Introduction

Surgical management of Adolescent Idiopathic scoliosis (AIS) remains controversial, particularly regarding optimal posterior fusion strategy [1, 2]. However, consensus appears to exist in the necessity to restore a global spinal alignment in the sagittal plane while correcting the thoracic hypokyphosis [3–7].

The recent development of deformity correction techniques via instrumented posterior fusion using only pedicular screws has been reported to provide excellent results in terms of axial correction and coronal plane restoration [8]. However, all-screw constructs have also been reported to create loss of thoracic kyphosis [9], and according to reciprocal interactions occurring in unfused segments, this loss of kyphosis can be related to a secondary loss of lumbar lordosis potentially leading to iatrogenic flatback [10].

Conversely, hybrid instrumentation strategies which do not employ pedicular screws in the thoracic region may have potential advantages such as reduction of the neurological risk due to screws misplacement [11]. Furthermore, these hybrid constructs may be more efficient for the restoration of thoracic kyphosis with little disadvantage in terms of coronal correction of the deformity compared to pedicular screws [5, 12].

According to these reported findings, we have formulated the hypothesis that a correction of the thoracic hypokyphosis would be associated to a compensatory increase of lumbar lordosis in unfused segments below instrumentation ending in the upper lumbar spine. The aim of this study was therefore to evaluate the reciprocal sagittal alignment changes between thoracic kyphosis and lumbar lordosis after posterior instrumented hybrid fusion with sublaminar bands and to determining if correlation exists between these reciprocal changes and the amount of correction in the coronal plane in AIS patients.

Methods

Study design and inclusion criteria

Thirty adolescent consecutive patients diagnosed with AIS were included in this retrospective single center and single operator study.

Inclusion criteria were defined as follows: AIS classified Lenke 1, 2 or 3 with coronal Cobb angle superior to 45°, surgical treatment by posterior instrumented fusion with sublaminar bands and a lower instrumented vertebra above or at L2, an absence of preoperative systemic disorder or neurologic deficit and a normal preoperative spinal axis MRI. Exclusion criteria included: lumbar major curve, the necessity to perform an anterior discectomy, congenital or neuromuscular scoliosis, an abnormal MRI or the impossibility to obtain preoperative neuro-monitoring.

Data collection and radiographic analysis

For each patient, full-spine postero-anterior and lateral radiographs were obtained in standing position preoperatively, at 3 months and 2 years postoperatively. On every radiograph the coronal Cobb angle, the T4–T12 thoracic kyphosis (TK) and the L1–L5 lumbar lordosis (LL) were measured by an independent observer.

Surgical procedure

During the preoperative evaluation, selection of the lower instrumented vertebra was done using lateral bending radiographs. The surgical procedure was conducted under general anesthesia in prone position using a Jackson table and under neuro-monitoring surveillance (SSEP and MEP). Posterior instrumented fusion was performed according to a protocol previously described [13] and summarized hereafter. After posterior exposure of the spine, an hybrid construct was systematically applied including: pedicular screws in the caudal area between T11 and the lower instrumented vertebra (superior or equal to L2); sub-laminar hooks in compression in the cranial area, two 5.5 mm titanium rods, and sublaminar bands (Universal Clamp, Zimmer, Bordeaux) on 3–6 levels on the concavity of the deformity and 1 or 2 on the convexity for stabilization purpose.

Reduction of the deformity was accomplished in a progressive and multi-level manner by applying a tensioning force on the sublaminar bands affixed to two pre-bent rods. Using the reduction tool, it was therefore possible to apply a postero-lateral translation force on the vertebrae, leading to a posterior traction effect of the spine toward the pre-bent rods (Figs. 1, 2).

