Patient Selection
Following study approval by our institution's investigational review board, a retrospective analysis of patients with thoracic scoliosis who had been operatively managed by a single surgeon (B.S.L.) between 2001 and 2004 was undertaken. In total, 148 patients with adolescent idiopathic scoliosis and a Lenke type-I curve were operatively managed during the study period. From these initial 148 patients, a matched-pair analysis of thirty-four consecutive patients (seventeen pairs) who underwent either video-assisted thoracoscopic surgery or posterior spinal fusion with thoracic pedicle screws for the treatment of thoracic adolescent idiopathic scoliosis was performed; this group included eight male and twenty-six female patients with an average age of 15.0 years. Pairs were matched with respect to curve type, curve magnitude (within 6°), age (within nine months), and sex. All patients had a minimum of twenty-four months of radiographic and clinical follow-up data.
Radiographic Outcomes
A complete radiographic dataset included preoperative standing posteroanterior, lateral, and right and left lateral supine bending radiographs. At the time of the most recent follow-up, the posteroanterior and standing lateral radiographs were reviewed. The preoperative and most recent postoperative radiographs were compared. Radiographic outcomes included major curve correction, cephalad thoracic curve correction, caudad compensatory lumbar curve correction, coronal balance (measured as the offset of a plumb line drawn from the centroid of T1 to the midpoint of the sacrum), lumbar lordosis (from T12 to S1), sagittal balance (measured as the offset of a plumb line drawn from the centroid of T1 to the posterior corner of the S1 end plate on the lateral radiograph), the tilt angle of the most caudad instrumented vertebra, thoracic kyphosis (from T5 to T12), and the number of levels fused. We also analyzed the location of the most caudad instrumented vertebra in relation to three radiographic landmarks: (1) Cobb angle end vertebra, (2) stable vertebra, and (3) neutral vertebra.
Clinical Outcomes
Angle of trunk rotation measurements were made with a Scoliometer (Orthopedic Systems, Hayward, California) preoperatively and postoperatively to assess rib hump deformity. Intraoperative and perioperative outcomes included the operative time, the estimated blood loss, the rate of blood transfusion, the use of a Cell Saver (Haemonetics, Braintree, Massachusetts), the length of hospital stay, and the rate and type of complications. The Scoliosis Research Society-22 (SRS-22) outcome questionnaire was administered to all patients preoperatively and at a minimum of twenty-four months postoperatively. Total scores and the individual domain scores for pain, self-image, function, mental health, and satisfaction were compared between the preoperative and twenty-four-month time points for the two groups.
Pulmonary function was also assessed for all patients. The standard protocol initially consisted of spirometry to assess vital capacity (as a measure of restrictive lung disease) and peak flow (as a measure of large airway function) at all preoperative and postoperative time points. Recently, however, the protocol shifted to include preoperative and postoperative forced vital capacity and forced expiratory volume instead. As a result, preoperative and postoperative peak flow and vital capacity data were available for all patients. Comparative data for forced expiratory volume and forced vital capacity, however, were only available at the twenty-four-month postoperative time point and are reported as percent predicted forced expiratory volume and percent predicted forced vital capacity.
Indications for Surgery
Surgery was recommended for patients with progressive thoracic scoliosis and a Cobb angle of =40°. Patients with a Cobb angle of =70° often underwent a thoracoscopic anterior release followed by posterior spinal fusion. Any patient with a Cobb angle of =70°, therefore, was excluded from the current study. Patients with hyperkyphosis measuring =40° (from T5 to T12) were advised to undergo posterior spinal fusion and therefore were not considered to be candidates for anterior spinal fusion or video-assisted thoracoscopic surgery. These patients also were excluded from the current study.
