Achieving proper sagittal plane balance is increasingly recognized as critical to the surgical management of adolescent idiopathic scoliosis. Although the emphasis in the past was on coronal plane correction, it has become evident that the long-term health of the spine may be related more to the status of the sagittal plane. Several studies with long-term follow-up demonstrate that the spine is not well equipped to handle sagittal plane imperfections. For example, segmentally flat lumbar fusions and distraction forces in the lumbar spine are associated with the development of fixed sagittal imbalance syndromes. 4,6 Also, stopping a fusion at the apex of a kyphosis is associated with the development of junctional problems. Based on these observations, establishing a harmonious sagittal contour seems at least as important as achieving well-balanced coronal curves.
Currently, anterior or posterior instrumentation can be used in the treatment of selected idiopathic curves. Proponents of anterior surgery cite the ability to save distal fusion levels, achieve greater correction, and avoid spinal extensor muscle disruption. 2,3,8 Advocates of posterior fusion note its proven reliability, avoidance of chest cage disruption, and ability to better control shoulder imbalance. 3 However, despite recognition of the importance of the sagittal plane on the long-term prognosis of spinal fusions, few studies have compared the effect of anterior versus posterior instrumentation on sagittal profiles. Given the importance of sagittal plane alignment to the structural health and prognosis of the spine, we sought to compare the effect of anterior versus posterior instrumentation on sagittal plane parameters in adolescent idiopathic scoliosis.
A total of 110 consecutive patients (11–21 years of age) with adolescent idiopathic scoliosis who underwent either anterior or posterior spinal instrumentation and fusion between 1996 and 1998 were included in the study. Patients with all curve patterns were included in the study. All surgeries were performed by one of two attending spinal surgeons at a single institution. Mean age at the time of surgery was 14 years. All patients had a minimum 24-month follow-up radiograph (range 24–60 months, mean 32 months).
Choice of surgical approach was at the discretion of the attending surgeon based on clinical examination, pulmonary function tests, and radiographs (standing anteroposterior and lateral, supine anteroposterior, bending, and push-prone). The patients were not randomized according to curve type. Double-major, double-thoracic, or triple-major curves were treated posteriorly, single thoracolumbar or lumbar curves were treated anteriorly, and single thoracic curves were treated with either anterior or posterior surgery. Patients underwent one of the following: 1) anterior thoracic instrumentation (n = 23); 2) posterior thoracic instrumentation (n = 40); 3) anterior thoracolumbar instrumentation (n = 27); or 4) posterior thoracic and lumbar instrumentation (n = 20). Of the 40 who had posterior thoracic instrumentation and fusion, one patient first underwent video-assisted thoracoscopic anterior release and uninstrumented fusion. Of the 20 who had posterior thoracic and lumbar fusions, 6 patients first underwent anterior release and uninstrumented fusion by open thoracotomy. Thus, seven patients in our series had combined anterior and posterior fusion, but with all of the instrumentation being performed posteriorly. All seven of these patients had large, stiff curves >72° (range 72–98°; mean 84°).
Anterior instrumentation was performed in the direct lateral decubitus position with single rod–single screw constructs. Autologous bone-filled titanium mesh cages were used for structural anterior column support and to promote lordosis when necessary. Cantilever and convex compression forces were used anteriorly. Posterior surgery was performed on the open OSI frame. Posterior constructs consisted of 5.5-mm rod-based CD-Horizon implants (Medtronic Sofamor-Danek, Memphis, TN) using some combination of screws, hooks, and Wisconsin wires. Pedicle screws or hooks were used as the distal anchors, most commonly hooks were used proximally, and Wisconsin wires were used for translational correction over the apex of the thoracic curve. Cantilever forces, translational apical correction with wires, and in situ rod contouring were the primary corrective maneuvers used posteriorly. Because the patients in this series had very mild thoracic hypokyphosis, maneuvers to enhance kyphosis (e.g., rod rotation, sublaminar wires) were not used. The increase in thoracic kyphosis achievable with these maneuvers was not considered necessary in light of the risks of junctional deformity and neurologic injury, respectively. Some compression forces were also used posteriorly in the lumbar spine as secondary corrective maneuvers in the coronal plane and to enhance lordosis. Distraction forces were avoided throughout both the thoracic and lumbar spine. Autologous bone graft (either iliac crest and/or rib) was used in every patient.
