INTRODUCTION
Osteoporosis has significant implications for lumbar fusion surgery, and reduced spinal bone mineral density (BMD) can result in a high risk for postoperative complications [1,2]. Thus, distinguishing poor bone quality before surgery has become increasingly important for spine surgeons to avoid devastating complications after degenerative lumbar surgery [1,3].
Currently, dual-energy x-ray absorptiometry (DEXA) is the gold standard for the radiographic diagnosis of osteoporosis [1,4]. However, the accuracy of DEXA is limited by the presence of superimposed calcified tissue and degenerative spinal pathology [2], and it has been reported that only 44% of surgeons would arrange preoperative DEXA for lumbar surgery, and DEXA screening rates could be as low as 27% [5]. In recent years, there has been increasing interest in alternative imaging techniques, including quantitative computed tomography (QCT) and magnetic resonance imaging (MRI) for assessing BMD [1,6,7]. QCT is more accurate method to measure the osteoporotic bone [8,9], however, the routine application of QCT is limited by its high cost and high risk of radiological hazards [10]. As MRI is routinely performed before modern lumbar surgery, the novel MRI-based vertebral bone quality (VBQ) score has emerged as an alternative for bone quality assessment [4].
Poor bone quality is associated with a high risk of postoperative complications such as cage subsidence and screw loosening in patients undergoing lumbar surgery [3,11,12]. The VBQ can reflect trabecular bone quality, and has been reported to be a good tool for assessing preoperative bone quality [13,14]. However, the clinical predictive value of the VBQ score for postoperative complications has not been well summarized. The meta-analysis was performed to comprehensively assess the predictive value of the VBQ for cage subsidence and screw loosening after degenerative lumbar surgery.
MATERIALS AND METHODS
1. Search Strategy and Screening
A comprehensive systematic literature review of original studies on the clinical effect of the VBQ score on postoperative complications of cage subsidence and screw loosening after degenerative lumbar surgery was performed by searching the PubMed, Embase, Cochrane Library, Web of Science and CNKI (China National Knowledge Infrastructure) databases (all to Jan. 31th, 2024). The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed to conduct this review. The search strategy for PubMed was “(MRI* tiab] OR magnetic tiab]) AND (marrow fat tiab] OR vertebral bone quality tiab] OR VBQ tiab]) AND (vertebra* tiab] OR spin* tiab] OR lumbar tiab])”. Furthermore, we reviewed the reference lists of the retrieved studies to obtain additional relevant articles.
After duplications were excluded, all studies were independently reviewed by 2 reviewers based on the titles and abstracts, and then the full texts of potentially eligible studies were retrieved for further assessment. We resolved disagreements by reaching a consensus through discussion. Clinical studies that explored the effect of the VBQ score on postoperative complications of cage subsidence and screw loosening after degenerative lumbar surgery were included. The exclusion criteria were: studies designed to be irrelevant to the VBQ effect for cage subsidence or screw loosening analysis; cervical VBQ or the thoracic VBQ analysis; insufficient outcome data could be obtained; improper patients included such as adolescent idiopathic scoliosis and review studies.
The VBQ score method was first introduced by Ehresman et al. [15] in 2019 using noncontrast T1-weighted images, and is based on the degree of fatty infiltration into the trabecular bone which occurs in osteoporosis [1,15]. As presented in Supplementary Fig. 1, the VBQ score was calculated by dividing the median midsagittal signal intensity (SI) for the L1 to L4 vertebral bodies by the SI of the L3 cerebrospinal fluid (CSF) [16,17]. The VBQ formula was as follows: VBQ score=SI (vertebral body)(L1−L4) median/SI (CSF)L3. In cases of scoliosis or poor visibility of midsection cuts, parasagittal sections through the trabecular bone and CSF signals at the L2 or L4 level could also be used for measurements [11].
