Keywords
spinal curvatures - spinal fractures - spinal injuries
Introduction
Thoracolumbar burst fracture (TBF) is characterized by comminution of the vertebral
body, thoracolumbar kyphosis, and bone fragment projected into the vertebral canal.[1]
Of all fractures involving the spine, 17% are TBFs.[2] This incidence is due to the transition anatomy between the rigid dorsal column
by rib summation and coronal orientation of the facets for the mobile lumbar spine
with sagittal facetal orientation.[3]
The objectives of TBF treatment are to provide stability, avoid deformity, and optimize
neurological recovery.[4] In cases with neural impairment, surgical intervention is mandatory, but in those
with normal neurological pattern, there is still doubt about the best treatment option.[5]
Advocates of surgical treatment state that the correction of thoracolumbar kyphosis
(TLC) generated by the fracture is fundamental to ensure better clinical and functional
results,[6] but the authors who recommend nonsurgical treatment suggest that it avoids surgical
complications, has lower cost, and, in the long run, exhibits similar clinical and
functional outcomes.[7]
The patient with TBF may evolve with gradual increase of TLC, creating deformities
in the sagittal plane with activation of compensation mechanisms in the thorax and
lower limbs in an attempt to remain balanced. Some authors have related severe deformities
in the sagittal plane with worse indices of function and quality of life.[8]
Most systematic reviews and meta-analyses on thoracolumbar fracture have linked the
increase in TLC with chronic pain, worsening function and quality of life;[5]
[6]
[7]
[9]
[10]
[11] however, few studies have analyzed global spinal parameters of patients with TBF.[12]
[13]
Our hypothesis is that the spinopelvic alignment of individuals who had TBF is altered
in relation to the asymptomatic population due to the increase in TLC.
The aim of the present study was to evaluate the spinopelvic alignment in patients
with TBF without neurological deficit treated nonsurgically and surgically in a reference
tertiary trauma hospital.
Materials and Methods
A cross-sectional study approved by the ethics committee of the institution (CAAE:
30745118.5.0000.5479) was conducted during routine consultations at the outpatient
clinic of the institution between March 2017 and March 2019. All patients who participated
signed a free and informed consent form.
The inclusion criteria were: patients with single-level TBF between T11 and L2 as
described by Denis,[14] age between 18 and 64 years old at the time of trauma, followed for at least 12
months after the beginning of treatment, without neurological deficit, of types A3
or A4 by the AOSpine classification. The exclusion criteria were the presence of vertebral
metastasis, presence of metabolic or endocrine disease, patients operated > 10 days
after the fracture, pathological fracture, firearm fracture, neurological deficit,
previous spinal surgery, and psychiatric disease.
Clinical data were collected on age, gender, fracture level, AOSpine classification,
trauma mechanism, type of treatment, and low back pain (visual analogue scale [VAS])
level, and questionnaires were applied on the Denis Pain Scale, the Denis Work Scale,
and the SF-36. Sagittal and spinopelvic radiographic parameters were measured. The
results were compared between the surgical and nonsurgical treatment groups, and,
in addition, the radiographic parameters of the groups were compared individually
with the control sample of the Brazilian population published by Pratali et al.,[15] which is considered normal.
The criteria for surgical indication of the group were TLC > 30°, loss of vertebral
body height of 50%, and spinal canal involvement > 50%.[16] The option for either short or long fixation or anterior route association and the
type of implant was at the discretion of the surgeon.
Lateral panoramic radiographs with a 36-inch chassis were taken and the patient was
placed in the standing position, with the upper limbs supported on a support, shoulders
resting at 30° flexion, and elbows slightly flexed.[17] Radiographic parameters were measured with the aid of a validated measurement tool,
Surgimap Spine Software (Surgimap, New York, NY, USA).
