Accuracy of intraoperative computed tomography image-guided surgery in placing pedicle and pelvic screws for primary versus revision spine surgery

View More View Less
  • 1 Departments of Neurosurgery and
  • | 4 Orthopaedics, Cedars-Sinai Medical Center, Los Angeles;
  • | 5 Department of Neurosurgery, University of California Davis Medical Center, Sacramento, California;
  • | 3 Department of Surgery, Duke University Medical Center, Durham, North Carolina; and
  • | 2 University of Texas Medical School at Houston, Texas
Free access

Object

Revision spine surgery, which is challenging due to disrupted anatomy, poor fluoroscopic imaging, and altered tactile feedback, may benefit from CT image-guided surgery (CT-IGS). This study evaluates accuracy of CT-IGS–navigated screws in primary versus revision spine surgery.

Methods

Pedicle and pelvic screws placed with the O-arm in 28 primary (313 screws) and 33 revision (429 screws) cases in which institutional postoperative CT scans were available were retrospectively reviewed for placement accuracy. Screw accuracy was categorized as 1) good (< 1-mm pedicle breach in any direction or “in-out-in” thoracic screws through the lateral thoracic pedicle wall and in the costovertebral joint); 2) fair (1- to 3-mm breach); or 3) poor (> 3-mm breach).

Results

Use of CT-IGS resulted in high rates of good or fair screws for both primary (98.7%) and revision (98.6%) cases. Rates of good or fair screws were comparable for the following regions: C7–T3 at 100% (good or fair) in primary versus 100% (good or fair) in revision; T4–9 at 96.8% versus 100%; T10–L2 at 98.2% versus 99.3%; L3–5 at 100% versus 99.2%; and pelvis at 98.7% versus 98.6%, respectively. On the other hand, revision sacral screws had statistically significantly lower rates of good placement compared with primary (100% primary vs 80.6% revision, p = 0.027). Of these revision sacral screws, 11.1% had poor placement, with bicortical screws extending > 3 mm beyond the anterior cortex. Revision pelvic screws demonstrated the highest rate of fair placement (28%), with the mode of medial breach in all cases directed into the sacral-iliac joint.

Conclusions

In the cervical, thoracic, and lumbar spine, CT-IGS demonstrated comparable accuracy rates for both primary and revision spine surgery. Use of 3D imaging of the bony pedicle anatomy appears to be sufficient for the spine surgeon to overcome the difficulties associated with instrumentation in revision cases. Although the bony structures of sacral pedicles and pelvis are relatively larger, the complexity of local anatomy was not overcome with CT-IGS, and an increased trend toward inaccurate screw placement was demonstrated.

Abbreviations used in this paper:

DRF = dynamic reference frame; IGS = image-guided surgery.

Object

Revision spine surgery, which is challenging due to disrupted anatomy, poor fluoroscopic imaging, and altered tactile feedback, may benefit from CT image-guided surgery (CT-IGS). This study evaluates accuracy of CT-IGS–navigated screws in primary versus revision spine surgery.

Methods

Pedicle and pelvic screws placed with the O-arm in 28 primary (313 screws) and 33 revision (429 screws) cases in which institutional postoperative CT scans were available were retrospectively reviewed for placement accuracy. Screw accuracy was categorized as 1) good (< 1-mm pedicle breach in any direction or “in-out-in” thoracic screws through the lateral thoracic pedicle wall and in the costovertebral joint); 2) fair (1- to 3-mm breach); or 3) poor (> 3-mm breach).

Results

Use of CT-IGS resulted in high rates of good or fair screws for both primary (98.7%) and revision (98.6%) cases. Rates of good or fair screws were comparable for the following regions: C7–T3 at 100% (good or fair) in primary versus 100% (good or fair) in revision; T4–9 at 96.8% versus 100%; T10–L2 at 98.2% versus 99.3%; L3–5 at 100% versus 99.2%; and pelvis at 98.7% versus 98.6%, respectively. On the other hand, revision sacral screws had statistically significantly lower rates of good placement compared with primary (100% primary vs 80.6% revision, p = 0.027). Of these revision sacral screws, 11.1% had poor placement, with bicortical screws extending > 3 mm beyond the anterior cortex. Revision pelvic screws demonstrated the highest rate of fair placement (28%), with the mode of medial breach in all cases directed into the sacral-iliac joint.

