In shoulder arthroplasty, humeral resurfacing or short stem devices rely on the proximal humeral bone for fixation and load transfer. For resurfacing designs, the fixation takes place above the anatomical neck, whilst for short stem designs the resection is made at the anatomical neck and fixation is achieved in the bone distal to that resection. The aim of the study is to investigate the bone density in these proximal areas to provide information for implant design and guidance on appropriate positions to place implant fixation entities.

CT scans of healthy humeri were used to map bone density distribution in the humeral head. CT scans were manually segmented and a solid model of the proximal humerus was discretised into 1mm tetrahedral elements. Each element centroid was then assigned an apparent bone density based on CT scan Grey values. Matlab was used to sort data in spatial groups according to element centroid position to map bone density distribution. The humeral head was divided into twenty 2mm thick slices parallel to the humeral neck starting from the most proximal region of the humeral head to distal regions beneath epiphyseal plate (Fig 1a). Each slice was then radially divided into 30 concentric circles and each circle was angularly divided into 12 regions (Fig 1b). The bone density for each of these regions was calculated by averaging density values of element centroid residing in each region.

Average bone density in each slice indicates that bone density decreases from proximal region to distal regions below the epiphyseal plate and higher bone density was measured proximal to the anatomical neck of the humerus (Fig2). Figure 3 shows that, both above and below the anatomical neck, bone density increases from central to peripheral regions where eventually cortical bone occupies the space. This trend is more pronounced in regions below the anatomical neck and above the epiphyseal plate. In distal slices below the anatomical neck, a higher bone density distribution in inferior (calcar) regions was also observed.

Current generation short stem designs require a resection at the anatomical neck of the humerus and a cruciform keel to fix the implant in the distal bone. In the example in Figure 3, the anatomical neck resection corresponds to the 18 mm slice, with the central cruciform keel engaging between slices 18 mm and 27 mm. The data indicates that this keel should make use of the denser bone by the calcar for fixation, suggesting a crucifix orientation as highlighted in Figure 3. The current generation of proximally fixed humeral components are less invasive than conventional long-stemmed designs, but the disadvantage is that they must achieve fixation over a smaller surface area and with a less advantageous lever arm down the shaft of the humerus. By presenting a spatial density map of the proximal humerus, the current study may help improve fixation of proximally fixed designs, with a simple modification of implant rotational orientation to make use of the denser bone in the calcar region for fixation and load transfer.