We obtained informed consent from all the patients and also approval from the Institutional Review Board of Seoul National University Bundang Hospital (approval No. B-1603/338-109). Consecutive adult patients (>18 year-old) who underwent PSO for fixed sagittal imbalance were sorted and analyzed from a single institutional database. All patients were operated by a single surgeon and observed for more than 2-year. Among total of 36 consecutive patients, five patients were excluded from this study because of the loss of follow-up or short of follow-up period (12.0-19.9 months of follow-up). Finally, 31 patients who underwent PSO between February 2012 and March 2016 were enrolled in this study. All patients suffered from persistent severe pain (visual analogue scale [VAS] ≥6), neurological deficit, or progressive deformity in spite of sufficient conservative treatment. Conservative treatment was comprised of physical therapy, medication, back muscle exercise, brace, and epidural steroid injection for more than three months. Decision for performing PSO was made for patients with type 2 sagittal imbalance (global sagittal imbalance) and lumbar curve without motion segment because of ankylosing spondylitis, severe degenerative change, or previous circumferential fusion along multiple vertebrae.

Long-cassette (14×36 inch) anteroposterior and lateral standing plain radiographs obtained before surgery, immediately following surgery, and at the most recent follow-up were analyzed retrospectively. Sagittal vertical axis (SVA) was determined as the distance from the C7 sagittal plumb line to the perpendicular line drawn from the superior posterior corner of the S1 vertebral body. The thoracic kyphosis (TK) was measured between the lower endplates of T5 and T12. LL was measured between the upper endplate of L1 and S1. The angle at the PSO site was defined between the upper endplate of the adjacent upper vertebra and the lower endplate of the adjacent lower vertebra. Proximal junctional kyphosis (PJK) was defined as a proximal junctional angle of 10° greater than the preoperative value.

Clinical data including sex, age, and primary or revision surgery status were collected by medical chart review, retrospectively. Operative details such as the level of osteotomy, number of fused segment, operation time, estimated blood loss (EBL), transfused volume of red blood cell (RBC), and length of hospital stay were obtained. Complications were evaluated and divided into three groups regarding time of onset; intraoperative, early-postoperative (<2 weeks), and late-onset (≥2 weeks). Clinical outcomes were assessed by Scoliosis Research Society-22 (SRS-22) questionnaire, prior to operation and at the ultimate follow-up (at least 2-year after operation). Patient questionnaire was filled out by every patient themselves. Then, physical assistant registered the data into the database immediately after the completion of the survey.

The patients were positioned prone on the radiolucent operative table with chest bolster and pelvic girdles (Jackson table), to achieve maximal lordotic curvature and to avoid abdominal pressure. Intraoperative neurophysiological monitoring using somatosensory and motor evoked potentials was applied to every patient. Intraoperative administration of tranexamic acid and installation of cell salvage system were done in order to reduce surgical bleeding and RBC transfusion. Subperiosteal dissection was performed to expose bony structure as far as the tip of transverse process. Placement of pedicle screws with or without pelvic screws (S2 alar-iliac screws) was performed using freehand technique [8]. Multi-level posterior column osteotomies (PCO) as a Ponte procedure were performed at every lumbar vertebra.

Removal of all posterior elements is conducted at the PSO segment including the spinous process, lamina, and the superior and inferior articular processes adjacent to the pedicle. In addition, cephalad and caudal laminae were undercut to avoid dural buckling or neural impingement during closure of posterior column. Following cutting of proximal junction of transverse processes bilaterally, lateral surface of the vertebral body was carefully dissected while protecting segmental vessels. Next, pediculectomy and wedge-shaped cancellous bone removal were done. Usually, a temporary rod was placed across the osteotomy site to prevent subluxation or premature closure. Lateral cortex was removed with wedge-shaped, and remnant posterior cortex was depressed and fractured with downward cutting curette. Consequently, pedicle subtraction was accomplished with downward force by an operator and secured with multiple rod construct. After completing the fixation procedure, autografts and allografts were placed over the laminae, facet joints and transverse processes.

Statistical analysis was performed using IBM SPSS version 22.0 software for Windows (IBM, Corp., Armonk, NY, USA). The distributions of the variables were demonstrated as the mean value and standard deviation. Paired t-test was used to analyze the paired variables before and after osteotomy. Statistical significance was set at p<0.05.

The patients included in this study were 27 females and four males, with a mean age of 66.0±9.3 years (Table 1). A mean follow-up period was 33.2±10.5 months. In terms of diagnosis, there were 16 cases of degenerative lumbar flatback, 11 cases of postoperative flatback, two cases of posttraumatic kyphosis, and two cases of tuberculous kyphosis. 64.5% (20/31) of patients had prior spinal surgeries, with a mean number of 1.2±1.2.

Operative details are summarized in Table 2. The most common level of osteotomy was L3 (15 cases), followed by L4 (eight cases), L1 (four cases), L2 (two cases), and T12 (two cases). The mean number of fused segments was 9.6±3.5. Pelvic fixation, by means of S2 alar-iliac screws, was conducted in 27 patients, and 19 patients received anterior column support by transforaminal lumbar interbody fusion at least one level. The mean operative time was 475.9±160.5 minutes. The amount of surgical bleeding was 1406.1±932.1 mL, and the volume of transfused RBC was 1063.0±568.1 mL. The length of hospital stay was 13.0±5.8 days. The preoperative SRS-22 score was 2.3±0.7, which improved to 3.2±0.8 at the final follow-up (p=0.000). All subscales in SRS-22 were significantly improved at the final follow-up, compared to those of preoperative value. The score of satisfaction with management was 3.7±0.9 (Table 3).

The mean pelvic incidence was 54.5±9.5° (Table 4). TK preoperatively and postoperatively was 16.2±18.5° and 27.0±12.8°, respectively. LL was changed from 7.0±28.9° to -50.2±13.2°. The correction angle of LL was 57.3±25.6° and sustained until last follow-up (loss of correction of 2.9±6.1°). The osteotomized vertebral angle was changed from 15.6±20.1° preoperatively to-16.1±19.4° postoperatively. The angle of correction in osteotomized vertebra (PSO angle) was 33.7±13.5°. The difference of correction angle between LL and PSO segment was 23.6°. The preoperative SVA was 189.5±93.0 mm, where the postoperative value was decreased to 12.4±40.8 mm. The loss of correction of SVA was 32.8±40.7 mm (last SVA of 44.8±48.2 mm).

Twenty out of 31 patients suffered from 28 complications (Table 5). Thus, the rate of total complications was 90.3% (28/31). Six cases of complications were occurred intraoperatively. Dural tear was most common (four cases), followed by screw malposition and chin sore (one case each).

There were eight complications during early postoperative period (<2 weeks). There were two cases of postoperative motor deficit, which required revision surgery in one patient. One patient suffered from postoperative voiding difficulty which required continuous intermittent catheterization. One patient suffered from postoperative wound dehiscence, and underwent revision surgery with local flap. Other complications were urinary tract infection, hydronephrosis, atrial fibrillation, and pulmonary effusion.

Late postoperative period complications (≥2 weeks) were identified in 14 cases. There were eight cases of PJK, one of which required revision surgery. There were four cases of rod fracture which required additional surgeries. Screw loosening was identified on routine follow-up in two patients with close observation. Additional operations were necessary in nine cases as a consequence of such complications.