diffuse pressure, pain, discomfort, and congestive edema that characterize the tight postinflammatory orbit and that respond well to decompression surgery (Fig. 2).

III. PLANNING SURGERY: STAGING THE DISEASE

The commencement of surgical rehabilitation is a major step in the life of a patient with Graves’ disease, and should be approached in a conservative and studied fashion. There is a long road ahead, and a good relationship with the physician and staff provides critical emotional support. There are rare severe risks of surgery and a likelihood of multiple stages of surgery, so informed consent should be thorough.

It is best to consider surgery in the stable phase of the disease. Operating on an inflamed orbit is characterized by intraoperative bleeding and a rocky postoperative course. Furthermore, planning surgical rehabilitation requires assessing orbital, strabismus, and eyelid parameters that can be a moving target during the acute phase of the disease. It is possible that things will improve spontaneously to the point that less surgery or even no surgery would be required. A rule of thumb is to wait for 6 months of stable, postinflammatory disease before embarking on surgical rehabilitation (which begins with consideration for orbital decompression).

There are, however, times when surgery is appropriately performed in the inflammatory stage of the disease. Although compressive optic neuropathy often responds to medical treatment in the inflammatory phase, there are cases in which persistent optic neuropathy (for example, in the range of 20/70 or worse) is unresponsive to aggressive steroid therapy or radiotherapy. If weeks have gone by, particularly if coexisting vascular disease is present, then surgical decompression of the nerve is appropriate and typically very effective. Also, some patients have a prolonged inflammatory course, or one characterized by exacerbations and remissions. They may be unable to work or function in daily life. Sometimes the best compromise is to accept the increased unpredictability of surgery, and

(A)

(B)

(C)

Figure 2 Congestive orbitopathy. (A) Patient with late postinflammatory, congestive orbitopathy: periorbital edema and pain are related to congestion of venous outflow at the apex. (B) Following orbital decompression: edema and pain resolve.

(C) Following eyelid repositioning surgery. ( 2001, Regents of the University of California.)

A

B

Figure 9 (A) The lateral orbital rim is internally thinned in the area of the lacrimal keyhole, providing space for decompression and allowing better visualization of the superior orbital fissure. (B) A groove is burred from the lacrimal keyhole at the orbital rim, to the superior orbital fissure, identifying the diploic space within the greater wing of sphenoid. (C) After all the diploe is removed from the lesser and greater sphenoid wing, a T-shaped groove is present, including a large diploic lake adjacent to the inferior orbital fissure. (D) The basin of the inferior orbital fissure is removed out to the buccal fat and maxillary sinus mucosa. ( 2001, Regents of the University of California.)

C

D

It leaves no visible scar (assuming an adequate hairline) and allows performance of a simultaneous upper facelift when desired. Through a coronal approach, the lateral rim can be left in place and thinned, augmented with specialized orbital rim onlay implants (3), or repositioned with osteosynthesis systems (16). After elevating the medial canthal tendon and lacrimal sac from their periosteal attachment, excellent exposure is obtained for medial and inferior orbital decompression.

Most cases do not require maximal bone removal, and the eyelid crease incision is considerably less time-consuming than the coronal approach (Fig. 8) (20). The eyelid crease incision is well hidden cosmetically and offers excellent exposure to the three areas of thick bone in the lateral wall, as discussed above.

Substantial bone in the deep lateral orbit is available for orbital expansion. The

lateral orbital wall is now my first wall for decompression: by removing bone from each of the three areas of deep bone, combined with excision of 2–5 cc intraconal fat, 5 or even 6 mm of proptosis reduction can be achieved. Consecutive strabismus is substantially reduced. I have found the rate of the new-onset strabismus to be in the range of 5%, compared to 30% using medial or balanced medial and lateral decompression (21).

