Keywords
lumbar vertebrae - spinal fusion - lordosis - pelvis
Introduction
Most degenerative lumbar diseases present good outcomes after nonsurgical treatments.[1] However, some groups of patients do not perceive such benefits from nonsurgical
treatments, therefore requiring surgical intervention.[2]
Several techniques can be applied to correct degenerative lumbar disorders, with decompressions,[3] arthrodesis,[4] and arthroplasties[5] being the most utilized approaches.
Among the arthrodesis group, we have lateral lumbar interbody fusion (LLIF), a minimally
invasive technique developed by Dr. Luiz Pimenta in the late 2000s apud Ozgur et al.,[6] that allows access to the lumbar discs through the psoas major muscle. Lateral lumbar
interbody fusion relies on its capacity to promote indirect decompression and on its
ability to maintain or correct sagittal parameters when needed, besides having a vast
literature regarding its clinical and radiological benefits.[7] However, arthrodesis alone is not enough. If, when performing the fusion technique,
the surgeon does not respect the spinopelvic parameters, even for degenerative conditions,
he might create a biomechanical disarrangement in the lumbar region that leads to
the overload of the discs and facet joints, culminating in the degeneration of adjacent
levels and even to reoperations.[8]
However, the LLIF technique is often regarded as one of the methods with the greater
capability of restoring sagittal lordosis.[9] Recent revisions of the literature show an extensive heterogeneity regarding the
capacity of LLIF to reestablish the lumbar lordosis, with authors pointing out significant
segmental lordosis gains[10] and other authors presenting small to no gain in some cases[11] Different cage properties and positions might explain some parts of this heterogeneity.[12]
However, the impact of preoperative spinopelvic parameters of patients in the gain
of segmental lordosis is poorly studied in the literature. Therefore, our study aimed
to assess whether preoperative spinopelvic parameters can increase segmental lordosis
after one-level LLIF.
Methods
This was a single-center, noncomparative, nonrandomized study approved by the ethics
commission (CAAE: 28761220.2.0000.8847). The study aimed to investigate the relationship
of some spinopelvic parameters and other radiological measurements with the improvement
of index level segmental lordosis in patients that underwent a LLIF procedure. All
patients included in the present study have given their free-consent to have their
data utilized in the study.
Inclusion Criteria
Patients who underwent LLIF surgery in our service had preoperative and postoperative
X-rays and signed a free and informed consent form.
Exclusion Criteria
Patients who received anterior column realignment or have the anterior longitudinal
ligament (ALL) unintentionally ruptured. Patients whose X-rays did not allow the correct
visualization and measurements of the proposed spinopelvic parameters.
Study Variables
The following radiological parameters were measured in the X-rays: pelvic incidence,
lumbar lordosis, pelvic tilt (PT), L4S1 lordosis, index level segmental lordosis,
intraoperative index segmental lordosis.
The following continuous variables were considered using the measured parameters:
pelvic mismatch (PI-LL), distal lordosis proportion (defined as the percentage of
L4S1 in the pelvic incidence), delta segmental lordosis (defined as the difference
between the standing X-ray preoperative index level segmental lordosis and the intraoperative
index level segmental lordosis). The following categorical variables were derived
from the measured parameters: PT > 20°, actual sacral slope (high, medium, or low),
and ideal sacral slope (ISS) (high, medium, or low). The ISS was defined by the following
operation ISS = (PT – 20) + actual sacral slope (SS), when patients had PT > 20°.
In cases of PT < 20°, the ISS is the actual SS. High, medium, and low SSs were defined
according to the Roussouly Columns Classification, with values < 35° being considered
low, > 45° being considered high, and values between these threshold values being
considered medium.[13]
[14] The 20° threshold for PT is derived from the SRS-Schwab classification.[15]
Study Outcomes
The primary goal of the present study was to evaluate the correlation between spinopelvic
parameters and the gain of segmental lordosis after one-level LLIF surgery.
The secondary goal of the present study was to assess how these spinopelvic parameters
could influence segmental lordosis after one-level LLIF surgery.
Statistical Analysis
The data were measured using Surgimap (Nemaris Inc., Toronto, Canada) software and
compiled using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) software,
and R software (R Foundation, Vienna, Austria) for the statistical analysis and graphs
elaboration. The following non-native packages were also used: ggpubr, ggplot2, tydeverse,
cluster, lsr.