Applying the Cobb method, we measured the following sagittal parameters: thoracolumbar
kyphosis (TLC) thoracic kyphosis (TC), and lumbar lordosis (LL).[18] The spinopelvic parameters were sagittal vertical axis (SVA), pelvic incidence (PI),
pelvic version (PV), sacral inclination (SI), and discrepancy between pelvic incidence
and lumbar lordosis (PI-LL).[19]
Qualitative variables were described as number and percentage. Data normality was
investigated using the Shapiro-Wilk test. The quantitative variables were summarized
by means of the mean (standard deviation [SD]) when the data were normally distributed;
otherwise, median (P50%), first and third quartiles (P25% and P75%, respectively)
were used, in addition to the minimum and maximum values.
To evaluate the association of qualitative variables with the type of treatment, the
chi-square test or the Fisher exact test was used when one or more cells presented
an expected value lower than five. The comparison between means was performed using
the Student t-test; for the medians, the Mann-Whitney nonparametric test was used.
To compare the radiographic findings of the present study with the reference values
for the Brazilian population presented by Pratali et al.,[15] the Student t-test was used because the data from the present study were expressed
as mean and SD.
All statistical analyses were performed using STATA/SE 15.1 for Windows software (StataCorp,
College Station, TX, USA). A significance level of 5% was adopted, that is, p-values < 0.05
were considered statistically significant. All tests were bilateral.
Results
The present study had a total of 50 participants, of which 39 (78%) were men. The
mean age of the participants was ∼ 49.8 years old, ranging from 20 to 81 years old
(±15.2 years). The most frequent trauma mechanisms were fall from height with 31 cases
(62%), followed by traffic accidents with 18 (38%). L1 was the most affected level,
observed in 21 (42%) patients, followed by T12 with 14 (28%). The average follow-up
was of 109 months, with a minimum of 19 and a maximum of 306 months. According to
the AOSpine classification, 37 (74%) were defined as A3, and the other 13 (26%) as
A4. Surgical treatment was performed in 26 participants (52%), and 24 (48%) received
nonsurgical treatment. [Table 1] shows the comparison of demographic and clinical data between the two types of treatment.
Table 1
|
Treatment
|
|
Total
(n = 50)
|
Surgical
(n = 26)
|
Nonsurgical
(n = 24)
|
p-value
|
|
Age, mean (SD)
|
49.8 (15.2)
|
49.2 (15.4)
|
50.4 (15.2)
|
0.848[a]
|
|
Sex
|
|
Female
|
11 (22%)
|
6 (23.1%)
|
5 (20.8%)
|
0.800[b]
|
|
Male
|
39 (78%)
|
20 (76.9%)
|
19 (79.2%)
|
|
Trauma mechanism
|
|
Fall from Height
|
31 (62%)
|
17 (65.4%)
|
14 (58.3%)
|
0.608[b]
|
|
Motor Vehicle Accident
|
19 (38%)
|
9 (34.6%)
|
10 (41.7%)
|
|
Fracture level
|
|
L1
|
21 (42.0%)
|
10 (38.5%)
|
11 (45.8%)
|
0.273c
|
|
L2
|
11 (22.0%)
|
8 (30.8%)
|
3 (12.5%)
|
|
T11
|
4 (8.0%)
|
1 (3.8%)
|
3 (12.5%)
|
|
T12
|
14 (28.0%)
|
7 (26.9%)
|
7 (29.2%)
|
|
AOSpine rating
|
|
A3
|
37 (74.0%)
|
15 (57.7%)
|
22 (91.7%)
|
0.006[b]
|
|
A4
|
13 (26.0%)
|
11 (42.3%)
|
2 (8.3%)
|
According to [Table 1], it can be verified that there was no difference between the type of treatment in
the mean age, gender, mechanism of trauma, and fracture level (p > 0.05). There was a higher percentage of participants with A3 classification in
those submitted to nonsurgical treatment (91.7%; p = 0.006). In relation to the 26 participants who underwent surgical treatment, 9
(34.6%) underwent VPC, 13 participants (50.0%) underwent VPL, and the other 4 participants
(15.4%) underwent VP + VA. The posterior implants found were PP in 18 (69.2%) participants,
pedicular screws with upper hooks (hybrid system) in 4 (15.3%), and Cotrel-Doubosset
system in 4 (15.3%). [Fig. 1] shows the distribution of implants.
