19.1 Introduction

The complexity and variety of curve patterns in adolescent idiopathic scoliosis (AIS) has made the creation of an algorithm for surgical correction difficult. As the surgical treatment for scoliosis has progressed from noninstrumented fusions to current pedicle screw instrumentation, surgeons have debated which vertebral levels need to be included in the fusion. The benefits of reducing the number of fusion levels have been enshrined in the principle of selective fusion. However, the reduction of fusion levels must not compromise balanced spinal alignment.

The impetus for King’s classification system was to create guidelines for deciding which levels need to be fused based on curve type. In addition to describing five types of curves, he defined “structural” and “compensatory” curves. Structural curves have underlying abnormalities which cause the deformity. Compensatory curves occur to compensate for the main structural curves. King recommended only fusing the structural curves as the compensatory curves would resolve once the structural curve was corrected. ¹ However, the King system did not specifically describe patterns with multiple structural curves. ² As surgical treatments for scoliosis evolved into pedicle screw fixation, there was a need for a newer classification system. The differentiation between structural and compensatory curves laid important groundwork for the Lenke classification system and the understanding of double and triple major curves.

Lenke and colleagues designed the AIS curve classification system in 2001 to improve reliability, validity, and reproducibility in the determination of all curve types for use in a clinical setting. ³ Today, this classification system is utilized worldwide by a majority of scoliosis surgeons. This system relies on measurements taken from standard radiographs from the front, side, and bending positions. The curve is classified by curve type based on the three regions of the spine; a modifier based on the distance of the center of the lumbar spine to the midline; and a sagittal thoracic modifier based on the amount of lateral curvature to the thoracic region. ⁴

Unlike the King system, the Lenke classification provides an independent evaluation of each of the scoliotic curves (proximal thoracic [PT], main thoracic [MT], and thoracolumbar [TL]) for structurality. Once a curve is defined as structural, the recommendation is to treat the curve. Cases in which there was more than one structural curve created the need for defined types of double and triple major curves. These multiple structural curves are labeled as Lenke types 3, 4, and 6. This classification provides the basis for recommendations regarding the treatment of double and triple major curves.

Some studies have suggested that the Lenke system describes only 6 of the 10 possible curves. ^{3 , 5} There is no Lenke type for an isolated structural PT curve. In addition, the Lenke type 2 curve, which describes a double thoracic curve, describes only a major main thoracic curve and a minor PT curve. It ignores a curve that might have a major PT curve and a minor main thoracic curve. The Lenke type 4 curve, which describes a triple major curve, has only a major main thoracic curve. There is no triple curve with the main PT or main TL curve in the Lenke system. Recently, a Brazilian cohort developed a three-dimensional classification for AIS with the aim of including curves that are not described by the Lenke system. ⁵ Three-dimensional assessment of the deformity may lead to new treatment options for patients with AIS. ^{6 , 7}

19.2 Curve Definitions

The Lenke classification system remains the most popular classification utilized by deformity surgeons. Its definition of double major and triple major curves has shaped current understanding and treatment strategies. For this reason, an intimate understanding of the system is required for any successful spine surgeon.

The Lenke type 3 or double major curve is defined by both a structural MT (major) and a structural but minor TL/lumbar (TL/L) curve (Fig. 19‑1a). In the Lenke system, these curves typically have a lumbar “C” modifier but may also have a lumbar “B” or even an “A” modifier. ⁸ There may be varying degrees of TL kyphosis between the two curves that can affect treatment decisions. As discussed earlier, TL kyphosis greater than 20 degrees from T10 through L2 also defines a curve as structural, even when the side-bending Cobb angle measurement is less than 25 degrees. A multicenter study of 606 patients demonstrated a prevalence of 11% for Lenke type 3 double major curves. ⁹

