CT and radiographic analysis of sagittal profile changes following thoracoscopic anterior scoliosis surgery

Surgical management of Adolescent Idiopathic Scoliosis (AIS) via the anterior approach has been shown to preserve motion segments, while producing major and compensatory curve corrections comparable to posterior approaches[1‐9]. However, patients with AIS also exhibit a reduced thoracic kyphosis or hypokyphosis[10‐13] accompanying the coronal and rotary distortion components. As a result, surgical restoration of the thoracic kyphosis while maintaining lumbar lordosis and overall sagittal balance is a critical aspect of achieving good clinical outcomes in AIS patients[5, 14‐16].

Anterior surgical approaches appear to be advantageous in this respect, with consistent reporting of increased thoracic kyphosis after surgery[4, 5, 7, 8, 14, 16‐19]. By contrast, previous literature has demonstrated flattening of thoracic sagittal contour and a corresponding decrease in lumbar lordosis following posterior pedicle screw instrumentation in the thoracic spine[7, 16, 20‐22]. Furthermore, the occurrence of proximal junctional kyphosis following posterior stabilisation is high, with reported incidence between 9.2% and 46%[23‐26].

When performing anterior approaches, thoracoscopic (keyhole) anterior spinal fusion (TASF) is an accepted alternative to open surgery in the instrumented correction of major thoracic curves[8, 27‐30]. The thoracoscopic approach reduces chest wall disruption, with less blood loss and soft tissue dissection than open procedures[4, 8, 31], and pulmonary function has been reported to recover as early as 12 months after surgery[32, 33]. To the best of our knowledge, sagittal profile changes have been reported for four existing single centre cohorts of TASF patients to date[4, 8, 18, 29, 34‐36], with reported increases of between 4 to 12° in thoracic kyphosis and 4 to 7° in lumbar lordosis on radiographs at minimum two years after TASF. Taken together, these studies suggest that TASF improves both thoracic kyphosis and lumbar lordosis in AIS patients.

However, the existing studies raise several questions relating to sagittal profile changes following AIS surgery. Firstly, standing sagittal radiographs are notoriously poor quality (Figure1) when attempting to identify thoracic endplates for sagittal Cobb angle measurement, adversely affecting measurement reliability[37‐39]. Secondly, there has been a lack of reporting of localised sagittal profile measures such as the instrumented levels or the thoracolumbar junction. A third question is whether TASF affects the sagittal profile of hypokyphotic patients differently to normokyphotic patients.

Accordingly, the aims of this study were to; (i) use the superior endplate clarity provided by low dose computed tomography (CT) scans to perform both overall and localised sagittal profile measurements before and after TASF for a group of AIS patients, (ii) compare these low dose CT sagittal profile measurements with standing radiograph sagittal profile measurements for the same patients, (iii) document the inter-observer variability associated with supine CT sagittal profile measurement, (iv) assess whether TASF affects hypokyphotic patients differently to normokyphotic patients, and (v) provide a quantitative comparison of previously reported sagittal profile results after posterior, open anterior and thoracoscopic anterior scoliosis correction literature to date.

27 females and 3 males consented to participate in the study. The mean age at the time of surgery was 15.4 ± 3.7 years (range 9.9-27.8). All 30 patients had right sided major thoracic Lenke 1 Type curves with 19 patients further classified as lumbar spine modifier A, 8 as lumbar modifier B, and 3 with lumbar modifier C. The mean major thoracic Cobb angle for the group before surgery was 51.3 ± 7.1 (range 40–66) and decreased to mean 21.9± 8.2 (range 8–33) on the fulcrum bending radiographs. The mean secondary lumbar Cobb angle before surgery measured 31.6 ± 9.0 (range 15–50) and decreased to mean 7.9 ± 7.6 (range 0–28) on active side bending radiographs. Mean T5-T12 kyphosis Cobb angle before surgery on radiographs was 15.6° ± 9.6 (range −8 - 28) such that 17 patients were classified as exhibiting a hypokyphosis, 13 were normokyphotic (Table 1) and of note the largest T5-12 kyphosis was 28°.

