Nonorthopaedic Manifestations

Although the musculoskeletal system is mainly affected by AS, there are other serious effects of the disease. The most common extraskeletal abnormality is anterior uveitis (approximately 25% of patients), which is usually treated with topical agents and rarely leads to vision loss. Ankylosis in the thoracic spine and at the costovertebral joints leads to markedly diminished chest wall expansion, resulting in restrictive lung disease with decreased lung volumes (vital capacity and total lung volume) and increased dependence on diaphragmatic excursion. Preoperative pulmonary function tests are recommended, as is smoking cessation and aggressive postoperative pulmonary toilet. Some AS patients (25% to 30%) also have a component of ileitis or colitis or both; regardless, all patients for whom surgery is being contemplated should have a preoperative nutritional assessment and be considered for perioperative nutritional supplementation. The incidence of aortic stenosis and aortic valve insufficiency is increased in these patients as well; a preoperative echocardiogram is recommended routinely.

Physical Examination and Diagnosis

A patient with AS is most often a young man who gives a history of vague nonlocalizing back pain, morning stiffness, and possibly increasing difficulty with activities of daily living. Although women are affected with AS, men often present earlier or with more advanced disease. Physical examination and diminished spinal mobility especially in the sagittal plane are usually present. The Schober test is used to evaluate lumbar spinal motion: Points 10 cm above and 5 cm below the lumbosacral junction in the midline are marked on the patient in the fully upright position. With full forward flexion, there should be at least 5 cm of excursion between these two points. Chest expansion is commonly limited to less than 2.5 cm of excursion and is typically measured at the fourth intercostal space. The modified New York diagnostic criteria for AS were outlined in 1984 and are as follows:

The presence of sacroiliac inflammation and one of the other three criteria is generally considered enough to establish the diagnosis of AS.

Sacroiliitis is usually identified on an anteroposterior pelvis film (with or without a Ferguson view). It is widely accepted that the presence of sacroiliitis is crucial for the diagnosis of AS. Sacroiliac joint destruction is the earliest manifestation of AS. The earliest stages of sacroiliitis show some blurring of the cortical margins; this progresses to subcortical erosions (more commonly on the iliac side because it is less robust than the sacral side). In advanced stages, the sacroiliac joints become completely fused, and the cortical erosions disappear. Sacroiliac joint involvement usually is symmetrical and bilateral. Studies have suggested that the use of bony pelvis computed tomography (CT) or magnetic resonance imaging (MRI) in conjunction with plain radiographs may lead to earlier diagnosis of AS.⁷ It has yet to be determined whether this early diagnosis favorably affects clinical outcomes.

Management of Acute Injury

The spinal surgeon is usually not the physician making the initial diagnosis of AS but rather is called on to address spinal deformity caused by AS in the clinic and spinal trauma in an AS patient in an emergency setting. A trauma patient with AS also presents a challenge to the spinal surgeon. The spine in AS functions as a long rigid beam, acting much like a long bone. This altered biomechanical state, plus the presence of osteoporosis and the lack of ligamentous constraints, significantly decreases the fracture threshold of the ankylosed spine. The key to detecting fractures in these patients is having a high index of suspicion, especially after minor trauma. The cervical and cervicothoracic regions are the most commonly affected. Plain radiography is neither sensitive nor specific in these instances, although it should be used as an initial screen, and the standard radiographic modality used for the diagnosis of fracture in all patients is CT. Epidural hematoma, spinal cord injury, and disc injury can be visualized with MRI techniques.

A patient with AS may present with a progressive neurologic deficit without obvious bony injury or with progression of the deformity and increased pain. Many patients with missed spinal column injuries present at a later time to the clinic or the emergency department with progressive neurologic deficit or worsening of deformity or both. The evaluating clinician must also be aware of possible hyperextension through a fracture at a kyphotic segment, which may result in relatively normal sagittal alignment; attempts to determine the patient’s preexisting deformity from history and prior radiographs should always be made. There have been reports of neurologic injury in patients strapped to spine boards in a position of hyperextension when compared with their preinjury alignment.^8–10 Because of the stiff and osteoporotic spine, minor trauma may result in acute angulation or moderately rapid deformity progression. One should refrain from attempting acute correction through such a fracture. The patient should be initially immobilized in a halo vest in the preinjury alignment.

