Clinical Outcomes of S2 Alar-Iliac (S2AI) Screw Technique in The Treatment of Severe Spinal Sagittal Imbalance: A Retrospective 2-Year Follow-Up Study

The treatment of adult spinal deformity (ASD) remains a signicant challenge, especially in elderly patients. This study aimed to evaluate the outcomes of the S2AI screw technique in the treatment of severe spinal sagittal imbalance with a minimum 2-year follow-up. Methods From January 2015 to December 2018, 23 patients with severe degenerative thoracolumbar kyphosis who underwent placement of S2AI screws for long segment fusion were retrospectively reviewed. Patients were divided into group A (no mechanical complications, 13 cases) and group B (with mechanical complications, 10 cases) according to the occurrence of mechanical complications at the last follow-up. Radiographic parameters were compared between groups preoperatively, 1 month postoperatively, and at the last follow-up. Risk factors for mechanical complications were analyzed.


Introduction
The number of elderly patients with spinal deformities is increasing with the growth in the aging population, with a reported incidence rate of 60%. 1 Over the past decade, spine surgeons have recognized the importance of restoring sagittal balance for the treatment of spinal deformities given the signi cant relationship between sagittal balance and health-related quality of life. 2,3 Patients with severe spinal sagittal imbalance may experience back pain and restrictions in daily activities. Moreover, a recent study reported an incidence of depression of 40.6% in patients with severe imbalances. 4 However, the treatment of spinal deformities remains challenging. Complication rates are high when long-segment posterior corrective surgery for spinal deformity ends at S1, 5 and long-term follow-up assessments have indicated that the rate of pseudarthrosis may be as high as 24%. 6 In 2007, Sponseller and Kebaish 7 rst reported the sacral-2 alar iliac (S2AI) screw technique.
Biomechanical test results of long-segment xation revealed that the S2AI screw technique signi cantly reduced the range of motion of exion, extension, lateral exion, rotation, and stress on the S1 screw. 8 Indications for this technique include long segment (≥ 5 vertebrae) spinal surgery ending at the sacrum, especially spinal corrective surgery, including degenerative scoliosis, rigid kyphosis with sagittal imbalance, and iatrogenic at back deformity. 9 Although the clinical application of the S2AI screw technique is growing, relevant literature regarding this technique is scarce. The purpose of this study was to observe the outcomes and complications of the S2AI screw technique in the treatment of elderly patients with severe spinal sagittal imbalance.

Study Participants
Patients with severe degenerative thoracolumbar kyphosis who underwent long-segment corrective surgery in our hospital between January 2015 and December 2018 were retrospectively reviewed. All patients were diagnosed with "sagittal imbalance syndrome" based on medical history and physical and imaging examinations. The inclusion criteria were as follows: (1) the patient had typical symptoms of spinal sagittal imbalance, the location of the pain was consistent with kyphotic segments, the trunk exhibited a clear forward lean when standing and walking; (2) PI-LL>30°; (3) structural kyphosis con rmed on dynamic radiographs or extensive multi dus and erector spinal muscle atrophy and fatty in ltration con rmed as myogenic kyphosis on magnetic resonance imaging (MRI); (4) kyphosis was clearly aggravated during activities; (5) combined scoliosis, lumbar disc herniation, lumbar spinal stenosis, lumbar spondylolisthesis, or old vertebral compression fractures; (6) postural kyphosis caused by nerve compression was excluded; and (7) distal xation extended to the pelvis using the S2AI technique. The exclusion criteria were as follows: (1) failure to complete the 2-year follow-up or incomplete data; (2) cervical spondylosis or cervical deformity; (3) infection, tumor, metabolic bone disease, or abnormal blood coagulation.
Before surgery, the patients underwent lumbar radiography, full-length spine X-ray, and thoracolumbar MRI examinations. At follow-up, the patients underwent lumbar X-ray and full-length spine X-ray examinations. Visual analog scale (VAS) score, Oswestry disability index (ODI) score, and type of complications were recorded. Computed tomography (CT) and MRI were performed as needed. This study was approved by the Ethics Committee of Peking University First Hospital (No. 2021-149). Written informed consent was obtained from all participants. All methods were carried out in accordance with relevant guidelines and regulations.
A total of 23 patients were recruited, comprising 2 men and 21 women (average age: 68.0±6.5 years, range: 60-84 years). The number of fused segments was 9.1±2.4, and the follow-up time was 32.2±6.2 months. Based on the occurrence of postoperative mechanical complications, the patients were divided into group A (no mechanical complications, n=13) and group B (with mechanical complications, n=10).
No signi cant differences were observed in sex, age, body mass index, bone mineral density, xed segment, operation time, method of osteotomy, intraoperative bleeding, and follow-up time between the two groups (P>0.05). The follow-up period for all patients was > 2 years. General patient information is presented in Table 1.

