Factors that influence long term instrumentation stability in patients with unstable thoracolumbar injuries

Abstract

BACKGROUND

Pedicle screw fixation is frequently used to treat unstable thoracolumbar injuries; however, the rate of instrumentation failure remains considerable. The primary contributing factor leading to instrumentation failure is poor bone quality. On the other hand, some evidence suggests that surgical tactics can influence long-term instrumentation stability.

AIM

To assess factors that influence the stability of spinal instrumentation in patients with thoracolumbar injuries.

METHODS

This study is a non-randomized single center ambispective evaluation of 204 consecutive patients (117 men; 87 women) with unstable thoracolumbar injuries. All patients underwent either stand-alone or combined with anterior column reconstruction instrumentation. In cases with spinal cord and nerve root injuries, either posterior or anterior decompression were performed. Patients with pedicle screw loosening were identified via computed tomography imaging. Out of those, cases with clinically significant instrumentation failure were registered.

RESULTS

The rate of pedicle screw loosening detected by computed tomography was inversely correlated with bone radiodensity figures and an increased association with the number of instrumented levels, residual kyphotic deformity, laminectomy, and lumbosacral fixation. Intermediate screws and anterior reconstruction were associated with a clinically relevant decreased risk of pedicle screw loosening development. Either complete or partial posterior fusion within instrumented levels was capable of decreasing instrumentation failure risk, while extensive decompression with laminectomy and at least one-level total facetectomy were associated with an increased risk of instrumentation failure. Anterior decompression does not have a negative impact on instrumentation stability.

CONCLUSION

Intermediate screws, anterior reconstruction and posterior tension band preservation are associated with decreased rates of instrumentation instability development. Posterior fusion is beneficial in terms of instrumentation failure prevention.

Key Words: Instrumentation stability; Unstable thoracolumbar injuries; Pedicle screw loosening; Anterior reconstruction; Posterior decompression

Core Tip: A single center ambispective non-randomized evaluation of 204 consecutive patients with unstable thoracolumbar injuries treated pedicle screw fixation. Patients with pedicle screw loosening were identified using computed tomography, and those cases with clinically significant complications were registered. Multivariate logistic regression and general discriminant analysis suggest that the rate of pedicle screw loosening depends on bone radiodensity, length of fixation, type and extensiveness of decompression. Auxiliary posterior fusion, intermediate screw and anterior reconstruction can provide a clinically significant decrease in complication rate.

INTRODUCTION

Traumatic injuries of the lumbar spine and thoracolumbar junction are the most common over the spine column because of the biomechanical specificities of the transitional zone[1,2]. Pedicle screw fixation is the most frequently employed technique to treat unstable high- and low-energy thoracolumbar injuries. On the other hand, the rate of instrumentation failure remains considerable, especially in patients with altered bone quality[3]. Bone quality is one of the most important factors that influence instrumentation stability; on the other hand, there is increasing evidence that surgical tactics and implant characteristics may help reduce the negative impact of the former[4].

Even though there has been a considerable amount of research on relevant topics, a standardized approach does not exist, and some conclusions remain controversial[5]. It has been stated that long fixation in the case of thoracolumbar injuries provides greater stability compared to short fixation, which is associated with more frequent loss of correction and instability development[6-10]. On the other hand, other studies suggest that short fixation with intermediate screws may provide results comparable to those achieved by long fixation[11]. One approach increase spinal instrumentation stability is through anterior column reconstruction[12-14]. However, the anterior approach is associated with greater blood loss and morbidity[15,16]. Taking into account the significant incidence of combined high-energy traumatic injuries and morbid obesity in the elderly adult population, not every patient is capable of tolerating additional surgery from the anterior approach.