The postoperative complication of cage subsidence could either be measured by midsagittal computed tomography (CT) imaging or by x-radiography, defined as more than 2 mm of cage migration into the vertebral endplate (superior, inferior, or both), and was graded according to the system proposed by Marchi [6,10,18,19]. Postoperative screw loosening was assessed by CT and was defined as the presence of a radiolucent zone of 1 mm or more around the pedicle screw, which is recognized as peri-screw osteolysis [20,21].
2. Data Extraction
Two independent authors reviewed the full text of each publication for study characteristics, study quality and outcome data. Any disagreements were resolved through consultation with a third reviewer. The outcome data included the incidence of postoperative complications, the mean VBQ score in the compared groups, the odds ratio (OR) for postoperative complications, and the number of true positives, false positive, false negatives, and true negatives for the diagnosis of postoperative complications. ORs with 95% confidence intervals (CIs) could be calculated by univariate analysis, multivariable Cox proportional hazards models or logistic regression analysis adjusted for other risk factors. When exploring the diagnostic effect for the VBQ score, the experimental group was the group with postoperative complications, and the control group was without the complications.
3. Assessment of Methodological Quality
The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool [22,23] was used to assess the methodological quality of the included studies. This tool helps to evaluate the principal methodological risk of bias in systematic reviews of test accuracy. The QUADAS-2 consists of 4 domains: patient selection, index test, reference standard, and flow timing. All 4 domains were evaluated for risk of bias, and only the first 3 domains were evaluated for applicability concerns. The quality of the included studies was independently assessed by the first 2 reviewers. Any disagreements were finally resolved through consultation with the third reviewer. In the assessment, we answered the 7 items with “yes,” “no,” or “unclear.” The application details for the VBQ score were previously detailed elsewhere [24]. If the answers to all signaling questions for a domain are “yes,” then the risk of bias can be judged to be low. If any signaling question is answered “no,” potential for bias exists. The “unclear” category should be used only when insufficient data are reported to permit a judgment.
4. Statistical Analysis
All the data were analyzed using Review Manager ver. 5.4.1 (Thomson Research Soft, Carlsbad, CA, USA) and Stata 17.0, (Stata Corp., College Station, TX, USA). The mean difference was utilized for continuous variables. ORs with 95% CIs were utilized to calculate the pooled risk of complications. Heterogeneity analysis was performed using the chi-square test, and the results were expressed in I2 statistic. Random-effect models (Dersimonian Laird method) were chosen over fixed-effect models. Subgroup analysis was performed to decrease substantial heterogeneity. The presence of heterogeneity due to threshold effects was explored by calculating the Spearman correlation coefficient. We calculated the following parameters and their 95% CIs: sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and area under the curve (AUC) of the summary receiver operating characteristic (SROC) curve. The cutoff value for the best diagnostic performance according to the Youden index was recorded. The data were presented as forest plots. Publication bias was assessed by funnel plots. Meta-regression analysis was performed to identify the source of heterogeneity. A probability of 2 tailed p<0.05 was regarded as statistically significant.
RESULTS
1. Included Studies
As presented in Fig. 1 of the stepwise literature review procedure, we eventually included 9 cohort studies. All the studies were retrospective in design. As presented in Table 1, 7 studies were performed in China and 2 studies were performed in the USA. The sample size ranged from 52 to 283 patients, with a total of 1,404 patients. For patient characteristics, the mean age was 60.4 years, the mean percentage of females was 57.0%, and the mean body mass index (BMI) was 25.9 kg/m². The mean follow-up period was approximately 1.0 to 2.0 years.
Of the 9 included studies, 6 explored the complication of cage subsidence [6,10,12,14,19,25] and 3 explored the complication of screw loosening [20,21,26]. The types of surgical fusion included posterior lumbar interbody fusion (PLIF) [11,20,21,26], transforaminal lumbar interbody fusion (TLIF) [6,11,14,19-21], oblique lumbar interbody fusion (OLIF) [10,25] and standalone lateral lumbar interbody fusion (LLIF) [12].