Surgical Procedure
The paravertebral muscles were dissected using a conventional posterior approach, and the distal end of the incision reached the level of the S2 spinous process. S2AI screws were inserted during the navigation.
The midpoint of the line connecting the lateral edges of the S1 and S2 foramina was selected as the entry point for the S2AI screw. The pedicle probe was directed 30-40° laterally in the axial plane and 20-30°c audally in the sagittal plane. When the pedicle probe penetrated the sacroiliac joint and entered the ilium, a ball-tip probe was used to con rm the integrity of the anterior, posterior, superior, and inferior walls of the ilium. The diameter and length of the S2AI screws were 7.5 mm and 60-80 mm, respectively.
Intraoperative CT was performed to con rm screw position. Seven patients underwent pedicle subtraction osteotomy (PSO) at the apical vertebrae and multi-segmental Ponte osteotomy, while the remaining patients underwent multi-segmental Ponte osteotomy (MPO). Decompression was performed for segments with spinal stenosis, lumbar disc herniation, or lumbar spondylolisthesis, and interbody fusion was performed if necessary. Bone grafting was routinely performed.

Data Collection and Radiographic Assessment
Types of complications were recorded. Patients' VAS and ODI scores preoperatively and at the last followup were recorded. Anteroposterior and lateral full-length spine radiographs were obtained preoperatively, at 1 month postoperatively, and at the last follow-up. Surgimap 2.0 software (Medicrea, New York) was used to measure the radiographic parameters, including pelvic incidence (PI), pelvic tilt (PT), SS, LL, thoracic kyphosis (TK), thoracolumbar kyphosis (TLK), TPA, SVA, Cobb angle (CA), and C7 plumb linecenter sacral vertical line (C7PL-CSVL). GAP score 10 was calculated 1 month post-operatively.

Data and Statistical Analysis
Statistical analyses were performed using SPSS26.0 (IBM, Armonk, NY, USA). Basic descriptive analysis included Student's t test and Fisher exact test for continuous and categorical variables, respectively. Results are presented as mean (+standard deviation) for continuous variables and n for categorical variables. Preoperative and postoperative parameters were compared using a paired sample t-test.
Groups A and B were compared using an independent sample t-test. The relationships between mechanical complications and parameters were evaluated using Pearson correlation analysis. Statistical signi cance was de ned as a P value<0.05.

Complications
One patient presented with a left foot dorsal exor strength of grade 0 postoperatively and was treated with methylprednisolone on the same day. On the rst postoperative day, surgical assessment revealed that the ligamentum avum of the L4/5 segment was hypertrophic, and the L4/5 segment was decompressed. Postoperatively, the patient was administered neurotrophic therapy and rehabilitation exercises. At 6 months postoperatively, the left tibialis anterior muscle strength had recovered to grade IV.
Two patients underwent revision surgery due to proximal junctional failure (PJF); one case involved proximal screw cap displacement, and spinal canal stenosis caused lower limb muscle weakness. One patient underwent revision surgery due to broken bilateral connecting rods (combined with loosening of the distal screws). The revision rate was 17.4%. One case of broken rod, one case of distal screw loosening, ve cases of proximal junctional kyphosis (PJK), and one case of super cial wound infection were observed, all of which were treated conservatively. Mechanical complications were observed in 10 patients, with an incidence rate of 43.5%.