The greatest controversies are associated with techniques used to decompress the spinal cord and nerve roots in patients with thoracolumbar trauma and neurological complications. Anterior and posterior decompression provide comparable results[5,17-19]. Therefore, some surgeons advocate for the posterior approach in patients with thoracolumbar injuries, while others claim that decompression from the anterior approach is more effective[16,20]. In addition, the anterior approach requires relevant skills that prevent routine application of those techniques. Posterior decompression may require a vast resection of posterior structures; however, the impact of the posterior column resection on instrumentation stability has been insufficiently studied. This has resulted in significant heterogeneity in the approaches to thoracolumbar trauma treatment. Furthermore, factors that influence instrumentation stability have been separately studied without assessing their relative contributions in a multivariate model[5,11]. The aim of this study was to assess factors that influence spinal instrumentation stability in patients with thoracolumbar injuries.

MATERIALS AND METHODS

Study population and data collection

This study was a non-randomized single center ambispective evaluation of 204 consecutive adult patients with unstable thoracolumbar injuries, including 117 men and 87 women with an average age of 40 years (σ = 15.01; range: 18-71 years). Patients with unstable thoracolumbar injuries were enrolled. All patients underwent either stand-alone or combined with anterior column reconstruction instrumentation during the 2016-2020 period. In cases with spinal cord and nerve root injuries, either posterior or anterior decompression was performed. Patients were followed up for 24 months, and both prospective and retrospective models were used to detect risk factors for implant instability development. The study was reviewed and approved by the local institutional review board, given that there were no anticipated risks associated with the study.

Inclusion and exclusion criteria

The inclusion criteria were unstable thoracolumbar injuries С, B1, B2, A4, A3 types[21]; levels of injury T11-L5 with a bed rest no longer than 1 month. The exclusion criteria were: (1) A and B American Spinal Injury Association (ASIA) groups because of difference in rehabilitation potential; (2) Low energy fractures associated with tumor lesion; (3) Spontaneous low energy fractures; (4) Multilevel injuries; (5) Cases with pedicle screws augmentation; (6) Cases with screw malposition and redirection detected on computed tomography (CT) images; (7) Admission for revision surgery; (8) Patients with comorbidities that may affect bone quality including decompensated liver and kidney disease, thyroid and parathyroid pathology, autoimmune diseases requiring application of corticosteroids; (9) Bedridden patients; and (10) Patients who received therapy to improve bone quality before surgery, including antiresorptive therapy.

Preoperative assessment

Upon admission, patients underwent neurological examination; out of total patients enrolled, 28%-13.7% cases were classified as C group ASIA (paresis with sensory deficit detected), 46%-22.6% as D group ASIA (only radiculopathy with sensory abnormalities detected), while 130%-63.7% cases were neurologically intact. All patients underwent computed tomography using single scanner (Aquilion 32, Toshiba Corporation), calibration using appropriate phantoms. Maintenance was performed by the authorized staff of the manufacturer according to the manual. Standard protocol was used to examine all patients with a 0.5 mm slice thickness, 50 cm scan area with 21.0 helical-pitch. Tube voltage accounted for 120 kV, while tube current was 300 mA with an auto mA range of 180-400. Integrated software used for measurements was Vitrea Version 5.2.497.5523 with a window width/window level ratio of 2.000/500. The CT images were assessed by two independent radiologists, and patients were classified according to AO Spine Classification[21]. During CT examination, radiodensity of the cancellous bone in Hounsfield units (HU) on standard levels of L2 or L3 was measured. Samples of the trabecular bone were selected without traversing into the cortical bone with a maximally achievable square in coronal, axial and sagittal plane. Out of those figures, mean radiodensity value was calculated for each patient.

Surgical techniques used

In all cases, pedicle screw fixation with a straight trajectory of screw placement to the anterior third of the vertebral body without penetration of the cortical bone into the vertebral body was used. Pedicle screw fixation with intermediate screws introduced into fractured vertebra was used in 93 cases (45.6%). Pedicle screws and rods were made of standard titanium alloy, and cobalt chromium rods were not used. Posterior fusion was performed using standard technique, including a decortication of articular processes, facet joint cartilage with adjacent removal. Finally, the gap formed during previous steps was filled with autologous bone with the additional bone apposed the decorticated articular processes. Patients in C and D group ASIA underwent posterior or anterior decompression. The amount of posterior column resection was classified as laminectomy (66 cases, 32.4%), total facet joint and ligamentum flavum removal (48 cases, 23.5%), or both (37 cases,18.1%). Reconstruction of the anterior column was performed from the anterior approach in more than 50% (76 patients, 37.2%) of patients who presented with vertebral body height loss. Titanium mesh with autologous and allogenic bone was used for anterior column reconstruction. Out of those, 31 patients (15.1%) underwent anterior decompression with the intracanal fragment of the vertebral body removal.

