Patients
We reviewed the cases of a consecutive cohort of 262 adult patients with a
spinal deformity who had a long arthrodesis (four levels or more) performed by
one of two senior spine surgeons (K.H.B. and L.G.L.) from 1985 to 2002 at a
single institution. We excluded thirty of these patients because they did not
have a minimum two-year radiographic follow-up examination. Therefore, the
study group consisted of 232 patients (192 women and forty men) with spinal
deformity who underwent a long instrumentation and arthrodesis. There were 150
primary and eighty-two revision procedures. Retrospective collection and
analysis of the data on patients with a minimum follow-up of two years
(average, five years; range, two to 16.8 years) were performed by an
independent spinal surgeon reviewer who did not participate in the operations.
The study was conducted after approval was obtained from the Human Studies
Committee of the institutional review board. Inclusion criteria consisted of
an age of greater than eighteen years at the time of surgery and deformities
treated with instrumented segmental spinal arthrodesis of four vertebrae or
more.
Spinal arthrodesis was indicated in each patient because of a combination
of factors: a major deformity (scoliosis of >40°, thoracic kyphosis of
>75°, coronal C7 plumb of >4 cm off the midline of the sacrum, and
sagittal C7 plumb of >5 cm forward of the posterior-superior S1 end plate),
documented progression of the deformity, pseudarthrosis, back pain, radicular
pain, and/or symptoms of spinal stenosis associated with the spinal deformity.
The average age of the patients at the time of surgery was 40.8 years (range,
18.1 to 77.3 years). The preoperative diagnosis was adult scoliosis in 131
patients, sagittal imbalance syndrome in ninety patients, and Scheuermann
kyphosis in eleven patients. Eighty-two of the 232 patients had prior
surgeries and were managed with a reoperation for either pseudarthrosis,
progressive junctional deformities, or coronal or sagittal imbalance.
Surgical Decision Making
Anterior and posterior combined arthrodesis was performed on patients who
were determined to be at high risk for pseudarthrosis, namely, those with a
long arthrodesis to the sacrum, large and stiff coronal and/or sagittal
deformities, and areas of pseudarthrosis from prior surgery with associated
kyphosis. The indications for preliminary anterior release and arthrodesis
(stage 1) included stiff coronal curves that did not correct to <50° on
bending or flexibility radiographs, or a rigid kyphosis defined as 20° of
kyphosis (relative to the normal sagittal Cobb angle for those regional
segments) persisting on a radiograph made with the patient supine and the
spine in hyperextension. An evaluation of the characteristics of the
lumbosacral fractional curve (L3 to the sacrum), such as the presence of
rotatory subluxations (>3 mm), and the presence of spinal stenosis
(central, in the lateral recess, or within the foramen), as determined on the
basis of magnetic resonance imaging scans and computed tomography myelograms,
was made to determine whether the arthrodesis should extend to L3, L4, L5, or
the sacrum. If the arthrodesis was extended to the sacrum, we augmented the
posterior instrumentation construct with "complete" sacropelvic
fixation, which consisted of bilateral iliac screw fixation and anterior
column support at L5-S1 in addition to the bilateral S1 screws (fifty-two of
eighty-one patients had an arthrodesis to S1). However, early in the series,
we did not always use complete sacropelvic fixation (twenty-nine of eighty-one
patients).
Operative Procedure
Anterior operations included thoracoabdominal and paramedian approaches.
For twelve patients with lumbar or thoracolumbar curves, anterior arthrodesis
with anterior segmental spinal instrumentation was placed exclusively.
Posterior instrumentation was always bilateral. All implants were stainless
steel, either Cotrel-Dubousset instrumentation or Cotrel-Dubousset Horizon
instrumentation (Medtronic Sofamor Danek). The constructs averaged 1.5
fixation points per level. Autogenous iliac-crest bone graft, in addition to
local bone harvested from the spinous processes and/or facet joints, was
routinely used for the posterior surgery.
Characteristics of Pseudarthrosis
The criteria used to define pseudarthroses were (1) loss of fixation, such
as implant breakage, dislodgment of rods or hooks, or a lucent halo around a
pedicle screw; (2) progression of the deformity clinically or
radiographically; (3) subsequent disc-space collapse at the distally
instrumented motion segment compared with the findings at the first
postoperative visit; and (4) motion during surgical exploration. One senior
spine surgeon (Y.J.K.), who was independent of the operative team, performed
the analysis.