For a patient with a neurologic deficit, MRI is imperative. MRI may reveal an epidural hematoma. Hematomas can occur in these patients owing to bleeding from minor trauma from the osteoporotic bone or from scarred epidural vessels adjacent to a fracture. Evacuation of a hematoma is essential in the presence of progressive neurologic deficit. The decompression required may significantly destabilize the AS patient, and so the surgeon should be prepared to stabilize the spine at the same setting. Usually rigid instrumentation is required, although rarely halo immobilization may be sufficient for some cervical cases. As with instrumentation for elective cases, the screw-bone interface is compromised because of osteoporosis; these constructs should generally be supplemented by external support such as with a halo vest. Laminar hooks may be more rigid in many patients, but external bracing should still be considered.

It is generally accepted that AS patients sustain more spinal fractures and dislocations than individuals without AS.^11–15 Cooper and colleagues¹⁶ retrospectively looked at 158 patients in Rochester, Minnesota, with AS and found a sevenfold increase in the incidence of spinal fractures over that of a cohort of patients without AS. They found no such increase in extremity fractures. The patients with spinal fractures tended to be older and had a greater preinjury involvement of the spine than patients without fractures. Cooper and colleagues¹⁶ also noted that this higher incidence was mainly during the first 5 years after diagnosis and suggested that this was due to a greater percentage of bone density loss during this period, resulting in a decreased fracture threshold. In addition, the dampening structures present in a normal spine have lost their load-absorbing qualities in the ankylosed spine. The intervertebral discs are stiff, as are the ligamentous structures, and the facet joints are ankylosed.

The incidence of neurologic injuries in these patients is quite high, owing to excessive bleeding at the fracture site leaking into the confined epidural space and translation (displacement) at the fracture site. This translation causes direct injury to the spinal cord and persistent bleeding owing to motion, resulting in an enlarging compressive hematoma. Most spinal injuries in AS patients are three-column injuries (owing to stiffening of the load-absorbing structures). These injuries are highly unstable because there are two long lever arms hinging at the fracture site. In addition, the presence of preinjury kyphosis increases the likelihood of translation at the level of the injury, which subsequently increases the likelihood of neurologic injury. Lastly, poor bone stock and difficult radiographic evaluation can lead to a delay in diagnosis. Most of these injuries (60% to 75%) are at the cervicothoracic junction, which is notoriously difficult to evaluate with plain radiographs.

Whang and colleagues¹⁷compared a cohort of 12 patients with AS who sustained spinal injuries with 18 patients with diffuse idiopathic skeletal hyperostosis (DISH) who sustained spinal injuries. The DISH group represents a group of patients of similar age whose spinal condition results in stiff segments above and below any spinal fracture. Falls from a standing position were the most common mechanism of injury. There was a greater likelihood that the DISH patients did not incur any neurologic deficit (44.4%) compared with AS patients (25% of whom did not have a neurologic deficit). Complication rates were higher in the AS group (42% vs. 33% in the DISH group). There were two deaths in each group related to the injury or its treatment, all of which were considered to be related to the use of the halo vest (aspiration [two deaths], respiratory failure, and multisystem organ failure). Several patients died of unrelated causes during the follow-up period; however, all surviving patients were contacted and were classified as having excellent or good outcomes.

Finkelstein and colleagues¹⁸ looked retrospectively at 21 AS patients with a diagnosis of spinal trauma. One third of these patients had a delay in diagnosis; three had complete spinal cord injuries on presentation, and three experienced neurologic deterioration to complete spinal cord injuries after admission. Finkelstein and colleagues¹⁸ recommended quick screening cervical and thoracic MRI (one film) and screening lumbosacral spine MRI (one film) for diagnosis, in addition to minimal transfers and immediate stabilization. They did not comment on their definitive protocol for treatment of these patients (operative vs. nonoperative).

Hitchon and colleagues¹⁹ retrospectively reviewed 11 patients with AS and thoracic and lumbar fractures. They found 10 of these patients had sustained three-column injuries; 9 patients had extension-type injuries. More than half of these patients had a neurologic deficit (the specifics of which the study authors did not mention); half of these neurologically injured patients had some improvement in function. Hitchon and colleagues¹⁹ recommended surgical intervention for stabilization of thoracic and lumbar three-column injuries because of their inherent instability.

Graham and van Peteghem²⁰ looked retrospectively at 15 patients over 6 years (1978-1984) comparing types of injuries and treatments. Of patients, 12 had cervical spine injuries; 9 of these had spinal cord injuries. The two patients with thoracic injuries had anterior cord syndromes. There were no compression-type injuries; most injuries resulted from a flexion-extension type of mechanism. The only patient treated with operative intervention was the patient with the lumbar injury, who had hardware failure and had to undergo revision. Two patients died, and three patients had pulmonary complications.

