The Necessity of Implant Removal after Fixation of Thoracolumbar Burst Fractures—A Systematic Review

Background: Thoracolumbar burst fractures are a common traumatic vertebral fracture in the spine, and pedicle screw fixation has been widely performed as a safe and effective procedure. However, after the stabilization of the thoracolumbar burst fractures, whether or not to remove the pedicle screw implant remains controversial. This review aimed to assess the benefits and risks of pedicle screw instrument removal after fixation of thoracolumbar burst fractures. Methods: Data sources, including PubMed, EMBASE, Cochrane Library, Web of Science, Google Scholar, and Clinical trials.gov, were comprehensively searched. All types of human studies that reported the benefits and risks of implant removal after thoracolumbar burst fractures, were selected for inclusion. Clinical outcomes after implant removal were collected for further evaluation. Results: A total of 4051 papers were retrieved, of which 35 studies were eligible for inclusion in the review, including four case reports, four case series, and 27 observational studies. The possible risks of pedicle screw removal after fixation of thoracolumbar burst fractures include the progression of the kyphotic deformity and surgical complications (e.g., surgical site infection, neurovascular injury, worsening pain, revision surgery), while the potential benefits of pedicle screw removal mainly include improved segmental range of motion and alleviated pain and disability. Therefore, the potential benefits and possible risks should be weighed to support patient-specific clinical decision-making about the removal of pedicle screws after the successful fusion of thoracolumbar burst fractures. Conclusions: There was conflicting evidence regarding the benefits and harms of implant removal after successful fixation of thoracolumbar burst fractures, and the current literature does not support the general recommendation for removal of the pedicle screw instruments, which may expose the patients to unnecessary complications and costs. Both surgeons and patients should be aware of the indications and have appropriate expectations of the benefits and risks of implant removal. The decision to remove the implant or not should be made individually and cautiously by the surgeon in consultation with the patient. Further studies are warranted to clarify this issue. Level of evidence: level 1.


Introduction
A new spinal fracture occurs every 22 s worldwide [1]. As a mechanical transition junction between the relatively rigid thoracic and the more flexible lumbar spine, the thoracolumbar region is the most common site of fracture to the spine, and burst fractures of the

Eligibility Criteria
Published studies were included if they met the following inclusion criteria: (i) Participants: adult patients who underwent internal fixation for thoracolumbar burst fractures; (ii) Intervention and/or comparison: removal or retention of the pedicle screw instrument after successful fixation of thoracolumbar burst fractures; (iii) Outcomes: clinical outcomes related to the benefits or harms of implant removal were considered. The primary outcomes were local kyphosis deformity after implant removal and pain intensity after implant removal. Secondary outcomes included improvement of segmental motion angle and removal-related complications; (iv) Study type: All types of studies that reported the benefits and risks of implant removal after thoracolumbar burst fractures were considered for inclusion, including but not limited to case reports and case series, cohort studies, case-control studies, cross-over studies, and randomized controlled trials.

Study Selection
Identified papers from each of the databases were imported into Endnote reference management software X9 (Clarivate Analytics). Two authors independently removed the duplicates, examined the titles and reviewed the abstracts for relevance, and then sorted the remaining records for "inclusion", "exclusion", or "potentially relevant". The full-text articles of eligible records rated "potentially relevant" were obtained, reviewed, and rated independently by the two reviewers. Any discrepancies were resolved by discussion between the authors.

Data Extraction
The data were extracted using a standardized data extraction form and entered into an excel sheet (Excel, Microsoft Corporation, WA, USA). The following study details were extracted where possible from included studies: first author, publication year, region, publication journal, type of study, year of study, sample size, participant demographic details, thoracolumbar fracture level, surgical approach, segmental fixation, time to implant removal, cause of implant removal, and clinical outcomes after implant removal. Data from the research were compared, and disagreements were resolved by consensus among researchers.

Quality Assessment
The Newcastle-Ottawa Scale (NOS) was used to assess the quality of non-randomized studies [29]. The quality of included studies was evaluated in the following three major components: selection of the study group (0-4 points); quality of the adjustment for confounding (0-2 points); and assessment of the outcome of interest in the cohorts (0-3 points). A higher score represented better methodological quality.

Statistical Analysis
Meta-analysis was performed only when there were at least three contrasts available for data synthesis. Risk ratios (RRs) with 95% confidence intervals (CIs) were calculated for dichotomous data, and the mean difference (MD) or standardized mean difference (SMD) along with corresponding 95% CIs were calculated for continuous outcomes. Heterogeneity was assessed using the Cochran Q statistic (p < 0.1) and measured with the I 2 statistic. Meta-analyses were conducted using a random-effects model regardless of heterogeneity. Two-sided p < 0.05 was considered statistically significant. We used Stata version 15 (Stata Corporation, College Station, TX, USA) for data analyses.