Conclusions

In the cervical, thoracic, and lumbar spine, CT-IGS demonstrated comparable accuracy rates for both primary and revision spine surgery. Use of 3D imaging of the bony pedicle anatomy appears to be sufficient for the spine surgeon to overcome the difficulties associated with instrumentation in revision cases. Although the bony structures of sacral pedicles and pelvis are relatively larger, the complexity of local anatomy was not overcome with CT-IGS, and an increased trend toward inaccurate screw placement was demonstrated.

Instrumentation is commonly used during spine surgery for stabilization and arthrodesis. The accuracy of instrumentation insertion is of paramount importance to reducing unintended bone, soft tissue, or neurological injuries, and the costs associated with postoperative morbidity, and even mortality. Accuracy rates with current techniques of pedicle instrumentation in which traditional fluoroscopic imaging is used have been previously reported in the literature to be as high as 5%–15%.2,5

Over the last 2 decades, intraoperative neuronavigation, particularly computed tomography image–guided surgery (CT-IGS), has emerged as an alternative to fluoroscopy-based techniques. The argument for CT-IGS has been most compelling in cases of brain and spinal cord tumors, trauma, complex deformity (acquired or congenital), obesity, osteoporosis, and revision surgery. In each of these instances, anatomy is significantly altered and difficulties are compounded by limitations in imaging visualization of bony landmarks. This study focuses on evaluating the performance of CT-IGS in the setting of primary versus revision spine surgery as it relates to pedicle and pelvic screw insertion.

Methods

Using postoperative CT scans, we studied the accuracy of all (742) C7–S1 pedicle and pelvic screws placed between 2009 and 2012 at Cedars-Sinai Medical Center with the aid of the intraoperative navigation software (StealthStation; Medtronic, Inc.) on a computer system (S7 Treon; Medtronic, Inc.) and intraoperative CT scanner system (O-arm; Medtronic, Inc.). Included in the study were all patients who had 1) O-arm Stealth-guided posterior pedicle and pelvic screws from C-7 to pelvis, and 2) postoperative CT scan evaluations. No eligible patients were excluded.

Study screws were divided into 2 groups: 1) “primary” screws (313) were defined as those placed in 28 consecutive cases in which the patient had not had prior spine surgery, and 2) “revision” screws (429) were those placed in 33 consecutive revision cases in which relevant anatomy had been distorted by prior surgery. In both groups postoperative CT scans were used to evaluate screw accuracy and safety. None of the authors or study technicians have any financial relationship to the manufacturer of the navigation technology (Medtronic, Inc.) (Figs. 13).

Fig. 1.
Fig. 1.

Postoperative CT scan demonstrating right “poor” screw (> 3-mm lateral breach) and left “good” screw.

Fig. 2.
Fig. 2.

Postoperative CT scan demonstrating right “fair” screw (1- to 3-mm medial breach) and left “good” screw.

Fig. 3.
Fig. 3.

Postoperative CT scan demonstrating left “fair” (> 3-mm bicortical breach) S-1 screw.

Surgical Technique

All patients were positioned prone on a Jackson table with appropriate padding, followed by sterile preparation and draping. A longitudinal midline open surgical incision exposed the levels of interest. The Stealth optical dynamic reference frame (DRF) was then rigidly affixed in 1 of 3 ways: 1) to an exposed spinous process; 2) bicortically through the posterior iliac crest through a percutaneous incision; or 3) if neither a spinous process nor pelvic entry site was available, a custom-bent rod attachment with DRF to existing pedicle screws was used. The DRF was affixed rigidly between the camera and the working surgical level to provide a line of sight to both the DRF and navigated instruments. The DRF was placed out of contact with soft tissue to prevent DRF drift due to tissue pull.

The O-arm was then brought into the protected sterile field. Anterior-posterior and lateral fluoroscopic scout images were obtained for localization of the region of interest. The Stealth optical camera was placed with a line of sight to the DRF. Intraoperative ventilatory arrest was then obtained for the duration of scan acquisition. Next, an intraoperative scan was performed that allowed automated registration of images spanning approximately 4–6 vertebral levels. After removing the O-arm from the field, StealthStation neuronavigation software was then used for intraoperative planning of screw placement, which included selection of screw diameter, length, and trajectory based on specific pedicle diameter and relevant anatomy.