Apple and Anson²¹ looked at AS patients with spinal fracture and spinal cord injury, comparing operative versus nonoperative treatments. This study was a retrospective, multicenter study of 59 patients. In the operative group, 37 patients were treated with a variety of procedures. Patients in the nonoperative group were placed in halo traction and then halo vests and placed on bed rest. There were no significant differences between the two groups with regard to motor recovery, fusion complications, or mortality rate (22% in both groups). The nonoperative group did have significantly shorter hospital stays. No analysis of the patients according to type of injury or treatment was done, and no discussion of the deaths was presented.

Hunter and Dubo¹² reviewed the cases of 19 AS patients who had sustained cervical spine fractures. Five of these patients had a complete spinal cord injury, and all of these patients died after their injury. All of these patients were treated nonoperatively. No patient developed neurologic deterioration, and all of the patients with incomplete cord injury regained some function. Hunter and Dubo¹² concluded that nonoperative treatment worked well in these patients, although they suggested that surgery be considered in patients with grossly unstable injuries.

Bohlman retrospectively reviewed 300 patients with cervical spine injuries.^21a He found only eight patients who carried a preinjury diagnosis of AS. Five of these patients died of pulmonary or gastrointestinal causes. Clinically significant epidural hematomas were found only in the AS patients. Bohlman recommended decompression for patients with progressive neurologic deficit. There was a delay in diagnosis in four patients, all of whom developed spinal cord injuries.

The generally accepted protocol with respect to the management of spine trauma in AS patients is as follows: If the clinician has even the slightest suspicion of spinal injury, the patient should be immobilized in the preinjury position. Plain radiography and fine-cut CT with reconstructions should be obtained. If the patient has a neurologic injury, MRI should be considered, looking for an epidural hematoma. If a fracture is detected and displacement or gross instability is noted, low-weight in-line traction should be used in an attempt to facilitate a reduction. If a reduction is obtained, the patient should be placed in a halo vest for definitive treatment. If a reduction cannot be obtained, internal fixation is recommended, with or without decompression as indicated by the patient’s neurologic status. Postoperative immobilization in the form of a halo vest is then recommended. In a patient with a progressive neurologic deficit, MRI is likely to reveal the presence of a hematoma. In a patient with a stable deficit and no hematoma, the cord injury likely occurred at the time of injury; as long as the spine is stable, management of the neurologic injury should be expectant. If the spine is unstable, reduction and stabilization either with traction and a halo vest or with surgery is recommended.

Complications can arise at the time of injury and from treatment of the injury. Deformity and neurologic injury can occur as a result of the injury; treatment with decompression and internal fixation carries risks of nonunion, hardware failure, failure of the bone-screw interface resulting in loss of fixation, and infection. Even halo management has complications. Skull fractures, pin tract infections, intracerebral hemorrhage, and intracranial air all have been reported with halo immobilization in these patients.^9,22 Taggard and Traynelis²⁴ described a posterior cervical fusion (lateral mass plating) they used in seven AS patients who had sustained fractures. The fusions were supplemented with autologous rib grafts. Postoperatively, the patients were immobilized in collars only, with the exception of one, who was placed in a sternal-occipital-mandibular immobilizer. Fusion occurred in all patients; there were two deaths in quadriplegic patients. Taggard and Traynelis²⁴ recommended operative intervention as a means of avoiding postoperative halo immobilization.

Deformity

The deformities seen in AS are a result of excessive kyphosis throughout the spine and excessive flexion at the hip joints. All areas of the spine can be affected, with the lumbar spine affected most often, followed by the thoracic and cervical spine. If the physical examination and radiographic examinations indicate that the hip flexion contracture plays a significant role in the overall deformity, hip arthroplasty should be performed before spinal surgical intervention. In this manner, the less morbid operation is performed first; in addition, the spine surgeon can better assess the actual amount of sagittal imbalance attributable to the spine.

The spine surgeon must carefully examine the patient in the standing, seated, and supine positions to determine the major component of the deformity. If a major portion of the deformity corrects on moving from a standing to a seated position, the deformity is mostly from the hip joints, and arthroplasty should be performed first. If the deformity persists on sitting but corrects in the supine position, the deformity is arising from the thoracic, thoracolumbar, or lumbar spine, and a lumbar osteotomy is usually indicated. If the deformity persists even in the supine position, the deformity is in the cervicothoracic area, and an osteotomy in this area is indicated.

