Since the 1950s, when Moe wrote his classic article defining curve patterns in adolescent idiopathic scoliosis (AIS) and their proposed treatment, controversy has continued over the treatment of the compensatory lumbar curve. Moe defined Type II curves as curves in which the nonstructural lumbar curve was smaller and more flexible than the thoracic curve. This type is distinct from the Type I, or double major curves, in which both the thoracic and lumbar curves are structural and therefore require fusion. In contrast, Moe suggested that the lumbar curve in a Type II curve pattern is compensatory and would undergo spontaneous correction with selective fusion of just the thoracic curve. 1,2
Subsequent studies that focused on the treatment of this Type II or “false double major” curve have had varying conclusions about fusing the lumbar curve. Some agree with Moe’s guidelines, whereas others have put forth limitations of curve magnitude, stating that a lumbar curve greater than 40° to 45° should be fused regardless of flexibility. 3–6 Lenke et al reported that the King-Moe definition of a Type II curve was not sufficient to recommend fusion of the thoracic curve alone and emphasized the relative differences of the thoracic and lumbar severity as guidelines in the decision to fuse the lumbar curve. They proposed that to fuse selectively in a false double major curve, the thoracic curve should be at least 20% bigger, have at least as much apical vertebral rotation, and have 20% more apical displacement than the minor lumbar curve. 7
The Lenke classification system is a more recently developed treatment-based classification that defines curve patterns by region, magnitude, and flexibility as well as sagittal profile and displacement of the lumbar apex from the central sacral vertical line. The system defines many of these false double major curves as Lenke Type 1B and 1C curves. In this curve pattern, the thoracic curve is the largest curve. The smaller lumbar curve is nonstructural (i.e., has a side-bending Cobb measurement of 25° or less) and has thoracolumbar kyphosis that is less than +20°. The system further classifies these curve patterns by the degree of apical displacement of the lumbar apex (A, B, or C) (Figure 1). 8–10
The treatment algorithm for Lenke Type 1 curve patterns is the same as Moe’s—in general, fuse only the thoracic curve. In clinical practice, however, some patients with Type 1B and 1C curve patterns do undergo fusion of both curves. In a recent report by Lenke et al, selective thoracic fusion was performed in 62% of patients with 1C curves. 8 This study focuses on the controversy surrounding the Lenke Type 1B and 1C “rule breakers” in which both the thoracic and lumbar curves are fused.
The purpose of this study was to: 1) identify the frequency of fusing the lumbar curve in Lenke Type 1B and 1C curve patterns for five sites of the DePuy AcroMed Harms Study Group (DAHSG); and 2) determine differences in the preoperative radiographic features of the curves fused selectively (thoracic curve only) versus those fused nonselectively (both thoracic and lumbar curves).
Methods and Materials
This was a retrospective evaluation of a subset of patients from the DAHSG database. Patients from five centers with either a Lenke Type 1B or 1C curve comprised the study cohort. Preoperative posteroanterior (PA) and side-bending radiographs were evaluated to classify the patients’ curve patterns according to the Lenke classification system. Curve patterns were defined as Lenke Type 1 if they had a major thoracic curve that was the largest curve and a minor lumbar curve that reduced to less than 25° on side-bending radiographs. The lumbar modifier B was applied to curve patterns in which the CSVL passed between the medial border of the concave pedicle and the concave edge of the lumbar apical vertebra. Curve patterns were given the lumbar modifier C if the CSVL passed completely medial to and did not touch the lumbar apical vertebra. None of the patients had a thoracolumbar kyphosis greater than 20°.
Postoperative radiographs were reviewed to identify the most distal level of instrumentation/fusion. Two groups of patients were defined depending on the most distal level of fusion. The “selective thoracic fusion” group included patients who underwent fusion distally to the L1 vertebra or above. The “nonselective fusion” group included patients with fusions of both curves, to the L2 vertebra or below. These data were used to calculate the incidence of fusion of the lumbar curve at each of the surgical centers.
Preoperative PA and side-bending radiographs were evaluated for radiographic parameters suggested by previous studies to be predictive in the decision to fuse the lumbar curve. These included: lumbar curve magnitude, side-bending lumbar curve magnitude, percentage lumbar curve flexibility, lumbar apical vertebral displacement from CSVL, the ratio of thoracic to lumbar curve magnitude, and the ratio of thoracic to lumbar apical vertebral displacement, as well as thoracic and thoracolumbar kyphosis. These radiographic parameters for both the selective and nonselective groups were compared by multivariate analysis of variance to evaluate which variables were associated with fusion of the lumbar curve. Statistical analysis was conducted using SPSS software (Chicago, IL) with significance set at α = 0.01 after a Bonferroni correction for multiple tests.