Study Selection
The initial search yielded 4051 records; after removing 1558 duplicates, 2493 articles were screened using the title and/or abstract. Of these, 2424 records were eliminated for being irrelevant to our analysis by screening titles and abstracts. The full texts of the remaining 69 articles were retrieved for further assessment. Finally, 35 studies were included in the systematic review . Figure 1 displays a flow diagram that shows the reasons for exclusion at each stage of the selection process.

Study Selection
The initial search yielded 4051 records; after removing 1558 duplicates, 2493 articles were screened using the title and/or abstract. Of these, 2424 records were eliminated for being irrelevant to our analysis by screening titles and abstracts. The full texts of the remaining 69 articles were retrieved for further assessment. Finally, 35 studies were included in the systematic review . Figure 1 displays a flow diagram that shows the reasons for exclusion at each stage of the selection process.

Risk of Bias
The pre-planned risk of bias was not assessed during this systematic review. Due to the present lack of high-quality evidence, case reports and case series studies were predetermined to be included to provide related information on our topic. Even if we also included retrospective cohort studies, retrospective case-control studies, and prospective cohort studies, the quality of the observational studies was not assessed due to the inherent biases associated with these study designs and the lack of a control group in many studies.
In addition, the pre-defined meta-analyses were unfeasible due to insufficient data for these clinical outcomes and considerable clinical heterogeneity and variations in outcome measures.

Pain Intensity after Implant Removal
Of the 27 observational studies, nine [39,41,44,46,49,53,56,59,63] reported significant pain relief after implant removal; of these, four studies [39,41,44,59] reported the decision to remove the pedicle screw instrument due to implant-associated symptoms such as pain or discomfort, while in three studies [46,53,59] the patients were asymptomatic before implant removal. One retrospective cohort study [47] found that 10 of 27 patients had increasing back pain after implant removal, while in another retrospective cohort study [60], 12 of 21 patients reported reduced back pain or discomfort after surgery. In addition, one study [64] found no significant changes after implant removal.

Removal-Related Complications
One case report [30] reported inadvertent screw migration into the retroperitoneal space, while one case series [37] reported that pedicular screws fractured and the threaded parts of the screws were, therefore, left in 1 patient.

Principal Findings
This systematic review showed that dozens of studies focused on the benefits and risks of implant removal after fixation of thoracolumbar burst fractures, and local kyphosis deformity was the most prevalent and most important sequelae after implant removal. However, some studies further confirmed reduced pain intensity and improved segmental motion angle after implant removal. In addition, implant removal-associated complications were not uncommon.

Comparison with Previous Studies
Kweh et al. published a systematic review and meta-analysis addressing a similar topic [65]. This study eventually included 13 articles for qualitative synthesis and six studies for quantitative synthesis. They found no statistically significant difference in sagittal correction loss between implant retention and removal cohorts, and suggested significantly improved pain intensity and ODI scores. They concluded that planned implant removal results in superior functional outcomes without significant differences in kyphotic angle correction loss compared to implant retention in younger patients with thoracolumbar burst fractures who undergo posterior surgical stabilization. In comparison, we included more types of studies to fully elaborate on this clinical dilemma. Although we did not perform a meta-analysis mainly due to the significant clinical heterogeneity among studies, we found similar benefits but also highlighted the potential risks. We further revealed conflicting evidence regarding the management of thoracolumbar burst fractures.

Implication for Clinical Practice
Although implant removal accounts for almost one-third of all elective operations in orthopedics, there remains an ongoing debate concerning the justification for such procedures [32]. The thoracolumbar junction is a transitional zone that constitutes the relatively fixed kyphotic thoracic area and the mobile lordotic lumbar region; therefore, it is a vulnerable region for injury. In theory, when natural bone healing and consolidation of fractured vertebrae has occurred, implant removal should allow complete motion segment preservation, but it is hard to decide for the thoracolumbar junction.

