Prospective Multicenter Evaluation of a New Predictive Model for the Progression of Adolescent Idiopathic Scoliosis

Overview

Scoliosis is a three-dimensional deformity affecting the orientation and position of the spine. Locally, the shape of the vertebra is also affected. The most common form is adolescent idiopathic scoliosis (AIS) with a prevalence of 1-3% affecting primarily young adolescent females. AIS can either be treated using a brace and in some cases necessitate surgical correction to prevent progressive deformity. Risk factors for progression include female gender, curve magnitude and location, skeletal maturity and growth velocity. However, these risk factors have been shown to be inconsistent in predicting curve progression. Over the past 6 years, the investigators have developed a predictive model of the final Cobb angle in AIS based on 3D spinal parameters. This analysis was based on a prospective cohort of 195 patients that were enrolled upon their initial visit and followed until maturity. This predictive model has a determination coefficient of 0.702. The proposed new study aims at refining and testing the external validity of this model in a larger cohort. The next step towards using the new model in the clinical setting is to redesign the model and to externally validate the model by measuring the agreement between the new method and the traditional Cobb angle at maturity in a larger multicenter study. The objective of this study is to characterize the risk of scoliosis progression based on local three-dimensional vertebral and pelvic measurements present on initial evaluation. Three-dimensional reconstructions will be derived from stereo-radiographs acquired with a new biplanar low-dose radiographic system installed in all 8 clinical sites (EOS system, EOS-Imaging, Paris). These calibrated radiographs will then be used to reconstruct the vertebrae and intervertebral disks at each level as well as the geometry of the pelvis. A series of local and regional parameters will then be calculated from these 3D reconstructions. Correlation analysis will help determine if intervertebral disk wedging, vertebral wedging, transverse plane rotation or pelvic geometry can be used as early predictors of curve progression in AIS. Identifying a new 3D measure of scoliosis associated with rapid curve progression could help predict which curves need early treatment to prevent further progression. The ultimate goal of this research project will be to validate this new predictive model and finally transfer this new predictive tool in the hands of clinicians treating AIS.