Fig. 1
figure 1

Illustration of a hybrid construct with sub-laminar bands in the concavity of the deformity. Once the pre-bent rods were positioned, reduction of the deformation was accomplished through progressive tensioning of the bands

Full size image
Fig. 2
figure 2

Postero-medial translation force applied on the apical vertebrae by progressive tensioning of the sub-laminar band in the concavity of the deformity

Full size image

A postero-lateral bone graft was then performed using autologous bone material from the spinous processes augmented with synthetic bone substitute (Biosorb, SBM, Lourdes).

Statistical analysis

Changes in radiographic parameters (coronal Cobb angle, TK and LL) between preoperative and postoperative evaluations were analyzed using a paired Student t test. Reciprocal changes between TK and LL, as well as the relationship between sagittal reciprocal changes and the amount of coronal correction were assessed by a Pearson correlation coefficient. A subgroup comparison analysis was then conducted on patients with a preoperative TK <20° or ≥20° using an impaired (>) t test. For all statistical analysis, the level of significance (alpha risk) was set as 5 % (i.e., p < 0.05) (Fig. 3).

Fig. 3
figure 3

Pre- and postoperative radiographs showing the coronal correction obtained with the hybrid construct using sublaminar bands and clamps

Full size image

Results

Demographic data and surgical parameters

Thirty consecutive AIS patients (28 females and 2 males, mean age 14.7 years, with closed triradiate cartilage) diagnosed with Lenke type 1, 2 or 3 curves were analyzed in this retrospective study. Full-spine coronal and lateral radiographs were available for preoperative, 3 months and 2 years postoperative follow-up for all patients (Fig. 4).

Fig. 4
figure 4

Sagittal radiographs (same patient): preoperative evaluation (left) showed a thoracic hypokyphosis. Surgical restoration of the TK is visible on the 3 months postoperative X-ray (middle) and reciprocal improvement of the lumbar lordosis appeared after the third postoperative month (2 year follow-up on the right)

Full size image

An average of six sub-laminar bands and clamps (range 4–8) were used in order to reduce the deformity during surgery. No intraoperative changes in neuromonitoring recordings were noted, however, one patient exhibited delayed motor activity of the lower extremities with full-recovery 1 h later.

Global results of radiographic measurements (Table 1)

Preoperative analysis across the study group demonstrated a mean coronal Cobb angle of 53.6° (SD = 7), a thoracic kyphosis of 19.7° (SD = 8) and a lumbar lordosis of 43.9˚ (SD = 11). According to Lenke’s Thoracic Sagittal modifier, 6 patients showed a preoperative thoracic hypokyphosis (TK <10°), 21 were normal (10 < TK < 40°) and 3 patients were hyperkyphotic (TK >40°).

Table 1 Summary of radiographic measurements on the whole series, p value was calculated between preoperative measurement and follow-up evaluations
Full size table

Three months after the surgical procedure, the mean coronal Cobb angle was significantly (p < 0.001) reduced by 68 % on average, without significant modifications between 3 months and 24 months postoperatively (p = 0.314).

Between the preoperative evaluation and 3 months postoperatively, the mean thoracic kyphosis on the study population was significantly increased (average 6.5˚, SD = 5; p = 0.005) without further significant modification between 3 and 24 months postoperatively (p = 0.275). At final follow-up, none of the patients showed a hypokyphosis, 28 had a normal TK and 2 had a TK >40° according to Lenke’s thoracic sagittal modifier.

The lumbar lordosis across the study population was increased (average 3.4°, SD = 6; p = 0.009) between preoperative evaluation and last follow-up, without significant change during the three first postoperative months (p = 0.845).

Results on the whole population are summarized in Table 1.

Subgroup analysis by preoperative thoracic kyphosis (Table 2)

Results from the subgroup based on preoperative TK showed a significantly (p < 0.05) improved coronal Cobb angle in both groups (preoperative TK <20° or ≥20°), but different postoperative radiographic sagittal measurements.