From a radiographic standpoint, a full spine lateral radiograph with the neck in a neutral position and the hips in a fully extended position is crucial for surgical planning. This radiograph allows measurement of the chin-brow angle, which is formed by a line from the chin-brow to the floor vertical angle. This measurement is helpful when planning any osteotomy. Ideally, the chin-brow angle should be zero. Suk and colleagues²³ looked at the significance of the chin-brow measurement in assessing the success of surgical intervention. These investigators evaluated 34 AS patients undergoing lumbar or thoracolumbar osteotomies for correction of sagittal imbalance. Preoperative and postoperative chin-brow angles were measured. Clinical outcome assessment involved the Modified Arthritis Impairment Scales (AIMS). This questionnaire consists of three simple questions plus numerous subscales: function, indoor activity, outdoor activity, psychosocial activity, pain, and overall subjectivity. Suk and colleagues²³ found improved postoperative AIMS scores for questions involving looking forward, going up stairs, and going down stairs. There was a negative correlation between chin-brow angles and correction obtained but no correlation between chin-brow angle and clinical outcome. The patients who were overcorrected (to an angle <−10 degrees) had worse scores with regard to looking forward and going down stairs; these results were found to be statistically significant.

When the location of the primary spine deformity is determined, the surgeon must decide what type of osteotomy would be most appropriate. It is preferable to place the osteotomy at the apex of the deformity, but this is not always possible. Thoracic and thoracolumbar osteotomies are limited by the rib cage, the spinal cord, and the conus medullaris. Deformities in these areas are almost always treated with a lumbar osteotomy. By moving the osteotomy inferiorly, one can obtain more sagittal plane alignment owing to a longer lever arm. By overcorrecting at the lumbar level, one can address the thoracic kyphosis and the lumbar kyphosis. If overcorrection is to be performed, the surgeon should take into account if a portion of the deformity is cervicothoracic because the patient’s horizontal gaze would be affected and may not be restored.

Preoperative Assessment

A careful history should yield information about the patient’s lifestyle, habits, and medications. Smoking cessation is imperative; many surgeons do not undertake the operation while a patient is actively smoking. Nonsteroidal anti-inflammatory drugs should be discontinued at least 2 weeks before surgery. Preoperative pulmonary function tests are indicated because many of these patients have restrictive lung disease. Preoperative echocardiography may also be indicated, as previously mentioned, although cardiac intervention is not often needed. Results of renal function tests should be obtained before surgery; an awareness of tenuous renal function would benefit intraoperative and postoperative fluid management. Many AS patients have a component of renal dysfunction because of long-term use of nonsteroidal anti-inflammatory drugs. Cervical spine flexibility should be assessed by the orthopaedist and the anesthesia service before the procedure; fiberoptic intubation is usually needed because of concomitant ankylosis of the cervical spine.

Historically, most osteotomies in AS patients were performed with the patient awake or using a Stagnara wake-up test for evaluation of spinal cord function before, during, and after a correction. Somatosensory evoked potentials and motor evoked potentials (epidural or transcranial) have become more reliable so that the wake-up test is less frequently needed. The anterior tracts are better monitored by motor evoked potentials. These tracts can be preferentially affected during an extension maneuver when addressing kyphotic deformities, either by direct compression or by impairing the vascular supply to the spinal cord.

A preoperative nutritional assessment (albumin, prealbumin, total protein) should be performed; perioperative nutritional supplementation (tube feedings or parenteral nutritional assessment) may be indicated. Klein and colleagues²⁵ noted a significant increase in complications such as deep wound infection in patients undergoing lumbar spinal fusion who were malnourished by nutritional parameters preoperatively. Hu and colleagues²⁶ and Lapp and colleagues²⁷ showed that supplementation in the form of parenteral nutrition is beneficial in reducing complication rates after reconstructive spine surgery.

Lumbar Osteotomies

The first lumbar osteotomy was described in 1945 by Smith-Petersen.³³ This osteotomy is an opening osteotomy, meaning that the apex of the wedge lies posteriorly, opening up the anterior column during correction (osteoclasis through ossified disc space and anterior longitudinal ligament). Smith-Petersen and colleagues performed multilevel osteotomies in six patients. These osteotomies were V-shaped in the coronal plane, with the point of the “V” at the midline in the interlaminar space. The osteotomies are carried out through the articular processes bilaterally at two or three levels. It is imperative that adequate amounts of lamina and flavum are resected before correction so that compression of the neural elements on closure does not occur.