Kyphosis Recurrence
Kyphosis recurrence after implant removal is not uncommon ( Table 2 and Supplementary Material Table S3). Previous studies have suggested that kyphotic recurrence is inevitable during the medium-to long-term period, regardless of the pedicle screw fixation with or without fusion, and the process of kyphotic recurrence may be accelerated after removal of the pedicle screw instrument, which has been reported in case reports and case series, and some of the observational studies [50,56]. However, there remains a lack of robust clinical evidence and long-term follow-up data, and our systematic review found that currently conflicting data was more present, highlighting this clinical dilemma.
In addition, these studies also investigated the mechanism of sagittal correction loss after implant removal. Some studies [39,44] have implicated that failure to support the anterior spinal column and vertebra collapse after implant removal lead to eventual loss of correction; however, more recent studies [38,46,[51][52][53] have found that intervertebral disc collapse and loss of disc height are the main factors contributing to postoperative kyphosis in patients with thoracolumbar burst fracture, no matter with or without vertebroplasty. Patients with incomplete and complete thoracolumbar burst fractures always suffer severely injured endplates and discs, so post-traumatic disc degeneration and height loss when loaded after implant removal are unavoidable. Thus, a mono-segmental fusion is better indicated in cases of expected disc injury to prevent secondary loss of reduction resulting from the collapse of the disc space, especially in younger patients. Removal of the implants may, therefore, not be necessary.
The relatively high incidence of kyphosis recurrence after implant removal may be caused by various factors. The surgical intervention for thoracolumbar burst fracture aims to restore stability, prevent neurological deterioration, attain canal clearance, prevent kyphosis, and provide rapid pain relief. Therefore, sufficient stability is important to avoid postoperative loss of segmental kyphosis correction, regardless of whether fusion is performed. Although the pedicle screw instrument is only to provide temporary fixation of the unstable spine and permanent restoration of spinal stability through achieving a solid fusion as the primary purpose, the pedicle screw instrument may still play an important role in maintaining the reduction, offering rigid fixation, and enhance bony union or fusion after bone healing. A previous study also suggested that the severity of the initial trauma also predicts the loss of correction after implant removal: the more severe the preoperative collapse of the fractured vertebral body is, the higher loss of correction after implant removal has to be expected [51]. In addition, other factors, such as the integrity of the posterior ligamentous complex, are also crucial, and implant removal in patients with non-healing of the posterior ligamentous complex would also induce instability and progressive kyphosis [53,56,66].

Segmental Range of Motion
Improvement of the segmental range of motion has been recognized as one of the major benefits of implant removal, especially in patients who received pedicle screw fixation without fusion. Several previous studies have confirmed the advantages of implant removal for the preservation of segmental motion, which can further alleviate pain and disability [38,[44][45][46][47]58] and lead to decreases in the pain intensity score and ODI score [39,41,44,56,59,63]. Therefore, these clinical benefits after implant removal are measurable and demonstrate that a subgroup of patients would benefit from implant removal, especially when there was no disastrous kyphosis deformity recurrence. Nonetheless, we should also realize that the actual mobility of the segment has possible implications-both positive and negative. The improved range of motion of the fractured segment in the thoracolumbar junction would unload the stress on the adjacent segments but put stress on the fractured vertebra and nearby discs. Hence, the improved segmental range of motion also means an unstable status after implant removal, with a potential risk of recurrent kyphosis deformity induced by destabilization after implant removal [37].

Decision-Making
Removing the pedicle screw instrument after posterior fixation of thoracolumbar burst fractures can effectively restore flexibility and relieve pain, but can also result in the progression of kyphosis. Moreover, it is impossible to predict the recurrence of kyphotic deformity before implant removal, and extra revision surgery might be needed later if patients have severe back pain due to severe kyphotic deformity. Thus, careful consideration should be made before removing the implant.
In most symptomatic cases, the patient is the initiator of pedicle screw removal. Many patients with persistent symptoms tend to blame the metallic implants; they often insist on implant removal and believe this will alleviate their symptoms [67,68]. However, in clinical practice, even in patients who have reported implant-related pain, removing the implant does not guarantee pain relief and may be associated with further complications (such as infection, re-fracture, and nerve damage) and worsening pain [31,32,37,41]. Therefore, patients should be notified of indications for implant removal and understand the uncertainty of expected benefits, potential complications, and inherent risks. On the contrary, implant retention would reduce costs and alleviate exposure to further surgery, but patients should also be informed of the possibility of screw breakage.
Surgeons are the decision-makers of implant removal [18,67]. The decision of implant removal should be predetermined as early as the initial treatment of the thoracolumbar burst fractures and dynamically adjusted according to the patient's clinical status (Figure 2). Careful preoperative evaluation and consideration should be made before removing the implant. First of all, surgeons should review details of the primary thoracolumbar burst fractures, such as the mechanism of injury, the morphology, and classification of the burst fractures, and learn about the first surgical management. Second, surgeons need to assess the fusion of the burst fractures, which is critical but challenging, and even intra-operative exploration demonstrates that a solid fusion cannot promise desired outcomes [67]. Next, for symptomatic patients, surgeons should try to figure out to what extent the patient's pain and discomfort are associated with the pedicle screw instrument, and how much pain relief can be expected from implant removal [69]. For example, postoperative pain may be attributed to instability, root pain, adjacent-level pathology, and factors related to the implant. Very often, the exact cause of post-instrumented pain remains difficult to determine. Finally, communication with patients is essential and crucial than ever before [41]. Patients should be informed thoroughly about the unpredictable outcomes of implant removal to avoid excessively high expectations [41]. Moreover, detailed preoperative evaluation before implant removal is also indispensable. For instance, a CT scan before implant removal would be beneficial for confirmation of posterolateral fusion and preoperative measurement of the bone mineral density of the fractured vertebral body and adjacent vertebral bodies to evaluate the risk of compression fracture after implant removal [31,32]. Based on these careful preoperative clinical evaluations and detailed communication, a decision to remove or retain the implants could be made. The timing of the removal of the implant remains an open question.

