General Session: Adult Spinal Deformity
Presented by: P. Passias - View Audio/Video Presentation (Members Only)
P. Passias(1), C. Oh(1), C. Jalai(1), N. Worley(1), R. Lafage(2), J. Scheer(3), E. Klineberg(4), R. Hart(5), H.J. Kim(2), J. Smith(6), V. Lafage(2), C. Ames(7), International Spine Study Group
(1) NYU Hospital for Joint Diseases, Department of Orthopaedic Surgery, NYC, NY, United States
(2) Hospital for Special Surgery, Department of Orthopaedics, New York, NY, United States
(3) Northwestern University, Department of Neurosurgery, Chicago, IL, United States
(4) University of California, Davis, Department of Orthopaedic Surgery, Sacramento, CA, United States
(5) Oregon Health & Science University, Department of Orthopaedic Surgery, Portland, OR, United States
(6) University of Virginia, Department of Neurosurgery, Charlottesville, VA, United States
(7) University of California, San Francisco, Department of Neurosurgery, San Francisco, CA, United States
Introduction: Cervical deformity (CD) after adult spinal deformity (ASD) surgical correction has been defined using the following measurements: CL>20°, C2-C7 SVA>40mm, and C2-C7 kyphosis>10°. While several studies have analyzed predictors of developing CD, few have defined and identified predictors of optimal cervical alignment (CA) following thoracolumbar surgery. This study aims to use advanced predictive modeling to identify patient characteristics, radiographic, and surgical variables that predict reaching an outcome threshold of sub-optimal cervical alignment following ASD surgery.
Materials/Methods: This study retrospectively reviewed a prospective multicenter database of surgical ASD patients with baseline and 2-year follow-up. Post-op CA at 2-years was defined according to the following radiographic criteria: 0°≤T1S-CL≤20°, 0mm≤C2-C7 SVA≤40mm, or C2-C7 lordosis>0°. Three thresholds of classifying CD were defined using these criteria: T1) only 1 criterion, T2) only 2 criteria, T3) all 3 criteria. Variables considered were baseline demographic, radiographic, and surgical factors. To establish a prediction model, multivariable logistic regression models augmented in a stepwise manner were performed using a bootstrap resampling procedure. Internal validation of the predictive model was performed by constructing the receiver operating characteristic (ROC), and the area under the curve (AUC).
Results: 225 surgical ASD patients were included. 208 patients (92.4%) were grouped in T3, while 17 (7.6%) fell outside all three CA criteria ranges. Patients in both groups were similar regarding mean age (56.02 vs. 61.47 years, p=0.150) and BMI (27.10 vs. 27.64 kg/m2, p = 0.716), but patients that met all 3 CA criteria had an increased prevalence of females (88.9% vs. 66.7%, p=0.017). The final predictive model had an AUC of 89.22% (DeLong) and included the following variables: C2 sacral slope, C2-T3 CL, T1S-CL, C2-C7 CL, Pelvic Tilt, C2-S1 SVA, PI-LL, and number of SPO´s during index. In this model, the following variables were identified as predictors of poor CA: number of SPO´s (OR: 1.336, p=0.017), and C2-T3 CL (OR: 1.048, p=0.005). Models for predictors of all thresholds are reported in Table 1. The AUC of the final model was 89.22% (95% CI: 97.49%-80.96%), indicating excellent predictive discrimination.
Conclusions: This study created a statistical model that predicts good 2-year post-operative CA in ASD surgical patients. Using T3 (patients meeting all 3 CA criteria) was the most effective model for predicting poor cervical alignment, and included increased baseline C2-T3 angle and increased Smith-Peterson osteotomies during index. This study could be used to aid surgeons in patient counseling efforts and to direct future research. Additional work is necessary to validate this model in practice.