For the analysis of the sample distribution, we applied the D'agostino normality test;
then, following this analysis, we investigated the correlation of the continuous variables
with the gain of segmental lordosis; we used the Pearson and the Spearman correlation
methods for normal and nonparametric distributed samples, respectively, and the Cramer
V test for correlation among categorical variables. Then, we applied the k-mean cluster
method to assign study patients to groups by similarity. To assess the ideal number
of clusters, the elbow method was chosen.
We used the T-test or the Wilcoxon Rank Sum Test to compare between groups, depending
on the sample distribution for continuous variables, and the chi-squared or the Fisher
exact test for discrete variables. When there were more than two groups, we performed
the Kruskal-Wallis test for nonparametric distributions and the Dunn Test method for
post-hoc comparison between groups. A p-value < 0.05 was rendered as a threshold for
statistical significance.
Results
Characteristics of the Study Population
A total of 104 patients were included in the present study, of which 75 (76%) presented
segmental lordosis gain. The mean segmental lordosis gain was 2.55° with L4L5 as the
index level in 84 (80%) surgeries. The frequency of the other studied variables is
shown in [Table 1].
Table 1
|
Continuous Variables
|
Frequency
|
|
Min
|
1st quarter
|
Median
|
Mean
|
3rd quarter
|
Max
|
Standard deviation
|
|
Pelvic incidence (°)
|
29.40
|
45.45
|
53.20
|
54.22
|
61.15
|
83.90
|
11.96
|
|
Lumbar lordosis (°)
|
9.30
|
39.17
|
49.90
|
49.17
|
61.15
|
87.00
|
15.98
|
|
Lordosis L4S1 (°)
|
3.10
|
22.50
|
27.95
|
29.15
|
35.23
|
59.20
|
11.70
|
|
Sacral slope (°)
|
14.00
|
26.30
|
33.35
|
33.77
|
41.27
|
70.60
|
11.14
|
|
Pelvic tilt (°)
|
0.20
|
12.57
|
20.20
|
20.22
|
26.73
|
44.90
|
10.35
|
|
Preoperative segmental lordosis (°)
|
−7.00
|
1.90
|
4.55
|
5.19
|
8.40
|
18.50
|
4.32
|
|
Intraoperative segmental lordosis (NA = 77) (°)
|
0.60
|
3.25
|
5.00
|
5.82
|
8.15
|
11.80
|
3.34
|
|
Delta Segmental Lordosis (NA = 77) (°)
|
−8.40
|
−3.75
|
−1.40
|
−0.02
|
3.40
|
8.80
|
4.71
|
|
Segmental lordosis gain (°)
|
−16.3
|
−0.12
|
2.10
|
2.42
|
6.20
|
12.5
|
5.35
|
|
Pelvic mismatch (°)
|
−26.90
|
−4.25
|
3.75
|
5.05
|
13.77
|
49.40
|
14.42
|
|
Distal lumbar lordosis proportion (%)
|
5.53
|
39.18
|
56.33
|
54.97
|
67.68
|
109.95
|
21.28
|
|
Categorical variables
|
Frequency
|
|
Pelvic tilt > 20°
|
yes: 39
|
no: 65
|
|
|
|
|
|
|
Segmental lordosis gain?
|
yes: 75
|
no: 29
|
|
|
|
|
|
|
Actual sacral slope (high, medium, low)
|
Low: 57
|
Medium: 33
|
High: 14
|
|
|
|
|
|
Ideal sacral slope (high, medium, low)
|
Low: 39
|
Medium: 38
|
High: 27
|
|
|
|
|
|
Operated levels
|
T12L1: 1
|
L1L2: 1
|
L2L3: 5
|
L3L4 : 13
|
L4L5: 84
|
|
|
|
Cage angulation
(NA = 34)
|
10: 69
|
12: 1
|
|
|
|
|
|
Correlations among Segmental Lordosis Gain and Studied Parameters
To assess the relationship between the parameters and the gain of segmental lordosis,
we performed a correlation analysis. The authors found that the parameters most correlated
with segmental lordosis gain were preoperative segmental lordosis (−0.50) and delta
intraoperative lordosis (0.51). The full correlation table is shown in [Table 2].
Table 2
|
Variables
|
p-value
|
Correlation
|
|
Delta segmental lordosis
|
0.002
|
0.536
|
|
Preoperative segmental lordosis
|
0.000
|
−0.500
|
|
Intraoperative segmental lordosis
|
0.046
|
−0.382
|
|
Pelvic tilt
|
0.001
|
0.304
|
|
Pelvic incidence
|
0.001
|
0.282
|
|
Pelvic mismatch
|
0.018
|
0.216
|
|
Distal lumbar lordosis proportion
|
0.049
|
−0.163
|
|
Sacral slope
|
0.486
|
0.052
|
|
Distal lumbar lordosis
|
0.648
|
−0.029
|
|
Lumbar lordosis
|
0.946
|
0.010
|
We also performed the Cramer V test to assess the association of categorical variables
with the gain or the absence of segmental lordosis. The ISS presented a medium correlation
with segmental lordosis gain ([Table 3]).
