Anterior versus posterior spinal fusion for Lenke type 5 adolescent idiopathic scoliosis: a systematic review and meta-analysis of comparative studies

To review and compare clinical and radiologic outcomes between anterior spinal fusion (ASF) and posterior spinal fusion (PSF) for the treatment of Lenke type 5 adolescent idiopathic scoliosis (AIS). A systematic review was performed according to Preferred reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines. All level I–III evidence studies investigating the clinical and radiologic outcomes of ASF and PSF for the treatment of Lenke type 5 AIS were included. Nine studies (285 ASF patients, 298 PSF patients) were included. ASF was associated with a significantly lower number of levels fused compared with PSF (p < 0.01) with similar immediate and long-term coronal deformity correction (p = 0.16; p = 0.12, respectively). PSF achieved a better correction of thoracic hypokyphosis in one study and lumbar hypolordosis in three studies. PSF was associated with a significant shorter length of stay (LOS) compared with ASF (p < 0.01). One long-term study demonstrated a significantly higher rate of proximal junctional kyphosis (PJK) with PSF compared with ASF. There were no significant differences in major complication or re-operation rates. For the treatment of Lenke type 5 AIS, there is moderate evidence to suggest that ASF requires a lower number of instrumented levels to achieve similar immediate and long-term coronal deformity correction compared with PSF. There is some evidence to suggest that PSF may achieve better thoracic and lumbar sagittal deformity correction compared with ASF. There is some evidence to suggest a higher incidence of PJK at long-term follow-up with PSF compared with ASF. ASF is associated with a longer post-operative LOS compared with PSF.


Introduction
Adolescent idiopathic scoliosis (AIS) is the most commonly acquired spinal condition among adolescents and presents with a variety of clinical presentation and progression of curvature [1]. The most commonly utilized classification system described by Lenke et al. defines thoracolumbar and lumbar major with a non-structural thoracic curvature as type 5 AIS [2]. Selective spinal fusion refers to the instrumentation and fusion of only the major curvature to maintain the mobility of the minor curvature.
In the past two decades, there has been a multitude of evidence that show adequate deformity correction and clinical results with both anterior and posterior selective spinal fusion for Lenke type 5 AIS [3][4][5][6][7][8][9][10]. Both single-and dual-rod instrumentation are commonly used for anterior spinal fusions (ASF), with recent evidence demonstrating no significant difference between the two instrumentation methods [11]. Posterior spinal fusions (PSF) can be also performed with either pedicle screw or hook instrumentation, or a hybrid method that combines these constructs, although recent evidence have demonstrated significantly better deformity correction with a pure pedicle screw construct [12,13]. However, the optimal methodology for selective fusion between ASF and PSF remains a subject of debate amongst spine surgeons.
There have been two systematic reviews in the recent literature that have evaluated the efficacy of anterior and posterior selective fusion methods of Lenke type 5 AIS [14,15]. However, these studies are limited by the inclusion of articles that are mostly non-comparative in nature within their analysis. The most recent meta-analysis by Lin et al. analyzed two comparative studies evaluating selective ASF versus PSF for thoracolumbar and lumbar major AIS; however, the analysis also included Lenke type 6 AIS and failed to include several earlier comparative studies and recently published studies [15]. Thus, the purpose of this study was to perform a comprehensive systematic review of comparative studies to review and compare clinical and radiologic outcomes between selective ASF and PSF for the treatment of Lenke type 5 AIS. The authors hypothesized that there were no significant differences in clinical and radiologic outcomes between the two interventions.

Design and search strategy
This systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 02/14/2021 (Registration ID: CRD42021237462). The authors were unable to identify any similar prior systematic reviews or meta-analyses within the PROSPERO. The search was conducted and reported using the protocol described in the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [16].
Two authors conducted separate searches using the following medical databases on August 14, 2021: PubMed (1966, SCOPUS (1966-present), and Ovid MED-LINE (1946-present). To ensure a stringent search strategy of relevant literature, keywords including "anterior", "posterior", "scoliosis", "Lenke 5", and "fusion" were combined with Boolean operators to develop a search protocol. A hand search of the included references was also performed to further minimize unintentional exclusion of relevant studies.

Eligibility criteria
All Level I, II, and III evidence (as defined by the Oxford Centre for Evidence Based Medicine [CEBM]) therapeutic studies published in the English language that investigated clinical outcomes between anterior and posterior instrumented fusions for the treatment of Lenke type 5 AIS were included [17].
Studies were excluded if they were studies of non-Lenke type 5 AIS, non-comparative studies (i.e., only one treatment method studied), cadaveric studies, basic science studies, animal studies, diagnostic studies, economic studies, prognostic studies, letters to editors, review articles, editorials, and surveys. In the situation of duplicate studies from the same author(s) and/or institution(s) reporting on the same or overlapping subjects, only the most recent study with the longest follow-up was included; however, certain data from the older studies were analyzed if not reported in the newer studies. These exceptions were disclosed within the results section.

