The accuracy of intraoperative O-arm imaging in assessing thoracolumbar pedicle screws placement

INTRODUCTION

Pedicle screw fixation is a common procedure in spinal surgery and is often considered the gold standard, especially for its adequate fixation throughout the vertebra.^[1] Successful fixation requires a stable implant-bone interface. The proximity of neural structures limits the available bone surfaces for instrumentation. In addition, the bone’s size and shape influence the material that can be used. Larger pedicles allow for the placement of larger pedicle screws, increasing fixation stability and strength. It is crucial to consider this variation among patients. At present, modern imaging systems enable pre-operative measurements of the optimal length and thickness of the screw.^[2]

In recent decades, the placement of transpedicular screws continues to be associated with complications of mispositioning.^[3,4] While it may be asymptomatic in most cases, it can significantly damage neurovascular structures. With various conventional placement techniques, a defined incidence rate of misplacement in the thoracolumbosacral region ranges from 15% to 30%.^[5] This can be prevented with proper placement using navigation techniques (StealthStation™) and verification using an intraoperative 2D/3D imaging system (O-arm™).^[6] If this technology is not available, the surgeon verifies screw placement with an intraoperative 2D fluoroscopy technique or post-operative computed tomography (CT) scan when a freehand technique is performed.^[1] However, another procedure is required for screw repositioning if necessary.

There are multiple classifications for assessing proper pedicle screw placement, evaluating this in the axial plane toward medial and lateral directions, whether they are within or outside the pedicle or body. However, there is no consensus on which classification to use or which is the most suitable.^[7,8]

Recent research demonstrates that perforations exceeding 4 mm in the medial wall of the pedicle could lead to neurological injuries. However, cadaveric studies suggest that perforations <2 mm in the medial pedicle wall are safe. Therefore, most surgeons consider this a safe perforation zone.^[9]

The Zdichavsky classification^[10,11] is a tool utilized to determine screw placement, comprising six types. Type Ia retains half or more of the screw within the pedicle and half or more within the vertebral body. In contrast, type Ib has over half of the screw laterally outside the pedicle and half within the vertebral body. Type IIa maintains half or more of the screw within the pedicle and more than half protruding laterally from the vertebral body, and type IIb is characterized by having half or more of the screw’s thickness within the pedicle, with its tip crossing the midline of the vertebral body. Type IIIa depicts more than half of the screw outside laterally, both the pedicle and the vertebral body, whereas type IIIb exhibits more than half of the screw outside the pedicle through the medial cortex, with its tip crossing the midline of the vertebral body [Figure 1].^[10,11]

Close

Figure 1:

Zdichavsky et al.’s classification (adapted from Zdichavsky et al.) depicting grades Ia to IIIb with schematic examples of each grade displayed alongside. The white arrow shows the position of the screw.

Export to PPT

The objective of this study was to assess the level of agreement between intraoperative O-arm imaging after placing the screws and post-operative CT scan for evaluating the accurate placement of thoracolumbar transpedicular screws using the Zdichavsky classification.

MATERIALS AND METHODS

A cross-sectional, analytical, observational, and retrospective study was conducted involving patients undergoing spinal surgery by three senior spine surgeons (CA, JF, and AP) at the Neurosurgery Department from March 2021 to May 2022. Sample size calculation utilized the formula for proportion with an infinite population, considering an expected proportion of 80%, a confidence level of 95%, and a precision of 7. The statistical software EPIDAT (version 3.5) was utilized, resulting in a minimum sample size of 126 screws.

Only sex, age, and images from the required studies were collected from the patient’s clinical record. The study included patients aged 15 years and older, of any sex, with complete intraoperative O-Arm™ (O-arm, Medtronic, Minneapolis, Minnesota, USA) images of all placed screws and complete post-operative CT scans visualizing all screws, the surgeon’s criteria for performing an O-arm scan before placing the screws and post-operative CT scan was not assessed. Patients were excluded if O-arm and/or CT scan images exhibited artifacts or were not visualized correctly.

Intraoperative O-arm images after placing the screws and post-operative CT scan images were collected in December 2022 from the imaging registry for all patients meeting the inclusion criteria. Subsequently, the position of all screws was determined according to the Zdichavsky classification in both studies, conducted by a single observer. Zdichavsky’s classification was chosen, among others, for its widespread use and reproducibility in determining the correct placement of transpedicular screws.^[8] Subsequently, the data were transferred to a spreadsheet, and the corresponding statistical analysis was conducted using Epi Info™ software version 7.2. Descriptive analysis was performed using absolute and relative frequency measures. Summary and dispersion measures were utilized for quantitative variables, and the Kappa index was calculated to assess the agreement between the two imaging methods.

