3.10 Idiopathic scoliosis

Terms used in the discussion of scoliosis are defined in Table 3.10.1. Curve patterns are summarized in Table 3.10.2.

About 80% of new cases of scoliosis have adolescent idiopathic scoliosis (AIS). Infantile and juvenile idiopathic scolioses comprise 0.5–4% and 7–10% respectively of all idiopathic scoliosis. In the 1950s the incidence of infantile idiopathic scoliosis in Europe was high compared with that in North America; current figures are comparable. Up to 10% of adolescents have minor scolioses of less than 10 degrees. The incidence of these small angles is equivalent in boys and girls. A Cobb angle of 10 degrees is generally accepted as the criterion above which a scoliosis exists. The larger the curve, the greater is the female-to-male ratio (Table 3.10.3).

The aetiology of idiopathic scoliosis remains elusive. Most work has focused on AIS, and it is clear that a complex and probably multifactorial process is involved. Naturally occurring scoliosis in vertebrates is seen almost exclusively in humans, although a number of animal models for the condition exist. There are many observed differences between convexity and concavity, but it is difficult to distinguish those factors that may be causing the curve from those that may result from it. Current thinking is that there is a defect of central processing or control which plays on a growing spine whose susceptibility to deformation varies from one individual to another. Girls may be more vulnerable to this process because they have a shorter and more rapid adolescent growth of spine than boys. Genetic studies have now identified areas of interest on various chromosomes that predispose to scoliosis and may prove to be the most important aetiological factor.

A spectrum of curve patterns has been classified, either to aid decisions of surgical management or (in one case) to define bracing decisions. There has been increasing interest in the sagittal as opposed to the coronal profile of the spine. Classification by surface topography is still poorly defined, although this is likely to be important in assessing outcome. The Cobb angle (the coronal profile of the spine) remains pre-eminent, although this angle is only one component of a complex change following treatment or observation.

A reliable method of classification is anatomical, describing the direction of the convexity of the curve and the level of the apical vertebra and is widely used whereas the King–Moe classification (Figure 3.10.1), which divides thoracic curves into five types, is now less used. Limitations of the King–Moe classification include poor reliability and the fact that only the coronal plane is considered and scoliosis is a three-dimensional deformity.

Lenke and coworkers have also developed a classification designed to help surgical decisions in response to dissatisfaction with the King–Moe system. This classification considers structural versus compensatory curves and introduces for the first time the sagittal plane and as important modifier. This system is now widely used but its complexity and poor reproducability often precludes day to day use in the clinical setting for all but the most common curve patterns.

There have been a number of retrospective reviews of untreated AIS, although most of these series are small. The detailed review by Weinstein in 1999 is an excellent source for these. It is unlikely that further reviews of the natural history of this condition will be possible, as treatment is widely available. It might be self-evident that the more severe the curve at skeletal maturity, the more problems arise, but this accurate generalization is often wrong in individual cases. At one time it was claimed that curves did not progress in adults, but this is clearly not the case for curves greater than 30 degrees. Even so, progression is unpredictable with one adult progressing by 8 degrees and another by 80 degrees. An annual increment of 1 degree can be expected for curves greater than 40 degrees at skeletal maturity. Double curves, lumbar curves (without thoracic cage support), and curves in association with imbalance tend to progress more rapidly.

This is not usually an issue in AIS, although an untreated early-onset curve may be bad enough to affect vital capacity. Curves of less than 70 degrees have no measurable effect on vital capacity. Significant reductions in vital capacity occur in curves greater than 100 degrees and, if severe enough, may eventually cause cor pulmonale and shortened life expectancy.