The upper level chosen was T5 in 13 cases, T6 in 15 cases and T7 in 2 cases. The lowest instrumented level was L1 in 3 cases, T12 in 20 cases, T11 in 5 cases and T10 in 2 cases. The mean number of levels fused and instrumented was 7.2 ± 0.7 (range 6–8). The postoperative low dose CT was performed at mean 2.2 ± 0.7 years (range 1.8-5.9) after surgery. The mean T5-T12 kyphosis Cobb angle at minimum 2 years after surgery was 27.4° ± 9.0 (range 8–45) showing a mean increase of 11.8° based on the standing sagittal radiographs. There was a single patient that remained by definition hypokyphotic on the 24 month follow-up radiograph (from 8° thoracic lordosis before surgery to 8° thoracic kyphosis after surgery).

Table2 shows changes in sagittal alignment measures on CT two years after surgery for all patients compared to those before surgery. With the exception of thoracolumbar alignment (T10-L2) and sagittal alignment of the motion segment distal to the instrumentation, all changes as a result of surgery were statistically significant. For direct comparison, Table2 also shows all possible sagittal Cobb angles as measured on standing radiographs (marked ‘x-ray’) for the same group of patients and indicates statistically significant differences between CT and X-Ray values.

Figure4 shows T5-T12 kyphosis before and after surgery measured on CT for each individual patient in the study. Table3 gives CT sagittal alignment results for the hypokyphotic and normokyphotic subgroups before and after surgery and indicates statistically significant changes as a result of the surgical correction for each subgroup. The changes of the various sagittal parameters of the NK and HK groups behaved similarly to the sagittal changes of all 30 patients with regards to statistical significance. Table4 shows the changes of all the sagittal parameters for the 4.5 and 5.5 mm rod groups and found there were no statistically significant differences between these groups of patients. Table5 gives the inter-observer measurement variability (mean, standard deviation and 95% confidence intervals) for measurements of T5-T12, T2-T12, T10-L2, and T12-S1 sagittal Cobb angles by the two observers.

Table6 compares the results of the current study with previous studies reporting changes in sagittal parameters on standing radiographs before and at a minimum two years after selective thoracic fusion surgery using anterior or posterior approaches.

Figures5 and6 present a comparison of supine CT versus standing x-ray Cobb angles for thoracic kyphosis (Figure5) and lumbar lordosis (Figure6) respectively, both before and two years after surgery. Linear regression equations are shown on the graph for each best fit line both before and after surgery in the form y = mx + c, where x is the sagittal plane Cobb angle measured on standing radiographs, and y is the Cobb angle measured on supine CT. These regression equations provide a useful means to convert between standing and supine sagittal profile measures. For example, a standing lumbar lordosis Cobb angle of 55º before surgery would be expected to reduce to 0.71 × 55° + 9.7° = 49° with the patient in a supine position. The standard errors of the slopes of the regression equations were 0.121 (T12-S1 lordosis post-op), 0.074 (T12-S1 lordosis pre-op), 0.093 (T5-T12 kyphosis post-op), and 0.071 (T5-T12 kyphosis pre-op), all of which were statistically significant at the P < 0.001 level.

Restoration of normal sagittal profile is an important goal of scoliosis correction surgery. The aim of this study was to provide a detailed analysis of sagittal profile correction following TASF, using both standing plane radiographs and supine low dose CT scans of the same patient group. CT was useful in addition to lateral radiographs due to the superior endplate clarity afforded by the reformatted CT images. The use of supine CT also potentially avoids the inherent variability in upright standing posture due to stance variations between subsequent sagittal radiographs[55]. However, the use of supine CT also raises questions about the applicability of the resulting sagittal profile measurements to clinically relevant standing postures, and in this study we provide a detailed comparison of standing versus supine sagittal profile measurements for the same patients. To our knowledge, no previous study has compared sagittal profile measurements before and after scoliosis surgery between standing radiographs and supine CT, nor reported on detailed changes in sagittal profile using the clearly defined vertebral endplate visualisation afforded by low dose CT. We also wished to compare the results of the current study with existing literature using plane radiographic measurements of sagittal profile following other selective thoracic fusion procedures (Table6).

