Introduction

Surgical correction of marked idiopathic scoliosis is deemed beneficial for the patient’s life satisfaction [1]. The hierarchy of therapeutic goals reflects its historical and still ongoing evolution:

  1. 1.

    Halting of curve progression

  2. 2.

    Solid bony fusion

  3. 3.

    Safe and solid instrumentation

  4. 4.

    Coronal plane correction and balance

  5. 5.

    Sagittal plane correction and balance

  6. 6.

    Surveillance and preservation of spinal cord and nerve root function

  7. 7.

    Better understanding the 3D biomechanics of the normal and scoliotic spine

  8. 8.

    3D spine and trunk correction

  9. 9.

    Preservation of function

  10. 10.

    Non-fusion techniques

The final common pathway of the first 100 years of operative treatment and of 50 years of instrumented correction of scoliosis is long bony fusion of a once-mobile polysegmental organ. However, the alternative, to leave a scoliotic spine to its natural fate, is cosmetically unacceptable (rib hump, waist asymmetries, shoulder imbalance, etc.) for many patients (50–80° curves) or leads to significant cardiovascular and pulmonary impairment in case of severe curves >80–90°. Moreover, severe scoliotic spines will also stiffen over time. True 3D correction, spinal and shoulder balance and an as-short fusion as possible are the cornerstones of modern scoliosis treatment. The last 100 years of development were rocky, and the final goal—correction with full preservation of function—is still under construction. This article highlights the developmental milestones of posterior corrective surgery for idiopathic scoliosis. A concise summary for anterior approaches is given in the article of Ilka Helenius, Anterior surgery for adolescent idiopathic scoliosis.

The first steps—from Hibbs to Harrington

The natural evil evolution of neuromuscular and post-tuberculosis deformities, which could not be controlled by external means or myotomies, were the starting point for invasive fusion strategies, which later were also applied to marked idiopathic scoliosis and still represent the fundament of any definitive corrective scoliosis surgery. Russell Hibbs (1911) was the first to publish on a series of patients who underwent long, uninstrumented in situ fusions and a long-lasting immobilization in a cast [2]. Above all, the goal was to halt curve progression. Limited correction resulted from the prone position during surgery. The spine was exposed from posterior and fusion was induced by crushing the spinous processes, facetectomies, decortication of the laminae and finally by local bone harvesting and placement along the exposed spine. However, despite the long immobilisation time, the lack of true correction, as well as high pseudarthrosis and infection rates rendered this pioneering work an unpredictable endeavour. Once exposed to gravity without instrumental support, the fusion mass often continuously re-bended and secondary curve progression occurred. Functionally and cosmetically the results were not satisfying. A combination of uninstrumented fusion and a so-called Risser cast improved the correction and lowered the pseudarthrosis rate, but was still not satisfying [3]. Probably the two world wars, which captured much of the intellectual and industrial power at that time, caused another 50 years of standstill in the development of scoliosis surgery. As the world war veterans with their infected pseudarthrosis and malunited fractures urged Gavril Ilizarov to develop ring fixators and callotasis, the poliomyelitis epidemics in the United States (worst in Houston, Texas in the early 1950s) left countless unsuccessfully treated neuromuscular scoliosis patients. Paul Harrington, who moved to Texas after the war, is acknowledged to be the first to use implants for scoliosis correction and to support fusion. First, concave-distraction over a stainless steel rod with a ratchet and a collar end attached to the spine at the curve top and bottom with hooks alone (nonsegmental instrumentation), and later a convex-compression rod in addition, were the cornerstones of his method. Originally the intention was mere instrumentation without fusion, but the complication rate was significant. However, pseudarthrosis, implant corrosion and breakage, as well as hook dislodgements, still led to frequent revisions, also with concomitant fusion since the constructs were not rigid enough. The limited stability required long-lasting postoperative immobilisation [4, 5]. The major drawback of unidirectional distraction is flattening in both the coronal and sagittal plane (Fig. 1). The latter leads to the typical flatback deformity, named “flatback syndrome” if it also causes pain [6]. As a further development for severe curves (100–200°), Pierre Stagnara from France obtained preoperative partial reduction by distraction casts or halo traction followed by one- or two-stage Harrington instrumentation (through the cast fusions) [7, 8]. Over two decades, the Harrington rods and technique remained the gold standard for scoliosis treatment. However, modern scoliosis surgery and the increasing surgeon’s and patient’s demands for 3D correction led to the Harrington rod phasing out.