The anatomy of deep lateral orbital decompression can be intimidating. This is the same bone that is removed through the classic neurosurgical approaches, and the surgeon must navigate around the dura of the anterior and middle cranial fossae. I have found that focusing on the river of diploe that runs through the sphenoid bone provides a reliable landmark that helps me to safely achieve maximal bone removal from the deep lateral orbit.

The orbit is widely exposed in the subperiosteal plane. Periorbital dissection over the zygomatic prominence and superiorly over the frontal bone is necessary to achieve maximal deep orbital exposure. The orbit is entered subperiosteally and exposure is taken back to the superior orbital fissure, dividing the meningolacrimal vessel as necessary. Inferior dissection exposes the inferior orbital fissure which is opened in its anterior 1 cm as described by Jack Rootman, exposing the orbital floor.

The first bone removed is the lacrimal keyhole, in the fossa of the lacrimal gland (Fig. 9). I remove a notch from the superolateral orbital rim both to achieve some proptosis reduction as the lacrimal gland prolapses outside the orbit, and also to help gain a better view of the superior orbital fissure. This dissection is performed with a side-cutting aggressive cutting burr. The orbital rim is left intact.

I then make a groove in the direction of the superior orbital fissure. By aiming for the fissure, the surgeon naturally encounters the beginning of the diploic space within the greater wing of the sphenoid. Superiorly, this dissection is limited by the thin bone of the orbital roof. Particularly in a young patient, I have no hesitation to expose some of the frontal dura as I delineate the superior most edge of the thick bone of the deep lesser wing. The surgeon may switch to a 3 mm diamond burr for increased control as the deepest bone is removed.

The diploic space within the greater wing is hollowed out using burrs and curettes. This leaves a ‘‘cliff ’’ of the orbital table of the diploe, which can be removed using the diamond burr, working back towards the superior orbital fissure.

I then follow the diploe as it branches off inferiorly towards the inferior orbital fissure. Again, a combination of burrs and curettes can be used. The combination of the lesser and greater sphenoid wing diploic space forms the shape of the letter ‘‘T.’’

Above the inferior orbital fissure, the diploic space typically widens to form a large lake of diploe that can be hollowed out along the edge of the inferior orbital fissure, creating a large cavity. The diploe in the greater wing of the sphenoid also leaves a cliff of bone in the deepest part of the greater wing, and this can be removed using the diamond burr.

Once the diploe has been removed, and the inner table and ‘‘cliff ’’ thinned, the remainder of the decompression is straightforward. The anterior lateral wall can be thinned over the temporalis muscle. I try to leave eggshell bone over the muscle, for fear that extensive removal of bone over the muscle may result in oscillopsia with chewing. No bone should be removed directly lateral to the globe. Lateral shift of the globe with increased pupillary distance may result. The basin of the inferior orbital fissure can be removed out to the buccal fat and maxillary sinus mucosa.

The periosteum is then opened widely over all of the areas of bone removal, and intraconal fat can be accessed for graded removal as the assistant retracts the lateral rectus muscle superiorly using a curved malleable retractor.

Preand postoperative CT scans demonstrate the significant orbital volume that can be achieved by removing the thick bone in the deep lateral orbit (Fig. 10). The risk of postoperative strabismus, based on my anecdotal experience, is lessened compared to medial approaches. With experience, the surgery can be performed in 45 min per side.

Bone is removed directly behind the globe, allowing proptosis reduction even in ‘‘woody’’ orbits that have little ability to expand their shape horizontally. Bone is removed from the deep orbital apex, so I hypothesize that it should be as efficacious for treating compressive optic neuropathy as the medial approach.