When a patient met the inclusion criteria, the decision regarding the use of video-assisted thoracoscopic surgery or posterior spinal fusion with thoracic pedicle screws was discussed with the patient and with the family. After careful consideration of the benefits and potential disadvantages of each option, the decision was made by the family19,23. It should be noted that the first video-assisted thoracoscopic surgery procedure performed in the current study was the lead surgeon's twenty-ninth case, which theoretically reduces the potential for learning curve-related influence on outcomes.
Surgical Technique
The surgical technique for video-assisted thoracoscopic surgery, and our modification of this technique from that originally described by Picetti et al., have been previously described in detail9,13,18. In order to optimize the fusion bed and minimize the risk for pseudarthrosis with this challenging technique13, autogenous bone from the rib was utilized as graft material. In most cases, a single 3 to 5-cm segment of apical rib was harvested through a thoracoscopic portal. Occasionally, two ribs were harvested, leaving an intact rib in between the two harvest sites. A single-rod video-assisted thoracoscopic surgery construct was used in fifteen of seventeen cases, and a dual-rod construct was used in two. Structural interbody grafting has not been routinely performed and was not done in any of the cases in this series.
For posterior spinal fusion with thoracic pedicle screws, no more than two hooks were placed throughout the entire construct and screws comprised a minimum of 80% of the anchors. Posterior spinal fusion was undertaken through a standard posterior approach with the patient in the prone position. Segmental polyaxial pedicle screw fixation was used unilaterally or bilaterally at each level. Polyaxial pedicle screws were used because they have more freedom at the screw-rod interface, which may facilitate easier seating of the rod into the head of the screw31. Recent evidence has demonstrated that polyaxial screws may be associated with reduced axial plane deformity derotation and increased cost when compared with monoaxial screws32. Pedicle screws were placed segmentally along the concavity of the curve and approximately at every other level along the convexity. A combination of cancellous allograft and demineralized bone matrix was used as the graft source. At the time of this study, the mean number of screws per level was compared with one screw per level for the video-assisted thoracoscopic surgery technique.
Vertebrae selected for fusion in patients undergoing video-assisted thoracoscopic surgery typically spanned the Cobb angle end vertebrae of the curve. The criteria for determining fusion levels in patients undergoing video-assisted thoracoscopic surgery are based on previous work on open anterior spinal fusion that has shown equivalent coronal plane curve correction, an improved ability to correct hypokyphosis, and the saving of 2.5 lumbar levels as compared with posterior spinal fusion when fusing the vertebrae that span the Cobb angle end vertebrae6. In contrast, posterior spinal fusion levels are typically based on principles previously set forth by Cotrel, Dubousset, and Shufflebarger, in which the fusion extends superiorly from the neutrally rotated vertebra cephalad to the cephalad Cobb angle end vertebra and inferiorly to the stable or neutral vertebra caudad to the distal Cobb angle vertebra10,33,34. In the current series, levels selected for fusion in patients undergoing posterior spinal fusion typically included T4 or T5 superiorly to the most cephalad vertebra caudad to the curve touched by a vertical line extended from the midportion of the sacrum (center sacral vertical line).
On the average, one or two cross-links were used in the current study. Although recent studies have demonstrated that perhaps cross-links do not add substantial structural rigidity to the construct when segmental pedicle screw fixation is used35,36, there are also substantial biomechanical data supporting the use of cross-links, which may increase the torsional rigidity of the construct, particularly if screws are not placed at each vertebra within the fusion37-40.
A left-sided rod was contoured to the desired sagittal contour and was seated in the pedicle screws on the concave side. A rod derotation maneuver was performed, and the rod was fixed by the tightening of set screws superiorly and inferiorly. Additional correction was achieved by performing a derotation maneuver by manipulating the concave and convex screws simultaneously. A second rod was placed, and additional derotation and distraction and compression maneuvers were performed to finalize correction.