Risk Factors for Pseudarthrosis
The effect of patient age, gender, and comorbidities (osteoporosis, alcohol
abuse, cigarette smoking, and cardiovascular, endocrine, neurologic,
gastrointestinal, or psychiatric disease) at the time of surgery were
evaluated. Revision surgery (compared with primary surgery) and the
performance of an osteotomy and laminectomy during the surgical procedure were
evaluated as risk factors. The number of vertebrae in the arthrodesis was also
assessed. A preexisting coronal Cobb angle of 70°, a preexisting sagittal
thoracic kyphosis angle (T5-T12) of >40°, and a thoracolumbar kyphosis
angle (T10-L2) of 20° as well as preexisting coronal (C7 plumb to the
center sacral vertical line of >20 mm) and sagittal imbalances (C7 plumb to
the posterior superior sacral end plate of >50 mm) were evaluated as risk
factors.
Patient Outcomes
The Scoliosis Research Society patient questionnaire (SRS-24) was used to
evaluate outcomes at the final follow-up examination. Two recent studies have
validated the use of this outcomes tool for adults with spinal
deformity10,11.
Completed questionnaires were available for 196 (84%) of the 232 patients.
Thirty-three (83%) of the forty patients with pseudarthrosis completed the
SRS-24 questionnaire at the final follow-up evaluation.
Statistical Analysis
The data were analyzed with use of SPSS software (version 10.0; SPSS,
Chicago, Illinois), and statistical analysis was performed. Dichotomous
variables were compared with use of the Fisher exact and chi-square tests. P
values were based on the Student t test for independent variables. The level
of significance was set at p < 0.05.
Characteristics of Pseudarthrosis
(Table I)
Pseudarthrosis was identified in forty (17%) of the 232 patients. The group
included thirty-three of the 192 women and seven of the forty men in the
study. The average age of the patients with pseudarthrosis at the time of
surgery was 46.9 years (range, 18.2 to 70.0 years). Seventeen patients were
greater than fifty-five years old. The patients with pseudarthrosis had an
average of thirteen vertebrae (range, six to seventeen vertebrae) included in
the arthrodesis; twenty-five patients had arthrodesis of thirteen to seventeen
vertebrae. Twenty of the forty patients had multiple levels of pseudarthrosis
(two to six segments). Sixteen (16%) of the 102 patients with a posterior
spinal arthrodesis alone had a pseudarthrosis. Two of the twelve patients who
underwent anterior spinal arthrodesis alone had a documented pseudarthrosis.
Twenty-two (19%) of the 118 patients who underwent combined anterior and
posterior arthrodesis had a documented pseudarthrosis. Six of the twenty-two
patients had a pseudarthrosis in the posterior construct at the level cephalad
to the anterior arthrodesis, and the other sixteen patients had pseudarthrosis
at levels where both anterior and posterior arthrodesis was attempted.
Pseudarthrosis occurred most commonly between T10 and L2, which was the
site of 58% (twenty-three) of the forty pseudarthroses, and at L5-S1, which
was the site of 25% (ten pseudarthroses). The eighty-one patients who had
arthrodesis to S1 had an average of 13.2 vertebrae involved, with a range of
seven (T12 to S1) to seventeen vertebrae (T2 to S1). Fifty-two of the
eighty-one patients who had arthrodesis to S1 had socalled complete
sacropelvic fixation (bilateral S1 screws, bilateral iliac screws, and
anterior column support at L5-S1). Ten of the eighty-one patients who had
arthrodesis to the sacrum had pseudarthrosis occur at the L5-S1 level. Of
these ten patients, five had complete sacropelvic fixation, resulting in a
9.6% rate of pseudarthrosis in the fifty-two patients who had been managed
with complete sacropelvic fixation. Five (17%) of the twenty-nine patients who
had been managed with so-called incomplete sacropelvic fixation had
development of a pseudarthrosis. Two had no anterior column support at L5-S1,
two had only unilateral iliac fixation, and one did not have either anterior
column support or bilateral sacropelvic fixation. With the numbers available,
no significant difference in the pseudarthrosis rate was found, with use of
the chi-square test, between the patients managed with complete sacropelvic
fixation and those managed with incomplete sacropelvic fixation (p =
0.32).
