3D models from EOS imaging to assess axial changes in the lumbar spine after selective thoracic fusion in adolescent idiopathic scoliosis (AIS)

Abstract Purpose Instrumented scoliosis correction to T12/L1 can be a full correction (FC) of a single thoracic curve or a selective thoracic fusion (STF) where there is a significant lumbar curve. This study aims to evaluate the utility and feasibility of 3-dimensional (3D) EOS modelling to quantify pre- and post-operative axial plane changes in the uninstrumented lumbar spine. Methods This study included patients undergoing primary surgery (2018–2021) for AIS Lenke 1 or 3, with the lowest instrumented level at T12-L1 and reconstructable EOS bi-planar images available pre-op, post-op, and at 1-year follow-up. EOS 3D modelling gives the apical rotation and mean rotation form T1-L5 from a neutral pelvis. Results Twenty patients (age 14.3; 7 Lenke 1 A (FC), 13 Lenke 1B–3 C (STF)) were included. Lumbar Cobb correction averaged 54% (1 A), 41% (1B), and 21% (1 C/3 C). Neither the STF nor FC achieved notable correction of apical lumbar rotation (1 A: −5%, p > 0.05, 1B; +0°, p > 0.05, 1 C/3 C; +2.5°, p > 0.05). Uninstrumented average L1-L5 rotation showed no significant change at 1 year for any curves. A significant correlation was observed between lateral bending Cobb angles and L1-L5 average rotation at one year ( p < 0.05). Notably, EOS imaging measured greater axial rotation than PA x-rays, with differences normalising after accounting for pelvic parameters. Conclusion EOS 3D modelling is valuable for visualising the mobile lumbar spine. Our models showed no significant correction of lumbar rotation, and a large impact from pelvic rotation on radiographic measurements. Increased curve flexibility may improve axial correction. The apparent reduction in lumbar rotation on plain radiographs is more likely attributable to pelvic rotation.