Determining the utility of three-column osteotomies in revision surgery compared with primary surgeries in the thoracolumbar spine: a retrospective cohort study in the United States
Article information
, Oluwatobi O Onafowokan1, Ankita Das1, Jamshaid Mahmood Mir1, Oscar Krol1, Peter Tretiakov1, Rachel Joujon-Roche1, Bailey Imbo1, Salman Ahmad1, Stephane Owusu-Sarpong1, Jordan Lebovic1, Shaleen Vira2, Andrew J Schoenfeld3, Muhammad Burhan Janjua4, Bassel Diebo5, Renaud Lafage6, Virginie Lafage7, Peter Gust Passias1
Abstract
Study Design
Retrospective cohort study.
Purpose
To determine the incidence and success of three-column osteotomies (3COs) performed in primary and revision adult spine deformity (ASD) corrective surgeries.
Overview of Literature
3COs are often required to correct severe, rigid ASD presentations. However, controversy remains on the utility of 3COs, particularly in primary surgery.
Methods
Patients ASD having 2-year data were included and divided into 3CO and non-3CO (remaining ASD cohort) groups. For the subanalysis, patients were stratified based on whether they were undergoing primary (P3CO) or revision (R3CO) surgery. Multivariate analysis controlling for age, Charlson comorbidity index, body mass index, baseline pelvic incidence–lumbar lordosis, and fused levels evaluated the complication rates and radiographic and patient-reported outcomes between the 3CO and non-3CO groups.
Results
Of the 436 patients included, 20% had 3COs. 3COs were performed in 16% of P3COs and 51% of R3COs. Both 3CO groups had greater severity in deformity and disability at baseline; however, only R3COs improved more than non-3COs. Despite greater segmental correction, 3COs had much lower rates of aligning in the lumbar distribution index (LDI), higher mechanical complications, and more reoperations when performed below L3. When comparing P3COs and R3COs, baseline lumbopelvic and global alignments, as well as disability, were different. The R3CO group had greater clinical improvements and global correction (both p<0.04), although the P3CO group achieved alignment in LDI more often (odds ratio, 3.9; 95% confidence interval, 1.3–6.2; p=0.006). The P3CO group had more neurological complications (30% vs. 13%, p=0.042), whereas the R3CO tended to have higher mechanical complication rates (25% vs. 15%, p=0.2).
Conclusions
3COs showed greater improvements in realignment while failing to demonstrate the same clinical improvement as primaries without a 3CO. Overall, when suitably indicated, a 3CO offers superior utility for achieving optimal realignment across primary and revision surgeries for ASD correction.
Introduction
The prevalence of adult spinal deformity (ASD) is increasing and concomitantly the frequency of corrective surgical procedures [1]. A consequent major challenge is the treatment of patients presenting with rigid ASD, including deformities such as hyperkyphosis, scoliosis, flatback syndrome, and L5 spondyloptosis [2,3]. This pathology generally warrants complex procedures in older, frail, multimorbid groups to achieve the desired deformity correction and is often in the context of revision surgery [4,5].
Surgical management of ASD aims to improve pain and function while preventing the progression of deformity and associated symptoms. Patients with ASD and rigid and significantly malaligned vertebral curves often require a three-column osteotomy (3CO) as part of their surgical management [6]. 3CO variations include the Smith–Peterson osteotomy, pedicle subtraction osteotomy, and vertebral column resection. 3COs are indicated for cases with the highest osteotomy grades, according to the classification system by Schwab et al. [7], and offer the greatest scope for deformity correction in rigid ASDs. Given the extent of bony resection and surgical invasiveness required, 3COs are also associated with a significant incidence of complications [4,5]. Of particular concern are mechanical complications such as postoperative kyphotic decompensation syndrome and proximal or distal junctional kyphosis and failure [8,9]. Nevertheless, the notable and sustained efficacy of severe deformity correction with 3COs has been repeatedly demonstrated [4,5,10,11].
Despite the extensive study of 3CO techniques, studies examining the performance of 3COs in primary compared with revision surgery are limited. In this context, we sought to compare the outcomes of 3COs when used in primary and revision correction procedures for ASD.
Materials and Methods
Data source and study design
A retrospective analysis of a prospectively-enrolled, single-center database of patients with ASD enrolled between 2012 and 2020 was performed. The approval from the Ethics Committee of NYU School of Medicine was obtained before the study commenced (protocol code S13-00422), and all patients provided informed consent. The general enrollment criteria and the definition of ASD for this dataset were described in previous publications, and this cohort was also used to analyze various aspects of ASD investigation and management [12]. Patients included in the dataset had at least one of the following radiographic deformity parameters: sagittal vertical axis (SVA) >5 cm, Cobb angle >20°, pelvic tilt (PT) >25°, or thoracic kyphosis (TK) >60°.
