Classifications in Brief: The Lenke Classification for Adolescent Idiopathic Scoliosis

History Adolescent idiopathic scoliosis (AIS) is a spine condition that causes deformity in the coronal, sagittal, and axial planes. It is defined as a coronal curvature > 10°. The condition affects individuals between 10 and 18 years of age [11]. AIS is a diagnosis of exclusion once neuromuscular, neural, congenital, and syndromic causes of scoliosis have been ruled out [9]. Although AIS occurs in approximately 2% to 3% of the adolescent population, treatment is only indicated in < 10% of patients. In 1983, King et al. [12] introduced the first AIS classification system. The system has five thoracic curve types with recommendations for specific vertebral levels for spinal arthrodesis. It soon became the standard classification system. In the late 1980s, with the evolution of screws and multiple hook systems that allowed increased coronal curve correction, several limitations with the King classification system became evident. The King classification system did not include triple curves nor did it take into account the central sacral line or sagittal plane. Type 2 and 3 curves in the King system became difficult to distinguish, leading to postoperative spinal imbalances with newer implant systems. Research at the time also placed more emphasis on the importance of sagittal plane balance, which was lacking in King’s description [3, 6]. Studies that looked at the reliability of King’s system also found only poor to fair reliability, validity, and reproducibility [5, 14]. One study that examined the validity of King’s system, led by Lawrence G. Lenke [14], concluded that the system was not sufficiently reliable and likely not helpful for guiding treatment with the spinal fixation methods available at that time. To address the limitations of King’s classification system, Lenke developed a new AIS system in 2001 [16]. Purpose Lenke et al. [16] designed their classification system around several parameters that they thought were essential for AIS. They wanted the system to improve on the poor/fair reliability, validity, and reproducibility. Several steps were taken to ensure this including a comprehensive system of all curve types and be easily understood to maximize practical value in a clinical setting. Unlike the King classification, the Lenke system additionally emphasized the sagittal plane along with the coronal plane rather than the coronal plane alone. The developers also wanted this system to better assist surgeons when considering operative intervention and where to use segmental spinal fixation. Lastly, to improve on King et al.’s system, their new system needed to have good-to-excellent intraobserver and interobserver reliability by the steps previously stated. Description The Lenke classification system [16] is organized according to curve type (1-6) with a sagittal thoracic modifier (-, N, or +) and a lumbar spine modifier (A, B, C). All definitions for the curve types were established by the Scoliosis Research Society (SRS). Structural curves are defined as those that are at least +25° on bending films or > +20° of kyphosis. When multiple curves are present, the major curve is defined as the largest curve; minor curves can still be defined as structural however. The basis of this classification system was to determine which vertebral levels were appropriate for an arthrodesis. Curve types are based on location (Table 1). Thoracic curves have an apex from T2 to the T11/T12 disc. Thoracolumbar curves have an apex at T12 to L1, whereas lumbar curves have an apex from the L1/L2 disc to L4.Table 1.: Lenke’s criteria for curve classificationType 1 is a main thoracic curve. There may also be proximal thoracic and thoracolumbar/lumbar nonstructural minor curves. Type 2 is a double thoracic curve with a proximal minor structural curve and a main thoracic major curve. There may also be a thoracolumbar/lumbar curve that is nonstructural and minor. Type 3 is also a double major curve with the main thoracic curve being the major curve. However, the second curve is in the lumbar region and both curves are structural. Type 4 is a triple major curve with structural curves in the proximal thoracic, main thoracic, and TL/L regions. Once again, the thoracic curve is the major curve. Type 5 is a single structural curve in the thoracolumbar/lumbar region surrounded by two minor nonstructural curves. Type 6 is a thoracolumbar/lumbar-main thoracic double curve. The thoracolumbar/lumbar is the major curve, but the thoracic curve is also structural. The lumbar modifiers Type A, B, and C are used to describe the relationship between the center sacral vertical line (CSVL) and the lumbar curve apex (Fig. 1). Type A has the line between the pedicles of the lumbar apex. For Type B, the CSVL touches the apical vertebral