2. Quality of the Included Studies
The methodological quality of the individual studies was judged by the QUADAS-2 tool. The result was presented in Fig. 2. The patient selection domain assessed whether the selection of cases was biased. Because a large number of consecutive patient patients were excluded from the studies (as high as 527 of 701 patients were excluded from the study of Chen et al. [26] and 563 of 679 patients were excluded from the study of Gao et al. [21]), as well as the number of patient exclusion was not reported [6,10,19,20,25], the patient selection bias was judged as high risk in 2 of 9 studies (unclear in 6 studies). The “index test” domain was used to assess whether the interpretation of the index test was without knowledge of the reference standard and whether the threshold was prespecified, and the index test bias was high in 2 of 9 studies (unclear in 2 studies). The reference standard domain assessed the accuracy and the blinding method of the reference standard, and the reference standard bias was high in 1 of 9 studies (unclear in 3 studies). The flow and timing domain assessed whether there was a reasonable time interval between the index test and the reference standard, whether all cases were subjected to the same reference standard, and whether all patients were included in the analysis. The flow timing bias was judged as high in 1 of 9 studies.
In the applicability concerns evaluation section, the patient selection domain assessed whether included patients matched the evaluation of the index test for the complications. A high risk for patient selection was detected in 1 of 9 studies. The index test domain evaluated whether the results of the index test matched the evaluation of the complications, and a high risk for the index test was detected in 2 of 9 studies. The reference standard domain evaluated the applicability for the reference standard to the evaluation of the complications, and 2 of 9 studies were judged as unclear. Overall, the quality of the included studies was relatively low and high risks of bias might exist. All the results of this review should be interpreted within this limitation.
3. A Higher VBQ Score in the Postoperative Complication Group After Lumbar Degenerative Surgery
Cage subsidence was reported by 6 studies [6,10,12,14,19,25] with a total of 958 patients. The total incidence of cage subsidence was 31.0% (297 of 958). As shown in Fig. 3, a significantly higher VBQ score was found in the complication group than in the control group (mean difference, 0.66; 95% CI, 0.44–0.88; p< 0.001; I2=81%; random-effect model).
Screw loosening was reported by 3 studies [20,21,26] with a total of 446 patients. The total incidence of screw loosening was 28.9% (129 of 446). As presented in Fig. 3, a significantly higher VBQ score was found in the complication group than in the control group (mean difference, 0.45; 95% CI, 0.25–0.66; p<0.001; I2=65%; random-effect model).
Subgrouped by surgery type and presented in Fig. 4, posterior lumbar surgery of PLIF or TLIF was reported by 6 studies with a total of 1,045 patients. The total incidence of postoperative complications was 30.8% (322 of 1,045), and a significantly higher VBQ score was found in the complication group (mean difference, 0.55; 95% CI, 0.36–0.75; p<0.001; I2=85%, randomeffect model). Lateral lumbar surgery involving OLIF and LLIF was reported in only 2 studies with a total of 257 patients. The total incidence of postoperative complications was 25.3% (65 of 257) and no difference could be found considering that not enough studies could be included.
For all 9 studies presented in Fig. 3, the total incidence of complications (including cage subsidence and screw loosening) was 30.3% (426 of 1,404), and a significantly higher VBQ score was found in the complication group (mean difference, 0.59; 95% CI, 0.41–0.77; p<0.001; I2=83%; random-effect model). No subgroup difference was found (p=0.18).
4. Risk Factor Analysis for Postoperative Complications Associated With a High VBQ Score
As presented in Fig. 5, a high VBQ was a significant risk factor for cage subsidence and screw loosening after degenerative lumbar surgery. The merged OR of a high VBQ score for cage subsidence was 5.37 (95% CI, 1.53–18.79; p=0.009; I2=93%; random-effect model) and that for screw loosening was 3.87 (95% CI, 2.26–6.62; p<0.001; I2=0%; random-effect model). As presented in Supplementary Fig. 2, subgrouped by surgery type, the merged risk of a high VBQ for postoperative complications after PLIF and TLIF surgery was OR 3.78 (95% CI, 1.65– 8.66; p=0.002; I2=90%; random-effect model).