This is a crucial consequence of AIS. Thoracic curves are associated with a rib hump dependent on the degree of rotation of the apical vertebra. The rib hump ranges from the trivial to the very conspicuous and often it is an oblique shoulder line or pelvis that causes a problem. The distress caused to the patient is very variable, and some will find the risks of surgery justified for very small humps. Others prefer to tolerate a deformity which can be concealed with appropriate clothing. The level of distress should not be underestimated (Figure 3.10.2). It is often the justification for prolonged non-operative treatment and much major surgery in this area. We studied the assessment of cosmesis and developed a scoring system which relates to assessment by independent reviewer. Patient perception may bear only limited relation to the severity of the curve. The Scoliosis Research Society Score attempts to quantify the cosmetic impact of the deformity allowing studies to be performed although little has been published in this important area.

Lumbar curves are usually less conspicuous than thoracic curves, although they can have an adverse effect on the waistline. Double curves tend to be inconspicuous, but are more prone to progression. Truncal imbalance may cause distress. Curve progression may be associated with loss of height in adults.

Back pain is common, and there is limited evidence that scoliosis is a major cause of back pain. Thoracic curves in general do not cause significant pain in adolescents, although some do experience apical pain. There is deterioration in the thoracic curve in some adults, especially during pregnancy, and they may experience significant pain.

Thoracolumbar and lumbar curves are more likely to cause back pain. Relief is often provided by lying and precipitated by standing or walking as the day goes on. This is one of the most common reasons for scoliosis surgery in adults, and occasionally in adolescents. Pain is more likely when translation (lateral subluxation) occurs, where one vertebra moves laterally on the one below. This may occur in both primary and secondary curves.

There is little evidence that the management of pregnancy should be altered in any way because of treated or untreated scoliosis. Epidurals may not be possible if the lumbar spine has been fused posteriorly. Pregnancy may be associated with additional curve progression, but studies have shown variable results.

Spinal cord compression is a not a problem in untreated AIS. Secondary degenerative changes in the lumbar spine may cause root compression.

Understanding the potential for progression is the key to management. The probability for progression depends on the following:

A growth chart is the easiest way to see growth potential. Menarche is an important landmark for spinal growth, which occurs most rapidly in the first year premenarche. The Risser sign is the best radiological indicator of growth potential. If the iliac apophysis has not appeared, or has only just appeared, curves are more likely to progress (Risser 0 or 1; see Table 3.10.1).

The larger the curve at presentation, the more inevitable is progression. Sixty-eight per cent of curves in the range 20–29 degrees will progress in children with Risser signs of 0 or 1. Curves of over 30 degrees almost invariably progress if there is significant growth potential. Curves of over 50 degrees at skeletal maturity will progress during adult life with an annual increment of at least 1 degree. Some curves in the range 30–50 degrees will progress in adults.

Double curves are more likely to progress than single curves as are lumbar curves and thoracolumbar curves without thoracic cage support.

Management options are observation, bracing, and surgery. Exercises and physiotherapy regimens are widely used but evidence for efficacy remains anecdotal. Plaster beds and localizer casts are of historical note.

Risks and benefits have to be carefully considered against the severity and potential for deterioration for each individual. The clinician has to be clear on the expectations of patients and their families, and on their capacity to comply with treatment. Careful counselling is essential, and this may require more than one visit to the clinic before treatment is initiated.

A full history includes birth history, age of menarche, childhood illnesses, and family history. Symptoms of the curve, including pain and deformity, and how it originally was discovered should be noted. Pain, if present, needs a full evaluation. Commonly this may be fatigue or apical pain. The pain may be remote from the curve and, especially in adults, irrelevant to it. Pain severity is assessed by its effects on daily life and analgesic intake (if any). Sometimes there may be neurological symptoms of significance such as paraesthesia or numbness. Where relevant, a history of interventions is essential.

Height, weight, sitting height, and span should be recorded on a growth chart. Look at the facies, eyes, and palate. The shape and relative height of the shoulders should be noted, as well as the position of the scapulae. A plumb line from the vertebra prominens allows measurement of any truncal imbalance. Lordosis and kyphosis should be reviewed. Examination should include an assessment of the deformity and its flexibility. The forward-bending test allows assessment of the site and severity of any rib hump (see Chapter 3.9; Figure 3.10.1). Leg length, limb asymmetry, and foot deformities should be noted, and a full inspection of the skin for hairy patches, skin tethering or dimples, and café au lait spots should be performed. Look for evidence of joint laxity. A careful neurological examination should be carried out, paying particular attention to lower-limb reflexes and abdominal reflexes. Cord anomalies may occur without neurological signs or with very subtle signs. Assess the stage of puberty. Pubic hair appears in boys before their growth spurt. Axial hair appears when there is little or no spinal growth in both sexes.

