9 Classification of Adolescent Idiopathic Scoliosis for Surgical Intervention
The first systematic classification of scoliosis was described by John R. Cobb in his classic article “Outline for the Study of Scoliosis,” published in 1948. However, descriptions of scoliosis and its treatment may be traced back to Hippocrates and his “De Articulationes” in Corpus Hippocraticum. Understanding of the nature of the deformity by Hippocrates was based purely on the subjective appearance of the unfortunate patient; the treatment recommendation was primarily traction for all types of deformity, with results recognized to be poor. Galen later used Hippocrates’ recommendations for treatment, and introduced the descriptive terms kyphosis, lordosis, and scoliosis. Despite the introduction of this appearance-based classification system, the treatment of scoliosis was largely unchanged until the late nineteenth century. After the discovery of roentgen rays in 1895, the description of scoliosis became more qualitative, as the deformed spine could now be visualized and more definitive treatment could be developed. With a better understanding of the geometry of the underlying spinal deformity, spinal fusion could then be attempted for certain types of scoliosis such as neuromuscular curves caused by polio, with some success.
When Cobb set forth to describe the classification and treatment of scoliosis, he was relying on only 30 years of experience in its surgical treatment, dating back to the first spinal fusion for scoliosis performed in 1914 by Hibbs. Cobb’s descriptions of major and minor curves, structural curves, types of scoliosis, and etiological classification, as well as treatment recommendations based on these classification parameters, continue to influence the classification and treatment of scoliosis to this day. As more surgeons began to treat scoliosis, it became obvious that etiology, such as tuberculosis and neuromuscular disease, had an impact on the patterns of deformity, and also that identifying these patterns helped in predicting the response to the developing treatment modalities. It soon became apparent that sharing observations about deformity patterns and their response to treatment was an important next step in organizing the treatment of scoliosis. The founding of the Scoliosis Research Society in 1966, owing in large part to the ideas and enthusiasm of David B. Levine, MD, was the next important step in providing a platform for rationalizing and discussing treatments for scoliosis treatment and creating a body of literature about these.
King Classification
In 1983, Howard A. King and colleagues published a review of the results of spinal fusion in thoracic idiopathic scoliosis. They reviewed and analyzed the cases of 405 patients treated by John Moe with Harrington rod instrumentation, attempting to define the criteria used by Moe to perform a selective thoracic fusion. Curves were defined and divided into five types (I to V). King’s classification system was the first to specifically describe the most common type of idiopathic scoliosis, occurring in the thoracic spine, and recommend treatment based on the type of spinal curve in the disorder. More importantly, King et al presented recommendations for selectively fusing the thoracic spine and allowing the lumbar spine to undergo correction through compensation, thus preserving motion.1
Description
The first concept that King et al described was that of the stable vertebra. This was defined as the vertebra most closely bisected by the center sacral vertical line (CSVL), a single line drawn through the center of the sacrum perpendicular to the iliac crests. Curve flexibility was an important concept in determining whether a curve was structural or compensatory. Flexibility was quantified by measuring the Cobb angle on films of the patient during bending, dividing it by the Cobb angle on the anteroposterior (AP) film of the spine in the erect position, and multiplying the result by 100%. The flexibility index was a concept defined as the percentage correction of the thoracic curve subtracted from the percentage correction of the lumbar curve on side-bending radiographs. King and Moe defined a type I curve as an S-shaped curve in which both the thoracic and lumbar curves cross the CSVL, with the lumbar curve larger than the thoracic curve, and with a negative flexibility index. A type II curve is also one in which both curves cross the CSVL, but the thoracic curve is greater than or of the same magnitude as the lumbar curve, with a positive flexibility index. A type III curve is a thoracic curve with a lumbar curve that does not cross the CSVL. Type IV curves are similar to type III, except that the fifth lumbar vertebra is centered over the sacrum and the fourth lumbar vertebra is tilted into the long thoracic curve. Type V curves are double thoracic curves, with the first thoracic vertebra tilted into the upper thoracic curve.
Application
Guidelines for selecting levels of fusion were developed for each type of curve. It was recommended that type I curves be fused to L4. Type II curves could be treated with a selective thoracic fusion, with the fusion stopping at the stable vertebra and leaving the lumbar curve flexible and able to spontaneously correct. Type III and IV thoracic curves could also be fused to the stable vertebra. Type V curves were to be treated by fusion of both thoracic curves, with the fusion ending at the stable vertebra.