Table 3
|
Variable
|
Correlation
|
|
Ideal sacral slope
|
0.22
|
|
Pelvic tilt > 20
|
0.11
|
|
Actual sacral slope
|
0.06
|
Differences between Patients with and without Segmental Lordosis Gain
We present the differences between the groups in [Figure 1]. The patients who gained segmental lordosis presented lower preoperative segmental
lordosis, higher PI, higher PT, and higher pelvic mismatch. These patients also showed
an increase in the index level segmental lordosis when positioned for the surgery
(delta segmental lordosis).
Fig. 1 Boxplots representing differences in preoperative spinopelvic parameters between
patients that gained segmental lordosis, and those who lost segmental lordosis. *p < 0.05,
** p < 0.01, ***p < 0.001.
Identifying Patient Clusters
Afterwards, a clustering analysis was performed to identify if patients with similar
characteristics could be prone to experience a higher gain of lordosis than other
patients. Four clouds were created based on the elbow method ([Figure 2]). The clusters can be explained as two groups within two subgroups each.
Fig. 2 Elbow plot to estimate the optimal number of clusters. Vertical Line: Optimal number
of clusters.
The first group comprised patients with low-medium PI (47.24 °) and was subdivided
into 1 cluster (3) with misaligned patients (PI-LL = −20.63°; PT = 27.24°) and 1 cluster
with aligned patients (1) (PI-LL = −6.15°; PT = 11.54°) ([Table 4]). The second group comprised patients with high PI (65.68°), also divided into 1
group (4) with misaligned parameters (PI-LL = 13.06°; PT = 28.90°) and 1 group (2)
with more aligned parameters (PI-LL = 0.73°; PT = 19.62°) ([Table 4]).
Table 4
|
Cluster
|
PI
|
Lumbar lordosis
|
Distal lordosis (L4S1)
|
Sacral slope
|
Pelvic tilt
|
Preoperative segmental lordosis
|
Segmental lordosis gain
|
Pelvic mismatch
|
Distal lumbar lordosis proportion
|
|
1
|
47.59
|
53.75
|
31.99
|
35.48
|
11.54
|
6.99
|
0.26
|
−6.15
|
68.93
|
|
2
|
69.20
|
68.47
|
48.04
|
49.58
|
19.62
|
5.94
|
4.00
|
0.73
|
70.14
|
|
3
|
46.91
|
26.28
|
15.23
|
19.67
|
27.24
|
4.49
|
1.77
|
20.63
|
33.73
|
|
4
|
62.17
|
49.10
|
24.89
|
33.26
|
28.90
|
2.43
|
5.52
|
13.06
|
40.83
|
When comparing the segmental lordosis gain between the clusters, we could see that
cluster 3 presented significantly more gain than the other clusters, except cluster
1 in the post-hoc analysis (p < 0.05) ([Figure 3]).
Fig. 3 Violin-plot showing the segmental lordosis correction between the clusters. *p < 0.05,
** p < 0.01, ***p < 0.001.
A chi-squared test to assess the loss of segmental lordosis showed that patients in
different clusters had different risks of losing segmental lordosis after the LLIF
procedure (p = 0.02). To investigate further, the clusters were aggregated into medium-low PI
clusters (1 and 3) and high PI clusters (2 and 4), showing that patients in the low-medium
PI clusters are more prone to present loss of segmental lordosis with an odds ratio
(OR) of 6.08 (95% confidence interval [CI] = 4.93–7.23).
Discussion
The literature argues that LLIF is a reliable approach to correct lumbar spinal degenerative
conditions.[16] However, some recently published works demonstrated a high heterogeneity of segmental
lordosis capacity in this technique.[10]
Parameters Associated with Segmental Lordosis Gain
Cage Conformity
Several studies have tried to identify possible parameters that could impact the correction
of segmental lordosis. Sembrano et al. showed that cages with angulation (lordotic
cages) provided better segmental lordosis correction than nonlordotic cages.[17] On the other hand, a biomechanical study conducted by Gambhir et al.[18] found that when treating L3-L4, the key factor influencing the gain of segmental
lordosis was the cage height, mainly the posterior cage height, and that cages with
0° provided better angular correction than lordotic cages.