Statistical Analysis
Repeated-measures analysis of variance was performed to determine if differences existed between the video-assisted thoracoscopic surgery group and the posterior spinal fusion group in terms of the change between the preoperative and two-year findings with respect to radiographic outcome measures, SRS-22 scores, the angle of trunk rotation, and pulmonary function. Analysis of variance was performed to assess if differences existed between the two surgical groups in terms of the percentage correction of the major curve angle and most caudad instrumented vertebra relative to the Cobb angle end vertebra, stable vertebra, and neutral vertebra or in terms of operative time, estimated blood loss, or length of hospital stay. Rates of transfusion and use of blood collected with the Cell Saver were compared with use of the chi-square test.
Source of Funding
No outside funding was received for this study.
Overall, the video-assisted thoracoscopic surgery and posterior spinal fusion groups were evenly matched at baseline with regard to age (14.9 compared with 14.9 years; p = 0.9), sex, Lenke curve type, and duration of follow-up (twenty-five compared with twenty-eight months; p = 0.21). In the video-assisted thoracoscopic surgery group, there were nine Lenke type-1AN curves, five type-1BN curves, and three type-1CN curves. In the posterior spinal fusion group, there were eleven Lenke type-1AN curves, one type-1BN curve, and five type-1CN curves. In the posterior spinal fusion group, an average of 3.2 ± 1.0 screws (range, two to five screws) were placed along the convexity of the curve, compared with 6.8 ± 1.9 screws (range, four to twelve screws) along the concavity. In the video-assisted thoracoscopic surgery group, an average of 1.3 ± 0.5 ribs (range, one to two ribs) were used for autograft.
Radiographic Results
Preoperative and postoperative radiographic results are displayed in Table I. Preoperatively, there was no difference between the video-assisted thoracoscopic surgery group and the posterior spinal fusion group in terms of the main thoracic curve (average, 49.1° compared with 47.4°; p = 0.34). There was no significant difference at baseline between the two groups with respect to cephalad thoracic curve, caudad compensatory lumbar curve, coronal balance, thoracic kyphosis, lumbar lordosis, sagittal balance, or tilt angle (p > 0.05).
Postoperative radiographic comparisons between the video-assisted thoracoscopic surgery and posterior spinal fusion groups are also shown in Table I. On the average, the number of levels fused was significantly smaller in the video-assisted thoracoscopic surgery group (5.9 compared with 8.9; p < 0.001). Postoperative curve magnitude was not significantly different between the treatment arms (21.2° for the video-assisted thoracoscopic surgery group, compared with 17.3° for the posterior spinal fusion group; p = 0.10). However, there was a trend toward a greater major curve percentage correction in the posterior spinal fusion group (57.3% for the video-assisted thoracoscopic surgery group, compared with 63.8% for the posterior spinal fusion group; p = 0.08). With the numbers available, no significant differences were detected between the two groups at the time of the most recent follow-up with respect to cephalad thoracic curve, caudad compensatory lumbar curve, coronal balance, thoracic kyphosis, lumbar lordosis, sagittal balance, or end vertebra tilt angle measurements.
On the average, the most caudad instrumented vertebra was 0.81 level more cephalad in the video-assisted thoracoscopic surgery group than in the posterior spinal fusion group relative to the Cobb end vertebra (p = 0.004). With reference to the stable and neutral vertebrae, the most caudad instrumented vertebra in the video-assisted thoracoscopic surgery group was also higher in comparison with that in the posterior spinal fusion group by 0.3 level and 0.7 level, respectively, although these differences did not achieve significance (Table II).
Clinical Results
Table III depicts the perioperative outcomes for the two groups. Operative times were significantly greater in the video-assisted thoracoscopic surgery group (p = 0.033), but estimated blood loss was significantly greater in the posterior spinal fusion group (p = 0.001). The rate of transfusion was not different between the posterior spinal fusion and video-assisted thoracoscopic surgery groups (29% compared with 18%; p = 0.69); however, intraoperative blood salvage was utilized in two-thirds of the patients in the posterior spinal fusion group and was not used in the video-assisted thoracoscopic surgery group. The average amount of blood from the Cell Saver that was used in the posterior spinal fusion group was 382 ± 267 mL (range, 20 to 800 mL). Table IV shows that there were no differences between the groups in terms of the SRS-22 questionnaire outcomes or the angle of trunk rotation measurements.