Preoperative, postoperative (within 2 months postoperative), and final follow-up (minimum 2-year) standing lateral radiographs were analyzed with the Cobb method in the sagittal plane. Measurements included the following: C7 plumbline (the horizontal distance of a plumbline dropped from the center of the C7 body to the posterosuperior corner of the S1 body), proximal junctional measurement (PJM; Cobb angle between the most proximal instrumented vertebra and the segment two levels cephalad), distal junctional measurement (DJM; Cobb angle between the most distal instrumented vertebra and the segment two levels caudal), thoracolumbar junction (TLJ) (T10–L2), thoracic kyphosis (T5–T12), and lumbar lordosis (T12–S1). Positive values were used to denote kyphosis. Negative values were used to indicate lordosis. Radiographs were also analyzed for evidence of instrumentation failure (e.g., broken implants, implant pullout). Pseudarthroses were identified when evident radiographically or their presence was inferred on the basis of failed instrumentation or progression of curvature over the instrumented levels. Statistical analysis was performed using a mixed-model, repeated-measures analysis of variance.
Four patients experienced implant failure and/or required revision for progression. Of these patients, two had broken rods (both had anterior instrumentation) and one had distal hook pullout (patient had selective posterior thoracic instrumentation of a double-major curve). The remaining patient progressed distal to a selective posterior thoracic fusion and required extension of fusion to include part of the lumbar curve.
Fusion levels varied according to the choice of anterior versus posterior instrumentation. The average number of instrumented levels was as follows: 6.8 for anterior thoracic (all of these patients had single thoracic curves), 10.7 for posterior thoracic (these patients had single or double thoracic curves), 4.3 for anterior thoracolumbar–lumbar, and 11.9 for posterior thoracic and lumbar. The most common proximal level was as follows: T5 for anterior thoracic instrumentation (16 of 23), T11 for anterior thoracolumbar–lumbar (10 of 27), T3 for posterior thoracic instrumentation (16 of 40), and T4 for posterior thoracic and lumbar instrumentation (10 of 20). The most common distal level was as follows: T11 for anterior thoracic instrumentation (12 of 23), L3 for anterior thoracolumbar–lumbar (22 of 27), L1 for posterior thoracic instrumentation (16 of 40), and L4 for posterior thoracic and lumbar instrumentation (14 of 20).
C7 Sagittal Plumbline
Selective anterior thoracic instrumentation led to the largest positive displacement in the C7 sagittal plumbline from preoperative to postoperative (+14 mm change) (Table 1). Posterior thoracic instrumentation and anterior thoracolumbar instrumentation also produced positive displacements in the C7 plumbline after surgery, but to lesser extents (+2 mm and +4 mm changes, respectively). Of the four groups, only those having both thoracic and lumbar curves instrumented posteriorly had a posterior displacement in C7 plumbline after surgery (−9 mm preoperative to −20 mm postoperative).
Between the postoperative period to final follow-up, however, patients in all four groups experienced positive displacements in their C7 plumbline. Nevertheless, those with anterior thoracic instrumentation still had the largest overall positive displacement from preoperative to final follow-up (+19 mm change). In all four groups, the absolute value of the C7 plumbline was negative preoperative and remained so at the final follow-up. No statistically significant difference was noted among the four groups with respect to the change in C7 plumbline.
Proximal Junctional Measurement
Patients undergoing anterior instrumentation for thoracic or thoracolumbar curves had a higher preoperative PJM than those undergoing posterior instrumentation (+8° for anterior thoracic, +6° for anterior thoracolumbar–lumbar, +4° for posterior thoracic, and +2° for posterior thoracic and lumbar) (Table 2). These preoperative differences are not comparable because different modes of instrumentation are associated with different proximal fusion levels.