RESULTS

Images were collected from a total of 19 patients, comprising 131 transpedicular screws. Among these patients, 7 were male (36.84%) and 12 were female (63.16%). The age of the patients ranged from 28 to 77 years, with a mean age of 57 ± 14 years. The number of screws per patient ranged from 4 to 10, with three patients having 4 screws (15.79%), seven patients with 6 screws (31.58%), six patients with 8 screws (31.58%), one patient with 9 screws (5.26%), and two patients with 10 screws (10.52%), with a mean of 7 ± 2 screws.

Using the Zdichavsky classification system, the 131 screws were categorized with post-operative CT and intraoperative O-arm images randomly by an observer. In the post-operative CT, 123 screws were identified as Type Ia (93.89%), seven screws as Type Ib (5.34%), and one screw as Type IIb (0.76%) [Table 1]. In intraoperative O-Arm images, 125 screws were classified as Type Ia (95.42%), five as Type Ib (3.82%), and one screw as Type IIb (0.76%) [Table 1]. A kappa value of 0.55 was observed for the concordance analysis, indicating a considerable level of agreement, with a confidence interval of 23–87% and P = 0.00 for statistical significance [Table 2].

DISCUSSION

Innovative navigation techniques in spinal surgery have greatly contributed to minimizing the incidence of incorrect pedicle screw placement, especially when employed by experienced spinal surgeons.^[12] The superiority of these systems is particularly beneficial when applied to patients with anatomical variations and in complex surgeries.^[13]

This study compared two diagnostic imaging studies, evaluating the placement of transpedicular screws in an axial plane and categorizing them using the Zdichavsky classification. The Landis and Koch Kappa interpretation system was utilized for statistical evaluation.^[14] When comparing the two studies, a moderate agreement between them was demonstrated. However, the confidence interval does not fall within the high-reliability standards, as there is a discrepancy of 23%, which indicates a continuing difference in a small percentage.

Ghodasara et al. show that following spinal instrumentation surgery, indications for performing a CT scan include assessing the condition and placement of the implant, observing the development and status of bone fusion, and identifying any issues such as peri-implant osteolysis, hidden fractures visible only through imaging, and the presence of soft-tissue accumulations.^[15] Previously, it was shown that performing only post-operative CT in symptomatic patients is advisable to verify the screw position; however, this would entail an additional procedure for screw repositioning.^[16] In contrast, with intraoperative images displayed using the O-Arm, there is the opportunity to reposition the implant during the same procedure, thereby reducing surgical team and patient exposure to radiation and other surgery.^[17]

It is worth emphasizing that in both tests, a high percentage of correct placement was demonstrated in our sample due to the navigation technique using StealthStation™ and O-arm, as type Ia, considered appropriate placement, was observed in 93.89% on CT and 95.42% on O-arm images, which is consistent with literature reports.^[18,19]

The most notable discrepancy was observed between type Ia and Ib, as these are the two most stable types within the classification without involving a violation of the medial cortex or screw instability with a similar screw direction. Similarly, adequate agreement was observed in type IIb in both tests. In this type, there is a more pronounced screw angulation, and it crosses the midline of the vertebral body, violating the medial wall of the pedicle by <50% of the screw thickness. Therefore, it is demonstrated that in higher types of classification where incorrect screw placement, especially toward the medial wall, is evident, it is more likely to find agreement.

It was observed that there is greater agreement between the two imaging studies in types of the Zdichavsky classification where incorrect placement is more evident, such as violating pedicle walls or crossing the midline of the vertebral body.

CONCLUSION

As mentioned earlier, a follow-up study such as CT is important in symptomatic patients postoperatively and when the control X-ray shows screw malposition. However, it was demonstrated that intraoperative O-arm images can be used as a post-operative control for screw position with minimal risk of inaccuracies and half radiation dose than CT scan.

AUTHORS’ CONTRIBUTIONS

FA conceived of the presented idea. FA and CGM collected the data. CMT and FA did the statistical analysis. FA wrote the manuscript with support from CML and CMT. CA reviewed the paper and made corrections with support from FA. All authors have critically reviewed and approved the final draft and are responsible for the manuscript’s content and similarity index.