Functional Outcomes by Group
In groups A and B, preoperative VAS scores were 7.5±1.9 and 7.1±1.1, respectively (P=0.573), and preoperative ODI scores were 64.2±7.0 and 66.5±10.3, respectively (P=0.523). In groups A and B, VAS scores at the last follow-up were 2.2±0.9 and 3.3±1.2, respectively (P=0.014), and ODI scores at the last follow-up were 28.6±7.8 and 49.5±9.6, respectively (P<0.01). At the last follow-up, the VAS and ODI scores were signi cantly improved compared with preoperative scores in both groups (P<0.01).

Radiographic Outcomes by Group
The radiographic parameters obtained preoperatively, postoperatively, and at the last follow-up are summarized in Table 2   Case Presentation Figure 1 depicts the mechanical complications of distal screw loosening in group B. The 66-year-old patient was diagnosed with degenerative kyphoscoliosis deformity and underwent T6-ilium correction and fusion. The sagittal correction was unsatisfactory. At 24 months follow-up, lumbar anteroposterior and lateral radiographs and lumbar CT indicated loosening of the S2AI screws.

Discussion
Long-segment corrective surgery is associated with a high complication rate. Scheer et al. 11 analyzed 138 patients aged 63.3±11.5 years with severe sagittal imbalance (PI-LL≥30°) and reported a postoperative complication rate of 74.6%, of which internal xation-related complications constituted 30.4%. In our study, 10 patients (43.5%) presented with mechanical complications, and the revision rate was 17.4%. This could be associated with advanced age, severe preoperative sagittal imbalance, obesity, low bone density, and/or other factors. PJK was the most common complication, with an incidence of 21.7%. Previously reported risk factors for PJK include advanced age, damage to the posterior tension band, combined anterior and posterior surgery, fusion to the sacrum/pelvis, choice of upper xed vertebrae, rod material, insu cient or excessive correction, and sarcopenia. 12 Unoki et al. 13 analyzed patients with longsegment corrective surgery and reported an incidence of PJK of 23.3%, 30.4%, and 29.2% for distal xation to L5, the sacrum, and to the pelvis with S2AI screws, respectively. No signi cant difference between techniques was observed, suggesting that the S2AI screw technique did not increase the risk of PJK. At the last follow-up, two patients (8.7%) developed symptoms of S2AI screw loosening in our study, which manifested as sacroiliac joint pain. Both patients were treated conservatively. Their BMDs were -2.91 and -2.84, respectively, which were considered osteoporotic. In the literature, the breaking rate of S2AI screws is reported to be 0.0-1.0%, and 7.8-10.4% of patients with S2AI screws demonstrated visible translucent bands around the screws on imaging at more than 1 year follow-up 14,15 , which was consistent with our ndings. Further, two patients presented with rod breakage. Because elderly patients often have osteoporosis, we used a titanium alloy rod with a 5.5-cm diameter rather than a 6.0cm diameter in order to reduce the risk of PJK and stress at the junction of the bone and screw; nevertheless, this reduced rod strength. Therefore, we recommend adding satellite rods to the kyphotic apex area to reduce the risk of rod breakage.
Compensation in patients with sagittal imbalance may be achieved via various mechanisms, including pelvic retroversion, reduction of thoracic kyphosis, hip hyperextension, and knee exion. In this study, PT was greater than SS preoperatively, con rming signi cant pelvic retroversion in both groups. At 1 month postoperatively, PT was decreased, SS was increased, and pelvic retroversion improved in both groups.
No signi cant differences were observed in PI, PT, and SS between 1 month postoperatively and at the last follow-up, suggesting that the S2AI screw technique maintained the relative position of the pelvis effectively. Accordingly, for patients with sagittal imbalance and severe retroversion of the pelvis, especially those with lumbosacral muscle atrophy, we recommend using S2AI screws for sacral pelvic xation if long-segment corrective surgery is performed.