All instruments including probe, pneumatic drill, awl, pedicle finder probes, screw tap, and the pedicle screw itself were navigated. Specifically, the pedicle entry hole was predrilled under image guidance, with or without an awl. Screws were then inserted with the navigated pedicle finder probe and tap. Each tapped hole was manually probed with a ball-tipped probe for sidewall or deep breach. The screw was finally inserted with a navigated screwdriver. System accuracy was checked following each screw insertion by placing the navigated probe onto a known anatomical landmark and visually confirming accuracy with the image guidance system. If accuracy deviated by > 1 mm, the O-arm was brought into the field and a scan performed on the same region. Once all screws were placed for the scanned vertebral levels, the DRF was repositioned as necessary and the O-arm was brought in for a scan of the next levels to be addressed.

To reduce radiation to the patient, an intraoperative CT scan with the O-arm was not routinely obtained after screw insertion unless there was concern for malposition. Motor evoked potentials, somatosensory potentials, and electromyography were routinely monitored for patients with thoracic exposure. Patients were evaluated postoperatively for any signs of neurological deficit or clinical symptoms associated with a malpositioned screw. Postoperative CT scans were routinely obtained within 3 months of surgery for verification of fusion.

Assessment of Pedicle and Pelvis Screws

To reduce interobserver error, a single reviewer blinded to the surgical group and individual patient was used to evaluate all screws. Only institution-obtained postoperative axial and sagittal CT images were used to review screw placement, to reduce variance in scan quality.

All screws were found to lie within the pedicle or pelvis in the rostral-caudal plane. Each instrumented pedicle was evaluated for medial or lateral breach on axial views. Screw depth was evaluated on axial images. Violations were defined as medial, lateral, or bicortical, and the degree of violation was noted for each screw (classified into 3 categories: good, fair, or poor).

“Good” screws were defined as those that were biomechanically optimal and safe. For good screws, neither the tip nor screw violated the cortex of the corresponding pedicle or pelvis by > 1 mm. In certain instances in which pedicle morphology or diameter precluded a screw from being contained purely within the pedicle, extrapedicular screws inserted with an “in-out-in” trajectory through the pedicle-rib complex were allowed. The “in-out-in” trajectory traverses “in” through the cortex of the transverse process, “out” through the lateral thoracic pedicle wall and in the costovertebral joint, and back “in” through the vertebral body. This “in-out-in” trajectory is considered biomechanically optimal because it allows for larger and longer screws (up to 60 mm in the thoracic spine) with tricortical or quadricortical purchase, although it does result in deliberate pedicle violation. It is considered safe because there are no vital structures within the pedicle-rib complex.

“Fair” screws were defined as screws that were safe but biomechanically suboptimal. Fair screws violated the cortex of the corresponding pedicle or pelvis by > 1 mm but < 3 mm. These screws, too long by 1–3 mm, were biomechanically suboptimal because the tip extended longer than necessary, but not to a degree to irritate ventral neural or vascular elements. In these cases of 1–3 mm of sidewall breach, the shaft was still invariably contained within 50% of the central axis of bone, did not cross the midline of the vertebral body, and did not significantly endanger the spinal cord in the central canal or thoracic contents.

“Poor” screws were defined as screws in which the tip or thread violated the cortex of the corresponding pedicle or pelvis by > 3 mm. These violations had the potential for spinal cord or other vital tissue injury, or had biomechanically unsound purchase.

Levels were evaluated based on overall region, divided by differences in anatomy. Regions were categorized as C7–T3, T4–9, T10–L2, L3–5, S-1, and pelvis. Because both surgeons were right-handed, an evaluation of left and right screws was also included. Surgeons inserted the instrumentation on the ipsilateral side (for example, the surgeon standing on the patient's left side inserted the left screws).

Patient charts were reviewed for intraoperative complications including neurological signal changes, as well as postoperative follow-up. All relevant morbidity was documented. Screw repositioning was recorded.

Results

In total, 742 (313 primary, 429 revision) pedicle screws in 61 (28 primary, 33 revision) cases were studied. Screws were subdivided by anatomical segment: C7–T3 (22 primary, 29 revision); T4–9 (62 primary, 64 revision); T10–L2 (113 primary, 151 revision); L3–5 (86 primary, 131 revision); S-1 (22 primary, 36 revision); and pelvis (8 primary, 18 revision). There was 1 reoperation for repositioning of a screw at the right L-2 due to a > 3-mm lateral breach with concern for pain, but without other neurological sequelae, in a patient undergoing primary surgery. There were no other repeat operations to reposition screws, and no other major morbidity or mortality associated with surgery.

