More severe thoracic idiopathic scoliosis is associated with a greater three-dimensional loss of thoracic kyphosis.

STUDY DESIGN Retrospective. OBJECTIVES Utilize three-dimensional (3D) measurements to assess the relationship between thoracic scoliosis severity and thoracic kyphosis in a large, multicenter cohort, and determine impact of 3D measurements on adolescent idiopathic scoliosis (AIS) curve classification. Research has demonstrated differences in two-dimensional (2D) and 3D assessment of the sagittal plane deformity in AIS. A prior smaller, single-institution study demonstrated an association between scoliosis severity and loss of 3D thoracic kyphosis. METHODS Data included retrospective compilation of prospectively enrolled bracing candidates and prospectively enrolled surgical candidates with thoracic AIS. Analysis included two groups based on thoracic curve magnitude: moderate (20-45°) and severe (> 45°). Imaging was performed using 2D radiographs. 3D thoracic kyphosis was calculated using a 2D to 3D conversion formula. Kyphosis was categorized according to the Lenke classification sagittal plane modifier. RESULTS Analysis included 3032 patients. 2D kyphosis was significantly less in the moderate group (21 ± 12 vs 23 ± 14, p = 0.028). However, estimated 3D kyphosis was significantly greater in the moderate group (13 ± 10 vs 5 ± 12, p < 0.001). In the moderate group, the rate of normokyphosis was 78% with 2D measures and 61% with 3D measures of T5-T12 kyphosis. In the severe group, this rate changed from 72 to 32% with use of 2D and 3D measures, respectively. In the moderate group, 16% of patients were classified as hypokyphotic using 2D measures while this rate increased 38% with 3D measures (p < 0.001). In the severe group, this rate changed from 18 to 68% using 2D and 3D measures, respectively (p < 0.001). CONCLUSIONS Increased coronal curve severity was associated with decreased thoracic kyphosis. Hypokyphosis was more pronounced in 3D. 2D radiographs increasingly underestimate kyphosis with increasing coronal severity. Assessment of sagittal alignment from 2D radiographs can be improved with a 2D-3D conversion formula. Findings indicate potential for classification system improvement with use of 3D sagittal plane measurements. LEVEL OF EVIDENCE IV.