All efforts should be made to minimize exposure of the growing child to radiation. There is good evidence that excessive radiography in children increases cancer risk in later life. A request for imaging should be justified by a defined therapeutic gain in every case. The spine is assessed by standardized erect anteroposterior and lateral radiographs. Long cassettes should be used if available. Both views should ideally show the head to the proximal femur. Various techniques, including rare-earth screening, have been developed to minimize radiation requirements. Some centres have used coned lateral views of the lumbosacral junction to identify a spondylolysis, but the number of exposures needed to pick up each case is not justified. Supine anteroposterior bending or traction radiographs are needed to define the flexibility of the curve if surgery is impending, but not as part of curve monitoring. In some cases, such as small infants, supine radiographs are best. Comparable views are needed to detect progress.

Follow-up anteroposterior radiographs are widely used, unless the unit is equipped with a device to measure surface topography such as ISIS (see Chapter 3.9; Figure 3.10.2) or Quantec. There is some evidence, at least for ISIS, that curve progression can be identified earlier than radiography, as the ‘rib hump’ leads the development of the scoliosis. This method can significantly reduce radiation exposure.

If the curve is painful, of early onset (under 10 years), morphologically unusual or there is any evidence of neurological abnormality (e.g. hairy patch, dimple, absent abdominal reflex, or lower-limb or foot asymmetry), the spinal cord should be investigated. Many centres routinely perform magnetic resonance imaging (MRI) of the entire neural axis in the presence of any spinal deformity to rule out spinal dysraphism which is present in 3% of so-called ‘idiopathic curves’. Before any surgery, MRI scan is mandatory. These abnormalities include Arnold–Chiari malformations, syrinx, tethered cord, and diastatomyelia. Computed tomography is rarely indicated in adolescent idiopathic scoliosis.

Patients and their parents will need a careful explanation of the condition. In many cultures the majority of people have never have heard of the condition. Parents require reassurance that they have not failed in their duty of care to their child. Curves can appear remarkably quickly, so that large curves can present in families familiar with the condition and with a sibling already affected. It is worth checking siblings with a forward-bending test if they are available in the clinic. Parents may rarely see their teenage children’s backs. Patients’ organizations for scoliotics exist in many countries, and some families obtain considerable support and comfort from them (examples are http://www.sauk.org.uk; http://www.srs.org/). Many surgeons have systems in which their experienced patients counsel new ones, particularly those considering surgical treatment. This system needs to be monitored with care to prevent inaccurate or insensitive advice being given.

Treatment options are observation, bracing, or surgery. The surgeon needs to consider the natural history of the curve, and the aspirations and expectations of the patient and his or her family. Treatment places considerable demands on all family members. Some interventions are difficult to sustain in a dysfunctional individual or family. Generally the child or adolescent will decide which treatment option is chosen. If this different from the expectations of the rest of the family, long-term resentments can arise. The family practitioner needs to be kept closely informed of the clinician’s advice.

Generally this option is safe. A small curve with a low probability of progression is easy to monitor. The younger child with a small curve may need to be viewed every 4–6 months, depending on growth rate. The passage of time allows the family to absorb the implications of other treatment options.

The timing of intervention, particularly bracing, may be difficult. In retrospect it is easy to see when treatment has been instituted too late.

Many braces have been designed for the treatment of scoliosis. Controversy exists as to what a brace is actually doing (see Chapter 3.14).

Much controversy surrounds brace treatment. There is one controlled trial comparing the patients of surgeons with a high threshold to bracing with those with a low threshold, which shows small but significant advantages in the treatment group. More recently, a meta-analysis of bracing and natural history papers came out in favor of bracing.

