Analysis of The Correction Capability of Oblique Lumbar Interbody Fusion Approach to Improving The Lenke-Silva Classication in First-Stage MIS Adult Spinal Deformity Corrections

A two-stage minimally invasive surgery (MIS) protocol with oblique lumbar interbody fusion (OLIF) combined with posterior instrumentation is becoming popular in the treatment of adult spinal deformities (ASDs) because of lower complication rates and less blood loss when compared with the open traditional posterior approach. The objective of this study is to determine the ecacy of correction and the causes of suboptimal correction in two-stage OLIF. This retrospective study included 27 patients who underwent both two-stage and single-stage OLIF with posterior instrumentation for treatment of ASD. Patients treated with two-stage OLIF were sorted into Group A (Improved Lenke-Silva classication after the rst-stage OLIF) and Group B (Unchanged Lenke-Silva classication after the rst-stage operation) to evaluate the correction ecacy of OLIF in a two-stage MIS protocol. Statistical analyses were performed to compare the clinical and radiological outcomes. The causes of complications and suboptimal corrections (Group B) in patients treated with the two-stage MIS protocol were analyzed. All 27 patient patients showed signicant improvement (P < 0.05) in the visual analog scale, the Oswestry Disability Index, and EuroQol. A total of 14 patient were treated with the two-stage protocol, with ten patients included in group A and four patients with insucient correction angles included in group B. The radiographic outcome of Group A showed signicant corrections (P < 0.05) in pelvic tilt, pelvic incidence-lumbar lordosis (PI-LL) mismatch, sagittal vertical axis, and max Cobb angles. In group B, surgical limitations of OLIF were the cause of unsatisfactory correction in two patients because the deformities involved the T-L junction, which was not accessible using the approach. Severe preoperative PI-LL mismatch (41°) and cage subsidence with an anterior endplate fracture were the cause of poor correction in two other patients. involving the T-spine or T-L junction, severe PI-LL mismatches, and cage subsidence are the possible causes of suboptimal corrections with OLIF using the two-stage protocol. position were obtained under three conditions: preoperative, post-OLIF (For the 2-stage protocol), and post-posterior instrumentation. We measured the following parameters with Surgimap (Nemaris, Inc.) when assessing the radiographs we assessed: the sagittal vertical axis (SVA), the max Cobb angle, coronal balance (coronal C7-plumb line), pelvic incidence (PI), pelvic incidence-lumbar lordosis (PI-LL), and lumbar lordosis (LL). All patients were classied using the Lenke-Silva classication preoperatively, post-OLIF (for the 2-stage protocol), and post-posterior instrumentation.[2] are the possible causes of suboptimal corrections with OLIF using the two-stage protocol.


Background
The incidence of adult spinal deformity (ASD) had been increasing as the elderly population grows. The traditional posterior approach for treatment of ASD may cause destabilizing effects on the spine and cause disruption of the posterior tension band. [1] Posterior osteotomies have been considered the standard method for surgical correction of ASD because the procedure can effectively improve coronal and sagittal alignment. [2][3][4] However, this method comes with a high risk of pseudoarthrosis, instrumentation failure, or rod breakage, and the complication rate can reach as high as 40%. [5,6] In the past decade, retroperitoneal approaches, such as anterior or oblique lumbar interbody fusion (ALIF or OLIF), have been developed to minimize the need for extensive dissection and trauma to muscles related to the posterior approach. Minimally invasive surgical (MIS) approaches are playing an increasing role in the treatment of ASD because such approaches spare the patient from severe complications associated with posterior osteotomies. [7][8][9][10] The advantages of MIS techniques include decreased blood loss, less postoperative pain, and a shorter length of stay in the hospital. Furthermore, both cadaver and MRI studies published by Davis et al. and Molinares et al. showed that entrance to the vertebral disc through the oblique lateral corridor is safe and simple to perform. [11,12] Thus, a variety of operative protocols have been adopted to reduce complications rates. In our hospital, we have been using a two-stage operative protocol over traditional posterior open surgery since 2018. The protocol involves OLIF during the rst stage followed by posterior instrumentation in the second stage, where the duration between the stages depends on the patient's physiological status and neurological symptoms.
The purpose of this study is to retrospectively analyze the e cacy of a two-stage OLIF correction and the causes of suboptimal correction or failure using the Lenke-Silva Classi cation as an objective tool. In previous studies, an MIS correction was considered appropriate for treating cases with relatively mild or moderate ASD, but the ability to achieve su cient corrections (changes in the Lenke-Silva classi cation level) and possible factors for failure of corrections when using this approach have not yet been thoroughly analyzed, which also directly affects whether to extend fusion to the thoracic spine (Lenke-Silva level V) or to perform an osteotomy (Lenke-Silva level VI) in the second stage. [2,13,14] Our hypothesis is that OLIF could contribute signi cantly to ASD correction during two-stage MIS surgery.

