Prediction of Screw Loosening After Dynamic Pedicle Screw Fixation With Lumbar Polyetheretherketone Rods Using Magnetic Resonance Imaging-Based Vertebral Bone Quality Score
Article information
Abstract
Objective
To investigate the correlation between magnetic resonance imaging-based vertebral bone quality (VBQ) score and screw loosening after dynamic pedicle screw fixation with polyetheretherketone (PEEK) rods, and evaluate its predictive value.
Methods
A retrospective analysis was conducted on the patients who underwent dynamic pedicle screw fixation with PEEK rods from March 2017 to June 2022. Data on age, sex, body mass index, hypertension, diabetes, hyperlipidemia history, long-term smoking, alcohol consumption, VBQ score, L1–4 average Hounsfield unit (HU) value, surgical fixation length, and the lowest instrumented vertebra were collected. Logistic regression analysis was employed to assess the relationship between VBQ score and pedicle screw loosening (PSL).
Results
A total of 24 patients experienced PSL after surgery (20.5%). PSL group and non-PSL group showed statistical differences in age, number of fixed segments, fixation to the sacrum, L1–4 average HU value, and VBQ score (p < 0.05). The VBQ score in the PSL group was higher than that in the non-PSL group (3.56 ± 0.45 vs. 2.77 ± 0.31, p < 0.001). In logistic regression analysis, VBQ score (odds ratio, 3.425; 95% confidence interval, 1.552–8.279) were identified as independent risk factors for screw loosening. The area under the receiver operating characteristic curve for VBQ score predicting PSL was 0.819 (p < 0.05), with the optimal threshold of 3.15 (sensitivity, 83.1%; specificity, 80.5%).
Conclusion
The VBQ score can independently predict postoperative screw loosening in patients undergoing lumbar dynamic pedicle screw fixation with PEEK rods, and its predictive value is comparable to HU value.
INTRODUCTION
The titanium rod-assisted lumbar fusion surgery, combined with pedicle screws, is currently considered the “gold standard” for treating degenerative diseases of the lumbar spine [1,2]. However, after lumbar interbody fusion, increased mobility and abnormal stress in adjacent segments ultimately lead to adjacent segment degeneration (ASD) [3,4]. With the advancement of biomechanical research and material development, various instruments and surgical techniques that preserve mobility have emerged clinically, such as the Dynesys system [5,6], Isobar TTL system [3,7], and polyetheretherketone (PEEK) rods system [8,9]. These dynamic fixation technologies aim to prevent or slow down the occurrence of ASD.
Postoperative screws in dynamic fixation are subjected to long-term stress at the bone interface, and pedicle screw loosening (PSL) is one of the most common complications following dynamic fixation, especially in patients with osteoporosis [10-12]. A precise preoperative assessment of the patient’s bone quality is clinically significant for identifying the critical threshold range of vertebral bone quality associated with dynamic fixation-related PSL.
Although dual-energy x-ray absorptiometry (DEXA) is currently recognized as the gold standard for diagnosing osteoporosis, it still has certain limitations [13]. Recently, the VBQ score based on magnetic resonance imaging (MRI) has been proposed as a more efficient and accurate method for detecting osteoporosis [14]. The latest research indicates that the VBQ score is an independent risk factor for PSL after interbody fusion and lumbar rigid fixation surgery [13,15,16].
Currently, there is no research assessing the correlation between VBQ scores and postoperative PSL in dynamic fixation with the PEEK rods system. This study retrospectively analyzes the correlation between dynamic fixation-related PSL and VBQ scores, aiming to explore the critical range of VBQ scores for predicting PSL.
MATERIALS AND METHODS
1. Patients
This study conducted a retrospective analysis of patients who underwent lumbar spine dynamic fixation surgery using the PEEK rods system due to degenerative diseases from March 2017 to June 2022. The study obtained approval from Institutional Review Board of Dongzhimen Hospital Affiliated to Beijing university of Chinese Medicine (2022DZMEC-085-04). As a retrospective study, informed consent was deemed unnecessary.