To evaluate differences in significant factors between patient care sites, individual site-specific analyses were conducted. Patients at each site were divided into selective and nonselective groups and compared to determine which radiographic factors were associated with nonselective fusion at that particular site.
This study included 203 patients with Lenke 1B or 1C curve patterns. There were 180 females and 23 males with an average age at the time of surgery of 14.6 ± 2.0 years with a range from 10.3 to 20.8 years. In this group, 150 patients underwent anterior spinal instrumentation and fusion, whereas 53 underwent posterior spinal instrumentation and fusion. Of the 203 patients, 126 (62%) had Lenke Type 1B curves and 77 (38%) had Lenke Type 1C curves.
This group was divided into two groups based on the most distally fused vertebra. The selective fusion group included 168 patients (83% of the total) who underwent fusion to the L1 vertebra or above. There were 150 females and 18 males with an average age of 14.6 ± 2.0 years. Of this selective fusion group, 145 (86%) had anterior instrumentation and 23 (14%) underwent posterior instrumentation. Sixty-nine percent of these patients had 1B Type curves, whereas 31% had Type 1C curves.
The nonselective fusion group included 35 patients (17% of the total) who underwent fusion to the L2 vertebra or below. There were 30 females and 5 males with an average age of 14.8 ± 1.9 years. In this group, 20% of patients underwent anterior instrumentation and 80% underwent posterior instrumentation. Twenty-nine percent of patients within the nonselective group had Type 1B curves and 71% had Type 1C curves.
The preoperative radiographic data of the selective and nonselective groups is summarized in Table 1. Patients who underwent nonselective fusion had larger preoperative lumbar curves (42 ± 10°vs. 37 ± 7°, P = 0.003) than selective fusion patients. There was a trend toward larger lumbar curves on side-bending radiographs (13 ± 8°vs. 10 ± 8°, P = 0.02) in nonselective patients. In addition, those patients who had both curves fused had greater lumbar apical displacement than those who had only the thoracic curve fused (3.1 ± 1.4 cm vs. 2.2 ± 0.8 cm, P < 0.001). The distribution of 1B and 1C curves within the two groups further supported the above findings with 32% of 1C curves being fused below L1, whereas only 8% of 1B curves had the lumbar curve included in the fusion (Figure 2). There were also smaller thoracic to lumbar curve magnitude ratios (1.31 ± 0.29 vs. 1.44 ± 0.30, P = 0.01) for the nonselective fusion group. There was no significant difference in the thoracic and thoracolumbar kyphosis between groups.
Site-specific analysis revealed different patterns of treatment at each scoliosis center. The incidence of fusion of the lumbar curves in Lenke Type 1B or 1C curves ranged from 6% to 33%. This variation was much less for the Type 1B curves, and the selective fusion rate in this subset ranged from 3% to 17%. This is in contrast to the larger discrepancy for the Type 1C curves. In these cases with the larger lumbar apical vertebral deviation, the frequency of applying selective fusion ranged from 6% to 67%. Furthermore, the characteristics that seemed to be determining factors for fusion of the lumbar curve differed from site to site. At three of the five sites, lumbar apical displacement was significantly increased in patients who had their lumbar curve fused. Significantly larger lumbar curve magnitudes were associated with lumbar fusion at site 3 (Table 2).
The fate of the lumbar spine in patients with scoliosis is to some degree dictated by the nature of the scoliosis, but is also dependent on the treatment chosen by the surgeon. Sparing the lumbar spine from fusion should be a goal whenever practical. Lumbar motion is important for function during the decades of life these adolescent patients have remaining. There is reason to believe that distal degeneration will be less problematic if more motion segments remain below a fusion. 11 Preserving motion is certainly a goal, but it must be recognized that this flies in the face of all our current fusion-based scoliosis correction techniques. Thus, a balance between motion and deformity correction must be struck for every surgically treated patient with scoliosis (Figure 3).
