- Email: [email protected]
Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty During Surgical Spine Reconstruction
Journal Pre-proof Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty during Surgical Spine Reconstruction Michael A. Bohl, MD, Rajiv Sethi, MD, Jean-Christophe Leveque, MD PII:
S1878-8750(19)32870-0
DOI:
https://doi.org/10.1016/j.wneu.2019.11.034
Reference:
WNEU 13703
To appear in:
World Neurosurgery
Received Date: 12 August 2019 Revised Date:
5 November 2019
Accepted Date: 6 November 2019
Please cite this article as: Bohl MA, Sethi R, Leveque J-C, Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty during Surgical Spine Reconstruction, World Neurosurgery (2019), doi: https://doi.org/10.1016/j.wneu.2019.11.034. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.
Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty during Surgical Spine Reconstruction
Michael A. Bohl, MD1,2 Rajiv Sethi, MD2 Jean-Christophe Leveque, MD2
1
Department of Neurosurgery
Barrow Neurological Institute St. Joseph’s Hospital and Medical Center 350 W Thomas Rd, Phoenix, AZ 85013
2
Neuroscience Institute
Virginia Mason Medical Center 1201 Terry Ave, Seattle, WA 98101
Correspondence: Michael Bohl, MD Department of Neurosurgery Barrow Neurological Institute St. Joseph’s Hospital and Medical Center Phoenix, AZ Tel: 602-509-2132 Email: [email protected]
Key Words: Adult; Bone Cements; Osteotomy; Vertebroplasty
Short Title: Vertebroplasty during Spine Reconstruction
Conflicts of Interest: The authors have no conflicts of interest relevant to this study.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Research Ethics Committee: This study was submitted for review by the Institutional Review Board, and review was deemed unnecessary given that the patient data was de-identified, reported in aggregate, and retrospectively reviewed.
Device Status/ Drug Statement: The Manuscript submitted does not contain information about medical device(s)/drug(s).
Prophylactic Vertebroplasty During Spine Reconstruction 1 1
INTRODUCTION
2
Complication rates following surgical reconstruction of adult spinal deformities (ASD) remain
3
high despite advances in procedural safety and techniques.[1-6] Many have advocated for the use of
4
targeted prophylactic cement vertebroplasty in these cases to reduce complications that are thought to
5
be related to poor bone quality, including proximal junctional kyphosis (PJK), pedicle screw pull-out, and
6
graft subsidence.[7-11] Although numerous studies have evaluated the effectiveness of prophylactic
7
vertebroplasty in preventing these complications, none have evaluated the safety of vertebroplasty
8
when performed prophylactically during extensive surgical reconstruction procedures.[7-11]
9
Prophylactic vertebroplasty in this context can broadly be defined as any cement augmentation of a
10
vertebral body for the purpose of preventing PJK secondary to vertebral body fracture or pedicle screw
11
failure. Although vertebroplasty is generally thought to be safe, there are numerous reports of
12
devastating complications arising from this technique when performed percutaneously for the isolated
13
treatment of vertebral body (VB) compression fractures.[12-38] These complications result from cement
14
extravasation into the spinal canal, neural foramina, or into the perivertebral venous system. Cement
15
leakage into the spinal canal or neural foramina has been reported to cause spinal cord or nerve root
16
injury.[19-26] Cement extravasation into the perivertebral venous plexus has reportedly caused
17
cardiopulmonary failure, impaired renal function, stroke, and even death.[27-38] See Table 1 for a list of
18
known complications resulting from vertebroplasty.
19
Numerous studies have evaluated the incidence of cement extravasation following
20
percutaneous cement kyphoplasty or vertebroplasty but study methodologies and results vary widely.
21
Reported incidences of cement extravasation range from 4% to 72%, with studies utilizing postoperative
22
computed tomography (CT) reporting higher incidences while those using postoperative fluoroscopy
23
report much lower incidences of extravasation.[10,12,18,39-41] The reported incidence of pulmonary
24
cement embolism ranges from 2.1% to 26%, with the large majority being asymptomatic emboli with no
Prophylactic Vertebroplasty During Spine Reconstruction 2 25
permanent clinical sequelae.[15] Given the increasing usage of prophylactic cement vertebroplasty in
26
complex spine patients undergoing spinal reconstruction, it is important to understand the risks of
27
cement injection in this unique population undergoing a much more invasive procedure. The purpose of
28
this study was to measure the incidence and severity of cement extravasation in a population of adult
29
patients undergoing prophylactic vertebroplasty as part of a spinal reconstruction procedure.
30 31 32
MATERIALS AND METHODS This study was submitted for review by the Institutional Review Board, and review was deemed
33
unnecessary given that the patient data was de-identified, reported in aggregate, and retrospectively
34
reviewed.
35 36 37
Data Collection Between January 2015 and October 2018, all consecutive adult patients treated with surgical
38
spine reconstruction involving prophylactic vertebroplasty were reviewed. All patients were treated by
39
the same attending neurosurgeon (author JCL) and a second attending orthopedic surgeon (author RS)
40
as part of a two-surgeon team. Any patient that did not have a postoperative CT was excluded from the
41
analysis. Surgical spine reconstruction was defined as any case involving the treatment of 6 or more
42
spine levels and/ or a 3-column osteotomy. Prophylactic vertebroplasty was defined as the injection of
43
bone cement into a VB for the limited purpose of complication avoidance, rather than treatment of a
44
pre-existing VB fracture.
45 46
Vertebroplasty Technique
47
The attending surgeons used the same vertebroplasty technique in all patients. Fenestrated
48
pedicle screws or Jamshidi needles were placed through bilateral pedicles at the levels to be treated
Prophylactic Vertebroplasty During Spine Reconstruction 3 49
under fluoroscopic navigation. Jamshidi needles were inserted through the paraspinal muscle fascia via
50
a skin incision extended above or below the open posterior surgical exposure. Fenestrated pedicle
51
screws were placed via an open posterior surgical approach and high-viscosity cement was injected
52
directly into the exposed pedicle screws. After waiting a sufficient time period for the cement to achieve
53
optimally high viscosity, each pedicle was injected at an approximate rate of 2 cc per minute while
54
monitoring a lateral fluoroscopic view every 2-5 seconds. Cement injection was stopped either when 2
55
cc of bone cement was injected through each pedicle (4 cc maximum for each vertebral body) or the
56
lateral fluoroscopy showed evidence of cement extravasation, whichever came first.
57 58 59
Data Analysis CT scans of each patient were evaluated for number of levels receiving prophylactic
60
vertebroplasty, the spine levels treated, the use of fenestrated pedicle screws or Jamshidi needles, the
61
presence of cement extravasation outside the treated VB, and the severity of cement extravasation if
62
present. Several cement extravasation grading systems have previously been reported in the literature,
63
but none include a clinically relevant grading scale for severity of extravasation.[42,43] A new grading
64
scale was therefore developed. Cement extravasation was given a simple grade from 0 to 5. Grade 0
65
meant no cement could be seen outside the treated VB on CT scan (Figure 1A). Grade 1 meant cement
66
was present outside the VB but within 5 mm of the treated VB (Figure 1B). Grade 2 meant cement was
67
greater than 5 mm ventral or ventrolateral to the VB but not extending above or below the treated
68
vertebral level (Figure 1-C). Grade 3 meant the extravasated cement extended one or more levels above
69
or below the treated VB but not to the vena cava (Figure 1D). Grade 4 meant the cement reached either
70
the vena cava or an end-organ such as the heart or lungs (Figure 1E). Grade 5 meant the cement
71
extravasated dorsally into the spinal canal and caused 20% or greater spinal canal stenosis (Figure 1F).
