Anterior spinal approaches have been indicated for thoracolumbar curves in adolescent idiopathic scoliosis (AIS).1,2 These approaches facilitate large coronal curve correction and derotation, short segment fusions, saving of distal levels, and reduced operative blood loss. Recent evidence suggests, however, that approach-related pulmonary function impairment may limit the utility of this approach.3–8 Consequently, the posterior-only approach for treating thoracic and, more recently, thoracolumbar scoliosis has been championed as a means to obviate the associated potential pulmonary compromise.8–12 Despite the universal popularity of posterior-only approaches for spinal deformities, the open anterior approach continues to be our surgical approach of choice for thoracolumbar scoliosis because of the potential for reduced blood loss and transfusions, excellent curve correction, dramatic improvement in clinical rib hump, and minimal impairment in pulmonary function. Although numerous previous reports on the efficacy of anterior spinal fusion (ASF) in the treatment of thoracolumbar scoliosis exist, no single study has simultaneously evaluated clinical outcomes including validated Scoliosis Research Society-22 patient questionnaire scores (SRS-22), radiographic, and pulmonary function outcomes in AIS.
In addition, most published series include multiple surgeons with little approach consistency.6–9 The purpose of this study is to report the radiographic, clinical, and pulmonary outcomes from a single surgeon for the thoracoabdominal approach to ASF in the treatment of thoracolumbar Copyright r 2010 by Lippincott Williams & Wilkins scoliosis. From the *Department of Orthopaedic Surgery, NYU Hospital for Joint Diseases, New York; and wDepartment of Orthopaedic Surgery, Bronx-Lebanon Hospital Center, Albert Einstein College of Medicine, Bronx, NY. This material has not been published or submitted for publication elsewhere. All the authors have read and provided approval for the content presented in the manuscript and for its submission to JPO. In consideration of JPO taking action in reviewing and editing my submission, the authors hereby transfer(s), assign(s), or otherwise convey(s) all copyright ownership to JPO in the event the article were to be published.
No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript. Reprints: Baron S. Lonner, MD, Department of Orthopaedic Surgery, NYU Hospital for Joint Diseases, 820 2nd Avenue, Suite 7A, New York, NY 10017. E-mail: firstname.lastname@example.org. ORIGINAL ARTICLE 664 | www.pedorthopaedics.com J Pediatr Orthop Volume 30, Number 7, October/November 2010 METHODS Patients After the institutional review board approval, we studied 31 consecutive thoracolumbar AIS patients (28 women and 3 men) who underwent single-rod (n = 3) or dual-rod (n = 28) anterior spinal instrumentation and fusion underwent surgery from 2000 to 2005. The average age of study participants was 15.3 ± 2.1 years (range: 11.6 to 21.6 y) with a mean major Cobb angle of 45 ± 6 degrees (range: 35 to 61 degrees). After approval from the Institutional Review Board Human Studies Committee, data were gathered retrospectively from a single-surgeon database.
Clinical, radiographic, SRS-22, and pulmonary function data was obtained at baseline preoperatively and 2 years postoperatively. Patients with prior spinal surgery or a medical diagnosis complicating pulmonary function were excluded from the study. According to the surgical classification of AIS by the Lenke et al13 system (Lenke 2001), all 31 patients in the study were of Lenke V classification with lumbar C modifier. Twenty-five patients had a normal thoracic sagittal modifier (T5-T12+10 to+40 degrees), 5 patients had a thoracic hyperkyphosis modifier (T5-T12>40 degrees), and 1 patient had a hypokyphosis modifier (T5-T12<40 degrees). Pulmonary Function Test All patients in this study had pulmonary function tests evaluating pulmonary volume and flow before surgery and at 2-years follow-up. Each measurement for all patients was taken using a digital spirometer (Renaissance II, Puritan Bennett, Boulder, CO) with the patient in the standing position by an experienced physician assistant. The measurements were performed in duplicate with the highest reading saved in the patient record.