body. In Type C, the CSVL is completely medial to the vertebral body. The thoracic sagittal modifier describes the thoracic kyphosis from T5 to T12. Less than 10° of kyphosis is designated with “-”, > 40° of kyphosis a “+”, and “N” (normal) is kyphosis in the range of 10° to 40° [16].Fig. 1 A-C: This figure illustrates Lenke’s lumbar modifiers. (A) Type A has the CSVL between the pedicles of the lumbar apex. (B) Type B has the CSVL touching the apical vertebral body. (C) Type C has the CSVL completely medial to the vertebral body.We show examples of patient radiographs with adolescent scoliosis categorized by Lenke’s classification system (Fig. 2).Fig. 2 A-B: This figure shows examples of AIS classified using Lenke’s system. Although this radiograph shows slight pelvic obliquity, images should aim to be taken with a level pelvis. (A) This shows an example of a Lenke curve type 5C. (B) This illustrates a Lenke curve type 1A.Validation In general, the Lenke classification achieved its goal of improved intra- and interobserver reliability compared with the King schema [7, 10, 16, 18, 20], although in one study the reliability of the Lenke system was lower [21]. There are few studies evaluating the clinical outcomes correlated with the Lenke system, but it has proven to further standardize surgical treatment approaches. In Lenke et al.’s [16] original description, five surgeons who were members of the SRS group that developed the system examined 27 radiographs (standing coronal and lateral as well as supine side-bending views). This group had previously studied these radiographs to assess the reliability of the King classification system. The surgeons interpreted all films on 1 day and then reinterpreted a day later to assess intraobserver reliability. The interobserver and intraobserver reliability κ values were 0.92 and 0.83, respectively. A second group of seven surgeons who also were SRS members evaluated the same set of radiographs. These results demonstrated κ values of 0.74 for interobserver reliability and 0.89 for intraobserver reliability (Table 2).Table 2.: Studies evaluating reliability of the Lenke classification system of adolescent idiopathic scoliosisA separate study performed by Lenke et al. [15] in 2001 presented seven preselected patients with operative AIS to 28 scoliosis surgeons at a spine surgery meeting. The multisurgeon review showed high agreement with curve classifications. Individual system components approached 90% agreement with the entire classification of curves approaching 80%. However, the surgeons’ surgical approaches and chosen fusion levels were highly variable. This reinforces the need for a classification system that allows the grouping of similar curves to objectively evaluate variable treatment options for each curve pattern. Ogon et al. [18] performed a reliability study on the Lenke system that included five surgeons who reviewed 51 patients’ radiographic films twice with a 3-week interval in between each reading. The researchers found an inter- and intraobserver rating of κ = 0.62 and κ = 0.73, respectively. They concluded that Lenke’s system was more reliable than King’s classification system [12], and the major reason for disagreement when using the Lenke classification was assessment of the lumbar modifier (Table 2). Some surgeons have advocated for fusion of a nonstructural thoracolumbar curve with a “C” modifier, whereas others think Lenke 3 curves with an “A” or “B” modifier could be amenable to a selective thoracic fusion [8]. In addition, determining the apical vertebrae can be challenging if the apex is located at the disc space. It has been reported that there are surgeons who also have difficulty differentiating between a type II and type IV curve [8]. In Richards et al. [21] (Table 2) four reviewers (two pediatric orthopaedic scoliosis surgeons and two orthopaedic fellows) interpreted radiographs of 50 patients with adolescent scoliosis. The participants reviewed all the images on two separate occasions, although the timeframe was unclear. They found interobserver to be κ = 0.50 and intraobserver reliability to be κ = 0.60. When the researchers examined the curve type separately, the reliability increased to κ = 0.64 for interobserver and κ = 0.76 for intraobserver with the sagittal modifier having a reliability ranging from κ = 0.41 to 0.47. Qiu et al. [20] asked five scoliosis surgeons to individually evaluate 62 patients’ radiographs. The radiographs were cleaned of marks before each review, and the surgeons reviewed the same films again 3 weeks later in random order. Interobserver reliability for the Lenke classification system in this report was κ = 0.81 and intraobserver reliability was κ = 0.83 (Table 2). In Duong et al.’s [7] study (Table 2) six reviewers who were SRS members evaluated 68 patients’ radiographs twice