Keywords
pedicle screws - 3D simulation study - minimally invasive - facet joint violation
- robotic
Introduction
Lumbar pedicle screw placement can be challenging due to the geometric complexity
of the pedicles and the proximity of the facet joints. Percutaneous instrumentations
can be especially demanding due to the limited visibility of anatomical landmarks
and a restricted working space.[1] Pedicle screws require good accuracy to achieve satisfactory placement without violation
of pedicle walls but also to maintain the integrity of the superior facet joints that
will not be included in the fusion.
However, violation of the superior facet joints is a frequent problem that could lead
to persisting symptoms and adjacent facet degeneration associated with higher reoperation
rates and diminished improvement in quality of life.[2] Rates reported range from 4 to 100% and are especially high in minimally invasive
percutaneous procedures guided by two-dimensional (2D) fluoroscopy.[3]
[4] In a previous patient study we compared three different pedicle screw insertion
2D techniques with regard to their frequency of facet joint violations (FJVs) and
revealed a significant lack of accuracy with the conventional minimal invasive percutaneous
procedure.[5] The comparison of a percutaneous robotic, percutaneous fluoroscopic, or standard
open technique for pedicle screw implantation revealed rates of 5%, 22%, and 6% for
upper FJVs, respectively.
Computer-assisted navigation technologies were developed for pedicle screw placement
to improve implantation accuracy.[6] Systems for conventional or robotic image guidance generally provide software tools
for three-dimensional (3D) planning of implant positions and some form of intraoperative
guidance (image based or physical by holding drill guides, etc.). Clinical studies
on these techniques demonstrated favorable radiologic results compared with solely
freehand techniques.[6]
[7]
[8]
In a clinical evaluation using a robotic navigation system, we recently showed that
the application of image guidance and 3D planning could help the surgeon perform pedicle
screw placement sparing upper facet joints.[5]
[9] However, the question remains whether the effect can be attributed rather to the
intraoperative guidance than to improved visibility of anatomical structures during
3D planning of implant positions. To date, no studies have directly and quantitatively
compared 2D and 3D planning as applied during conventional percutaneous or navigated
percutaneous pedicle screw placement.
This experimental study was conducted to investigate whether the mode of planning
(using 2D or 3D imaging tools) alone could have a significant effect on implant positions,
and whether 3D visualization during planning might help spare the adjacent upper facet
joint in pedicle screw placement. Differences between the two techniques in the location
of the entry point and trajectory to the pedicle axis were assessed.
Methods
Experimental Setup
We performed a simulated instrumentation on a software reconstruction. Informed consent
and ethical approval were waived by the corresponding ethical board, based on the
nonclinical nature of this study.
Imaging data of 250 consecutive patients who underwent lumbar fixation for degenerative
spine disease at our institution between 2012 and 2016, and who met the inclusion
criteria were included in the study. [Fig. 1] depicts the inclusion criteria. Patients with infection, tumor, revision surgery,
congenital deformity, and/or trauma or those who did not succeed in registration of
the navigation software due to technical problems (e.g., unsatisfactory imaging of
the lumbar levels involved, metal artifacts from prior surgeries causing noise in
3D reconstruction images) were excluded.
Fig. 1 Trial flowchart.
The patients' imaging data sets (1 mm thickness axial computed tomography [CT] scans
from L2–S1, Aquilion RXL multislice scanner, Toshiba, Zoetermeer, Netherlands) were
processed with Renaissance planning software (Mazor Robotics, Caesarea, Israel). The
software provided 2D (anteroposterior [AP] and lateral translucent 2D reconstructions
of the CT images resembling fluoroscopy images) and 3D visualization of the CT (CT
cuts in axial, sagittal, and coronal reconstructions) of the lumbar spine. First,
all data sets underwent conventional pedicle screw placement using 2D fluoroscopic
images in AP and lateral views (2D group). We subsequently performed screw placement
using the 3D reconstruction and planning mode of the same spinal navigation-planning
software (3D group). From the axial source images, sagittal and coronal images were
reconstructed. All levels between L1and L5 were virtually instrumented in each data
set. All simulated procedures were performed by the same surgeon, experienced in both
techniques. After completion of the simulation, all data sets were subsequently assessed
by two surgeons blinded to the applied technique.