72
See Table 2 for a summary of this new grading scale.
Prophylactic Vertebroplasty During Spine Reconstruction 4 73
After all the treated VBs were evaluated and graded, descriptive statistics were calculated for
74
patient population demographics, the number of VBs treated, the number of VBs demonstrating cement
75
extravasation, and the number of VBs demonstrating each grade of extravasation. Statistical
76
comparisons were then performed using chi-square analyses and Fisher’s exact tests to determine if
77
there was a correlation between grade of extravasation and treated spinal level and grade of
78
extravasation and use of fenestrated pedicle screws. Medical charts of patients who had grade 4 or 5 CT
79
findings were evaluated to determine what, if any, clinical sequelae resulted. Finally, imaging was used
80
to determine how many patients had suffered postoperative PJK/ PJF and the length of radiographic
81
follow-up available for each patient.
82 83
RESULTS
84
Population Demographics
85
Between January 2015 and October 2018, 52 patients underwent prophylactic vertebroplasty
86
during a surgical procedure. Thirteen of these patients did not have postoperative CT scans and so were
87
eliminated from the study. Of the remaining 39 patients, 5 did not meet the criteria for a surgical spinal
88
reconstruction procedure (6 or more levels fused and/ or a 3-column osteotomy), leaving 34 patients for
89
analysis. Collectively, these 34 patients comprised a total of 112 VBs treated with prophylactic
90
vertebroplasty as part of a spinal reconstruction procedure. The mean radiographic follow-up for this
91
patient population was 8.8 months (standard deviation, 10.6 months). The mean (range) levels fused for
92
the study population was 9.3 (3-16), and the number of 3-column osteotomies performed was 15. The
93
mean number of VBs injected per patient was 3.3 (range 1-10, standard deviation 2.08). Table 3
94
summarizes the surgical pathologies indicating spinal reconstruction for all 34 patients. The most
95
common pathology was adult degenerative spinal deformity, comprising 15 of 34 patients (44.1%). The
Prophylactic Vertebroplasty During Spine Reconstruction 5 96
next most common indication was tumor (26.5%), followed by trauma (11.8%), PJK/ PJF (8.8%), infection
97
(5.9%), and adult idiopathic scoliosis (2.9%).
98 99 100
Incidence of Cement Extravasation All 34 evaluated patients (100%) had at least grade 1 cement extravasation present on
101
postoperative CT scan, and 103 out of 112 VBs (92.0%) demonstrated at least grade 1 cement
102
extravasation. Nine vertebral bodies had no (grade 0) extravasation (8.0%). Thirty-one vertebral bodies
103
(27.7%) had a grade 1 extravasation, 24 (21.4%) were grade 2, 35 (31.3%) were grade 3, 8 (7.1%) were
104
grade 4, and 5 (4.5%) were grade 5. If grades 0 to 3 are combined into the same group of clinically non-
105
worrisome events, and grades 4 and 5 are grouped into clinically worrisome events, we have 99 VBs
106
(88.4%) with non-worrisome extravasation and 13 VBs (11.6%) with clinically worrisome extravasation.
107
Eighty-five VBs (75.9%) had cement injected via a fenestrated pedicle screw, and 27 VBs (24.1%) were
108
injected via a Jamshidi needle. Of note, all 9 grade 0 VBs (100%) were injected via fenestrated screws.
109
See Figure 2 for a summary of these results.
110
A full report of the spinal levels treated and extravasation grades at each of those levels can be
111
found in Figure 3. In summary, all spinal levels from C7 to L5 (with the exception of T1) were treated
112
with prophylactic vertebroplasty in this study population. The most frequently treated levels were T9
113
and T10. Levels with 2 or fewer treated VBs included C7, T1, L4, and L5.
114 115 116
Statistical Comparisons For comparative testing of spinal levels and risk of cement extravasation, the data were divided
117
into groups of upper thoracic (C7-T6) and lower thoracic-lumbar (T7-L5) levels, as well as clinically
118
worrisome extravasation events (grades 4 and 5) and clinically non-worrisome events (grades 0 to 3).
119
Fisher’s exact test of independence demonstrated a significantly higher rate of grade 4 and 5
Prophylactic Vertebroplasty During Spine Reconstruction 6 120
extravasation events for upper thoracic levels than for lower thoracic-lumbar levels (p = 0.009). See
121
Table 4 for this analysis.
122
Statistical comparison for predictability of fenestrated screws to Jamshidi needles for high grade
123
cement extravasation was performed using Fisher’s exact test. Extravasation grades were again grouped
124
into clinically worrisome (grades 4-5) and clinically non-worrisome (grades 0-3) categories, and
125
compared to groups of fenestrated screws and Jamshidi needles. No significant difference in the risk of
126
high-grade cement extravasation was found between fenestrated pedicle screws and Jamshidi needles
127
(p=0.29). See Table 4 for this analysis.
128 129 130
Clinical Sequelae A total of 9 patients (26.5%) suffered 13 grade 4 or grade 5 cement extravasation events,
131
comprising 11.6% of all injected VBs. Manual chart review was performed for these 9 patients to
132
determine what, if any, clinical sequelae resulted. Table 5 provides a summary of these results. Of note,
133
no permanent or symptomaticclinical sequelae were identifiable from these events at the time of
134
hospital discharge. The most clinically worrisome event involved a patient who suffered a temporary
135
desaturation into the low 80s after cement injection of T3 and T4. The procedure was aborted at that
136
time and postoperative CT scan of the chest showed cement in the superior vena cava, right atrium,
137
right ventricle, right main pulmonary artery, and several branches of the right and left pulmonary
138
arteries. The patient awoke hemodynamically stable. The decision was made not to anticoagulate the
139
patient, and the procedure was successfully completed 5 days later. Several weeks postoperative, the
140
patient resumed the preoperative dose of warfarin for treatment of atrial fibrillation.
141
Finally, only 1 out of the 34 evaluated patients had radiographic evidence of PJK (2.9%) on
142
followup imaging (Figure 4). It is suspected that this patient’s lumbar hypolordosis was over-corrected at
143
the time of surgery, leading to an increased risk of PJK.
Prophylactic Vertebroplasty During Spine Reconstruction 7 144 145 146
DISCUSSION Prophylactic vertebroplasty is becoming a commonly performed technique for complication
147
avoidance in surgical spine reconstruction procedures. The results of this study suggest that although
148
this technique appears effective (2.9% PJK rate in this population with a limited followup), it certainly
149
comes with an increased potential risk related to cement extravasation. 100% of patients demonstrated
150
some amount of extravasation, and 92% of all injected VBs had CT evidence of cement extravasation.
151
These extravasation rates are much higher than those reported in the percutaneous kyphoplasty and
152
vertebroplasty literature. There are two possible reasons why our study found a much higher rate of
153
extravasation. First, our study method excluded any patients who did not have a postoperative CT scan.
