Is the Combination of Convex Compression for the Proximal Thoracic Curve and Concave Distraction for the Main Thoracic Curve Using Separate-rod Derotation Effective for Correcting Shoulder Balance and Thoracic Kyphosis?

BACKGROUND: Posterior correction of the proximal thoracic curve in patients with adolescent idiopathic scoliosis has been recommended to achieve shoulder balance. However, finding a good surgical method is challenging because of the small pedicle diameters on the concave side of the proximal thoracic curve. If the shoulder height can be corrected using screws on the convex side, this would appear to be a more feasible approach. QUESTIONS/PURPOSES: In patients with adolescent idiopathic scoliosis, we asked: (1) Is convex compression with separate-rod derotation effective for correcting the proximal thoracic curve, shoulder balance, and thoracic kyphosis? (2) Which vertebrum is most appropriate to serve as the uppermost-instrumented vertebra? (3) Is correction of the proximal thoracic curve related to the postoperative shoulder balance?
METHODS: Between 2015 and 2017, we treated 672 patients with scoliosis. Of those, we considered patients with elevated left shoulder, Lenke Type 2 or 4, or King Type V idiopathic scoliosis as potentially eligible. Based on that, 17% (111 of 672) were eligible; 5% (6 of 111) were excluded because of other previous operations and left-side main thoracic curve, 22% (24 of 111) were excluded because they did not undergo surgery for the proximal thoracic curve with only pedicle screws, 21% (23 of 111) were excluded because the proximal thoracic curve was not corrected by convex compression and separate rod derotation, and another 3% (3 of 111) were lost before the minimum study follow-up of 2 years, leaving 50% (55 of 111) for analysis. During the study period, we generally chose T2 as the uppermost level instrumented when the apex was above T4, or T3 when the apex was T5. Apart from the uppermost-instrumented level, the groups did not differ in measurable ways such as age, sex, Cobb angles of proximal and main thoracic curves, and T1 tilt. However, shoulder balance was better in the T3 group preoperatively. The median (range) age at the time of surgery was 15 years (12 to 19 years). The median follow-up duration was 26 months (24 to 52 months). Whole-spine standing posteroanterior and lateral views were used to evaluate the improvement of radiologic parameters at the most recent follow-up and to compare the radiologic parameters between the uppermost-instrumented T2 (37 patients) and T3 (18 patients) vertebra groups. Finally, we analyzed radiologic factors related to shoulder balance, defined as the difference between the horizontal lines passing both superolateral tips of the clavicles (right-shoulder-up was positive), at the most recent follow-up.
RESULTS: Convex compression with separate-rod derotation effectively corrected the proximal thoracic curve (41° ± 11° versus 17° ± 10°, mean difference 25° [95% CI 22° to 27°]; p < 0.001), and the most recent shoulder balance changed to right-shoulder-down compared with preoperative right-shoulder-up (8 ± 11 mm versus -8 ± 10 mm, mean difference 16 mm [95% CI 12 to 19]; p < 0.001). Proximal thoracic kyphosis decreased (13° ± 7° versus 11° ± 6°, mean difference 2° [95% CI 0° to 3°]; p = 0.02), while mid-thoracic kyphosis increased (12° ± 8° versus 18° ± 6°, mean difference -7° [95% CI -9° to -4°]; p < 0.001). Preoperative radiographic parameters did not differ between the groups, except for shoulder balance, which tended to be more right-shoulder-up in the T2 group (11 ± 10 mm versus 1 ± 11 mm, mean difference 10 mm [95% CI 4 to 16]; p = 0.002). At the most recent follow-up, the correction proportion of the proximal thoracic curve was better in the T2 group than the T3 group (67% ± 10% versus 49% ± 22%, mean difference 19% [95% CI 8% to 30%]; p < 0.001). In the T2 group, T1 tilt (6° ± 4° versus 6° ± 4°, mean difference 1° [95% CI 0° to 2°]; p = 0.045) and shoulder balance (-14 ± 11 mm versus -7 ± 9 mm, mean difference -7 mm [95% CI -11 to -3]; p = 0.002) at the most recent follow-up improved compared with those at the first erect radiograph. The most recent shoulder balance was correlated with the correction proportion of the proximal thoracic curve (r = 0.29 [95% CI 0.02 to 0.34]; p = 0.03) and change in T1 tilt (r = 0.35 [95% CI 0.20 to 1.31]; p = 0.009).
CONCLUSION: Using the combination of convex compression and concave distraction with separate-rod derotation is an effective method to correct proximal and main thoracic curves, with reliable achievement of postoperative thoracic kyphosis and shoulder balance. T2 was a more appropriate uppermost-instrumented vertebra than T3, providing better correction of the proximal thoracic curve and T1 tilt. Additionally, spontaneous improvement in T1 tilt and shoulder balance is expected with upper-instrumented T2 vertebrae. Preoperatively, surgeons should evaluate shoulder balance because right-shoulder-down can occur after surgery in patients with a proximal thoracic curve. LEVEL OF EVIDENCE: Level III, therapeutic study.