Planning of Pedicle Screw and Assessment of Entry Point and Trajectory
At first, the planning surgeon positioned the screws using AP and lateral radiographic
images only (2D group) as used in fluoroscopy-guided instrumentation. After that,
the surgeon simulated the pedicle screw using the complete 3D planning tools of the
software (3D group) as applied in navigation and robot-guided procedures. For both
techniques, the surgeon's aim was to position the screw with a 1-mm safety margin
to all pedicle walls and respecting the superior articular process and articular facet
with a safety margin of 2 mm to the screw shaft and screwhead. The priority was focused
on respecting the pedicle walls and placing pedicle screws as accurately as possible.
In the 2D group, pedicle screws were planned to be inserted according to the method
described by Weinstein et al using an entry point at the “nape of the neck” of the
superior articular process and an inward trajectory.[10] On the AP radiograph of the lumbar spine, this point corresponds to the lateral
ridge of the pedicle projection ([Fig. 2]). From this entry point the fictive pedicle probe is placed in the center of the
pedicle, and the trajectory is controlled in the lateral view to confirm that the
fictive probe reaches the pedicle-vertebral body junction. Keeping this sagittal angulation,
the definite screw position is planned with the distal end of the trajectory placed
in the anterior vertebral body and the middle part seen within the boundaries of the
pedicle on the AP view but lateral to the medial pedicle wall.
Fig. 2 Virtual lumbar pedicle screw placement using the anteroposterior radiograph for the
2D technique und the three dimensionally reconstructed model for the 3D technique.
In the 3D group, screw positions were reviewed and optimized in all three planes (axial,
sagittal, and coronal) ([Fig. 3]).
Fig. 3 Measurements of the radiologic parameters of the two planning techniques presented
in a single case: (a) vertical offset, (b) horizontal offset of the 3D entry point from the 2D-placed pedicle screw, and (c) transverse angle on the reconstructed axial image with greater angulation angle
in the 3D-placed trajectory.
Finally, differences of the screw entry points were quantified. The deviation of the
3D entry point was measured from the 2D entry point as lateral and craniocaudal deviation
in millimeters ([Fig. 2]). The convergence angle was also compared between the two groups.
Evaluation of Pedicle Screws with Respect to Facet Joints
The 3D planning software was also used to evaluate resulting FJV from L1–L2 to L4–L5.
This was performed by two surgeons independently who were blinded to the applied planning
mode. The simulated FJV included not only the trajectory but also the screw head/tulip
position and the screw diameter of 6 mm.
Abutment of the polyaxial screw head on the dorsal surface of the facet joint and
interposition of screw threads between the superior and inferior articular processes
of the facet joint were classified as an FJV.
Additionally, facet joint osteoarthritis was graded on a 4-point scale according to
Pathria.[11]
[12] We then analyzed whether the degree of structural facet disease represented a risk
factor for upper FJV.
Evaluation of Pedicle Screws in Respect to Pedicle Walls
Screw placement was considered correct if the screw was completely surrounded by the
pedicular cortex. An incorrect screw position was categorized as cortical encroachment
if the pedicle cortex could not be visualized or as frank penetration when the screw
was outside the pedicular boundaries.
Evaluation of Time Required for the Planning
Apart from accuracy data, the time required for planning of the procedures was recorded.
Statistical Analysis
All data were analyzed using SPSS v.22.0 (IBM Inc., Armonk, New York, United States).
For each technique, we evaluated the overall frequency of FJV and the relative risk
of violation caused by facet joint osteoarthritis. Then we performed two comparative
analyses. First, we compared the overall frequency of FJVs by 2D- versus 3D-planned
screws in the entire population. Second, the same comparison was performed for the
subset of patients with severe facet joint osteoarthritis (Pathria grades 3 and 4).
The frequency of facet joint osteoarthritis was compared using the Fisher exact test
(for independent proportions) or the McNemar test (for paired proportions). The level
of significance was set at p < 0.05 for all statistical analyses.
