Preoperative thoracic hypokyphosis is part of the typical presentation in patients with adolescent thoracic idiopathic scoliosis, and it is well described in the literature. 4,17 In a previous study comparing the results of anterior and posterior spinal fusions, the Harms Study Group 1 reported the advantage of anterior instrumentation’s ability to save an average of 2.5 distal fusion levels while obtaining coronal correction and balance equal to that obtained by posterior instrumentation. However, these authors noted excessive postoperative hyperkyphosis in 40% of the patients undergoing anterior spinal fusion (ASF) when the preoperative kyphosis was greater than 20°. They also observed that this was more prevalent in immature patients than in mature patients. In addition, the kyphosis appeared to increase over time from first erect radiograph to 2-year follow-up radiograph (Figure 1).
This type of postsurgical deformity is not well described. Discussions of crankshaft in young children after posterior spinal fusion for scoliosis usually focus on coronal and axial plane deformity. 6,9,12,14,16 Sagittal profile change after posterior spinal fusion has been described as a loss of disc height, which usually is balanced by increased vertebral body height resulting in no significant change in the kyphotic angle. Hallock et al 8 noted increased kyphosis after a Hibbs type of posterior spinal fusion for tuberculosis infection in 15 young children with an average long-term follow-up period of 21 years. They found that the normal vertebrae included in the fusion grew 23% less anteriorly and 36% less posteriorly than the adjacent unfused vertebrae. In such cases, increased kyphosis after surgery is significant. Connolly et al 3 demonstrated a relation between a lower spine score and increased thoracic kyphosis after long-term instrumentation extending to the lumbar spine. Because skeletal immaturity may be an indication for thoracic anterior instrumentation in patients requiring surgery, analysis of this kyphosis appears warranted.
This study proposed the following questions: 1) With progression of the kyphotic deformity after surgery, is there a difference in skeletally immature and mature patients? 2) Could these changes in the sagittal profile be the result of continued growth of the posterior elements after ASF?
A total of 109 patients who had idiopathic thoracic curves of King Types II to V underwent ASF using flexible 3.2-mm rod-screw instrumentation and bone grafting as part of the Harms-MOSS study. 1 For this retrospective study, 47 patients from that study group met the inclusion criteria requiring radiographic evidence of solid fusion by 2 years after surgery, no rod breakage, adequate anteroposterior and lateral radiographs available for review and measurement, and a minimum follow-up period of 2 years. Of the 62 patients excluded, 23 had inadequate follow-up evaluation; 11 were fused too short (less than the Cobb angle); and 14 had loss of correction 10° or more in the coronal plane, which may have given a false appearance to the sagittal profile. Two criteria for exclusion were met by 12 of these patients. There was no significant statistical difference between the numbers of patients deleted from each group.
At the time of surgery, the participants were divided into two groups: Risser Grade 0 (group 1) and Grades 1 to 5 (group 2). This was the only marker of skeletal maturity consistently available for this retrospective study. Risser 0 patients presumably would show greater growth potential than the higher Risser stage patients. Of the 47 patients, 10 (8 girls and 2 boys) were Risser 0, and 37 patients (28 girls and 9 boys) were Risser 1 to 5. The average age at surgery was 12.4 years for group 1 and 15.4 years for group 2. The preoperative age range was 9 to 15 years for the 10 patients in group 1 and 12 to 20 years for the 37 patients in group 2.
Standard radiographic measurements had been performed previously at each of the participating centers on preoperative, immediately postoperative, and latest follow-up radiographs. The coronal measurements were made according to the Cobb method, and sagittal measurements from T5 to T12 were used to determine the degree of kyphosis. In group 1, the average preoperative primary coronal curve was 60°, and the preoperative sagittal measurement from T5 to T12 averaged 11.8°. In group 2, the average preoperative primary coronal curve was 54.7°, and the preoperative sagittal measurement from T5 to T12 averaged 21.5°. The average follow-up period was 33.3 months (range, 24–48 months) for group 1 and 34.3 months (range, 24–69 months) for group 2.
Fusion levels were determined by including all vertebrae in the Cobb angle measurement of the primary curve, and proximal and distal compensatory curves had to reduce to 25° or less on bending radiographs. In both groups, an average of 7.6 levels per patient were fused. A single skin incision was made paralleling the seventh or eighth rib. When instrumentation of five to six levels was planned, a single thoracotomy was performed at the interspace proximal to the upper level of instrumentation (in this study 9 of 37 patients). For patients who required instrumentation of six or more vertebrae, a double thoracotomy usually was made through the single skin incision (in 28 of 37 patients). All disc and endplate material was removed, and rib heads were removed to facilitate disc exposure and screw placement. Thoracoplasty was performed as warranted by the rib prominence. Autologous rib graft was placed firmly into each clean disc space, except for the three apical disc spaces. Harms-MOSS anterior instrumentation was inserted. The threaded rod was flexible, and the sagittal contour was not determined by rod bending, but by the amount of rib graft placed into the disc spaces.