Cauda equina syndrome has been reported by Simmons²⁸ as a result of a decrease in canal dimensions. The posteriorly based closing wedge type of osteotomy results in anterior opening at the level of the disc space. This anterior opening can be better achieved in an AS patient than a patient without AS because of the stiffness of the disc space. Complications of an opening wedge osteotomy include superior mesentery artery syndrome and aortic rupture owing to stretching of the abdominal vasculature.^29,30 Vascular complications are rare and tend to occur in older patients with calcific, adherent abdominal vessels.

Lichtblau and Wilson²⁹ described a patient who underwent closed osteoclasis followed by cast placement. This patient died in the immediate postoperative period of an aortic rupture. His history was significant for a large dose of radiation that was used to treat the ankylosed spine. Fazl and colleagues³⁰ described an AS patient who sustained a fracture through the T12-L1 disc space. He was treated with Harrington rod instrumentation and fusion but died 2 days postoperatively from an aortic rupture at the level of the injury. Aortic necrosis was present at autopsy, as were adhesions of the vessels to the spine. More common complications reported include ileus, pneumonia, and root traction injury. Cauda equina syndrome with flaccid paralysis below the level of injury, although rare, was reported in these studies as well.

Patients originally were immobilized in plaster; segmental instrumentation currently is indicated for these patients. Many investigators have reported their results after multilevel Smith-Petersen osteotomies.^31–33 Nonunion rates resulting in recurrence and progression of deformity were significant. Soon after the original description of this technique, reports of “plugging up” the open disc spaces with interbody fusions showed increased fusion rates and decreased complications.³⁴ The complications associated with opening wedge osteotomies led to modifications of Smith-Petersen’s techniques. In 1949, Wilson and Turkell^34a described a procedure similar to the Smith-Petersen procedure in which less bone is removed but more osteotomies are created. The anterior longitudinal ligament is not ruptured; the anterior column length is not changed. In 1962, McMaster³² described the addition of Harrington compression instrumentation to Smith-Petersen osteotomies in 14 patients. This instrumentation was used to close the wedges produced after osteotomy. Postoperatively, the patients were placed in casts for 9 months. Mean correction was 33 degrees at final follow-up. Subjective improvement was found in horizontal gaze and height and posture. McMaster³² suggested that a slow controlled osteotomy closure was beneficial in terms of overall stability and protection of neural elements. Püschel and Zielke^34b also performed multiple wedge-shaped Smith-Petersen type osteotomies and used Zielke instrumentation to close the osteotomies. They also recommended a slow correction with a gradual lordosis.

After reports of nonunions and concerns about stretching of the abdominal vasculature and viscera,^35,36,39 Thomasen³⁷ described a closing wedge osteotomy. He reported on 11 patients in whom he preformed a complete laminectomy at L2, transected the transverse processes, and resected the ankylosed facets at L2-3. The pedicles of L2 were removed in their entirety down to the posterior aspect of the vertebral body. The entire vertebral body was decancellated, followed by removal of the posterior cortex and osteotomies of both lateral cortices. After careful mobilization of the dura above and below L2, Thomasen³⁷ closed the wedge by gradual flexion of the table. Internal fixation (plates and wiring) was used in six patients; all patients were placed in casts. One patient had a fracture-dislocation above the level of the osteotomy resulting in a cauda equina syndrome; this patient had almost complete return of neurologic function after revision decompression and internal fixation. Correction ranged from 12 to 50 degrees. All patients had subjective improvement of posture and horizontal gaze.

Heinig’s eggshell procedure was described in 1984 as a monosegmental osteotomy to be used in the same situations in which one would use Thomasen’s procedure.³⁷ Thomasen leaves the anterior vertebral body cortex intact, whereas Heinig actually describes fracturing this cortex, which decreases the length of the anterior column and the posterior column. As long as more bone is removed posteriorly, restoration of lordosis occurs.

Bradford and colleagues³⁸ reported in 1987 on a series of 21 patients with AS who underwent single-level or multilevel lumbar or thoracic osteotomies with or without anterior discectomies. All patients had internal fixation posteriorly with a thoracolumbosacral orthosis. Average corrections ranged from 9 to 36 degrees. Complications were noted more frequently in the closing wedge–type osteotomies (neurapraxias and fracture during hook placement). Wake-up tests were used in all patients. In addition, Bradford and colleagues³⁸ recommended closing-type osteotomies to avoid traction on the spinal cord, wide decompression, and internal fixation to avoid neurologic complications.