The Lenke type 4 or triple major curve pattern has a structural curve in all three regions of the spine, consisting of a PT (minor), MT (major), and TL/L (minor) curve (Fig. 19‑1b). The lumbar apical vertebrae will typically lie lateral to the center sacral vertical line (CSVL) with a lumbar modifier “C” for curves of this type; however, this is not always the case. ⁹ The type 4 curve is the least common of the curves with a reported frequency in the literature of 1.4 to 4%. ^{9 , 10 , 11 , 12}

In contrast to a Lenke type 3, the type 6 curve pattern resembles the King type I curve pattern, with the TL/L curve being the larger or major curve (Fig. 19‑1c). All Lenke type 6 curves have a lumbar “C” modifier. ⁴ Lenke type 6 curves are also rare, with a prevalence just above 3%. ⁹

19.3 Treatment Principles

The overall guiding treatment principles for double and triple major curves are to correct the deformity; to improve body shape and self-image; and to minimize the number of levels in the fusion. The complexity and severity of the three-dimensional deformity of double and triple major curves can pose challenges for the spine surgeon.

Correction of deformity: Three-dimensional deformity correction includes improvement of the Cobb angle; decrease in apical vertebral rotation (AVR) and translation; restoration of coronal and sagittal global alignment; and restoration of kyphosis in these typically hypokyphotic deformities. ¹³ Optimal radiographic correction will result in improved body shape as has been shown in studies utilizing surface topography.

Minimization of the number of fusion levels: Although an important goal of surgery for the AIS patient is to minimize fusion levels and preserve distal motion segments, this is a challenging aspiration in the treatment of double major or triple curvatures. With the introduction of instrumentation by Harrington in the late 1950s, fusions were long and typically extended into the lower lumbar region. ^{14 , 15} Long-term follow-up studies of these fusions found greater than average back pain and risk of lumbar degeneration below the end-instrumented vertebra. ^{16 , 17 , 18 , 19 , 20 , 21} Reducing the number of fused vertebral levels maximizes spinal flexibility and distributes stress across a larger number of remaining distal lumbar motion segments. ²² Theoretically, this may diminish the long-term risk of disc degeneration at the adjacent distal levels. ²³

Beyond these basic principles, little has been written to guide surgeons in treating double and triple major curves. Because these curves are less common than others, they have typically been discussed as part of larger, more comprehensive series of AIS patients. The traditional treatment for double and triple major curves has been arthrodesis of all curves. ^{24 , 25 , 26 , 27 , 28 , 29 , 30 , 31} Currently, pedicle screw instrumentation is used; however, their benefits are not truly known at this time. New methods are currently being investigated to better analyze the three-dimensional deformity associated with scoliosis. The Lenke classification recommendation is the fusion of all the structural curves to prevent coronal decompensation through the uninstrumented curve. ⁴

19.4 Recent Trends in the Surgical Decision-Making Process

Owing to the rarity of double and triple major curves, there is a paucity of literature on the topic. However, insight into treatment strategies can be found in studies that include subsets of patients with these deformities. Important considerations for surgical planning include the proximal extent of the fusion (upper instrumented vertebra [UIV)], type of construct, the distal extent of the fusion (lower instrumented vertebra [LIV]); selective versus nonselective thoracic fusion; strategies for axial plane correction; and choice of implants. These recent trends in the surgical decision-making process have the goals of better accomplishing the aforementioned principles of treatment: correction of deformity, improvement of cosmetic appearance, and minimization of fusion levels.

Extension to L3 or L4: AIS patients with major TL or major lumbar curves should have these curves included in the fusion because they are, by definition, structural curves. However, there has been significant controversy as to whether the distal fusion can be stopped at L3 without causing a significant risk of decompensation. Studies have suggested that fusions extending past L3 can cause higher rates of disc degeneration, greater functional loss, or more severe low back pain. ^{16 , 17 , 32}

King et al suggested that fusion should extend to the last “stable vertebra.” He defined the “stable vertebra” as the most proximal lumbar vertebra that is bisected by the CSVL. ¹ The Lenke classification uses radiographic and clinical evaluations to determine the extent of fusion, UIV, and LIV. The selection of the UIV in a structural thoracic curve is based on the shoulder height inequality seen clinically and on the upright radiographs. The LIV is determined by the lumbar modifier of the curve types. Lenke recommended fusion to the vertebra that is at least intersected by the CSVL. Thus, the LIV could be one or two levels cephalad to the true stable vertebra.