Inclusion criteria were as follows: (1) clinically symptomatic and radiologically diagnosed with degenerative diseases of the lumbar spine (such as lumbar disc herniation, lumbar spinal stenosis, degenerative spondylolisthesis or instability of the lumbar spine), with ineffective conservative treatment for over 6 months, or effective conservative treatment for over 3 months but experiencing recurrent symptoms severely impacting daily activities and normal work; (2) patients retaining a certain degree of mobility in the lumbar spine segments involved; (3) undergoing posterior lumbar surgery using the pedicle screw system with PEEK rods for dynamic fixation, with lumbar MRI performed in our hospital within one month before surgery, and lumbar computed tomography (CT) scan within 3 months before surgery.
Exclusion criteria included: (1) a history of previous lumbar spine surgery; (2) lumbar spondylolysis, structural retroversion, severe intervertebral space narrowing; (3) severe osteoporosis, coronal imbalance of the spine, lateral displacement of the vertebral body > 1 cm, and coronal vertebral body wedge compression > 1/3; (4) patients with severe menopausal symptoms, mental disorders.
2. Evaluation of Clinical and Radiological Parameters
All Individual data recorded patient age, sex, body mass index (kg/m2), history of hypertension, diabetes, hyperlipidemia, and long-term smoking and drinking. Additionally, VBQ score and L1–4 average HU values were documented. Surgical data included the length of surgical fixation (number of segments) and the lowest instrumented vertebra (LIV).
3. Calculation of the VBQ Score
Following the method proposed by Ehresman et al. [17], VBQ scoring was conducted using T1-weighted images from lumbar spine MRI. On the midsagittal plane of the lumbar spine, the region of interest (ROI) was placed between the trabecular part of the L1–4 vertebral bodies and the cerebrospinal fluid (CSF) gap at the level of L3. The average signal intensity (SI) within each ROI of the vertebral bodies and the average SI of CSF at the L3 level were recorded (Fig. 1). The VBQ score was calculated as the median SI of L1–4 vertebral bodies divided by the average SI of CSF [VBQ score = median (SIL1-4)/SICSF]. In cases where the midsagittal plane could not accurately measure the ROI (vascular tumors, changes in spinal lateral curvature), a parasagittal plane was used as a substitute for the midsagittal plane ROI. If the entire vertebral body could not be measured, that vertebra was excluded. If the CSF at the L3 level was completely obstructed, the ROI for CSF was placed at the L2 or L4 level.
In the context of T1-weighted images, vertebral bone quality (VBQ) scores are computed through the delineation of regions of interest. The VBQ score shown in this example is 2.97. VBQ indicates vertebral bone quality.
4. Measurement of HU Value
Three transverse sections parallel to the upper and lower endplates and the midhorizontal plane of the vertebral body were selected from CT images. ROI were delineated within the trabecular bone, avoiding cortical bone, areas of local bone hyperplasia, and the vertebral venous plexus. The HU value within the ROI was then obtained (Fig. 2). The mean values from the 3 planes were calculated to represent the HU value of the vertebral body. The L1–4 average HU value was calculated as the average HU value of the L1–4 vertebrae.
Measure the Hounsfield unit value of the L3 vertebral body in computed tomography images.
5. Pedicle Screws Loosening
We conducted lumbar spine x-ray examinations (anterior-posterior, lateral, flexion-extension views) for follow-up patients to assess screw loosening. Screw loosening was defined as the presence of a radiolucent zone of 1 mm or more around the pedicle screw [18]. For suspected loosening cases, lumbar spine CT scans were performed, and in the coronal reconstruction images with bone window settings, the presence of a ring-shaped low-density shadow around the screw was considered indicative of screw loosening [19]. Based on the follow-up results, patients were categorized into the PSL group and the non-PSL group.