Removal of deep bone in the lateral orbit requires sculpting away layers of cortical and marrow bone in an anatomically complex area. It is possible to injure the globe or apical neurovascular structures, and it is also possible to enter the intracranial cavity and cause central nervous system (CNS) injury. Knowledge of anatomy and experience in the cadaver lab are prerequisites for safe surgery. Good illumination and retraction are paramount. It is not uncommon to create a small dural exposure, and cerebrospinal fluid (CSF) leaks can occur rarely. These are managed by packing the area of the leak with tissue grafts (with or without tissue glue) and by postoperative observation. The leak has no long-term egress route and is self-limited. A substantial dural tear or intracranial entry could cause bleeding or brain tissue injury. The surgeon should use utmost care to avoid extensive intracranial disruption, and neurosurgical backup should be available. In our series of over 100 deep lateral orbital decompressions, we have had 5 self-limited CSF leaks and no serious intracranial injury.

VI. REMOVAL OF INTRACONAL FAT

Intraconal orbital fat can be removed from either the eyelid crease or coronal incisions, and also through an inferior fornix conjunctival approach. A significant change in surgical philosophy and technique involves a new enthusiasm to remove intraconal orbital fat in cases characterized by enlargement of the fat compartment (as opposed to primary extraocular muscle enlargement). I shared many surgeons’ reluctance to enter the muscle cone and remove fat when I first read the reports in the plastic surgery literature (22). I first timidly and then more aggressively began removing orbital fat. To my surprise I have not noted complications related to removal of as many as 6 cc intraconal fat. Trokel and Kazim have published a large series reflecting the safety and efficacy of intraconal fat removal (23) and more recently reported successful treatment of compressive optic neuropathy (24). Removing 2–5 cc intraconal fat between the lateral rectus and inferior rectus, and if needed from the superiomedial and inferomedial compartments, provides additional proptosis reduction in patients with nonwoody, freely flowing fat.

VII. MEDIAL DECOMPRESSION

The lamina papyracea of the medial wall can be removed through an endonasal, coronal, or transcaruncular (Baylis) approach (25,26). The latter was first used by Henry Baylis in the late 1980s and further developed by Norman Shorr and myself. We have used it extensively for a multitude of medial orbital surgeries in the extraperiosteal and intraperio-

steal spaces. The approach provides wide, rapid exposure of the entire medial wall from the roof to the floor through a cosmetically hidden incision, and is therefore ideal for medial orbital decompression (Fig. 11).

For medial decompression, the lamina papyracea is removed from the sphenoid– ethmoid junction posteriorly to the equator of the globe anteriorly, and from the frontoethmoid suture superiorly to the maxilloethmoidal strut inferiorly. The roof of the ethmoid sinus (which is properly called the fovea ethmoidalis, not, as I often hear, the lamina cribrosa) adjoins the anterior cranial fossa. Severe intracranial complications can occur if it is breached, ranging from CSF leak, which is usually treatable by intranasal or rarely intracranial repair, to vascular injury to the anterior communicating artery, which can be

A

B

Figure 10 Preoperative and postoperative CT scan (negatives, with right orbit outlined) demonstrate removal of thick bone in the deep lateral orbit. ( 2001, Regents of the University of California.)

A

B

C

Figure 11 The transcaruncular (Baylis) incision provides wide, rapid access to the entire medial wall for medial and inferomedial decompression. ( 2001, Regents of the University of California.)

D

E

Figure 11 Continued

fatal. It is not possible to avoid dural exposure or CSF leak in every case, but knowledge of the anatomy and study of preoperative CT scans to observe anatomical variations such as a sloped fovea ethmoidalis can minimize complications.

As an alternative to the transorbital approach, the medial orbit can be decompressed transnasally. Most centers utilize the videoendoscope for visualization (27). There may be advantages in tight orbits, or if the surgical team is inexperienced in transorbital approaches, but the transcaruncular orbital approach provides equivalent exposure of the medial wall back to the optic canal ring, is rapid, and requires no special endoscopic equipment.

When the floor is added as a third wall, it is best to preserve the maxilloethmoidal strut (27) to minimize postoperative dystopia (sunset syndrome) and strabismus (Fig. 12).

Figure 12 By leaving intact the strut of bone between the maxillary and ethmoid sinuses, dystopia and strabismus can be minimized in inferomedial decompression. ( 2001, Regents of the University of California.)