Anterior thoracic instrumentation had no appreciable change on the PJM either after surgery or at the ultimate follow-up (8° preoperative, 9° postoperative and at final follow-up). Anterior thoracolumbar instrumentation caused a 3° increase in the PJM from 6° preoperative to 9° postoperative, with no further change at final follow-up. Posterior thoracic instrumentation led to a relatively larger 6° increase in PJM postoperative (from 3° preoperative to 9° postoperative), but there was a minimal increase from postoperative to final follow-up (9° to 10°). Posterior instrumentation of double major curves also led to a relatively larger increase of 7° in PJM from 2° preoperative to 9° postoperative, which increased mildly to 11° at final follow-up. Therefore, posterior instrumentation was associated with greater increases in PJM than anterior instrumentation. These changes from preoperative to final follow-up were statistically significant (anterior thoracic vs. posterior thoracic, P = 0.02; anterior thoracic vs. posterior thoracic and lumbar, P = 0.002; anterior thoracolumbar vs. posterior thoracic and lumbar, P = 0.03). Most of the increase occurred immediately after surgery, with little progression at the final follow-up.
No patient who underwent anterior thoracic instrumentation had a change in PJM ≥10° or ≥15° at final follow-up compared with preoperative (Table 3); 19% of those with anterior thoracolumbar instrumentation had a change in PJM of ≥10°, but none had a change ≥15°. Overall, 35% of those with any form of posterior instrumentation experienced an increase in PJM ≥10°, and 17% had an increase of ≥15°. No patient developed clinically problematic increase in PJM or underwent revision for that reason.
All patients were relatively hypokyphotic before surgery (+24° for all patients undergoing anterior instrumentation and +25° for those undergoing posterior instrumentation) (Table 4). There was a trend for anterior thoracic instrumentation to be minimally kyphogenic after surgery (27° preoperative to 28° postoperative), whereas posterior thoracic instrumentation had a trend toward being mildly lordogenic (24° preoperative to 21° postoperative) (change in thoracic kyphosis with anterior vs. posterior thoracic instrumentation;P = 0.09). These findings are consistent with the fact that convex compression forces were applied with anterior instrumentation. The lordogenic effect of posterior instrumentation may result from the fact that distraction forces, rod rotation, and sublaminar wires were avoided with posterior instrumentation in this series. In addition, positioning a patient prone on the open OSI frame promotes less thoracic kyphosis. 9
Over time, these trends became more statistically significant, with a 4° increase in thoracic kyphosis at final follow-up using anterior thoracic instrumentation (27° preoperative to 31° at final follow-up) versus a 2° decrease in kyphosis using posterior thoracic instrumentation (24° preoperative to 22° at final follow-up) (change in thoracic kyphosis from preoperative to final follow-up with anterior vs. posterior thoracic instrumentation;P = 0.04). Patients having both thoracic and lumbar curves fused posteriorly also became less kyphotic in the thoracic spine (28° preoperative to 25° at final follow-up).
Patients undergoing selective thoracic instrumentation had a relatively straight TLJ before surgery (+1° preoperative in those having anterior instrumentation and −2° preoperative in those having posterior instrumentation) (Table 5). No significant changes were noted in either of these groups postoperative or at final follow-up (+2° at final follow-up in the anterior thoracic group and −2° at final follow-up in the posterior thoracic group; not significant). Similarly, patients with thoracolumbar curves treated anteriorly had no change in their TLJ measurements (+2° preoperative and at final follow-up). However, patients with double major curves having posterior instrumentation of both curves began with a relatively kyphotic TLJ (+11° preoperative), which was corrected to −3° postoperative and did not change significantly at final follow-up (−1°).
In curve patterns in which the lumbar spine was not primarily instrumented, the compensatory lumbar response to either selective thoracic anterior or posterior instrumentation was a decrease in lumbar lordosis after surgery (−62° preoperative to −57° postoperative for anterior thoracic instrumentation; −64° preoperative to −57° postoperative for posterior thoracic instrumentation) (Table 6). However, at the final follow-up, most of the lordosis present preoperative had been regained (−62° preoperative to −61° at final follow-up for anterior instrumentation; −64° to −62° at final follow-up for posterior instrumentation).