Fig. 1 Distribution of implants.
Regarding functional outcomes, according to [Table 2], it can be observed that there was a significant difference between treatments,
according to the Denis Work Scale (p = 0.046): the median presented by the participants who underwent surgical treatment
was higher than that of those who underwent nonsurgical treatment. There was no significant
difference between the treatments for lower pain VAS and Denis Pain Scale (p = 0.468 and p = 0.623, respectively).
Table 2
|
Treatment
|
|
Surgical
(n = 26)
|
Nonsurgical
(n = 24)
|
p-value
[#]
|
|
VAS low back pain
|
|
Median (P25–P75)
|
6 (3–7)
|
5 (3–7)
|
0.468
|
|
Minimum–maximum
|
0–9
|
0–8
|
|
Denis Pain
|
|
Median (P25–P75)
|
3 (2–4)
|
3 (2–3)
|
0.623
|
|
Minimum–maximum
|
1–5
|
1–4
|
|
Denis Work
|
|
Median (P25–P75)
|
5 (3–5)
|
3 (3–4.5)
|
0.046
|
|
Minimum–maximum
|
1–5
|
1–5
|
Regarding quality of life, [Table 3] shows that there was no significant difference between treatments in any of the
domains evaluated (p > 0.05).
Table 3
|
Treatment
|
|
Surgical
(n = 26)
|
Nonsurgical
(n = 24)
|
p-value
[#]
|
|
Functional capacity
|
|
Average
|
44.2
|
38.0
|
0.853
|
|
Median (P25–P75)
|
37.5 (15–75)
|
35 (20–55)
|
|
Minimum–maximum
|
0–100
|
5–100
|
|
Physical limitation
|
|
Average
|
27
|
31
|
0.323
|
|
Median (P25–P75)
|
0 (0–25)
|
0 (0–12.5)
|
|
Minimum–maximum
|
0–100
|
0–50
|
|
Pain
|
|
Average
|
50.3
|
50.7
|
0.830
|
|
Median (P25–P75)
|
45 (22.5–67.5)
|
43.5 (41–64.8)
|
|
Minimum–maximum
|
0–100
|
0–100
|
|
General state of health
|
|
Average
|
80
|
66.1
|
0.090
|
|
Median (P25–P75)
|
55 (30–75)
|
71 (48.5–82)
|
|
Minimum–maximum
|
25–95
|
25–97
|
|
Vitality
|
|
Average
|
63.6
|
58.4
|
0.552
|
|
Median (P25–P75)
|
55 (50–75)
|
55 (40–75)
|
|
Minimum–maximum
|
30–100
|
30–95
|
|
Social aspects
|
|
Average
|
65.4
|
59.2
|
0.357
|
|
Median (P25–P75)
|
62 (38–75)
|
56 (37.5–75)
|
|
Minimum–maximum
|
25–100
|
12–100
|
|
Emotional limitations
|
|
Average
|
25.5
|
21.5
|
0.872
|
|
Median (P25–P75)
|
0 (0–33)
|
0 (0–33)
|
|
Minimum–maximum
|
0–100
|
0–100
|
|
Mental health
|
|
Average
|
64.6
|
57.7
|
0.246
|
|
Median (P25–P75)
|
72 (52–80)
|
64.5 (48–78)
|
|
Minimum–maximum
|
32–100
|
24–96
|
[Table 4] indicates that there was no statistically significant difference between treatments
in any of the radiographic parameters evaluated (p > 0.05).