154
Most low grade extravasation events are unlikely to be identifiable on fluoroscopy or postoperative
155
radiographs, and many published studies use either CT scan or fluoroscopy to retrospectively identify
156
cement extravasation. Secondly, a number of important procedural and anatomical differences exist
157
between prophylactic vertebroplasty during spinal reconstruction procedure and percutaneous
158
vertebroplasty for a single level compression fracture. For example, it is possible that VBs with
159
compression fractures have disrupted venous drainage, and are therefore less likely to permit cement
160
extravasation during vertebroplasty. Patients undergoing spinal reconstruction also are frequently
161
subject to high volume blood loss, coagulopathy, and major shifts in body fluid volumes. These
162
differences in patient physiology may also play a role in the higher risk of cement extravasation during
163
prophylactic vertebroplasty.
164
Interestingly, the large majority of extravasation events (88.4%) were lower grade (grade 0-3)
165
and therefore clinically non-worrisome. No reports have been made in the literature of local cement
166
vascular extravasation or non-compressive dorsal extravasation causing any short-term or long-term
167
complications. This absence makes sense given the relatively minor potential impact that cement can
Prophylactic Vertebroplasty During Spine Reconstruction 8 168
have if it remains in perivertebral veins and doesn’t cause spinal canal or foraminal stenosis. However,
169
26.5% of all patients in this study were found to have 1 or more grade 4 or 5 (clinically worrisome)
170
extravasation events. Although none of these patients went on to develop overt long-term clinical
171
sequelae from these events, the potential for devastating complications in similarly severe episodes of
172
cement extravasation has been well reported.[19-38] We furthermore cannot rule out the possibility
173
that these patients have suffered long-term subclinical sequelae from these high grade extravasation
174
events as cardiopulmonary functional studies were not routinely completed postoperatively.
175
It is important to consider these findings on the potential risk of using prophylactic cement
176
vertebroplasty within the context of the known potential risk of PJK. Failure of the spine at the proximal
177
and distal junctions of long-segment spinal fusions is also a widely known risk with potentially
178
devastating consequences. Although prophylactic cement vertebroplasty clearly comes with an
179
additional risk, larger prospective studies may discover that the actual risk of prophylactic
180
vertebroplasty is offset by the improved outcomes of reduced PJK risk.
181
Interestingly, the risk of high grade cement extravasation was highest in the upper thoracic
182
spine. Of the 32 upper thoracic VBs injected in this study, 8 VBs (25%) had CT evidence of high grade
183
extravasation. This is compared to 5 out of 80 (6.3%) lower thoracic-lumbar VBs with high grade
184
extravasation. There are several reasons why upper thoracic VBs may have a higher risk for cement
185
injection. First, it is more difficult to visualize these VBs on a lateral fluoroscopic view during an open
186
spinal reconstruction procedure, making it more difficult to identify when cement may be extravasating
187
outside the VB. Secondly, there is a shorter distance in the upper thoracic spine for a cement embolus to
188
travel from the VB to the vena cava, heart, and lungs. Finally, upper thoracic VBs are smaller and may
189
therefore tolerate a smaller amount of cement. These factors, and potentially others, likely contributed
190
to the higher association between clinically worrisome cement extravasation and upper thoracic VBs.
191
Because of these findings, the surgeons involved in this study have transitioned from performing
Prophylactic Vertebroplasty During Spine Reconstruction 9 192
prophylactic vertebroplasty in the upper thoracic spine for PJK prevention, to using transverse process
193
hooks at the top 2 upper instrumented vertebrae.
194
It is also important to note the lack of association between extravasation grade and the use of
195
fenestrated screws or Jamshidi needles. One might predict that given the larger bore size of the Jamshidi
196
needle, cement extravasation might be higher in this subset as a larger bolus of cement can be delivered
197
more quickly. Although we did find that the only VBs with no cement extravasation (grade 0) were
198
cemented through fenestrated pedicle screws, we found overall no significant difference in the rate of
199
high grade extravasation between fenestrated screws and Jamshidi needles. Perhaps the smaller bore of
200
the fenestrated pedicle screws produces smaller cement fragments that have an easier time entering
201
the peri-vertebral veins. Larger studies will need to be conducted to determine if there truly is no
202
difference in the extravasation risk between fenestrated pedicle screws and Jamshidi needles.
203
Finally, the results of this study have motivated a new effort by the authors to study in detail the
204
numerous variables that are thought to contribute to cement extravasation risk. For example, cement
205
viscosity has been identified in the literature as a contributing factor to cement extravasation risk.[44]
206
Bone mineral density (BMD), rate of cement injection, volume of cement injection, presence or absence
207
of vertebral body fracture or tumor, and level of cement injection also likely affect the risk of cement
208
extravasation. It is our practice to wait until the bone cement has achieved a subjectively high viscosity
209
before injecting it into the vertebral body, but we utilize no standard waiting period nor objective
210
measure of viscosity, and therefore were unable to evaluate the impact of cement viscosity on this
211
patient population. We also considered evaluating the impact of bone mineral density on risk of cement
212
extravasation in our study population, as more osteoporotic bone may be more prone to permitting
213
cement extravasation. Preoperative DEXA or CT scans were not included in our study inclusion criteria.
214
Postoperative CT scans are furthermore inadequate for evaluating BMD (via Hounsfield unit
215
measurement) because the presence of spinal hardware and cement affects the relative radiodensity of
Prophylactic Vertebroplasty During Spine Reconstruction 10 216
surrounding VBs. We were therefore unable to evaluate bone mineral density as a predictive variable for
217
cement extravasation risk in this study population. Finally, underlying condition was difficult to evaluate
218
for impact on extravasation risk as our patient population had a high amount of heterogeneity in
219
number of levels fused, number of levels cemented, spinal levels treated, and underlying condition
220
indicating spinal reconstruction. As such, we felt that any attempt to reach conclusions about the impact
221
of underlying condition on extravasation risk would be severely limited.
222
Limitations
223
This is a retrospective study and all the limits of retrospective radiographic reviews are therefore
224
relevant to these results. The study is also relatively small, limiting the power of the study to determine
225
lack of significant differences between study groups. Although the studied patient population is small at
226
34 patients, these patients comprised 112 VBs, making us more confident in the results. Finally, the
227
results are likely dependent on the cement injection technique used, and other surgeons may find
228
different results if their vertebroplasty technique significantly differs from that described herein. For
229
example, it is possible that usage of biplanar fluoroscopy during cement injection could minimize lateral
230
cement extravasation into the perivertebral venous plexus, thereby reducing the risk of high grade
231
extravasation events to larger, more distal venous structures and end-organs. It is also likely that by
232
changing the volume of cement injected at each VB, one could significantly change the extravasation
233
risk. At our institution, we have limited the volume of cement injection to 2 ml per pedicle as this seems
234
like an adequate volume of cement to fill the vertebral body. It is possible that 1.5 ml or 1.0 ml of
235
cement per pedicle would provide sufficient biomechanical support while reducing the risk of
236
extravasation. No biomechanical studies have yet provided the data needed, however, to make this
237
determination.
238
Finally, postoperative CT scans were not dedicated chest and abdomen views, and it is therefore
239
possible that distant cement emboli were missed on the postoperative CT scans obtained, especially for
Prophylactic Vertebroplasty During Spine Reconstruction 11 240
lumbar levels. We attempted to minimize the potential bias this might introduce into the data by
241
considering the inferior vena cava an end-organ and including this structure in our grade 4
242
extravasations. Nevertheless, it is possible that cement could have traveled from a vertebral body all the
243
way to the heart or lungs without leaving a trace behind in the perivertebral venous plexus or inferior
244
vena cava. This could potentially skew the data toward identifying more emboli for thoracic levels, as
245
thoracic CT scans include views of the heart and lungs.