In the curve patterns in which a structural lumbar curve was instrumented, lumbar lordosis increased slightly postoperative and continued to increase at final follow-up (−57° preoperative to −60° at final follow-up for anterior thoracolumbar instrumentation; −54° preoperative to −58° at final follow-up for posterior instrumentation of double major curves). With posterior thoracic and lumbar instrumentation, the segmental lumbar lordosis (i.e., from T12 to the lowest instrumented level, usually L4) at final follow-up was −21°, which was an improvement in lordosis of −8° from a value of −13° preoperative. With anterior thoracolumbar instrumentation, the segmental lordosis from T12 to the lowest instrumented vertebrae (usually L3) increased minimally by 1° from −6° preoperative to −7° at final follow-up. Segmental lumbar lordosis was not calculated in patients undergoing selective anterior or posterior thoracic instrumentation because the lowest instrumented vertebrae in these patients were T12 and L1, respectively.
Distal Junctional Measurement
Anterior thoracic instrumentation led to a 2° increase (i.e., more kyphotic measurement) in the DJM at final follow-up compared with preoperative (−6° preoperative to −4° at final follow-up) (Table 7). Posterior thoracic instrumentation led to a 1° increase in DJM (−21° preoperative to −20° at final follow-up). Anterior thoracolumbar–lumbar instrumentation produced a 2° decrease (i.e., more lordotic measurement) in DJM at final follow-up (−36° preoperative to −38° at final follow-up). Posterior instrumentation of both the thoracic and lumbar spine led to a 4° increase at final follow-up (−39° preoperative to −35° at final follow-up). None of these changes was statistically different among the four groups.
Therefore, large changes in DJM were not observed from preoperative to final follow-up in any of the four groups evaluated. However, with all forms of instrumentation, the DJM initially increased substantially postoperative (i.e., became more kyphotic) and then decreased toward preoperative values at final follow-up.
The purpose of this study was to determine the effect of anterior versus posterior instrumentation on sagittal plane profiles in patients undergoing surgical correction of adolescent idiopathic scoliosis. The goals of scoliosis correction in the sagittal plane include achieving normal ranges of thoracic kyphosis and lumbar lordosis and producing a harmonious sagittal contour with the patient in slightly negative or at least neutral sagittal balance. At a minimum 2-year follow-up, we found that both anterior and posterior instrumentation could be used to meet these goals using the surgical techniques described. The overall radiographic results using both forms of instrumentation were very similar in the sagittal plane. There were some slight differences in both the direction and magnitude of changes in sagittal plane parameters (i.e., from preoperative to postoperative or final follow-up) according to whether anterior or posterior instrumentation was used. Although some of these differences achieved statistical significance, most were small in magnitude. In addition, the sagittal parameters created depend not only on the approach (i.e., anterior vs. posterior) but also on the method of instrumentation. For example, the lordogenic effect of posterior instrumentation observed in this series is not necessarily generalizable to other methods of instrumentation capable of enhancing thoracic kyphosis (e.g., rod rotation, sublaminar wires, distraction forces). Nevertheless, surgical approach is a major determinant on the sagittal plane; thus, differences resulting from anterior versus posterior surgery should be considered before surgery to design operations that will optimize postoperative sagittal parameters.
The numerically largest difference between anterior versus posterior instrumentation in the present study was the association of posterior instrumentation with the greatest increases in the PJM. In other words, posterior instrumentation was associated with the development of proximal junctional kyphosis. However, the magnitude of the increase in PJM was small in our series. Although 35% of all patients with posterior instrumentation had an increase in PJM of ≥10° at final follow-up, only 17% (10 of 60) had an increase of ≥15°. Importantly, no patient required revision for proximal junctional kyphosis, and no one was noted to have a clinically problematic proximal junctional kyphosis. The development of proximal junctional kyphosis with posterior instrumentation in our series is consistent with a previous report by Lee et al. 7
Although the numerical values of PJM at final follow-up were similar among the four groups in our study, the importance of the PJM differs with the proximal instrumented level. According to Bernhardt and Bridwell, 1 the “normal” sagittal measurement between T1 and T3 is 4° and between T2 and T4 is 6.5°. Thus, the observed values of PJM of 10° for posterior thoracic instrumentation (with the most common proximal instrumented vertebra being T3) and 11° for posterior thoracic and lumbar instrumentation (with the most common proximal instrumented vertebra being T4) are 6° and 4.5° greater than “normal,” respectively. In contrast, the PJM with anterior thoracic instrumentation at final follow-up (9°) was nearly identical to the “normal” value (8.5°) between T3 and T5. The PJM with anterior thoracolumbar instrumentation at final follow-up was 9°, which is 3° more than “normal” between T9 and T11. Potential reasons for the increase in PJM with posterior instrumentation include the following: 1) disruption of the posterior tension band with posterior surgery; 2) posterior compression forces (e.g., as part of proximal claw constructs, or to correct proximal thoracic curves from the convexity) might promote junctional kyphosis; and 3) the increase in PJM may be a compensation for the decrease in thoracic kyphosis associated with posterior instrumentation. Consistent with the last hypothesis is a report by Hilibrand et al, 5 who noted development of cervical kyphosis in patients with adolescent idiopathic scoliosis experiencing a decrease in thoracic kyphosis postoperative after posterior instrumentation. In the present study, however, the converse was not true; namely, anterior thoracic instrumentation (which was kyphogenic to the thoracic spine) did not lead to a compensatory decrease in PJM.