After selective thoracic fusion, the lumbar spine needs to adapt to the altered shape of the thoracic spine to maintain coronal and sagittal balance[16]. This spontaneous correction of the lumbar compensatory curve in the coronal plane has been evaluated for various surgical approaches with varying reports as to the superiority of correction between anterior and posterior approaches[6, 7, 17, 20, 56, 57]. The post-operative response of the lumbar spine in the sagittal plane is thought to be a consequence of the change in thoracic kyphosis achieved during surgery. A number of recent studies[4, 5, 7, 14, 16] have found that anterior techniques for the correction of thoracic scoliosis are more kyphogenic than posterior approaches. Multiple discectomies and compression along the rod lead to shortening of the anterior column and immediate increases in the thoracic kyphosis at the first erect radiograph after surgery, with further increases reported two years after anterior selective thoracic fusion[5, 7, 8, 14, 16, 18]. Prior studies by our group on the larger cohort have reported complications associated with this type of surgery[8, 41, 58]. In the current cohort of 30 patients, there were 3 rod fractures and 3 top screw pullouts found by the most recent follow-up. Note that as previously reported, rod fracture is associated with a minimal loss of correction and tended to occur in the earliest patients in the series with only 3 rod fractures found from the most recent 150 cases in the larger series.

A number of prior studies have noted the poor quality of sagittal radiographs with regard to the visualisation of vertebral endplates, especially in the mid and upper thoracic regions of the spine. For example, Dang et al[37] reported excellent intra-observer reproducibility for coronal plane radiograph measurements but for sagittal radiographs, examiners were found to have only fair to good reproducibility for angles measured from upper thoracic vertebrae, such as T2 or T5, and poor inter-observer agreement when measuring spinal levels below T9. Dang et al’s paper concluded that sagittal parameters measured on traditional radiographs do not provide valuable information because they cannot be measured reproducibly or reliably. The difficulties with measuring sagittal parameters on lateral radiographs[37‐39] are avoided with reformatted sagittal CT reconstructions due to the superior endplate clarity afforded by this imaging modality. In the current study, the 95% confidence intervals for inter-observer variability of sagittal Cobb angle measurements (range 5-8°, Table5) are comparable with previously published 95% confidence intervals for coronal Cobb angle measurement from supine CT scans[59]. This suggests that the use of CT allows equivalent clarity for either sagittal or coronal plane Cobb angle measurements.

This study confirms that TASF is a kyphosing technique which has a similar corrective effect on patients who are hypokyphotic or normokyphotic before surgery (Table3). Those receiving a 4.5 mm rod had a slightly greater increase of their kyphosis across the instrumented segment than the group receiving the 5.5 mm rod which is in contrast to an earlier study using posterior approaches where the use of larger diameter titanium rods (6.35 vs 5.5 mm) resulted in larger thoracic kyphosis after surgery[50]. However, a recently published paper on 49 TASF cases[36] found similar results to the current study reporting a greater increase in kyphosis when using a smaller diameter rod (4.0 mm stainless steel in earlier patients vs. 4.75 mm titanium alloy) but rather than interpret the difference as being the result of the different implant types, suggested evolving surgeon experience in patient selection was the most likely factor influencing the different sagittal changes. The 4.5 mm rod group in the current study were also the earlier cases in our larger series undergoing TASF so may also have been affected by a similar patient selection issue, although our differences were not statistically significant. All 30 patients in the study had some increase in thoracic kyphosis following TASF surgery according to CT (Figure4) and X-Ray measures, with 26 patients found to be in the normokyphotic range on the minimum 24 months after surgery radiographs. One patient was classified as being hyperkyphotic two years after surgery (T5-T12 kyphosis 45° on CT and X-Ray) and continues to be monitored six years later and to date has not required additional surgery. AIS is a triplanar deformity and in Lenke type 1 scoliosis, the results presented here suggest that anterior correction is capable of addressing the sagittal component of both the instrumented and adjacent non instrumented segments. The corrective forces exerted by single rod anterior constructs results in a flexion moment which increases the kyphosis across the instrumented levels. The un-instrumented lumbar spine must in turn balance the kyphotic curve above so any increase in thoracic kyphosis will see a corresponding increase in the lumbar lordosis of the patient. This is evidenced in the current study (Table2) where T2-T12 kyphosis increased by a mean 8.4° and the T12-S1 lordosis increased by mean 6.2°.