By identifying the patients for whom the model predicts a significant progression the investigators could focus their efforts on being more pro-active towards brace wear for those at risk of progression. With the advent of new growth modulation techniques, these patients could also potentially receive early management of their curves to prevent further progression and diminish the need for surgical correction and fusion of their curve. Alternatively, by identifying the patients that will not progress immediately at the first visit, the investigators could arguably decrease their number needed to treat from 4 to 2 by removing those patients that won't progress no matter what their treatment is. So, the prediction model will not only have an impact for the patients at risk of progression but will also impact those that are braced despite having a low risk of progression. Patient enrollment: Patients who meet the inclusion criteria will be offered enrollment in the project. If the family wishes to enroll in the project, the informed consent process will be undertaken and the patient will be enrolled. If Patients need to be treated with brace No changes in the usual treatment and monitoring will be instituted. Therefore, patients will be allowed to undergo brace treatment. The investigators are aware that brace treatment could have an impact on curve progression but these patients are still included as progression despite brace treatment may give meaningful information regarding these difficult-to-treat curves. Patient under brace treatment will have to remove brace at least the night before their appointment and follow-up spine radiographs. Also the protocol requires that patients have at least one spine radiograph out of brace every 12 months. Bracing compliance will be evaluated as a specific covariable in the upcoming study. To better understand the impact of bracing on progression, the investigators will be using brace monitors, pressure sensitive transducers that can monitor brace wear and brace effectiveness, in their multicenter study and use the data generated by these monitors as a co-factor in an improved prediction model. If patients need to have surgery Although patients showing sufficient curve progression to undergo surgical correction will not be kept in the prospective cohort after surgery, their data will still be analyzed to assess the correlation between local 3-D measurements and curve progression. -Data collection Data collected at first visit will include: * Demographic data * past medical history * gender * body-mass index (weight and height) * family history * sexual maturity (start of menses for girls) * rib prominence * initial radiographic evaluation. Data collected at each follow-up visit (ideally every 6 months) * body-mass index (weight and height) * menarchal status, * skeletal maturity * rib prominence. * Radiographic evaluation (at least every 12 months during the study period, without brace). * Image acquisition and transfer to central measurement site. Image acquisition is standardized between sites according to standard operating procedures. All images are acquired with the EOS™ system in the same fashion based on the position proposed by Horton et al. that standardized lateral radiographs by using a hand-on-clavicle position. The radiographs are all taken in the same way thus minimizing variability. For the centers without an EOS™ system for part or for the entire study, utilization of calibration belts will allow radiographs to be taken in a calibrated environment. The position used for image acquisition is the same as for the EOS™ system. * Three-dimensional reconstructions of the spine The stereo radiographic images are used to create an external three-dimensional representation of the vertebral body using a specific algorithm. First, a spline is fitted through the centers of the vertebral bodies both on the PA and lateral views. The information from both images is then used to reconstruct a 3-D spline or curve which will act as a rough three-dimensional scaffold onto which the local vertebral and intervertebral reconstructions will take place. The vertebral endplates are represented by a crude preliminary model of the spine by using a set of cubic templates roughly representing each vertebral body stacked on top of one another to form the spinal column. A global configuration of the deformable spine model is thus described for each cubic template associated with each vertebral level. The final reconstruction can then be completed using a priori knowledge from a database of scoliotic and normal vertebrae that were measured by the investigator's research group. The a priori knowledge model relies on the description of each vertebra by a deformable model, which incorporates statistical knowledge about its geometrical structure and its pathological variability. The statistically optimized reconstructions will be used to determine the intervertebral disk shape at each level. * Three-dimensional stereoradiographic reconstructions of the pelvis The pelvis will be reconstructed in 3-D for all normal and AIS subjects using a combination of the Non-Stereo Corresponding Points (NSCP) and Non-Stereo Corresponding Contours (NSCC) methods, which were successfully by the investigator's team to reconstruct the spine in 3D. Preliminary evaluation of the technique gave an overall accuracy of 1,6 mm, which is adequate for the calculation of clinical geometric indices of the pelvis. Four steps are required with this 3D reconstruction technique: 1. Identification of seven specific regions of the pelvis; 2. Display of a preliminary model with 45 control points. The control points can be modified in real time by the user (NSCP); 3. Interactive identification of regional contours (NSCC); 4. Generation of the personalized 3D pelvic model by deforming a generic 3D pelvic model using 3D geometrical kriging. * Local three-dimensional measurements of the vertebrae and disks: The calculated parameters were divided into six categories. Each category refers to global (whole spine), regional (scoliotic segment) and local (vertebra) descriptors. Vertebra centroid is understood as the halfway point between the centers of the two endplates of the vertebra. The local vertebra axis system is defined by the Scoliosis Research Society (SRS) 3D terminology group: the origin is at the centroid of the vertebral body, the local 'z' axis passes through the centers of the upper and lower endplates, and 'y' axis is parallel to a line joining similar landmarks on the bases of the right and left pedicles. 1. Cobb Angles: Cobb angles defined as the angle between the upper and lower endplate of the respective end vertebrae of a curve. Cobb angle was measured in the frontal plane, in the plane of maximal deformation in 3D and in the sagittal plane for thoracic kyphosis (T4-T12) and lumbar lordosis (L1-S1). 2. Plane of maximal deformation: Axial angle of the plane in which the Cobb angle is maximal. 3. Three-dimensional wedging of vertebral body and disk: Wedging of the apical vertebral body in the plane of maximal deformation (3D plane) and mean maximal 3D wedging of the two apical intervertebral disks. Maximal 3D wedging represents the wedging measured in the plane, wherein the wedging value is maximal around the vertical axis. If apex was a disk, then the mean of the 3D wedging of both apical vertebral bodies was calculated and only the 3D wedging of the apical disk was documented. 3D disk wedging was analyzed for all levels of the spine (from T1-T2 to L4-L5). 4. Axial intervertebral rotation of the apex, upper and lower junctional level and thoracolumbar level: Rotation between two adjacent vertebrae at upper, apical, and lower curve levels and thoracolumbar junction (T12-L1) in the axial plane according to the inferior local vertebrae reference. 5. Torsion: Mean of the sum of intervertebral axial rotation (measured according to the local referential of the inferior vertebrae) of the two hemicurvatures of the curve (between upper end vertebra and apex and between lower end vertebra and apex). 6. Slenderness (local T6, T12 and L4 and regional T1-L5): Ratio between the height (distance between the superior and inferior endplates at the center of the vertebrae) and the width (measured at the center of the vertebrae using a line perpendicular to the height line in medio lateral direction) of the vertebral body for T6, T12 and L4 vertebrae. Ratio between the length of the spine from T1 to L5 and the mean of the width of vertebral bodies of T6-T12 and L4. The same measurement was made by replacing the width by the depth (a line perpendicular to the height line at the center of the vertebra in an anteroposterior direction). * Geometric pelvic indices: Geometric pelvic indices will be calculated automatically with the generation of the 3D pelvic model. Indices describing the orientation of the pelvis in 3D (positional indices), based on the line joining the center of both femoral heads (hip axis), will be calculated first. Pelvic axial rotation is the orientation of the hip axis around the vertical axis (gravity line) as viewed in the transverse plane. Pelvic obliquity is the orientation of the hip axis around the horizontal line as viewed from the coronal plane. Pelvic tilt is the orientation of the pelvis around the medio-lateral axis of the pelvis as viewed from the sagittal plane. In this last case, at least another landmark from the pelvis (such as the center of the upper sacral plate) needs to be identified in addition to the hip axis. The sacral slope (angle between the upper sacral plate and the horizontal line is also calculated since it is highly correlated with lumbar lordosis. Sacral obliquity is the angle between the upper sacral plate and the horizontal line. Morphological indices of the pelvis (not dependent on the patient's position) are computed in a reference coordinate system based on the hip axis orientation, in order to eliminate the effect of pelvic axial rotation and obliquity. This means that transformation into the reference coordinate system allows calculation of morphological indices in the true coronal, axial and sagittal planes of the pelvis. The investigators selected the pelvis as their reference system based on the pelvic vertebra principle from Dubousset, which states that the pelvis can be viewed as a separate vertebra. As described earlier, numerous morphologic parameters of the pelvis have been used in the past, as detailed in Tables I to III. Of these, after a thorough review of the literature, the investigators selected what they consider the 7 most pertinent specific indices. The investigators based their decision on the potential association of the pelvic indices with the pathogenesis of AIS. Since they demonstrated the close relationship between pelvic incidence and lumbar lordosis in AIS, the investigators selected morphological indices of the pelvis that are correlated with the spine geometry.