In 1990, Hehne and colleagues³⁹ reported on 177 patients with AS in whom multisegmental opening wedge lumbar and thoracolumbar osteotomies were performed. Hehne and colleagues³⁹ were the first to report on the use of pedicle screw fixation. Casting and bracing were used postoperatively. Average correction at follow-up (18 to 42 months) was 43 degrees, with horizontal gaze subjectively restored in all cases. Complications included deaths, transient paresis, transient and permanent nerve root injuries, implant failures, and infections. These authors suggested that pedicle screw fixation with multisegmental osteotomies can produce a smoother lordosis than that produced by a monosegmental osteotomy.

In 1992, Jaffray and colleagues^39a presented three patients in whom a decancellation closing wedge osteotomy was performed. They did not remove the entire pedicle; rather, the inferior aspect of the pedicle was preserved to ensure protection for the exiting nerve root. Pedicle screw fixation plus a postoperative cast was used. Jaffray and colleagues recommended two-level osteotomies (L2 and L4) for patients who needed more correction. Horizontal gaze was corrected in two patients; one patient required a cervicothoracic osteotomy for complete gaze correction. Complications were not discussed.

More recently, Van Royen and De Gast⁴⁰ mathematically analyzed the sagittal plane corrections of two patients and determined that the amount of correction needed depends on three parameters: sacral endplate angle, C7 plumb line, and chin-brow angle. The sacral endplate angle reflects the amount of sagittal plane deformity that can be attributed to the hip joints: As the flexion contracture at the hip increases, the pelvis must rotate posteriorly to keep the body center of mass over the pelvis, decreasing the sacral endplate angle. The chin-brow angle has been shown to be a quantifiable parameter that reflects the restoration of horizontal gaze.⁴⁰ A mathematical formula was found that determines the ideal location and angle for each particular patient for a closing wedge–type osteotomy centered on the anterior longitudinal ligament.

There have been many retrospective reviews of AS patients treated with various osteotomies for sagittal imbalance. Van Royen and De Gast⁴⁰ performed a meta-analysis of 856 AS patients. They found three different techniques described: multisegment (two to three levels) opening wedge osteotomies with rupture of the anterior longitudinal ligament (i.e., Smith-Petersen), multisegment closing wedge osteotomies (Wilson-Turkell), and closing wedge–type osteotomy with pedicle resection and an anterior hinge (Thomasen). After a thorough and careful review, Van Royen and De Gast⁴⁰ concluded that although no single technique was clearly superior to the others, the complications associated with closing wedge osteotomies were less serious than the complications associated with the other two groups. In addition, loss of correction was more prevalent in patients treated with opening wedge and polysegmental wedge osteotomies and the closing wedge types.

A handful of studies have attempted to quantify results in terms of patient outcomes using a standardized grading system. In 1995, Halm and colleagues⁴¹ used the modified AIMS questionnaire to evaluate 175 patients retrospectively after lumbar osteotomy. Treatment groups were multisegment Smith-Petersen with Harrington compression instrumentation (n = 34), multisegment Smith-Petersen with transpedicular fixation (n = 136), and monosegmental Thomasen with segmental fixation (n = 4). The investigators found statistically significant improvement in 47 of 60 items. Kim and colleagues⁴² used the AIMS questionnaire prospectively in 45 patients with AS who were treated with Thomasen osteotomies at one or two levels. Osteotomies were mainly performed in the lumbar spine (usually L3). Average increase in lumbar lordosis was 34 degrees, with no significant increase in thoracic kyphosis. All parameters measured were significantly improved. Clinical outcome scores were significantly improved in all five categories; no correlation was found between the amount of radiographic correction obtained and clinical outcome as measured by the questionnaire.

Berven and colleagues⁴³ looked at 13 patients undergoing transpedicular wedge resection. Three of these patients had AS and were having spine surgery for the first time. These investigators also used outcome measures (modified Scoliosis Research Society questionnaire) in a retrospective manner (Figs. 35-1 to 35-4). After 2 years, most of these patients was satisfied and would have the surgery again. The changes in C7 plumb line and lumbar lordosis were statistically significant. Complications included dural tear, transient nerve root injury, pulmonary embolus, and loss of sagittal balance. None of the AS patients showed a loss of sagittal balance at follow-up.

FIGURE 35–1 A and B, Clinical photographs of a patient with ankylosing spondylitis with typical loss of lumbar lordosis, forward sagittal balance, and flexed hips and knees to maintain stance.

FIGURE 35–2 Lateral (A) and posteroanterior (B) radiographs. Note ankylosis of sacroiliac joints, bamboo spine with syndesmophyte formation, and forward sagittal balance. The patient had hip replacement previously for severe hip disease.

FIGURE 35–3 Lateral (A) and posteroanterior (B) radiographs after pedicle subtraction osteotomy. Note improved sagittal balance. Closing wedge of L3 is seen on lateral radiograph and can be appreciated by widely divergent screws in L2 compared with L4.