Patient enrollment
After approval from the Institutional Review Board (IRB No: CE20218B), we retrospectively reviewed our prospectively collected MIS ASD database for patients who had received staged OLIF (ALIF used for pathologies involving L5-S1) as the approach to ASD correction surgery between May 2018 and June 2021. The inclusion criteria were patients with adult spinal ASDs treated using OLIF or ALIF for correction involving L5-S1 as the main correction method. The exclusion criteria were surgical correction without the use of the OLIF approach, deformities caused by infections or malignancies, patient age less than 20 years old, and incomplete imaging data. A owchart of the patient enrollment process is provided in Fig. 1. Patients with changes in the Lenke-Silva level after rst-stage surgery were grouped into group A, and patients with unchanged Lenke-Silva levels after rst-stage surgery were grouped into group B to assess the ability of OLIF to correct their deformity using the two-stage MIS protocol.

Operative techniques
The surgery protocol for staged ASD correction in this study was OLIF (ALIF was used for pathologies involving L5-S1) for the rst stage of surgery, then posterior instrumentation implanted in the second stage of the surgery. There were no strict guidelines for the duration between the rst-and second-stage surgeries.
For the rst stage of the correction surgery, OLIF procedures were performed depending on the patient's clinical condition. After general anesthesia was administered, the patient was securely braced in the right lateral decubitus position with silicone padding and lateral supports to prevent pressure sores.
Before sterile draping, the desired level of correction was mapped out with a mobile C-arm system to ensure the proper approach to the spine. Polyether ether ketone devices with anteroposterior widths of 18 mm or 22 mm were used (6° lordosis cage, CLYDESDALE™ Spinal system, Medtronic, Inc). Robotic assistance (Mazor Robotics Renaissance®) was optional for posterior instrumentation in the second stage.

Functional outcome assessment
Assessments of functional scores were performed preoperatively (Pre-op) and after completion of the staged operation protocol (Pre-op and 1, 3, 6, 12, and 24 months after operation). We used the visual analog scale (VAS) to assess back and leg pain, the Oswestry Disability Index (ODI), and the EuroQoL 5dimension (EQ-5D). Complications observed during the study period were carefully recorded and classi ed into major complications and minor complications. We modi ed the classi cation methods for complications presented by Xu et al. [15] Major complications included vascular injury, deep venous thrombosis, pseudoarthrosis, instrument failure, proximal junctional kyphosis, and proximal junctional failure. Minor complications were approachrelated transient left thigh numbness, muscle weakness, cerebrospinal uid leakage, and super cial infections.