Reliability
King and his coworkers noted that 4 of the 405 patients in their study had progression of the lumbar curve of their deformity, requiring a second operation to extend fusion into the lumbar spine. According to their treatment guidelines, these patients’ spines were “inappropriately” fused either caudad or cephalad to the stable vertebra. From these findings, King et al concluded that for type II, III, IV, and V curves, selecting the stable vertebra rather than the neutral vertebra for the distal fusion level gave the most reliable results.
Updates and Revisions
As segmental instrumentation systems began to gain favor over Harrington rods among surgeons, many patients with idiopathic scoliosis exhibited decompensation in the lumbar spine when King et al’s recommendations for the distal level of instrumentation were followed. In 1992, Richards2 examined 24 patients with type II idiopathic scoliosis and a flexible lumbar curve >40 degrees. All 24 patients underwent selective thoracic fusion with segmental spinal instrumentation (Cotrel-Dubousset or Texas Scottish Rite Hospital instrumentation). Despite the amount of preoperative lumbar-curve flexibility, these lumbar curves remained larger after surgery than did the instrumented thoracic curves, resulting in spinal imbalance. Richards concluded that lumbar-curve flexibility was not a reliable predictor of compensatory lumbar-curve correction with the selective fusion of King type II curves when the lumbar curve was >40 degrees.
Roye et al3 published their results with treating scoliosis classified according to their King type with segmental instrumentation. They found significant decompensation in cases of King type II and III curves, whereas King type I, IV, and V curves had no decompensation. Also, in 1992, Knapp et al4 published their results in which fusion levels in 253 patients were based on the classifications and recommendations of King and coworkers. They recommended including part of the lumbar curve in cases of King type II curves, and that King type IV curves could be safely fused at one level proximal to the stable vertebra.4 Collectively, these studies suggest that the King system, developed during the era of Harrington-rod fixation, may not reliably predict the response of curves corrected with more powerful segmental instrumentation systems.
In addition to concerns about postoperative decompensation, concerns began to be raised about the reliability of the King classification system and its reproducibility among surgeons. In 1998, Cumming et al5 published an article on both the inter- and intraobserver reliability of the King classification for idiopathic scoliosis. They found that the median kappa coefficient for interobserver reliability was 0.44 and that the kappa coefficient for intraobserver reproducibility was 0.64. They therefore concluded that the reproducibility of the system was fair and the reliability of the system was poor. Behensky et al,6 assessing the reliability of the King classification in an article published in 2002, calculated a kappa coeficient of 0.45, indicating poor interobserver reliability. It was becoming more obvious that classifying curves according to the King classification system was somewhat unreliable. In combination with the instances of decompensation suggesting flaws in the King classification, concerns about lack of reproducibility of the King classification led to the development of a new classification system.
The Lenke Classification
The Lenke classification system was devised as a project by Lawrence Lenke and the Harms Study Group (HSG) to enhance the ability to accurately compare similar types of spinal curves among different treatment centers. The classification system was devised from the beginning to be descriptive, comprehensive, and reproducible with excellent inter- and intraobserver reliability. It sought to accomplish this goal by devising objective criteria for each type of curve, incorporating data on coronal deformity, flexibility, and sagittal alignment toward the goal of consistently classifying patterns of deformity and developing standardized treatment protocols for them, with reliable outcomes.
Description
To fully define a curve by the Lenke system, one must identify its type, lumbar modifier, and thoracic sagittal profile. The types of curve defined by the system were based on the features of a major curve and the structural characteristics of the minor curves accompanying it. A few terms need to be defined to use the system. The major curve is the curve of greatest magnitude, and is always considered structural. Minor curves can be structural or nonstructural. A nonstructural curve is defined as one that bends to less than 25 degrees on a radiograph of the bending patient. With these terms, curves can be classified as being of one of six types. In a type 1 pattern (main thoracic [MT] curve), the major curve is in the thoracic spine and the proximal thoracic (PT) and thoracolumbar (TL) curves are minor and nonstructural. In a type 2 pattern (double thoracic curve) the MT curve is the major curve, with the PT curve being minor and structural and the TL curve being minor and nonstructural. The type 3 pattern (double major curve) has the MT curve as the major curve, the PT curve as nonstructural, and the lumbar curve as minor and structural. The type 5 pattern (triple major curve) describes the PT, MT, and TL curves as all being structural, with either the MT or the TL curve as the major curve. In the type 4 pattern (thoracolumbar/lumbar [TL/L] curves) the lumbar curve is the major and only structural curve, with the PT and MT curves being minor and nonstructural. In the type 6 pattern (TL/L and MT curves), the TL/L curve is the major curve, measuring at least 5 degrees more than the MT curve, which is minor but structural. A pattern in which the difference between the lumbar and thoracic curves is less than 5 degrees can be categorized as being of type 3, 4, or 5 on the basis of structural characteristics in the MT and TL/L regions.