246 247
Conclusions
248
Prophylactic cement vertebroplasty is an increasingly common technique employed in spinal
249
reconstruction procedures for the avoidance of common complications related to bone quality.
250
Although no patients in this study suffered known long-term consequences of this technique, 26.5% of
251
patients had cement extravasation that threatened end-organs or neural elements. These results should
252
serve as a reminder to those performing this technique that cement injection comes with very real risks,
253
and should therefore only be performed when the potential benefit of use is thought to outweigh the
254
potential risks. Based on our results, upper thoracic vertebrae come with an elevated risk of high grade
255
cement extravasation. These results prompted our transition to using transverse process hooks in the
256
upper thoracic spine for PJK prevention. Larger prospective studies will be needed to verify these
257
results.
FIGURE LEGEND Figure 1: Computed tomograms demonstrating examples of different grades of cement extravasation. From left to right and top to bottom, grade 0 extravasation with all the cement contained in the index vertebral body, grade 1 with the extravasated cement less than 5 mm from the index vertebral body, grade 2 with extravasated cement greater than 5 mm from the index vertebral body but not extending above or below the index level, grade 3 with extravasated cement extending 1 or more levels above or below the index vertebral body, grade 4 with extravasated cement in the vena cava or, in this case, the heart and lungs, grade 5 extravasation with cement causing 20% or greater stenosis of the spinal canal. Figure 2: A stacked bar chart depicting the number of vertebral bodies at each cement extravasation grade by the cement injection technique, fenestrated pedicle screws or Jamshidi needles. Figure 3: A stacked bar chart depicting the number of injected vertebral bodies at each spinal level by the severity of cement extravasation, clinically worrisome or clinically non-worrisome. A dashed line is placed between T6 and T7 as this was the division point between upper thoracic and lower thoraciclumbar test groups for statistical comparisons. Figure 4: Postoperative standing lateral radiograph of the single patient who suffered proximal junctional kyphosis in this patient population.
Table 1: Known Complications Following Vertebroplasty Complications Reported Following Vertebroplasty Foraminal stenosis and radiculopathy Central canal stenosis and spinal cord injury Anterior spinal artery syndrome Intradural cement leakage Cement vascular emboli to inferior vena cava Cement emboli causing stroke Cardiopulmonary cement emboli Intra-atrial thrombus Cardiac perforation Tricuspid regurgitation Acute respiratory distress syndrome and respiratory failure Renal cement embolism Fatal pulmonary embolism
Table 2: Novel Grading System for Clinical Severity of Cement Extravasation Grade 0
Computed Tomogram Findings No cement is visible outside the cortical walls of the index vertebral body
1
Extravasated cement is within 5 mm of the nearest cortical wall of the index vertebral body Extravasated cement is greater than 5 mm from
2
nearest cortical wall of the index vertebral body, but does not extend a spinal level above or below the index level Extravasated cement extends 1 or more levels
3
above or below the index level, but does not reach the superior or inferior vena cava
4
Extravasated cement reaches the superior or inferior vena cava or an end-organ Dorsally extravasated cement causes spinal canal
5
stenosis, defined as reduction in spinal canal diameter of 20% or more
Abbreviations: mm, millimeters
Table 3: Surgical Pathology Indicating Spinal Reconstruction Pathology
Number (%) of Patients
Adult Idiopathic Scoliosis (AIS)
1 (2.9%)
Tumor
9 (26.5%)
Infection
2 (5.9%)
Adult Degenerative Deformity (ASD)
15 (44.1%)
Trauma
4 (11.8%)
Proximal Junctional Kyphosis or Failure
3 (8.8%)
Table 4: Results of Statistical Comparisons Patient Groups
Grades 0-3 (clinically
Spine Levels
P-value
C7-T6
T7-L5
24
75
8
5
non-worrisome) Grades 4-5 (clinically
0.009
worrisome)
Cement Injection Technique Fenestrated Pedicle
P-value
Jamshidi Needles
Screws Grades 0-3 (clinically
77
22
8
5
non-worrisome) Grades 4-5 (clinically
0.290
worrisome) Results of statistical comparisons between groups of cement extravasation severity and spinal levels or cement injection technique. Fisher’s exact tests for independence were performed to account for the low number of patients with grade 4 or 5 cement extravasation.
Table 5: Clinical Outcomes of Patients with Grade 4 or 5 Cement Extravasation Patient Number
Clinical Outcome Grade 4 extravasation at T12 and L2 with cement in inferior vena cava. Patient
1
suffered transient episode of expressive aphasia postoperatively. MR brain negative. Neurology attributed aphasia to delirium.
2
Grade 5 extravasation at T2 and T7. No neurological deficit postoperatively. Tumor progression and death 7 months postoperative.
3
Grade 4 extravasation at T9 with cement in superior vena cava. No pulmonary embolus on CT scan, no treatment required. Attempted T4-Pelvis aborted after cement injection and pedicle screw insertion at T3 andT4 with subsequent loss of motor evoked potentials. Postoperative CT scan
4
showed T3 cement in superior vena cava, and non-compressive T4 cement in spinal canal. Pedicle screws removed, patient awoke motor intact. Loss of evoked potentials attributed to medial pedicle screws at T4. Procedure successfully performed 9 days later. Attempted T4-Pelvis aborted after T3 and T4 vertebroplasty with subsequent oxygen desaturation to the low 80s. Patient awoke hemodynamically stable.
5
Postoperative CT scan showed cement in the superior vena cava, right atrium, right ventricle, bilateral pulmonary arteries, and distally in the lungs. No anticoagulation deemed necessary, procedure successfully performed 5 days later.
6
Routine postoperative CT scan shows cement in left main pulmonary artery and left lower lobe artery. No clinical manifestation, no treatment required. Grade 5 extravasation at T4 along with medially placed pedicle screw. Not
7
identified intraoperatively. No changes in evoked potentials. Patient awoke motor intact. No treatment required.
8
Grade 5 cement at T2 with no monitoring changes. Patient awoke in delirium. Slowly returned to motor baseline. Asymptomatic, no treatment required. Grade 5 cement extravasation at T3. Noted intraoperatively on fluoroscopy, but no
9
monitoring changes. Procedure completed, patient awoke at baseline motor strength. No treatment required.
Figure 1 Grade 0
Grade 3
Grade 1
Grade 2
Grade 4
Grade 5
Figure 2 Number of Vertebral Bodies at Each Grade of Extravasation by Cement Injection Technique Cement Extravasation Grade
5
3
4
2 5
3
3
26
2
9
20
1
4 22
0
9 0
9
0 5
10
15
20
25
Number of Vertebral Bodies Fenestrated Screw
Jamshidi Needle
30
35
40
Figure 3 Spinal Level By Extravasation Severity Number of Vertebral Bodies
16 14 12 10 8 6 4 2 0
C7
T1
T2
T3
T4
T5
T6
T7
T8
T9
T10 T11 T12
L1
L2
L3
L4
Spinal Level grade 0-3 extravasation (clinically mild)
grade 4-5 (clinically worrisome)
L5
Figure 4
ABBREVIATIONS ASD = Adult Spinal Deformity CT = Computed Tomography PJK = Proximal Junctional Kyphosis VB = Vertebral Body
S1878-8750(19)32870-0
DOI:
https://doi.org/10.1016/j.wneu.2019.11.034
Reference:
WNEU 13703
To appear in:
World Neurosurgery
Received Date: 12 August 2019 Revised Date:
5 November 2019
Accepted Date: 6 November 2019
Please cite this article as: Bohl MA, Sethi R, Leveque J-C, Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty during Surgical Spine Reconstruction, World Neurosurgery (2019), doi: https://doi.org/10.1016/j.wneu.2019.11.034. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.