Certain curve patterns leave little doubt about the ability to preserve lumbar motion. Those curves with only a thoracic deformity (King-Moe III, V and Lenke 1A, 2A) have long been treated with the distal level of fusion generally a level or two proximal to the stable vertebra. This rarely results in a fusion distal to L2. On the other hand, there are curve patterns that routinely require fusion to L3 or L4. When the primary curve is in the lumbar or thoracolumbar region, lumbar fusion is unavoidable if surgical correction is undertaken. The debate to include the lumbar spine in the fusion has always focused on those curves with both a thoracic and lumbar component. The original recommendation of Moe for selective thoracic fusion of the King-Moe II curve from the Harrington instrumentation era has since been practiced by some and modified by others. 3–5,12–14 Following the introduction of more powerful instrumentation systems, coronal decompensation with postoperative trunk shift to the left was recognized as a risk if selective fusion was performed in some King-Moe II curves. One of the primary methods of limiting the risk of lumbar decompensation seems to be limiting the degree of thoracic curve correction. 15–17 Thus, the two options for many of these King-Moe II/Lenke 1B, 1C curves are: 1) selective thoracic fusion limiting the degree of direct correction of the thoracic curve to maintain coronal balance; or 2) extend the fusion to include the lumbar curve achieving maximal correction of both curves. Although not the purpose of this analysis, the postoperative curve magnitudes for both the thoracic and lumbar curves were larger in the patients treated by selective thoracic fusion compared to those in which both curves were instrumented (thoracic postoperative curves: selective, 21 ± 8°vs. nonselective, 17 ± 9°, P = 0.009; lumbar postoperative curves: selective, 23 ± 8 vs. nonselective, 15 ± 8, P < 0.001).
Utilizing the DAHSG database, the authors have been able to demonstrate substantial variation in their approach to the Lenke 1B and 1C curves. By definition, in all of these cases, there was a lumbar curve, but it was smaller in magnitude than the thoracic curve. Additionally, the flexibility of the lumbar curve was such that on side bending, it was reduced to less than 25°. Overall, for these curve patterns, the study group highly favored selective fusion of only the thoracic component, done so for 83% (168 of 203) of the cases. However, the differences in this rate between the five centers varied from 67% to 94%.
Identifying the preoperative radiographic features of the cases that were treated selectively compared to those fused into the lumbar curve was the second goal of the analysis. The Lenke classification lumbar curve modifier, “B” compared to “C,” was an important determinant of ultimate treatment. Selective thoracic fusion was more likely (92%) if the CSVL intersected some part of the lumbar apical vertebra (B modifier) compared to those with complete deviation of the lumbar curve from CSVL (C modifier) (68%). This is similar to the subclassification of King-Moe II curves based on the position of the CSVL relative to the lumbar apex proposed by Ibrahim et al. They and others have suggested King-Moe IIA curves (CSVL intersects lumbar apex) would be amenable to selective fusion compared to King-Moe IIB curves (CSVL misses lumbar apex) that require inclusion of the lumbar curve. 18,19 However, there must be other important variables, as the current authors resorted to lumbar fusion in only 32% of the Lenke IC (or King-Moe IIB) curves. Again, this rate of selective fusion of Lenke IC lumbar curves varied between 6% and 67% for the 5 centers included in this series.
The related radiographic measure of absolute magnitude of the lumbar apical vertebral deviation from the CSVL was also significantly greater for the patients who were fused into the lumbar curve. This variable was different between the selective and nonselective fusion groups at three of the five centers, suggesting this to be the most agreed on radiographic determinant for choosing the extent of fusion. The apical lumbar deviation was 2.2 ± 0.8 cm in the selective fusion group compared to 3.1 ± 1.4 cm for the nonselective group. Despite this difference, there did not appear to be a critical value that would predict surgeon decision-making. The range of lumbar apical displacement varied between 0.3 cm and 5.0 cm for both groups (Figure 4).
Another lumbar curve characteristic suggested in the literature as an important determinant of when to include the lumbar curve is the Cobb angle magnitude of the lumbar curve. The upper “limit” for successful selective fusion has been suggested to be as low as 40° to 45°. 3,4 Lumbar curve magnitude was significantly different between the two groups (37 ± 7° for selective fusions vs. 42 ± 10° for nonselective fusions); however, the mean difference in the lumbar Cobb angle between the two groups was only 5°. The range for the lumbar curve magnitude was as high as 61° in the selective fusion cases (Figure 5). Therefore, many lumbar curves greater than 40° to 45° have been treated with selective thoracic fusion by the study group members. The long-term results are not known, but it appears that at least some of the authors are comfortable in some cases exceeding this 40° to 45° “limit” for the lumbar curve when choosing a selective fusion for their patients.