Two experienced surgeons independently measured bone quality parameters (HU score, VBQ score) and extracted other clinical data. To assess reliability, 1 month later, each observer remeasured the same 30 cases.
6. Statistical Analysis
All data were statistically analyzed using IBM SPSS Statistics ver. 25.0 (IBM Co., Armonk, NY, USA). The intraclass correlation coefficient (ICC) was employed to assess intraobserver and interobserver consistency. All continuous variables are presented as mean± standard deviation and analyzed using Student t-test. Categorical variables are reported as frequencies and percentages, and evaluated using the chi-square test. Logistic regression analysis was employed to identify independent risk factors for screw loosening. Receiver operating characteristic (ROC) curve analysis was utilized to assess the predictive value of VBQ scores for PSL, and the Youden index was used to establish the optimal threshold for VBQ scores. A p-value < 0.05 was considered significant.
RESULTS
This study included 117 patients undergoing dynamic fixation with PEEK rods for lumbar spine diseases, including 69 males and 48 females, with an average age of 59.4± 9.1 years (43–71 years). The follow-up period was 37.5± 11.4 months (14–55 months). All patients were divided into the PSL group (24 cases, 20.5%) and the non-PSL group (93 cases, 79.5%). There were no revision surgeries required for either group during the 12 months postoperatively.
1. Demographics and Operative Profiles
The age, range of fixation, LIV at S1, VBQ score, and HU value exhibited statistically significant differences between the PSL group and the non-PSL group (p< 0.05). Additionally, patients in the PSL group had lower HU values (p < 0.001) and higher VBQ scores (p< 0.001) than those in the non-PSL group (Table 1).
Comparison of patient characteristics between the PSL and non-PSL groups
2. Prediction of PSL Using VBQ Scores
Variables with p< 0.05 in Table 1 were considered potential risk factors influencing screw loosening and were included in the multivariate logistic regression analysis (Table 2).
Logistic regression analysis of factors influencing PSL after lumbar dynamic fixation surgery
The results indicated that fixation length (odds ratio [OR], 3.749; 95% confidence interval [CI], 1.312–9.136), LIV at S1 (OR, 3.338; 95% CI, 1.655–7.091), average vertebral CT value of L1–4 (OR, 1.031; 95% CI, 1.014–1.263), and VBQ score (OR, 3.425; 95% CI, 1.552–8.279) were independent risk factors for screw loosening; age was not a predictive factor for screw loosening. Furthermore, there was a significant negative correlation between VBQ score and HU values (Fig. 3). The VBQ score and HU value was shown to have excellent interrater reliability with an ICC > 0.8.
Correlation between the vertebral bone quality (VBQ) score and mean Hounsfield unit (HU) value of L1–4.
3. Evaluation of Predictive Scoring Measures
The area under the ROC for VBQ scores and HU values in predicting screw loosening were 0.819 and 0.793, respectively (Fig. 4). The optimal threshold for predicting screw loosening was determined by the maximum Youden index, and it was found to be 3.15 for VBQ scores (sensitivity, 0.831; specificity, 0.805).
Receiver operating characteristic (ROC) curve analysis was used to evaluate vertebral bone quality (VBQ) scores and the mean Hounsfield unit (HU) values of the L1–4 for the prediction of screw loosening.
DISCUSSION
PSL is a common complication after lumbar fixation surgery using the pedicle screw system, particularly with a significantly higher incidence in patients with osteoporosis [20-22]. Studies have shown that the proportion of screw loosening in patients with normal bone density ranges from 1% to 27%, while in osteoporotic patients, the proportion can be as high as 60% [23,24]. Therefore, accurate bone quality assessment in patients undergoing lumbar fixation with the pedicle screw system is crucial for guiding preoperative planning and selecting postoperative measures to mitigate bone loss.
