Abstract
Introduction
Hemiarthroplasty is a treatment option for comminuted fractures and non-unions of the distal humerus. Unfortunately, the poor anatomical fit of off-the-shelf distal humeral hemiarthroplasty (DHH) implants can cause altered cartilage contact mechanics. The result is reduced contact area and higher cartilage stresses, thus subsequent cartilage erosion a concern. Previous studies have investigated reverse-engineered DHH implants which reproduce the shape of the distal humerus bone or cartilage at the articulation, but still failed to match native contact mechanics. In this study, design optimization was used to determine the optimal DHH implant shape. We hypothesized that patient-specific optimal implants will outperform population-optimized designs, and both will optimize simple reverse-engineered designs.
Methods
The boney geometries of six elbow joints were created based on cadaver arm CT data using a semi-automatic threshold technique in 3D Slicer. CT scans were also obtained with the elbows denuded and disarticulated, such that the high contrast between hydrated cartilage and air could be exploited in order to reconstruct cartilage geometry. Using this 3D model data, finite element contact models were created for each elbow, where bones (distal humerus, proximal ulna and radius) were modelled as rigid surfaces covered by non-uniform thickness layers of cartilage. Cartilage was modelled as a Neo-Hookean hyperelastic material (K = 0.31 MPa, G = 0.37 MPa), and frictionless contact was assumed. In order to simulate hemiarthroplasty, the distal humerus cartilage surface was replaced by either a rigid surface in the shape of the subchondral bone (bone reverse engineered or BRE design), or a surface offset from the bone by some distance, which was defined parametrically and modified by an optimization algorithm. Simple flexion-extension with constant balanced muscle loads was simulated in ABAQUS (Fig 1), and resulting contact areas and contact stresses were calculated. For each specimen, the contact mechanics of the intact and DHH reconstructed joints were calculated. A design optimization algorithm in Matlab was used to determine the optimal offset distance which resulted in contact stress distributions on the ulna and radius which most closely resembled their intact conditions. This procedure was repeated in order to generate specimen-optimal offsets, as well as population-optimal offsets.
Results
The population-optimal offset distance was 0.72 mm; whereas the specimen-optimal offsets ranged from 0.52 to 1.04 mm. Compared to the BRE design, which is effectively an offset distance of 0 mm, contact area generally increased at both the ulna (Fig 2) and radius (Fig 3) when either optimized design was used. On average, the specimen-optimal implant designs yielded only slightly larger contact areas than the population-optimal offsets, and only at mid-flexion (40–60 deg). Neither optimization strategy increased contact areas to those of the intact joint.
Conclusions
Design optimization is a promising technique for improving patient-specific implants by offering customization in terms of contact mechanics, instead of simply reproducing osseous geometry. In this study, our models predict a large increase in contact area if optimal offsets are used when designing subject-specific DHH, and a population-optimal offset distance seems to be just as good as a subject-optimal offset.
For any figures or tables, please contact authors directly (see Info & Metrics tab above).