Pulmonary function test results were performed as absolute values and as percent predictive values using height to approximate predictive pulmonary function. Preoperative and postoperative lung function was compared using forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1). These 2 parameters have been well established as adequate assessors of pulmonary function.7,8 Before surgery, complete pulmonary function evaluations were also performed by a licensed pulmonologist but were not used for this analysis to maintain measurement consistency. Radiographic Measurements Thoracolumbar/lumbar curve correction was determined by a radiographic comparison by a single researcher (coauthor) who was independent of the surgical team. The apical trunk rotation (ATR) was assessed using a manual scoliometer (Orthopedic Systems Inc, Union City, CA) by the operating surgeon. All measurements were taken at baseline and on 2-years follow-up.
Preoperative and postoperative coronal and sagittal plane measurements of the spinal curvature were made on 36 inch long cassette coronal and sagittal radiographs with the patient in the standing position. From the coronal radiographs, the main lumbar and thoracic curves were estimated using the Cobb method.14 End instrumented vertebral tilt angle and vertebral disc wedge angle were measured at the lower endplate of the most distal instrumented vertebrae. Instrumented segmental (IS) angle (Cobb angle of fusion) was measured in coronal and sagittal views from the upper endplate of the proximal instrumented vertebrae to the lower endplate of the distal instrumented vertebrae. From sagittal radiographs, kyphosis of the thoracic curve (thoracic kyphosis angle) was measured from the upper endplate of T2 to the lower endplate of T12. Proximal junctional kyphosis was measured from the lower endplate of the proximal instrumented vertebra to the upper endplate of the vertebrae above it. Similarly, distal junctional kyphosis was measured from the upper endplate of the distal instrumented level to the lower endplate one level below it. Lumbar lordosis was estimated from the upper endplate of T12 to the lower endplate of S1. Sagittal balance was accessed by determining the horizontal distance from the midpoint of C7 to the midpoint of the upper S1 endplate. Surgical Technique All 31 consecutive thoracolumbar AIS patients underwent single-rod (n= 3) or dual-rod (n= 28) anterior spinal instrumentation (stainless steel or titanium) and fusion via rib resection thoracotomy (ninth or tenth rib) with peripheral division of the diaphragm and a retroperitoneal approach to the lumbar spine. Dual-rod constructs offered improved stability and alleviated the need for postoperative bracing.
Bicortical screw placement was employed with less than one thread extending beyond the opposite cortex. In some cases, screw placement was just short of bicortical. In either case, screw placement did not require additional surgical exposure. A thoracoplasty was not performed. Thoracic spine exposure was performed by a single experienced thoracic surgeon. A structural graft (n= 3) or cage (n= 28) was used on the caudal 1 to 2 segments to create lordosis and increase distal construct stiffness. For one patient, only morcellized rib was used for bone grafting. Before wound closure, a chest tube was placed for wound drainage. Patients were gradually returned to full activity 3 months postoperatively. Clamshell Thoracolumbosacral Orthosis bracing was used for the first 3 months postoperatively in patients instrumented with single rod constructs. Statistics The distribution of variables at baseline and at 2-years follow-up was given as means, SDs, and ranges. P values were derived from a paired analysis using a student t test for comparison of individual variables within the study group.
A P value <0.05 signified the baseline and follow-up measurements were statistically significant. J Pediatr Orthop Volume 30, Number 7, October/November 2010 ASF for Thoracolumbar Curves r 2010 Lippincott Williams & Wilkins www.pedorthopaedics.com | 665 RESULTS Radiographic Changes Thoracolumbar/lumbar curve correction at 2 years averaged from 45 to 11 degrees (74%) (P<0.0001) and spontaneous correction of thoracic curves averaged from 26 to 15 degrees (42%) from baseline (P<0.0001) (Table 1, Fig. 1). On average, 3.7 vertebrae were fused to achieve adequate correction. IS coronal angle improved by 33.1 degrees (P<0.0001), whereas instrumented segment lordosis increased by 11 degrees (P<0.0001). End instrumented vertebral tilt angle decreased from 11 to 0 degrees (P= 0.01), whereas disc wedge angle increased from 1 to 4 degrees (P= 0.002). Sagittal balance changed from 5.7 to 2.1 cm (P<0.0001) and proximal junctional kyphosis changed from 3 to 6 degrees (P= 0.01). No significant changes were noted in T2 to T12 kyphosis, distal junctional kyphosis, T12-S1 lumbar lordosis, or coronal balance. The lowest instrumented vertebra was on average 0.2 levels cephalad to the distal Cobb end vertebra.