In addition to the specific features of the lumbar curve (e.g., magnitude and apical deviation), the relative characteristics compared to the thoracic curve have also been suggested as criteria for choosing a short versus long instrumentation approach. Lenke et al have proposed consideration for selective fusion when the thoracic curve is 20% larger and with 20% greater apical vertebral displacement compared to the lumbar curve. 7 The ratio of thoracic to lumbar curve magnitude was significantly different in this series of patients; however, the difference in distribution patterns between the selective and nonselective groups was relatively small (Figure 6).
Thoracolumbar kyphosis before surgery may also drive a surgeon to perform a more distal fusion in patients who might otherwise have been candidates for selective fusion. The Lenke classification criterion states that if the T10–L2 kyphosis measures >20°, the thoracolumbar/lumbar region is considered “structural” and fusion is suggested across these levels. The 20° cutoff may be too high, at least in the minds of some, and may be a partial explanation for extending the fusion distally in some of these patients. In this series of patients, there was no significant difference in thoracolumbar kyphosis between the two groups and the percentage of patients with thoracolumbar kyphosis >10° was similar (9% in the selective group vs. 11% in the nonselective group). Global thoracic hyperkyphosis might also require correction by instrumentation distally to L2 or L3 in some cases. Although there was no significant difference in T5–T12 kyphosis between the two groups, 14% of patients in the nonselective group had kyphosis greater than 40° as compared to 8% of patients in the selective group. The sagittal plane deserves equal attention to the coronal plane, and the distal level of fusion must be appropriate for the deformity in both planes.
Relative axial rotation of the thoracic and lumbar regions has not been included in this analysis; yet it may be one of the most important determinants in choosing to include the lumbar curve in the fusion. This is difficult to assess radiographically (Nash-Moe or Perdriolle methods) with questionable accuracy. 20–22 Clinical evaluation of the patient during a forward bend seems more helpful than the radiographs in making this assessment. A selective fusion seems inappropriate if the lumbar rotational prominence is dominant compared to the thoracic region.
This analysis does not allow assessment of whether the “right” choice (selective vs. nonselective fusion) was made for a given patient, but it does provide insight into which radiographic features the surgeons may have valued in making the choice. The decision may or may not have been done consciously based on analysis of the radiographic features of the deformity, and the actual thought process of the surgeon in each case is not known. These approaches have very different short-term and long-term risks. If the choice to perform selective fusion is “wrong,” and decompensation occurs, the need for early revision by extending the fusion or reducing the correction is difficult for the patient and surgeon alike. The nonselective approach rarely leads to early troubles that require a second procedure and is often perceived as being “safer.” Though this approach may be safer in the short-term, it may be more difficult in the long-term as distal degeneration is more likely. This raises the question: is it “better” to be “safe” in the short-term (with a long fusion) or take a chance at avoiding later degenerative problems (with a shorter motion sparing fusion)? Debate will continue regarding fusion of the lumbar curve in the King-Moe II, Lenke 1B and 1C cases largely because the long-term outcomes of the two surgical options are unknown. Opinion and philosophy dominate in the face of this limited understanding. This study does, however, confirm the substantial variation in the frequency of lumbar fusion especially for the 1C curves. This occurred despite participation of these surgeons for many years in a study group where selective thoracic fusion has been an important concept and goal. The concept of selective fusion was nearly always practiced by the group for the Lenke 1B curves and, in this subset, selective thoracic fusion (if the sagittal plane allows) was the standard. The majority of the 1C curves also had selective fusion, though surgeons varied substantially in their use of this approach for these more apically deviated lumbar curves.
- Members of the DePuy-AcroMed Harms Study Group performed selective thoracic fusions for the vast majority of (83%) Lenke 1B and 1C curves.
- Substantial variability in the rate of selective fusion existed within the Study Group members, especially for the Lenke 1C curve pattern (6% to 67%).
- Selective fusion should be “considered” for all Lenke 1B and 1C curves, although variables beyond the classification system may ultimately lead one to perform a longer fusion.
1. King HA, Moe JH, Bradford DS, et al. The selection of fusion levels in thoracic idiopathic scoliosis. J Bone Joint Surg Am 1983; 65: 1302–13.
