Over the past few years, the surgical treatment of curves in adolescent idiopathic scoliosis (AIS) has evolved significantly. With the advent of modern instrumentation techniques, including robust pedicle screws and direct vertebral rotation systems, a paradigm shift has occurred toward a posterior-only approach [1]. A multitude of options can be used individually or in combination, including posterior release, Ponte osteotomy (PO), vertebral column resections (VCR), perioperative skeletal traction, and staged procedures [2].

Preoperative axial traction has been consistently used to facilitate correction in AIS [3]. Intraoperative traction (IOT) was first mentioned by Cotrel et al. [2] in 1988. Subsequently, IOT has been widely utilized as an adjunct to posterior release for the correction of AIS curves, particularly with a flexibility index (FI) <0.35 [5]. It reduces the vertebral translation and thereby the magnitude of the curvature, facilitating easier exposure on the convex side, helps in the placement of screws by correction of vertebral body rotation to some extent, and overall helps in avoiding morbidities associated with anterior surgery (staged or single procedure) [4,6]. However, it does not significantly contribute to a sagittal alignment correction.

Scoliosis is a three-dimensional deformity with the coronal deformity being only one aspect of the broader picture. With anterior column overgrowth, a significant component of hypokyphosis is documented in most patients with a curve of >65°. A true correction of the deformity entails not only a reduction in the coronal deviation but also an overall correction in the coronal axial and sagittal dimensions [7]. PO results in an overall increase in spinal flexibility. The systematic closure of the osteotomy gaps allows the surgeon to establish the desired thoracic kyphosis (TK) [1].

On a comprehensive review of the literature, the choice of technique adopted for curves between 65° and 100° remains controversial with no major consensus favoring one technique over the other. Furthermore, data on the combined use of IOT and PO for correcting high-magnitude curves are very limited. Thus, this study aimed to address this gap by treating a series of patients with AIS using a combination of PO and IOT. The hypothesis tested was that multisegmental PO combined with IOT provides better correction than IOT alone and that these techniques can be safely combined without the risk of neurological injury.

After the Institutional Ethical Clearance Committee at Bombay Hospital and Medical Research Centre, Mumbai (approval no., BH-EC-0138), a retrospective comparative study was performed, comprising 24 patients with AIS with curve >65° and FI <0.5, in whom deformity correction was performed by a single experienced spine surgeon between January 2014 and December 2021. The posterior-only approach with the segmental pedicle screw-rod fixation system was utilized in all patients. Informed consent for inclusion in the study and subsequent publication of research data, including photographs, was obtained from all patients and their families.

IOT, using halo-femoral skeletal traction, was employed for all patients. In the first 10 patients, IOT was used to facilitate correction without additional PO (T group). In the subsequent 14 patients, traction was supplemented with multilevel PO (TP group). This shift occurred primarily because the authors started to use this technique for rigid curves in the later years of their practice. The standard approach for osteotomy involved starting at the apex of the curve and then proceeding to one level above and one level below it. Additional osteotomies were performed in both directions, spanning across Cobb to Cobb as needed, based on the severity of the curve and the extent of correction required. This depended on the clinical judgment of the senior surgeon who assessed the flexibility of the residual curve after each osteotomy. Posterior release, facetectomy, and pedicle screw-rod system were utilized in all patients to aid correction and fusion. Triplanar deformity correction was achieved by in situ contouring, derotation, and selective posterior distraction/compression maneuvers. The allograft was applied over the decorticated posterior elements before closure. The entire surgery was performed under intraoperative neurophysiological monitoring (IONM) in the form of somatosensory and transcranial motor evoked potentials (SSEP and MEP, respectively) by a trained neurophysiologist.

Preoperative clinicoradiographic evaluation of all patients included the following: (1) Clinical images (erect-from back and sidewards, bending forward without knee flexion to visualize the rib hump [Adam’s forward bending]). (2) Radiographs: Anteroposterior (AP) and lateral views of the whole spine scanogram (standing erect), traction, and convex side bending AP radiographs in the prone position.