Radiological outcome assessment
Whole spine radiographs of the patients standing in the upright position were obtained under three conditions: preoperative, post-OLIF (For the 2-stage protocol), and post-posterior instrumentation. We measured the following parameters with Surgimap (Nemaris, Inc.) when assessing the radiographs we assessed: the sagittal vertical axis (SVA), the max Cobb angle, coronal balance (coronal C7-plumb line), pelvic incidence (PI), pelvic incidence-lumbar lordosis (PI-LL), and lumbar lordosis (LL). All patients were classi ed using the Lenke-Silva classi cation preoperatively, post-OLIF (for the 2-stage protocol), and postposterior instrumentation. [2] Statistical analysis A statistical analysis was completed with IBM SPSS statistics version 23 (IBM Corp., Armonk, New York, USA). Data were presented as mean ± standard deviation. The functional outcomes were assessed with the Wilcoxon signed-rank test. A paired t-test was used to compare the radiological outcomes of all patients because the normality of the data was con rmed with the Kolmogorov-Smirnov test. The radiological outcomes for group A and group B were analyzed with the Mann-Whitney U test for the continuous variables. For the demographics, the Mann-Whitney U test was used for the continuous variables, and Fisher's exact test was used for the categorical variables.

Demographic Data
In this study, a total of 27 patients (8 men and 19 women) treated with OLIF correction surgery were included. Their average age was 65.2 ± 7.9 years old, and their mean postoperative follow-up duration was 10.1 ± 5.2 months. Twenty-six patients (96.3%) had de novo type ASDs, and one patient (3.7%) had a residual-type ASDs. All twenty-seven patients had lumbar main curve deformities. The demographics data are presented in Table 1.  Table 2 provides the intraoperative data for all twenty-seven patients included in this study. Thirteen patients in this study went through single-stage surgery, and fourteen patients received two-stage surgery. The average OLIF/ALIF levels were 2.8 ± 0.6, and an average of 4.2 ± 2.2 levels of posterior instrumentation were applied. Furthermore, fteen patients (56%) had robotic-assisted instrumentation. The mean total operating time for all patients was 10.0 ± 3.3 hours, and the average total blood loss was 892 ± 552 ml.

Clinical Outcomes
There were statistically signi cant improvements (P < 0.001) between the pre-operative and post-operative scores for VAS-back, VAS-leg, ODI, and EQ-5D; their improvements were 7.2 ± 2.4 to 2.9 ± 1.7, 6.8 ± 2.6 to 1.4 ± 2.1, 57.9 ± 9.2 to 33.8 ± 14.0, and 11.7 ± 0.7 to 8.0 ± 1.4, respectively. The clinical outcomes are provided in Table 3. preoperatively and at the nal follow-up, respectively. Even though there were no signi cant improvements in the SVA postoperatively, the average SVA was 25.4mm, which was within the normal value. There was signi cant difference preoperatively and at the nal follow-up in PI-LL and LL, 15.2 ± 11.9 to 9.8 ± 9.2 (P = 0.004), and 32.0 ± 12.9 to 37.2 ± 9.5 (P < 0.001), respectively. Comparison of the radiological outcomes of Group A and Group B Ten patients with improved Lenke-Silva levels were included in Group A, and four patients with unchanged Lenke-Silva levels were included in Group B. The rate of successful correction was 71%. The max Cob angle of group A was 22.9 ± 8.6 and 16.4 ± 5.3 (P = 0.005) preoperatively and after the rst stage OLIF, respectively ( Table 6). The PI-LL mismatch was 13.9 ± 14.7 preoperatively, which improved to 3.2 ± 9.6 (P = 0.005) after the rst stage OLIF. The SVA was 34.7 ± 33.5 mm preoperatively and − 0.1 ± 22.3 mm (P = 0.022) after the rst-stage OLIF. The PT, LL, PI-LL mismatch, SVA, max Cobb angle in group A were signi cantly improved after the rst stage OLIF, where area, group B did not show signi cant improvement.

Complications
The overall complication rate was 37% (ten of the twenty-seven patients); major complications occurred in four of the twenty-seven patients (14.8%), and minor complications occurred in six of the twenty-seven patients (22.2%). No patients in this study required revision surgery (Table 7).