Fig. 1
figure 1

40-year-old woman with a 26-year follow-up after single Harrington distraction rod instrumentation. Uneventful fusion occurred and the uninstrumented lumbar spine remained stable over time. Slight loss of balance to the left in the coronal plane (a). In the lateral view (b) loss of balance in posterior direction, flattening of the instrumented thoracic spine and a significant rib hump are present

Full size image

From stability to 3D biomechanical understanding

The next development was initiated by the treatment of neuromuscular scoliosis: Eduardo Luque, from Mexico City, started to use sublaminar steel wires in combination with L-shaped rods in the 1970s. After partial removal of the ligamentum flavum to gain access to the spinal canal, the prebent wires are manually guided around the front side of the laminae through the space between the lamina and spinal cord. The two ends are passed around the rods and then tightened sequentially. As this mechanism is applied multisegmentally, the spine is gradually translated and reduced to the rod [9, 10]. This cheap, low-profile and effective segmental fixation method is particularly apt for neuromuscular scoliosis since it provides load distribution in osteoporotic bone. Later it was also applied to idiopathic curves since its high primary stability omits the need for postoperative plaster or brace immobilization and the costs are lower than for pedicle screw constructs [11]. The major concern is the risk of neurologic injury. However, in clinical series sublaminar placement of wires in idiopathic scoliosis proved to be safe [12, 13]. The corrosive combination of steel wires with modern titanium, TAN and cobalt chrome rods is not recommended, but modern sublaminar implants like titanium cables and polyester bands (universal clamps, Zimmer Spine, Bordeaux/France) meet the material requirements [14]. Bands provide more contact area than wires or cables and hence bear a lower risk for cut out through the lamina. A versatile and modular hook-and-screw system for translation correction emerged in the 1990s, the Universal Spine System (USS, Synthes, Oberdorf, Switzerland). Instead of accessing the spinal canal with sublaminar anchors, the reduction manoeuvre relies on anchors (hooks or screws) with side openings (side loading) for the reception of the rod, which is provided by a strong instrument—the “persuader”—for posterior pull and medial translation of the vertebral anchors. The next practice altering instrumentation was that of Cotrel-Dubousset (CD, Medtronic, Memphis/USA former Sofamor-Danek) released at the beginning of the 1980s which for the first time aimed at a 3D correction. Later, many similar devices were manufactured (Moss Miami, Isola, Texas Scottish Rite, etc.). They all incorporate a frame construct of two rods and transverse connectors, and multiple point fixations including laminar hooks, pedicle hooks, but also pedicle screws (thoracolumbar or lumbar level) at selected vertebrae. This implant versatility provides the flexibility to create custom-made constructs according to the requirements of the deformity (3D curve pattern, stiffness) [15, 16]. Strategic segmental anchor placement is the first step, followed by placement of the bent, concave rod into the anchor heads coming from their tops (top loading). The rod matches the deformity. Hence, its material properties needs to allow segmental bending and its pliability prevention of anchor pullout during the reduction derotation manoeuvre: held by special forceps, the rod is slowly rotated back from the frontal into the sagittal plane, thereby translating the apex towards the midline. In contrast to the initial belief, though convincing in theory, the 3D effect in the sagittal and transverse plane is very limited: rod rotation does not lead to significant periapical detorsion of the spine as shown with CT scans [17, 18]. Some change in axial plane alignment is usually attributed to the placement of the second, differently bent rod. In addition, rod derotation in case of selective thoracic fusion can lead to increased rotation of the non-instrumented lumbar spine and to loss of coronal balance (Fig. 2).

Fig. 2
figure 2

a 13-year-old girl with fast progressive adolescent idiopathic scoliosis. 95° main right thoracic and compensatory 60° left lumbar curve with 3 cm loss of trunk balance to the right and shoulder imbalance. Clinically 30° rib hump and 20° lumbar prominence. b The lateral view displays a hypokyphotic thoracic profile and the significant rib hump. c To prevent primary long fusion to L4, it was decided to selectively instrument and fuse the thoracic spine from T3 to L1 with a hook-wire-screw hybrid construct. In view of the dysplastic concave mid-thoracic pedicles sublaminar classic Luque wires were used in combination with periapical convex pedicle screws. The right shoulder was pulled down with a pedicle hook-transverse process hook-claw construct and the left shoulder lifted with pedicle hooks and slight distraction. Solid thoracolumbar pedicle screw foundation bilaterally. Two years postoperatively, well balanced trunk, with a 30-30° right thoracic-left lumbar double curve and leveled shoulders. d The lateral view shows a residual, but improved rib hump and a normalized sagittal profile