Rates of good (biomechanically optimal and safe) screws for both primary (94.2%) and revision (91.6%) cases were relatively high and not significantly different in aggregate (p = 0.17). This was particularly true for rates of good screws above the sacrum, including C7–T3 at 100% primary versus 100% revision; T4–9 at 85.5% versus 90.6% (p = 0.37); T10–L2 at 92.9% versus 90.1% (p = 0.42); and L3–5 at 100% versus 97.7% (p = 0.16), respectively.

Rates of good and fair screws combined (safe) were equivalent for both primary (98.7%) and revision (98.6%) cases in aggregate (p = 0.89). Rates of good or fair screws were comparable for all spine regions except sacrum: C7–T3 at 100% (good or fair) in primary versus 100% (good or fair) in revision; T4–9 at 96.8% versus 100%; T10–L2 at 98.2% versus 99.3%; L3–5 at 100% versus 99.2% (Tables 13).

TABLE 1:

Accuracy of screw placement in 28 patients with primary spine surgery—overall percentages as well as per side of the spine*

CategoryC7–T3T4–9T10–L2L3–5S-1PelvisTotalRtLt
total screws226211386228313155158
good screws
 w/in bone829827747207107100
 bicortical13699180552233
 EP11311000251015
 EP-bicortical053000862
fair screws
 lat000000000
 lat-bicortical000000000
 medial0660001266
 medial-bicortical010000110
 bicortical000000000
 pelvis-SIJ violation000001101
poor screws
 lat002000220
 medial020000211
 medial-bicortical000000000
 bicortical000000000
accuracy
 good (%)22 (100)53 (85.5)105 (92.9)86 (100)22 (100)7 (87.5)295 (94.2)145 (93.5)150 (94.9)
 fair (%)07 (11.3)6 (5.3)001 (12.5)14 (4.5)7 (4.5)7 (4.4)
 poor (%)02 (3.2)2 (1.8)0004 (1.3)3 (1.9)1 (0.6)

EP = extrapedicular; SIJ = sacroiliac joint.

TABLE 2:

Accuracy of screw placement in 33 patients with revision spine surgery—overall percentages as well as per side of the spine

CategoryC7–T3T4–9T10–L2L3–5S-1PelvisTotalRtLt
total screws29641511313618429214215
good screws
 w/in bone13461071071213298153145
 bicortical1061021170642638
 EP251500022814
 EP-bicortical414000972
fair screws
 lat005100624
 lat-bicortical005100624
 medial044000871
 medial-bicortical010000110
 bicortical010030413
 pelvis-SIJ violation000005532
poor screws
 lat001000110
 medial000000000
 medial-bicortical000000000
 bicortical000140532
accuracy
 good (%)29 (100)58 (90.6)136 (90.1)128 (97.7)29 (80.6)13 (72.2)393 (91.6)194 (90.7)199 (92.6)
 fair (%)06 (9.4)14 (9.3)2 (1.5)3 (8.3)5 (27.8)30 (7.0)16 (7.5)14 (6.5)
 poor (%)001 (0.6)1 (0.8)4 (11.1)06 (1.4)4 (1.9)2 (0.9)
TABLE 3:

Comparison of primary and revision percentages for good and safe screws in 61 patients who underwent spine surgery

CategoryC7–T3T4–9T10–L2L3–5S-1PelvisTotalRtLt
good
 primary100.0%85.5%92.9%100.0%100.0%87.5%94.2%93.5%94.9%
 revision100.0%90.6%90.1%97.7%80.6%72.2%91.6%91.1%92.1%
safe (good + fair)
 primary100.0%96.8%98.2%100.0%100.0%98.7%98.7%98.1%99.4%
 revision100.0%100.0%99.3%99.2%88.9%98.6%98.6%98.6%98.6%

The rate of good S-1 screws was significantly lower in revision (100% primary vs 80.6% revision, p = 0.027). Of the 36 S-1 revision screws, 29 were considered good, 3 (8.3%) fair, and 4 (11.1%) poor. Revision sacral screws had the highest rate of poor placement (0% primary vs 11.1% revision, p = 0.107). Each of the screws categorized as fair and poor in revision cases were due to longer length rather than sidewall breach.