FIGURE 35–4 Patient back (A) and side (B) views showing clinical correction.

Bridwell and colleagues⁴⁴ looked at 27 patients undergoing pedicle subtraction osteotomy, also in a retrospective fashion. Two of these patients had AS. Outcome data (Oswestry and SRS-24) were also obtained retrospectively. Bridwell and colleagues⁴⁴ found a significant improvement in sagittal balance and lumbar lordosis and a high level of patient satisfaction. Complications included deep vein thrombosis, myocardial infarction, compartment syndrome, visual field loss, pseudarthrosis, loss of correction, urinary retention, and neurologic deficits (root lesions). The patients with the latter two complications all responded to a central canal decompression.

Thoracic Osteotomies

Even when the deformity has been localized to the thoracic spine primarily, lumbar osteotomies are usually recommended. These can be performed below cord and conus level, and they have the advantage of large degrees of correction owing to a long lever arm. Osteotomies performed in the thoracic spine are usually Smith-Petersen, although there have been a handful of studies describing closing wedge osteotomies in the thoracic and thoracolumbar spine.

Kawahara and colleagues⁴⁵ described a closing-opening wedge osteotomy in the thoracic and thoracolumbar spine. They used this procedure on seven patients with sagittal imbalance. The osteotomy consisted of a partial vertebrectomy with a large posterior wedge that is performed in a manner similar to a costotransversectomy. After bony resection, pedicle screw instrumentation plus temporary correction rods are used to facilitate a closing wedge correction of about 30 degrees. An opening wedge–type maneuver is facilitated, again through the instrumentation, and a spacer or allograft is inserted. Kawahara and colleagues⁴⁵ noted good improvement in kyphosis, lordosis, and plumb line. They had no neurologic complications, no nonunions, and no loss of correction (follow-up of 2.2 to 7.5 years).

Cervicothoracic Osteotomy

If the primary deformity has been determined to be in the cervical spine, an osteotomy at the cervicothoracic junction can be done. Patients with these deformities, in addition to the problems with horizontal gaze, also can experience dysphagia and problems related to poor oral intake. The chin-brow angle is of paramount importance when planning a corrective osteotomy in the cervicothoracic region. A key point is not to overcorrect the horizontal gaze because this can lead to inability of patients to see the floor ahead of them. These patients may function better when corrected to a chin-brow angle of about 10 degrees. In 1958, Urist⁴⁶ described an osteotomy in the cervicothoracic region, noting that the canal at this level is quite large and that the C8 nerve root is quite mobile compared with the upper cervical roots. In addition, potential loss of the lower cervical roots is less morbid than loss of the upper cervical roots. Lastly, the vertebral arteries are typically extraosseous at these levels, making resection of the lateral masses less risky.

Careful preoperative planning with radiographic studies is of paramount importance when performing a cervicothoracic osteotomy. Full-length standing lateral spine radiographs are needed to measure the chin-brow angle. Lateral tomography or fine-cut CT with sagittal and coronal reconstructions is performed to delineate the anatomy.^28,46 Axial CT scans can be very helpful in characterizing the distorted anatomy often seen in patients with AS, especially with regard to placement of instrumentation. MRI can help to rule out occult fractures, which should be suspected if there is recent onset of pain or rapid progression of deformity; MRI should be obtained if there is any neurologic deficit. In addition, flexion and extension lateral cervical spine radiographs should be obtained to look for any instability occurring at the occipitocervical and atlantoaxial levels. A subset of AS patients develop instability in these areas as a result of excessive stiffness of the entire spinal column because the stress placed on these upper cervical areas can be quite high. Although the number of AS patients who have occipitocervical or atlantoaxial instability is not as high as the number of AS patients with rheumatoid arthritis, missing this instability can be catastrophic.

The surgeon must carefully examine the preoperative radiographs and CT scans to determine the amount of correction needed. As described by Simmons,²⁸ the measured chin-brow angle should be transposed onto a neutral cervicothoracic film, with the apex of the angle centered on the posterior longitudinal ligament at the C7-T1 level. By extrapolation, the extent of posterior elements to be resected can be determined.

As originally described by Urist⁴⁶ and popularized by Simmons, the procedure was performed under local anesthesia only, with the patient awake in a seated position, in seated halo traction. The seated position carries with it the risk of air embolus in a patient with a patent foramen ovale so that continuous cardiac monitoring is indicated during the procedure.^28,46 With the widespread use of neurophysiologic monitoring, the awake seated position is almost never used for these procedures at the present time except in the case where the severe spinal deformity precludes prone positioning. A baseline set of somatosensory evoked potentials and motor evoked potentials is obtained, and the patient’s head is placed in a rigid head holder (three-pin Mayfield). After the flip, a repeat run of neurophysiologic monitoring is obtained.