Table 4
|
Treatment
|
|
Surgical
(n = 26)
|
Nonsurgical
(n = 24)
|
p-value
[#]
|
|
Thoracolumbar cyphosis (°)
|
|
Median (P25–P75)
|
15.0 (11.8–22.9)
|
17.5 (12.2–28.2)
|
0.299
|
|
Minimum–maximum
|
1.9–31.0
|
0.1–39.5
|
|
Thoracic cyphosis (°)
|
|
Median (P25–P75)
|
41.6 (35.2–47.0)
|
38.3 (31.1–57.2)
|
0.861
|
|
Minimum–maximum
|
24.9–70.1
|
12.2–66.8
|
|
Lumbar lordosis (°)
|
|
Median (P25–P75)
|
52.4 (40.3–57.6)
|
47.4 (40.0–56.0)
|
0.587
|
|
Minimum–maximum
|
14.2–82.2
|
23.0–72.0
|
|
Sacral vertical axis (mm)
|
|
Median (P25–P75)
|
19.5 (8.9– 34.5)
|
15.5 (- 4.2– 27.8)
|
0.207
|
|
Minimum–maximum
|
- 12.0–101.1
|
- 36.0–104.2
|
|
Sacral slope (°)
|
|
Median (P25–P75)
|
31.1 (23.6–33.7)
|
31.7 (27.1–34.1)
|
0.771
|
|
Minimum–maximum
|
14.6–52.2
|
19.7–45.5
|
|
Pelvic version (°)
|
|
Median (P25–P75)
|
17.4 (8.8–23.1)
|
16.6 (13.6–24.0)
|
0.655
|
|
Minimum–maximum
|
0.2–47.7
|
0.6–34.4
|
|
Pelvic incidence (°)
|
|
Median (P25–P75)
|
50.1 (39.7–63.1)
|
47.1 (43.0–58.7)
|
0.946
|
|
Minimum–maximum
|
27.9–77.0
|
26.2–65.5
|
|
Discrepancy between lumbar lordosis and pelvic incidence (°)
|
|
Median (P25–P75)
|
10.1 (5.4–13.2)
|
10.5 (3.8–17.1)
|
0.946
|
|
Minimum–maximum
|
2.3–43.7
|
0.0–33.3
|
According to [Table 5], it can be verified that all sagittal and spinopelvic radiographic parameters had
a significant difference between the group submitted to nonsurgical treatment and
the sample of Pratali et al., except for PI (p = 0.674).
Table 5
|
Study
|
|
Nonsurgical treatment
(n = 26)
|
Pratali et al.15
(n = 130)
|
p-value
[#]
|
|
Lumbar lordosis (°)
|
|
Average (SD)
|
47.5 (12.4)
|
56.8 (8.0)
|
< 0.001
|
|
Sacral vertical axis (mm)
|
|
Average (SD)
|
14.2 (30.2)
|
- 5.4 (27.0)
|
< 0.001
|
|
Sacral slope (°)
|
|
Average (SD)
|
30.9 (5.9)
|
37.2 (6.7)
|
< 0.001
|
|
Pelvic version (°)
|
|
Average (SD)
|
18.6 (8.6)
|
12.4 (5.8)
|
< 0.001
|
|
Pelvic incidence (°)
|
|
Average (SD)
|
49.5 (10.2)
|
49.4 (8.2)
|
0.674
|
|
Discrepancy between lumbar lordosis and pelvic incidence (°)
|
|
Average (SD)
|
12.0 (10.1)
|
- 7.4 (7.7)
|
< 0.001
|
For lumbar lordosis, it was found that the participants of the study by Pratali et
al. presented, on average, 9.3 ± 2.0° more than the participants of the nonsurgical
group (p < 0.001). The participants of the present study (nonsurgical group) presented, on
average, 19.6 ± 6.1 mm more in the sacral vertical axis when compared with those in
the study by Pratali et al. (p < 0.001). For sacral inclination, it was observed that the participants of the study
by Pratali et al. presented, on average, 6.3 ± 1.5° higher when compared with the
participants of the nonsurgical group. The participants of the present study (nonsurgical
group) presented higher mean pelvic version and discrepancy between lumbar lordosis
and pelvic incidence when compared with the values presented by the study by Pratali
et al. (p < 0.001), and the mean differences were estimated at 6.2 ± 1.4° and 19.4 ± 1.8°,
respectively.
According to [Table 6], it can be observed that all sagittal and spinopelvic parameters showed a significant
difference between the group submitted to surgical treatment and the control sample
of the normal Brazilian population published by Pratali et al.,[15] except for pelvic incidence, which did not present statistical significance (p = 0.949).