Results
CT scans of 379 consecutive patients were used in this study. Overall, 250 patients
fulfilled the selection criteria; their data sets were included. We performed virtual
instrumentation in each data set, so a total number of 500 adjacent cranial facet
joints and pedicle screws could be assessed. The trial flowchart in [Fig. 1] offers an overview of the steps performed. Imaging data originated from 153 women
and 97 men with an average age of 57 years (range: 35–81 years). There were 330 facet
joints with an osteoarthritis grade 1/2 und 170 with grade 3/4 among the 500 assessed
screw insertion sites.
Comparison of 2D- versus 3D-Planned Percutaneous Pedicle Screw Insertion for Upper
Facet Joint Violation Rates
Screw trajectories were evaluated bilaterally, for a total of 500 possible screw insertion
sites for each type of technique. Overall, 2D-planned screws violated upper facet
joints in 28% (140/500 screws). Among the 170 facet joints with an osteoarthritis
grade 3/4, the 2D-planned screws violated the facet joint in 110 (93.5%). Among the
330 possible screw insertion sites with low-grade osteoarthritis grade 1/2, the 2D-planned
screws violated the upper facet joint in only 30 (9.1%). Therefore, the frequency
of FJVs by 2D-planned screws was significantly higher in high-grade facet joint osteoarthritis
than in low-grade osteoarthritis (p < 0.001), with a relative risk of 10.27.
The 3D-planned screws violated the upper facet joints in 4.8% of cases (24/500 screws).
Among the 170 possible screw insertion sites with facet joint osteoarthritis 3/4,
the 3D-planned screws violated the facet joint in 19 (11.1%). Among the 330 possible
screw insertion sites with low-grade osteoarthritis grade 1/2, the 3D-planned screws
violated the upper facet joint in only 5 (1.5%). Therefore, the frequency of FJVs
by 3D-planned screws was also higher in high-grade facet joint osteoarthritis than
in low-grade osteoarthritis (p < 0.001), with a relative risk of 7.4. The difference in frequency of upper FJVs,
28% in 2D-planned screws versus 4.8% in 3D-planned screws, was statistically significant
(p < 0.001) ([Table 1]). Among the 170 possible insertion sites with facet joint osteoarthritis grade 3/4,
FJV was significantly lower with 3D-planned screws (19 violations, 11.1%) than with
2D-planned screws (110 violations [64.7%])
Table 1
Upper facet joint violations by 2D- versus 3D-planned screws
|
Variants
|
Violations by 2D-planned screws % (n)
|
Violations by 3D-planned screws %
|
p
[a]
|
|
Overall (n = 500)
|
28 (140)
|
4.8 (24)
|
< 0.0001
|
|
Facet joint osteoarthritis Pathria grades 3 and 4 (n = 170)
|
65 (110)
|
11 (19)
|
< 0.0001
|
a McNemar test is used.
Comparison of 2D- versus 3D-Planned Percutaneous Pedicle Screw Insertion for Pedicle
Wall Violation Rates
In terms of pedicle wall integrity, no statistically significant difference between
the two groups was found. Among the 500 pedicle screws simulated in this study, 463
(92.6%) of the 2D group and all the screws of the 3D group were interpreted as correctly
inserted within the pedicle. Cortical encroachment was found for 35 screws (7%), and
frank penetration (> 2 mm) for 2 (0.4%) of the 2D group.
Comparison of 2D- versus 3D-Planned Percutaneous Pedicle Screw Insertion for Entry
Points and Trajectories
The difference between the entry point locations for the 2D- and 3D-planned screws
are depicted in [Tables 2] and [3]. A more lateral (mean distance: 3 mm) and inferior (mean distance: 2.5 mm) offset
of the pedicle entry point on the lateral ridge of the pedicle projection on the AP
radiograph ([Table 2]) and a larger medial angulation of the trajectory (mean angle: 9 degrees) were observed
in the 3D group at all levels ([Table 3]). No statistically significant difference was observed inside the 3D group between
mild and severe degenerated facet joints.