Operative and Clinical Data On average, operative time including surgical exposure was 219 minutes, estimated blood loss was 343 mL, and length of hospital stay was 5.3 days. From baseline to 2-years follow-up, ATR improved dramatically from 12 to 3 degrees for thoracolumbar curves (P<0.0001) (Table 2). For thoracic curves, ATR resolved spontaneously from 9 to 5 degrees (P = 0.13). Average SRS scores improved from 3.9 to 4.4 (P<0.0001). Specifically, SRS assessments of self-image and pain improved from 3.6 to 4.5 (P = 0.0002) and from 4.1 to 4.6 (P = 0.0092), respectively, and there were no significant differences in the activity and mental health domains. Pulmonary Function Tests There were no significant changes in pulmonary function from baseline to 2 years (Table 3). Accounting for changes in patient height, percent predicted FVC and FEV1 both remained unchanged from 91.3% to 87.2% (P = 0.97) and 89.9% to 85.4% (P= 0.40), respectively (Fig. 2). Overall, changes in pulmonary function seemed to occur within the normal range. A power analysis was performed to determine if our sample size was adequate to detect a clinically significant difference in pulmonary function from baseline to 2-years follow-up.
It was determined that a sample of 25 patients was sufficient to detect a difference in FEV1 or FVC as small as 0.4 L. This study of 31 patients is, therefore, powered adequately. Undetected differences in FVC and FEV1<0.4 L could also be related to patient or measurement variability. These small differences are also unlikely to be clinically significant (Wong et al and Kumano and Tsuyama). TABLE 1. Radiographic Data Preoperative (Degrees) 2-year Follow-up (Degrees) P Thoracolumbar/lumbar curve 45 11 <0.0001* Mean thoracic curve 26 15 <0.0001* Thoracic kyphosis T2-T12 38 37 0.73 Proximal junctional kyphosis 3 6 0.01* Distal junctional kyphosis 13 16 0.1 Lumbar lordosis T12-S1 64 61 0.26 Instrumented segmental lordosis 6 4 <0.0001* Instrumented segmental coronal 41 8 <0.0001* Disc wedge 1 4 0.002* End instrumented tilt 11 1 <0.0001* Sagittal balance 6 2 <0.0001* *Statistically significant. FIGURE 1. Mean TL/L and MT curve magnitude before surgery and on 2-years follow-up. Seventy-four percent and 42% curve correction was achieved for the TL/L and MT curves, respectively (P<0.0001). MT indicates mean thoracic; post-op, postoperative; Pre-op, preoperative; TL/L, thoracolumbar/lumbar. TABLE 2. Clinical and SRS-22 Data* Preoperative 2-year Follow-up P Apical trunk rotation (TL/L) (degrees) 12 3 <0.0001w Average SRS-22 Score 3.9 4.4 <0.0001w SRS activity score 4.3 4.5 0.0664 SRS pain score 4.1 4.6 0.0092w SRS self-image score 3.6 4.5 0.0002w SRS mental health score 4.1 4.2 0.30 SRS satisfaction score 4.6 wStatistically significant. *A lower SRS-22 score indicates a worse score.