Incidence and Clinical Risk of Cement Extravasation in Adult Patients Undergoing Prophylactic Vertebroplasty during Surgical Spine Reconstruction
Michael A. Bohl, MD1,2 Rajiv Sethi, MD2 Jean-Christophe Leveque, MD2
1
Department of Neurosurgery
Barrow Neurological Institute St. Joseph’s Hospital and Medical Center 350 W Thomas Rd, Phoenix, AZ 85013
2
Neuroscience Institute
Virginia Mason Medical Center 1201 Terry Ave, Seattle, WA 98101
Correspondence: Michael Bohl, MD Department of Neurosurgery Barrow Neurological Institute St. Joseph’s Hospital and Medical Center Phoenix, AZ Tel: 602-509-2132 Email: [email protected]
Key Words: Adult; Bone Cements; Osteotomy; Vertebroplasty
Short Title: Vertebroplasty during Spine Reconstruction
Conflicts of Interest: The authors have no conflicts of interest relevant to this study.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Research Ethics Committee: This study was submitted for review by the Institutional Review Board, and review was deemed unnecessary given that the patient data was de-identified, reported in aggregate, and retrospectively reviewed.
Device Status/ Drug Statement: The Manuscript submitted does not contain information about medical device(s)/drug(s).
Prophylactic Vertebroplasty During Spine Reconstruction 1 1
INTRODUCTION
2
Complication rates following surgical reconstruction of adult spinal deformities (ASD) remain
3
high despite advances in procedural safety and techniques.[1-6] Many have advocated for the use of
4
targeted prophylactic cement vertebroplasty in these cases to reduce complications that are thought to
5
be related to poor bone quality, including proximal junctional kyphosis (PJK), pedicle screw pull-out, and
6
graft subsidence.[7-11] Although numerous studies have evaluated the effectiveness of prophylactic
7
vertebroplasty in preventing these complications, none have evaluated the safety of vertebroplasty
8
when performed prophylactically during extensive surgical reconstruction procedures.[7-11]
9
Prophylactic vertebroplasty in this context can broadly be defined as any cement augmentation of a
10
vertebral body for the purpose of preventing PJK secondary to vertebral body fracture or pedicle screw
11
failure. Although vertebroplasty is generally thought to be safe, there are numerous reports of
12
devastating complications arising from this technique when performed percutaneously for the isolated
13
treatment of vertebral body (VB) compression fractures.[12-38] These complications result from cement
14
extravasation into the spinal canal, neural foramina, or into the perivertebral venous system. Cement
15
leakage into the spinal canal or neural foramina has been reported to cause spinal cord or nerve root
16
injury.[19-26] Cement extravasation into the perivertebral venous plexus has reportedly caused
17
cardiopulmonary failure, impaired renal function, stroke, and even death.[27-38] See Table 1 for a list of
18
known complications resulting from vertebroplasty.
19
Numerous studies have evaluated the incidence of cement extravasation following
20
percutaneous cement kyphoplasty or vertebroplasty but study methodologies and results vary widely.
21
Reported incidences of cement extravasation range from 4% to 72%, with studies utilizing postoperative
22
computed tomography (CT) reporting higher incidences while those using postoperative fluoroscopy
23
report much lower incidences of extravasation.[10,12,18,39-41] The reported incidence of pulmonary
24
cement embolism ranges from 2.1% to 26%, with the large majority being asymptomatic emboli with no
Prophylactic Vertebroplasty During Spine Reconstruction 2 25
permanent clinical sequelae.[15] Given the increasing usage of prophylactic cement vertebroplasty in
26
complex spine patients undergoing spinal reconstruction, it is important to understand the risks of
27
cement injection in this unique population undergoing a much more invasive procedure. The purpose of
28
this study was to measure the incidence and severity of cement extravasation in a population of adult
29
patients undergoing prophylactic vertebroplasty as part of a spinal reconstruction procedure.
30 31 32
MATERIALS AND METHODS This study was submitted for review by the Institutional Review Board, and review was deemed
33
unnecessary given that the patient data was de-identified, reported in aggregate, and retrospectively
34
reviewed.
35 36 37
Data Collection Between January 2015 and October 2018, all consecutive adult patients treated with surgical
38
spine reconstruction involving prophylactic vertebroplasty were reviewed. All patients were treated by
39
the same attending neurosurgeon (author JCL) and a second attending orthopedic surgeon (author RS)
40
as part of a two-surgeon team. Any patient that did not have a postoperative CT was excluded from the
41
analysis. Surgical spine reconstruction was defined as any case involving the treatment of 6 or more
42
spine levels and/ or a 3-column osteotomy. Prophylactic vertebroplasty was defined as the injection of
43
bone cement into a VB for the limited purpose of complication avoidance, rather than treatment of a
44
pre-existing VB fracture.
45 46
Vertebroplasty Technique
47
The attending surgeons used the same vertebroplasty technique in all patients. Fenestrated
48
pedicle screws or Jamshidi needles were placed through bilateral pedicles at the levels to be treated
Prophylactic Vertebroplasty During Spine Reconstruction 3 49
under fluoroscopic navigation. Jamshidi needles were inserted through the paraspinal muscle fascia via
50
a skin incision extended above or below the open posterior surgical exposure. Fenestrated pedicle
51
screws were placed via an open posterior surgical approach and high-viscosity cement was injected
52
directly into the exposed pedicle screws. After waiting a sufficient time period for the cement to achieve
53
optimally high viscosity, each pedicle was injected at an approximate rate of 2 cc per minute while
54
monitoring a lateral fluoroscopic view every 2-5 seconds. Cement injection was stopped either when 2
55
cc of bone cement was injected through each pedicle (4 cc maximum for each vertebral body) or the
56
lateral fluoroscopy showed evidence of cement extravasation, whichever came first.
57 58 59
Data Analysis CT scans of each patient were evaluated for number of levels receiving prophylactic
60
vertebroplasty, the spine levels treated, the use of fenestrated pedicle screws or Jamshidi needles, the
61
presence of cement extravasation outside the treated VB, and the severity of cement extravasation if
62
present. Several cement extravasation grading systems have previously been reported in the literature,
63
but none include a clinically relevant grading scale for severity of extravasation.[42,43] A new grading
64
scale was therefore developed. Cement extravasation was given a simple grade from 0 to 5. Grade 0
65
meant no cement could be seen outside the treated VB on CT scan (Figure 1A). Grade 1 meant cement
66
was present outside the VB but within 5 mm of the treated VB (Figure 1B). Grade 2 meant cement was
67
greater than 5 mm ventral or ventrolateral to the VB but not extending above or below the treated
68
vertebral level (Figure 1-C). Grade 3 meant the extravasated cement extended one or more levels above
69
or below the treated VB but not to the vena cava (Figure 1D). Grade 4 meant the cement reached either
70
the vena cava or an end-organ such as the heart or lungs (Figure 1E). Grade 5 meant the cement
71
extravasated dorsally into the spinal canal and caused 20% or greater spinal canal stenosis (Figure 1F).