There was a trend for anterior thoracic instrumentation to lead to the greatest postoperative anterior displacement in the C7 sagittal plumbline, with a +19 mm increase from −26 mm preoperative to −7 mm at final follow-up. This finding was not statistically significant. The clinical importance of this finding is also unclear, especially given the low magnitude of the change and the fact that the absolute value of the C7 plumbline at final follow-up remained negative in this group as it did in all of the other groups. No form of instrumentation was associated with overt sagittal vertebral axis problems during the follow-up period. However, in three of the four groups, the C7 plumbline underwent an anterior displacement or stayed the same from preoperative to postoperative and then sustained further anterior displacement from postoperative to final follow-up. The anterior displacement in C7 plumbline could occur naturally with time (i.e., even in the absence of surgery). We can conjecture as to why anterior thoracic instrumentation produced the greatest anterior displacement in C7 plumbline. The kyphogenic force of anterior convex compression in the thoracic spine may create a moment that brings the upper body forward. In the absence of a reciprocal increase in lumbar lordosis (as was the case in this study), the result might be an anterior displacement in the C7 plumbline. Although theoretically possible, this hypothesis seems somewhat unlikely as there was only a 4° increase in thoracic kyphosis and a 1° decrease in lumbar lordosis at final follow-up using anterior thoracic instrumentation. Potential reasons why the C7 plumbline did not sustain as large an anterior displacement in the other groups include the following: 1) kyphogenic forces were avoided with our methods of posterior instrumentation because distraction forces, rod rotation, and sublaminar wires were not applied; and 2) patients undergoing anterior thoracolumbar instrumentation had convex compression forces applied anteriorly, but the kyphotic force in these patients was converted into a lordogenic effect by the placement of anterior structural cages or bone graft.
Anterior thoracic instrumentation was kyphogenic over the thoracic spine (+4° increase at final follow-up), whereas posterior instrumentation was lordogenic (−2° more lordosis at final follow-up for posterior thoracic instrumentation; −3° more lordosis at final follow-up for posterior thoracic and lumbar instrumentation). Several factors may be involved in this result: 1) anterior compression forces were used with anterior thoracic instrumentation; 2) prone positioning on the OSI table has a lordogenic effect, as noted in previously published reports; 3) the rod-rotation maneuver was not used, thus limiting the amount of kyphosis produced posteriorly; and 4) posterior correction forces consisted of in situ contouring and translational rather than distraction forces. Anterior thoracolumbar instrumentation was also slightly kyphogenic to the thoracic spine, but the overall thoracic kyphosis was less than that seen with anterior thoracic instrumentation. Importantly, the average thoracic kyphosis at final follow-up in all of the groups was within the normal range (20–40°).
Lumbar lordosis could be preserved and even enhanced when a structural lumbar curve was instrumented either anteriorly or posteriorly. Lordosis was promoted with anterior thoracolumbar instrumentation through the use of structural anterior cages or bone graft. With posterior thoracic and lumbar instrumentation, lordosis was promoted by prone positioning on the OSI table as well as compression on the lumbar convexity. For these patients the segmental lordosis from T12 to the lowest instrumented vertebra also improved −8° to a final follow-up value of −21°, which compares favorably with the “normal” value of −24° between T12 and L4. 1 In contrast, when thoracic instrumentation was used anteriorly or posteriorly, some loss of lumbar lordosis occurred. There was a loss of 5° of lordosis after surgery with anterior thoracic instrumentation and a loss of 7° of lumbar lordosis with posterior thoracic instrumentation. However, between postoperative and final follow-up, some recovery of the lost lordosis occurred, such that the ultimate loss of lordosis was only 1° for anterior thoracic and 2° for posterior thoracic instrumentation. In these young patients with presumably normal lumbar segments, compensation in the sagittal plane appeared to occur over time but was not entirely complete.
