As the course of the disease progress, the rigid TLKD of AS patients may deteriorate, causing the center of gravity shifts anteriorly accordingly, which triggers backward pelvic rotation associated with the maximal hips’ extension to compensate for the sagittal malalignment[4, 12, 15]. Currently, PSO has been widely recognized as the effective and preferable technique to reconstruct sagittal alignment, placing the pelvis in a neutral position and restoring the ability to see ahead[4, 8–10, 16, 17]. In operation, the closure of osteotomy gap is obtained by collective rotation of the cranial and caudal vertebral column at the hinge[17]. Therefore, suitable adjustment of ROC should be taken into consideration based on the individual preoperative deformity features or osteotomy level.
In the surgical decision-making of one-level PSO for AS, the option of the ideal osteotomy level was a pivotal factor contributing to the results of correction. There were usually two options being suggested: osteotomy at the apical vertebra or osteotomy distal to apex[3, 18]. Generally speaking, the apical vertebra was at the angular pathology where it was easiest to operate in the prone position, and a favorable sagittal thoracolumbar curve could be obtained. However, most osteotomies tend to be performed over more caudal vertebra for the followed reasons: first, it decreased the incidence of injury to the neural elements; second, a more caudal osteotomy level might generate more profound correction in AS patients, due to a larger part of trunk being moved more posteriorly. In AS patients, thoracolumbar and lumbar regions are usually the most impacted regions by kyphosis, and the junction of superior and inferior body and the stress concentration in prone position commonly lied at L2 or L3, so facilitating easier and more efficient correction following osteotomy. Furthermore, for AS patients with severe TLK deformity, the goal of PSO is not only the reconstruction of LL but also the improvement of kyphotic deformity to avoid the postoperative inability of lying flat, which was distinct from ASD patients. As Chen et al[3] observed, the maximal correction might be achieved when performing PSO at L2 or L3 in AS-related thoracolumbar kyphosis. Diao et al[18] also pointed out that patients with lordotic lumbar profiles are suitable for L1/L2 PSO while L2/L3 PSO is appropriate for patients with kyphotic lumbar profiles. Hence, the upper lumbar vertebra (L1-L3) was the most commonly used osteotomy site, and currently this study mainly focused on the impact of the upper lumbar PSO level on ROC.
There has been a general assumption that a greater OVA might achieve more correction amount, whereas the relationship between the osteotomized level and the magnitude of correction has been debated extensively[8–10, 19]. Several studies revealed that a more caudal osteotomy level might generate more profound correction in AS patients, due to a larger part of trunk being moved more posteriorly[7–9]. While Lafage et al[10] confirmed that the level of osteotomy correlated with the variation of PT in a line fashion, which didn’t impact the absolute sagittal correction (SVA). In the present study, the average ROC value was 57.89%/42.11%, which means 57.89% of the osteotomy closure applied in the truncal restoration and 42.11% of that was used in pelvic retroversion. Nevertheless, there was no significant difference between the osteotomy level and ROC (P = 0.155, Table 2). Specifically, a more caudal osteotomy level didn’t decrease the percentage of truncal closure, which was different from the results of Diebo et al for ASD patients[11]. Because the rebalance of the spine couldn’t be simplified as a “lever arm” above or beyond the PSO site but involved the spontaneous pelvic adaptation[20]. However, for AS patients with limited hip function or complete ankylosed hip joint, the correction of TLKD is accomplished by the closure of upper part of the OV absolutely due to the difficulty of postoperative adaptive change of pelvis. Hence, osteotomy distal to apex could be considered to avoid the risk of negative spinal imbalance caused by excessive osteotomy amount. In a general way, the selection of OV is not as straightforward and the option is based on other factors, such as the preoperative sagittal lumbar profile, the location of kyphotic apex, hip joint function and surgeon’s preference[3, 18].
The role of osteotomy level itself is not the driver for ROC based on the present findings. Therefore, surgeons should pay more attention to preoperative characteristic deformity. For example, Van Royen et al[9] adopted the combined use of SVA, CBVA and SS for planning corrective osteotomies in AS. However, this method was not individual since the SS was set as an identical value for each patient. Subsequently, Min et al[21] reported the significance of the whole-body kyphosis angle in surgical planning but ignored the role of pelvic parameters. Similarly, Le Huec et al[22] described a method based on a global evaluation of the whole-body balance to predict the desired correction amount for an appointed patient, whereas Le Huec et al didn’t consider the osteotomy level. The above-mentioned methods either undervalued the response of pelvis adapted to the sagittal imbalance or achieved equivalent pelvic correction regardless of osteotomy level. Song et al[23] in 2013 defined postoperative PT on the basis of an equation proposed by Vialle and calculated the needed OVA for individual patient. It’s worth noting the method raised by Song et al considered both pelvic compensation and spinal alignment[23].
Further analysis of patients by closure group (TC, EC, PC) disclosed that there were statistical significances among three groups concerning preoperative SVA, ST, PT and PI (Table 3). The patients (Figure. 2) in the TC group presented with a significant greater pre-op SVA than patients in the EC group which in turn had a significant greater SVA than patients (Figure. 3) in the PC group. The inverse result could be noticed in regard to PT and ST. This phenomenon revealed that the patients with more pelvic retroversion (larger PT), up to a desired SVA, achieved more pelvic closure to eliminate the compensation of pelvis and lower extremities as much as possible, as illustrated in Table 4. Theoretically, AS patients with a high PI would be more able to compensate for TLKD by regulating pelvic retroversion than one with low PI, as the maximal retroversion produced by a high PI pelvis is superior to a low PI one. However, Roussly et al[5] observed that the mean value of SS was easier to reach 0° in low PI than in high PI. They accredited this observation to the restricted hip extension which limited a greater backward rotation of the pelvis. Qian et al[24] investigated the impact of preoperative sagittal parameters on the postoperative sagittal alignments and revealed that there was more likely trend toward failed sagittal realignments for AS patients with either larger preoperative SVA or PI following one-level lumbar PSO. Similarly, Schwab et al[25] concluded that ASD patients with failed realignments had greater preoperative spinopelvic deformity (large SVA, PI and PT). Apparently, AS patients with larger PI require more correction of LL to generate the harmonious relationship between the pelvis and the spine. The current study demonstrated that patients with larger PI (TC vs. PC, P < 0.001) achieved more pelvic closure (Table 4) and each group of patients obtained satisfactory postoperative sagittal alignments. Therefore, it’s essential to reach a more elevation of lower extremities for AS patients with higher PI intraoperatively.
There are several limitations of the present study. Firstly, the patient might not stand in a standard neutral postures due to the patients’ functional recovery period when taking radiographic imaging postoperatively. Secondly, the present study couldn’t take parameters of lower extremities into consideration on account of incomplete imaging of lower extremities, whereas it is worthy of further intensive study.