VIII. ORBITAL RIM ONLAY GRAFT

Another surgical option is use of the orbital rim onlay graft designed for me by Porex Corporation (College Park, GA) (Fig. 13) (3) and in which I have no financial interest. The rim onlay graft is placed in a subperiosteal plane through a coronal, canthoplasty, or eyelid crease incision. It advances the position of the inferolateral rim and lateral canthus, reducing the disparity between these support structures and the prominent globe (28). In cases that cannot be maximally decompressed because of patient desires or medical considerations, the rim onlay graft is a valuable technique to correct bony structural disproportions in the globe–eyelid relationship. It can be added to decompression, or used as an alternative.

IX. COMPLICATIONS

The most worrisome complications of orbital decompression are loss of vision or loss of life. Visual loss can occur intraoperatively, related to vascular or pressure damage to the optic nerve or globe, and postoperatively related to orbital hemorrhage or vasospastic ischemia. On our service we have had one patient lose vision the day after surgery, and I am aware of other cases of postoperative visual loss. Fortunately this is extremely rare. It is not completely preventable, but gentle surgical technique, good hemostasis, and rapid evaluation for evacuable hematoma if visual loss is recognized postoperatively, are appropriate. I do not use a postoperative drain and I do not believe there is any evidence that this can reduce the risk of visual loss.

The primary risk of stroke or death is related to intracerebral complications. All of the cases of postdecompression stroke or death that I have reviewed relate to vascular injury of the anterior cerebral circulation. This occurs when instruments pass through the

A

B

Figure 13 (A) Medpor orbital rim onlay graft designed to advance the lateral orbital rim. The implant should cover the rim. (B) Incorrect placement: the implant is too far outside the orbit and will not adequately advance the lateral canthus. (C, D) Patient with stable thyroid-related orbitopathy, before and after orbital rim onlay graft and eyelid repositioning surgery. No decompression was performed. ( 2001, Regents of the University of California.)

orbital or ethmoid roof and damage the vessels that lie on the floor of the anterior cranial fossa. Early or late cerebral vasospasm results. The vasospasm can occur more than a week later, so patients who have had violation of the dura need to be monitored carefully for problems, perhaps with neurosurgical consultation.

Fortunately most dural violations do not involve vascular or brain parenchymal injury, and a fatal outcome of the more common isolated CSF leak after orbital decompression is unlikely. CSF leak after medial decompression leads to CSF rhinorrhea, and there is a risk of meningitis or permanent fistula. If the leak is observed intraoperatively, it might be possible to patch the leak using free tissue plugs of fat or fascia, mucosal flaps, with or without tissue glue. The sinus should be firmly packed. If the leak is noted postoperatively because of persistent clear fluid nasal discharge that worsens with leaning forward and positive test for glucose (usually 2/3 of serum glucose), it is reasonable to observe for up to 1 week. The rate of spontaneous closure is high. However, a persistent leak with a nasosinus egress route will require additional surgical intervention, either with an attempt at identification and packing from below through the sinus, or by patching the leak from above via a craniotomy approach (29).

I have managed half a dozen lateral CSF leaks after deep lateral orbital surgery. There is no egress route for the CSF and in all cases, after several days of a somewhat boggy orbit with CSF variably present at the wound, the leak has closed. This is no surprise to our neurosurgical colleagues, who of course routinely see some self-limited leakage after intradural procedures.

Double vision is the most common significant complication of orbital decompression surgery. It is more common after unbalanced inferomedial decompression (30–32) and more likely in patients with type 2 disease (33). Patients must be informed of the risk of double vision and prepared for a second stage of eye muscle surgery, if needed. I prefer to wait 3 months after decompression before the second stage, because many cases of early strabismus will improve spontaneously.

