Modern Concepts in Sagittal Curve Measurement: Comparison of Spline-Based and Fixed Landmark Measurements in a Cohort of 1520 Healthy Subjects.

Abstract

Retrospective multicenter study. Our objective was to compare the spline-based measurement of sagittal spinal curvatures to fixed landmarks in a normative population. Recent research has stressed the importance of considering sagittal curvature in its entirety using a spline reconstruction. To date, no data supports the superiority of this method in comparison to classic measurement methods. Full-spine biplanar radiographs of subjects over 20 years old who had normal radiographs were analyzed. Thoracic kyphosis (TK) and lumbar lordosis (LL) were measured following two modalities: either using predefined landmarks (TK T1T12 , TK T5T12 , and LL L1S1 ) or spline-based measurement (TK Spline and LL spline ). A total of 1520 subjects were included (mean 54yo). The mean difference between TK spline and TK T1T12 was 1.4° and between TK spline and TK T5T12 was 11.7° ( P <0.001). LL spline was significantly larger than LL L1S1 (55° vs . 54°, P<0.001 ). LL spline and LL L1S1 were correlated ( R =0.950, P <0.001). Pelvic incidence had no influence on the difference between LL spline and LL L1S1 ( R =-0.034, P= 0.184). Using LL L1S1 measurements, LL was underestimated in 17% of the cases. The comparison of outlier distributions according to age groups ( P =0.175), sex ( P =0.937), or pelvic incidence groups ( P =0.662) found no difference. There were significantly more outliers in Roussouly type 1 compared with other types (56%, P <0.001). Our results suggest that the use of TK T1T12 and LL L1S1 is acceptable to assess spinal sagittal curvatures. However, TK T5T12 is not accurate for the thoracic curve and should be used with caution. LL L1S1 can be used to accurately assess the lumbar curve, except in Roussouly type 1.