Abstract
Background: Chin-on-chest deformity is a disabling manifestation of
ankylosing spondylitis. Surgical treatment consists of extension osteotomy at
the cervicothoracic junction. The purpose of this study was to characterize
the clinical presentation of this deformity and to determine the long-term
functional and radiographic outcomes of treatment.
Methods: The medical records and radiographs of all twenty-six
patients treated with cervicothoracic extension osteotomy by one of us between
1976 and 2001 were retrospectively reviewed. Three patients died during the
two-year-minimum follow-up period. The remaining twenty-three patients were
followed for an average of 4.5 years (range, two years to twenty-one years and
ten months).
Results: The mean sagittal correction was 38°. Delayed union in
two patients and additional cervical trauma in two others resulted in partial
loss of the initial correction. Quadriplegia developed in one patient, who
died as a result of subluxation at the osteotomy site. Five patients had
irritation of the eighth cervical nerve root postoperatively.
Conclusions: Extension osteotomy can reliably improve sagittal
alignment and horizontal gaze as well as decrease neck pain, eating
difficulties, and neurologic abnormalities. Internal fixation is recommended
to prevent subluxation, delayed union, nonunion, loss of correction, or
neurologic injury. There is a risk of death or catastrophic neurologic injury
from the procedure.
Level of Evidence: Therapeutic Level IV. See Instructions
to Authors for a complete description of levels of evidence.
Ankylosing spondylitis is often complicated by kyphotic spinal
deformity. Usually the hyperkyphosis is distributed throughout the thoracic
and lumbar spine, but occasionally a patient will present with a
disproportionate kyphosis in the cervicothoracic spine. The resulting
chin-on-chest deformity may contribute to problems with horizontal gaze,
chewing and swallowing, hygiene, neck pain, and myelopathy.
The surgical treatment to correct symptomatic chin-on-chest deformity
secondary to ankylosing spondylitis is an extension osteotomy and arthrodesis
at the cervicothoracic junction. To the best of our knowledge, there has been
no long-term follow-up study of a large series of patients with chin-on-chest
deformity who were treated surgically with one technique. Mason et al.
established the importance of performing the osteotomy at the cervicothoracic
junction, below the entrance of the vertebral arteries into the transverse
foramina at the sixth cervical
vertebra1. In 1958,
Urist described the same surgical technique, with use of a local anesthetic in
a seated, awake
patient2. Several
other authors have reported on series of patients treated with minor
modifications of Urist's
technique2-7.
However, in those series, the duration of follow-up was no more than three
years for any given patient or was not specified, preoperative neurologic
abnormalities were not described, and the sagittal alignment of the cervical
spine was not clearly defined. The purpose of this study was to characterize
the clinical presentation of chin-on-chest deformity secondary to ankylosing
spondylitis and to determine the long-term functional and radiographic
outcomes of treatment with cervicothoracic extension osteotomy.
Between 1977 and 2001, the senior author (H.H.B.) performed
twenty-six spinal osteotomies for chin-on-chest deformity
(Figs. 1-A, 1-B, and 1-C) in
patients with ankylosing spondylitis (see Appendix). The medical records and
radiographs of each patient were evaluated retrospectively. An attempt was
made to contact each patient by telephone to request that they return for an
additional follow-up evaluation, but only two patients actually returned. Each
patient was evaluated preoperatively; at six weeks, three months, six months,
and one year postoperatively; and then annually for a minimum of two
years.
Three patients died during the two-year-minimum follow-up period. Two of
the deaths were late deaths that were believed to be unrelated to the
osteotomy. One patient (Case 6; see Appendix), who was diagnosed with an
esophageal carcinoma two months after the osteotomy, died six months after the
osteotomy secondary to respiratory failure and congestive heart failure. The
other patient (Case 7) died of sudden cardiac arrest, while in a nursing home,
nine months after the osteotomy. No autopsy was performed. The third patient
(Case 14) became quadriplegic after the osteotomy and died of pneumonia and a
cardiac arrest in the immediate postoperative period. The average duration of
follow-up for the remaining twenty-three patients was four years and six
months (range, two years to twenty-one years and ten months). There were
twenty-five men and one woman. Their average age at the time of surgery was
fifty-one years (range, thirty-one to seventy-three years).
The medical history of each patient had been recorded by the senior author
and included information about trauma related to the onset of the deformity
and the patient's subjective descriptions of neck pain, swallowing
difficulties, field of vision, and neurologic symptoms. The severity of each
of these problems was rated by the patient at each visit as mild, moderate,
severe, or none. Twelve patients had a history of trauma preceding noticeable
progression of neck pain and/or deformity. At the time of presentation, all
twenty-six patients had difficulty with horizontal gaze, twenty-five had neck
pain, twenty had difficulty swallowing, and thirteen patients had difficulty
shaving.