Table 2
Evaluation of horizontal and inferior vertical offsets from conventional entry point
of 2D-planned screw after application of 3D planning technique for each level
|
Horizontal offset, mm
|
|
Vertical offset, mm
|
|
|
Level
|
3D
|
p value
|
3D
|
p value
|
|
Facet joint osteoarthritis Pathria 1–2
|
Facet joint osteoarthritis Pathria 3–4
|
|
Facet joint osteoarthritis Pathria 1–2
|
Facet joint osteoarthritis Pathria 3–4
|
|
|
L5
|
3.5 ± 1.5
|
4.5 ± 0.5
|
0.722
|
3 ± 1
|
4 ± 1
|
0.722
|
|
L4
|
3.5 ± 1
|
4 ± 1
|
0.722
|
2.5 ± 1
|
3 ± 1.5
|
0.714
|
|
L3
|
3 ± 1.5
|
4 ± 1
|
0.722
|
2.5 ± 0.5
|
3.5 ± 1
|
0.714
|
|
L2
|
2.5 ± 1.5
|
3 ± 2.5
|
0.714
|
1.5 ± 2
|
2 ± 1
|
0.7
|
|
L1
|
2 ± 1.5
|
2.5 ± 2.5
|
0.8
|
1.5 ± 2
|
2 ± 1
|
0.7
|
Data are presented as mean and standard deviation.
Table 3
Evaluation of transverse medial angulation angle from conventional trajectory of 2D-planned
screw after application of 3D planning technique for each level
|
Transverse angles, degrees
|
|
|
Level
|
3D
|
p value
|
|
Facet joint osteoarthritis Pathria 1–2
|
Facet joint osteoarthritis Pathria 3–4
|
|
|
L5
|
14 ± 5
|
16 ± 6
|
0.725
|
|
L4
|
10 ± 7
|
13 ± 5
|
0.696
|
|
L3
|
7 ± 2
|
8 ± 3
|
0.735
|
|
L2
|
5 ± 3
|
6 ± 2
|
0.730
|
|
L1
|
5 ± 5
|
4 ± 6
|
0.727
|
Data are presented as mean and standard deviation.
Time for Planning of Surgery
In the statistical analysis, the mean time for the completion of the 3D planning was
23 minutes (95% confidence interval [CI], 15–28 minutes), whereas the time spent in
the 2D group was 12 minutes (95% CI, 5–19 minutes). This difference was statistically
significant (p < 0.05).
Discussion
The study presented here is the continuation of a previous retrospective report that
assessed the performances of the 3D-planned and percutaneous robot-guided pedicle
screw placement when compared with a percutaneous fluoroscopic-guided freehand technique
in clinical practice.[5]
Impact of 3D Planning on Upper Facet Joint Violation Rates
The results presented here indicate the pedicle screw accuracy in terms of the integrity
of the upper facet joints is higher in the 3D-planned percutaneous pedicle screw simulation
than in procedures where only 2D images are available. It appears that the 3D-guided
technique leads to a more lateral and inferior entry point compared with the conventional
2D fluoroscopy-guided technique. This could be explained by the fact that the facet
joints are only poorly visualized in the AP and lateral fluoroscopic view but rather
in the oblique view, which is not popular among spine surgeons because orientation
is very difficult.[3]
Our results showed that 3D planning reduced the rates of upper FJV in a simulation
setting. Several studies reported on the incidence of FJV with percutaneously placed
pedicle screws.[13]
[14] Although many studies investigated the rate of pedicle perforations in percutaneous
and/or navigated procedures, < 10% of these trials and most of them since 2010 report
the results on upper FJV rates.[3]
[4]
[5]
[8]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20] The development of 3D navigation systems during the last few years and the increasing
interest in the prevention of adjacent-level disease due to increased spine instrumentations
may be the reason for this new focus.
Facet joint abutment through the screw head is possible, but in clinical practice
it is clearly associated with the insertion depth of the pedicle screw, a fact that
can be modified intraoperatively according to the anatomical conditions.
