Periosteal versus true cross-sectional geometry: A comparison along humeral, femoral, and tibial diaphyses

Cross-sectional geometric (CSG) properties of human long bone diaphyses are typically calculated from both periosteal and endosteal contours. Though quantification of both is desirable, periosteal contours alone have provided accurate predictions of CSG properties at the midshaft in previous studies. The relationship between CSG properties calculated from external contours and "true" (endosteal and periosteal) CSG properties, however, has yet to be examined along the whole diaphysis. Cross-sectional computed tomography scans were taken from 21 locations along humeral, femoral, and tibial diaphyses in 20 adults from a late prehistoric central Illinois Valley cemetery. Mechanical properties calculated from images with (a) artificially filled medullary cavities ("solid") and (b) true unaltered cross-sections were compared at each section location using least squares regression. Results indicate that, in this sample, polar second moments of area (J), polar section moduli (Z _p), and cross-sectional shape (I_max/I_min) calculated from periosteal contours correspond strongly with those calculated from cross-sections that include the medullary cavity. Correlations are high throughout most of the humeral diaphysis and throughout large portions of femoral and tibial diaphyses (R² = 0.855-0.998, all P < 0.001, %SEE ≤ 8.0, %PE ≤ 5.0), the major exception being the proximal quarter of the tibial diaphysis for J and Z_p. The main source of error was identified as variation in %CA. Results reveal that CSG properties quantified from periosteal contours provide comparable results to (and are likely to detect the same differences among individuals as) true CSG properties along large portions of long bone diaphyses.