Three-dimensional analysis of the proximal part of the humerus: relevance to arthroplasty.

Abstract

Recreation of normal anatomical relationships may be important to optimize the outcome of proximal humeral arthroplasty. With use of computerized tomographic data and three-dimensional computer modeling, we concurrently studied both extramedullary and intramedullary humeral morphology, including canal shape, and related these findings to the design of proximal humeral prostheses. Sixty cadaveric humeri (thirty pairs: fifteen from male donors and fifteen from female donors) were studied. Three-dimensional computer models were built from canal and periosteal contours extracted from computerized tomographic data and multiple measured anatomical parameters, including humeral canal axis, humeral head center, and hinge point offset; greater tuberosity and bicipital groove offset; humeral head center, radius, thickness, retroversion, and inclination; and size and torsion of sections of the canal. On the average, the humeral head center was offset both medially (seven millimeters) and posteriorly (two millimeters) from the humeral axis. The humeral head hinge point did not line up with the axis but instead was laterally offset by an average of seven millimeters. The average humeral head thickness was nineteen millimeters. The humeral head thickness and length were proportionately linked. There was marked variability in all of these parameters. Humeral head inclination averaged 41 degrees but was less variable than previously described, with 95 percent of our sample within the range of 35 to 46 degrees. The proximal section of the humeral canal was retroverted, and the retroversion was found to be similar to that of the humeral head on statistical analysis. Version of the middle and distal sections of the canal, however, was dissimilar to that of the proximal section of the canal. Proximal humeral retroversion was found to be extremely variable and averaged 19 degrees. The accuracy, reliability, and repeatability of the computer-based-model measurements were found to be excellent. Measurements of external proximal humeral morphology made with three-dimensional computer models of cadaveric specimens derived from the Midwestern United States agreed, in general, with those described for different populations evaluated with different measuring techniques. Proximal humeral morphology was extremely variable as highlighted by the large ranges of measurements seen for all variables. Examination of the intramedullary morphology showed that there is an internal version, with measurements dependent on the canal distance distal to the anatomical neck. Because of the marked variabilities seen in proximal humeral morphology, newer prosthetic designs are now allowing surgeons to control multiple prosthetic variables. An understanding of the normal values for proximal humeral morphology can serve as an important guideline for component selection, especially when the normal anatomy is distorted. Additionally, variations in intramedullary version may have important consequences for future designs of press-fit proximal humeral replacement.