72
See Table 2 for a summary of this new grading scale.
Prophylactic Vertebroplasty During Spine Reconstruction 4 73
After all the treated VBs were evaluated and graded, descriptive statistics were calculated for
74
patient population demographics, the number of VBs treated, the number of VBs demonstrating cement
75
extravasation, and the number of VBs demonstrating each grade of extravasation. Statistical
76
comparisons were then performed using chi-square analyses and Fisher’s exact tests to determine if
77
there was a correlation between grade of extravasation and treated spinal level and grade of
78
extravasation and use of fenestrated pedicle screws. Medical charts of patients who had grade 4 or 5 CT
79
findings were evaluated to determine what, if any, clinical sequelae resulted. Finally, imaging was used
80
to determine how many patients had suffered postoperative PJK/ PJF and the length of radiographic
81
follow-up available for each patient.
82 83
RESULTS
84
Population Demographics
85
Between January 2015 and October 2018, 52 patients underwent prophylactic vertebroplasty
86
during a surgical procedure. Thirteen of these patients did not have postoperative CT scans and so were
87
eliminated from the study. Of the remaining 39 patients, 5 did not meet the criteria for a surgical spinal
88
reconstruction procedure (6 or more levels fused and/ or a 3-column osteotomy), leaving 34 patients for
89
analysis. Collectively, these 34 patients comprised a total of 112 VBs treated with prophylactic
90
vertebroplasty as part of a spinal reconstruction procedure. The mean radiographic follow-up for this
91
patient population was 8.8 months (standard deviation, 10.6 months). The mean (range) levels fused for
92
the study population was 9.3 (3-16), and the number of 3-column osteotomies performed was 15. The
93
mean number of VBs injected per patient was 3.3 (range 1-10, standard deviation 2.08). Table 3
94
summarizes the surgical pathologies indicating spinal reconstruction for all 34 patients. The most
95
common pathology was adult degenerative spinal deformity, comprising 15 of 34 patients (44.1%). The
Prophylactic Vertebroplasty During Spine Reconstruction 5 96
next most common indication was tumor (26.5%), followed by trauma (11.8%), PJK/ PJF (8.8%), infection
97
(5.9%), and adult idiopathic scoliosis (2.9%).
98 99 100
Incidence of Cement Extravasation All 34 evaluated patients (100%) had at least grade 1 cement extravasation present on
101
postoperative CT scan, and 103 out of 112 VBs (92.0%) demonstrated at least grade 1 cement
102
extravasation. Nine vertebral bodies had no (grade 0) extravasation (8.0%). Thirty-one vertebral bodies
103
(27.7%) had a grade 1 extravasation, 24 (21.4%) were grade 2, 35 (31.3%) were grade 3, 8 (7.1%) were
104
grade 4, and 5 (4.5%) were grade 5. If grades 0 to 3 are combined into the same group of clinically non-
105
worrisome events, and grades 4 and 5 are grouped into clinically worrisome events, we have 99 VBs
106
(88.4%) with non-worrisome extravasation and 13 VBs (11.6%) with clinically worrisome extravasation.
107
Eighty-five VBs (75.9%) had cement injected via a fenestrated pedicle screw, and 27 VBs (24.1%) were
108
injected via a Jamshidi needle. Of note, all 9 grade 0 VBs (100%) were injected via fenestrated screws.
109
See Figure 2 for a summary of these results.
110
A full report of the spinal levels treated and extravasation grades at each of those levels can be
111
found in Figure 3. In summary, all spinal levels from C7 to L5 (with the exception of T1) were treated
112
with prophylactic vertebroplasty in this study population. The most frequently treated levels were T9
113
and T10. Levels with 2 or fewer treated VBs included C7, T1, L4, and L5.
114 115 116
Statistical Comparisons For comparative testing of spinal levels and risk of cement extravasation, the data were divided
117
into groups of upper thoracic (C7-T6) and lower thoracic-lumbar (T7-L5) levels, as well as clinically
118
worrisome extravasation events (grades 4 and 5) and clinically non-worrisome events (grades 0 to 3).
119
Fisher’s exact test of independence demonstrated a significantly higher rate of grade 4 and 5
Prophylactic Vertebroplasty During Spine Reconstruction 6 120
extravasation events for upper thoracic levels than for lower thoracic-lumbar levels (p = 0.009). See
121
Table 4 for this analysis.
122
Statistical comparison for predictability of fenestrated screws to Jamshidi needles for high grade
123
cement extravasation was performed using Fisher’s exact test. Extravasation grades were again grouped
124
into clinically worrisome (grades 4-5) and clinically non-worrisome (grades 0-3) categories, and
125
compared to groups of fenestrated screws and Jamshidi needles. No significant difference in the risk of
126
high-grade cement extravasation was found between fenestrated pedicle screws and Jamshidi needles
127
(p=0.29). See Table 4 for this analysis.
128 129 130
Clinical Sequelae A total of 9 patients (26.5%) suffered 13 grade 4 or grade 5 cement extravasation events,
131
comprising 11.6% of all injected VBs. Manual chart review was performed for these 9 patients to
132
determine what, if any, clinical sequelae resulted. Table 5 provides a summary of these results. Of note,
133
no permanent or symptomaticclinical sequelae were identifiable from these events at the time of
134
hospital discharge. The most clinically worrisome event involved a patient who suffered a temporary
135
desaturation into the low 80s after cement injection of T3 and T4. The procedure was aborted at that
136
time and postoperative CT scan of the chest showed cement in the superior vena cava, right atrium,
137
right ventricle, right main pulmonary artery, and several branches of the right and left pulmonary
138
arteries. The patient awoke hemodynamically stable. The decision was made not to anticoagulate the
139
patient, and the procedure was successfully completed 5 days later. Several weeks postoperative, the
140
patient resumed the preoperative dose of warfarin for treatment of atrial fibrillation.
141
Finally, only 1 out of the 34 evaluated patients had radiographic evidence of PJK (2.9%) on
142
followup imaging (Figure 4). It is suspected that this patient’s lumbar hypolordosis was over-corrected at
143
the time of surgery, leading to an increased risk of PJK.
Prophylactic Vertebroplasty During Spine Reconstruction 7 144 145 146
DISCUSSION Prophylactic vertebroplasty is becoming a commonly performed technique for complication
147
avoidance in surgical spine reconstruction procedures. The results of this study suggest that although
148
this technique appears effective (2.9% PJK rate in this population with a limited followup), it certainly
149
comes with an increased potential risk related to cement extravasation. 100% of patients demonstrated
150
some amount of extravasation, and 92% of all injected VBs had CT evidence of cement extravasation.
151
These extravasation rates are much higher than those reported in the percutaneous kyphoplasty and
152
vertebroplasty literature. There are two possible reasons why our study found a much higher rate of
153
extravasation. First, our study method excluded any patients who did not have a postoperative CT scan.