Table6 compares the results of the current study with previous studies reporting changes in sagittal contour after scoliosis correction surgery. This table shows that posterior approaches either exacerbate the existing thoracic hypokyphosis (at worst 12 degrees[20]), or only achieve small increases in kyphosis in the order of 1-2°. By contrast, anterior thoracic fusion procedures consistently increase T5-T12 kyphosis by between 4 - 12° at two years after surgery. With respect to lumbar lordosis, Table6 reports lumbar lordosis flattening as much as 7.4° following posterior selective thoracic procedures, whereas again by contrast anterior approaches report a deepening of the lumbar lordosis by as much as 8.6° in response to the kyphosing surgical effect in the thoracic spine. The results of the meta-survey of prior studies in Table6 suggest that anterior correction is more capable of addressing the sagittal component of both the instrumented and adjacent non instrumented segments for AIS patients.

Use of the supine position for CT-based sagittal profile measurement clearly changes the geometry of the spine relative to the standing posture, but the comparative results in this study (Table2, Figures5,6) show that there is a predictable (linear) relationship between supine and standing sagittal profile measurements. Of note is that mean kyphosis across the instrumented levels after surgery changed minimally between supine and standing, whereas the uninstrumented lumbar lordosis and T5-T12 kyphosis each demonstrated significant differences (mean 4.8°) due to the change of posture. It is not being suggested that CT scans should replace standing radiographs for scoliosis assessment, but for the group of patients examined here for research purposes, the paired CT data before and after surgery uniquely provided a superior imaging modality (in terms of image contrast and endplate clarity) for analysing the effects of TASF on sagittal plane deformity. There are both advantages and disadvantages to supine measurement of sagittal profile. Use of the supine position provides an ‘unloaded’ configuration of the spine which is not subject to variations in standing posture due to arm positioning[46, 55, 60], time of day[61], or muscle activation strategy[62, 63], all of which can affect sagittal Cobb measurements. Further, relative rotation between the pelvis and ribcage can vary between subsequent standing radiographs whereas the supine position standardises many of these variables. Supine measurements are also valuable in biomechanical modelling of scoliosis, since the supine position provides an approximate zero load configuration for the spine which can be used as a starting point for biomechanical simulations. A disadvantage of supine imaging and a limitation of this study is that sagittal balance and the role of pelvic incidence in the standing position cannot be assessed. Further, the standing position is relevant to a condition such as scoliosis where gravity is known to affect the magnitude of the deformity. Recent advances in multi-slice CT are allowing lower radiation doses and faster acquisition times which will make CT an increasingly useful research tool for three-dimensional biomechanical studies of scoliosis correction. Also, low dose standing biplanar systems (such as EOS) are expected to play an important future role in scoliosis imaging and surgical planning.

Thoracoscopic anterior instrumented fusion significantly improves global thoracic kyphosis (T2-T12), thoracic kyphosis (T5-T12), lumbar lordosis (T12-S1) and instrumented segment kyphosis while simultaneously correcting and stabilising the coronal and rotational plane deformities. The results of this study show that the technique reliably increases thoracic kyphosis and lumbar lordosis while preserving proximal and distal junctional alignment in thoracic AIS patients.