The most commonly used tool in clinical practice for assessing bone mineral density (BMD) is DEXA [25]. However, in patients with degenerative lumbar diseases, factors such as osteophyte formation, facet joint degeneration, and aortic calcification can make DEXA less accurate in evaluating the trabecular bone BMD in the vertebral bodies. Additionally, not all patients undergo routine DEXA examinations before surgery [26]. Therefore, there is a need for a convenient and efficient alternative to DEXA for bone quality assessment.
Previous research has found that CT-HU value can be used for bone quality assessment and have good predictive value for PSL after lumbar rigid fixation surgery [27]. A recently developed MRI-based method predicts PSL, offering a new tool for evaluating PSL after lumbar rigid fixation fusion surgery, alongside HU values [28].
Histological analysis reveals that osteoporotic bone is typically characterized by the replacement of local fat cells and the predominant feature of trabecular bone atrophy [29]. Additionally, an increase in fat infiltration and excessive trabecular bone atrophy can result in higher SI on T1-weighted MRI images. Based on this theory, Ehresman et al. [30] first proposed the VBQ score, which introduces CSF signal to eliminate differences in individual patient and scanner baseline signals compared to previous MRI-based scores. Due to its simplicity, feasibility, and resistance to confounding factors, VBQ score has rapidly become a practical tool for assessing vertebral bone quality in clinical settings. Previous studies have indicated that VBQ score serves as an independent predictor for PSL after lumbar rigid fixation surgery [13,28,31].
Currently, instrumentation for posterior lumbar fixation via the pedicle screw system mainly comprises 2 categories: dynamic rods, which allow for some degree of motion, and titanium rods, which offer minimal to no motion. PEEK rods belong to the category of instruments for dynamic lumbar fixation and are applied in nonfusion lumbar surgeries in this study, whereas titanium rods are commonly utilized in fusion lumbar surgeries. These 2 types of instrumentation are guided by distinct core principles, resulting in variations in screw loosening rates. Titanium rods prioritize robust fixation of the operative segment to promote fusion, while dynamic rods aim to preserve lumbar mobility while ensuring segmental stability to mitigate complications arising from diminished lumbar mobility. Presently, consensus is lacking regarding the comparative postoperative screw loosening rates between dynamic and rigid fixation [32]. We underscore the importance of addressing postoperative screw loosening in dynamic fixation procedures. Following complete fusion of the operative segment in rigid fixation surgeries, the function of screws and titanium rods is essentially fulfilled. In contrast, dynamic rods continue to provide lumbar support and withstand sustained mechanical stress. Should screw loosening occur, it could significantly compromise long-term lumbar stability and potentially expedite degeneration of adjacent segments. Consequently, compared to fusion fixation, greater attention should be directed toward bone quality assessment in patients undergoing dynamic lumbar fixation surgeries.
To the best of our knowledge, this is the first study to assess the correlation between VBQ score and PSL after dynamic fixation surgery. In this study, PSL was identified using a combination of x-ray and CT methods, with the highest incidence occurring approximately 1 year postoperatively during follow-up. Interestingly, as the follow-up period lengthened, the rate of screw loosening gradually decreased, with some loosening patients showing gradual improvement and eventual nonloosening status, a phenomenon consistent with a recent report by Shu et al. [33].
Patients in the PSL group were older, had more fixed segments, had a higher proportion of fixation to S1, lower HU values of L1–4 vertebral bodies, and significantly higher VBQ scores. Multivariable logistic regression analysis indicated that the number of fixed segments, LIV at S1, HU values of L1–4 vertebral bodies, and VBQ scores were independent risk factors for screw loosening, while age was not an independent risk factor for screw loosening in this study, which may be related to the relatively small number of cases included. The predictive value of HU values and VBQ scores was relatively ideal. These results are generally consistent with previous studies on screw loosening after lumbar rigid fixation surgery [13,27,29,34].
