23 – PREDICTIVE ANTHROPOMETRIC MEASUREMENTS FOR HUMERAL HEAD CURVATURE

Abstract

Purpose: Osteochondral allograft transplantation for the treatment of osseous defects to the humeral head has recently grown in popularity. Because only a portion of the articulating surface of the humeral head is replaced, conformity of the allograft to the native surface is imperative to restore the natural geometry of the joint. To achieve proper conformity, it is essential that the curvature of the humeral head of the allograft tissue match that of the native tissue. Curvature determination is also important for shoulder replacement procedures. Curvature of the humeral head is difficult to directly measure in allograft specimens. As a result, predictive measurements, such as the maximum length of the humerus are used to predict this curvature. The purpose of this study was to investigate the value of various anthropometric measurements for predicting humeral head curvature. We hypothesized that the maximum length of the humerus would be the most predictive of humeral curvature.

Method: 60 (28 female, 32 male) cadaveric humeri were obtained from the Hamann-Todd Human Osteological Collection. Specimens ranged from 20 to 35 years of age at the time of death (27.9 ± 4.5, mean ± SD). Specimens from this collection include height and weight as collected at the time of death. All specimens were scanned with a 3-dimensional laser scanner (NextEngine, Santa Monica, California, USA). This scanner has been shown to be accurate to within 0.005 inches. Linear measurements (maximum humeral length, epicondylar breadth) were made according to the recording standards for skeletal remains. Both measurements were made by choosing points on the 3-dimensional scan, rather than the traditional osteometric board. Humeral head curvature was determined by a custom computational code to fit a sphere to the articulating surface of the humerus. Data analysis was performed in Minitab (version 13, State College, PA, USA). A linear regression was performed for each predictive measurement. A stepwise linear regression with forward and backward substitution was performed for the most predictive variables from the initial linear regression.

Results: The most predictive factors (R^2 > 0.5) were epicondylar breadth, height, maximum humeral length, and gender. Based on the linear regression coefficients, these four factors (all normalized) were included in a forward and backward stepwise regression (alpha to enter and remove = 0.15). The resulting equation (shown below) had an R^2 values of 0.807. Humeral Diameter = 0.894 + 0.048*(epicondylar breadth) + 0.043*height – 0.020*gender

Conclusion: Of the predicted measurements evaluated, patient height, epicondylar breadth, and gender were most correlated with humeral head curvature. Including these three factors in a linear regression model increased the R2 value to 0.807. If only a single measurement can be used to size the humeral curvature, patient height will give approximately the same accuracy as epicondylar breadth, and can more easily be obtained.