Postoperative assessment

The follow-up period accounted for 24 months. Examination using Visual Analogue Scale, Oswestry Disability Index questionnaire was used at 3 months, 6 months, 12 months, 18 months, and 24 months. CT scanning was employed at 6 months and 12 months after surgery and at 18 months and 24 months if no stable fusion was detected previously.

Anterior fusion was classified according to the modified Tan classification with merging 1 and 2 grades into one group because those cases do not require revision surgery due to instrumentation failure[22]: (1) Bipolar fusion - no radiolucent zone detected by CT examination between bone graft and both endplates of the adjacent vertebral bodies, bridging bone trabeculae are evident within the whole square of graft-adjacent vertebral body interface; (2) Unipolar pseudoarthrosis - no fusion detected between bone graft and one endplate of the adjacent vertebral bodies confirmed by radiolucent zone detected by CT; and (3) Complete pseudoarthrosis - a radiolucent zone detected between bone graft and both adjacent endplates. Christensen classification was used to assess posterior fusion: (1) Total fusion - complete ankylosing of facet joints with evident trabeculation detected on CT images; (2) Partial fusion - bone bridging within facet joint was present, however only partial ankylosing of the facet joint was achieved; and (3) Total non-union - no bone bridging detected within facet joint[23].

During the follow-up period, cases with pedicle screw loosening were detected and registered. The CT criteria for screw loosening were at least a 1-mm zone of bone resorption around the screw or a double halo sign (Figure 1). Surgical outcomes were classified in a dichotomized scale, either complication regardless of the number of loosened screws or without screw loosening detected. Out of the cases with signs of pedicle screw loosening, patients with clinically significant instrumentation instability were registered.

Figure 1 Computed tomographic image of a lumbar spine in a coronal plane. Bilateral double halo sign is evident on L3 Level (radiolucent zone surrounded by sclerotic bone).

Statistical analysis

Power analysis was performed to assess the required sample size using Monte-Carlo method, 2000 simulations were performed. Logistic regression and Kernel-Fisher discriminant analysis were used to assess factors associated with pedicle screw loosening and clinically significant instrumentation failure. PASS 2008 and Statistica 12 were used for statistical analysis.

RESULTS

The basic data concerning enrolled patients are presented in Table 1. By the end of the follow-up period, 52 (25.5%) patients presented with signs of pedicle screw loosening; out of those, 27 (13.2%) presented with axial pain Visual Analogue Scale of over 4 and Oswestry Disability Index scores exceeding 40. Patients with persistent pain syndrome after spinal instrumentation presented with either multiple pedicle screw instability or bilateral one-level screw loosening, along with either unipolar or bipolar anterior pseudoarthrosis and a lack of posterior fusion. Out of those with persistent pain syndrome, 19 patients underwent revision surgery. The relatively high prevalence of pedicle screw loosening detected on CT images can be explained by the considerable number of patients with a radiodensity values lower than a threshold for osteoporosis detection, which accounts for 63 cases (30.9%) in the enrolled group[3].

Table 1 Characteristics of the enrolled group.