at 6-month intervals to assess the reliability of the Lenke classification system’s lumbar modifier. Duong et al. were uncertain of the lumbar modifier’s importance and proposed to use their results as a basis for a new three-dimensional classification system; however, to our knowledge, this system has not been published. Their interobserver agreement at the first trial was κ = 0.56, which improved to κ = 0.64 during the second interval. The mean intraobserver reliability in this study was κ = 0.69. Duong et al. [7] also reconstructed the lumbar region of each spine using a stereoradiographic technique and a computer algorithm that correctly identified the curve 93% of the time. Hosseinpour-Feizi et al. [10] compared the interobserver reliability of the King et al. [6] and the Lenke classification systems (Table 2). They found Lenke’s system to be more reliable with a κ coefficient of 0.97 on curve type and 0.92 on the lumbar modifier, whereas with King’s system, the κ coefficient ranged from 0.16 to 0.52. To improve the reliability of the Lenke classification system, Zhang et al. [25] conducted a study using a computer-aided system to help measure angles and identify apical vertebrae. Five operators reviewed 62 patients; the interobserver reliability was κ = 0.81 and intraobserver reliability was κ = 0.89 (Table 2). Aside from evaluating the reliability of the Lenke classification system, there is a large number of studies that use the system to evaluate clinical outcomes [1, 2, 13, 17, 22-24]. However, there are only a few studies that explicitly measure the utility for the system to guide treatment with mixed results [4, 8, 19]. Despite continued disagreement between treatment decisions, a study by Clements et al. [4] showed that since the introduction of the Lenke system, there has been a reduction in the variation of treatment approaches. Future efforts to improve the classification of adolescent scoliosis may choose to better evaluate curve types in three dimensions using CT reconstructions or specifically look at the reliability of the modifiers. There is also more need for future research to evaluate how successful the Lenke system is at guiding treatment. Overall, most of these studies indicated that the newly developed Lenke classification system has better reliability compared with the established King classification. In addition, the studies seemed to note the greatest variability in the selection of the lumbar modifier. Limitations Although Lenke’s system shows reliability ranging from 0.5 to 0.97, it has some limitations. Lenke’s design allows for 42 potential curve patterns, a level of complexity that may not be ideal for a busy orthopaedic surgeon [16]. However, the curve types should be well known to surgeons who treat scoliosis, and modifiers can only be added to provide additional information. As research in adolescent scoliosis advances, greater emphasis is placed on evaluating curves in three dimensions, which cannot be done with the Lenke classification. This would likely be done with a CT scan, exposing patients to more radiation, a consequence that should only be done in patients with severe AIS. This classification system also relies on side-bending films. Although many surgeons use traction films or lateral films with patients laying over a bolster, most surgeons do not regularly obtain bending films until surgical treatment is considered. It is also important to understand that this classification is only validated for AIS. Applying the system to other conditions such as syndromic scoliosis or congenital scoliosis is improper, although they have no suitable classifications. Curves that have more rotation generally require more sophisticated surgical planning and intraoperative manipulation. Lastly, there remain a subset of “rule-breaker” curves, in which the treatment did not follow the surgical recommendations made by the Lenke system [4]. After the introduction of the Lenke classification, curves 3, 4, and 6 have a higher percentage of “rule-breakers” than before, although these are the least common AIS curves [4]. Conclusion Overall, the Lenke classification system for AIS has improved reliability and ease of use compared with the previous system. The system’s reliability κ value ranges from 0.5 to 0.97 in terms of interobserver and intraobserver agreement. A comprehensive description of all curve types allows the system to have value in both guiding treatment options and future research on AIS. However, compared with other AIS classifications, the system is complex, which may deter use in clinical practice. The Lenke classification system has improved usefulness compared with prior systems, but there is still controversy over the utility when guiding treatment. Future studies should further evaluate how the system accurately guides surgical decisions.