General Session: Adult Spinal Deformity-2

Presented by: J.J. Varghese - View Audio/Video Presentation (Members Only)


J.K. Henry(1), J.J. Varghese(2,3), F. Schwab(2), T. Protopsaltis(1), J. Smith(4), C. Ames(5), M. Kelly(6), J. Gum(7), G. Mundis(8), R. Hostin(9), R. Hart(10), K. Kebaish(11), D. Burton(12), S. Bess(13), V. Lafage(2), International Spine Study Group

(1) NYU Hospital for Joint Diseases, New York, NY, United States
(2) Hospital for Special Surgery, New York, NY, United States
(3) State University of New York, Downstate Medical Center, Brooklyn, NY, United States
(4) University of Virginia, Charlottesville, NH, United States
(5) UCSF, San Francisco, CA, United States
(6) Washington University in St. Louis, St. Louis, MO, United States
(7) Norton Leatherman Spine Center, Louisville, KY, United States
(8) San Diego Center for Spinal Disorders, La Jolla, CA, United States
(9) Baylor Scott & White Health, Plano, TX, United States
(10) OHSU, Portland, OR, United States
(11) Johns Hopkins Hospital, Baltimore, MD, United States
(12) University of Kansas Hospital, Kansas, KS, United States
(13) Rocky Mountain Scoliosis & Spine, Denver, CO, United States


Purpose: To investigate the predictability of post-operative alignment in primary vs. revision adult spinal deformity (ASD) patients.

Introduction: Studies have developed prediction models for corrected alignment based on pre-op parameters and age, but the predictability of post-op alignment in primary vs revision patients (pts) given their baseline (BL) and post-op fusion length has not been investigated.

Methods: ASD pts undergoing ≥5 level fusion were classified according to fusion length (None/Short [NS]: < 5 levels below L1; Lower Thoracic [LT]: ≥5 levels, UIV at/below T7; Upper Thoracic [UT]: ≥5 levels, UIV above T7) at BL and 1yr. Pts were propensity matched by BL global alignment and grouped by change in fusion length/location from BL to 1yr. Regression models used age, PI, BL thoracic kyphosis (TK), and post-op lumbar lordosis (LL) to predict 1yr alignment (PTSVA). Radiographic parameters and variance were compared.

Results: From 789 pts, 222 were included after propensity matching with 74 in each group (NS, LT, UT). Of the 3 BL fusion types, LT and UT fit better to the PT prediction model (R2=0.72, 0.65) than NS (R2=0.59). In sub-group analysis, LT-to-LT and LT-to-UT had the strongest PT predictions (R2=0.81, 0.75); NS-to-UT had the worst (R2=0.51). SVA was best predicted by LT-to-LT (R2=0.56). There were significant differences among groups in the variability of correction from BL to post-operative in LL, TK, and SVA (all P< 0.04). NS-to-UT had the most variability for all parameters, while LT-to-LT had the narrowest range. Predicted PT/SVA using change in fusion length were similar to post-op values (Table).

Conclusions: Postoperative SVA and PT can be predicted with variable accuracy depending on fusion length and termination. Pts with previous fusion to the LT spine have the most reliable post-operative alignment; pts with no/short previous fusion undergoing fusion to the UT region have the greatest variability. Reduced variability and improved technology to guide intra-op surgical alignment changes may improve complex realignment surgery globally.