Spino-Pelvic Parameters After Surgery Can be Predicted

Abstract

Prospective and retrospective radiographic study of adult patients with spinal deformities. Construct predictive models for pelvic tilt (PT) and global sagittal balance (sagittal vertical axis [SVA]) and evaluate the effectiveness of these predictive models against a group of patients after pedicle subtraction osteotomy. Spinal balance involves a complex interaction between the pelvis and vertebral column. In the setting of adult spinal deformity, prediction of postoperative alignment can be challenging. The study included 219 adult patients treated for spinal deformity. Full-length standing films were available for all subjects. Multilinear models with a stepwise condition were used on the first group of patients (n = 179) to predict PT and global sagittal balance (measured by the SVA). Prediction models were then applied on a second group of patients (n = 40) to estimate postoperative radiographic parameters after pedicle subtraction osteotomy surgery. Differences between estimated parameters and real values were evaluated. Multilinear regression analysis applied on the first group of patients led to a predictive formula for PT (r = 0.93, standard error = 4.4°) using the following parameters: pelvic incidence, maximal lordosis, and maximal kyphosis. These parameters with the addition of the predicted PT were then used to predict the SVA (r = 0.89, standard error = 32 mm). Validation of predictive models (second group of patients) used pelvic incidence and postoperative sagittal curves. Postoperative PT was predicted with a mean error of 4.3° (SD 3.5°) and postoperative SVA was predicted with a mean error of 29 mm (SD = 23 mm). This is the first study to develop and validate pragmatic predictive models for key spino-pelvic parameters (PT and SVA) in the setting of adult spinal deformity. Using a morphologic pelvic parameter (pelvic incidence) and spinal parameters modifiable through surgery (lumbar lordosis and thoracic kyphosis), postoperative sagittal alignment can be predicted.