Pulmonary function outcomes data are presented in Table V. Two years postoperatively, no differences were detected between the groups in terms of the percent predicted forced vital capacity and forced expiratory volume parameters. Both groups demonstrated improvement from the preoperative status with respect to peak flow and vital capacity. However, there was a significantly greater improvement in peak flow from preoperatively to postoperatively in the posterior spinal fusion group when compared with the video-assisted thoracoscopic surgery group (p = 0.04). Analysis of the change in vital capacity from preoperatively to postoperatively did not reveal any differences between the two groups.
There was one complication in the video-assisted thoracoscopic surgery group; specifically, a distal set screw dislodged from the construct after the patient lifted a 50-lb (22.7-kg) box. This resulted in partial loss of correction at four months. The set screw did not migrate in the chest cavity and therefore was not removed. The loss of correction was subsequently treated with a posterior spinal fusion with thoracic pedicle screws (Fig. 1). There were three complications in the posterior spinal fusion with thoracic pedicle screws group: (1) a superficial wound infection, which resolved with a course of oral antibiotics; (2) a misplaced L1 pedicle screw causing radicular pain that required revision on the twelfth postoperative day, with complete symptom resolution (Fig. 2); and (3) one case of minor respiratory depression as a result of narcotic analgesia, which resolved with supportive care.
In the treatment of structural thoracic adolescent idiopathic scoliosis curves, anterior spinal fusion maintains several possible advantages over standard posterior spinal fusion constructs. With the increasing acceptance and use of thoracic pedicle screw constructs for posterior spinal fusion, however, the notion that anterior spinal fusion can routinely save caudad fusion levels in comparison with posterior spinal fusion constructs has been brought into question and may no longer be true. Several studies have demonstrated that the potential advantages of posterior spinal fusion with thoracic pedicle screws as compared with conventional posterior spinal fusion with hooks, sublaminar wires, or hybrid techniques include a shorter fusion length, improved major curve correction, reduced blood loss, and improved pulmonary function4,25,28. It is possible, therefore, that the more powerful correction that results from thoracic pedicle screw constructs may negate one of the main advantages of anterior spinal fusion over posterior spinal fusion, that is, the ability to save caudad fusion levels. Another recently described advantage of pedicle screw constructs over anterior spinal fusion is the ability to safely correct large curves of >90° without the need for an anterior release27. Video-assisted thoracoscopic surgery, on the other hand, is not recommended as a routine stand-alone treatment in curves of >70°. In what we believe to be the only other study in which anterior spinal fusion has been compared with posterior spinal fusion with thoracic pedicle screws, Potter et al. compared the radiographic outcomes of open anterior spinal fusion with those of posterior spinal fusion with thoracic pedicle screws and found that posterior spinal fusion with thoracic pedicle screws provided significantly greater main thoracic curve correction (62% compared with 52%) and spontaneous thoracolumbar-lumbar curve correction41. Those authors also found improved correction of thoracic torsion and rotation with posterior spinal fusion with thoracic pedicle screws, at the expense of an additional 1.2 levels fused in comparison with anterior spinal fusion. However, they did not evaluate the position of the most caudad instrumented vertebra in reference to the stable or end vertebrae, nor did they report on intraoperative parameters, clinical outcomes, or pulmonary function. To our knowledge, no previous study has compared the clinical and radiographic outcomes of posterior spinal fusion with use of thoracic pedicle screws with those of video-assisted thoracoscopic surgery for the treatment of structural thoracic curves.