For type 3 double major (DM) curves, the LIV needs to be either L3 or L4. If the apex of the TL/L curve is at L2 or caudal, the L3–L4 disc is either convex or open to the convexity of the TL/L curve; and if the L4 vertebral rotation is 1 or more, then the instrumentation should be extended to L4. L3 can be chosen as the LIV if the apex of the TL/L curve is at the L1–L2 disc or cephalad, the L3–L4 disc is either neutral or closed on the convexity of the TL/L curve, and the L3 vertebral rotation is 1.5 or less. Similar operative principles are used for the treatment of type 4 and type 6 curves.

Lee et al performed a retrospective study that compared AIS patients with major TL/L curves who had distal fusion extending to either L3 or L4. ³³ The goal was to determine if the distal extension to L3 was sufficient. They found there was no difference in the radiological or clinical outcomes in AIS patients according to the distal fusion level. However, they noted a trend where patients who had fusion to L3 had the lower end vertebra at L4 or below, and the last touched vertebra at L5 showed less correction of the deformity. In addition, the subjacent disc angle was increased, suggesting possible disc degeneration and an adding-on phenomenon. Therefore, they postulated that major TL or L correction would be sufficient distally at L3 with the lower end vertebra at L3 or higher and the last touching vertebra at L4 or higher.

Lonner et al retrospectively reviewed prospectively collected data from an AIS registry with a minimum follow-up of 10 years. ²³ A composite radiographic score was calculated to determine the disc degeneration status caudal to the LIV. Various risk factors which potentially cause increased disc degeneration distal to the LIV were evaluated. The factors evaluated included the following: the number of levels fused, LIV translation from the CSVL, residual lumbar curvature, LIV tilt, location of the LIV, wedging of the disc below the LIV, surgical approach, and type of construct used for surgery. LIV translation greater than 2 cm from the CSVL and LIV below L3 were found to be associated with an increased risk of disc degeneration distal to the LIV.

Hamzaoglu et al modified the Lenke recommendations for determining the lowest instrumented vertebra. ³⁴ They looked at lateral bending films toward the concavity of the curve, as opposed to neutral films, to determine the most proximal lumbar segment intersected by the CSVL. They also ensured that the LIV’s rotation was corrected by one or two Nash-Moe grades in the convex lateral bending films. Utilizing this criterion allowed Hamzaoglu et al to end the fusion at L3 more often; however, some curves decompensated despite the use of this criterion. The authors evaluated Lenke type 3C and 6C curves and compared traction X-rays under general anesthesia (TrUGA) to traditional neutral radiographs to determine the lowest instrumented vertebra. In 46 cases (52%), traditional radiographs determined L4 to be the lowest instrumented vertebra, while TrUGA indicated stopping the fusion at L3. In all of these patients, fusion was stopped at L3. All patients had successful radiographic outcomes despite the discrepancy between the LIV on traditional versus traction radiographs. Favorable radiographic outcomes were defined as the CSVL to be within 2 cm of T1, tilt angle less than 10 degrees, and the L3–L4 disc wedge less than 10 degrees at final follow-up (average: 5.4 years). The study concluded that TrUGA may be an alternative method for the selection of fusion levels in Lenke type 3 and 6 curves and may help save L4 compared to traditional radiographic assessments.