Full size image

Three column fixation and direct vertebral manipulation

The use of pedicle screws was first described by the French spine surgeon Roy-Camille [19]. Constructs with bilateral pedicle screws on all instrumented scoliotic levels became a benchmark operative strategy in the 1990s. They provide stable, three-column vertebral fixation and safe manipulation of single vertebral bodies [20, 21]. They are believed to provide superior 3D correction in comparison to all previous methods [17]. Screw placement may be demanding, mainly on the concave curve side of significant curves with dysplastic pedicles [22]. The proximity of the spinal cord medially and the aorta anteriorly, particularly at the mid thoracic spine, warrants upmost caution, anatomic knowledge about entry points and pedicle directions, experience and imaging support [23]. The probability of non-optimal screw placement (medial, lateral, superior, inferior breaches or overlength) was found to be about 20 % in postoperative CT scans [24, 25]. However, most do not require revision [26]. Neurological, vascular or visceral compromise is rare [27]. All pedicle screw constructs offer an à la carte application of all available corrective techniques, such as derotation, translation, segmental distraction–compression and in situ bending. Moreover, during the last decennium, recent developments and implants have provided the additional option of vertebral column manipulation (VCM), also called direct vertebral derotation (DVD): the connection of two screws of one single vertebra with a transverse bar (triangulation) or several levels with longitudinal and transverse bars allow for reduction of single vertebral bodies or an apical multilevel block in the transverse plane [28]. Proper screw placement is mandatory for safe derotation and prevention of screw plowing [29, 30]. Though full derotation is only possible in flexible spines, there are fewer indications for thoracoplasties due to the direct transmission of reduction forces to the vertebrae. However, in most cases some degree of asymmetry remains which is attributed to the physiologic lack of full symmetry, intrinsic rib deformities on the convex site of severe curves and intravertebral torsion between the upper and lower endplate [31, 32]. For larger rib prominences thoracoplasty alone or in combination with DVD may therefore yield better results than an attempt of DVD only [33]. Although a severe rib hump mimics hyperkyphosis in the lateral clinical view, the true apical sagittal profile is mostly hypokyphotic or even truly lordotic. Powerful rotation of those segments back into the anatomic sagittal plane may lead to its flattening. This point of criticism towards vertebral column manipulation does not hold true in recent clinical series [34] but may become relevant if derotation is combined with a “push down” manoeuvre on the convex side which itself may accentuate the lordosing effect [35]. In order to restore or at least preserve the sagittal profile, the derotation manoeuvre should be combined with posterior translation to a precontoured rod. The latter should be strong enough to withstand the high forces [36]. Effective derotation leads to perfect overlap of the two rods on a lateral radiograph. Another concern with rigid, all pedicle screw constructs and flattening of the thoracic spine is junctional kyphosis at the upper end of the fusion. Some spine surgeons believe that a hybrid construct consisting of pedicle screws around the apex for a forceful 3D restoration and a pedicle hook-transverse process hook-claw at the upper end provide optimal correction and a smooth transition from the uppermost instrumented to the first cranial uninstrumented segments.

Additional improvements over time

Not only have implants become more powerful, safe and reliable, and instrumentation more comfortable for the surgeons, but different adjunct modalities have boosted the striving for excellence in scoliosis surgery: one- or two-stage anterior releases are almost completely substituted by a case-based selection of preoperative halo traction, intraoperative skull-femoral traction (slow stretch out and staged correction, easier determination of fusion level), posterior release techniques (ribs, supra-, interspinous ligaments, facets, ligamentum flavum), osteotomies (Smith-Peterson SPO, pedicle subtraction PSO) or even vertebral column resections (VCR) for severe stiff deformities [37–40]. As all those supportive tools and strategies are burdened with some hazards, they should be applied with care [41–43]. The risk of a new neurological deficit (nerve root, cauda equine and spinal cord) after surgery for idiopathic scoliosis during growth is reported to be around 0.7 % according to the most recent report of the Scoliosis Research Society Morbidity and Mortality database [44]. Since the early 1970s until the 1990s the Stagnara wake-up test was the only option for intraoperative detection of surgery-related neurologic deficits. A direct view of the (hopefully moving) patient’s feet at the end of the instrumentation eventually revealed the truth, sometimes after a long period of waiting, depending on the anaesthetist’s capabilities to reach an adequate level of patient wakefulness and responsiveness [7]. In contrast, modern multimodal spinal cord monitoring offers more reliable, timely, faster and safe detection over the whole course of the intervention [45–47]. Use of synthetic antifibrinolytic drugs (tranexamic and epsilon-aminocapronic acid) may reduce the perioperative blood transfusion requirement [48, 49]. Pre- and postoperative 3D low dose whole body radiographic imaging (e.g., sterEOS) enhances the biomechanical understanding and supports preoperative planning [50, 51]. Intraoperative use of navigation to place pedicle screws steepens the learning curve and aids instrumenting in difficult anatomic situations [22, 52–54].

Conclusions

Our biomechanical understanding of curve patterns and their behaviour over time with and without surgery, adequate choice of fusion levels, and the power, safety, stability and reliability of instrumentation have tremendously improved over the last 100 years, although the basic principle—multisegmental bony fusion—persisted [55]. The modern patient’s and surgeons’ demands on optimal 3D scoliosis and trunk correction is reflected by high implant density, extensive use of pedicle screws, and shorter hospital stays, but with rising costs [56]. It remains to be proved if the overall higher expenditure is equally and justifiably transmitted into higher satisfaction and quality-of-life improvement [57–59]. In view of the many function-preserving and -restoring innovations in modern orthopaedic surgery over the last decades, above all in total joint replacement, scoliosis surgery is still stigmatized with a rather archaic way of sacrificing function in young and otherwise healthy individuals. The path of instrumented fusion seems to end at a high but still not fully satisfying level. It is to be hoped and to be stipulated with sound scepticism that new function-preserving technologies—preventive or curative—will emerge soon to take us to the next level of astonishment.