We found that posterior spinal fusion with thoracic pedicle screws had a trend toward greater major curve correction in comparison with video-assisted thoracoscopic surgery (64% compared with 57%), which is similar to the findings of Potter et al.41. With respect to percent curve correction with use of video-assisted thoracoscopic surgery, our results were comparable with those reported by Picetti (50.2%), Newton (60%), Al-Sayyad (68%), and Wong (62%) and colleagues12,13,15,21. In the present study, the percentage of curve correction in the posterior spinal fusion group, however, was slightly lower than those in the more recent reports of all-screw constructs by Suk (69%), Kim (70%), Kim (76%), Cheng (68%), and Lonner (68%) and colleagues4,25,28,42,43. This difference is most likely due to the use of fewer screws in the present series than in the other reported series. There were no other radiographic differences between the groups at a minimum of twenty-four months of follow-up with respect to coronal balance, thoracic kyphosis, lumbar lordosis, sagittal balance, and end vertebral tilt angle. We also found that video-assisted thoracoscopic surgery afforded no additional advantage with respect to spontaneous correction of the cephalad thoracic and caudad lumbar compensatory curves. The patients managed with video-assisted thoracoscopic surgery did have significantly fewer fused levels in comparison with those managed with posterior spinal fusion with thoracic pedicle screws (5.9 compared with 8.9). We did not find any increased tendency to produce thoracic kyphosis in the video-assisted thoracoscopic surgery group in comparison with the posterior spinal fusion group. This finding is contrary to those of other reports, which have demonstrated a kyphogenic effect of anterior constructs and a poor ability of thoracic pedicle screw constructs to restore or increase kyphosis44-46, although the clinical benefit of slight additional kyphosis restoration for curves within the normal range remains unclear44. Both groups had relatively hypokyphotic thoracic spines preoperatively, typical of adolescent idiopathic scoliosis, with a slight increase in kyphosis noted in both groups following surgery.
Although a reduction in the total number of levels fused has been shown to be an advantage of anterior spinal fusion over standard posterior spinal fusion, the stable and lower end vertebrae are also routinely evaluated in the selection of fusion levels in order to achieve maximum curve correction, to optimize sagittal and coronal plane balance, and to prevent curve decompensation1,2,8. In a retrospective multicenter study, Kuklo et al. compared open anterior spinal fusion with standard posterior spinal fusion with use of a variety of fusion techniques2. They found that in 298 "single overhang" (main thoracic) curves, the most caudad instrumented vertebra was at the level of the stable vertebra, or more cephalad, in 97% of Lenke type-1A and 1B curves that were treated with anterior spinal fusion but in only 65% of such curves that were treated with posterior spinal fusion. The authors pointed out, however, that the main limitation of their study was that the data did not include all-pedicle-screw constructs2. In the present study, we found that, compared with posterior spinal fusion with thoracic pedicle screws, the most caudad instrumented vertebra in video-assisted thoracoscopic surgery is approximately 0.81 level above the Cobb angle end vertebrae, 0.7 level above the neutral vertebra, and 0.3 level above the stable vertebra. These results reflect the fact that the selection of caudad fusion levels with use of contemporary and established criteria continues to be different between anterior and posterior spinal fusion constructs. Nonetheless, on the basis of these strict criteria, our results suggest that, despite the ability to achieve greater percentage curve correction with use of posterior spinal fusion with thoracic pedicle screws, it is still possible to spare an average of approximately one lumbar level with use of video-assisted thoracoscopic surgery. While the long-term importance of sparing one lumbar segment remains unknown, the rationale behind the preservation of the maximum number of lumbar motion segments stems from previous studies, which have shown an increased prevalence of low-back pain in patients with extension of the fusion into the lower lumbar spine2,47-52.
