A new noninvasive mechanical bending test accurately predicts ulna bending strength in cadaveric human arms

High error rates in the prediction of fragility fractures by bone mineral density have motivated searches for better clinical indicators of bone strength, and the high incidence of non-hip, non-spine fractures has raised interest in cortical bone. The aim of this study was to assess the accuracy of Cortical Bone Mechanics Technology™. CBMT is a new non-invasive 3-point bending technique for measuring the mechanical properties of cortical bone in the ulnas of living humans.

35 cadaveric human arms were obtained from small women and large men ranging widely in age (17 < Age < 99 years) and body size (14 < BMI < 40 kg/m²). Noninvasive CBMT measurements of the flexural rigidity of the ulna bones within these arms (EI_CBMT) were compared to measurements of EI by Quasistatic Mechanical Testing in the ulnas excised from those arms (EI_QMT). Ulna bending strength was also measured by QMT as the peak moment before fracture (M_peak). The open source BoneJ plugin to ImageJ image processing software was used to calculate cortical porosity (CP) in micro-computed tomography images of a 2 mm length of the mid-shaft of each fractured ulna, and the interosseous diameter (IOD) of each ulna was also measured in those images.

EI_CBMT measurements (13 < EI_CBMT < 97 Nm²) explained 99% of the variance in QMT measurements of ulna bending strength (11 < M_peak < 90 Nm), but EI_CBMT was biased high by 30% (p < 0.0001) relative to EI_QMT (11 < EI_QMT < 69 Nm²). After correcting this bias, EI_CBMT and EI_QMT measurements lay along the identity line (y = 1.00x, R² = 0.99, SEE = 3.1 Nm²). Predictions of M_peak by EI_CBMT were less accurate than predictions by EI_QMT (both R² = 0.99; SEE_CBMT = 5.9 Nm vs SEE_QMT = 4.5 Nm, F = 2.92, p = 0.001), but EI_CBMT predictions were substantially more accurate than those by IOD (R² = 0.79; SEE_IOD = 10.6 Nm, F = 3.30, p < 0.001) and CP (R² = 0.35; SEE_CP = 18.9 Nm, F = 10.45, p < 10^-9). Predictions by EI_CBMT were also more accurate than predictions by arm donor height (R² = 0.63; SEE = 14.3 Nm, F = 5.87, p < 10^-6), body weight (R² = 0.77; SEE = 11.1 Nm, F = 3.54, p < 0.001) and BMI (R² = 0.64; SEE = 14.1 Nm, F = 2.39, p < 0.01). In forward stepwise multiple regression beginning with EI_CBMT, only age explained any additional variance in ulna bending strength (ΔR² = 0.3%, F = 8.03, p = 0.008).

Noninvasive CBMT measurements of ulna EI explain 99% of individual differences in QMT measurements of ulna bending strength in cadaveric human arms.