Quality assessment and data extraction
Two authors independently reviewed all studies using previously recommended methodology [18]. If two or more separate studies utilized the same patient population, the study with the longer follow-up, higher level of evidence, greater number of patients, and/or greater clarity of methods and results was included. The level of evidence (CEBM), study design, and the methodological quality of each study were graded using the Modified Coleman Methodology Score (MCMS) [17,19]. For all included studies, the overall Strength-of-Recommendation Taxonomy (SORT) score and the Grading of Recommendations Assessment, Development and Evaluation (GRADE) score were calculated [20,21]. Patient demographics (age, gender, diagnosis), re-operation rates, complication rates (major and minor), patient-reported outcome scores, radiologic outcomes, and the study authors' overall conclusion were extracted from each individual study. For extracting data from digital plots, WebPlotDigitizer version 4.4 (Ankit Rohatgi, Pacifica, CA, USA, https:// autom eris. io/ WebPl otDig itizer) was utilized to best estimate the reported data using prior described methods [22,23]. Metaanalysis was performed using outcome data without significant heterogeneity; a systematic review with best-evidence synthesis was chosen as the synthetic review type for outcome measures with significant heterogeneity [24].

Statistical analysis
Data analysis was performed using SPSS statistical software (Version 25.0; SPSS, Inc, Chicago, IL). The mean and standard deviation (SD) were integrated into the weighted mean difference (WMD) for continuous parameters. Two-tailed student t test was used to analyze continuous data and the Chi-Square Test was used to analyze categorical data. Revman 5.2 software (Cochrane Collaboration, Copenhagen, Denmark) was used for meta-analysis. A p value ≤ 0.05 was considered statistically significant.

Risk of bias assessment
Two authors used the Revised Cochrane Risk of Bias Tool for Randomized Trials (RoB 2) and the Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) assessment tool to perform risk of bias assessment of each included study [25,26].

Study characteristics and patient demographics
The preliminary search identified 1260 studies with 47 found to be duplicates. Of the remaining 1213 studies, nine met all inclusion and exclusion criteria (Fig. 1). One randomized study was level II evidence, and eight non-randomized studies were level III. According to MCMS, one study was rated as good (scores between 70 and 84) and eight studies were rated as fair (scores between 55 and 69) [21]. The overall SORT score was A and GRADE score was B [22,23]. According to RoB2 and ROBINS-I, the overall risk of bias was low for one study and moderate for eight studies ( Table 1). These nine studies contained: 583 patients with Lenke type 5 AIS treated with either (i) ASF (285 patients) or (ii) PSF (298 patients) [27][28][29][30][31][32][33][34][35].
One study investigated changes in pre-and post-operative apical vertebral rotation (AVR) and found no significant differences between ASF and PSF [28].