Before positioning, two well-padded bolsters were placed horizontally on the table. These resilient custom-made bolsters were designed according to the principles of the four-poster system. The proximal bolster aligned with the patient’s sternal prominence, and the distal bolster was positioned at the level of the anterior superior iliac spine to support the pelvis and avoid direct pressure on the abdomen, thereby reducing the risk of venous compression and rise in intra-abdominal pressure. Subsequently, Gardner Wells tongs (GWT) were applied to the head to ensure symmetrical placement of the calipers for maintaining the head in the neutral position and a uniform traction on the cervical spine. Smooth Steinmann pins (4 mm in diameter) were placed transversely in the bilateral distal femoral metaphysis, parallel to the knee joint, taking utmost care not to damage any neurovascular structures.

The patient was then positioned carefully in the prone position on a radiolucent operating table. Traction was applied using weights not exceeding 50% of the body weight, half of which was suspended from both lower limbs and one-third from the skull. The GWT were adjusted to maintain the head in a slightly flexed position, with the traction cable suspended from a pulley system to ensure continuous and controlled traction. Adequate lubrication of the pulley was maintained to minimize friction.

Distally, the traction cables were connected to a stirrup and suspended from an L-bar system attached directly to the traction apparatus, providing a vertical suspension of the weights. Although we used the L-bar system, the traction cables can also be suspended using a pulley system instead. The L-bar system allows for direct force application and stable, fixed traction throughout the surgery. It also has a more compact setup than the overhead pulley system, which requires a frame for frictionless suspension—this can be a limitation in smaller operating rooms. However, the angle and balance of traction with the L-bar system must be monitored, as it cannot be adjusted intraoperatively unlike the pulley system.

The arms of the patient were on the sides with proper padding to avoid brachial plexus injury. To facilitate access to the intravenous cannulas, the anesthetist’s equipment was placed at the distal end of the operating table (Fig. 1A–C).

Postoperative evaluations included erect standing clinical images from the back and erect whole spine scanogram–AP and lateral views. IONM alerts were recorded in both groups systematically, before and after positioning, after application of traction, during pedicle screw insertion, after performing osteotomies and after manipulation. Immediate rectification was performed in the case of a transient IONM signal-loss alert at every incidence by following the institutionally adopted care pathway. The postoperative rehabilitation protocol remained uniform for all patients. They were made to sit on postoperative day 1 and stand on the next day. Further mobilization was gradually started from postoperative day 2 as per their comfort.

Patient-related data including the curve pattern and demographics were recorded. Radiographic data including coronal Cobb’s angle, FI, apical vertebral rotation (AVR; Nash and Moe grading), apical vertebral translation (AVT), TK, lumbar lordosis (LL), and sagittal vertical axis (SVA) length were calculated preoperatively and postoperatively for all patients. Postoperatively, correction indices (CI) were calculated by analyzing the preoperative and postoperative films (Fig. 2A–2D, 3A–3D) [5,8]. Statistical analysis was conducted using the major thoracic or thoracolumbar curve measurements. Comparison studies were conducted using the two-sample t-test and Mann-Whitney U test for nonparametric data. A p-value of <0.05 was considered significant.