We observed that the PI increased by 3.7±3.9° in group B at 1 month postoperatively. In this regard, S2AI screws may stabilize the sacroiliac joint and change the morphology of the pelvis, thus changing the PI.
Ishida et al. 16 used S2AI screws for pelvic xation in 46 adult patients aged 61.5±10.7 years with spinal deformity. In their study, the preoperative PI was 63.4±12.3°, and the postoperative PI decreased by 6.0±12.5°. Wei et al. 17 reported that the use of S2AI screws for pelvic xation reduced PI after surgery in patients with high PI (PI≥60°); conversely, PI increased after surgery in patients with normal or low PI (PI<60°). In this study, preoperative PI was 43.4±9.9° and the degree of kyphosis was severe in group B, which could partly explain the increase in postoperative PI.
Although postoperative VAS and ODI scores in both groups were signi cantly improved compared with preoperative values, VAS and ODI scores were signi cantly better in group A than in group B at the last follow-up. This indicates that the occurrence of mechanical complications signi cantly worsens surgical outcomes. Preoperatively, LL in group B was -0.9±10.2°, indicating a kyphotic curve, whereas that in group A was 12.8±11.6°, with a signi cant difference between groups. As patients in group B had more severe preoperative sagittal imbalance, greater correction was required during surgery. This suggests that the risk of postoperative mechanical complications is higher in patients with severe sagittal imbalance and lumbar kyphosis. Therefore, sagittal alignment should be carefully evaluated preoperatively, and a detailed surgical plan should be designed to correct this imbalance.
At 1 month postoperatively, SS, LL, and PI-LL in group B were 20.2±8.6°, 24.4±9.9°, and 22.7±12.5°, respectively, indicating that sagittal reconstruction was insu cient. In group A, SS, LL, and PI-LL were 28.8±8.5°, 39.6±6.7°, and 11.4±11.3°, respectively, indicating that the sagittal alignment was signi cantly better than that in group B. Pearson correlation analysis demonstrated that mechanical complications were signi cantly associated with SS, LL, and PL-LL at 1 month postoperatively, highlighting the importance of restoring lumbar lordosis. In our study, the best cutoff values of SS, LL, and PI-LL were 24.1°, 32.75°, and 12.0°, respectively. SS represents lower lumbar lordosis, and normal lower lumbar lordosis accounts for approximately two-thirds of the total lordosis curve. Therefore, the recovery of L4-S1 lordosis during operation is important. In the L4/5 and L5/S1 segments, we opted to release and distract the intervertebral space and subsequently placed a cage with a large lordosis angle to restore the local lordosis angle. Based on our experience, we propose that for the upper lumbar segments, lordosis can be reconstructed via intervertebral space release and posterior multi-segment Ponte osteotomy, whereas for areas with large local kyphosis, PSO should be considered.
At the last follow-up, LL was signi cantly reduced, and PI-LL, TPA, and SVA were signi cantly increased in both groups. Patients underwent long-segment xation, and the pelvis was in a relatively stable position.
Therefore, the loss of sagittal correction was predominantly caused by the reduction in lumbar lordosis during the follow-up period. LL decreased by 4.4±4.7° in group A and by 8.1±7.3° in group B, but the change in LL was not signi cantly different between the two groups (P>0.05). In group B, the restoration of LL was insu cient at 1 month postoperatively, and the further reduction in LL may have partly underpinned the occurrence of mechanical complications. Nakazawa et al. 18 and Im et al. 19 reported that insu cient restoration of the PI-LL could lead to progressive worsening of sagittal imbalances after longsegment orthopedic surgery with sacropelvic xation. Therefore, we should also consider the reduction of lumbar lordosis in the preoperative planning.
