An intramedullary prosthesis is a stemmed device, such as the femoral component in hip replacement surgery, that is anchored in the marrow cavity of a long bone. As the total number of prostheses im planted has increased dramatically due to their surgical success, so has the number of revision surgeries involving the replacement of failed devices. It is well known that intramedullary prostheses reduce mechanical loading of the surrounding bone. This effect is exaggerated in cementless prostheses, because they have larger cross-sections than their cem ented counterparts. Stress reduction— also known as stress shielding— causes bone loss through disuse atrophy, which results in a decrease in bone density and the degradation of bone structural properties. In some cases, the prosthesis loosens as its support is undermined, and must ultimately be replaced. The purpose of this study was to improve understanding of load transfer in cementless intramedullary devices and to determine whether the alteration of fundamental design variables can improve mechanical loading of the peri-prosthetic bone without jeopardizing stable anchorage of the prosthesis.

A series of straight-stem m ed, axisymmetric models was used to capture the essential features of intramedullary prostheses while maintaining generality in the results and permitting isolation of the effects of variations in individual design variables. Closed-form solutions for bone stresses in these models are restricted to the middle region of long, well-fixed implants. For such implants, the bone stress maintained was put in simple analytical form and the limited effect o f a reduction of prosthesis modulus, which is a function of the relative thickness of the implant, was demonstrated.

The primary contribution of this thesis is a detailed analysis of the effect of a proximal collar (a flange intended to mate with the cut bone surface, provide an insertion stop, and transfer loads directly to bone). The relative merits o f flat-collared versus collarless designs has been a long-standing controversy in hip replacement surgery. In this context, finite element analysis was used to study the influence of conical collars on stress transfer, with flat-collared (0° collar) and collarless (80° collar) implants taken as limiting cases. In the absence of bone ingrowth (e.g., during the early post-surgical period, when bone ingrowth fails to take place, or subsequent to loosening), conical collars increased proximal bone stresses by creating hoop stresses under axial and oblique loads. Representative values of proxim al bone strain energy density (SED) under oblique loading ranged from 53% of normal for the flat collar to 270% for the 80° collar. However, maximum relative motion betw een implant and bone, which can threaten the eventual establishment of fixation by bone ingrowth, also increased with collar angle, from 63 to 163 pm. Under bending loads, or with proximal bone ingrowth, hoop stresses were minimal, and bone stress shielding was substantial, with the 80° collar being most severe and little difference am ong the other models. Overall, collars in the 20°-50° range probably provide the best com bination o f appropriate proxim al bone mechanical stim ulus and limited micromotion.

The effects of stem length and extent of porous coating for bone ingrowth were exam ined under oblique loading. Shortening the implant to 38% of its original length reduced m axim um relative motion by more than 50% and improved (i.e., increased) bone SED, w ithout increasing interface stresses at the distal im plant tip. Nevertheless, the long im plant with the proximal 38% coated had less relative motion (17 pm) at the distal coating edge than the short implant with full coating (32 pm), suggesting that the uncoated distal stem improves initial stability in exchange for increased stress shielding.

Based upon the understanding of collar function developed previously, a spherical-collared implant was analyzed under axial and oblique loads. Some reduction in relative motion, together with collar interface shear stresses that were 25% to 35% lower, im plied improved stability compared with a com panion conical-collared model. Small initial m ism atches in bone and implant collar radii caused a substantial local redistribution of stresses, demonstrating that designed collar mismatches could potentially be used to establish more physiologic stress patterns.

Predictions of internal bone remodeling were made for the conical-collared im plants using an existing adaptive bone remodeling algorithm. Proximal bone m aintenance was better under conditions of no bone ingrowth, com pared with proximal o r full ingrowth, and improved with an increase in collar angle. For ingrown stems, differences by collar type were small, and all m odels showed proximal bone resorption. Proxim ally ingrown m odels performed better than fully ingrown ones between the distal edge of ingrowth and the distal tip of the prosthesis, as has been observed clinically.