Numbness is common for the first 3 months after decompression, and mild permanent numbness is not rare because some small sensory branches are necessarily cut. The lacrimal, zygomaticotemporal, and zygomaticofacial senstory branches are at risk in lateral decompression. The infraorbital nerve is at risk from the inferolateral or inferomedial decompression. The inferior orbital nerve in particular is a large sensory trunk. If it is irreversibly damaged, symptomatic numbness of the cheek and upper lip will occur.

Recurrent proptosis occurs both early, over the first 3 months, and late, after months or years. I have suspected that early recurrences relate to contraction of the periosteal vault, although I have not been able to prove this with imaging studies. Some early and all late recurrences relate to the instability of the underlying disease. Since these recurrences can be self-limited, it is generally best to take a conservative approach and allow 6 or more months to pass before considering additional orbital decompression.

X. SECONDARY DECOMPRESSION

Secondary (repeat) decompression presents some special challenges, but overall the effectiveness and safety is comparable to primary decompression. Orbital imaging studies, obviously, are paramount in surgical planning: evaluation of bony anatomy will suggest the most valuable source for additional bony expansion. In my experience there is frequently additional room in the deep lateral and deep medial orbit even after prior decompression. Of course, any virgin areas are logically approached first. If a previously operated

Figure 14 Postoperative CT scan shows complete medial decompression. The ethmoid air cells have been removed back to the sphenoid–ethmoid junction (arrow). ( 2001, Regents of the University of California.)

area has to be approached, the surgeon should anticipate scar tissue. Fortunately in orbital decompression the surgical planes are bony, providing reliable guidance for re-exploration.

XI. RESULTS OF SURGERY

Various compilations of surgical results have been published. One must always interpret outcome studies of treatment of thyroid-related orbitopathy with the perspective of the natural history of the disease: the strong tendency for things to improve over time can be mistaken for treatment effect. Another variable is surgical technique. In my practice I frequently evaluate and obtain orbital imaging studies from patients who have had previous decompression surgery by different surgeons. I have noted a wide variation in bony removal. Sometimes even when the operative report suggests that the entire medial wall was removed, for example, the CT scan shows only removal of the anterior and middle ethmoid cells (Fig. 14).

XII. SUMMARY

Orbital decompression surgery is effective in reducing the pressure pain of congestive orbitopathy, improving the eyelid–globe relationship to restore corneal protection, and addressing disfiguring proptosis (Fig. 15). It is very effective in treating compressive optic neuropathy, but mild neuropathy can often be treated medically. Surgery for optic nerve compression is now performed only rarely in my practice, for patients whose condition

A

B

Figure 15 Patient with postinflammatory thyroid-related orbitopathy and marked proptosis, before and after combined lateral and medial orbital decompression. ( 2001, Regents of the University of California.)

fails to respond to medical therapy or who have congestive orbital pressure in the postinflammatory phase. There are significant complications, including rare severe ones, and patients must be well informed. The benefits are great and patients who have been through surgical rehabilitation for thyroid-related orbitopathy are some of my most grateful. Patients must be thoroughly informed, and accept the rare severe risks of surgery.

I now use graded removal of the three areas of thick bone in the lateral wall as my first approach in cases without compressive optic neuropathy, adding intraconal fat removal in most cases. This represents a significant departure from the traditional approach that began with inferomedial decompression. However, inferomedial decompression, even with creation of a strut at the ethmoid maxillary junction, has a significant risk of consecutive diplopia and globe displacement. The sinuses are violated and chronic sinusitis can result, especially if the osteomeatal complex is compromised. The dura above the fovea ethmoidalis is not easily visualized, and laceration can result in intracranial bleeding or chronic CSF leak. By contrast, lateral decompression has less risk of inducing consecutive diplopia since the muscle cone is not shifted inferomedially. We have found that decompression only of the lateral wall has a rate of consecutive strabismus of approximately 5%, compared to a rate of 30% in medial decompression (29). If dural exposure is necessary for maximal decompression, this is accomplished under excellent visualization and the risk of CSF leak is minimized both by better exposure and by the lack of a potential external egress pathway. Up to 5 cc bony volume, and at least 6 mm of globe retrodisplacement, is available in lateral orbital decompression. Woody orbits with little ability to expand their shape or respond to fat excision can be decompressed by removing the deep lateral wall, allowing the orbit to move directly posteriorly. The lateral orbit can be approached through hidden incisions either through a coronal flap, which I use for cases in which maximal decompression is needed, or through an upper eyelid crease incision, which I find useful for the majority of cases in which I intend to move the globe back 3–5 mm.