A posterior release consisting of subperiosteal dissection, excision of posterior ligaments, and facet joint capsules across segments to be fused anteriorly was performed in patients with a rigid curve (i.e., curve greater than 50° or curve demonstrating less than 50% correction on bending radiographs). The releases were believed to be necessary to increase flexibility for stiff curve correction.
Statistical analyses were performed using the StatMost statistical software package, version 3.5 (DataMost Software, Inc., Salt Lake City, UT). Data were analyzed for significance by means of the Mantel-Haenszel Χ2 formula. Statistical significance was set at P value less than 0.05.
The data were analyzed by group for preoperative primary coronal curve and sagittal contour, immediate and latest postoperative primary coronal and sagittal contour, and percentage of postoperative correction of both. The criteria used to denote significant postoperative progression from the immediate postoperative period to the latest follow-up visit was 10° or more of change in the sagittal contour (measured from T5 to T12). This figure was chosen to ensure that the risk of error because of measurement would be low. 2,8
For group 1 (Risser 0), the average immediate postoperative primary coronal curve was 18.7°, as compared with 20.6° at the latest follow-up visit. The average immediate postoperative sagittal (T5–T12) segment was 25.4°, as compared with 36.4° at the latest follow-up visit. The average number of levels fused was 7.6 per patient.
For group 2 (Risser 1 to 5), the average postoperative primary coronal curve was 18.1°, as compared with 21.8° at the latest follow-up visit. The immediate postoperative sagittal (T5–T12) segment was 29.5°, as compared with 34.2° at the latest follow-up evaluation. The average number of levels fused was 7.6 per patient.
Using the study criteria of 10° change for significant sagittal progression, 6 of 10 patients (60%) in group 1 had 10° or more of sagittal progression (average, 15°). Five patients progressed 10° to 19°, and one patient progressed 25° (Figure 1). The four patients who did not progress more than 10° averaged only 4° of sagittal progression. Of the 37 patients in group 2, 10 (27%) had sagittal progression of 10° or more (average, 15°). Nine patients progressed 10° to 20°, and one progressed 33°. The other 27 patients in group 2 did not progress more than 10°, averaging 1.1° of progression.
The Χ2 analysis of group 1 as compared with group 2 for significant sagittal progression showed a statistically significant difference with patients in Group 1 (Risser 0) having a higher incidence than group 2 (Risser 1 to 5) (P < 0.004). When the actual degrees of progression in each patient of the two groups were compared, the same statistical difference was again found (P = 0.02, Mann-Whitney test analysis).
To reduce the interobserver error in the use of the Risser classification, the group 2 patients were subdivided into Risser 1 to 2 and Risser 3 to 5. The group 1 (Risser 0) patients then were compared with the Subgroup 2 (Risser 3 to 5) patients. A Χ2 analysis showed that the Risser 0 patients were more likely to have statistically significant sagittal progression than the Risser 3 to 5 patients (P < 0.001).
The data were analyzed further by dividing patients according to chronological age and comparing them with their study group. Age values were selected according to the mode age (14.2 years) of the study patients: group 1 mean age was 12.4 and group 2 mean age was 15.4. The only consistent finding was that patients 14 years or older were less likely to show significant sagittal progression (P = 0.002).
The data also were cross-analyzed for the effect of preoperative coronal and preoperative sagittal deformities on the postoperative significant sagittal progression. No consistent statistically significant influence could be identified.
The effect of the posterior release was analyzed in both groups. According to Χ2 analysis, the group 1 patients who had a posterior release were more likely to have significant sagittal progression than group 2 (P = 0.01). Nine of 10 patients in group 1 had a posterior release, and 5 of these 9 (56%) had 10° or more of progression. The one patient who did not have a posterior release also progressed more than 10°. In group 2, 21 patients had a posterior release; 6 (29%) progressed 10° or more; and 4 of the 15 (27%) who did not have a posterior release progressed. This suggests that posterior release is not the cause of increasing kyphosis.
With an ASF, the anterior body growth is surgically arrested by removal of the cartilaginous endplates. If the anterior vertebral body growth is halted, then how can the thoracic segment contour continue to change? The posterior elements can continue to grow by apposition and at cartilaginous centers located in the spinous, transverse, and facet joint processes until the child matures. 6 The continued growth in the facet joints may have contributed to the higher rate of increased kyphosis in the immature patients of the Harms Study, 1 and this is one of the questions raised in this article.