Brace compliance may be poor. A review suggests that the full conventional regimen of wearing the brace for 23h per day is best. This is difficult for most teenagers, although younger patients usually comply better. Most regimens allow the brace to be removed for sports. Brace wearing from a very young age to adolescence is not usually needed. This regimen may be bitterly resented by the child, and can become a cause of emotional warfare between the child and his or her parents. Blame is then assigned if the regimen fails and surgery is required. Bracing compliance varies with culture and is much more difficult in hot countries. Some studies of brace compliance have cast doubt on how much the brace is actually worn (a conspiracy between children and their parents), while other studies show good accord between compliance and diary. Much depends on the quality of the brace maker and the support services for the families. Braces are expensive, and may need to be changed frequently in a rapidly growing child.

The objective in most cases is to fuse the spine in the best possible position. In the past this was achieved using either plaster beds or localizer casts, and required prolonged immobilization (up to a year). These operations were dogged by high pseudarthrosis rates, and were confined to a few dedicated surgeons and patients. The first successful spinal implant was the Harrington rod, which was developed in Texas in the late 1940s and early 1950s during a horrendous poliomyelitis epidemic. Many other implant systems have followed.

Modern surgical treatment allows early mobilization and often avoids the need for postoperative jacketing and bracing. The pseudarthrosis rate is low; indeed, it is now almost unheard of in AIS.

This is major surgery with a small but significant risk of spinal cord damage. The Stagnara wake-up test was an important advance, but has been superseded by the clonus test and, more generally, by various forms of spinal cord monitoring. The wake-up test involves voluntary movement of the feet and hands during the surgery. The clonus test exploits the normal phenomenon that clonus can be elicited during the normal waking process. If it is absent, it suggests an interruption of the normal reflex pattern.

Spinal cord monitoring can be sensory, motor, or both. Sensory monitoring is easier and is widely used. Potentials evoked by peripheral nerve or distal spinal cord stimulation are detected by an epidural or cortical electrode. Motor evoked potentials are detected peripherally following magnetic transcranial stimulation or direct stimulation of the spinal cord. Anaesthetic techniques may have to be adjusted to make this possible.

None of these methods are foolproof. The incidence of complication is difficult to determine, although in the 1970s the Scoliosis Research Society collected data suggesting that the risk of paraplegia was 1%. This figure may be too high with current methods, although it is widely used by scoliosis surgeons when counselling their patients for surgery. Early infection is fortunately rare, although a high incidence of unexpected infection has been found with normally non-pathogenic organisms such as Propionibacterium acnes when rods are removed late for pain. Patients should therefore be warned of this late complication as part of the consent process.

Up to the 1950s, none of the various attempts that had been made to develop implants for the stabilization of the spine had been successful. The majority of spinal fusions were obtained by prolonged immobilization in plaster beds or with the use of plaster localizer jackets. Spinal fusion was first described in 1911 by Hibbs and Albee independently in the United States. Fusion, usually using autograft, is an essential component of spinal surgery. If fusion is not used, or fails, the implants will inevitably fail eventually. In the 1940s Allen, in the United Kingdom, developed a bottle-screw device for the instrumentation of scoliosis.

This device was developed in Texas in Harrington’s garage. It was the surgeon’s response to the overwhelming poliomyelitis epidemics of the time, which resulted in aggressive and progressive spinal deformities in small children. It consists of a distraction hook, where the upper hook can be moved away from the lower hook on a ratchet. The hooks were initially placed under the laminas. It was immediately successful, although Harrington did not publish until he was sure of the results. Although some modifications were made to the system, it has remained a gold standard for subsequent spinal implants (Figure 3.10.3).