In a similar study, Erdem et al conducted a study of 93 patients comparing traction X-rays under general anesthesia and traditional radiographs for selection between the L3 and L4 vertebrae. ³⁵ In 56 patients (60%), traditional radiographs indicated the L4 as the LIV, while TrUGA indicated the L3 as the LIV. All these patients had fusion to L3. There was no difference in radiologic or clinical outcomes or the need for additional surgeries. Erdem et al concluded that traction radiographs are an efficient alternative in the selection of the lowest instrumented vertebra in patients with Lenke type 3C, 5C, and 6C curves.

Minimizing fixation : Double and triple major curves represent large structural deformities with significant complexities when compared to single major curves. The uncommon incidence of double and triple major curves further complicates the development of a single surgical treatment strategy to treat these deformities.

Recent trends in the surgical treatment of scoliosis have sought to minimize the use of instrumentation. There are definable costs and surgical risks associated with placing bilateral pedicle screws at each segment of the deformity. This density of instrumentation may not be necessary to maintain curve correction.

Tsirikos et al presented a surgical technique limiting the number of pedicle screws placed during the correction of double and triple major curves in 191 patients with AIS. ³⁶ Their convex segmental pedicle screw technique placed pedicle screws in each segment on the convexity of the curve. However, they placed only one or two screws at each caudal and cephalad level on the concavity side of the curve. The decreased number of pedicle screws showed a low rate of neurological and vascular complications. Despite the reduced fixation, the study showed that the technique achieved satisfactory correction of scoliosis, improved thoracic kyphosis, and normal sagittal balance.

Implant options : Recently, there has been interest in the benefits of nonstandard rod geometric configurations and materials. In a retrospective two-center study of AIS patients, Ohrt-Nissen et al used a pedicle screw fixation system with a modified rod with the goal of better restoration of thoracic kyphosis in the typically hypokyphotic thoracic spine associated with AIS. ³⁷ The modified rod used in this study was beam-shaped, fashioned to increase its biomechanical strength especially when resisting bending and shear loads. Two consecutive cohorts of AIS patients received one of three different types of rod constructs and pedicle screw placement at every level. A standard construct used bilateral beam-like rods; a hybrid construct used a circular rod on the convex side of the curve and a beam-like rod on the concave side; and a modified construct used beam-like rods that transitioned to circular rods in the cranial two to three fusion levels. The modified construct was designed to provide better rod contouring and decrease junctional stress. Improved restoration of the kyphosis was achieved with the bilateral modified rods compared to the standard rods. The modified construct and hybrid constructs also had a significantly lower postoperative loss of thoracic kyphosis.

Hwang et al studied the differences in hybrid constructs consisting of caudal pedicle screws combined with rostral hooks and/or sublaminar wires versus all-pedicle-screw anchors to achieve maintenance of thoracic kyphosis in 127 patients with Lenke types 1 to 4 curves. ³⁸ The outcomes were comparable between the two groups at 2 and 5 years of follow-up. However, the hybrid constructs were associated with more concurrent anterior releases and thoracoplasty procedures. These procedures have the potential for increased morbidity, with all pedicle screw constructs providing an advantage in patients with more advanced curve patterns.

Rod-rotation technique : Double and triple major curves are technically challenging when rod-rotation techniques and sequence are considered. Basic scoliosis correction principles need to be followed to achieve optimal coronal Cobb correction, sagittal and coronal balance, and shoulder balance. The rods should be precontoured to match optimal sagittal profile in the thoracic and lumbar spine. The rod insertion is started first by engaging the lumbar vertebral screws on the convexity (predominantly left), particularly if the lumbar curve is large. Compression forces are applied on the convexity to produce lordotic alignment in the lumbar segment. The rod is then sequentially reduced in the thoracic pedicle screw tulips on the concavity (left). The bottom portion of the rod is captured with the set screws and set screws are tightened. The rostral end of the rod is loosely engaged in the tulips with the help of set screws. The apical tulips are sequentially engaged with the help of screw tab extenders/reduction towers. Once all the screws are fully engaged, the rod-derotation maneuver is carried out with the help of vice grips. The rod-rotation maneuver is carried out from left to right side with an assistant applying counterpressure on the rib prominence. This leads to coronal Cobb correction as well as helping restore the thoracic kyphosis. Contralateral rod is then placed and captured in the tulips with the help of set screws. Fine corrective manipulations can then be carried out to restore the shoulder balance and T1 tilt. This could be achieved with segmental compression and distraction maneuvers. Attention should also be given to the coronal tilt of the LIV. Axial plane individual vertebral derotation can be achieved with the help of coupling devices that link the tulip tab extenders on both sides. This can be carried out at each individual periapical vertebra or at the adjacent apical group of vertebrae.