Rates of good pelvic screws trended toward better in primary than in revision cases, but were not significantly different (87.5% primary vs 72.2% revision, p = 0.393). Each of the fair pelvic screws (27.8% revision vs 12.5% primary) was classified as fair due to medial breach into the sacral-iliac joint. There were no poor pelvic screws placed in either primary or revision cases.

There were a total of 4 (1.3%) poor screws in primary cases. Two screws (in separate patients) had a medial breach of > 3 mm between T-4 and T-9 (1 right T-6, 1 left T-6) without neurological sequelae, and were left in place. Two screws (in separate patients) had a lateral breach of > 3 mm between T-10 and L-2 (1 right T-10, 1 right L-2). The right L-2 screw was ultimately revised because of pain.

There were a total of 6 poor screws in revision (1.4%). One screw had a lateral breach of > 3 mm from T-10 to L-2 (right L-1) without sequelae. One screw had a tip breach of > 3 mm from L-3 to L-5 (right L-5) without sequelae. Four screws had a tip breach of > 3 mm beyond the sacral promontory at S-1 (2 right S-1, 2 left S-1) without sequelae.

In aggregate, there was no significant difference in screws placed into the right or left pedicle for the right-handed surgeons. In total, good screws in the right and left pedicle were comparable for both primary (93.5% and 94.9%, respectively) and revision (90.7% and 92.6%, respectively) surgeries. Poor right and left screws were equally comparable for both primary (1.9% and 0.6%, respectively) and revision (1.9% and 0.9%, respectively) cases, suggesting that the surgeon's handedness does not affect placement.

Discussion

Fluoroscopy has long been used in spine surgery. The literature contains numerous articles describing many fluoroscopic images as less than ideal and/or difficult to obtain. These include cases in which normal anatomy is distorted; cases of tumor with both lytic and blastic components within the bone; trauma cases due to fracture and soft-tissue aberrancy; patients with deformity (both congenital and acquired) in which remodeled bone can cause small and easily violated pedicles; obese patients; patients with osteoporosis, which entails loss of bone resolution on fluoroscopy; and revision surgery in patients in whom the anatomy has been altered.5 Even in patients without these special problems, certain regions of the spine and the nearby anatomy (subaxial cervical pedicle, high thoracic spine, pelvis) provide a challenge for fluoroscopy due to the 2D nature of the images generated. Over the last 2 decades, intraoperative neuronavigation has emerged as an alternative to fluoroscopy-based techniques and CT-IGS has been used for screw insertion by many spine surgeons, with good results.11

In the treatment of spinal deformity, Larson et al.8 noted that only 1 of 142 screws placed with O-arm navigation required revision intraoperatively due to malposition (99.3% screw accuracy rate), versus the 94.9% accuracy rate reported for nonnavigated screws in all children undergoing pedicle screw fixation.

There are patients in whom initial surgical intervention fails and who require revision surgery. These spine revision cases present special challenges due to the fact that the operative site is compromised.3,7,9 No matter what caused the failure, subsequent surgery is complicated by the presence of scar tissue, which can cause compression on nerve roots and complicate a surgical approach. Furthermore, scar tissue is often heavily vascularized, causing higher estimated blood loss intraoperatively, and this leads to an obstructed surgical view or a postoperative hematoma.6 In addition, varied structural compromise of previously instrumented and/or fused vertebrae may make same-site instrumentation risky.9 Patients who fail to improve secondary to pseudarthroses, for example, may require more extensive instrumentation at adjacent segments.3,9

The initial step in planning revision spine surgery is identifying the root cause of the failure.1,3,9 Potential causes could be organic (adjacent-site disease, same-site disease, infection); technical (poor instrumentation, operator error, hardware failure); or a combination.3,9 Evaluation requires imaging.

Intraoperative imaging during the revision surgery is important for accurate localization of instrumentation and its relationship to vital neurovascular structures, the achievement of adequate decompression, and maintaining structural integrity within the spinal column. Currently, many centers routinely use the C-arm (fluoroscopy); however, fluoroscopy is limited in resolution and depth of field. Improved CT-IGS navigational functions of systems like the O-arm provide 2D and 3D resolution of structures in real time to ensure proper instrument placement and alignment.11 Neuronavigation reduces the risk that tactile feedback cues from previously violated or weakened bone or dense scarring will mislead the surgeon, and also provides the option of acquiring an intraoperative postinstrumentation CT image to confirm correct placement of screws.10 This option is exceedingly useful in an existing revision, because further revision would place the patient in added jeopardy.