The findings of the physical examination of each patient, which included a
thorough assessment of the neurologic status, had been recorded by the senior
author at each office visit. A Nurick grade was assigned to classify the
neurologic status of each patient at each
time-point8. The
severity of the deformity was assessed qualitatively, but consistent,
standardized measurements were not made. Of the twenty-six patients, sixteen
were neurologically normal at presentation and ten had neurologic deficits,
including seven with hyperactive reflexes and a minor gait abnormality (Nurick
grade 1 or 2) and three with a major gait disturbance or quadriparesis (Nurick
grade 3 or 4) (see Appendix). Two of the three patients with a Nurick grade of
3 or 4 had a history of trauma that had caused a fracture of the lower
cervical spine or the cervicothoracic junction. The fractures had been treated
with laminectomy and halo immobilization, with worsening of the neurologic
status prior to presentation. The third patient had been treated with a halo
vest for more than a year prior to presentation.
The radiographs of each patient were evaluated by orthopaedic surgeons who
were not involved in the care of the patients. The deformities were measured
on plain radiographs, tomograms, sagittal reconstructions of computerized
axial tomography scans, or sagittal magnetic resonance images of the
cervicothoracic spine. The initial degree of correction and the subsequent
loss of correction were determined by comparing measurements from identical
reference levels for each patient. The reference levels were determined for
each patient individually, depending on which levels were clearly visible on
all preoperative and postoperative radiographic studies. Generally, the
reference levels were from the superior end plate of C5 to the inferior end
plate of T5.
Cervical alignment was measured preoperatively in the sagittal and coronal
planes from the inferior end plate of C2 to the inferior end plate of C6. The
average preoperative sagittal alignment of the cervical spine was 12° of
lordosis, with a range of 2° of kyphosis to 31° of lordosis. Thus, the
deformity was mainly in the upper thoracic spine or at the cervicothoracic
junction. Fusion status was determined with serial radiography and with either
plain or computerized tomography of the cervicothoracic junction. The absence
of a persistent lucent line at the osteotomy site or the presence of
continuous trabecular bone across the level of the osteotomy was considered to
be evidence of a solid fusion.
Three patients had had a previous atlantoaxial arthrodesis for the
treatment of instability, and twenty-four patients had complete ankylosis of
the cervical spine caudad to C2. No patient had evidence of scoliosis in the
cervical spine.
Surgical Procedure
Surgery was performed, with the use of local anesthesia, while the patient
was awake and seated and in halo
traction9
(Figs. 2-A through 2-D). Spinal
cord monitoring was not used since we believe that an alert patient serves as
the most accurate spinal cord monitor. Complete laminectomy was performed at
the C6 and C7 levels. Complete foraminotomies were carried out bilaterally
over the C8 nerve roots at the C7-T1 joints. A portion of the C7 or T1 pedicle
was removed bilaterally to make room for the C8 nerve roots. If internal
fixation was to be used, it was applied at this point, but it was not secured.
After administration of a brief general anesthetic, the patient's head was
then slowly and carefully extended while the osteotomy site was observed
posteriorly. The osteotomy was completed anteriorly by means of osteoclasis.
Excessive force or sudden movement was avoided by using overhead traction and
an assistant to control the halo ring anteriorly. The dura generally buckled
slightly on extension of the osteotomy
(Fig. 1-C).
The osteotomy site was then secured with internal fixation; a wide variety
of fixation methods was used (Table
I). During the earlier years in which this series of patients was
treated, the only internal fixation available was wire. Many early procedures
were performed without internal fixation. As wiring became more common, it was
incorporated into the technique. When the Luque rectangle (Medtronic Sofamor
Danek, Minneapolis, Minnesota) became available, it was used for most of the
recent operations. Once the correction and fixation had been achieved, local
bone autograft generated by the laminectomies and foraminotomies was
morselized and placed over the osteotomy site prior to closure. Other sources
of bone graft material were generally not utilized.
The postoperative treatment included external immobilization with a halo
ring and a molded plaster jacket for three months. Removal of the plaster
jacket was followed by use of a rigid two-poster cervicothoracic orthosis for
an additional three months. Pin-site or cast-related problems resulted in
earlier conversion to the brace for four patients, but generally the orthosis
was not used before two months postoperatively. Plain or computerized
tomography was performed for all patients at three-month intervals to
determine the position of the osteotomy site and the fusion status until the
osteotomy site was observed to be fused radiographically.