According to our results, application of 3D planning tools moved the entry point more
laterally and inferiorly. Thus surgeons need to modify the insertion technique of
the percutaneous technique when based only on 2D fluoroscopic images. If surgeons
are not equipped with navigation tools, robotic assistance, and 3D fluoroscopic devices,
we recommend the following: (1) Plan the screws thoroughly according to the preoperative
CT data. (2) Palpate the anatomy of the transverse and lateral facet through a mini
open extension of the incision, avoiding penetration of the joint capsule. (3) Use
a more lateral (mean distance: 3 mm) and inferior (mean distance: 2.5 mm) offset of
the pedicle entry point on the lateral ridge of the pedicle projection on the AP radiograph.
(4) Consider a sufficient medial angulation so the screw tip reaches the ventral third
and also reaches the medial third of the vertebral body.
Effect of Facet Degeneration
Although many reports have proposed the optimal entry points for avoiding penetration
of facet joints, these studies did not consider the technique of implantation and
the degree of facet joint degeneration.[21]
[22]
[23] Hypertrophy and osteophytes of the facet joints are observed with aging and degenerative
disease. Thus changes of the entry points and angulation angle in situ in cases of
open approach and in the conventional fluoroscopy images in cases of minimally invasive
percutaneous approach may be necessary. However, until now no studies have compared
entry points and trajectories of lumbar pedicle screws according to their 2D- or 3D-planned
routes. Thus understanding the entry point and trajectories of pedicle screws is critical
to avoid violation of upper unfused facet joints. According to our results, surgeons
need to modify the insertion technique of the 2D-planned pedicle screws when considering
inserting screws at the upper vertebra, especially in the degenerative lumbar spine,
to maintain the integrity of the upper unfused facet joint. In fact, pedicle screw
insertion during fluoroscopy-guided percutaneous surgery with a minimum of 3 mm lateral
to the outer pedicle margin on the AP radiograph might be a good option. However,
it is technically demanding and could lead to perforation of the lateral pedicle wall.
Therefore, a 3D-guided technique as provided by robotic or image guidance could be
considered an option if degenerative changes are severe.
Clinical Significance of Upper Facet Violation
FJV does not always cause clinical symptoms. But the violation of the most upper adjacent
facet joint could increase the risk of facet arthritis. Lumbar fusion is reported
to accelerate degenerative changes of the adjacent facet joints and disks.[24] In addition, facet degeneration could be a significant factor in the occurrence
of low back pain and adjacent-level disease, which is more prominent in the proximal
adjacent segment.[9] In a retrospective cohort study of 240 patients, FJV was independently associated
with a higher reoperation rate and diminished improvement in quality of life.[2] At 2-year follow-up, patients in the FJV group were less likely to make a significant
improvement on the EQ-5D questionnaire (p = 0.041). Also, the reoperation rate in the FJV group was significantly higher than
in the control group at 2 years (15.2% versus 6.3%, respectively; p = 0.024) and 3 years (19.6% versus 9.4%; p = 0.023). Thus violation of the upper unfused facet joints by the pedicle screws
might accelerate the degenerative process of the adjacent level. Hence the surgeon
should (1) plan the screws preoperatively, and pay attention to morphological changes
of the spine that could affect the entry point and trajectory; and (2) consider intraoperatively
the inherent risk of FJV of the percutaneous 2D fluoroscopy-guided insertion technique.
The role of soft tissue violation surrounding the facet joints, causing capsular impingement,
for example, has not been studied yet, and it is unclear whether it is promoting osteoarthritic
changes.
Time for Planning of Surgery
We found that the process of 3D-based planning required significantly more time compared
with conventional 2D planning (∼ 23 versus 12 minutes). This is a drawback, of course,
because time is scarce in clinical routine. However the 11 or 12 additional minutes
may be well spent if they facilitate surgery and render the process safer. One single
revision will quite outweigh several planning procedures. Furthermore, current 3D
planning (and navigation) software generally allow archiving the preoperative plan.
Although this is not yet obligatory for spinal instrumentation, it is mandatory in
other fields like knee or hip replacement surgery and definitely helpful if questions
arise later on.
Pedicle Wall Integrity
In this simulation study we found good accuracy of the pedicle screws as far as the
pedicle wall integrity is concerned when using 2D planning during simulated surgery.