Our retrospective study found that CT-IGS provided safe screw placement in both revision (98.6%) and primary (98.7%) spine surgeries. This was true in nearly all locations of the spine, with all levels exceeding 96% in safe placement (for example, a < 3-mm cortical breach in any dimension) in the cervical, thoracic, and lumbar spine. There were no statistical differences found for safely placed screws in revision or primary spine for each spine region evaluated (C7–T3, T4–9, T10–L2, and L3–5). Rates of poor placement were almost identical for both primary and revision screws. Of all screws placed, only 1 patient (primary) returned to surgery for a laterally misplaced L-2 screw due to pain without neurological morbidity.

The known limits of CT-IGS are well documented and include concerns of cost, time, and radiation exposure.8 Revision surgery, however, presents its own set of concerns. In cases in which the spinous process or other rigid bony structure is unavailable, it may become a challenge to find a purchase in which to adequately affix the reference frame. In our experience, there are several alternatives besides attachment of the reference frame to the spinous process. The more common solutions to cervical and high thoracic spine registration may be accomplished by attachment of the frame to a Mayfield head holder, whereas lumbar and sacral registration may be accomplished by frame attachment to a pelvic pin through the posterior-superior iliac spine. In cases of existing instrumentation, attaching the reference frame through domino instrumentation to a temporary rod affixed to pedicle screws can often provide good stabilization. Attempts to navigate an instrument that abuts dense scar can easily lead to false navigation, particularly if the navigated device bends during the process. Care must be taken to limit aggressive attempts to navigate against scar. Conversely, pseudarthroses or facet joints made incompetent by the initial surgery may provide undue laxity and false registration or inaccuracy. This ultimately requires reacquisition of the navigated images and thus additional radiation.

As with any tool, the success of CT-IGS is affected by the surgeon's judgment and intrinsic patient factors. For instance, although nearly all pedicle screws in the cervical, thoracic, and lumbar spine were placed safely (< 3-mm cortical violation), rates of good placement (either with < 1-mm medial breach or in a planned extrapedicular manner) were significantly lower in aggregate within the midthoracic T4–9 (85.5% primary vs 90.6% revision, p = 0.37) and thoracolumbar T10–L2 (92.9% vs 90.1%) regions compared with the high thoracic C7–T3 and low lumbar L3–5 (approaching 100% for revision and primary cases) regions.

This finding may be explained by thoracic pedicle anatomy that often precludes a completely intrapedicular screw placement, primarily due to either small pedicle diameter or instances in which the pedicle orientation is such that insertion of the screw would result in the tip breaching the vertebral body laterally. In such situations, our preferred option was to perform an extrapedicular approach due to anatomical necessity. White et al.12 found that transpedicular screws display a minimally greater mechanical stability than extrapedicular screws, but that the latter remain a biomechanically sound alternative to transpedicular screws in cases in which intrapedicular screws are not an option.

The thoracic pedicle violations that changed the classification from good to fair within our study were almost uniformly due to medial breach, again with no differences between revision and primary surgeries. These represented cases in which the surgeon believed that intrapedicular placement was possible, although there was a narrow-diameter pedicle. These screws remained safe, with no compromise of neural elements.

Perhaps the most telling difference between revision and primary cases was the lower rate of good screws (100% primary vs 80.6% revision, p = 0.03) in sacral pedicle screw placement. This difference is probably because revision S-1 screws were specifically targeted for tricortical purchase at the S-1 promontory to maximize screw pullout strength; biomechanical optimization was particularly key in revision, to reduce further pseudarthrosis or hardware failure that may have been the cause for the revision surgery. A slightly longer than standard screw implantation was therefore specifically desired, thereby increasing the number of resultant fair and even poor screws in the S-1 pedicle. That the violations of the S-1 pedicle were uniformly due to distal breach (rather than medial or lateral) is not surprising given the wide sacral pedicle diameter.