The mean age group in the T group was 19.4 (13–36) years and that in the TP group was 16.8 (10–34) years. In the T group, two patients had Lenke type 1, six had Lenke type 2, and two had Lenke type 5 curves. In the TP group, two patients had Lenke type 1, five had Lenke type 2, one had Lenke type 3, four had Lenke type 5, and two had Lenke type 6 curves. Both groups were matched in terms of demographic data (Table 1). The mean preoperative Cobb angle on the AP radiograph in the T and TP groups were 89.35°±6.05° and 92.32°±9.28°, respectively (p=0.59). The mean FI on bending radiographs were 0.31±0.016 in the T group and 0.36±0.03 in the TP group (p=0.41). The mean FI on traction films were 0.40±0.11 in the T group and 0.41±0.08 in the TP group (p=0.38). In both groups, traction rendered the curve more flexible than side bending. In the TP group, PO was performed at a mean of 5.8 (5–7) levels. The mean postoperative Cobb angle in the T and TP groups were 40.25°±5.95° and 19.1°±3.20°, respectively (p=0.041). The mean CI achieved were 55.29% and 70.64% in the T and TP groups, respectively (p=0.03). The AVR improved from mean grade 3.2 (2–4) to grade 2.6 (1–3) in the T group and from a mean grade of 3.6 (2–4) to a mean grade of 1.8 (1–3) in the TP group. In terms of the mode, AVR improved from grade 3 to grade 1 in the TP group and from grade 3 to grade 2 in the T group (p=0.041). The AVT improved from a mean of 2.29±0.02 cm preoperatively to 0.98±0.01 cm postoperatively in the T group and mean of 3.22±0.041 cm preoperatively to 0.51±0.02 cm postoperatively in the TP group (p=0.046). The mean TK was 9.82°±1.67° and 9.21°±1.79° in T and TP groups preoperatively, respectively. This was corrected to means of 13.84°±2.10° and 21.02°±1.68° in the T and TP groups postoperatively, respectively (p=0.044). Preoperatively, the mean LL was 32°±5.24° and 36°±3.84° in the T and TP groups, respectively, which changed to a mean of 38°±4.21° and 43.4°±7.98° postoperatively, respectively (p=0.042). Postoperatively, the sagittal balance was maintained in all patients with SVA <4 cm in both groups. At the last follow-up (minimum 2 years), no significant difference was noted in all radiological parameters in both groups (p>0.05) (Table 2).

The mean intraoperative blood loss in the T group was 690.72±10.28 mL and that in the TP group was 880.44±9.80 mL (p=0.042). In all patients, surgery was performed under IONM. The incidences of IONM alerts were similar between the groups. No signal-loss alerts were noted at the time of application or after traction removal. In each group, one case of transient SSEP signal loss occurred intraoperatively, which recovered subsequently after following the standard care pathway adopted by our team. No patients developed a neurological deficit postoperatively. All alerts were either due to hypoperfusion or erratic pedicle screw trajectory and were eluded by achieving hemostasis, maintaining adequate blood pressure, changing the screw trajectory, and giving warm saline wash to the operating field.

No patients needed critical care beyond 12 hours. Both groups were mobilized on postoperative day 1 within the limits of comfort. No differences were found in the postoperative protocol in terms of diet, medications, physiotherapy, or rehabilitation between the groups. At the end of 2 years, no incidences of pseudarthrosis or implant-related complications were reported in either group.

The correction of rigid and high-magnitude curves is challenging. Delay in presentation, compact musculoligamentous envelope, posterior tether, complex facet orientation, and alteration in anatomy contribute to rigidity in such patients. Literature reveals many tried and tested options over the years. The popular techniques utilized to tame rigid scoliosis are multilevel PO, VCR, application of perioperative traction, internal distraction, and intervertebral space releases, either in single or staged procedures [9]. The aforementioned methods are usually employed in combination as per the severity of the deformity. This study followed a “posterior-only, single surgery” approach using IOT in all patients with a curve magnitude of >65°.

The safety, virtues, applications, and advantages of intraoperative skeletal traction are well known [5,10]. Using the principles of ligamentotaxis, traction derotates the spine and facilitates correction in all axes (Fig. 4). It also makes exposure easier and minimizes the stress on the implants. PO helps achieve correction of the curve in all three dimensions by increasing the overall flexibility of the spine and helps in establishing the normal TK. This has been consistently documented in the literature [7,11–16].

Nonetheless, a study combining the virtues of the two techniques has not yet been conducted. Moreover, to the best of our knowledge, no studies have commented on the safety profile of this combination. An increase in flexibility by 69% compared with the intact spine after three Ponte osteotomies and axial rotation by 2.8° following a single osteotomy was reported in a cadaveric study by Sangiorgio et al. [13]. We found a significant difference in the CI of both groups postoperatively. A mean CI of 55.29% was achieved in the T group, whereas it was 70.64% in the TP group (p=0.03).