Although we did not observe any pedicle wall violations in the 3D group, this difference
was found to be statistically insignificant. Although published clinical series described
a lower rate of pedicle violations,[6]
[7]
[8]
[24]
[25] the failure to show this in our study might be attributed to the relatively low
number of cases (250 patients). Or, more probable, the difference observed in clinical
series was due to the intraoperative guidance provided by navigation systems rather
than an improved planning tool.
Usefulness of 3D Visualization for Educational Purposes
The use of 3D monitoring systems in medical education offers the advantage of stereopsis
and contributes to surgical training.[26] In spine surgery, 3D visualization can be extremely useful for navigating complex
deformities and improving anatomical understanding for training. Furthermore, 3D visualization
software could be useful in understanding space interval and depth of implants in
relation to spine structures. Medical students, medical support staff, and patients
who are unfamiliar with the anatomy of the pathology can easily understand anatomical
relationships with 3D models. And 3D printing is a further development and a growing
transformative technology with a potentially wide range of applications in the field
of spine surgery. Life-size 3D models can not only allow observation but also actual
cutting and drilling using surgical instruments, which in turn could considerably
enhance a surgeon's skill and improve risk management for complex surgical procedures.
Limitations of the Study
The presented study only assesses the planning process. Upon translation of the planning
results into clinical practice, a certain number of additional FJVs will occur due
to imprecise intraoperative positioning of screws. However, actual placement of screws
into bone in experienced hands is not a completely radiographic procedure and is enabled
through tactile feedback. The surgeon can palpate the anatomy of the transverse and
lateral facet even in percutaneous procedures through a mini open extension of the
incision and can merge the information gleaned from the fluoroscopy. Nevertheless,
the difference observed here is relevant because clinical results could presumably
worsen if only 2D fluoroscopic data are relied on. The application of a navigation
system will also be beneficial at this stage of the operation.
Until recently the significance of upper FJVs was only hypothesized.[9]
[24] In 2018, Levin et al presented the first clinical evidence highlighting the potential
morbidity associated with FJV at the superior-most fusion level.[2] Further studies are needed to prove that facet osteoarthritis through pedicle screw
violation is a relevant clinical entity. Nevertheless, apart from this point, the
data presented show that careful planning using 3D data may improve the precision
of pedicle screw placement in general.
Another point is the applied software. Although several types of spinal navigation
and planning software exist, every such tool can be prone to some errors, and results
should eventually be verified in a clinical series. Especially those patients who
had to be excluded from the study because of technical difficulties with the registration
process might be problematic cases that are not covered by our results.
The screws in our study were all placed by a single surgeon, experienced in both conventional
and navigated techniques. This makes the data more coherent, but of course it also
depends on the individual capabilities of this surgeon. If the study was conducted
by another individual with his or her set of experiences, the results might be different.
This is a clear limitation because the reliability between more than one observer
could not be assessed.
Although rapidly developing, spinal navigation and robotic surgical spinal technology
have not achieved their full potential owing to some limitations. Cost effectiveness
is a major issue that is not clarified in the literature. Other drawbacks to robotic
surgery include the lack of tactile feedback to the surgeon and the bulkiness of the
robotic equipment currently in use.
Nevertheless, regardless of the limitations just described, this is the first study
in a consecutive group of patients to assess the value of 3D versus 2D planning quantitatively
in regard to pedicle screw accuracy and adjacent facet joint integrity. The results
presented here suggest that 3D reconstruction and planning is a useful tool for preoperative
planning of pedicle screw positions.
Conclusion
This study demonstrates that the application of only 2D fluoroscopic images for pedicle
screw insertion has a more inherent anatomical risk of upper FJV. Application of 3D
planning tools moved the entry point more laterally and increased the rate of radiographically
intact upper facet joints. Thus surgeons need to modify the insertion technique of
the percutaneous technique when based on only 2D fluoroscopic images. In the presence
of severe facet joint arthropathy, simulated placement of 3D-planned screws is significantly
safer than the placement of 2D-planned screws.
Complementary prospective randomized in vivo studies could be performed to fully validate
the results observed here in terms of accuracy, repeatability, and ergonomics.