Two patients did not have neck pain preoperatively and remained
pain-free at the time of final follow-up. Of the twenty-four patients with
preoperative neck pain, twenty-one had less pain at the time of final
follow-up, one who had had mild neck pain preoperatively had no change, one
died in the immediate postoperative period, and one with moderate neck pain
preoperatively had (unexplained) initially severe neck pain postoperatively,
which was moderate at the time of final follow-up
(Table II). Of nineteen
patients who had difficulty swallowing preoperatively, eighteen reported an
improved swallowing ability at the time of final follow-up and one, who had no
improvement, was ultimately diagnosed with esophageal carcinoma. Transient
dysphagia occurred in ten patients in the immediate postoperative period, and
it resolved completely by the six-week follow-up evaluation in all cases.
Horizontal gaze and the field of vision were not systematically measured, but
they were subjectively improved in all patients.
Of the sixteen patients who had been neurologically normal preoperatively,
one became quadriplegic, one had transient arm dysesthesias that quickly
resolved, and fourteen remained neurologically normal. Of the ten patients
with preoperative neurologic deficits, one still had hyperactive reflexes and
clonus at the time of follow-up but the remaining nine patients had
improvement of one, two, or three Nurick grades (see Appendix).
One patient died one week postoperatively. Of the twenty-five remaining
patients who were evaluated with plain or computerized tomography at three
months postoperatively, twenty-two were seen to have fusion at the osteotomy
site. Two patients were found to have fusion when they were re-evaluated at
six months, and one patient underwent repeat bone-grafting and fixation with a
single interspinous wire four months postoperatively and subsequently had a
successful fusion. Four patients had 5 mm of anterior subluxation at the
osteotomy site at the time of the initial follow-up, without adverse sequelae.
The anterior opening wedge correction (the osteoclasis) was at the C7-T1 level
in twenty-three patients, at the C6-C7 level in two, and at the C5-C6 level in
one. There were no adverse consequences related to the corrections that were
inadvertently performed cephalad to C7-T1.
The mean initial sagittal correction was 38°, with a range of 15°
to 84°, and there was a mean loss of correction of 2.6°, with a range
of 3° of improvement to a 20° loss of correction, at the time of final
follow-up. Of the five patients who lost =5° of the initial sagittal
correction, two had a delayed union and the other two had sustained additional
trauma to the cervical spine that had resulted in some loss of correction in
the interval before the final follow-up evaluation. In all remaining patients,
the correction seen at the time of final follow-up was within 5° of the
initial correction.
Neurologic Complications
A variety of neurologic complications occurred. The most devastating
occurred in a patient (Case 14, see Appendix) who was found to be quadriplegic
while he was still in the operating room immediately after the osteotomy. He
underwent urgent reexploration with repositioning of the obviously subluxated
osteotomy site and had almost complete neurologic recovery over the next three
days. However, he ultimately died of pneumonia and cardiac arrest while still
in the intensive care unit. Radiographs made at the time suggested that the
osteotomy site had resubluxated despite immobilization in a halo cast.
Five patients had new upper-extremity pain, dysesthesias, or weakness in
the immediate postoperative period. These symptoms completely resolved within
six months in three patients, whereas they decreased over a period of about
two years in the remaining two patients, who never regained a normal status.
One patient (Case 1, see Appendix) had a seizure in the immediate
postoperative period and had altered mental status with left-sided weakness
that persisted for five days; she recovered without any permanent neurologic
sequelae.
Most patients with ankylosing spondylitis do not have a loss or
reversal of cervical lordosis alone. Rather, the most severe kyphotic
deformity is found at the cervicothoracic junction and in the upper thoracic
spine. Thus, extension osteotomy should be carried out not in the cervical
spine per se, but at the cervicothoracic junction. Such intervention does not
change the thoracic kyphosis, but it does correct the sagittal alignment of
the cervical spine and the head in relation to the thoracic spine.
Neurologic abnormalities are frequent in patients with ankylosing
spondylitis who have a chin-on-chest deformity. In this series of twenty-six
patients, ten had neurologic deficits preoperatively, with findings ranging
from mild myelopathy in seven patients to severe myelopathy in three. Spinal
cord dysfunction may be due to the exaggerated cervicothoracic kyphosis
causing relative lengthening or distraction of the posterior spinal column and
draping of the spinal cord over the posterior vertebral bodies. Neurologic
signs and symptoms decreased with successful cervicothoracic extension
osteotomy in nine of the ten patients.
McMaster reported a mean sagittal correction of 54° (range, 30° to
71°) with an average loss of correction of 5.7° in a series of fifteen
patients treated with a cervicothoracic extension osteotomy for chin-on-chest
deformity secondary to ankylosing
spondylitis6. In our
series, the mean sagittal correction was 38° (range, 15° to 84°)
with a subsequent mean loss of correction of 2.6° at the time of final
follow-up. The excellent maintenance of correction in our patients may have
been a result of the increased use of instrumentation, aggressive external
immobilization, and achievement of a high fusion rate. The initial correction
of sagittal alignment was maintained over time, with the notable exceptions of
patients who had a postoperative fracture, delayed union, or subluxation.