Overall, the overall risk of postoperative complications (including cage subsidence and screw loosening) after degenerative lumbar surgery in patients with a high VBQ was OR 4.64 (95% CI, 2.03-10.61; p<0.001; I2=89%; random-effect model).
5. Diagnostic Value of a High VBQ for Postoperative Complications After Degenerative Lumbar Surgery
As presented in Fig. 6, 6 studies reported the diagnostic sensitivity and specificity of the VBQ score between 223 patients with postoperative complications and 660 controls. The VBQ cutoff value varied greatly from 2.87 to 4.10, with a mean of 3.34±0.45, and little heterogeneity could be found by the threshold effect (p=0.21). The sensitivity and specificity ranged from 0.65 to 0.88 and from 0.65 to 0.89, respectively. The pooled sensitivity and specificity were 0.75 (95% CI, 0.67–0.82) and 0.75 (95% CI, 0.68–0.81) respectively. As presented in Supplementary Fig. 3, the PLR was 3.05 (95% CI, 2.34–3.96), the NLR was 0.33 (95% CI, 0.25–0.44), and the DOR was 9 (95% CI, 6–15).
6. SROC Curves of High VBQ Scores for the Diagnosis of Postoperative Complications
The summary ROC curve indicating the overall diagnostic accuracy of the VBQ score was synthesized by Stata 17.0 and Review Manager 5.4.1 software. As shown in Fig. 7A, the AUC of a high VBQ for the diagnosis of postoperative complications (including cage subsidence and screw loosening) after degenerative lumbar surgery was 0.82 (95% CI, 0.78–0.85).
7. Publication Bias
A Deeks’ funnel plot of high VBQ scores for the diagnosis of postoperative complications was presented in Fig. 7B. A p-value of 0.50 indicated little risk for small study effects/publication bias. The funnel plot for publication bias detection was also presented in Supplementary Fig. 4, and the shape of the funnel plot did not have obvious asymmetry, suggesting little significant publication bias.
8. Meta-Regression Analysis
To determine the impact of moderator variables on heterogeneity, a meta-regression analysis including sample size, complications, VBQ cutoff value, surgery, patient age, sex, and BMI was performed. As presented in Fig. 7C, the results showed that the possible source of heterogeneity including VBQ cutoff value (p=0.03 for specificity), surgery type (p=0.02 for specificity), female gender (p=0.05 for sensitivity) and complications of cage subsidence or screw loosening (p=0.04). Considering little difference was found between the subgroup analysis of surgery type (PLIF/TLIF or OLIF/LLIF) and complications (cage subsidence or screw loosening), the heterogeneity could be mainly sourced from the VBQ value.
DISCUSSION
The meta-analysis was performed to assess the predictive value of the VBQ score for postoperative complications of cage subsidence and screw loosening after degenerative lumbar surgery. By including 9 studies with 1,404 patients, we found that a high VBQ was associated with cage subsidence and screw loosening. Risk factor analysis revealed that a high VBQ was a risk factor for cage subsidence (OR, 5.37) and screw loosening (OR, 3.87). With a VBQ cutoff value of 3.34±0.45, the pooled sensitivity and specificity for the diagnosis of postoperative complications (cage subsidence and screw loosening) were 0.75 and 0.75, respectively, and the AUC was 0.82 (95% CI, 0.78–0.85). We speculated that the VBQ could be considered for identifying high-risk patients for further evaluation of postoperative complications after degenerative lumbar fusion surgery.
1. The VBQ Score to Measure Bone Quality
Because bone quality is associated with screw pullout strength, insertional torque and vertebral body loading properties [3,27], it is well recognized that poor bone quality increases the risk of postoperative complications for patients undergoing degenerative lumbar surgery. Currently, the DEXA method is the gold standard for categorizing osteopenia and osteoporosis according to the World Health Organization, and QCT is also reported to be an accurate method for assessing bone quality [1,4,28]. While DEXA measurements are influenced by superimposed calcified tissue and degenerative changes in the lumbar spine, false elevated measurements by DEXA may lead to an underestimation of osteoporosis [29]. Because bone quality is crucial for surgical outcomes, a convenient and effective screening tool for the preoperative assessment of bone quality is necessary not only to develop a reasonable surgical plan, but also to take relevant preventive measures to reduce complications related to osteoporosis [30].