Correction of triple major curve follows similar principles with rod placement starting at the convex side of the L/TL curve. The PT curve can then either be reduced to the left side rod or partially corrected into the main thoracic curve with the help of right sided rod placement. This decision is based on preoperative sagittal profile of the proximal curve. This technique converts the triple major curve into a DM curve and the rest of the correction maneuvers can then be repeated as mentioned earlier.

Axial plane correction : In the 1980s, Cotrel–Dubousset used special instrumentation combined with a derotation maneuver that attempted to correct all three planes of deformity. Since then, surgeons have sought to develop new techniques to address rotational deformity. The axial plane correction has traditionally been achieved using rod-rotation maneuvers. Surgical techniques have been further modified to achieve better axial plane correction. Lee et al used screw derotators in addition to rod-rotation maneuvers and achieved significantly better axial plane correction. ³⁹

Yang et al placed vertebral column manipulator devices over the apical vertebrae to achieve a direct vertebral rotation technique in DM curves. A synchronous derotation was carried out to achieve convex correction followed by concave rod placement. This technique provided safe and excellent axial and coronal plane correction, in addition to avoiding the pitfall of worsening the apical rotation of the minor curve when the major curve is being derotated. ⁴⁰ Differential rod contouring is another technique that allows significant axial derotation of the apical vertebrae with good correction of the rib prominence. ⁴¹

Selective fusion : In keeping with the principle of minimizing fusion levels, the concept of selective thoracic fusion was introduced by Moe and further elaborated on in the King–Moe classification. ⁴² The term “selective fusion” refers to a surgical strategy of selectively fusing the primary structural curve while leaving the minor structural thoracic/lumbar curves untreated even if they cross the midline. Selective fusion can provide balanced curve correction while also maximizing the number of mobile vertebral segments. The performance of selective fusion in which minor structural curvatures are left untreated is controversial for Lenke 3, 4, and 6 curvatures. Selective thoracic fusion has generally been defined for Lenke 1 curvature and has been to some extent accepted for this type of curve.

In the decision-making process concerning whether or not to limit the inclusion of minor structural curvatures in the fusion, the patient’s lifestyle can be an important factor: athletes and dancers require lumbar mobility and may be inclined to prefer a selective fusion. It is of utmost importance to analyze each curve individually using stricter criteria for structural curves. Clinical assessment of the deformity and the patient’s skeletal maturity are equally important. The magnitude of a thoracic rib prominence and a lumbar prominence should be carried out, and the patient’s willingness to accept a lumbar prominence should be assessed when a selective thoracic fusion is contemplated. The sagittal TL profile should be carefully analyzed, as selective thoracic fusion with a kyphotic TL profile could lead to the progression of the deformity postoperatively. Intraoperatively, special attention should be given to the tilt of the LIV and its correlation with the lumbar modifier. An optimal tilt may be left in the LIV for “B” and “C” lumbar modifier curves. However, it is the surgeon’s responsibility to ensure that patients understand the risk of complications associated with selective fusions. These include the following: a risk for continued curve progression of the unfused structural curvature, junctional issues, coronal imbalance, and the need for revision surgery to extend the fusion. ⁴³

Studer et al reported on 30 patients who had selective fusion: 16 having selective thoracic fusion (curve types 1–3) and 14 having selective TL/L fusion (curve types 5C and 6C). ⁴⁴ Coronal decompensation occurred in 25% of the selective thoracic fusions and in 29% of the TL/L selective fusions. An adding-on phenomenon occurred in 25% of selective thoracic fusions and in 36% of the TL/L fusions. Despite the high complication rates, there were no revision surgeries at 11 years of follow-up.