Often, sacral fixation is insufficient due to long cantilever forces at the lumbosacral junction, and pelvic instrumentation becomes essential. However, given the long trajectory of optimal screw placement, traditional iliac screw placement results in significant morbidity due to poor fluoroscopic imaging of the pelvis or extensive dissection along the lateral iliac crest and sciatic notch. Therefore, accurate pelvic screw placement performed using CT-IGS represented a significant finding in our study.

We found that CT-IGS could be used to safely insert pelvic instrumentation in both primary (98.7%) and revision (98.6%) cases. There were more than twice as many pelvic screws placed in patients undergoing revision than in those with primary surgery (19 vs 8). This is not surprising, given that pelvic supplementation was necessary in cases of failure at the caudal end of a construct terminating in S-1. Rates of good pelvic screws trended toward better in primary surgeries than in revision, but were not significantly different (87.5% primary vs 72.2% revision, p = 0.393). We believe this was seen because revision pelvic screws were often mated to an existing construct. Therefore, the entry point and trajectory of the pelvic screw are more determined by the existing screw alignment. It is conceivable that a less than ideal pelvic screw position was chosen in these revision cases to best connect the lumbosacral instrumentation.

Sidewall breach of < 3 mm through the sacral-iliac joint was tolerated and classified as fair. Indeed, each of the fair pelvic screws (27.8% revision vs 12.5% primary) was tolerated by all patients without reports of sacral-iliac joint pain. The number of pelvic screws inserted in this series was low, and therefore our findings are preliminary at best. To our knowledge the case report by Garrido and Wood4 is the only other published study regarding navigated pelvic screws; in that case report of 10 such screws, they were able to identify ideal iliac bone screw placement projecting within 2 cm over the sciatic notch, between pelvic tables, with less surgical exposure and complete accuracy by using image navigation techniques with a percutaneous reference frame.

Limitations of the Study

There are several limits to our study. Revision spine surgery is a broad category and encompasses a wide range of procedures. For the purposes of this study, we limited the definition of revision surgery to operations specifically in the region of spine undergoing repeat operation, without distinguishing between the presence or lack of prior instrumentation. Isolating cases of prior instrumentation itself leads to certain difficulties, because prior placement of instrumentation biases the likelihood of new instrumentation finding the same trajectory. Another concern was our classification of screw accuracy. This study was retrospective and therefore reported on patients with available postoperative CT scans; patients who lacked postoperative imaging or who were lost to follow-up may have introduced a measurement bias. Finally, this study was purely an imaging-based study without a focus on functional outcome.

Conclusions

Advantages offered by CT-IGS seem to be shared between revision and primary surgeries in which cannulation of the pedicles is required. For screws placed in the larger bony structures (sacrum and pelvis) there is a trend toward greater inaccuracy in revision surgery compared with primary surgery.

Disclosure

Dr. Kim is a consultant for DePuy Synthes. The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author contributions to the study and manuscript preparation include the following. Conception and design: Kim, Johnson. Acquisition of data: Hsieh, Drazin, Pashman. Analysis and interpretation of data: Hsieh, Drazin. Drafting the article: Hsieh, Drazin, Firempong. Critically revising the article: Drazin. Reviewed submitted version of manuscript: Kim, Drazin, Pashman, Johnson. Statistical analysis: Hsieh. Study supervision: Kim, Johnson.

References

  • 1

    Bertagnoli R, , Zigler J, , Karg A, & Voigt S: Complications and strategies for revision surgery in total disc replacement. Orthop Clin North Am 36:389395, 2005

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Castro WH, , Halm H, , Jerosch J, , Malms J, , Steinbeck J, & Blasius S: Accuracy of pedicle screw placement in lumbar vertebrae. Spine (Phila Pa 1976) 21:13201324, 1996

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Eichholz KM, & Ryken TC: Complications of revision spinal surgery. Neurosurg Focus 15:3 E1, 2003

  • 4

    Garrido BJ, & Wood KE: Navigated placement of iliac bolts: description of a new technique. Spine J 11:331335, 2011

  • 5

    Hicks JM, , Singla A, , Shen FH, & Arlet V: Complications of pedicle screw fixation in scoliosis surgery: a systematic review. Spine (Phila Pa 1976) 35:E465E470, 2010

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Kim MC, , Chung HT, , Cho JL, , Kim DJ, & Chung NS: Factors affecting the accurate placement of percutaneous pedicle screws during minimally invasive transforaminal lumbar interbody fusion. Eur Spine J 20:16351643, 2011