Recently, the VBQ score has emerged as an alternative tool for assessing preoperative bone strength in patients undergoing lumbar surgery [31]. Histological studies have demonstrated that osteoporotic bone is characterized by trabecular atrophy and replacement by local adipocytes [27]. Theoretically, the VBQ, which measures the T1-weighted MRI image of the lumbar spine, estimates fat infiltration within the vertebral body and assesses trabecular bone quality [24]. To normalize the baseline signal differences across MRI scanners, the median T1 SI of L1–4 was adjusted by the L3 CSF SI. By the internal adjustment of the L3 CSF signal, the VBQ score has been shown to be generalizable across multiple MR systems from different manufacturers [29], however, the thresholds still varied among the included studies. Although Spearman correlation coefficient analysis in the present meta-analysis revealed little influence of the threshold effect, further research is needed to explore the exact diagnostic threshold.
Bone strength is composed of bone quantity and quality [16], and the DEXA and QCT methods measure only bone quantity. Population-based studies have indicated that bone strength might not be solely determined by bone density [24,29], as reported in the Rotterdam Study in which the majority of elderly fractures were classified as nonosteoporotic individuals according to the DEXA. Although controversial as an assumed measurement of bone “quality,” the VBQ score was claimed be a sensitive and precise risk assessment tool [14,16]. Current evidence suggests that the VBQ score has a moderate ability to predict osteopenia/osteoporosis [19,28,29], and a meta-analysis showed that the AUC for the VBQ score was 0.84 for the diagnosis of bone loss, with a sensitivity of 0.81 and a specificity of 0.64 [24]. Nevertheless, it is still important to note that the VBQ cannot replace the QCT or DEXA at the current stage [2]. The QCT was a more accurate method not only to measure the osteoporotic bone [8,9], but also to predict postoperative cage subsidence [9,32] and screw loosening [33,34]. The opportunistic MRI-based VBQ score can be used as an additional screening tool for identifying patients with low bone mass who need further evaluation [3]. In combination with routine DEXA, the VBQ score might play a complementary role in bone quality examination and postoperative complication prediction [17,30].
2. The VBQ Score to Predict Postoperative Complications After Lumbar Surgery
Cage subsidence is a common complication following lumbar fusion surgery and low bone density is considered to play an important role [14,19]. In addition to surgical factors, a high VBQ score has been reported to be a significant risk factor [6,10,14,19,25]. Pedicle screw loosening is another commonly reported complication [21]. The incidence of screw loosening ranges from 0.6% to 19.5% in nonosteoporotic individuals, while in osteoporotic patients, it is as high as 60% [21]. According to the literature, the VBQ is an independent factor affecting screw loosening [20,21]. In the present study, our merged results confirmed that a high VBQ was significantly associated with cage subsidence (OR, 5.37) and screw loosening (OR, 3.87).
The AUC is always used to evaluate the accuracy of diagnostic tests. To demonstrate excellent accuracy, the AUC should be as high as 0.97 [35]. An AUC of 0.93 to 0.96 is considered to be very good and 0.75 to 0.92 is considered to be good [35]. Our meta-analysis revealed that with the mean VBQ cutoff value of 3.34±0.45, the pooled sensitivity and specificity for the diagnosis of postoperative complications (cage subsidence and screw loosening) were 0.75 and 0.75, respectively, and the merged AUC was 0.82 (95% CI, 0.78–0.85). We recommended a VBQ threshold of 3.3 for a high risk, however, the optimal threshold had better be adjusted to the local institution. Overall, although not promising, the present meta-analysis demonstrated that the VBQ score had a good role in the prediction of postoperative complications of cage subsidence and screw loosening. The cutoff value of the meta-analysis could be used for reference.