Call for Future Studies
The currently available evidence for removing or retaining the pedicle screw instrument in thoracolumbar burst fractures is heterogeneous, limited, and insufficient. Thus, more prospective cohort studies and clinical trials with long-term follow-ups are strongly warranted to provide additional details about the advantages and disadvantages of each option, which would help mitigate the trade-off between the benefits and harms of different treatment options. Second, there is a desperate need to explore the biological mechanisms and clinical determinants of symptomatic and asymptomatic implants, as well as the risk factors and predictive parameters for the recurrence of kyphotic deformity, which will contribute to developing clinical decision rules that may determine which patient subgroup will benefit most from implant removal and which patient subgroup will face more risks [69,70]. Next, future studies should compare the same types of fractures (e.g., incomplete vs. complete burst fractures) when evaluating the outcomes of removing or retaining pedicle screw instruments after thoracolumbar burst fractures, which would help to observe actual clinical outcomes and avoid confusing the effects of fracture types. Additionally, pedicle screw removal is a second surgery performed under general anesthesia, which has substantial economic implications; therefore, a cost-effectiveness analysis should also be performed for policymakers, decision-makers, and other stakeholders [52,69,71].

Limitations
This study has several weaknesses. First, there was substantial clinical heterogeneity among the included studies, including the patient populations (e.g., symptomatic or asymptomatic), the morphology and classification of thoracolumbar burst fractures (e.g., incomplete or complete burst fractures), the severity of injury (e.g., the degree of injury to the discs, the integrity of the posterior ligamentous complex), the treatment strategies of thoracolumbar burst fractures, criteria for implant removal, follow-up duration, etc. These discrepancies reflect the lack of consensus on thoracolumbar burst fractures and compromise the quality of evidence. Second, this study was predetermined to include all kinds of studies, including case reports and case series, which may induce remarkable publication bias, since studies with positive results (e.g., unexpected complications) are more likely to be published in peer-reviewed journals [71]. Third, 25 of 35 included studies were from Asia, mainly from China, Japan, and South Korea, which may also induce bias.

Conclusions
In conclusion, the removal of implants after successful fusion of thoracolumbar burst fractures may be performed effectively to restore flexibility and relieve pain, but it may also lead to the progression of kyphotic deformity and surgical complications. Both surgeons and patients should be aware of the indications and have appropriate expectations of the benefits and risks of implant removal. There was no robust evidence to support the routine removal of pedicle screw instruments after the successful fusion of thoracolumbar burst fractures, which may expose the patients to unnecessary complications and costs. The potential benefits and possible risks should be weighed to support patient-specific clinical decision-making. Further research is warranted to provide more evidence to clarify this issue.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm12062213/s1, Table S1: Search strategies for primary databases; Table S2: Baseline characteristics of the included case reports and case series; Table S3: The reported clinical outcomes in case reports and case series.
Author Contributions: X.W.: Contributed substantially to conception and design, acquisition of data, analysis, and interpretation of data; drafted the article; gave final approval of the version to be published; agreed to act as a guarantor of the work. X.-D.W.: Contributed substantially to conception and design, acquisition of data, analysis, and interpretation of data; drafted the article; gave final approval of the version to be published; agreed to act as a guarantor of the work. Y.Z.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. Z.Z.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. J.J.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. G.L.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. J.L.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. J.S.: Contributed substantially to the acquisition and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. Y.S.: Contributed substantially to conception and design, acquisition of data, analysis, and interpretation of data; revised it critically for valuable intellectual content; gave final approval of the version to be published; agreed to act as a guarantor of the work. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement:
This was a systematic review, and no ethics approval and consent to participate were required.

Informed Consent Statement: Not applicable.
Data Availability Statement: The datasets used and analyzed during the study will be available from the corresponding authors on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.