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Kostuik JP: Complications and surgical revision for failed disc arthroplasty. Spine J 4:6 Suppl 289S291S, 2004

  • 8

    Larson AN, , Polly DW Jr, , Guidera KJ, , Mielke CH, , Santos ER, & Ledonio CG, et al.: The accuracy of navigation and 3D image-guided placement for the placement of pedicle screws in congenital spine deformity. J Pediatr Orthop 32:e23e29, 2012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Rihn JA, , Harrod C, & Albert TJ: Revision cervical spine surgery. Orthop Clin North Am 43:123136, ixx, 2012

  • 10

    Santos ER, , Ledonio CG, , Castro CA, , Truong WH, & Sembrano JN: The accuracy of intraoperative O-arm images for the assessment of pedicle screw postion. Spine (Phila Pa 1976) 37:E119E125, 2012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Sembrano JN, , Santos ER, & Polly DW Jr: New generation intraoperative three-dimensional imaging (O-arm) in 100 spine surgeries: Does it change the surgical procedure?. J Clin Neurosci 21:225231, 2014

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    White KK, , Oka R, , Mahar AT, , Lowry A, & Garfin SR: Pullout strength of thoracic pedicle screw instrumentation: comparison of the transpedicular and extrapedicular techniques. Spine (Phila Pa 1976) 31:E355E358, 2006

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Postoperative CT scan demonstrating right “poor” screw (> 3-mm lateral breach) and left “good” screw.

  • View in gallery

    Postoperative CT scan demonstrating right “fair” screw (1- to 3-mm medial breach) and left “good” screw.

  • View in gallery

    Postoperative CT scan demonstrating left “fair” (> 3-mm bicortical breach) S-1 screw.

  • 1

    Bertagnoli R, , Zigler J, , Karg A, & Voigt S: Complications and strategies for revision surgery in total disc replacement. Orthop Clin North Am 36:389395, 2005

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Castro WH, , Halm H, , Jerosch J, , Malms J, , Steinbeck J, & Blasius S: Accuracy of pedicle screw placement in lumbar vertebrae. Spine (Phila Pa 1976) 21:13201324, 1996

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Eichholz KM, & Ryken TC: Complications of revision spinal surgery. Neurosurg Focus 15:3 E1, 2003

  • 4

    Garrido BJ, & Wood KE: Navigated placement of iliac bolts: description of a new technique. Spine J 11:331335, 2011

  • 5

    Hicks JM, , Singla A, , Shen FH, & Arlet V: Complications of pedicle screw fixation in scoliosis surgery: a systematic review. Spine (Phila Pa 1976) 35:E465E470, 2010

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Kim MC, , Chung HT, , Cho JL, , Kim DJ, & Chung NS: Factors affecting the accurate placement of percutaneous pedicle screws during minimally invasive transforaminal lumbar interbody fusion. Eur Spine J 20:16351643, 2011

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Kostuik JP: Complications and surgical revision for failed disc arthroplasty. Spine J 4:6 Suppl 289S291S, 2004

  • 8

    Larson AN, , Polly DW Jr, , Guidera KJ, , Mielke CH, , Santos ER, & Ledonio CG, et al.: The accuracy of navigation and 3D image-guided placement for the placement of pedicle screws in congenital spine deformity. J Pediatr Orthop 32:e23e29, 2012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Rihn JA, , Harrod C, & Albert TJ: Revision cervical spine surgery. Orthop Clin North Am 43:123136, ixx, 2012

  • 10

    Santos ER, , Ledonio CG, , Castro CA, , Truong WH, & Sembrano JN: The accuracy of intraoperative O-arm images for the assessment of pedicle screw postion. Spine (Phila Pa 1976) 37:E119E125, 2012

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Sembrano JN, , Santos ER, & Polly DW Jr: New generation intraoperative three-dimensional imaging (O-arm) in 100 spine surgeries: Does it change the surgical procedure?. J Clin Neurosci 21:225231, 2014

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    White KK, , Oka R, , Mahar AT, , Lowry A, & Garfin SR: Pullout strength of thoracic pedicle screw instrumentation: comparison of the transpedicular and extrapedicular techniques. Spine (Phila Pa 1976) 31:E355E358, 2006

    • Crossref
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 948 61 9
PDF Downloads 443 87 11
EPUB Downloads 0 0 0