The average age of the patients in our study was 68.0±6.5 years. As natural degeneration of the spine is accompanied by an increase in positive sagittal balance, age is a key factor affecting sagittal alignment. In elderly patients with sagittal imbalance, the intervertebral discs, articular processes, paravertebral ligaments, muscles, and other structures have adapted to a positive balance for a long time. Immediate correction to the "ideal" state after surgery did not seem to improve clinical outcomes, but increased the incidence of mechanical complications. 20 Lafage et al. 21 emphasized that for elderly patients aged over 75 years, the ideal sagittal parameters were PT=28.5°, PI-LL=16.7°, and SVA=7.8 cm. In our study, PT, PI-LL, and SVA in group A were 25.1±11.8°, 17.3±9.6°, and 5.8±3.1 cm at the last follow-up. According to the sagittal modi ers of SRS-Schwab classi cation, 22 these parameters were classi ed as positive if 20°≤PT≤30°, 10°≤PI-LL≤20°, and 4 cm≤SVA≤9.5 cm. These results indicated that the sagittal alignment was in a moderate positive balance, and the VAS and ODI scores of the patients were signi cantly improved. Zhang et al. 23 performed long-segment corrective surgery on 44 patients aged 65.1±2.8 years with spinal deformities and reported that postoperative PI-LL was between 10° and 20°, which was similar to the results of our study. As such, we did not consider it necessary to perform corrections to the state of LL = PI ± 9° in elderly patients.
The GAP score was used to predict mechanical complications in patients undergoing long-segment corrective surgery for adult spinal deformities. The GAP score includes ve categories: relative pelvic version, relative lumbar lordosis, lordosis distribution index, relative spinopelvic alignment, and age.
The GAP score enables the calculation of the ideal sagittal parameters based on PI and emphasizes the signi cance of lordosis distribution in the lumbar spine. The PI value in uences the morphology of the spine and pelvis. Patients with a small PI have less sagittal compensation ability, and the requirements for reconstruction of lumbar lordosis are stricter. 24 In our study, mechanical complications were signi cantly correlated with GAP scores. The average GAP scores were 6.9±2.0 and 10.9±2.4 in groups A and B, respectively. In a study by Yilgor et al., 10 the ideal GAP score after surgery was ≤4, and the incidence of mechanical complications at 7 points and 11 points or above was 72.7% and 100%, respectively. These results are discordant with our ndings, which could be related to individual differences, such as age and degree of deformity. Ham et al. 25 retrospectively evaluated 84 patients aged 70.5 ± 7.7 years old with severe sagittal imbalances and reported that the area under the curve for predicting mechanical complications based on the GAP score was 0.839, similar to our study. Moreover, 82% of patients with mechanical complications had a GAP score ≥ 7. For elderly patients with severe sagittal imbalance of the spine, a postoperative GAP score ≥ 10 indicated a high risk of mechanical complications, which should be carefully monitored during follow-up.
This study has several limitations. First, it was a retrospective study and lacked a randomized control group with distal xation to the S1 or ilium. Second, the sample size was small, and selection bias may have occurred. Due to the lack of a uni ed standard for evaluating the strength of the lower back muscles, this study did not include muscle-associated factors. Further, full-length radiographs of the lower extremities were lacking; hence, the compensation status of the hip and knee joints could not be analyzed in detail. At present, there is a paucity of studies discussing the surgical treatment of elderly patients with severe kyphosis, and long-term follow-up with a large sample is warranted to con rm the effects of long-segment corrective surgery using the S2AI screw technique.

Conclusion
The incidence of mechanical complications in long-segment corrective surgery with the S2AI screw technique for severe spinal sagittal imbalance remains high. Inadequate sagittal correction is a risk factor for the development of mechanical complications. Our ndings suggest that the ideal sagittal balance for elderly patients is LL>30° and 10°≤PI-LL≤20 °.

Availability of data and materials
The datasets analyzed during the current study are available from the corresponding author on reasonable request.  Figure 1