TRABECULAR BONE RESPONSE TO IMPLANT MEDIATED LOAD

Participants: N.J. Caldwell, D.G. Orton, K.C. Anderson, J.M. Taboas, M.M. Moalli, S.A. Goldstein, S.J. Hollister, K.A. Sweet, B. Nolan, D.C. Kayner, J. Baker, R. Taylor

Keywords: bone adaptation, porous coating, trabecular bone

Introduction

The goal of porous-coated orthopaedic implants is to attain fixation by trabecular adaptation at the implant interface. It has been hypothesized that the early trabecular bone response at the implant interface is dominated by biologic factors (wound healing) while the later response is influenced by the mechanical environment. Previous work using our in vivo loading model supported this hypothesis. In that study, an increase in type I pro-collagen production at five weeks was seen in all tissue that had undergone surgical trauma, with no difference between loaded and non-loaded tissue. The purpose of this present study was to investigate the long term response of trabecular bone adjacent to porous-coated implants at a time significantly removed from surgical trauma, specifically focusing on the relative influence of mechanical stimuli on equilibrium bone adaptation.

Materials and Methods

The experimental model consisted of a custom servohydraulic implant placed in the distal femur of adult mongrel dogs. Five separate porous-coated cylindrical platens (6 mm diam. and length) were interference fit within the femoral trabecular bone. A common hydraulic manifold then fits over the platens and can be activated via an external pump. The model provides a mechanically controlled environment, with the only loads delivered to the trabecular bone being from the five actuators.

Three experimental groups were examined; animals that received mechanical stimuli at either a high (35.6 N peak load) or low (8.9 N peak load) magnitude and those that underwent the same surgical procedure but whose implants were never activated (no load), with six animals in each group. Mechanical stimulation of the load groups commenced one week post-surgery and consisted of a trapezoidal waveform with either a 8.9 N or 35.6 N peak load, 178 N/s rate, at 1 Hz for 1800 cycles per day, every day, for 24 weeks. This experimental duration was chosen to represent a time when an equilibrium state was approached with respect to the biologic healing response to surgical trauma.

At sacrifice, the platen-bone constructs and site-matched contralateral controls were cored out of the femurs using a 9 mm trephine to a depth of 15 mm and fixed. The constructs were scanned on a 3D micro-computed tomography (micro-CT) system at a 50 micron/voxel resolution. Stereologic analyses were then performed on the 5 mm cube of bone directly under the platen and controls to obtain measures of bone architecture, which included the following: bone volume fraction (BVF); connectivity (CON); trabecular plate separation (TPS), number (TPN), and thickness (TPT); and the degree of anisotropy (DA).

Following scanning, the platen-bone specimens (and controls) were sectioned transversely to separate the bone directly under the platen. This portion was then sectioned longitudinally. One half of the bone was decalcified, embedded in paraffin and examined by immunocytochemical analysis of type-I procollagen synthesis. The percent of trabecular surfaces producing type-I procollagen was quantified with a square grid. The other half-cylinder of bone was embedded in glycomethacrylate and stained for alkaline phosphatase and acid phosphatase.

Progress

The in vivo phase has been completed for all eighteen dogs in the study. Micro-computed tomography scanning, stereologic analyses, and immunohistochemical analysis for eleven of the dogs has been completed, with preliminary results presented below.

Results

Preliminary results of stereologic analyses are presented in Table 1. Comparison of experimental to contralateral specimens reveals a dramatic loss of bone in all study groups. Surgically treated tissue had substantial decreases in bone volume fraction, connectivity, trabecular plate number, and an increase in trabecular plate separation. Figure 1 demonstrates this result in a 35.6 N load case.

Figure 1.

Analysis of variance between the three treatment groups revealed only one significant difference due to load protocol. A smaller loss of connectivity was evident in the low load group when compared to the higher load and no load groups. However; this may be due in part to the lower connectivity of the contralateral specimens in this group, thus resulting in a smaller relative loss in the experimental limbs.

Osteoblastic biosynthetic activity as measured by type I pro-collagen (%Active Surface, Table 1) revealed no significant differences between experimental and contralateral control specimens for all groups.

Table 1.

Discussion

The immunocytochemistry results suggest that at 25 weeks the wound healing response is no longer dominant. The surgically manipulated tissues (experimentals) do not have significantly different osteoblastic biosynthetic activity compared to contralateral controls.

The bone architecture showed a generalized decrease in amount of bone via loss of whole trabeculae (trabecular plate thickness did not change) in all experimental limbs. This result was not affected, in either magnitude or direction, by application of daily mechanical stimulus. It should be noted that the morphologic changes measured in the 35.6 N loaded specimens are similar to those found in a previous study utilizing the same experimental model for the same duration, but with porous-coated platens of different designs. This may be an indication that less bone structure is required to support the uniaxial load applied in this model.

The unexpected lack of response when comparing load magnitudes and no loading may be explained by several factors. The six month time frame may not be long enough for the mechanically mediated adaptation response to be complete. The mature, lamellar bone may be less sensitive to the loading regime than highly reactive woven bone. Finally, the adaptation from a multiaxial to a uniaxial effective load may be more dominant than any response due to the relative magnitude of the uniaxial load.