Patients with ASD were included in this study if they were undergoing surgical intervention with complete demographic, radiographic, and patient-reported outcomes data up to at least 2 years postoperatively. In addition, patients were included if deemed to have a severe global deformity, which was defined as having a pelvic incidence–lumbar lordosis (PI–LL) mismatch of ≥20°. All surgical procedures were performed with at least two surgeons present. Surgical indications included neurological deficit, persistent severe pain despite conservative measures, spondylotic myelopathy, and intolerable, functionally-limiting postural deformity. 3CO was also indicated for rigid thoracolumbar deformity with PI–LL of ≥30°. Patients undergoing primary surgery were stratified according to whether they were undergoing a 3CO (P3CO+) or a lower-grade osteotomy (P3CO−). Similarly, patients undergoing revision surgery were stratified by undergoing a 3CO (R3CO+) or lower-grade osteotomy (R3CO−).
Data collection and radiographic assessment
Demographic data collected included age, sex, race, and body mass index (BMI). Collected health-related quality of life (HRQL) metrics included the Oswestry disability index (ODI) and the Scoliosis Research Society 22-item questionnaire (SRS-22). The minimum clinically important difference in ODI was set at 11% as reported previously [13]. Lateral spine radiographs were used to assess radiographic parameters at baseline and all follow-up timepoints. All images were analyzed with SpineView (ENSAM; Laboratory of Biomechanics, Paris, France) [14–16]. Spinopelvic radiographic parameters assessed included measures commonly used in ASD assessment, such as PT, PI–LL mismatch, SVA, T1 pelvic angle (T1PA), and L1 pelvic angle (L1PA) [17,18]. These parameters are illustrated in Fig. 1 [18].
Illustration of regional, lower extremity, and global spinal radiographic parameters (SVA, PT, PI, LL, SFA, KA, AA, PS, and GSA) [18]. SVA, sagittal vertical axis; PT, pelvic tilt; PI, pelvic incidence; LL, lumbar lordosis; SFA, sacrofemoral angle; KA, knee angle; AA, ankle angle; PS, posterior pelvic shift; GSA, global sagittal angle.
The global alignment and proportion (GAP) score was used to assess deformity correction [19]. This realignment schema includes various spinopelvic components to assess the proportionality of alignment. The GAP score has been validated in patients with ASD, with lower scores being associated with poorer postoperative outcomes [20,21]. The segmental utility ratio (SUR) assessed the relative segmental correction achieved. This was defined as the segmental correction at the intervention level divided by the overall correction in lumbar lordosis (L1–S1) divided by the number of thoracolumbar interventions (interbody fusions + Smith–Peterson osteotomy/3-CO). The SUR was developed to assess segmental harmony, which is described as the absence of any significant differences in lordosis between adjacent disc pairs [22].
Complication assessment
Radiographic, mechanical, and neurological complications were the focus of this analysis. Radiographic complications included proximal junctional kyphosis (PJK) and proximal junctional failure (PJF). PJK was defined by a PJK angle of <–10° and a PJK angle difference of <–10° from the baseline at any time point up to 2 years. PJF was defined using the criteria of Lafage et al. [17]: a PJK angle of <–28° and a difference in the PJK angle of <–22° from the baseline at any follow-up time point up to 2 years. Mechanical complications included any complications related to the implant: implant prominence or malposition, implant failure, interbody dislocation, screw nerve impingement, screw fracture, rod dislocation, and rod fracture. Mechanical complications were classified as major if they required invasive interventions for correction or caused prolonged or permanent morbidity/mortality. Neurological complications were diagnosed if any new postoperative neurological deficits persisted beyond 6 weeks.
Statistical analysis
The primary outcome was postoperative radiographic alignment. Baseline demographic, radiographic, surgical, and clinical data were compared between the cohorts using chi-square test and t-test for categorical and continuous variables, respectively. Patients were first separated into groups based on whether they were undergoing a 3CO during the procedure or not and then whether they were undergoing a primary or revision surgery. Multivariable linear and regression analyses accounting for age, sex, BMI, osteoporosis, baseline deformity (via PI–LL mismatch), and fused levels surgically evaluated the complication rates and radiographic and patient-reported outcomes between groups at 2 years. All p-values <0.05 were considered significant. All statistical analyses were conducted using IBM SPSS Statistics ver. 28.1 (IBM Corp., Armonk, NY, USA).