Table 6
|
Study
|
|
Surgical treatment
(n = 26)
|
Pratali et al.15
(n = 130)
|
p-value
[#]
|
|
Lumbar lordosis (°)
|
|
Average (SD)
|
48.8 (16.2)
|
56.8 (8.0)
|
< 0.001
|
|
Sacral vertical axis (mm)
|
|
Average (SD)
|
23.5 (22.0)
|
- 5.4 (27.0)
|
< 0.001
|
|
Sacral slope (°)
|
|
Average (SD)
|
31.2 (9.0)
|
37.2 (6.7)
|
< 0.001
|
|
Pelvic version (°)
|
|
Average (SD)
|
18.1 (11.8)
|
12.4 (5.8)
|
< 0.001
|
|
Pelvic incidence (°)
|
|
Average (SD)
|
50.2 (13.7)
|
49.4 (8.2)
|
0.949
|
|
Discrepancy between lumbar lordosis and pelvic incidence (°)
|
|
Average (SD)
|
11.2 (8.6)
|
- 7.4 (7.7)
|
< 0.001
|
For lumbar lordosis, it was found that the participants of the study by Pratali et
al. presented, on average, 8.0 ± 2.1° more than the participants of the surgical group
(p < 0.001). The participants of the surgical group presented, on average, 28.9 ± 5.6 mm
more in the sacral vertical axis when compared with those in the study by Pratali
et al. (p < 0.001). For sacral inclination, it was observed that the participants of the study
by Pratali et al. presented, on average, 6.0 ± 1.5° degrees more when compared with
the participants of the surgical group. The participants of the surgical group presented
higher mean pelvic version and discrepancy between lumbar lordosis and pelvic incidence
when compared with the values presented by the study by Pratali et al. (p < 0.001), and the mean differences were estimated at 5.7 ± 1.5° and 18.8 ± 1.7°,
respectively.
Discussion
According to the studies surveyed,[2]
[6]
[9] we obtained a sample composed mostly of men, with a mean age of 49.8 years old,
who had fall from height as the most common mechanism of trauma. However, we presented
a mean follow-up in 109 months (minimum of 19 and maximum of 306 months), a follow-up
time that was considered long-term among the studies on TBF.
Our study showed no differences between low back pain measured by VAS for patients
treated surgically and nonsurgically. On the contrary, data from Shen et al.[20] suggested that the surgery resulted in better VAS scores compared with nonsurgical
treatment in the 1st month, although there were no differences between the groups 6 months after the injury.
Using the Denis Pain Scale, the median of the surgical group was 3 (2 to 4) and that
of the nonsurgical scale was 3 (2 to 3); there was no statistically significant difference
(p = 0.623) between the operated and nonoperated groups. In both groups, the highest
prevalence was of patients with moderate pain making occasional use of medication.
According to the Denis Work Scale, the median of the surgical group was significantly
worse (5 [3 to 5]; p = 0.046), with a higher prevalence of patients unable to work. The statistical difference
found in the Denis Work Scale in our study does not necessarily establish a better
result with the use of thoracolumbar orthosis, because the operated fractures tended
to be more severe by the surgical criterion of the group. In addition, the incidence
of type A4 lesions, with greater comminution, was significantly higher in the operated
group.
Regarding quality of life, we did not find any of the parameters of the SF-36 with
a significant difference between groups; however, they were worse when compared with
the sample of the Brazilian population considered normal in the study by Laguardia
et al.[21] We believe that TBF usually compromises previously healthy people who undergo trauma
followed by long treatment, many of whom present with outcomes such as chronic pain,
movement limitations, and post-traumatic deformities, a situation that can alter life
habits and generate disability for activities of daily living, worsening the quality
of life.
Regarding radiographic parameters, we did not observe significant differences in any
of the variables analyzed. Most studies on TBF without neurological deficit have related
the increase in TLC with outcomes such as chronic pain and worsening function and
quality of life.[5]
[6]
[7]
[9]
[10]
[11] However, the clinical relevance of the radiological outcome is a matter of debate,
and there is no study that demonstrates an unequivocal association between radiological
and clinical results after TBF of Denis, A3 and A4 AOSpine.
