Surface vs. skeleton: the relationship between surface topographic and radiographic measurements of shoulder symmetry in patients with scoliosis.

Abstract

This study evaluates the relationship between existing radiographic measurements of shoulder asymmetry with novel surface topographic (ST) measurements, hypothesizing that these relationships will be weak. Data were gathered from a prospectively collected registry of patients who underwent ST scanning at a single tertiary care institution. Inclusion criteria were diagnosis of juvenile or adolescent idiopathic scoliosis, age 11-21, same-day ST and EOS radiographic evaluation. Twelve radiographic variables that evaluate shoulder height were measured, as well as curve magnitudes and vertebral axial rotation. ST data were collected using the 3dMDbody scanning system. Three ST measurements of shoulder symmetry were evaluated: ST-based AC angle (the angle between a line made between the surface of the AC joints and a line parallel to the ground), Shoulder Normal Asymmetry angle (the angle between the mirrored normals to the planes defined by the jugular notch, vertebral prominence, and AC joint), and Shoulder Volume Asymmetry Index (difference in volumes between the right and left shoulder). Univariate, followed by a stepwise multivariate linear regression was performed to evaluate the correlations of the radiographic measurements to the ST-based measurements. Correlation categories: weak (x < 0.4), moderate (0.4 ≤ x < 0.6), strong (0.6 ≤ x < 0.8), and very strong (x ≥ 0.8). 141 patients with idiopathic scoliosis were evaluated (65.2% female, mean age 15.1years, mean BMI 20.5kg/m2, with mean maximum curve 44.7 degrees). ST-based AC angle had moderate-to-strong correlations with several radiographic measurements, while ST-based Shoulder Normal Asymmetry angle and Shoulder Volume Asymmetry Index had weak or no correlation with individual radiographic measures. Multivariate models created using a combination of radiographic variables demonstrated a strongly positive correlation between radiographic variables and ST-based AC angle (R = 0.678) and moderately positive correlations with ST-based Shoulder Normal Asymmetry Angle (R = 0.488), and ST-based Shoulder Volume Asymmetry Index (R = 0.514). Radiographic measurements may be acceptable stand-ins for two-dimensional ST measurements such as AC angle, but not for more complex shoulder measurements based on three dimensions. This study demonstrates the inadequacy with which radiographic assessments evaluate shoulder height asymmetry and highlights the use of ST measurements.