2. Moe J. A critical analysis of methods of fusion for scoliosis. J Bone Joint Surg Am 1958; 40A: 529–54.
3. McCall RE, Bronson W. Criteria for selective fusion in idiopathic scoliosis using Cotrel-Dubousset instrumentation. J Pediatr Orthop 1992; 12: 475–9.
4. Richards BS. Lumbar curve
response in type II idiopathic scoliosis after posterior instrumentation of the thoracic curve. Spine 1992; 17: S282–6.
5. Kalen V, Conklin M. The behavior of the unfused lumbar curve
following selective thoracic fusion for idiopathic scoliosis. Spine 1990; 15: 271–4.
6. McCance SE, Denis F, Lonstein JE, et al. Coronal and sagittal balance in surgically treated adolescent idiopathic scoliosis
with the King II curve pattern. A review of 67 consecutive cases having selective thoracic arthrodesis. Spine 1998; 23: 2063–73.
7. Lenke LG, Bridwell KH, Baldus C, et al. Preventing decompensation in King type II curves treated with Cotrel-Dubousset instrumentation. Strict guidelines for selective thoracic fusion. Spine 1992; 17: S274–81.
8. Lenke LG, Betz RR, Clements D, et al. Curve prevalence of a new classification of operative adolescent idiopathic scoliosis
: does classification correlate with treatment? Spine 2002; 27: 604–11.
9. Lenke LG, Betz RR, Haher TR, et al. Multisurgeon assessment of surgical decision-making in adolescent idiopathic scoliosis
: curve classification, operative approach, and fusion levels. Spine 2001; 26: 2347–53.
10. Lenke LG, Betz RR, Harms J, et al. Adolescent idiopathic scoliosis
: a new classification to determine extent of spinal arthrodesis. J Bone Joint Surg Am 2001; 83A: 1169–81.
11. Large DF, Doig WG, Dickens DR, et al. Surgical treatment of double major scoliosis. Improvement of the lumbar curve
after fusion of the thoracic curve. J Bone Joint Surg Br 1991; 73: 121–4.
12. Knapp DR Jr, Price CT, Jones ET, et al. Choosing fusion levels in progressive thoracic idiopathic scoliosis. Spine 1992; 17: 1159–65.
13. Lenke LG, Betz RR, Bridwell KH, et al. Spontaneous lumbar curve
coronal correction after selective anterior or posterior thoracic fusion in adolescent idiopathic scoliosis
1999;24:1663–71; discussion 72.
14. van Rhijn LW, Plasmans CM, Veraart BE. No relationship exists between the correction of the thoracic and the lumbar curves after selective thoracic fusion for adolescent idiopathic scoliosis
King type II. Eur Spine J 2002; 11: 550–5.
15. Patwardhan AG, Rimkus A, Gavin TM, et al. Geometric analysis of coronal decompensation in idiopathic scoliosis. Spine 1996; 21: 1192–200.
16. Shufflebarger HL, Clark CE. Fusion levels and hook patterns in thoracic scoliosis with Cotrel-Dubousset instrumentation. Spine 1990; 15: 916–20.
17. Arlet V, Marchesi D, Papin P, et al. Decompensation following scoliosis surgery: treatment by decreasing the correction of the main thoracic curve or “letting the spine go”. Eur Spine J 2000; 9: 156–60.
18. Burton DC, Asher MA, Lai SM. The selection of fusion levels using torsional correction techniques in the surgical treatment of idiopathic scoliosis. Spine 1999; 24: 1728–39.
19. Ibrahim K, et al. Cotrel-Dubousset instrumentation for double major right thoracic left lumbar scoliosis: the relation between frontal balance, hook configuration, and fusion levels [abstract]. Orthop Trans 1991; 15: 114.
20. Nash CL Jr, Moe JH. A study of vertebral rotation. J Bone Joint Surg Am 1969; 51: 223–9.
21. Richards BS. Measurement error in assessment of vertebral rotation using the Perdriolle torsionmeter. Spine 1992; 17: 513–7.
22. Ho EK, Upadhyay SS, Ferris L, et al. A comparative study of computed tomographic and plain radiographic methods to measure vertebral rotation in adolescent idiopathic scoliosis
. Spine 1992; 17: 771–4.
Keywords:© 2003 Lippincott Williams & Wilkins, Inc.
adolescent idiopathic scoliosis; Lenke classification; lumbar curve; fusion ] Spine 2003;28:S217–S223