Results
Baseline characteristics
Of the 436 patients included in this study, 88 (20.2%) underwent a 3CO (16.3% pedicle subtraction osteotomy and 2.8% vertebral column resection), and 52.1% of the remaining cohort underwent a lower-grade osteotomy. Of the total cohort, a 3CO was performed in 16.1% of primaries (n=27, P3CO) and 50.8% of revisions (n=61, R3CO). The mean age in this cohort was 53.8±20.2 years, 68% were female, and the mean BMI was 26.0±5.9 kg/m2. Baseline characteristics are displayed in Table 1.
Baseline characteristics between primary and revision three-column osteotomy groups
P3COs
The P3CO+ group had double the EBL of the P3CO− group (2,635 mL versus 1,245 mL, p=0.001), despite similar operative times (475.3 minutes versus 465.7 minutes, p=0.35), and were more likely to be accompanied by a posterior support rod (odds ratio, 20.8; 95% confidence interval [CI], 4.2–122.5; p=0.003). The P3CO+ group demonstrated higher rates of major complications and were more likely to endure a neurological complication (OR, 12.1; 95% CI, 7.36–25.49; p=0.032). No significant differences in HRQL outcomes were found the between P3CO+ and P3CO− groups. Despite similar correction in deformity (by PI–LL and T1PA correction), the P3CO+ group demonstrated higher rates of achieving alignment in the GAP score lumbar distribution index (68% versus 49%, p=0.01), GAP spinopelvic parameters (69% versus 50%, p=0.03), and overall proportioning in GAP (58% versus 41%, p=0.045).
R3COs
The R3CO+ group demonstrated greater deformity correction than the R3CO− group (by PI–LL magnitude: −29.4° versus −17.1°, p=0.025). No significant differences were found in complications between the two groups.
Similar to the P3CO+ group, multivariable regression analyses controlling for levels fused, baseline deformity, and disability revealed that the R3CO+ group less often achieved the MCID in ODI by 2 years compared with the R3CO− group (OR, 0.44; 95% CI, 0.21–0.78; p=0.021). No differences in GAP score proportionality were found between the R3CO+ and R3CO− groups.
P3COs versus R3COs
The P3CO+ and R3CO+ groups had greater severity of deformity and disability at baseline than their non-3CO counterparts (Table 2). Both groups did not differ in the use of interbodies/support rods and segmental or overall correction in GAP lumbopelvic and global parameters. The P3CO+ group had higher rates of aligning in GAP LDI (65% versus 36%, p=0.015) (Table 3). The P3CO+ group also had significantly longer lengths of stay (9.6 days versus 7.1 days, p=0.043), admissions to the surgical intensive care unit (76% versus 57%, p=0.025), and higher neurological complications rates (34% versus 13%, p=0.001). The R3CO group had higher mechanical complication rates (41% versus 21%, p=0.005).
Spinopelvic radiographic outcomes from baseline to early follow-up and early follow-up to 2 years
Focal and overall correction by revision status
Discussion
Despite the notable associated morbidity risk, numerous studies have demonstrated the efficacy of 3COs at achieving significant deformity correction in patients with ASD presenting with severe, rigid malalignments in the sagittal, coronal, and/or axial planes [7,8]. The ability to perform these osteotomies asymmetrically can allow for deformity correction in the sagittal and coronal planes simultaneously [7]. The vast majority of literature on 3COs involves patients who have previously undergone spine surgery, with some patients having previously undergone multiple surgical interventions before 3COs [8]. In these patients, 3COs can be challenging, requiring extricating previous instrumentation and the potential obliterated/variable anatomy caused by scarring from prior surgery. 3COs are more recently employed in primary surgery. A potential reason for this may be the desire to avoid anterior spinal approaches in patients with severe deformity and thus avoid the notable associated risks of vascular, genitourinary, and bowel injuries [23–25].
Several published studies have reported on patients undergoing 3COs as part of their primary procedure. Raad et al. [26] reported a series of 197 patients undergoing 3COs for ASD, in which 69 patients (35%) underwent P3COs. Kelly et al. [27] reported on 132 patients undergoing 3COs for ASD, in which 17 patients (13%) underwent P3COs. In the study by Smith et al. [28] of 82 patients who underwent 3COs, 12 (15%) underwent a P3CO. In a series of 105 patients undergoing 3COs for ASD, Auerbach et al. [29] reported 16 (15%) undergoing a P3CO. However, none of these studies isolated the patients undergoing P3COs and assessed their specific outcomes.
