Comparison of Radiographic And Clinical Outcomes Observed After Adult Spinal Deformity Correction Using Iliac Screw Versus S2-Alar-Iliac Screw

Background: To date, there is a paucity of reports clarifying the change of spinopelvic parameters in patients with adult spinal deformity (ASD) who underwent long segment spinal fusion using iliac screw (IS) and S2-alar-iliac screw (S2AI) xation. Methods: A retrospective review of consecutive patients underwent deformity correction surgery for ASD between 2013 and 2017 was performed. Patients were divided into two groups based on whether IS or S2AI xation was performed. All radiographic parameters were measured preoperatively, immediately postoperatively, and the last follow-up. Demographics, intraoperative and clinical data were analyzed between the two groups. Additionally, the cohort was subdivided according to the postoperative change in pelvic incidence (PI): subgroup (C) was dened as change in PI ≥ 5° and subgroup (NC) with change <5°. In subgroup analyses, the 2 different types of postoperative change of PI were directly compared. Results: A total of 142 patients met inclusion criteria: 111 who received IS and 31 received S2AI xation. The IS group (65.6 ± 26, 39.8 ± 13.8) showed a signicantly higher change in lumbar lordosis (LL) and upper lumbar lordosis (ULL) than the S2AI group (54.4 ± 17.9, 30.3 ± 9.9) (p<0.05). In subgroup (C), PI signicantly increased from 53° preoperatively to 59° postoperatively at least 50% of IS cohort, with a mean change of 5.8° (p<0.05). The clinical outcomes at the last follow-up were signicantly better in IS group than in S2AI group in terms of VAS scores for back and leg. The occurrence of sacroiliac joint pain and pelvic screw fracture were signicantly greater in S2AI group than in IS group (25.8% vs 9%, p<0.05) and (16.1% vs 3.6%, p<0.05). Conclusions: IS xation showed a greater change in LL and ULL than S2AI xation in ASD


Background
Spinopelvic xation (SPF) is becoming an increasingly important avenue for degenerative conditions as the aging population grows. This instrumentation technique can also be applied in high-grade spondylolisthesis, trauma, tumors or infection. However, achieving solid xation in lumbosacral junction continues to be a challenge for spine surgeons because of the tremendous biomechanical forces demand across the junctional area, complex regional anatomy, and a high pseudarthrosis rate, especially in patients with adult spinal deformity (ASD). [1][2][3][4][5] With the emergence of advanced spinal instrumentation, multiple options have been described for additional SPF over the past decades. Currently, iliac screw (IS) and S2-alar-iliac (S2AI) screw xations are the most popular method of SPF in clinical practice.
The IS had a long-established history of improving stability, with comparative advantages of greater diameter and length bolts could be used, higher pullout strength, and easier application.
[6] However, there are some disadvantages regarding the IS, including the need for lateral connector, more extensive tissue dissection, prominent hardware, and wound dehiscence etc. [3,7,8] Tsuchiya et al. [9] reported that up to 34% ASD patients treated via IS technique necessitating reoperations due to prominence.
In response to above drawbacks, the S2AI has recently become an increasingly popular technique as an alternative method of SPF, which was initially described by Dr. Sponseller. [10] This technique prevented hardware prominence owing to its deeper and more medial entry point compared to IS. [7] Additional, current literature reported the S2AI had several potential advantages over IS with lower rates of reoperation, wound dehiscence, and implant failure. [4,[11][12][13] However, there are limited data regarding comparative data between these techniques. To date, few studies exist speci cally focused on the radiographic change of spinopelvic parameters after SPF utilizing IS and S2AI techniques. Therefore, the aim of this study was to compare the radiographic change of sagittal spinopelvic parameters between these two techniques for the treatment of ASD, and assess the complication rate of our substantial case series.