154
Most low grade extravasation events are unlikely to be identifiable on fluoroscopy or postoperative
155
radiographs, and many published studies use either CT scan or fluoroscopy to retrospectively identify
156
cement extravasation. Secondly, a number of important procedural and anatomical differences exist
157
between prophylactic vertebroplasty during spinal reconstruction procedure and percutaneous
158
vertebroplasty for a single level compression fracture. For example, it is possible that VBs with
159
compression fractures have disrupted venous drainage, and are therefore less likely to permit cement
160
extravasation during vertebroplasty. Patients undergoing spinal reconstruction also are frequently
161
subject to high volume blood loss, coagulopathy, and major shifts in body fluid volumes. These
162
differences in patient physiology may also play a role in the higher risk of cement extravasation during
163
prophylactic vertebroplasty.
164
Interestingly, the large majority of extravasation events (88.4%) were lower grade (grade 0-3)
165
and therefore clinically non-worrisome. No reports have been made in the literature of local cement
166
vascular extravasation or non-compressive dorsal extravasation causing any short-term or long-term
167
complications. This absence makes sense given the relatively minor potential impact that cement can
Prophylactic Vertebroplasty During Spine Reconstruction 8 168
have if it remains in perivertebral veins and doesn’t cause spinal canal or foraminal stenosis. However,
169
26.5% of all patients in this study were found to have 1 or more grade 4 or 5 (clinically worrisome)
170
extravasation events. Although none of these patients went on to develop overt long-term clinical
171
sequelae from these events, the potential for devastating complications in similarly severe episodes of
172
cement extravasation has been well reported.[19-38] We furthermore cannot rule out the possibility
173
that these patients have suffered long-term subclinical sequelae from these high grade extravasation
174
events as cardiopulmonary functional studies were not routinely completed postoperatively.
175
It is important to consider these findings on the potential risk of using prophylactic cement
176
vertebroplasty within the context of the known potential risk of PJK. Failure of the spine at the proximal
177
and distal junctions of long-segment spinal fusions is also a widely known risk with potentially
178
devastating consequences. Although prophylactic cement vertebroplasty clearly comes with an
179
additional risk, larger prospective studies may discover that the actual risk of prophylactic
180
vertebroplasty is offset by the improved outcomes of reduced PJK risk.
181
Interestingly, the risk of high grade cement extravasation was highest in the upper thoracic
182
spine. Of the 32 upper thoracic VBs injected in this study, 8 VBs (25%) had CT evidence of high grade
183
extravasation. This is compared to 5 out of 80 (6.3%) lower thoracic-lumbar VBs with high grade
184
extravasation. There are several reasons why upper thoracic VBs may have a higher risk for cement
185
injection. First, it is more difficult to visualize these VBs on a lateral fluoroscopic view during an open
186
spinal reconstruction procedure, making it more difficult to identify when cement may be extravasating
187
outside the VB. Secondly, there is a shorter distance in the upper thoracic spine for a cement embolus to
188
travel from the VB to the vena cava, heart, and lungs. Finally, upper thoracic VBs are smaller and may
189
therefore tolerate a smaller amount of cement. These factors, and potentially others, likely contributed
190
to the higher association between clinically worrisome cement extravasation and upper thoracic VBs.
191
Because of these findings, the surgeons involved in this study have transitioned from performing
Prophylactic Vertebroplasty During Spine Reconstruction 9 192
prophylactic vertebroplasty in the upper thoracic spine for PJK prevention, to using transverse process
193
hooks at the top 2 upper instrumented vertebrae.
194
It is also important to note the lack of association between extravasation grade and the use of
195
fenestrated screws or Jamshidi needles. One might predict that given the larger bore size of the Jamshidi
196
needle, cement extravasation might be higher in this subset as a larger bolus of cement can be delivered
197
more quickly. Although we did find that the only VBs with no cement extravasation (grade 0) were
198
cemented through fenestrated pedicle screws, we found overall no significant difference in the rate of
199
high grade extravasation between fenestrated screws and Jamshidi needles. Perhaps the smaller bore of
200
the fenestrated pedicle screws produces smaller cement fragments that have an easier time entering
201
the peri-vertebral veins. Larger studies will need to be conducted to determine if there truly is no
202
difference in the extravasation risk between fenestrated pedicle screws and Jamshidi needles.
203
Finally, the results of this study have motivated a new effort by the authors to study in detail the
204
numerous variables that are thought to contribute to cement extravasation risk. For example, cement
205
viscosity has been identified in the literature as a contributing factor to cement extravasation risk.[44]
206
Bone mineral density (BMD), rate of cement injection, volume of cement injection, presence or absence
207
of vertebral body fracture or tumor, and level of cement injection also likely affect the risk of cement
208
extravasation. It is our practice to wait until the bone cement has achieved a subjectively high viscosity
209
before injecting it into the vertebral body, but we utilize no standard waiting period nor objective
210
measure of viscosity, and therefore were unable to evaluate the impact of cement viscosity on this
211
patient population. We also considered evaluating the impact of bone mineral density on risk of cement
212
extravasation in our study population, as more osteoporotic bone may be more prone to permitting
213
cement extravasation. Preoperative DEXA or CT scans were not included in our study inclusion criteria.
214
Postoperative CT scans are furthermore inadequate for evaluating BMD (via Hounsfield unit
215
measurement) because the presence of spinal hardware and cement affects the relative radiodensity of
Prophylactic Vertebroplasty During Spine Reconstruction 10 216
surrounding VBs. We were therefore unable to evaluate bone mineral density as a predictive variable for
217
cement extravasation risk in this study population. Finally, underlying condition was difficult to evaluate
218
for impact on extravasation risk as our patient population had a high amount of heterogeneity in
219
number of levels fused, number of levels cemented, spinal levels treated, and underlying condition
220
indicating spinal reconstruction. As such, we felt that any attempt to reach conclusions about the impact
221
of underlying condition on extravasation risk would be severely limited.
222
Limitations
223
This is a retrospective study and all the limits of retrospective radiographic reviews are therefore
224
relevant to these results. The study is also relatively small, limiting the power of the study to determine
225
lack of significant differences between study groups. Although the studied patient population is small at
226
34 patients, these patients comprised 112 VBs, making us more confident in the results. Finally, the
227
results are likely dependent on the cement injection technique used, and other surgeons may find
228
different results if their vertebroplasty technique significantly differs from that described herein. For
229
example, it is possible that usage of biplanar fluoroscopy during cement injection could minimize lateral
230
cement extravasation into the perivertebral venous plexus, thereby reducing the risk of high grade
231
extravasation events to larger, more distal venous structures and end-organs. It is also likely that by
232
changing the volume of cement injected at each VB, one could significantly change the extravasation
233
risk. At our institution, we have limited the volume of cement injection to 2 ml per pedicle as this seems
234
like an adequate volume of cement to fill the vertebral body. It is possible that 1.5 ml or 1.0 ml of
235
cement per pedicle would provide sufficient biomechanical support while reducing the risk of
236
extravasation. No biomechanical studies have yet provided the data needed, however, to make this
237
determination.
238
Finally, postoperative CT scans were not dedicated chest and abdomen views, and it is therefore
239
possible that distant cement emboli were missed on the postoperative CT scans obtained, especially for
Prophylactic Vertebroplasty During Spine Reconstruction 11 240
lumbar levels. We attempted to minimize the potential bias this might introduce into the data by
241
considering the inferior vena cava an end-organ and including this structure in our grade 4
242
extravasations. Nevertheless, it is possible that cement could have traveled from a vertebral body all the
243
way to the heart or lungs without leaving a trace behind in the perivertebral venous plexus or inferior
244
vena cava. This could potentially skew the data toward identifying more emboli for thoracic levels, as
245
thoracic CT scans include views of the heart and lungs.