Characteristics	Value
Age (years)	m = 40 (σ = 15.01; range: 18-71 years)
Male to female ratio	117/87
Patients with radiodensity of cancellous bone below 110 HU	63%-30.9%
Intermediate fixation used	93%-45.6 %
Laminectomy	66%-32.4%
Total facet joints and ligamentum flavum removal	48%-23.5%
Extensive decompression with laminectomy and at least one level total facetectomy and ligamentum flavum removal	37%-18.1%
Reconstruction of the anterior column	76%-37.2%
Anterior decompression	31%-15.1%
Monosegmental short fixation	10%-4.9%
Bisegmental fixation	139%-68.1%
Three segment fixation	21%-10.3%
Four segment fixation	34%-16.7%
Lumbosacral fixation	6%-2.9%
Residual kyphotic deformity	38%-18.6%

HU: Hounsfield units.

A general logistic regression analysis was used to assess the significance and relative contribution of patient- and surgery-related factors to the complication rate detected by CT. Variables considered were cancellous bone radiodensity measured in HU, number of instrumented levels (length of fixation), extensiveness of posterior structure resection, including laminectomy and bilateral facetectomy, anterior reconstruction, anterior decompression performed, intermediate screws used, lumbosacral fixation, and residual kyphotic deformity of over 10°. The estimated parameters of the logistic regression with the best achievable discriminant function are presented in Table 2.

Table 2 Parameters of general logistic regression.

Components of regression model	Regression coefficient	OR per unit change (95%CI)
Intercept	-1.5227, P = 0.0537	-
Number of instrumented levels	0.6215, P = 0.0076	1.8617 (1.1819-2.9324)
Lumbosacral fusion	2.3039, P = 0.0292	10.0136 (1.2655-79.2374)
Residual deformity over 10°	1.3772, P = 0.0028	3.9636 (1.6161-9.7213)
Intermediate fixation	0.8377, P = 0.0457	2.3110 (1.0162-5.2553)
Anterior reconstruction	-1.2827, P = 0.0095	3.7491 (1.3725-3.6062)
Radiodensity HU	-0.0142, P = 0.0029	0.9869 (0.9768-0.9951)
Laminectomy	0.8809, P = 0.0448	2.4130 (1.0210-5.7029)
Anterior decompression performed	-0.0726, P = 0.9071	0.9299 (0.2729-3.1688)

OR: Odds ratio; CI: Confidence interval; HU: Hounsfield units.

The overall goodness of fit of the estimated model was χ² = 49.8090; P < 0.0001. The classification accuracy achieved 79.9% of cases, with 92.7% specificity and 42.3% sensitivity. In terms of cases with clinically significant instrumentation failure, the best discriminative characteristics were achieved using a retrospective kennel-Fisher model with the inclusion of variables relevant to anterior and posterior fusion assessment according to CT results. The parameters of the estimated discriminant model are presented in Table 3.

Table 3 Parameters of general discriminant model for the assessment of risk factors for clinically significant instrumentation failure.

Components of regression equation	Regression coefficient	Statistical significance	Beta coefficient
Intercept	0.5679	P = 0.0008	-
Residual deformity of over 10°	0.0529	P = 0.0518	0.1216
Anterior fusion bipolar, Tan 1-2	0.0027	P = 0.9323	0.0063
Partial or complete ankyloses of facet joints within instrumented area	-0.1684	P = 0.0006	-0.4612
Anterior decompression performed	0.0219	0.5109	0.0463
Anterior reconstruction	-0.0780	0.0040	0.0207
Extensive posterior decompression with laminectomy and at least one level total facetectomy	0.1194	P = 0.0003	0.2716
Lumbosacral fusion	0.0799	0.2133	0.0796
Radiodensity in HU	0.0002	0.6894	0.0239
Length of fixation (number of levels)	0.0362	0.6894	0.0239
Intermediate fixation	-0.1262	0.0025	-0.2036

HU: Hounsfield units.

The additional dichotomized variables used in the discriminant model were the presence of partial or complete posterior fusion within instrumented levels and the presence of bipolar anterior fusion, corresponding to Tan 1-2 grade. Goodness of fit of the estimated model: Eugen value of successive root = 0.5950, r = 0.6108, Wilks λ = 0.6169, χ² = 91.277; P < 0.0001. The estimated model correctly classified 89.7% of cases, with 93.8% specificity and 63.0% sensitivity.