SRS-22 indicates Scoliosis Research Society-22; TL/L, thoracolumbar/lumbar. TABLE 3. Pulmonary Function Data Preoperative 2-year Follow-up P Absolute FVC (L) 3.2 3.1 0.67 Absolute FEV1 (L) 2.8 2.7 0.67 Percent predicted FVC (%) 91.3 87.2 0.97 Percent predicted FEV1 (%) 89.9 85.4 0.4 FEV1 indicates forced expiratory volume in 1 second; FVC, forced vital capacity. Verma et al J Pediatr Orthop Volume 30, Number 7, October/November 2010 666 | www.pedorthopaedics.com r 2010 Lippincott Williams & Wilkins Complications Two patients in the study suffered intercostal neuralgia postthoracotomy, but did not require longterm pain management. No intraoperative complications, postoperative infections, nonunions, pseudoarthroses, or other complications were noted in any of the patients leading to an unplanned second operation or prolonged hospital stay. DISCUSSION ASF continues to be a viable treatment option for Lenke types I (main thoracic) and V (thoracolumbar) AIS deformities.1,2,12,15 Several purported advantages of ASF over posterior spinal fusion (PSF) include shorter fusions, reduced blood loss and transfusion requirements,16–18 ability to maintain thoracic kyphosis,2,19–21 improved spontaneous lumbar curve correction,22,23 and the ability to save on average 1 to 3 distal motion segments by fusing shorter than the Cobb end vertebra.2,18 ASF also facilitates kyphosis restoration without performing Ponte osteotomies, which may further increase blood losses from the PSF approach.10
Finally, overall SRS scores and patient satisfaction have been improved with ASF over PSF for AIS.16 Our group has also recently shown that for Lenke type I curves, the minimally invasive video-assisted thoracoscopic surgical patients scored higher in the self-image, mental health, and total domain scores, despite smaller curve corrections when compared with a matched group treated with PSF.18 This may be attributed, at least in part, to a smaller surgical incision and improved cosmetic result. In recent years, however, many surgeons have championed posterior segmental pedicle screw techniques over ASF even for large thoracic or thoracolumbar curves.8–12 However, the notion that all chest wall violations result in significant long-term pulmonary demise is not true. The curve type and magnitude, inclusion of a rib thoracoplasty, location of the thoracoplasty, thoracotomy versus thoracoabdominal approach, age, and preoperative lung performance all likely contribute. Despite the trend toward posterior-only approaches, we feel strongly that ASF offers several distinct advantages over PSF in the treatment of primary thoracolumbar spinal deformity, without detrimental long-term pulmonary sequelae.
Our results demonstrate that thoracolumbar scoliosis can be successfully treated with ASF performed via the thoracolumbar approach, without long-term pulmonary sequelae. We report excellent radiographic correction of the main thoracolumbar curve (average 74%), unfused thoracic curve (average 42%), clinical ATR, SRS scores, and maintenance of the sagittal profile. At 2-years follow-up, there were no significant differences in the percent predicted FEV1 and FVC, suggesting an overall maintenance of pulmonary function. Although studies have shown reductions in pulmonary function after ASF in the immediate postoperative period, pulmonary status at 2 years postoperatively has been inconsistently reported. Further, the clinical significance of the reported small declines, if any, remains to be seen with longer term follow up. Zhang et al6 and Newton et al17 both similarly reported modest pulmonary declines immediately after ASF that were most strongly related to preoperative pulmonary status rather than surgical approach. Kim et al8 (2008) recently reported a decline in absolute and percent predicted pulmonary function test scores at 2 years for patients treated with open thoracotomy, but no such decline was found in patients with thoracolumbar curves treated with a thoracoabdominal approach.
The authors provide several potential explanations for the lack of pulmonary impairment with the thoracoabdominal approach, including the more caudal position of the curve apex in thoracolumbar curves compared with midthoracic curves, thereby inducing less chest cage disruption, lower rib resections, fewer fusion levels, and greater coronal plane curve correction.8 Our results corroborate the above findings and provide evidence that the thoracoabdominal approach may mitigate pulmonary involvement compared with open thoracotomy.8 Graham et al24 also reported a return to baseline pulmonary function at 2-years follow-up after correction of AIS with ASF. The authors found a significant decline in absolute and percent predicted FEV1, total lung capacity, and FVC at 3 months follow-up, but all measures of pulmonary function returned to 97% of baseline at 2 years.24 Another recent study within our group demonstrated that among main thoracic curves, the addition of a thoracoplasty to the open anterior approach and to a lesser extent video-assisted thoracoscopic surgical approach led to a decline in pulmonary function at 2 years, whereas the thoracoabdominal approach for thoracolumbar curves did not cause this impairment.15 Further, Wong et al3 and Kumano and Tsuyama4 have argued that minor changes in percent predicted pulmonary function have little or no deleterious effects on an otherwise healthy patient. With respect to radiographic outcomes, both ASF and PSF are capable of achieving excellent curve FIGURE 2.