Pseudarthroses were detected radiographically at an average of forty-two
months (range, twelve to 131 months) after the operation; detection occurred
at least three years postoperatively in seventeen patients and at least five
years postoperatively in nine patients. Revision was performed at an average
of eighteen months (range, one to forty-six months) after radiographic
detection. The most common radiographic findings with pseudarthrosis were rod
breakage, which occurred in twenty-eight (70%) of the forty patients with
pseudarthrosis (Fig. 1);
progression of deformity, in twenty-four (60%); disc space collapse, in
sixteen (40%); hook pull-off, in five (13%); and a halo lucency around screws,
in five (13%). Pseudarthrosis was recognized in seventeen (11%) of the 151
patients who had arthrodesis to L5 or a cephalad level and in twenty-three
(28%) of eighty-one patients who had arthrodesis to S1. A significant
difference in the rate of nonunion was detected, with the chi-square test,
between patients in whom the lowest instrumented vertebra was at L5 or a
cephalad level and those in whom the lowest instrumented vertebra was S1 (p =
0.002; Table II).
Radiographic Results
The radiographic measurements of the various deformities made
preoperatively, immediately postoperatively, and at the final follow-up
examination are available in the Appendix. The measurements include the
relevant coronal and sagittal Cobb angles and the coronal and sagittal balance
parameters relative to the C7 plumb to the sacrum.
Risk Factors (Table
II)
Seventeen (35%) of the forty-nine patients who were more than fifty-five
years old and twenty-three (11%) of the 213 patients who were between eighteen
and fifty-five years old demonstrated a pseudarthrosis. An age of fifty-five
years or more at the time of surgery was associated with a significantly
higher rate of pseudarthrosis (p = 0.001). The rate of pseudarthrosis in men
was similar to that in women. Patients who had arthrodesis of more than twelve
vertebrae had a significantly higher rate of nonunion than did those who had
arthrodesis of twelve or fewer vertebrae (p = 0.037). Smoking history was not
found to significantly increase the nonunion rate, with the numbers available
(p = 0.06). Pseudarthrosis was not found to be significantly associated with
comorbidities, with the numbers available (p = 0.25). Fifteen (18%) of the
eighty-two patients who had a revision and twenty-five (17%) of the 150
patients who had primary surgery had development of a pseudarthrosis. The
nonunion rate associated with revision surgery was not found to be increased
compared with that associated with primary surgery, with the numbers available
(p = 0.91). The nonunion rate for patients who had an osteotomy was not found
to be increased compared with that for patients who had not had an osteotomy,
with the numbers available (p = 1.00). Although there was no apparent
relationship between laminectomy site and pseudarthrosis area, patients who
had laminectomy of the posterior elements had an increased nonunion rate
compared with those who had not had posterior decompression (p = 0.003). A
higher preoperative major Cobb angle (70°) was not found to be associated
with a significantly higher nonunion rate, with the numbers available (p =
0.779). Preoperative thoracolumbar kyphosis (T10-L2 of 20°) was associated
with a significantly higher nonunion rate (p < 0.0001). Preoperative
thoracic hyperkyphosis (T5-T12 of 40°) was not found to be associated with
a higher nonunion rate, with the numbers available (p = 0.130). Preoperative
coronal global imbalance (C7 plumb to center sacral vertical line of >20
mm) did not appear to increase the nonunion rate (p = 0.85). Preoperative
sagittal global imbalance (C7 plumb to sagittal sacral vertical line of >50
mm) was not found to be associated with an increased nonunion rate, with the
numbers available (p = 0.74).
Patient Outcomes (Table
III)
SRS-24 functional outcome
data9 were obtained
for patients with union at the final follow-up examination and for patients
who had a nonunion but had not yet had subsequent revision surgery to treat
it. Less than one-half of the patients had SRS-24 outcomes data
preoperatively, as the instrument did not exist at that time; therefore,
preoperative SRS-24 data are not reported. The total score has a maximum of
120 points and includes the postoperative score (the final nine questions),
which has a maximum of 45 points; pain, 25 points; self-image, 25 points;
function, 25 points; and satisfaction, 10 points. The last nine questions
include the two satisfaction questions, so the satisfaction score (10 points)
is also included in the score for the final nine questions (45 points). A
higher score is a better result, and a lower score implies more abnormality.
Patients with a nonunion had significantly lower SRS-24 scores (average, 79
points) compared with patients with union (average, 93 points; p < 0.001).
The total score on the last nine questions (p = 0.005) and the pain (p =
0.003), self-image (p = 0.002), and function (p < 0.001) subscales for the
patients with a nonunion were significantly lower than the scores for the
patients with union. The score on the satisfaction subscale (two questions)
for patients with a nonunion was not significantly different (p = 0.09) from
that for the patients with union, with the numbers available.