The results detected slightly different factors associated with pedicle screw loosening rate found during CT examination and clinically significant instrumentation instability. The rate of pedicle screw loosening detected by CT was inversely correlated with bone radiodensity and positively correlated with the number of instrumented levels. Additional risk factors for pedicle screw loosening development were residual kyphotic deformity, laminectomy, and lumbosacral fixation, while intermediate screws and anterior reconstruction application were associated with a decrease in rate of pedicle screw loosening development. The results of retrospective general discriminant analysis demonstrate that factors related to surgical tactics were the most important in relation to clinically significant instrumentation failure development. Either complete or partial posterior fusion within all instrumented levels was capable of decreasing the risk of instrumentation failure, while extensive decompression with laminectomy and at least one-level total facetectomy were associated with an increased risk of instrumentation failure. In contrast, anterior decompression does not have a negative impact on instrumentation stability. According to the analysis results, anterior reconstruction and intermediate screws have a significant clinical effect, resulting in a decrease in the rate of clinically significant instrumentation failure.

DISCUSSION

Even though a considerable amount of research has been published, approaches to the surgical treatment of unstable thoracolumbar trauma are not standardized[5]. The greatest controversies are associated with anterior reconstruction, the required length of fixation and decompression technique application. The applied technique of decompression and fusion may depend on the surgical skills and preferences of the particular surgeon. Furthermore, there is no evidence that one kind of decompression technique is advantageous over another[18,19].

Pedicle screw fixation is the most frequently applied technique to treat unstable thoracolumbar spine injuries; on the other hand, the most frequently reported complication is associated with pedicle screw loosening[24]. Even though a radiolucent zone of over 1 mm around the screw and a double halo sign are associated with pedicle screw loosening, it does not have clinical relevance in a significant number of cases[24]. For that reason, risk factors for pedicle screw loosening were assessed twice. All potential factors associated with the pedicle screw loosening rate detected by CT were assessed using a logistic regression model. Out of those contributing to clinically significant instrumentation failure, they were detected using general discriminant analysis, with an emphasis on potential interaction between factors. The most frequently discussed factor associated with implant-related complications is bone quality. It has recently been demonstrated that bone radiodensity in HU can be used as a predictive factor for the development of nonunion or pedicle screw loosening[3,25]. On the other hand, certain evidence suggests that the applied surgical technique and implant characteristics contribute to long-term instrumentation stability[4,26,27]. For now, the greatest evidence is provided concerning the results of anterior reconstruction and intermediate screw application that provide considerable stability to the instrumentation[12,15]. According to the analysis, both techniques can provide clinically relevant effects, resulting in the decline of pedicle screw loosening[15]. However, considering the significant incidence of combined high-energy traumatic injuries and morbid obesity in the elderly adult population, not every patient is capable of tolerating additional surgery from the anterior approach[15]. The results demonstrate that the discussed techniques are comparable, resulting in a decline in the rate of pedicle screw loosening, supporting intermediate screw placement in patients with morbid obesity.