Accounting for changes in patient height, percent predicted forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1) both remained unchanged from 91.3% to 87.2% (P = 0.97) and 89.9% to 85.4% (P = 0.40), respectively. J Pediatr Orthop Volume 30, Number 7, October/November 2010 ASF for Thoracolumbar Curves r 2010 Lippincott Williams & Wilkins www.pedorthopaedics.com | 667 correction in both the coronal and sagittal planes. Geck et al25 recently compared posterior shortening pedicle screw-based constructs with ASF for thoracolumbar curves. Patients with AIS undergoing posterior pedicle fixation surgery demonstrated better coronal curve correction and shorter hospital stays than patients undergoing dual-rod ASF. Yet in a prior study, the same authors reported 80% thoracolumbar/lumbar Cobb angle improvement using posterior only techniques, which is comparable to this case series. In contrast, a recent prospective study by Wang et al26 reported comparable coronal curve correction for thoracolumbar/lumbar curves treated with either ASF or PSF.
However, ASF was beneficial over PSF in terms of reduced operative time, blood loss, transfusion rate, implant cost, overall hospital expense, and shorter fusion constructs. Our results also show a relative maintenance of sagittal kyphosis and lordosis, which matches that previously published for Lenke types 5, 3C, and 6 curves collectively.10 Of note, IS lordosis increased by 10.5 degrees, which is higher than typically reported for PSF. This case series is limited by the absence of a control group necessitating comparisons to historical controls. Although pulmonary function appears to be stable at 2 years postoperatively, the long-term sequelae are not fully known.24 Although patient compliance may be a limiting factor, future studies should include follow-up data beyond 2 years. Finally, our data does not account for confounding factors that may also affect pulmonary function during adolescence including sex, race, body mass index, smoking history, and age of peak growth.27 However, this limitation is shared by most retrospective studies and difficult to control for with a small population of patients treated by a single surgeon. In conclusion, the thoracoabdominal anterior approach for thoracolumbar scoliosis is able to facilitate excellent clinical and radiographic outcomes, minimal blood loss, powerful clinical ATR correction, relative maintenance of lordosis, relatively short fusion constructs, and improved SRS-22 scores, without detectable pulmonary function impairment at 2 years.
Although prior literature has recognized declines in pulmonary function with significant chest wall violation, recent literature including our current findings, suggest that the thoracoabdominal approach affects pulmonary function significantly less than that traditionally seen with open thoracotomy for thoracic scoliosis. ASF provides outstanding clinical and radiographic results comparable to posterior-only techniques without significant pulmonary impairment, and continues to be an efficacious option to the spinal deformity surgeon in the treatment of thoracolumbar scoliosis.
1. Dwyer AF, Schafer MF. Anterior approach to scoliosis. Results of treatment in fifty-one cases. J Bone Joint Surg Br. 1974;56:218–224.
2. Betz RR, Harms J, Clements DH, et al. Comparison of anterior and posterior instrumentation for correction of adolescent thoracic. Spine. 1999;24:225–239.
3. Wong CA, Cole AA, Watson L, et al. Pulmonary function before and after anterior spinal surgery in adult idiopathic scoliosis. Thorax. 1996;51:534–536.
4. Kumano K, Tsuyama N. Pulmonary function before and after surgical correction of scoliosis. J Bone Joint Surg Am. 1982;64: 242–248.
5. Chen SH, Huang TJ, Lee YY, et al. Pulmonary function after thoracoplasty in AIS. Clin Orthop Relat Res. 2002;399:152–161.
6. Zhang JG, Wang W, Qiv GX, et al. The role of preoperative pulmonary function tests in the surgical treatment of scoliosis. Spine. 2005;30:218–221.
7. Kim YJ, Lenke LG, Birdwell KH, et al. Pulmonary function in AIS relative to the surgical procedure. J Bone Joint Surg Am. 2005;87:1534–1541.
8. Kim YJ, Lenke LG, Birdwell KH, et al. Prospective pulmonary function comparison of anterior spinal fusion in adolescent idiopathic scoliosis: thoracotomy versus thoracoabdominal approach. Spine. 2008;33:1055–1060.
9. Di Silvestre M, Bakaloudis G, Lolli F, et al. Posterior fusion only for thoracic adolescent idiopathic scoliosis of more than 80 degrees: pedicle screws versus hybrid instrumentation. Eur Spine J. 2008;17: 1336–1349.
10. Shufflebarger HL, Geck MJ, Clark CE, et al. The posterior approach for lumbar and thoracolumbar adolescent idiopathic scoliosis: posterior shortening and pedicle screws. Spine. 2004; 29:269–276; discussion 276.