Apparently, patients undergoing 3COs as part of their primary surgical intervention are a significant minority of all patients undergoing 3COs. In this study, patients undergoing P3COs and R3COs, as well as those undergoing primary and revision lower-grade osteotomies, were investigated. Similar to previously reported rates, patients undergoing a P3CO also make up the minority in the present study and comprised 31% of the 3CO groups. The P3CO+ group had similar HRQL outcomes to their P3CO− counterparts, as reported in previous studies [30]. Despite similar HRQL outcomes and deformity correction, the P3CO+ group had greater blood loss and higher complication rates and was more likely to develop a neurologic complication than the P3CO− group. The finding of increased blood loss and higher complication rates is perhaps not surprising because of the much higher invasiveness level associated with 3COs [8,31]. Given the lack of significant differences in HRQLs and radiographic outcomes, our findings indicate that the use of 3COs in primary surgery should be minimized wherever possible.
When comparing the P3CO and R3CO groups, no differences in alignment at 2 years were found. In this study, 3COs were predominantly performed at the L3 level, in line with previous studies [8–10]. After analyzing correction at various osteotomy sites, no differences were found between the overall lordosis correction when 3CO was performed as primary or revision surgery. The R3CO+ group had higher rates of mechanical complications than the P3CO+ group. The increased infection rates in revision cases is well-reported in the literature and likely has a multifactorial etiology, including higher blood loss, surgical invasiveness, and increased operative times [32–34]. Increased risk of mechanical complications in the R3CO+ group is perhaps unsurprising, and a potential reason for this is the decreased bone purchase for re-instrumentation, potentially contributing to higher rates of delayed union and/or non-union [31,35]. However, the P3CO+ group had longer hospital stays, higher rates of intensive care unit admissions, and higher rates of neurological complications.
3COs are undoubtedly a significant technical challenge to execute [31]. Studies have demonstrated that even patients with postoperative complications have still displayed significant improvements in quality-of-life metrics after 3COs, even those requiring reoperation [10,11,29]. Although 3CO implementation in primary surgery has increased, overall 3CO usage and associated mechanical complications has declined, reflecting improved understanding of 3CO utility and technique enhancement [8]. Our study’s findings indicate that patients who underwent revision surgery may demonstrate greater benefit from 3COs than those who underwent primary surgery.
This study has limitations. First, the utilization of a retrospective analysis of patients undergoing operation by a single surgeon incurs the risk of confounding by indications and expertise bias. Second, retrospective studies cannot determine causality, and we recommend delving into the effect of other factors, such as age, frailty, BMI, and mineral bone density at the operative levels, in the context of lower extremity compensation [17,36]. Given the exploratory, retrospective approach of this study, we do not recommend stratifying surgical candidacy or solely determining a treatment plan based on these results. They may be used as an integrative tool for surgical planning outside of the current parameters and factors currently utilized. Furthermore, we were unable to explain the higher incidence of neurologic deficits or longer length of stay in the P3CO group when compared with the R3CO group. Future studies isolating and analyzing the P3CO group may shed some light on this. We also did not perform a cost analysis, which may have contributed greater clarity to delineating the utility of 3COs in primary and revision surgeries. Finally, we encourage that these findings be replicated and validated to fully determine their true effects on clinical practice before more definitive substantiations are possible.
Conclusions
3COs possess significant ability for deformity correction in patients with severe spine deformity. However, patients experience greater benefit from these highly invasive techniques when they are performed as part of revision surgeries. Therefore, it may be more prudent to utilize lower-grade osteotomies when planning primary surgery for patients with rigid thoracolumbar deformities because of the inherent risks.
Key Points
The utility of three-column osteotomies (3CO) remains controversial, especially in primary adult spine deformity (ASD) realignment surgery.
This study investigated 3CO use in patients undergoing primary or revision surgery for ASD.
Patients undergoing 3CO demonstrated greater improvements in alignment metrics, but did not experience the same significant improvements in patient-reported outcomes.
Notes
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
Author Contributions
Study design: TKW, OOO, AD, JMM, OK, PT, RJR, BI, SA, SOS, JL, SV, AJS, MBJ, BD, RL, VL, PGP. Data acquisition: TKW, OOO, AD, JMM, OK, PT, RJR, BI, SA, SOS, JL, SV, AJS, MBJ, BD, RL, VL, PGP. Analysis and interpretation of data: TKW, OOO, AD, JMM, OK, PT, RJR, BI, SA, SOS, JL, SV, AJS, MBJ, BD, RL, VL, PGP. Active involvement in drafting and critical revision of manuscript: TKW, OOO, AD, JMM, OK, PT, RJR, BI, SA, SOS, JL, SV, AJS, MBJ, BD, RL, VL, PGP. Provided final approval of version to be published: TKW, OOO, AD, JMM, OK, PT, RJR, BI, SA, SOS, JL, SV, AJS, MBJ, BD, RL, VL, PGP.