With respect to perioperative outcomes, patients undergoing video-assisted thoracoscopic surgery also had a significantly reduced estimated blood loss in comparison with those managed with posterior spinal fusion with thoracic pedicle screws, although the proportion of patients requiring a blood transfusion was not different. The lack of a difference in the transfusion rate may reflect the fact that intraoperative blood salvage was used for the posterior spinal fusion group but not for the video-assisted thoracoscopic surgery group. All patients received autologous blood only. In our study, the video-assisted thoracoscopic surgery group had an average estimated blood loss of 371 mL, which is comparable with the results from other series9,15,19,23. There was a significantly greater operative time associated with video-assisted thoracoscopic surgery (326 compared with 246 minutes), which is also consistent with other reports in the literature9,15. The additional time for rib autograft harvest with video-assisted thoracoscopic surgery may have contributed to these differences. The substantial learning curve associated with video-assisted thoracoscopic surgery has been previously described, with significant reduction in operative time and improvement in the percentage curve correction noted after the surgeon's first twenty-eight cases19,23.
Analysis of patient-based outcomes with use of the SRS-22 outcome questionnaire revealed no significant differences between the groups. This finding is in contrast with those of our previous study comparing video-assisted thoracoscopic surgery with posterior spinal fusion with use of hook or hybrid instrumentation, in which we found that the video-assisted thoracoscopic surgery group, on the average, had significantly improved mean total subscores at the time of follow-up9. We also found no differences in the angle of trunk rotation measurements between the groups, indicating that both video-assisted thoracoscopic surgery and posterior spinal fusion with thoracic pedicle screws were able to correct a clinically appreciable rib-hump deformity.
Diminished pulmonary function following anterior spinal surgery is one of the main sources of morbidity and has served as an impetus for the development of more minimally invasive anterior approaches such as video-assisted thoracoscopic surgery11,14,27,53. In the present study, patients in the video-assisted thoracoscopic surgery group experienced an improvement from baseline in terms of both peak flow and vital capacity at two years. However, while the patients in the posterior spinal fusion group experienced significantly greater improvement in peak flow, there was no difference in the degree of improvement in vital capacity. Analysis at the two-year time point (without baseline preoperative values for comparison) for the percentage predicted values of forced expiratory volume and forced vital capacity showed no differences between the groups.
The complication rate for both groups was acceptably low. In the video-assisted thoracoscopic surgery group, one patient had instrumentation failure related to an improper activity level in the early postoperative period. While the reported rate of rod breakage associated with video-assisted thoracoscopic surgery as reported in the literature is still about 2%, no rod breakages have occurred to date in our series, although the potential for this complication is another limitation of this technique. In an earlier reported series, rod breakage occurred with titanium implants9. Stainless steel implants were used in the present series9. In the present study, there were three complications in the posterior spinal fusion group. Two minor complications (superficial wound infection and temporary respiratory depression) resolved without the need for any surgical intervention. One patient had a misplaced pedicle screw that resulted in postoperative radicular pain, which resolved after revision of the instrumentation. A freehand technique in which fluoroscopy was not employed was utilized for the placement of all pedicle screws29,54. This case further highlights the challenges associated with upper lumbar pedicle screw placement. A recent anatomic study of patients without deformity revealed that the L1 pedicle has a smaller pedicular isthmus than T10, T11, T12, and L2, emphasizing the caution that must be taken with upper lumbar screw placement55. No complications resulted in persistent problems for the patients in either group.
One of the major limitations of the present study is that the subjects were not randomized to either treatment option; however, they were matched by curve size and type, age, and sex. In this series, polyaxial screws were utilized. Currently, many scoliosis surgeons utilize monoaxial or uniplanar screws that may or may not result in more powerful axial plane derotation.
In conclusion, for structural thoracic adolescent idiopathic scoliosis curves, video-assisted thoracoscopic surgery is a viable alternative to open anterior and posterior spinal fusion techniques. Although it requires a substantial training period and learning curve, video-assisted thoracoscopic surgery offers some distinct advantages over standard posterior spinal fusion and thoracic pedicle screw-based constructs and warrants additional study as a minimally invasive surgical alternative for the treatment of adolescent idiopathic scoliosis. Instrumentation rigidity and adequacy may be a concern with the current state of the video-assisted thoracoscopic surgery implants. 