Lumbar-spine modifiers in the Lenke classification system are defined by the location of the center sacral vertical line (CSVL) on the apical vertebra of the lumbar curve. The CSVL is defined as a vertical line bisecting the cephalad aspect of the sacrum and perpendicular to the true horizontal. An “A” is used as a modifier when the CSVL runs between the lumbar pedicles of the lumbar apical vertebra. The curve must have a thoracic apex at or cephalad to the level of the 12th thoracic disc. Therefore, the modifier A can be used only for MT curves of types 1 through 3. The modifier B is used when the CSVL touches the apex of the lumbar curve between the medial border of the lumbar concave pedicle and a concave lateral margin of the apical vertebral body or bodies. These curves were defined as having an apex in the main thoracic region. The modifier C is used when the CSVL falls completely medially to the margin of the vertebra at the apex of the lumbar curve. By their designation, curves with the modifier C would seem to be simply the next step in a progression of lateral deviations from curves designated by modifiers A and B; however, a curve with the modifier C may represent a distinct pathological entity and probably requires a treatment plan that deviates from those for curves with modifiers A and B. Lumbar curves designated by modifier C are more likely to have significant rotation at the apex and to deviate more from the midline than curves designated by the modifiers A or B. These features could result in a clinical deformity with a significant lumbar prominence. They may also contribute to the curve behaving more like a structural curve with continued deformity that does not improve in the same manner as that of a flexible, nonstructural curve designated by modifier A or B. Thus, curves with the modifier C, despite being considered strictly nonstructural, may occasionally be included at least partly in the levels of the spine that are instrumented and fused in treating a case of scoliosis.
For the first time in any classification system for scoliosis, the sagittal profile of the spine is included in the Lenke classification system. Sagittal thoracic modifiers are defined as normal (N) if sagittal thoracic alignment ranges from 10 to 40 degrees as measured from T5 to T12. Curves with less than 10 degrees of kyphosis from T5 to T12 are given a minus (-) modifier, and curves with more than 40 degrees of hyperkyphosis are given a plus (+) modifier.
In the original article describing the Lenke classification and its rationale, Lenke and colleagues presented their evaluation of the reliability of the system. The kappa value for the new system was noted to be 0.92, with interob-server reliability for determining a curve at 93%. When compared with the King classification, the intraobserver reliability of the Lenke classification system was calculated at 85%, with a kappa value of 0.83. The intraobserver reliability among five surgeons using the King classification was 69%, with a mean kappa value of 0.69. The interobserver reliability for the new classification system was 85%, with a mean kappa value of 0.83. From these data, Lenke and coworkers concluded that their new system was more reproducible and reliable than the King system.
A follow-up article published by Lenke et al and reporting a multi-surgeon assessment of the new system for surgical decision-making in idiopathic scoliosis showed a high level of agreement (84 to 90%) in curve classification, although choices for operative approaches and fusion levels still varied widely.7 Subsequent reviews have focused on the intra- and interobserver reliability of the King and Lenke classifications. In an article published in 2006, Niemeyer et al8 concluded that both classifications had good reliability. On nonmeasured radiographs, the higher the level of orthopedic training and experience of the measurers, the better the inter- and intraobserver reliability Likewise, Ogon et al9 published a similar review of the reliability of the Lenke classification system, reporting that it was more reliable than the King classification, although proper classification of high thoracic and lumbar curves could be problematic. Most of these studies indicate that the Lenke classification provides a more reliable and reproducible way in which to communicate curve patterns than does the King classification, thus allowing surgeons to begin to speak the same language and compare results of different treatments in patients with similar curve patterns.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