After a wide and lengthy exposure, a wide cervical decompression at C7 and T1 is performed, including a dorsal unroofing of bilateral C8 nerve roots. The lateral masses at C7 are also resected; in addition, the pedicles at C7 are partially removed to ensure that the C8 nerve roots are not compressed after the correction. Sometimes a portion of the superior aspect of the T1 pedicles must be removed as well. Before the widespread use of instrumentation, postoperative immobilization consisted of a halo vest or cast that was custom measured or fitted to the patient’s torso preoperatively. Currently, sublaminar hooks or lateral mass and pedicle screw instrumentation can be used for rigid stabilization, which greatly reduces the likelihood of translation at the osteoclasis level and subsequent neurologic injury. Lateral mass screws generally do not hold well in these patients with their osteoporotic spines and should be supplemented by halo vest immobilization if the surgeon prefers to use them. Instrumentation is placed three levels above and three levels below C7-T1 after the decompression but before osteoclasis.

After the instrumentation has been placed, a temporary rod is prepared. One surgeon breaks scrub and manipulates the head and neck via the halo ring, facilitating the osteoclasis. The neck is slowly extended about the C7-T1 level, while the scrubbed surgeon watches the decompression site for excessive dural compression. Neurophysiologic signals are carefully monitored during the correction. An audible crack is often heard; the manipulating surgeon should appreciate a decrease in resistance when the osteoclasis has been completed. If the decompression has been planned and executed properly, any residual lateral mass of C7 should be opposed to T1, the C8 nerve roots should be free, and the dura should not be excessively compressed. The halo or Mayfield is resecured to the operative table in the corrected position. The rods are secured into position.

Even with neurophysiologic monitoring, a wake-up test is often performed after correction. Adjustments can be made intraoperatively after the main correction has been obtained. After the instrumentation has been secured, the local bone resected during the osteotomy is used for grafting, supplemented by allograft if the surgeon believes this is indicated, and the wound is closed. The anterior soft tissues are usually tight after being in a shortened position for a long time. The patient is usually kept intubated until the soft tissue edema and the postoperative anterior hematoma lessen. Even with rigid segmental instrumentation, the patient’s osteopenia may necessitate placement in a rigid halo vest for the duration of the healing period. Dysphagia is common and usually resolves with time. If the patient’s caloric intake is borderline, as is common, it is imperative that these patients receive nutritional supplementation during the healing period, in the form of either tube feedings or parenteral supplementation.

The literature concerning cervicothoracic osteotomies in AS patients is sparse. Simmons^46a in his original article in 1972 reported on 42 patients who underwent cervicothoracic osteotomy as described by Urist. The operations were performed in the seated position under local anesthesia, and postoperative immobilization consisted of a halo vest. Simmons reported two nonunions successfully treated with anterior fusion, one pulmonary embolus, two myocardial infarctions (one fatal), and one root injury treated with repeat decompression. The patients who did not experience complications all were quite satisfied with their outcomes and had their horizontal gaze restored.

McMaster³² reported retrospectively on 15 patients with abnormal horizontal gaze who were treated with an extension osteotomy at the cervicothoracic junction. These surgeries were performed in the prone position with a halo jacket in place before the osteotomy. Only three patients had internal fixation. All patients had their horizontal gaze restored. Complications included one patient with delayed postoperative quadriparesis, two nonunions, a C8 nerve root lesion, and subluxation at the osteotomy site. McMaster³² also treated their nonunions with anterior fusion with subsequent good results.

The remainder of the literature dealing with cervicothoracic osteotomies in AS patients is in the form of case reports. Sengupta and colleagues^46b addressed the complication of overcorrection resulting in the inability of the patient to look down. They performed a same-day four-stage procedure in the lateral decubitus position with transparent drapes and reported restoration of horizontal gaze in one patient. They recommended this procedure only in extreme cases.

Summary

Spinal reconstructive surgery in a patient with AS is a complex and high-risk procedure. These patients have significant disability from their spinal deformities, however, and can experience significant benefit from correction of their alignment and sagittal imbalance. Careful preoperative planning, a clear understanding of the characteristics of the spines of AS patients, and meticulous intraoperative and postoperative care can lead to measurable improvement in quality of life for these patients.