The mechanisms of posterior element growth have been scantily reported in the literature. The neurocentral cartilage contributes to the growth of the vertebral body’s posterior aspect and to the three-dimensional growth of the pedicles. It also contributes to the cord canal diameter. 6,15 The time of actual closure is highly varied from the age 5 years to the late teens. 10,15 Growth of the laminae, transverse processes, and facet joints occur through enchondral ossification. Periosteal appositional growth contributes to the length of the lamina and articular processes until a child is fully mature. 7,10,11 Roaf 13 believed that in the thoracic segment, the posterior elements grow faster than the anterior elements, and that the opposite occurs in the cervical and lumbar segments. A study by Lord et al 11 found that the contributions of the anterior and posterior elements varied by location and age. The longitudinal growth of the T1–S1 spinal segment is 0.9 cm annually from age 5 to 10 years, but increases to 1.8 cm per year during the adolescent growth spurt. 5 It is not unreasonable to think that a skeletally immature child who had an ASF might develop a secondary kyphotic spine deformity. Further work is needed to determine the etiology of this postoperative spinal deformity.
Skeletally immature patients with a posterior release to increase curve flexibility appeared to have a statistically significantly higher risk for increased thoracic kyphosis after surgery as compared with the mature patients in this study. The etiology of this effect is unknown. Facet joint debridement and removal of cartilage should induce fusion and should prevent kyphosis progression. Perhaps the removal of ligamentous supporting structures at the facet joints and interspinous ligaments allows the development of hyperkyphosis. In addition, overgrowth may occur similarly to limb overgrowth after a femur fracture in a child. However, posterior release as a sole factor causing the kyphosis likely is not related due to the fact that equal kyphotic deformity occurred in group 2 (the mature patients) whether or not they underwent posterior release.
In conclusion, some patients with thoracic adolescent idiopathic scoliosis treated with anterior instrumentation may be at risk for progressive sagittal kyphosis with growth. The question remains as to whether this is secondary to posterior overgrowth or anterior subsidence. Risk factors include Risser 0 or evidence of significant skeletal immaturity. Posterior release may be an additional factor in these immature patients. Preserving the sagittal profile with intervertebral spacers, rigid rods, and bone graft to allow for an average anticipated increase of 15° of kyphosis with growth may be appropriate.
- • Some patients with thoracic adolescent idiopathic scoliosis treated with anterior instrumentation may be at risk for progressive sagittal kyphosis secondary to growth.
- • Skeletal immaturity (Risser 0) appears to be a risk factor.
- • In these immature patients, preserving the sagittal profile to allow for an average 15° increase of kyphosis with growth may be appropriate.
The authors thank Gail Huss, RN, and Kathy Blanke, RN, for data collection and analysis. They also thank the DePuy-AcroMed Corporation for an educational grant for research.
1. Betz RR, Harms J, Clements DH, et al. Comparison of anterior versus
posterior instrumentation for correction of adolescent thoracic idiopathic scoliosis
. Spine 1999; 24:225–39.
2. Carman DL, Browne RH, Birch JG. Measurement of scoliosis and kyphosis radiographs: Intraobserver and interobserver variation. J Bone Joint Surg [Am] 1990; 72:328–33.
3. Connolly PJ, Von Schroeder HP, Johnson GE, Kostuik JP. Adolescent idiopathic scoliosis
: Long-term effect of instrumentation extending to the lumbar spine. J Bone Joint Surg [Am] 1995; 77:1210–16.
4. Dickson RA, Lawton JO, Archer IA, Butt WP. The pathogenesis of idiopathic scoliosis
. J Bone Joint Surg [Br] 1984; 66:8–15.
5. DiMeglio A. Growth of the spine before age 5 years. J Pediatr Orthop B 1993; 1:102–7.
6. Dubousset J, Herring JA, Shufflebarger H. The crankshaft phenomenon. J Pediatr Orthop 1989; 9:541–50.
7. Dubousset J, Katti E, Seringe R. Epiphysiodesis of the spine in young children for congenital spinal deformations. J Pediatr Orthop B 1993; 1:123–40.
8. Hallock H, Francis KC, Jones JB. Spine fusion in young children: A long-term end-result study with particular reference to growth effects. J Bone Joint Surg [Am] 1957; 39:481–91.
9. Hefti FL, McMaster MJ. The effect of the adolescent growth spurt on early posterior spinal fusions in infantile and juvenile idiopathic scoliosis
. J Bone Joint Surg [Br] 1983; 65:247–54.