Posterior thoracic and lumbar instrumentation produced the greatest increase in DJM after surgery. As with lumbar lordosis, DJM did return toward preoperative values in these patients, but not completely. At final follow-up, DJM increased 4° in these patients compared with preoperative. The clinical significance of this finding is unclear, not only because the magnitude of the change is small but also because the total lumbar lordosis in these patients increased by 4° during the same time period. Thus, the increase in DJM may be a compensation for the improved lordosis over the instrumented levels.
Properly performed, acceptable sagittal plane profiles can be achieved with either anterior or posterior instrumentation in adolescent idiopathic scoliosis. However, anterior and posterior instrumentation do have differential effects on the sagittal plane, and these should be kept in mind during preoperative planning to design an operation that will optimize sagittal parameters after surgery. Posterior instrumentation can lead to proximal junctional kyphosis. When performed according to the methods described in the present study, anterior thoracic instrumentation is more kyphogenic to the thoracic spine than is posterior thoracic instrumentation. Lumbar lordosis can be enhanced when structural lumbar curves are instrumented either anteriorly or posteriorly. Neither anterior nor posterior instrumentation led to distal junctional kyphosis in our series. Some anterior displacement of the C7 sagittal plumbline can be expected in most patients, but the overall balance should remain negative.
- Properly performed, acceptable sagittal profiles can be achieved with either anterior or posterior instrumentation.
- Proximal junctional kyphosis was associated with posterior instrumentation. Distal junctional kyphosis was not associated with either anterior or posterior instrumentation.
- Thoracic kyphosis was more enhanced with anterior instrumentation.
- Lumbar lordosis was enhanced when structural lumbar curves were instrumented either anteriorly or posteriorly.
- The C7 sagittal plumbline sustained the largest anterior displacement with anterior thoracic instrumentation. Some anterior displacement can be expected in most patients, but the overall balance should still remain negative.
The authors thank Karen Steger-May, MA, Division of Biostatistics, Washington University School of Medicine, for her statistical analysis.
1. Bernhardt M, Bridwell KH. Segmental analysis of the sagittal plane
alignment of the normal thoracic and lumbar spines and thoracolumbar junction. Spine 1989; 14: 717–21.
2. Betz RR, Harms J, Clements DH III, et al. Comparison of anterior and posterior instrumentation for correction of adolescent thoracic idiopathic scoliosis
. Spine 1999; 24: 225–39.
3. Betz RR, Shufflebarger H, McLain RF, et al. Anterior versus
posterior instrumentation for the correction of thoracic idiopathic scoliosis
. Spine 2001; 26: 1095–100.
4. Booth KC, Bridwell KH, Lenke LG, et al. complications and predictive factors for the successful treatment of flatback deformity (fixed sagittal imbalance). Spine 1999; 24: 1712–20.
5. Hilibrand AS, Tannenbaum DA, Graziano GP, et al. The sagittal alignment of the cervical spine in adolescent idiopathic scoliosis
. J Pediatr Orthop 1995; 15: 627–32.
6. Lagrone MO, Bradford DS, Moe JH, et al. Treatment of symptomatic flatback after spinal fusion. J Bone Joint Surg Am 1988; 70: 569–80.
7. Lee GA, Betz RR, Clements DH III, et al. Proximal kyphosis after posterior spinal fusion in patients with idiopathic scoliosis
. Spine 1999; 24: 795–9.
8. Lenke LG, Betz RR, Bridwell KH, et al. Spontaneous lumbar curve coronal correction after selective anterior or posterior thoracic fusion in adolescent idiopathic scoliosis
. Spine 1999; 24: 1663–71.
9. Marsicano JG, Lenke LG, Bridwell KH, et al. The lordotic effect of the OSI frame on operative adolescent idiopathic scoliosis
patients. Spine 1998; 23: 1341–8.