3. Source of Heterogeneity
The statistical heterogeneity was relatively high among the included studies. By meta-regression analysis and subgroup analysis, we found that the heterogeneity could mainly be sourced from the VBQ value. Although adjusted by the internal control of the CSF signal, the VBQ values still varied among the included studies. For example, in the study of Pu 2023, the VBQ score in the complication group and control group were 5.10± 1.70, 3.31±0.94, respectively; in the study of Jones 2023, the VBQ score in the complication group and control group were 2.67±1.08, 2.39±0.44, respectively. The VBQ cutoff value for ROC curve varied greatly from 2.87 to 4.10, with a mean of 3.34±0.45. Although it was reported that the VBQ score was not significantly influenced by the different MR manufacturers and parameters (magnetic fields of 1.5 T or 3.0 T), we still considered it too arbitrary to apply the cutoff value too extensively. Maybe the cutoff value of the VBQ score could not be easily defined as DEXA and QCT, and the results should be considered according to race, heredity, gender, hyperlipidemia, abdominal visceral fat, BMI, long-term opioid use, steroid use, smoking status, et al. Although efforts were made, the heterogeneity could not be completely eliminated and a random-effect model was utilized.
4. Limitations and Future Researches
To our knowledge, this is the first systematic analysis to summarize the predictive value of the VBQ score for the postoperative complications of cage subsidence and screw loosening in patients who underwent degenerative lumbar surgery. However, this study had limitations common to systematic reviews of observational studies. First, all of the included studies were retrospective cohort studies that are susceptible to selection bias and observer expectations, which might have led to an overestimation of the diagnostic advantage ratio, and the quality of the included studies was relatively low. Second, the study population was limited to elderly patients undergoing spinal surgery, and the results might not be generalizable to young healthy individuals. Third, the clinical and statistical heterogeneity was relatively high among the included studies. Although efforts were made, the heterogeneity could not be completely eliminated, and a random-effect model was utilized. Fourth, the thresholds of the VBQ varied among the included studies, and further research was still needed to explore the exact threshold. Fifth, because limited publications could be retrieved, the prognostic value of VBQ scores for other osteoporosis-related complications, such as proximal junctional kyphosis (PJK), adjacent segment degeneration (ASD) and reoperations, as well as the difference between open surgery and minimal invasive surgery would need further analysis. Finally, the exclusion of unpublished data might have resulted in publication bias. All the results of this review should be interpreted within the above limitations.
Future prospective cohort studies with long follow-up periods are necessary to explore the prognostic value of VBQ scores, and the following issues are of particular concern: whether the clinical significance of the VBQ score was influenced by the race, gender, hyperlipidemia, BMI, long-term opioid use, steroid use, smoking status, et al.; the prognostic value for other osteoporosis-related complications, such as PJK and ASD; the exact VBQ cutoff value and whether the cutoff value could be used extensively.
CONCLUSION
The meta-analysis was performed to assess the predictive value of the VBQ for postoperative complications of cage subsidence and screw loosening in patients who underwent degenerative lumbar surgery. Although the quality of the included studies was not high and heterogeneity still existed, by including 9 studies with 1,404 patients, we found that a high VBQ was associated with cage subsidence and screw loosening. Risk factor analysis revealed that a high VBQ was a risk factor for cage subsidence (OR, 5.37) and screw loosening (OR, 3.87). With a VBQ cutoff value of 3.34±0.45, the pooled sensitivity and specificity for the diagnosis of postoperative complications (cage subsidence and screw loosening) were 0.75 and 0.75, respectively, and the AUC was 0.82 (95% CI, 0.78–0.85). We concluded that a high VBQ was associated with a high risk of postoperative complications of cage subsidence and screw loosening. Because it is free of radiation and easy to analyze, the VBQ could be considered for identifying high-risk patients for further evaluation to develop a reasonable surgical plan and take relevant preventive measures. Further prospective studies are still needed to evaluate this issue.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement
Print