Analysis of suboptimal correction
In terms of the patient characteristics of group A and B, there were no signi cant differences in age, gender, and BMI. There were also no signi cant betweengroup differences in type of ASD, preoperative Lenke-Silva classi cation level, main curve type, sagittal and coronal alignment, or convexity. (Table 5). In Table 8. The causes of suboptimal corrections in four patients were carefully analyzed. In group B, the approach with OLIF was the main cause of unsatisfactory corrections in two patients (2/4, 50%) because the deformities involved the T-L junction, which was not accessible using the approach. Severe preoperative PI-LL mismatches (41°) and cage subsidence with anterior endplate fractures were the cause of poor corrections in two patients. Patient A had kyphosis located at the T-L junction (T10-L2 kyphosis: 21°), which was impossible to correct, and L1/2 was di cult to access and reconstruct using the OLIF approach. For patient B, the major curve was located at the TL-L junction, and the patient had residual type ASD with a stiff curve of L1/2 and L2/3 (L1-L3 Cobb angle from 35° to 34°). In this case, the disc height of L1/2 was di cult to restore with OLIF because of blockage by the rib cage due to convex bending to the left side. In patient C, there was unsuccessful correction of lumbar kyphosis because of a severe preoperative PI-LL mismatch with insu cient change in LL (6°->30°). In patient D, correction of SVA failed (post 1st stage SVA: 96mm) because of cage subsidence due to an endplate fracture.

Discussion
In recent years, minimally invasive retroperitoneal approaches such as OLIF have been recognized as an alternative MIS choice for the treatment of ASDs. [13,14,16] Such approaches can achieve a correction rate comparable to that of a posterior osteotomy through restoration of disc height and result in lower complication rates and reduced blood loss. [17][18][19] In our study, 14 patients with ASD were treated using a 2-stage MIS protocol, and 10 patients (10/14, 71.4%) achieved effective correction of coronal and sagittal plane deformity after rst stage OLIF (Fig. 2). Overall, there was less blood loss, and the rate of major complications in patients treated with the 2-stage MIS protocol was lower than with a posterior osteotomy.
Kim et al. demonstrated that OLIF could be an effective approach for ASD correction in patients with severe sagittal deformity. [20] The author used an L1 to S1 oblique retroperitoneal approach (OLIF) in the treatment of 32 patients with ASD. The mean SVA decreased from 13.6 cm before surgery to 2.9 cm at the nal postoperative follow-up; LL increased from 5.8° before surgery to 46.5° at the nal postoperative follow-up, and the Cobb angle improved from 21.6°b efore surgery to 9.6° at the nal postoperative follow-up. The mean preoperative PI-LL mismatch in these patients was 44.5°. Yang et al. used OLIF combined with unilateral posterior instrumentation through a Wiltse approach to treat degenerative lumbar scoliosis and achieved signi cant correction of the coronal and sagittal plane, where the post-operative MRIs showed there was signi cantly increased cross-sectional dural sac area. [21] In treating a severe sagittal deformity, greater correction e cacy can be achieved with lateral lumber interbody fusion (LLIF) combined with the anterior column resection (ACR) technique. [17] Leveque et al. compared LLIF combined with ACR(LLIF-ACR) to PSO in treating patients with severe PI-LL mismatches. [22] Their results showed comparable correction rates among these two groups, with lower pseudoarthrosis and re-operation rates due to pseudoarthrosis in the LLIF-ACR group. Additionally, the results of a study conducted by Li et (Table 4) in the rst stage, effectively reducing the need for fusion to the thoracic spine and posterior osteotomy in the second stage, and the correction rates were similar to those of previous studies. [2,16] There were four patients in which ineffective correction in the rst stage was not achieved (Table 6). Their post-rst stage Lenke-Silva classi cation levels were, IV, V, and VI. The causes were classi ed as patient factors in three cases and as surgical technique factors in one case. Patient A had severe degeneration of T12-L1 and L1-2 at the thoracolumbar (T-L) junction, which led to kyphosis of the T-L junction (T10-L2 kyphosis: 21°). However, T12-L1 and the L1-2 disk space were impossible to approach through OLIF. After using the OLIF approach to the L2-3, L3-4, and L4-5 during the rst-stage operation, the PI-LL mismatch was improved from 19° to 8° postoperatively. However, there was still focal kyphosis of the T-L junction, and global alignment could not be restored to the normal range. Thus, we believe for patients with deformities involving the T-L junction, spinal fusion must be done because deformities in the T-L junction cannot be corrected with OLIF. Patient B was a case of residual type ASD (Fig. 3), in which the main curve was at the T-L junction. Patient B had stiff coronal plane scoliosis curves at L1-2 and L2-3.
The left side was the convex side, and the OLIF approach to the L1-2 disc from the left side was blocked by the rib cage. Therefore, the degree of Cobb angle correction in the rst stage was not ideal. These two cases of suboptimal correction implied that MIS correction using the OLIF approach is insu cient for deformities involving the T-L junction or the T-spine. Patient C had severe lumbar kyphosis (PI-LL: 41°) due to multilevel lumbar disc collapse. Despite the increment of the LL by 24° after performing OLIF at L2-3, L3-4, L4-5, and ALIF at L5-S1, there was still a PI-LL mismatch of 17° after the rst stage surgery; therefore, a posterior osteotomy was still required to achieve optimal lumbar lordosis. Although previous studies have suggested that OLIF is more suitable for management of relatively minor deformities as compared to other alternatives (SVA < 60 mm and PI-LL < 30°), we believe OLIF may also be suitable for achieving sagittal alignment correction in patients with lumbar kyphosis if a hyperlordotic cage or LLIF-ACR is used. [13,14,16] Patient D had a global imbalance with a preoperative SVA of 92 mm, where, after surgery, L2-3 segmental lordosis changed from 16° preoperatively to only 9° after the rst-stage operation due to L2-3 cage subsidence (36% of the vertebral height), making correction of lumbar lordosis and global imbalance impossible (Fig. 4). This risk factor was also cited in a study by Li et al. In this patient, the cage was placed in the anterior third of the disc space, where an anterior endplate fracture cause more signi cant loss of lumbar lordosis. [16] In the future, a larger prospective study is needed to validate the risk factors for failure of MIS OLIF correction.