For patients with more proptosis than can be reduced through lateral wall decompression alone, the medial wall is added as a second wall, through a transcaruncular (Baylis) approach (or through the coronal approach if this has been chosen). Carrying this paradigm further, the floor (with preservation of the maxillary ethmoid strut) is now, for me, the third wall for decompression in cases that require maximal retrodisplacement. It is utilized only rarely, for maximal proptosis (more than 9 mm). When all the surfaces are combined, as much as 10 mm retrodisplacement can be obtained. The floor can be approached through the caruncular incision, through a separate fornix incision, through a coronal incision, or through a transantral or endoscopic nasal approach. The transantral and transnasal approaches do not allow easy sparing of the ethmoidal maxillary strut, and are now only rarely used in my practice for inferomedial decompression.

REFERENCES

1.Ogura JH, Walsh TE. The trans-antral orbital decompression operation for progressive exophthalmos. Laryngoscope 1962; 72:1078–1097.

2.Goldberg RA, Hwang MM, Garbutt MV, Shorr N. Orbital decompression for non-Graves’ orbitopathy: a consideration of extended indications for decompression. Ophthal Plast Reconstr Surg 1995; 11:245–252.

3.Goldberg RA, Weinberg DA, Shorr N. Management of the patient with a relatively prominent eye: non-Graves’ orbital decompression and orbital rim onlay porous polyethylene implants. Facial Plast Surg Clin North Am 1998; 6:11–19.

4.Goldberg RA. The evolving paradigm of orbital decompression. Arch Ophthalmol 1998; 166: 95–96.

5.Lyons CJ, Rootman J. Orbital decompression for disfiguring exophthalmos in thyroid orbitopathy. Ophthalmology 1994; 101:223–230.

6.Shorr N, Neuhaus RW, Baylis HI. Ocular motility problems after orbital decompression for dysthyroid ophthalmopathy. Ophthalmology 1982; 89:323–328.

7.Shorr N, Seiff SR. The four stages of surgical rehabilitation of the patient with dysthyroid ophthalmopathy. Ophthalmology 1986; 93(4):476–483.

8.Rootman J, Stewart B, Goldberg RA. Decompression for thyroid orbitopathy. In: Orbital Surgery: A Conceptual Approach, Philadelphia: Lippincott-Raven, 1995:353–384.

9.Neigel JM, Rootman J, Belkin RI, et al. Dysthyroid optic neuropathy. The crowded orbital apex syndrome. Ophthalmology 1988; 95:1515–1521.

10.Nunnery WR, Martin RT, Heinz GW, Gavin TJ. The association of cigarette smoking with

clinical subtypes of ophthalmic Graves’ disease. Ophthal Plast Reconstr Surg 1993; 9:

77–82.

11.Kennerdell JS, Maroon JC. An orbital decompression for severe dysthyroid exophthalmos. Ophthalmology 1982; 89:467–472.

12.Naffziger HC. Exophthalmos: some surgical principles of surgical management from the neurosurgical aspect. Am J Surg 1948; 75:25–41.

13.Leone CR, Jr, Piest KL, Newman RJ. Medial and lateral wall decompression for thyroid ophthalmopathy. Am J Ophthalmol 1989; 108:160–166.

14.McCord CD Jr. Current trends in orbital decompression. Ophthalmology 1985; 92:21–33.