Table 2 Summary of radiographic measurements by subgroups (TK <20°, TK ≥20°)
Full size table

At final follow-up, patients with a preoperative hypokyphosis (TK <20°), exhibited on average a significant increase of the instrumented TK of 10.3° (p < 0.001), with a reciprocal LL increase of 5.5° (p = 0.005). On the other hand, for patients with a normal preoperative kyphosis (TK ≥20°), the instrumentation maintained a normal TK and secondary a normal LL.

Results of the subgroup analysis are summarized in Table 2.

Correlation between radiographic parameters (Table 3)

Based on initial and latest follow-up, across the study population, a significant correlation was found between restoration of thoracic kyphosis and increase in lumbar lordosis (p = 0.037, r = 0.382). However, no significant correlation was found between sagittal reciprocal changes and amount of correction of the Cobb angle in the coronal plane (p = 0.442 and p = 0.691 for kyphosis and lordosis, respectively).

Table 3 Values of Pearson correlation coefficient between the different radiographic parameters, for the study group
Full size table

These results were also found in the group of patients with a preoperative TK <20°, for whom the improvement of TK was also associated with an increased LL (p = 0.05, r = 0.423), without significant correlation to correction of the Cobb angle (p = 0.232).

Discussion

Objectives and technical considerations in the surgical management of AIS bear several differences when compared to adult spinal deformity treatment.

Westrick and Ward [14] summarized the five main objectives in the surgical treatment of AIS: arrest deformity progression through a solid fusion, obtain a permanent correction of the deformity, improve appearance of the patient, improve functional outcomes and decrease the risk of degenerative disorders during adulthood.

According to these objectives, AIS treatment strategy will be mainly based on the radiographic characteristic of the deformity while pain and disability are the major parameters of the treatment strategy in adult deformity [15]. Of note, sagittal malalignment (loss of lumbar lordosis, anterior malalignment and pelvic retroversion) have been reported as the main drivers of disability in adults with spinal deformities [16, 17].

Restoration of a satisfactory postoperative sagittal alignment therefore appears a crucial goal, in the management of AIS, whatever the surgical strategy or implant system applied to obtain a solid arthrodesis.

The aim of this study was to analyze the sagittal reciprocal interactions between surgical correction of thoracic kyphosis and improvement of lumbar lordosis, and to search for a correlation between these reciprocal changes and the amount of correction of the coronal Cobb angle.

Hybrid constructs and kyphosis correction

Lehman et al. [9] reported in 2008, on a series of 114 patients, with an average loss of thoracic kyphosis of 9.9° when performing reduction by all-screw constructs but without using rod-derotation maneuvers. When applied, the rod-derotation maneuvers can lead to an improvement of the thoracic kyphosis with an average gain reported to be 5° by Suk et al. [8], but this can also lead to coronal hypercorrection of the deformity with trunk inclination and shoulder asymmetry [18, 19]. Results from this study confirm the ability of hybrid constructs using sub-laminar bands and clamps to maintain TK for normokyphotic patients, and to restore a satisfactory TK for hypokyphotic patients, by applying a posterior traction force on the spine toward pre-bent rods, with an average gain of 6.5° at last follow-up. Furthermore, this technique can limit the risk of hypercorrection due to the absence of rod-derotation maneuvers. A crucial point is the necessity to choose the appropriate lower instrumented vertebra. In this study, we used supine side-bending radiographs in order to take into account curve flexibility but recent reports revealed that the use of fulcrum-bending radiographs was more likely to approximate the coronal postoperative outcomes [20].

The impact of vertebral derotation on sagittal alignment and rib hump is also an important parameter. In our experience, no direct vertebral derotation forces were applied to vertebrae. However, the postero-medial translation forces applied on the concavity of the curve can decrease vertebral rotation [21], but this parameter was not measured in our series. It has been recently reported that direct vertebral body derotation was associated with a significant decrease of rib hump [22], even if for large rib prominences better inclinometer readings were achieved with thoracoplasty alone [23].