246 247
Conclusions
248
Prophylactic cement vertebroplasty is an increasingly common technique employed in spinal
249
reconstruction procedures for the avoidance of common complications related to bone quality.
250
Although no patients in this study suffered known long-term consequences of this technique, 26.5% of
251
patients had cement extravasation that threatened end-organs or neural elements. These results should
252
serve as a reminder to those performing this technique that cement injection comes with very real risks,
253
and should therefore only be performed when the potential benefit of use is thought to outweigh the
254
potential risks. Based on our results, upper thoracic vertebrae come with an elevated risk of high grade
255
cement extravasation. These results prompted our transition to using transverse process hooks in the
256
upper thoracic spine for PJK prevention. Larger prospective studies will be needed to verify these
257
results.
FIGURE LEGEND Figure 1: Computed tomograms demonstrating examples of different grades of cement extravasation. From left to right and top to bottom, grade 0 extravasation with all the cement contained in the index vertebral body, grade 1 with the extravasated cement less than 5 mm from the index vertebral body, grade 2 with extravasated cement greater than 5 mm from the index vertebral body but not extending above or below the index level, grade 3 with extravasated cement extending 1 or more levels above or below the index vertebral body, grade 4 with extravasated cement in the vena cava or, in this case, the heart and lungs, grade 5 extravasation with cement causing 20% or greater stenosis of the spinal canal. Figure 2: A stacked bar chart depicting the number of vertebral bodies at each cement extravasation grade by the cement injection technique, fenestrated pedicle screws or Jamshidi needles. Figure 3: A stacked bar chart depicting the number of injected vertebral bodies at each spinal level by the severity of cement extravasation, clinically worrisome or clinically non-worrisome. A dashed line is placed between T6 and T7 as this was the division point between upper thoracic and lower thoraciclumbar test groups for statistical comparisons. Figure 4: Postoperative standing lateral radiograph of the single patient who suffered proximal junctional kyphosis in this patient population.
Table 1: Known Complications Following Vertebroplasty Complications Reported Following Vertebroplasty Foraminal stenosis and radiculopathy Central canal stenosis and spinal cord injury Anterior spinal artery syndrome Intradural cement leakage Cement vascular emboli to inferior vena cava Cement emboli causing stroke Cardiopulmonary cement emboli Intra-atrial thrombus Cardiac perforation Tricuspid regurgitation Acute respiratory distress syndrome and respiratory failure Renal cement embolism Fatal pulmonary embolism
Table 2: Novel Grading System for Clinical Severity of Cement Extravasation Grade 0
Computed Tomogram Findings No cement is visible outside the cortical walls of the index vertebral body
1
Extravasated cement is within 5 mm of the nearest cortical wall of the index vertebral body Extravasated cement is greater than 5 mm from
2
nearest cortical wall of the index vertebral body, but does not extend a spinal level above or below the index level Extravasated cement extends 1 or more levels
3
above or below the index level, but does not reach the superior or inferior vena cava
4
Extravasated cement reaches the superior or inferior vena cava or an end-organ Dorsally extravasated cement causes spinal canal
5
stenosis, defined as reduction in spinal canal diameter of 20% or more
Abbreviations: mm, millimeters
Table 3: Surgical Pathology Indicating Spinal Reconstruction Pathology
Number (%) of Patients
Adult Idiopathic Scoliosis (AIS)
1 (2.9%)
Tumor
9 (26.5%)
Infection
2 (5.9%)
Adult Degenerative Deformity (ASD)
15 (44.1%)
Trauma
4 (11.8%)
Proximal Junctional Kyphosis or Failure
3 (8.8%)
Table 4: Results of Statistical Comparisons Patient Groups
Grades 0-3 (clinically
Spine Levels
P-value
C7-T6
T7-L5
24
75
8
5
non-worrisome) Grades 4-5 (clinically
0.009
worrisome)
Cement Injection Technique Fenestrated Pedicle
P-value
Jamshidi Needles
Screws Grades 0-3 (clinically
77
22
8
5
non-worrisome) Grades 4-5 (clinically
0.290
worrisome) Results of statistical comparisons between groups of cement extravasation severity and spinal levels or cement injection technique. Fisher’s exact tests for independence were performed to account for the low number of patients with grade 4 or 5 cement extravasation.
Table 5: Clinical Outcomes of Patients with Grade 4 or 5 Cement Extravasation Patient Number
Clinical Outcome Grade 4 extravasation at T12 and L2 with cement in inferior vena cava. Patient
1
suffered transient episode of expressive aphasia postoperatively. MR brain negative. Neurology attributed aphasia to delirium.
2
Grade 5 extravasation at T2 and T7. No neurological deficit postoperatively. Tumor progression and death 7 months postoperative.
3
Grade 4 extravasation at T9 with cement in superior vena cava. No pulmonary embolus on CT scan, no treatment required. Attempted T4-Pelvis aborted after cement injection and pedicle screw insertion at T3 andT4 with subsequent loss of motor evoked potentials. Postoperative CT scan
4
showed T3 cement in superior vena cava, and non-compressive T4 cement in spinal canal. Pedicle screws removed, patient awoke motor intact. Loss of evoked potentials attributed to medial pedicle screws at T4. Procedure successfully performed 9 days later. Attempted T4-Pelvis aborted after T3 and T4 vertebroplasty with subsequent oxygen desaturation to the low 80s. Patient awoke hemodynamically stable.
5
Postoperative CT scan showed cement in the superior vena cava, right atrium, right ventricle, bilateral pulmonary arteries, and distally in the lungs. No anticoagulation deemed necessary, procedure successfully performed 5 days later.
6
Routine postoperative CT scan shows cement in left main pulmonary artery and left lower lobe artery. No clinical manifestation, no treatment required. Grade 5 extravasation at T4 along with medially placed pedicle screw. Not
7
identified intraoperatively. No changes in evoked potentials. Patient awoke motor intact. No treatment required.
8
Grade 5 cement at T2 with no monitoring changes. Patient awoke in delirium. Slowly returned to motor baseline. Asymptomatic, no treatment required. Grade 5 cement extravasation at T3. Noted intraoperatively on fluoroscopy, but no
9
monitoring changes. Procedure completed, patient awoke at baseline motor strength. No treatment required.
Figure 1 Grade 0
Grade 3
Grade 1
Grade 2
Grade 4
Grade 5
Figure 2 Number of Vertebral Bodies at Each Grade of Extravasation by Cement Injection Technique Cement Extravasation Grade
5
3
4
2 5
3
3
26
2
9
20
1
4 22
0
9 0
9
0 5
10
15
20
25
Number of Vertebral Bodies Fenestrated Screw
Jamshidi Needle
30
35
40
Figure 3 Spinal Level By Extravasation Severity Number of Vertebral Bodies
16 14 12 10 8 6 4 2 0
C7
T1
T2
T3
T4
T5
T6
T7
T8
T9
T10 T11 T12
L1
L2
L3
L4
Spinal Level grade 0-3 extravasation (clinically mild)
grade 4-5 (clinically worrisome)
L5
Figure 4
ABBREVIATIONS ASD = Adult Spinal Deformity CT = Computed Tomography PJK = Proximal Junctional Kyphosis VB = Vertebral Body