Limitations
Our institution has used OLIF as the MIS approach for correction of ASD since 2018. However, this retrospective study involved a relatively small sample size, and a subsequent statistical analysis of speci c risk factors could not be conducted. We focused on patients who did not achieve satisfactory correction after the rst stage. These patients were individually investigated to determine the possible factors leading to failure of correction. In the future, we plan to conduct a larger prospective study to analyze the factors contributing to the failure of MIS correction of ASD. The present study had a relatively short followup period. However, some complications, such as proximal junctional kyphosis or proximal junctional failure, could appear after a longer follow-up period.
These complications also affect alignment and functional outcomes and will also be included in our future studies with long-term follow up.   Case of suboptimal correction This 56 y/o female was a case of residual type ASD (3A and 3B) in which the main curve was at the thoraco-lumbar junction (Cobb angle: 53°). There were stiff coronal plane scoliosis curves at L1-2 and L2-3(3C and 3D). Flexion-extension X-ray showed convexity at the left side. (3E and 3F) The access to the L1-2 disc space through left side OLIF approach was blocked by the rib cage due to the convexity of left side. Therefore, the correction of coronal plane cobb angle was less than ideal (Cobb angle: 41°)(3G and 3H). The level of LS remained at level VI after the rst stage OLIF correction. Therefore, posterior facetectomy and extended fusion to T-spine were required in the second stage operation. ASD: Adult spinal deformity, TL:

Figure 4
This 70 y/o female was a case of ASD with severe sagittal imbalance (SVA=92mm, PI-LL=13 degree)(4A and 4B). L23, L34 and L45 OLIF correction was done in the rst stage operation(4C and 4D). However, L2 lower endplate was violated during operation and endplate fracture with cage subsidence was occurred after operation, which caused 7° loss of segmental lordosis. Therefore, LL and SVA restoration were failed after rst stage correction and the LS level was remained at level V after rst stage correction, because LL and SVA restoration were ineffective. Extended posterior fusion to T-spine was necessary in the second stage operation to achieve optimal fusion. ASD: Adult spinal deformity, LL: Lumbar lordosis, SVA: Sagittal vertical axis, LS: Lenke-Silva classi cation.