Furthermore, impact of direct vertebral derotation (DVD) on sagittal alignment is still controversial. According to Kadoury et al. [24] there is a 74 % axial rotation correction and a better 3D correction achieved with DVD compared to Harrington–Luque, and Cotrel–Dubousset instrumentation but all of these techniques were associated to a loss of lumbar lordosis and loss of correction has been described at 2-year follow-up especially in skeletally immature patients [25]. More recently, Hwang et al. [26] showed on 323 patients that DVD did not worsen the sagittal profile compared to rod-derotation technique but both groups showed a postoperative decrease of thoracic kyphosis, highlighting the importance of appropriate rod contouring. Similar findings were also reported in another study based on 30 patients [27] where the authors concluded that DVD was responsible for a loss of thoracic kyphosis and lumbar lordosis. The same authors therefore recommended pulling posteriorly the concave side of the spine in order to preserve sagittal alignment, which is the equivalent of the technique we used in this study with sublaminar bands and clamps.

Another concern with hybrid construct is the possible occurrence of postoperative crankshaft phenomenon as described by Dubousset et al. [28] as a consequence of a continued anterior spinal growth in the presence of a posterior fusion. Tao et al. [29] have suggested that hybrid construct were not likely to prevent crankshaft phenomenon when compared to pedicle screws construct. This complication is essentially encountered with skeletally immature patients, and was not seen in our series as all patients showed a closed triradiate cartilage at the time of surgery.

Reciprocal sagittal interactions and correlations

Results from this study confirm the initial research hypothesis on the presence of sagittal reciprocal interactions in the unfused segments of the spine. Conversely to surgical constructs where the loss of thoracic kyphosis is followed by a secondary loss of lumbar lordosis [10], our results demonstrate a significant correlation between correction of thoracic kyphosis and improvement of lumbar lordosis for hypokyphotic patients, corresponding to a positive reciprocal change. These reciprocal changes have also been reported in adult deformity patients in the unfused spinal segments, potentially leading, according to the type of compensation, to favorable or unfavorable radiographic outcomes [30, 31].

Such interactions have also been described in AIS patients but in the coronal plane with a spontaneous correction of the lumbar curve [32]. Considering the sagittal plane, Hilibrand et al. [33] have reported presence of such reciprocal interactions on the upper spine, with a significant correlation between hypokyphosis and loss of cervical lordosis.

Interestingly, the results from this study showed that improvement of lumbar lordosis secondary to thoracic kyphosis correction occurred after the third postoperative month, reflecting, in our opinion, the necessary adaptation of posture to the new imposed alignment.

Notably, no correlation was found between sagittal reciprocal interactions and the amount of coronal correction. This finding leads to questioning the need to obtain maximal coronal plane correction of scoliotic deformities in the setting of AIS. Hybrid constructs with sub-laminar bands may not offer the same amount of coronal correction than all-screw constructs, however, this difference will not lead to inferior sagittal reciprocal changes. This consideration supports other reports questioning the necessity to obtain a perfect coronal correction to the detriment of sagittal alignment [10, 19].

The long-term impacts of the radiographic results noted in this study are difficult to evaluate. While the presence of a thoracic hyperkyphosis has been correlated with a decrease of clinical outcomes (SRS-score) [34], no modifications of clinical outcomes has been reported at 2-year follow-up in AIS patients with a postoperative hypokyphosis [4]. However, the lack of differences at short-term follow-up does not indicate the long-term prognosis for these patients. Further studies will be necessary in order to evaluate the clinical impact of sagittal reciprocal changes, including aspects of cervical spinal alignment and pelvic parameters (spino-pelvic chain of correlation).

Conclusion

Despite limitations due to the retrospective design of this study, and a limited follow-up, sagittal reciprocal changes between improvement of thoracic kyphosis and improvement of lumbar lordosis was determined. These results underline the necessity to restore optimal thoracic kyphosis, whatever the surgical strategy used, in order to offer adolescents suffering from AIS optimal sagittal alignment as they mature into adulthood.