Materials And Methods
After institutional review board approval, we performed a retrospective review of consecutive patients who underwent long-segment (≥ 6 levels) spinal fusion and pelvic xation using IS or S2AI technique at a single institution between 2013 and 2017. The indications for surgery included frequent recurrent lowback and/or leg pain, neurological de cits, severe disability and/or progressive deformity that failed to better with conservative treatment for more than 6 months. We included ASD patients with age ≥ 60 years and at least one of the following: C7 sagittal vertical axis (SVA) > 50mm, pelvic tilt (PT) > 25°, pelvic incidence (PI) -lumbar lordosis (LL) > 10°. Patients with a history of neuromuscular diseases, malignancy, infection, and postoperative follow-up less than 24 months were excluded. Furthermore, patients were also excluded from this study if they had ankylosing spondylitis, Parkinson's disease, or incomplete radiographic and clinical records. Eligible patients were divided into 2 groups according to the surgical method: IS group (iliac screw xation) and S2AI group (S2-alar-iliac screw xation).
The clinical data for this current cohort were obtained from the electronic medical records and operative database of our institution. Standard demographic data (e.g., age, gender, primary diagnosis, medical comorbidities, and American Society of Anesthesiologists [ASA] [14] classi cation) were extracted.
Surgical parameters included the operated levels, estimated blood loss (EBL), surgical duration, use of an osteotomy, and length of stay (LOS). The major complications requiring revision included proximal junctional kyphosis (PJK), pseudarthrosis, deep infection, SIJ pain, or other implant-related complications (pelvic screw loosening, fracture, and wound dehiscence etc.) were compared. Clinical results were evaluated using the Visual Analog Scale (VAS) [15] of back and leg pain and Oswestry Disability Index (ODI)[16] at preoperative and 3 months postoperative, and the last follow-up.

Surgical procedures
The IS procedure was performed base on the technique described previously. [1,6,[17][18][19] With patients in the prone position, suprafascial dissection was performed via a midline incision. Routine exposure of the spinous process, lamina, facet joints, and transverse process at the index levels. The bilateral posterior superior iliac spine (PSIS) were then identi ed. Adequate subperiosteal dissection was performed and splitted fascia longitudinally over the PSIS midline, then stripped to the sides for fascial integrity. The ideal entry point of IS was about 10 mm between the posterior edge of the iliac crest and posterior sacrum. An osteotome was used to remove a small tri-cortical wedge around this point. A straight probe was gently and cautiously inserted into the cancellous channel of ilium, avoiding penetration of inner and outer cortex. Next, the trajectory was tapped with a ball-tip probe to verify the integrity of cortical bone. Then the screw was deeply inserted toward PSIS until its head was ush with the cortex of the ilium to lower the risk of prominence. Finally, the screw was attached to the rod underneath S1 pedicle screw using a lateral connector (Fig. 1).
The procedure of S2AI xation was performed using a free hand technique described previously. [7,[20][21][22][23][24] A routine posterior exposure was performed similar as IS. S2AI technique has a sacral ala entry point locating distal to the S1 foramen with the angulation of trajectory approximately 30-45° laterally and 25-40° distally, aiming toward the greater trochanter of the femur. Using uoroscopy to ensure position cephalad to the greater sciatic notch and tap drilling until the sacroiliac joint (SIJ), then reverse drilling until the iliac cortex was reached. Finally, the screw was inserted until its head placed in-line with S1 pedicle screw head (Fig. 2).
We commonly used the pelvic screw with a diameter of 8.5mm and a length of 7.0mm. All cases were performed as 2-stage anteroposterior fusion with an interval of 1 week by the two senior authors (KT and YC) as a team with > 20 years of experience in ASD.
[26] Kyphosis was indicated by a (+) value whereas lordosis was indicated by a (-) value. Change of parameter was calculated by subtracting the preoperative value from the postoperative value. All the parameters were measured by two surgeons who did not participate in the operation and the mean value were adopted Additionally, we de ned proximal junctional kyphosis (PJK) was considered present when proximal junctional angle was > 10° and at least 10° greater than the preoperative measurement, which was a severe complication after deformity surgery with radiographic evidence of acute PJK deformity and mechanical failure at the upper instrumented vertebra (UIV). [27] Fusion was con rmed by the presence of bridging bone connecting the adjacent vertebral bodies either through the implants or around the implants or < 3mm of translation on exion or extension radiographs.[28-30] Screw loosening was de ned as a radiolucent area ≥ 2mm surrounding the screw on radiographic images, also known as the "double halo sign". [31,32] Results were expressed as mean ± SD. We used the Student's t-test for continuous variables, and the Chisquare test used for categorical variables. All statistical analyses were performed using SPSS 21.0 software (SPSS Inc., Chicago, IL, USA) with p values of < 0.05 considered to be statistically signi cant.