PO aided in three-dimensional curve correction as reflected by the higher mean TK and LL in the TP group than in the T group (p=0.044 and 0.042, respectively) postoperatively. However, at the end of 1 and 2 years postoperative follow-ups, although the LL was still higher in the TP group, the difference was no longer significant. The higher immediate postoperative TK and LL in the TP group could be attributed to the more aggressive three-dimensional correction facilitated by the complementary action on IOT and PO. The subsequent reduction in the difference in LL in both groups can be explained by multifaceted compensations in the body (such as pelvic retroversion and adjustments in TK) to maintain a physiological sagittal balance. This could also be due to the greater natural adaptability of the lumbar spine than the thoracic spine. The initial difference in TK between the groups was greater than the difference in LL. Subsequent changes in kyphosis and lordosis appeared to be proportional, with the difference in TK remaining greater than that of LL in both groups. Furthermore, the SVA consistently remained <4 cm at each postoperative follow-up, indicating a balanced sagittal alignment in the patients.

IONM enabled the authors to document changes in the somatosensory (SSEP) and MEP at various stages of the surgery. Transient signal-loss was reported intraoperatively in two patients, distributed equally in both groups. This negates the age-old fear of the risk of neurological insult when utilizing a combination of IOT and multilevel PO. The traditional belief that osteotomies in the presence of IOT leads to neurological injury as a result of possible sudden and additional creep and distraction appears ill-founded, particularly in PO. This mishap may happen when IOT is combined with a three-column osteotomy such as a pedicle subtraction osteotomy or VCR where the spinal segments are totally disconnected and a tug and pull may occur on the spinal cord. Because Ponte osteotomies involve the posterior column only, its benefits can be exploited along with the benefits of IOT, thus promoting synergy without compromising the safety profile of the surgery.

Concerns were noted about higher blood loss with the Ponte technique, which directly correlates to the number of levels at which the osteotomy needs to be performed [12,13]. In this study, the mean intraoperative blood loss in the T group was 690.72±10.28 mL and that in the TP group was 880.44±9.80 mL (p=0.042). Blood loss was on the higher side for patients in the TP group and was managed with transfusion(s). However, this did not reflect as increased postoperative morbidity as no significant difference was noted in both groups in terms of stay in critical care postoperatively.

Preoperative radiographic evaluation should include traction AP radiographs for all scoliotic curves. It is the traction radiograph that matches with the postoperative radiographs better than the bending films [5]. Besides correcting the derotation, it gives an idea of the shoulder balance on correction and enables the surgeon to fine tune the surgical technique using PO, compression, and distraction maneuvers to achieve leveled shoulders. Bending films are devoid of this advantage. In this study, evaluation comprised both bending and traction films. In both groups, the mean FI was higher in traction radiographs than in bending radiographs. The difference between traction and bending radiographs can be attributed to inadequate opening of the curve in bending films because the iliac crest restricts further bending movement of the costal margin in severe rigid deformities [5].

To the best of our knowledge, this is the first study where PO is coupled with IOT for the correction of AIS. Supplementing IOT with PO enables successful correction of the deformity comparable with staged surgeries (utilizing both anterior and posterior approaches) (Table 3) [17–20] but with much less morbidity, shorter operative time and hospital stay, semi-rigid implants, and early mobilization of the patients.

One of the major limitations of the present study was the non-inclusion of a third set of patients in whom only PO was performed. In addition, the cohorts were not randomized. The first 10 patients underwent surgery using IOT, and in the subsequent 14 patients, IOT was supplemented with PO at multiple levels. This was because the authors started to use this technique for high-magnitude curves in the later years of their practice.

The management of rigid and high-magnitude curves, particularly in delayed cases, is a challenge. The combination of IOT and PO is deemed to complement one another, and the two can be safely combined without an attributable risk of neurological injury. The cumulative effect, in addition to de-stressing the tissue dissection, hastens the ease of deformity correction maneuvers.