King et al discussed the selective fusion concept and recommended selective thoracic fusion for their type II curves. ¹ Lenke et al further refined this recommendation by listing parameters that would maximize success with selective thoracic fusions. ²⁹

In 2017, Lenke et al presented the outcomes of 26 patients with type 3 and type 4 curves who had selective thoracic fusion (fused to L1 or proximal). ⁴⁵ In addition to the curve types, Lenke also relied on the coronal Cobb ratio, the AVR ratio, and the apical vertebral translation (AVT) ratio to determine the suitability for selective thoracic fusion in these patients. These patients had good outcomes and required no revision surgery at 5 years of follow-up. Future studies must seek out additional radiographic and clinical indicators for those with double curves who may benefit from selective thoracic fusion.

A selective thoracic fusion can be considered for Lenke type 3 curves if the TL curve is small, flexible, and has a minimal axial deformity. For Lenke type 3C and 6C curves, the Cobb angle, AVR, and AVT are important radiographic parameters. In type 3C curves, selective thoracic fusion may be appropriate if the ratios for these parameters between the thoracic and TL/L curves are greater than 1.2.

Shulz et al studied patients with Lenke 1C to 4C curves to determine additional radiographic guidelines for selective fusion in order to obtain an optimal outcome. ⁴⁶ They recommended performing a selective thoracic fusion in patients with a preoperative lumbar Cobb angle less than 45 degrees and/or a preoperative lumbar bend less than 25 degrees. However, this suggests that most structural TL/L curves would be at risk of “less-than-ideal” outcomes since curves less than 25 degrees are not considered structural.

Not all studies have found success with selective fusion for Lenke type 3 curves. Singla et al ⁴⁷ completed a retrospective multicenter study of 25 patients who had selective thoracic fusion compared to 49 patients who had a fusion of both curves. At 2 years of follow-up, the majority of patients with selective fusion showed significantly more coronal imbalance, less lumbar curve correction, less lumbar apical vertebra correction, and less correction of the lumbar prominence.

At 10 years postoperatively, Large et al ⁴⁸ found that patients who had a successful selective fusion of a DM curve had significantly less back pain and stiffness than those who had both curves fused. Because their study predated the current Lenke classification, it is unclear whether all of the lumbar curves in the study of Large et al were structural and would have required fusion under the current guidelines.

When considering the fusion of the PT curve in a Lenke type 4 curve pattern, indications for arthrodesis can be extrapolated from data for double thoracic or King type V curve. Although spontaneous correction of the upper thoracic region can occur, both the proximal extension of the instrumentation used in treating these curves and the failure to correct a PT curve may have significant effects on shoulder balance. ^{49 , 50 , 51} One may consider making the proximal instrumented vertebra the vertebra at the apex of the PT curve if the T1 tilt is less than 5 degrees. Other literature has suggested that the best predictor of postoperative shoulder balance is the clavicle angle rather than the absolute or bending Cobb angle or the T1 tilt. ⁵⁰ The clavicle line is defined by the intersection of a tangential line connecting the two highest points of each clavicle with a horizontal reference line. If the clavicle angle is positive (left clavicle higher than right clavicle), then fusion of the PT curve is likely to result in improved shoulder symmetry.