15.Wolfe SA, Hemmy D. How much does moving the lateral wall help in expanding the orbit? Ophthal Plast Reconstr Surg 1988; 4:111–114.

16.Thaller SR, Kawamoto HK. Surgical correction of exophthalmos secondary to Graves’ disease. Plast Reconstr Surg 1990; 86:411–418.

17.Goldberg RA, Kim AJ, Kerivan KM. The lacrimal keyhole, orbital door jamb and basin of the inferior orbital fissure. Three areas of deep bone in the lateral orbit. Arch Ophthalmol 1998; 116:1618–1624.

18.Mourits MP, Koornneef L, Wiersinga WM, et al. Orbital decompression for Graves’ ophthalmopathy by inferomedial, by inferomedial plus lateral, and by coronal approach. Ophthalmology 1990; 97:636–641.

19.Goldberg, RA, Weinberg DA, Shorr N, Wirta D. Maximal, three-wall, orbital decompression through a coronal approach. Ophthalmic Surg Lasers 1997; 28(10):832–843.

20.Antoszynk JH, Tucker N, Codere F. Orbital decompression for Graves disease: exposure through a modified blepharoplasty incision. Ophthal Surg 1992; 23:516–521.

21.Kennedy DW, Goodstein ML, Miller NR, Zinreich SJ. Endoscopic transnasal orbital decompression. Arch Otolaryngol Head Neck Surg 1990; 116:275–282.

22.Olveri N. Transpalpebral decompression of endocrine ophthalmopathy (Graves’ disease) by removal of intraorbital fat: experience with 147 operations over 5 years. Plast Reconstr Surg 1991; 87:627–641.

23.Trokel S, Kazim M, Moore S. Orbital fat removal. Decompression for Graves’ orbitopathy. Ophthalmology 1993; 100:674–682.

24.Kazim M, Trokel SL, Acaroglu G, Elliott A. Reversal of dysthyroid optic neuropathy following orbital fat decompression Br J Ophthalmol 2000, 84(6):600–605.

25.Shorr N, Baylis HI, Goldberg RA, Perry JD. Transcaruncular approach to the medial orbit and orbital apex. Ophthalmology 2000; 107(8):1459–1463.

26.Balch KC, Goldberg RA, Green JP, Shorr N. The transcaruncular approach to the medial orbit and ethmoid sinus. Facial Plast Surg Clin North Am 1998; 6:71–77.

27.Goldberg RA, Shorr N, Cohen MS. The medial orbital strut in the prevention of postdecompression dystopia in dysthyroid ophthalmopathy. Ophthal Plast Reconstr Surg 1992; 8:32–34.

28.Goldberg RA, Relan A, Hoenig J. Relationship of the eye to the bony orbit, with clinical correlations. Aust NZ J Ophthalmol 1999; 27(6):398–403.

29.Wilson JL, Deschler DG, Kaplan MJ, Pilsbury HC. Complications of cranial base surgery. In: Weissler MC, Pillsbury HC, eds. Complications of Head and Neck Surgery. New York: Thieme, 1995:336–351.

30.Goldberg RA, Perry JD, Hortaleza V, Tong JT. Strabismus after balanced medial plus lateral wall versus lateral wall only orbital decompression for dysthyroid orbitopathy. Ophthal Plast Reconstr Surg. 2000; 16(4):271–277.

31.Hurwitz JJ, Birt D. An individualized approach to orbital decompression in Graves’ disease. Arch Ophthmol 1985; 103:660–665.

32.Garrity JA, Fatourechi V, Bergstralh EJ, et al. Results of transantral orbital decompression in 428 patients with severe Graves orbitopathy. Am J Ophthalmol 1993; 116:533–547.

33.Nunery WR, Nunery CW, Martin RT, Truong TV, Osborn DK. The risk of diplopia following orbital floor and medial wall decompression in subtypes of ophthalmic Graves’ disease. Ophthal Plast Reconstr Surg 1997; 13:153–160.