11. Luhmann SJ, Lenke LG, Kim YG, et al. Thoracic adolescent idiopathic scoliosis curves between 70 degrees and 100 degrees: is anterior release necessary? Spine. 2005;30:2061–2067.
12. Dobbs MB, Lenke LG, Kim YG, et al. Anterior/posterior spinal instrumentation versus posterior instrumentation alone for the treatment of adolescent idiopathic scoliotic curves more than 90 degrees. Spine. 2006;31:2386–2391.
13. Lenke LG, Betz RR, Harms J, et al. AIS: a new classification to determine extent of spinal arthrodesis. J Bone Joint Surg Am. 2001;83-A:1169–1181.
14. Cobb JR. Outline for the Study of Scoliosis. Instr Course Lectures. Vol 5. Ann Arbor, Mich: The American Academy of Orthopaedic Surgeons; 1948:261–275.
15. Lonner BS, Auerbach J, Estreicner MB, et al. Pulmonary function changes following various anterior approaches in the treatment of adolescent idiopathic scoliosis. Presentation, HIBBS Society meeting, Scoliosis Research Society 42nd Annual Meeting, Edinburgh, Scotland, September 4, 2007; JSDT. 2008:551–558.
16. Lonner BS, Kondrachov D, Siddiqi F, et al. Thoracoscopic spinal fusion compared with posterior spinal fusion for the treatment of thoracic adolescent idiopathic scoliosis. J Bone Joint Surg Am. 2006;88:1022–1034.
17. Newton PO, Perry A, Bastrom TP, et al. Predictors of change in postoperative pulmonary function in adolescent idiopathic scoliosis: a prospective study of 254 patients. Spine. 2007;32:1875–1882.
18. Lonner BS, Auerbach J, Estreicher M, et al. Video-assisted anterior thoracoscopic spinal fusion versus posterior spinal fusion: a comparative study utilizing the SRS-22 outcome instrument. Spine. 2009;34:193–198.
19. Vora V, Crawford A, Babekhir N, et al. A pedicle screw construct gives an enhanced posterior correction of adolescent idiopathic scoliosis when compared with other constructs. Spine. 2007;32: 1869–1874.
20. Lowenstein JE, Matsumoto H, Vitale MG, et al. Coronal and sagittal plane correction in adolescent idiopathic scoliosis: a comparison between all pedicle screw versus hybrid thoracic hook lumbar screw constructs. Spine. 2007;32:448–452.
21. Sucato DJ, Agrawal S, O’Brien MF, et al. Restoration of thoracic kyphosis after operative treatment of adolescent idiopathic scoliosis: a multicenter comparison of three surgical approaches. Spine. 2008;33:2630–2636.
22. Lenke LG, Betz RR, Birdwell KH, et al. Spontaneous lumbar curve coronal correction after selective anterior or posterior thoracic fusion in adolescent idiopathic scoliosis. Spine. 1999;24:1663–1671; discussion 1672. Verma et al J Pediatr Orthop Volume 30, Number 7, October/November 2010 668 | www.pedorthopaedics.com r 2010 Lippincott Williams & Wilkins
23. Potter BK, Kuklo TR, Lenke LG, et al. Radiographic outcomes of anterior spinal fusion versus posterior spinal fusion with thoracic pedicle screws for treatment of Lenke Type I AIS curves. Spine. 2005;30:1859–1866.
24. Graham EJ, Lenke LG, Lowe TG, et al. Prospective pulmonary function evaluation following open thoracotomy for anterior spinal fusion in adolescent idiopathic scoliosis. Spine. 2000;25:2319–2325.
25. Geck MJ, Rinella A, Hawthorne D, et al. Anterior dual rod versus posterior pedicle fixation surgery for the surgical treatment in Lenke 5C adolescent idiopathic scoliosis: a multicenter matched case analysis scoliosis research society (SRS) 43rd Annual Meeting. 2008.
26. Wang Y, Fei Q, Qiv G, et al. Anterior spinal fusions versus posterior spinal fusion for moderate lumbar/thoracolumbar adolescent idiopathic scoliosis: a prospective study. Spine. 2008;33:2166–2172.
27. Kishan S, Bastrom T, Betz RR, et al. Thoracoscopic scoliosis surgery affects pulmonary function less than thoracotomy at 2 years postsurgery. Spine. 2007;32:453–458.