10. Knutsson F. Growth and differentiation of the postnatal vertebra. Acta Radiol 1961; 55:401–8.
11. Lord MJ, Ogden JA, Ganey TM. Postnatal development of the thoracic spine. Spine 1995; 20:1692–8.
12. Mullaji AB, Upadhyay SS, Leong JC. Vertebral growth after posterior spinal fusion for idiopathic scoliosis
in skeletally immature adolescents. J Bone Joint Surg [Br] 1994; 76:870–6.
13. Roaf R. The treatment of progressive scoliosis by unilateral growth arrest. J Bone Joint Surg [Br] 1963; 45:637–51.
14. Sanders JO, Herring JA, Browne RH. Posterior arthrodesis and instrumentation in the immature (Risser Grade 0) spine in idiopathic scoliosis
. J Bone Joint Surg [Am] 1995; 77:39–45.
15. Vital JM, Beguiristain JL, Algara C, et al. The neurocentral vertebral cartilage: Anatomy, physiology, and physiopathology. Surg Radiol Anat 1989; 11:323–8.
16. Winter RB, Moe JH. The results of spinal arthrodesis for congenital spinal deformity in patients younger than five years old. J Bone Joint Surg [Am] 1982; 64:419–32.
17. Xiong B, Sevastik J, Hedlund R, Sevastik B. Sagittal configuration of the spine and growth of the posterior elements in early scoliosis. J Orthop Res 1994; 12:113–18.
Point of View
John A. Ogden MD
Director of Orthopaedics
Atlanta Medical Center
Clinical Professor, Orthopaedics
D’Andrea et al have presented essential data showing that anterior fusion in the absence of posterior instrumentation and fusion in young (i.e., skeletally immature) patients may lead to the development of a phenomenon comparable to the crankshaft phenomenon when surgery is restricted to the posterior elements. The “unexpected” development of the posterior postoperative deformities reflects our incomplete understanding of growth in the spine.
The portion of the spine that grows in a manner most analogous to a long bone is through the growth plates that are situated on the superior and inferior portions of the vertebral body. However, growth also occurs through the neurocentral synchondroses. This latter growth defines the size of the canal and probably plays a role in the development of narrowing or stenosis that is present in certain skeletal dysplasias. However, this region of growth essentially has ceased to function by the time most patients undergo anterior scoliosis fusion procedures. Accordingly, it probably played a small role in the development of the deformity.
Perhaps the least understood aspect of vertebral growth relates to the presence of a small growth plate associated with each of the facets and growth that occurs in each spinous process. 1 The facets undergo progressive reorientation as the child grows and obviously are not oriented in a normal manner at many levels in children with scoliosis. This is readily evident in analyzing computed tomography or magnetic resonance imaging data. These facet growth plates are essential to keep pace with growth in the anterior column through the vertebral end plates. After procedures in which facet capsules are released and articular cartilage is removed, it is feasible that the adjacent growing portions are left undisturbed and capable of continued growth. The spinous processes elongate through growth cartilage at the very end. However, they also grow appositionally through the periosteum. Continued growth longitudinally and latitudinally could also have an impact on the development of a kyphotic deformity. It is certainly feasible that there may be a hyperemic growth response to surgery in these posterior column regions, similar to that occurring in immature long bones after a fracture. Abnormal facet orientation and spinous process growth could result in kyphotic deformity.
It is extremely important that we continue to study the nuances of growth of the vertebra at different levels. The presence of multiple regions of growth potential in each of the three columns of the spine creates a situation that perhaps encourages postoperative deformity even when well-planned and implemented operative interventions are used to control deformity. Future research, particularly that involving postmortem material, is essential to furthering our knowledge of the variations in these growth plates in our patients. Comparable histologic analyses in animal models have little relevance because of variations in morphology of the vertebra and variations in the overall anatomy. In particular, most animals develop significantly sized secondary ossification centers within the superior and inferior end plates of the vertebral body. Secondary ossification centers also develop adjacent to the apposition of ribs into the margins of the vertebral bodies, near the neurocentral synchondroses. 1 Interestingly, this latter observation may also be an unknown factor in the development of deformities after surgery in humans. Perhaps an ideal animal model for postnatal vertebral development might be obtained through studies of large primates such as chimpanzees and baboons. Such material is limited, but in centers that have large numbers of primates that are often killed at various stages of growth, incidentally obtaining the skeletal elements offers a unique opportunity to learn more about developmental patterns.
1. Ogden JA, Ganey TM. Development and maturation of the axial skeleton. In Weinstein S, ed. The Pediatric Spine: Principles and Practice. 2d ed. NY, Raven Press, 2000.