Unfortunately, although impressive results were obtained with the Harrington rod, it was not ideal for treating the severe postpolio deformities. Again, it was this condition which, in the late 1960s, led Dwyer, in Australia, to develop his screw and cable system for use on the front of the spine. This was one of the first orthopaedic implants to be made of titanium. This approach revolutionized spinal surgery, although it was many years before anterior spinal surgery became widely adopted. Credit for this goes to Hodgson, a British expatriate surgeon in Hong Kong, who combined Dwyer instrumentation with the Harrington rod. The anterior approach proved to be a most effective method in the management of deforming spinal tuberculosis. Figure 3.10.4 shows the use of a Webb–Morley implant which evolved from the Dwyer system.

Once again it was poliomyelitis which precipitated Eduardo Luque of Mexico City to develop a technique based on an earlier Portuguese idea of sublaminar wires. The wires were passed under the lamina of each segment and over a pair of stainless steel rods, which were contoured to the desired shape of the spine. Each rod was bent at right angles at one end (L-rod). This proved to be a productive, economical, and powerful technique for the management of neuromuscular scoliosis. It has been used for all types of scoliosis.

It is a powerful method, but has gained a reputation for a high risk of spinal cord damage. This relates not only to the sublaminar wires, but also to the possibilities offered for overvigorous correction. The technique was modified by adapting it to Harrington rods. This allows the spine to be pulled onto a bent Harrington rod to obtain correction without too much distraction—the Harri–Luque technique. This depends on a modification of the Harrington rod where the rod locks into a square rather than a round socket in the lower hook. This has proved effective in the management of AIS, although it has been superseded by later methods.

Luque rods were also modified to allow pelvic fixation—the Galveston technique. This is still used in the management of neuromuscular curves. Another modification is the joining of the two rods into a rectangle. A variety of versions of this are available, including the Hartshill rectangle.

Cotrel and Dubousset (CD) in France developed a whole new concept of correction of scoliosis. Cotrel tried various modifications of the Harrington system including a link between the distraction rod on the concavity and a compression rod on the convexity. This produced a better correction of the curve. He came to realize that this had been obtained at the expense of increasing the rotation of the curve. Although others had appreciated this before, scoliosis surgeons began to realize the importance of rotation of the spine. Dickson in Leeds revitalized the concept of lordosis as an important component of the thoracic deformity of idiopathic scoliosis. He promoted the Harri–Luque concept on the contoured rod.

The CD system of instrumentation was the outcome of Cotrel’s ideas. Here the spine is connected to a contoured knurled rod by a series of hooks and screws in its neutral position. The rod on the concavity is rotated away from the curve, the normal sagittal profile of the spine is recovered, and the implant is locked into position. Distraction and compression can be used subsequently. This system is widely used.

CD has been superseded by the so-called third-generation systems, including Texas Scottish-Rite Hospital (TSRH), the Universal Spine System (USS), Colorado, Legacy, and others. These rely on two rods and increasingly an emphasis is placed on apical translation and derotation with more advance rod connection mechanisms. The concave rod can be contoured and the spine is then pulled towards the rod. There is some evidence (notably for the USS system) that this method improves derotation compared with the CD system (Figures 3.10.5 and 3.10.6) although the benefit of more recent innovations and the use of pedicle screws in isolation are under evaluation.

The revolution in spinal implants has had some important consequences:

This condition is distinct from AIS. It is much less common, and is more frequent in boys than girls. The direction of the curve is more frequently to the left. Many of the curves improve or even resolve with observation. Curves with marked rotation are much less likely to resolve, as was demonstrated by Mehta. She found that a measurement reflecting the difference of angulation of the ribs at the apex of the curve was higher in those whose scoliosis progressed (Figure 3.10.7).

There is a diversity of views amongst scoliosis surgeons on decision management in the large ‘grey area’ between patients presenting with small curves and little growth potential (where most parties agree that no intervention is needed) and large curves with much growth potential (where surgical intervention is beneficial). Honest discussion with the family is essential, with a careful weighing of risk and benefit of natural history versus intervention. Surgical treatment and bracing both carry serious risks. Future developments require large trials of operative and non-operative management, the development of outcome measures relevant to appearance and cosmesis, and the longer-term issues of pain and disability. Technology to assess surface topography needs further development to enable the effects of time and treatment to be presented to patients and their families.