Discussion
The surgical treatment of AIS aims to correct the coronal and sagittal deformities while preventing the progression of the scoliotic curvatures. Due to its less invasive nature and the relative maintenance of spinal flexibility, selective fusion is often more favorable than non-selective fusions when properly indicated. However, anterior versus posterior approach to selective fusions have historically been debated. To our knowledge, this is the first comprehensive systematic review of comparative studies that compare clinical and radiologic outcomes between selective ASF and PSF for the treatment of Lenke type 5 AIS. The authors hypothesized that there were no significant differences in clinical and radiologic outcomes between the two interventions. However, the extracted evidence supported significant advantages and disadvantages to each intervention. ASF was found to be superior compared to PSF in terms of the number of levels fused with less instrumented levels required to obtain similar amounts of major and compensatory curvature correction. This was partially consistent with the most recent meta-analysis by Lin et al. which also found similar coronal deformity correction between the two interventions; however, they did not report a significant difference in the number of instrumented levels [15]. This discrepancy may stem from their analysis including merely two comparative studies with Lenke type 5 within their analysis compared to our more comprehensive analysis with eight comparative studies. One of the included studies by Dong et al. chose fusion levels based on the Hall fusion selection principles for the ASF group, while the upper end vertebra (UEV) and lower end vertebra (LEV) were utilized for the PSF group, which may have also contributed to the higher overall fusion levels for the PSF group [32]. Prior studies comparing ASF and PSF for the treatment of thoracic major curvatures also reported partially similar results. The most recent comparative study by Nohara et al. found mean instrumented levels of 10.7 ± 1.1 with PSF compared to 7.2 ± 0.6 with ASF while achieving a significantly better initial deformity correction with ASF. However, they also found a greater loss of correction with ASF at 10-year follow-up [36]. Although our analysis also demonstrated less number of instrumented levels required with ASF for Lenke type 5 compared with PSF, we found no differences in immediate or long-term post-operative correction between the two treatment methods.
Prior studies have also found no differences in the incidence of PJK between ASF and PSF for Lenke type 5 AIS [14,15]. However, merely one study in the literature performed a 10-year comparative follow-up after ASF and PSF, which reported a significantly lower incidence of PJK at 10-year follow-up compared with PSF (16.0% vs. 45.0%, p < 0.01) [33]. Although further long-term comparative studies are necessary to truly elucidate the absolute difference in PJK incidence between the two interventions, the current PSF, on the other hand, was found to be partially superior to ASF in terms of correction of thoracic hypokyphosis and lumbar hypolordosis. This was consistent with the prior study by Lin et al. who found similar results with both thoracic major and thoracolumbar/lumbar major curvatures [15]. One of the nine studies included in our analysis reported a significantly higher correction of thoracic hypokyphosis with PSF [32], and three studies reported a better correction of lumbar hypolordosis with PSF [32,34,35]. Although there is limited evidence, these results may demonstrate that patients with thoracic hypokyphosis and lumbar hypolordosis that may require a larger sagittal correction may better benefit from PSF; however, further longterm studies are necessary to better elucidate the efficacy of correcting the sagittal deformity between ASF and PSF.
Our analysis excluded one classic study by Suk et al. due to their inclusion of both Lenke type 5 and 6 AIS. This study compared the use of Zielke ventral derotation system (VDS) versus the Cotrel-Dubousset Instrumentation (CDI) and also found that a lower number of instrumented levels is required with ASF to achieve similar coronal deformity correction compared with PSF [37]. Interestingly, the authors  Table 5 Complications/re-operations Study Hee et al. [27] Wang et al. [28] Geck et al. [29] Li et al. [30] Abel et al. [31] Dong et al. [32] Lim et al. [33] Li et al. [34] Miyanji et al. also demonstrated an overall loss of lumbar lordosis with ASF using VDS. This was in contrast to our included studies which demonstrated neither loss nor significant improvement in lumbar lordosis with ASF. This may be reflective of the ability of newer anterior instrumentation systems to better maintain lumbar lordosis. Our analysis also found that PSF is superior compared to ASF in terms of a shorter post-operative LOS and a lower incidence of post-operative transient lumbar sympathetic deficit. These findings have yet to be reported in previous metaanalyses. However, a shorter LOS has been reported among patients undergoing selective PSF compared with ASF for thoracic major curvatures mainly due to the need for longer post-operative monitoring for the more intra-thoracically and intra-abdominally invasive nature of ASF [38]. Although there was a significantly higher rate of post-operative transient sympathetic deficit seen with ASF (p < 0.01), all 16 cases were seen in merely one study by Wang et al., which all resolved by six months after surgery [28].
Another study by Demura et al. that was excluded from our analysis due to inclusion of both Lenke types 5 and 6 AIS compared pulmonary function between the two treatment methods and found no long-term difference in forced vital capacity (FVC) or forced expiratory volume in 1 s (FEV 1 ) [39]. However, they also found that ASF is associated with decreased pulmonary function during the immediate postoperative months. The authors speculate that this is largely due to the initial chest cage disruption involved during the anterior approach. Although this led to no major related difference in complication rates, these results suggest the necessity for careful pre-operative patient selection, particularly with patients with pre-existing pulmonary conditions.
There are several limitations to this systematic review. First, the heterogeneity of the included studies with variable recorded outcome measures precluded a meta-analysis of a majority of the outcome scores. Although high-quality comparative studies were included in this review, a majority of the studies were level III evidence, and no randomized controlled trials were identified. While four studies reported long-term functional outcomes without significant difference between the interventions, none of the included studies reported short-term functional outcomes. Furthermore, these results may also be surgeon specific and certain outcomes may vary at other institutions not reported within this analysis. At last, our stringent search protocol and limiters may have possibly led to the unintentional omission of other relevant studies on this topic, including those published in the non-English language. Wang et al. [28] Geck et al. [29] Li et al. [30] Abel et al. [31] Dong et al. [32] Lim et al. [33] Li et al. [34] Miyanji et al. [35] Treatment groups

Conclusion
For the treatment of Lenke type 5 AIS, there is moderate evidence to suggest that ASF requires a lower number of instrumented levels to achieve similar immediate and longterm coronal deformity correction compared with PSF. There is some evidence to suggest that PSF may achieve better thoracic and lumbar sagittal deformity correction compared with ASF. There is some evidence to suggest a higher incidence of PJK at long-term follow-up with PSF compared with ASF. ASF is associated with a longer postoperative LOS compared with PSF.