The prevalence of pseudarthrosis following instrumentation and arthrodesis
for the treatment of adult spinal deformity in our series was 17% (forty of
232 patients). A pseudarthrosis developed in sixteen (6%) of 102 patients who
had posterior arthrodesis alone, two of twelve who had anterior arthrodesis
alone, and twenty-two (19%) of 118 who had combined anterior and posterior
arthrodesis. These findings demonstrate that pseudarthrosis occurred in adults
with spinal deformity even though circumferential arthrodesis had been
performed.
Only twenty-three (58%) of the forty patients with pseudarthrosis had
instrumentation failure during the first three years after surgery. Nine of
the forty patients had instrumentation failure more than five years after
surgery. Therefore, the definitive determination of osseous union may require
as many as seven years of follow-up. If all of our patients were followed for
ten years, the pseudarthrosis rate would likely be higher than the 17%
determined in this analysis.
The thoracolumbar junction had the highest nonunion rate (58% of the forty
patients with pseudarthrosis) compared with other anatomical regions of the
spine. This finding suggests that particular attention should be given to
arthrodesis bed preparation, and the application of a larger quantity of
autogenous bone graft at the thoracolumbar area might be appropriate.
Factors such as older age (greater than fifty-five years) and the number of
vertebrae in the arthrodesis, which was perhaps related to the use of a
smaller amount of iliac-crest bone graft at each level, had notable effects on
the pseudarthrosis rate in this study. This result is consistent with that in
the report by Martin et
al.12, who found
that implanting only 50% of the normal volume of autogenous bone graft lowered
the fusion rate from 70% to 33% in a rabbit model. In our study, preoperative
anatomic factors, such as a larger coronal Cobb angle, thoracic kyphosis, and
preoperative coronal and sagittal imbalance, were not found to have a
significant effect on pseudarthrosis rate, with the numbers available. In
contrast, the presence of a thoracolumbar kyphosis (T10 to L2 of 20°) had
a significant effect on the rate of pseudarthrosis (p = 0.0001). An
association was found between laminectomy and pseudarthrosis, but not
pseudarthrosis at the laminectomy levels, which is not easily explainable.
Forty patients had pseudarthrosis. Twenty-three (28%) of the eighty-one
patients who had been managed with an arthrodesis to S1 had pseudarthrosis
compared with seventeen (11%) of 151 patients who had been managed with
arthrodesis to L5 or to a cephalad level (p = 0.002). Arthrodesis to S1 was
associated with a high rate of nonunion at the L5-S1 level (43% [ten] of the
twenty-three patients with pseudarthrosis who had had arthrodesis to S1 had
nonunion). We believe that thoracolumbar kyphosis and arthrodesis to S1 in
addition to older age (greater than fifty-five years) and a large number of
vertebrae in the arthrodesis (greater than twelve) should be considered risk
factors for pseudarthrosis in patients with adult spinal deformity who are
undergoing instrumentation and arthrodesis.
To reduce the rate of nonunion at L5-S1 when performing an arthrodesis to
the sacrum, we currently use several methods. Although complete sacropelvic
fixation (bilateral S1 screws, bilateral iliac screws, and anterior column
support at L5-S1) did not have a significant effect on union compared with
incomplete fixation, we strongly recommend complete sacropelvic fixation. We
also place a larger quantity of autogenous bone graft posteriorly at this
level. Currently, our common practice is to do anterior surgery on only the
distal segments and to approach them through a straight anterior
retroperitoneal approach. This allows for a more complete discectomy at L5-S1
and better placement of a load-sharing cage at that level.
Studies on the surgical treatment of degenerative
spondylolisthesis13,14
have demonstrated a frequent lack of association between a solid fusion and
clinical outcome. Furthermore, there has been a lack of consensus as to the
appropriate indications for arthrodesis in degenerative spondylolisthesis.
Other
studies15,16
have shown an association between a higher rate of union and a higher rate of
better clinical outcomes. Our study demonstrated significantly lower total
outcome scores on the SRS-24 (p = 0.001 to 0.005) in all subscales, except the
satisfaction score (p = 0.09), among the patients with pseudarthrosis. One
limitation of our study is that we did not have the opportunity to obtain SRS
outcome scores on many patients preoperatively. Therefore, it is possible that
the patients with a pseudarthrosis may have had lower scores before the
surgery than the patients with a solid union, but this does not seem
likely.
Finally, although a significant association between pseudarthrosis and
smoking was not demonstrated, with the numbers available, we found a strong
trend (32% of smokers compared with 15% of nonsmokers had a pseudarthrosis).
It is our policy to request that any patient with spinal deformity quit
smoking prior to surgery.
A table showing the radiographic measurements of the study patients is
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). ?