Patient Characteristics and Perioperative Data
In total, 142 patients with a mean follow-up duration of 32.3 months were included in the study and analysis: 11 who received IS technique and 31 received S2AI technique. The mean age was 67.3 years and the BMI 28.3 kg/m 2 , 82.4% of patients were female. With regard to surgical data, the mean number of levels fused (8.8 vs. 8.7, p = 0.084) between the 2 groups were not signi cantly different. The mean estimated blood loss (EBL) were 2421ml and 2368ml in IS and S2AI groups, respectively (p > 0.05). Mean surgical duration in IS group was slightly more than in S2AI group (388.2 ± 99.1 vs. 377.6 ± 102.1, p = 0.602). As shown in Table 1, the comparative analyses revealed no signi cant intergroup differences in age, sex, primary diagnosis, comorbidities, LOS, and other perioperative data.

Radiographic Outcomes
Changes in the radiological spinopelvic parameters of all patients are summarized in Table 2. At enrollment, there were no differences in any preoperative parameters between the two groups. Except PI, the improvement achieved after surgery were maintained in all parameters, and subsequently maintained at the last follow-up. LL signi cantly changed from 11.4 ± 23.9° preoperatively to -54.2 ± 10°p ostoperatively in IS group, ULL was 7.1 ± 8.4° preoperatively and − 32.7 ± 9.7° after surgery. LL signi cantly changed from 12.1 ± 5.3° preoperatively to -42.3 ± 7.8° postoperatively in S2AI group, ULL was 6.6 ± 11.9° preoperatively and − 23.7 ± 9.3° after surgery. In IS group, the mean preoperative SVA of 155.5mm to 20.9mm after surgery and was 46.8mm at the last follow-up. In S2AI group, the mean preoperative SVA of 139.3mm to 32.2mm after surgery and was 67.2mm at the last follow-up. The mean PT was 37.1° preoperatively, improved to 17.4° postoperatively, and changed to 20.3° at the last follow-up in IS group. It was 38.3° before surgery, corrected to 26.2° after surgery, and changed to 27.7° at the last follow-up in S2AI group. For the further analysis, the cohort was subdivided based on the postoperative change in PI (Table 3): a change ≥ 5° subgroup (C) (IS = 64, S2AI = 4) and a change < 5° subgroup (NC) (IS = 47, S2AI = 27). In subgroup (C), PI signi cantly increased from 53.3° preoperatively to 59.1° postoperatively at least 50% of IS cohort, with a mean change of 5.8° (p = 0.032).

Clinical Outcomes and Complications
The preoperative back and leg VAS scores and the ODI scores were similar between two groups (Table 4). Although the scores for back and leg, and the ODI scores were greater in IS group than in S2AI group at 3 months postoperatively, the differences between the two groups were not statistically signi cant. At the last follow-up, the back and leg VAS scores were signi cantly lower in IS group than in S2AI group (back VAS score: 3.39 ± 1.26 vs 3.92 ± 1.36, p = 0.044; leg VAS score: 3.52 ± 1.03 vs 3.95 ± 1.12, p = 0.046). The ODI showed no signi cant difference between two groups at the last follow-up.  Table 5 described postoperative complications. The occurrence of SIJ pain was signi cantly higher in S2AI group than in IS group (p = 0.029). The S2AI group had signi cantly more pelvic screw fracture than IS group (S2AI 16.1%% vs IS 3.6%; p = 0.035). There were four postoperative infections in IS group compared with two in S2AI group (p > 0.05). Rates of PJK, pseudarthrosis, wound dehiscence, or screw loosening were similar between two groups.