Successful fusion of the osteotomy site was readily achieved in this series
of twenty-six patients; one patient died immediately postoperatively and the
remaining twenty-five had a solid fusion at six months, including one who had
a reoperation because of delayed union. Simmons reported two nonunions in
forty-two patients who had been treated with anterior
arthrodesis5.
McMaster reported nonunion in two of fifteen patients who had been treated
with anterior
arthrodesis6. All
forty-two of the patients in Simmons' series and twelve of the fifteen
patients in McMaster's series were treated without internal fixation. Delayed
union occurred in three patients in our series, two of whom had been treated
without internal fixation and one of whom had been treated with a single
interspinous wire for fixation. One of the patients treated without internal
fixation underwent subsequent repeat bone-grafting and interspinous wiring,
whereas the other two did not require additional surgery.
Anterior subluxation of =5 mm occurred in five of the twenty-six
patients in our series, with one patient sustaining a severe neurologic
injury. McMaster reported subluxation in four of the fifteen patients in his
series6; nonunion
developed in two of them and irritation of the eighth cervical nerve root,
which resolved, developed in the other two. All of the subluxations in the
current series and in McMaster's series occurred in patients who had been
treated without internal fixation. We therefore recommend the use of secure
internal fixation to prevent subluxation. A Luque rectangle with Drummond
wires (Zimmer, Warsaw, Indiana) appears to be sufficient. Other methods, such
as lateral mass or pedicle screw fixation, may be suitable, but generally
there is not enough room posteriorly to expose the second thoracic vertebra
with the plaster jacket in place. External immobilization with a halo is
advisable in the setting of rigid internal fixation because of the
biomechanical difficulty of immobilizing the stiff, long segments of the
ankylosed spine and the generally osteoporotic bone that is characteristic of
these patients.
Post-osteotomy complications related to irritation of the eighth cervical
nerve root in these patients are usually dysesthesias, which tend to resolve
uneventfully over time. However, when intrinsic hand-muscle weakness occurs it
can persist. Care must be taken during closure of the posterior osteotomy
site, and a wide foraminotomy and pediculectomy are essential. As emphasized
by Simmons, it is helpful to perform the osteotomy in an awake patient so
that, if necessary, adjustments can be made in the final position of the head
in the controlled surgical
environment5. A
disadvantage of rigid internal fixation in this setting may be that it
prevents such minor adjustments.
When considering cervicothoracic osteotomy for a given patient, the surgeon
must weigh the risks of neurologic injury against the functional improvement
and pain relief that can be achieved with successful correction of the spinal
deformity. The likelihood of relieving neck pain, improving altered neurologic
status, achieving fusion at the osteotomy site, improving horizontal gaze, and
improving swallowing ability is high.
Tables presenting details of the patients' history, radiographic
characteristics, neurologic function, and postoperative complications are
available with the electronic versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?
Mason C, Cozen L, Adelstein L. Surgical
correction of flexion deformity of the cervical spine. Calif
Med.1953;79:
244-6.79244
1953
[PubMed]
Urist MR. Osteotomy of the cervical
spine; report of a case of ankylosing rheumatoid spondylitis. J Bone
Joint Surg Am.1958;40:
833-43.40833
1958
Law WA. Osteotomy of the spine. J
Bone Joint Surg Am.1962;44:
1199-206.441199
1962
Simmons EH. The surgical correction of
flexion deformity of the cervical spine in ankylosing spondylitis. Clin
Orthop Relat Res.1972;86:
132-43.86132
1972
[CrossRef]
Simmons EH. Kyphotic deformity of the
spine in ankylosing spondylitis. Clin Orthop Relat Res.1977;128:
65-77.12865
1977
[PubMed]
McMaster MJ. Osteotomy of the cervical
spine in ankylosing spondylitis. J Bone Joint Surg Br.1997;79:
197-203.79197
1997
[PubMed][CrossRef]
Mehdian SM, Freeman BJ, Licina P.
Cervical osteotomy for ankylosing spondylitis: an innovative variation on an
existing technique. Eur Spine J.1999;8:
505-9.8505
1999
[PubMed][CrossRef]
Emery SE, Bohlman HH, Bolesta MJ, Jones
PK. Anterior cervical decompression and arthrodesis for the treatment of
cervical spondylotic myelopathy. Two to seventeen-year follow-up. J
Bone Joint Surg Am.1998;80:
941-51.80941
1998
[CrossRef]
Bohlman HH, Dabb B. Anterior and
posterior cervical osteotomy. In: Bradford DS, editor. Master
techniques in orthopaedic surgery: the spine. Philadelphia:
Lippincott-Raven; 1997.
1997