The criteria for fusion in the Lenke classification include a side-bending Cobb angle greater than 25 degrees or T2 to T5 kyphosis greater than 20 degrees, or both. Cil et al attempted to evaluate these criteria by retrospectively evaluating patients who had undergone fusion before the development of the Lenke classification. ⁵² They divided patients who had nonstructural PT curves into two groups: Group 1 had a proximal fusion to T3 or above and Group 2 had an UIV at T4 or below. They compared postoperative bilateral coracoid heights, clavicle angles, and T1 tilt in the two groups and found similar outcomes. This suggested that the Lenke criteria for leaving a PT curve unfused were safe. When considering fusion of a PT curve, we use the bending Cobb angle but also evaluate the patient’s radiographic clavicle angle and clinical shoulder height. A patient with either a level or slightly lower left shoulder height may need instrumentation of the upper thoracic curve if correction of a large MT curve is anticipated. It is important to include the curve flexibility with the radiographic and clinical shoulder-height measurements when determining the need to instrument an upper thoracic curve.

Selective TL/L fusion is also a consideration in attempting to minimize fusion levels. In a study of 49 patients with major TL/L and “partially structural” thoracic curves, Sanders et al ⁵³ assessed compensatory thoracic curve correction when anterior fusion was done only on the TL/L scoliosis. They determined that a lumbar-to-thoracic Cobb angle ratio greater than 1.25 and a thoracic curve of less than 20 degrees on bending films, as well as a closed triradiate cartilage, were the best predictors of successful selective fusion. Lenke et al ⁵⁴ expanded on this, utilizing the specific curve patterns of the Lenke classification as a template. Their radiographic criteria for the selective fusion of type 6C curves included a ratio greater than 1.25 for lumbar-to-thoracic curve Cobb angle, AVT and AVR, a flexible thoracic curve (ideally < 25 degrees), and TL junctional kyphosis of less than 10 degrees. Their clinical criteria for selective fusion included a level or high left shoulder (for a right thoracic and left TL curve pattern), a lumbar truncal shift exceeding the thoracic truncal shift, a lumbar-to-thoracic curve ratio greater than 1.2 by scoliometric measurement, and an acceptable rib prominence.

19.5 Surgical Approaches

The traditional approach to the treatment of double and triple major curves is posterior instrumentation and fusion. ^{30 , 55 , 56 , 57} Whether using pedicle screws, hooks, wires, or hybrid constructs, a posterior approach allows access to all curves through a single incision. Sagittal deformities can be addressed with multilevel posterior osteotomies. ⁵⁸

Anterior approaches, on the other hand, can be used in combination with a posterior approach for rigid curves, prevention of the crankshaft phenomenon, and the treatment of severe sagittal deformities. ^{59 , 60 , 61} We consider anterior release for curves greater than 90 degrees or particularly stiff curves that do not bend to less than 50 degrees. However, with pedicle screw instrumentation and wide posterior osteotomies, the role of an anterior release has become less clear. Historically, as compared with posterior instrumentation, anterior instrumentation has been shown to save fusion levels and result in better correction of the uninstrumented lumbar curve. ^{62 , 63 , 64 , 65} As new surgical instrumentations and techniques continue to be evaluated, this may not remain the case. When instrumentation is used in compression, anterior surgery of the thoracic curvature has more reliably resulted in the restoration of kyphosis than has posterior surgery. ^{63 , 66}

When considering anterior approaches, one must understand the associated morbidity. An anterior fusion through an open thoracotomy has been shown to adversely affect pulmonary function. ^{67 , 68 , 69} More recently, video-assisted thoracoscopic surgery (VATS) has become a popular alternative to open thoracotomy. ^{66 , 70 , 71 , 72} Studies have demonstrated a reduced effect of VATS on pulmonary function testing. ^{73 , 74 , 75} Whether this translates into improved clinical outcomes remains unclear. Interestingly, no significant diminution in pulmonary function was found for the thoracoabdominal approach despite the associated disruption of the diaphragm. ⁷⁶ In patients with a severe pulmonary compromise, anterior thoracic surgery should be avoided because of a possibly significant risk of further deterioration.