Cage Position
Moreover, some other studies showed that the cage position might play a critical role
in the final segmental lordosis. Park et al.[12] showed that a cage within the anterior third of the vertebral body was the best
position to restore segmental lordosis while not losing indirect decompression potential.
Similarly, Kepler et al.,[19] showed that placement of the cage in a more anterior position resulted in a mean
7.4° gain, while a more posterior position led to a 1.2° decrease (kyphotic effect)
in segmental lordosis gain. Otsuki et al.[20] also analyzed the cage position and its impacts on segmental lordosis correction.
Functional Spinal Unit Mobility
Another important parameter to be considered in the gain of segmental lordosis is
the mobility of the operating level. Two common ways to improve the mobility of a
spinal level are posterior osteotomies (facetectomies or Smith Petersen) or releasing
the anterior longitudinal ligament.[21] Although not related to a technical maneuver to increase mobility, our study also
found indications that functional spinal unit (FSU) mobility may play a significant
role in the gain of segmental lordosis, as patients with higher (positive) delta segmental
lordosis presented increased amounts of segmental lordosis gain (p < 0.05).
Segmental Lordosis and Spinopelvic Parameters
Other studies also pointed out that one key factor for estimating the amount of gain
of segmental lordosis is the preoperative segmental lordosis. In a systematic review
published by Uribe et al.,[22] the authors showed that the preoperative segmental lordosis was inversely associated
with the increase of segmental lordosis after the procedure. Similarly, our study
showed that preoperative segmental lordosis had a −0.50 correlation with segmental
lordosis gain. Also, patients presenting an increase in segmental lordosis had significantly
lower preoperative segmental lordosis (p < 0.001).
Our work also showed that patients who gained segmental lordosis had higher PI, PT,
and PI-LL values (p < 0.01; p < 0.05; and p < 0.05, respectively). Moreover, the present study showed that patients in the high
PI values cluster (> 60°) not only gained more lordosis but were also less likely
to experience loss of segmental lordosis ([Figure 3] and [Table 4]). Another interesting factor in our study was that the ISS had a medium to strong
correlation with the gain of segmental lordosis, corroborating the results that the
PI value might play a role in segmental lordosis correction.
However, the authors could not find any literature directly correlating these spinopelvic
parameters with segmental lordosis gain. Therefore, based on other studies showing
the effect of spinal parameters in the biomechanics of the spine, the authors hypothesize
the possible impacts of spinopelvic parameters on segmental lordosis correction. As
demonstrated by the literature, patients with high PI and SS (type 4) have hyperextended
lumbar morphology.[23]
[24] Moreover, Roussouly et al = ,[25] hypothesized that due to its more angulated conformity, type 4 Roussouly columns
might possess smaller posterior vertebral elements when compared with type 2 Roussouly
columns, which could make this type of patient more capable of posterior extension.
A recently published study reported that asymptomatic patients with type 4 Roussouly
columns had significantly larger intradiscal lordosis than patients with type 1 in
L5-S1 and that Type 4 patients also had more significant intradiscal lordosis than
Type 2 patients in L4-L5 and L2-L3.[26] For that, the authors think that it might be fair to speculate that due to the higher
amount of "biomechanical resources" that patients with high PI might recruit, this
group of patients can mitigate some of the impacts of both preoperative segmental
lordosis and positional/conformational aspects of the cage to accommodate the FSU
and avoid the loss of segmental lordosis.
The limitations of the present study include its retrospective design and the fact
that only the spinopelvic parameters were analyzed, not including in the work other
important factors such as cage position, cage angulation, and vertebral body angulation
and shape. Another limitation of our study is that the L4L5 level counted for more
than half of the included patients, impacting the generalizability our findings to
other spine levels. Finally, the last limitation is a philosophical one: the amount
of correction is achieved because of the spinopelvic parameters independent from the
surgeon or is it an effect achieved by the previous knowledge of the sagittal parameters
of the patient? This is a question that might be answered in the future by matching
patients with different column types or PI values and similar cage conformities and
positions.
Conclusion
Our work shows that surgeons might use the studied spinopelvic parameters to plan
their surgical options goals when performing one-level LLIF, mainly in patients with
low PI profile, because they presented higher odds of loss of segmental lordosis.
The index level of the segmental lordosis might also play a role in increasing or
not segmental lordosis after surgery.
However, the preoperative spinopelvic parameters do not seem to play a critical role
in the fate of segmental lordosis gain but act more as coadjutants in a complex set
of factors, as most of the parameters showed a moderate to weak correlation with segmental
lordosis gain.
