The kyphotic deformity caused by AS can impair the ability to look forward and lie flat, which might compromise cardiopulmonary and digestive function in severe cases. [21, 22] Osteotomy surgery can not only correct the sagittal imbalance of the spine but also improve cardiopulmonary and digestive function. [5, 23, 24] At present, several corrective osteotomies were reported for the treatment of AS-related kyphosis: SPO, PSO, Ponte, and VCR, among which SPO and PSO are the most widely used procedure. SPO is a posterior chevron-shaped osteotomy that obtains 10 ° correction with a single level. [7, 25] Ponte osteotomy and SPO are often mistakenly used in scientific articles. The notable differences between the two osteotomies are in Schwab’s Osteotomy Classification and the range of resections. An adequate resection of laminae was required in Ponte osteotomy to correct thoracolumbar kyphosis by substantially shortening the posterior column. PSO is a closing wedge osteotomy without lengthening the anterior column, which will accomplish approximately 30° to 40° correction. [8, 12, 26] Neither single-level osteotomy is sufficient to correct severe thoracolumbar kyphotic deformities. One-stage two-level PSO, which can achieve almost 100° correction,is thought to be an effective treatment of AS-related severe kyphosis. However, such a procedure is technically demanding, and requires longer operating time as well as greater blood loss, which increases surgical difficulty and risk. [27, 28] Therefore, we suggest staged osteotomy surgeries to treat AS-related severe and rigid kyphosis. Zhong et al, reported that the mean blood loss was 2560±1109ml in 10 severe kyphotic AS patients, who were treated with one-stage two-level PSO. Zhang et al, also reported that the mean blood loss was 3311±523ml in 9 severe kyphotic AS patients treated with one-stage procedure. In our study, the total blood loss in staged osteotomies was 1998±613ml, which was less than one-stage two-level PSO. Additionally, patients with severe kyphosis usually cannot tolerate one-stage two-level osteotomies because of poor physical and nutrition conditions. Thus, we believe that staged surgery is a relatively safe way in the treatment of severe kyphosis in AS patients, which reduces the surgical difficulty and risk.
In the first-stage, 22 patients underwent a single-level Ponte osteotomy in the lateral position, which is Grade 2 osteotomy in Schwab’s Osteotomy Classification. We performed the Ponte osteotomy from pedicle to pedicle (Fig. 3), which achieved a mean correction of 29.1 °. The correction is mainly obtained by shortening the posterior column and lengthening the anterior column, which might lead to injury of the major vessels, particularly the abdominal aorta.  However, no major vascular injury was recorded in our study. This may be due to the fact that the preoperative CT angiography showed no abdominal aorta calcification in all cases; thus, the elastic aorta could well accommodate to elongation of the spine while the correction procedure. According to our experience, the width of osteotomy is safe within 25mm in Ponte osteotomy.
It is difficult to place AS patients with severe kyphosis in a prone position, which often requires a specially prepared reverse “V” shaped folding bed. In some cases, patients cannot even undergo corrective surgery in a prone position because of the attachment of chest and abdomen. (Fig. 5) Blindness as a complication, possibly caused by local extrusion and ischemic optic neuropathy, was not uncommon during the operation of spinal surgery in the prone position. Qian et al, reported that brachial plexus palsy may occur due to local compression and excessive abduction of the shoulder during the spinal osteotomy in the prone position. In this study, we performed first-stage osteotomy in the lateral position for the first time, which not only solved the problem of setting intraoperative position but also avoided the complications of postoperative blindness and brachial plexus palsy. The complication incidence of the first-stage lateral surgery was 4.3%, which was lower than the average complication rate (13.4%) of osteotomy reported by Qian et al. 
In this study, GK, TK, LL, SVA, and CBVA were used to evaluate the sagittal balance of the spine. Our results showed all parameters were significantly improved after surgery and at the last follow-up. Qian et al, explored the feasibility of single-stage skipping two-level PSO in 10 patients with AS-related severe kyphosis (Cobb> 100 °). Their results showed that GK, LL, and SVA were corrected from 113.4 °, 41.9° and 25.2cm preoperatively to 71.6 °, -44.1°and 5.8cm postoperatively respectively. Zhong et al, showed that the kyphosis angle, CBVA and SVA were improved from 92.0°, 37.6° and 24.1cm preoperatively to 30.0°, –0.6°and 7.5cm postoperatively respectively. In our study, the kyphosis angle correction is similar to previous studies, but superior in SVA restoration.
In this study, 23 patients underwent a total of 46 operations in two stages. The complication rate of the first-stage, second-stage, and the total were 4.3%, 13%, and 8.7% respectively. Because of the adhesive lesions between the dura and the ligamentum flavum, facet, or lamina, cerebrospinal fluid leak is the most common complication.  Additionally, postoperative blindness and brachial plexus paralysis are avoided due to the lateral position.
The present study also has limitations. The sample size was relatively small. Also, retrospective studies suffer from several inherent limitations, such as selection bias and data availability. The sagittal alignment parameters of the spine were evaluated in this study (GK, TK, LL, SVA), however, the spinopelvic alignment parameters such as pelvic incidence, pelvic tilt, and sacral slope were not fully assessed because of the quality of X-ray imaging.