Discussion
The present study demonstrates three major ndings: (1) Historically, both spinopelvic fusion techniques reported in this paper have demonstrated similar effects in terms of stiffness and stability. [24,33] However, when the differences from baseline to immediately postoperative were compared in the current study, the IS group had signi cantly greater amounts of correction in LL and ULL. Our study is the rst to report that the IS technique provides more cantilever force to obtain more LL and ULL in patients with ASD in comparison with the S2AI technique.
There may be two reasons for this: one is that the S2AI technique insertion point is reported to be 15mm deeper than that in the IS technique. [34] Furthermore, we could observe the curved rod application in the distal area in the coronal plane of S2AI group, which showed worse resistance against distal implant pullout and decreased strain between the screw head and shaft. The S2AI technique is enhanced with iliac buttressing, but the location of the rods is probably too medial to correct more LL. These differences can be further appreciated in Fig. 3 (a graphical rendering of the IS vs. S2AI xation).
(2) PI is a unique anatomical parameter and often considered a constant value after maturity. [ in uenced by various forces due to joint motion or position during supine, sitting, or locomotion behaviors. Thus, the laxity of the SIJ serves as one basis for the change in PI. Nutation increases PI and counternutation decreases PI due to motion at the SIJ. [39] The changes in PI occur in patients who undergo lumbosacral xation using the IS technique while the changes associated with the S2AI technique have not been well established. Five degrees was determined as a cutoff threshold because deviation may occur owing to varied postures during imaging and measurement and the magnitude of SIJ movement in adulthood has been reported to be 1-4° of rotation. [40][41][42] Our hypothesis is that mobility in the SIJ could determine the difference in the nal shape of the pelvis after surgery, and different surgical techniques would lead to different outcomes in this parameter. The SIJ is six times more resistant to lateral forces than the lumbar spine, and approximately one-half as resistant to axial direction and rotation forces. [43] Hence, stress on the SIJ could increase after spinal fusion that could accelerate degenerative changes, resulting in an increase in motion, particularly that which occurs more frequently after SPF. [44,45] Speci cally, we hypothesize that forced extension of the hips and lumbar spine on the operating table could cause anterior rotation of the sacrum into the pelvis, resulting in an intraoperative increase in PI. If the IS technique was used, this change would be xed, with a consequent postoperative increase in the observed PI.
Pelvic retroversion generates a reaction force on the SIJ in patients with sagittal malalignment, destabilizing the joint, particularly if combined with degeneration. During surgery, the S2AI screws were placed through the SIJ, reconstructing the morphology of the pelvis. As a consequence, we observed a dramatic decrease in PI. The discrepancy between our postoperative change in PI and that reported by other studies [46] indicates that the potential mechanism may be related to multiple factors. According to cadaveric data [21], over half of S2AI surgeries violated the SIJ, which could potentially affect PI. In addition, postoperative changes in PI are a complicated biological phenomenon in vivo, which are challenging to test in cadavers or biomechanically. However, this result provided some insight into how spinal alignment may be affected by instrumentation and how it is translated into clinical outcomes, though further prospective study is warranted.
(3) The S2AI Group showed a signi cantly higher rate of pelvic screw breakage than the IS Group. There may be two reasons for this: one is that the acute angle that develops between the screw head and shaft of the screw, may be more prone to failure in the extremes of head-shaft angulation. The second reason may be that the distribution of stresses into two connections with the IS as opposed to a single one with the S2AI are different. Further research is needed to develop a better understanding of these particular types of failure.
In addition, hardware prominence and subsequent pain in the buttocks have been established as complications of the IS technique. [1] In this study, no patients were identi ed as having screw head prominence that either required revision surgery or resulted in pain, wound dehiscence, or poor cosmesis, which was inconsistent with previously reported results. [12,47] The difference in lower rate of wound complications associated with the IS technique may be attributed to the fact that we minimized the screw head pro le when we deeply recessed pelvic screws into the PSIS speci cally to avoid the issue of prominence. Our results suggest that IS xation is an effective method for SPF. Ishida et al.[48] reported signi cantly higher rates of symptomatic implant prominence in an IS xation group. In their study, the screw placement technique was not described and no mention was made of recessing the screw head into the ilium to minimize prominence. Rates of pseudarthrosis at L5-S1 and reoperation due to PJF were similar between their groups.
This study involves some limitations. Selection bias might have in uenced the results. Additional limitations of this study include the fact that the data were obtained from a single-center in a retrospective case-control design with a relatively small sample size, and bone specimen quality. We believe that this study, although with low evidence, has presented useful results for many spinal

Declarations
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.