The greatest controversies observed in the literature are those associated with the required fixation length and the applied decompression technique. For a long period, it was thought that burst fractures of the lumbar spine require at least four fixation segments, especially in the thoracolumbar region because of relatively high rate of the implant failure in patients who underwent short fixation[10,28]. On the contrary, a considerable amount of research demonstrates that short-bisegmental fixation may provide comparable results to long fixation, especially if intermediate screws are introduced[7,29-35]. According to the results of our study, the fixation length is not associated with clinically significant instrumentation failure detection; furthermore, the rate of pedicle screw loosening positively correlated with the instrumented segments number. The same effect was detected in patients with degenerative disease of the lumbar spine. The potential explanation is that the load on marginal points of fixation grows with the increase of cantilever arm in patients with longer fixation[24]. Furthermore, anterior reconstruction is performed in the middle of the fixed area, while marginal segments remain without anterior support. Long fixation is preferred in patients with C type AO Spine Classification injuries and significant kyphotic deformity because this strategy may provide appropriate deformity correction[15,36]. It has been reported that understored correction and loss of kyphotic correction are associated with worse results, while long fixation provides better outcomes[6-9]. The results of our study demonstrate that understored kyphotic deformity is associated with an increased risk of pedicle screw loosening. On the other hand, the effect of instrumentation extension and the potential risk associated with residual kyphotic deformity was clinically insignificant. An additional detected risk factor for pedicle screw development was lumbosacral fixation; however, its clinical effect remains uncertain, likely because of the minority of those who underwent instrumentation of this transient zone. The potential risk of lumbosacral fixation can be explained by biomechanical features and the anatomy of the sacrum[26]. The least studied effect on instrumentation stability is the amount of posterior tension band structure resection. The main source of relevant controversies is that no evidence is provided that any decompression technique has a considerable advantage over another[16,20,37]. Both anterior and posterior decompression may provide comparable rates of neurological recovery, while the anterior approach is associated with increased access-related morbidity[5,17-19,38]. In contrast, posterior structures may provide additional stability to the instrumented levels, especially in the lumbosacral area. In previously published biomechanical studies, resection of posterior structures substantially increases range of motion in the corresponding segments, while increased micro-motion may favor pedicle screw loosening[26,39,40]. Additional effects on long-term stability can be provided by the additional fusion area achieved by posterior fusion. The results of the current study demonstrate that even incomplete posterior ankylosing at every instrumented level may have a clinical impact, resulting in a decline in clinically significant instrumentation failure. Conversely, extensive resection of posterior structures may prevent posterior fusion formation, causing deleterious effects in patients at risk of implant instability development. Furthermore, anterior nonunion was not clinically significant in terms of complication rate, likely because the biomechanical effect of load distribution is greater than that associated with the anterior fusion achieved. According to the results, decompression from the anterior approach does not have a negative impact on instrumentation stability; therefore, its application can be suggested in patients with an increased risk of instrumentation failure development.

Less expectedly, radiodensity in HU was significant for only signs of pedicle screw loosening detected by CT, while this factor turned out to be insignificant in the discriminant model for clinically relevant instrumentation failure risk assessment. The results can be explained by the heterogeneity of the enrolled group in terms of the applied surgical tactics and are consistent with previously reported studies[3,4,25,27]. The results of recent studies clearly demonstrate that surgical technique and implant characteristics may have a strong influence on instrumentation stability, preventing implant-related complications. Our study demonstrates that optimal surgical tactics may help diminish unfavorable factors related to altered bone quality, neglecting the significance of the radiodensity parameter in the model for clinically significant instability development risk assessment.

To conclude, surgical tactics may provide a substantial increase in long-term instrumentation stability. Even though anterior and posterior decompression can provide comparable results in terms of neurologic recovery, the former can be recommended for patients at risk of instrumentation instability development. Avoiding unnecessary extensive posterior decompression may provide a supplementary healing surface for further posterior fusion formation capable of decreasing instrumentation failure development. Both anterior reconstruction and intermediate screws may result in a decrease in instrumentation failure rate with a comparable impact on complication rate, thus supporting the latter in patients not capable of tolerating the anterior approach. Long fixation does not have a great effect on instrumentation stability and is likely to have a rational application for deformity correction.

Limitation

The suggested study has certain notable limitations. The enrolled sample size is heterogeneous and relatively small in terms of the number of studied factors. Also, the collinearity of some data may impact the results of the analysis. Variables relevant to the injury type were not taken into account; however, both patients with fair and poor bone quality were enrolled. Those cases should be assessed using different classification systems that do not have total agreement. Also, variables associated with drinking and smoking status were not used in this study, on the other hand the discussed parameters could be extremely heterogeneous in terms of quantity.

CONCLUSION

Intermediate screws, anterior reconstruction and posterior tension band preservation are associated with a decrease in rate of instrumentation instability development. Posterior fusion is beneficial in terms of instrumentation failure prevention.

ACKNOWLEDGEMENTS

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.