The meta-analysis by Rometsch et al.[11] found no difference in the results of disability or pain between surgical and nonsurgical
treatment, which are similar to those of our study. The radiographic analysis of the
review was not performed due to the wide variety of different TLC measurement techniques.
Based on the current knowledge on sagittal alignment of the spine, analysis only in
focal TLC is not adequate. Analyzing the whole spinopelvic alignment ensures more
faithful results of the consequence of the increase in TLC caused by traumatic injury.
However, there are few studies that performed a global analysis of the spinopelvic
parameters of patients with TBF.[12]
[13]
[22]
Koller et al.[12] retrospectively analyzed the spinopelvic parameters of 21 patients, including burst-type
fractures of the thoracolumbar and lumbar region without neurological injury (A3 AOSpine)
treated nonsurgically with a 9-year follow-up. They found a strong correlation between
age and worse closure regarding lumbar VAS in TBF. The elderly were well in the first
years after the injury, but then noticed an increase in pain at the fracture level;
we understand that this is due to loss of the potential for compensation of sagittal
alignment due to sarcopenia and degenerative changes of the spine.
Mayer et al.[22] conducted a retrospective study of 36 adults with exclusive TBF treated surgically.
The results demonstrated the interdependence between sagittal alignment and clinical
outcomes. They also support the assumption that stable restoration of thoracolumbar
alignment to normality and its maintenance has a positive impact on clinical outcome.
The statistical analysis, however, did not reach the significance level for the differences
between the groups treated posteriorly and by the posterior and anterior combined
route. The intergroup analysis revealed no significant differences in global sagittal
spinal and pelvic radiographic measurements or clinical outcome measurements.
Regarding sagittal alignment, we observed that the parameters were significantly worse
in these patients, both in those who were operated on and in those who were not, compared
with the control group of the asymptomatic Brazilian population.[15] One hypothesis is that these patients, although aligned globally, already present
some initial factor of compensation of the pelvic version due to residual kyphosis
of the thoracolombosacral transition. Possibly, as our sample was composed by young
patients (mean age: 49.8 years old; SD: ± 15.2), there was a greater compensation
capacity of the overall alignment as a function of the lumbar and pelvic stabilizing
muscles being more efficient than in the elderly. Future studies can be conducted
to analyze pelvic compensation capacity in elderly patients when physiological lordosis
(natural increase in SVA and PV) and sarcopenia occurs.
Currently, there is no consensus regarding the influence of muscles on sagittal misalignment.
Research in individuals without spinal deformities has shown an increase in fat infiltration
of almost 15% with aging.[23] Similarly, some authors have observed decreased volume and increased fat infiltration
of the spine ethers in patients with loss of lumbar lordosis.[24] In a prospective cohort, Ferrero et al.[25] related muscle quality with spinopelvic parameters. Sagittal misalignment was associated
with increased fatty infiltration and decreased muscle volumes with poor clinical
results.
In comparison with the literature, our study presents a homogeneous group of patients
(A3 and A4) and a minimum follow-up time of 19 months, which is considered long-term.
However, the design of the present cross-sectional study allows us to formulate only
hypotheses that can be confirmed or not with prospective studies and with a long-term
control group. Moreover, the patients evaluated were all who returned to the outpatient
clinic. We know that the loss to follow-up in the context of the Brazilian Unified
Health System (SUS, in the Portuguese acronym) is generally relevant. However, it
seems that these patients use compensatory mechanisms to maintain the aligned spine
globally and locally. We suggest conducting long-term prospective studies to improve
the level of evidence of this hypothesis.
Conclusions
The spinopelvic alignment in patients with TBF without neurological deficit treated
nonsurgically and surgically after a minimum follow-up of 19 months was normal; however,
the patients presented higher mean pelvic version and discrepancy between lumbar lordosis
and pelvic incidence when compared with the reference values of the Brazilian population.
