Three-dimensional classification of thoracic scoliotic curves.

STUDY DESIGN Three-dimensional (3D) characterization of the thoracic scoliotic spine (cross-sectional study). OBJECTIVES To investigate the presence of subgroups within Lenke type-1 curves by evaluating the thoracic segment indices extracted from 3D reconstructions of the spine, and to propose a new clinically relevant means (the daVinci representation) to report 3D spinal deformities. SUMMARY OF BACKGROUND DATA Although scoliosis is recognized to be a 3D deformity of the spine its measurement and classification have predominantly been based on radiographs which are 2D projections in the coronal and sagittal planes. METHODS Thoracic segment indices derived from 3D reconstructions of coronal and sagittal standing radiographs of 172 patients with right thoracic adolescent idiopathic scoliosis, reviewed by the 3D Classification Committee of the Scoliosis Research Society, were analyzed using the ISOData unsupervised clustering algorithm. Four curve indices were analyzed: Cobb angle, axial rotation of the apical vertebrae, orientation of the plane of maximum curvature of the main thoracic curve, and kyphosis (T4-T12). No assumptions were made regarding grouping tendencies in the data nor were the number of clusters predefined. RESULTS Three primary groups were revealed wherein kyphosis and the orientation of the PMC of the main thoracic curve were the major discriminating factors with slight overlap between groups. A small group (G1) of 22 patients having smaller, nonsurgical (minor) curves was identified. Although the remaining patients had similar Cobb angles they were split into 2 groups (G2: 79 patients; G3: 71 patients) with different PMC (G2: 65 degrees -81 degrees ; G3: 76 degrees -104 degrees ) and kyphotic measures (G2: 23 degrees -43 degrees ; G3: 7 degrees -25 degrees). CONCLUSION Two distinct subgroups within the surgical cases (major curves) of Lenke type-1 curves were found thus suggesting that thoracic curves are not always hypokyphotic. The ISOData cluster analysis technique helped to capture inherent 3D structural curve complexities that were not evident in a 2D radiographic plane. The daVinci representation is a new clinically relevant means to report 3D spinal deformities.