Adult spinal deformity is a broad category that encompasses a diverse group of spinal malalignment patterns. It may range from a simple bi-planar deformity to more complex three- dimensional deformities with significant loss of coronal and sagittal alignment. For patients presenting significant rigid deformity, pedicle subtraction osteotomies can be utilized to create substantial changes in local and global alignment. This surgical technique requires multiple fixation points above and below the level of osteotomy. Extension of fusion beyond the adjacent levels is frequently required due to compensatory rigid deformity, additional deformity in adjacent regions, and the need to prevent adjacent level failure and further deformity. However, when possible, surgeons may attempt to correct the most rigid and deformed portions of the spine while leaving long segments unfused. This has the advantage of allowing continued flexibility and more normal motion in the nonfused portions. However, these unfused segments are dynamic and have the potential for ongoing changes in alignment to occur.
In our retrospective series, the majority of surgeons chose to include distal or proximal fixation points in the thoracic or lumbar spine as part of the major deformity correction. These extended fusions commonly spanned from T4 to the pelvis (78% of the patients were fused to the sacrum). In the thoracic PSO group, 17 of 29 (59%) patients underwent a “long” fusion while 71 of 105 (68%) lumbar PSO patients had fusions that extended into the thoracic spine. This difference in preference for a limited fusion in the thoracic group is likely related to the surgical indication and pathology that was treated. Short-segment fixation may be acceptable for a PSO that plans on addressing a primary coronal curve with associated suboptimal sagittal alignment. The thoracic group was, of note, 10 years younger (mean age of 45 years old versus 55 years old, P < 0.001) than the lumbar group, which also may have influenced decision making to maximize the sparing of lumbar motion segments.
Patients who received selective thoracic fusions also tended to have a larger lumbar lordosis than those receiving long fusions (70° versus 52°). Conversely, the selective lumbar fusion group had a lower preoperative thoracic kyphosis, 22° versus 33°. It appears that in the lumbar fusion group long fusions were selected for patients with more substantial sagittal plane mismatch between the thoracic and lumbar spines: lower lumbar lordosis and greater thoracic kyphosis. Long fusions were also selected in patients with thoracolumbar kyphosis, most likely to avoid accelerated junctional failure.
Surgeons commonly choose to perform longer fusions in the setting of adult spinal deformity in an effort to prevent accelerated degeneration, deformity, or kyphosis of the adjacent spinal segments and regions. Proximal junctional failure after lumbar PSO is a well-recognized entity, while distal junctional failure after thoracic PSO is less well understood [19
]. Most of our literature regarding distal fixation points is derived from the Schueurmann's kyphosis literature and recommends that the distal extent of the fusion extends to the last lordotic disc [20
]. Additionally, while avoiding extended lumbar fusions is felt to be critical for normal motion and to prevent adjacent level disease, extending fusions up into the thoracic region does not seem to elicit the same concerns of functional limitations or accelerated degeneration.
Interestingly, the findings in this study demonstrate that fusions extended into the upper thoracic spine after lumbar PSO, only increased thoracic kyphosis minimally (+6°). However, in limited lumbar fusions, the unfused thoracic region demonstrated mean reciprocal changes of increased thoracic kyphosis by 13° (significantly greater than in long fusion group, P
= 0.002). This reciprocal correction resulted in improved regional sagittal alignment (thoracic kyphosis post-op. mean 35°) and did not deter from the improved global alignment (SVA from 14
cm to 4
cm post-op.) and pelvic version (PT 32° to 25° post-op.).
Following thoracic PSO, when the fusion extended into the lumbar spine, there was a mean increase in lumbar lordosis (+8°). On the other hand, when the lumbar spine was left unfused, the spontaneous reciprocal change led to a decrease of the lumbar lordosis 70° to 62° (P < 0.05). These changes allow a normalization of the lumbar lordosis in regard to the pelvic incidence.
The etiology of the reciprocal change is likely multifactorial. The preoperative spinal alignment represents the efforts of a maximally compensated spine, despite global malalignment. The compensatory portions of the thoracic or lumbar spine act to maximally increase or decrease their curvature to allow the head to be centered over the pelvis. It appears that with the correction of abnormal regional alignment in one portion of the spine, the unfused regions accommodate by “relaxing” the compensatory preoperative alignment to a more normalized one. The reciprocal changes may also reflect the central nervous system using this portion of the spine to once again center the head over the pelvis. The necessary degree of kyphosis or lordosis for optimal global spinal alignment can thus be achieved.
Kim et al. evaluated parameters that predicted optimal lumbar lordosis and sagittal alignment in adult spinal deformity (ASD) patients fused from the thoracolumbar spine (T9-L2) to L5 or S1 [23
]. Patients with postoperative lumbar lordosis that exceeded thoracic kyphosis by 20°or more demonstrated optimal sagittal balance (defined as C7 plumb line falling within 3
cm of the posterior aspect of the L5-S1 disc) at minimum two-year followup. Patients with optimum sagittal balance, in turn, demonstrated superior health-related quality of life (HRQL) scores compared to patients with C7 plumb line greater than 3
cm from the posterior L5-S1 disc. However, all patients demonstrated postoperative progression of thoracic kyphosis over time. Kyphosis progression was similar between the optimal group and suboptimal groups; however, the suboptimal group had lower final postoperative lumbar lordosis and a greater percentage of patients with UIV ending in the lumbar spine, and, therefore, less ability to control reciprocal changes in the cephalad segments, compared to the optimal group. This phenomenon was recognized by Rose et al. who found that thoracic kyphosis did not change from preoperative to postoperative in patients fused to T5 or cephalad after lumbar PSO; however, patients fused caudal to T5 demonstrated significantly increased postoperative thoracic kyphosis (reciprocal change) [9
]. The authors recommended the formula: lumbar lordosis ≤ 45°, thoracic kyphosis, pelvic incidence to predict optimal lumbar lordosis and sagittal alignment needed for lumbar PSO. However for selective fusion, reciprocal changes do occur, and thoracic kyphosis is not a static measurement. Consequently, formulas that do not appreciate this reciprocal change can underestimate the amount of PSO required to provide appropriate global sagittal alignment.
This is a retrospective review of PSO's that were performed at multiple medical centers, which is an inherent limitation of the study. Variations in the specific technique of PSO as well as surgical indications are not accounted for. An additional limitation is the short followup for these patients with postoperative radiographs measured at three months. Further investigation into detailed analyses of reciprocal change patterns and evolution over time will be undertaken.