The longer the fixed segment stabilized within the lumbar spine through pedicle screw fixation, the more likely the occurrence of PSL. In this study, the proportion of 3-segment and 4-segment PSL groups was 45.8%, significantly higher than the proportion in the non-PSL group, which was 19.4% (Fig. 5). This may be attributed to the increased stress on the screws at the proximal and distal ends with a greater number of fixed segments, leading to a higher likelihood of screw loosening [35].
Fixation segment distribution percentage in the pedicle screw loosening (PSL) group and non-PSL group.
The unique location and anatomical structure of the sacrum may contribute to a higher postoperative loosening rate of sacral screws. Literature reports the occurrence rate of sacral PSL to be in the range of 15.6%–46.5%, significantly higher than the loosening rate observed in lumbar PSL [36,37]. L5–S1, serving as the transitional zone between the lumbar spine and sacrum, not only bears a substantial portion of the body weight but also experiences increased shear forces at the lumbosacral angle, further intensifying local loads [38,39]. The uneven distribution of trabecular bone density within the sacrum poses challenges, with higher trabecular bone density near the anterior cortex region and lower density in the posterior cortex regions adjacent to the sacral wings [40]. Consequently, sacral screws often require sufficient depth of insertion to obtain an adequate grip. The first sacral segment exhibits a wider pedicle inner diameter and weaker cortical support for screw fixation. Therefore, identifying predictive indicators for sacral screw loosening after sacral fixation becomes particularly crucial.
Current research indicates that osteoporosis is one of the risk factors influencing screw loosening [41]. Bone quality is a primary determinant of stability for pedicle screws since it governs the strength of the screw-bone interface [42]. The HU values of lumbar vertebral bodies can accurately reflect the strength of trabecular bone density, demonstrating good practicality in assessing bone density. Some literature has confirmed that HU values can serve as one of the predictive factors for postoperative screw loosening in the lumbar spine [43,44].
While bone density is generally considered a component of skeletal strength, it cannot be the sole indicator of skeletal mechanical performance. During the process of osteoporosis, local fat cells gradually replace trabecular bone, weakening the overall bone structure [45]. VBQ score reflects the extent of fat infiltration in vertebral trabecular bone, offering another perspective on bone quality. This study reveals that higher VBQ scores in most patients correlate with a higher occurrence rate of screw loosening. Additionally, there is a moderate negative correlation with HU values (p< 0.001), indicating that patients with higher VBQ scores tend to have correspondingly lower HU values. The predictive value of VBQ scores for screw loosening is generally comparable to HU values. Furthermore, this study suggests that VBQ score is consistently reliable, easy to operate, less influenced by confounding factors, and serves as a practical tool for auxiliary assessment of bone quality.
This study also has some limitations. Different field strengths (1.5/3.0 T) may have a significant impact on VBQ scores. The study of Lin et al. [46] showed that VBQ1.5T scores were more sensitive for identifying osteoporosis compared to VBQ3.0T scores. Considering these factors, all patient images included in this study were obtained under a 1.5 T magnetic field, which may introduce bias into the final results. All cases in this study were sourced from a single center, and the relatively limited number of cases may render some risk factors influencing PSL, as reported in previous studies, statistically insignificant in this study. Additionally, some patients in this study exhibited lower HU values ( < 50 HU) or higher VBQ scores ( > 3.5), but did not experience screw loosening during the postoperative follow-up. PSL is a result influenced by multiple factors, and the included influencing factors in this study are still not comprehensive enough.
CONCLUSION
The VBQ score can independently predict postoperative PSL in patients undergoing dynamic fixation with lumbar pedicle screws and PEEK rods, and its predictive value is comparable to HU values.
Notes
Conflict of Interest
The authors have nothing to disclose.
Funding/Support
This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Author Contribution
Conceptualization: GJ, XY; Data curation: GJ, WZ, JS, NF; Formal analysis: YM, WL, JS, NF, YZ; Methodology: JG, YY, ZQ; Project administration: LX, SZ, YQ, XY; Visualization: NF, ZL, LM; Writing – original draft: GJ, LX; Writing – review & editing: JG, XY.