Feature Article

FEATURE

FEA and orthopaedics

Computer simulations have been used historically to solve complex structural problems in fields such as civil and mechanical engineering. The finite element analysis (FEA) approach is based on numerical modelling of physical systems and evaluation of deformation and stresses in solid bodies or structures. As with physical experiments, FEA can be a powerful tool in product design and development, for meeting regulatory objectives, and for proof of concept. From the regulatory perspective, FE models are used in evaluating metallic implants for the purpose of predicting fatigue loads (ASTM International WK8881 Guide for FEA of Metallic Orthopaedic Implants, www.astm.org). FEA can also be used to complement physical testing, particularly when extensive parametric studies are desired. This article will explore the benefits and the limitations of applying numerical/computational simulation techniques in orthopaedic product design and evaluation and the value of using these techniques for understanding clinical performance.

Table I. (click to enlarge) The process of developing a FE model.

Performing a FEA

Figure 1. Example of a FE model of a resurfaced hip generated from QCT data.

To understand the relevance of FEA to clinical performance, it is important to be aware of the steps involved in performing a FEA analysis, particularly for orthopaedic applications. In general, there are three phases to FEA: preprocessing, analysis and postprocessing. Briefly, preprocessing involves construction of the structure such as the bone implant system. This requires acquisition of the geometry; creation of the FE model; and definition of material properties, applied loads, boundary conditions and interface (contact) conditions (Table I). In particular, quantitative computed tomography (QCT) scans of bone are recommended to obtain realistic anatomy, as well as to allow the extraction of physiological bone properties (Figure 1). Commercial FE software is available to analyse the FE model and allow the determination of deformation, strains and stresses that result from the applied loads. This software also includes a postprocessing component that allows visualisation of the results and identification of specific values. At times, further postprocessing is required to convert the available results into output data that is relevant to orthopaedics, for example, bone remodelling potential, interface relative motion, wear depth and wear volume.

The value to orthopaedics

With any well designed physical experiment (cadaver testing or simulator studies) or computational simulation (FEA), proper and clinically relevant inputs must be given to produce meaningful outputs. The FE model must be verified or validated by comparing it with

• simultaneous physical experiments (FE model of experiment)
• results from previously published experiments or models, or
• clinical outcomes.

If closed-form solutions are also available, those should be utilised to verify that the model produces reasonable results. For example, a simple composite beam bending equation can be used to approximate the stresses for an analysis of portions of a hip stem implant.

Given the multifactorial nature of implant performance, FEA is particularly useful in ranking different designs. For example, design A can be compared with design B under identical conditions. This cannot be achieved in clinical studies or even cadaveric experiments. It allows for fundamental design exploration and provides a means to assess designs. FE models can also be used to evaluate the same design under specific controlled changes to conditions. This helps, in turn, to determine the upper and lower bounds of the performance by considering best case and worst case conditions such as comparing the effects of extreme valgus and varus implantation or different extreme loading activities. As part of understanding short-term versus long-term outcomes, bone implant fixation conditions can be altered to examine the effects of bony ingrowth on load transfer and bone remodelling potential.

Table II. (click to enlarge) Matrix of possible success measures from FEA.

Parameters such as strain transfer, excessive stress, wear, deformation, anatomic fit, relative motion and range of motion can be quantified from FEA (Table II). Clinical failure modes such as pain, implant loosening, osteolysis, periprosthetic fracture, dislocation and infection may not be directly measured, although FE models can be used to help provide some mechanical explanation to many of these outcomes. For example,

• implants with excessive relative bone implant micromotion may present a higher likelihood of pain
• increased wear volume may increase the risk of osteolysis
• greater interface stress promotes loss of fixation and implant loosening
• smaller range of motion may induce dislocation.

Potential pitfalls

FEA, when developed with scientific/technical rigour and used in an appropriate manner, can be a powerful tool. However, it is also critical to recognise the limitations of FE models. Biological structures such as bone, ligaments, tendons and cartilage have highly complex mechanobiology. The biological interactions at the cellular and molecular levels have not been incorporated into conventional bone implant models.

Because bone density and architecture varies dramatically between anatomic sites and also spatially within each site, FE models that do not incorporate the nonhomogeneity of bone properties from QCT data are less able to capture the complex interactions between the implant and supporting bony tissue. Even though models may be generated from QCT data, these models tend to be representative of a single “patient” in terms of anatomy and bone quality. Therefore, designers should be cautious of extrapolating the outcomes from a single model to the entire patient population.

Figure 2. (click to enlarge) A retrieved hip resurfacing component exhibiting interfacial fibrous tissue (B), between the cement mantle (A), and cancellous bone. Neocortical bone (C) was also identified.

However, that does not mean that a researcher cannot understand the fundamental biomechanics of an implant design from performing an analysis using FEA. Parametric studies can help designers better understand an implant’s performance under various conditions. Currently, it is also unclear which success measures are sensitive to these limitations. Some of these measures may also be design-dependent, that is, some implant designs may be more “robust” to variability in the model input. There follows a case study, which demonstrates the value of FEA.

FEA in practice

Figure 3. (click to enlarge) FE model of the intact (left) and resurfaced (middle) femurs generated from QCT data. A cross-section of the resurfaced femur (right) is shown with the spatially varying CT data (in Hounsfield units).

A proportion of retrieved hip resurfacing components have shown resorption of the proximal femoral bone or formation of a fibrous membrane at the bone–cement interface (Figure 2). A FE model of a resurfaced femur was developed using QCT data to determine whether the presence of fibrous tissue (2 MPa) exacerbates offloading, that is, to reduce loading to the proximal bone and promoting resorption (Figure 3). Bone remodelling stimulus was determined using changes in strain energy.

Figure 4. (click to enlarge) Effect of fibrous tissue (right) on (A) initial bone remodelling and (B) bone–implant interface pressure.

Analysis found that the presence of the fibrous tissue resulted in increased proximal medial bone resorption and slightly greater bone formation surrounding the stem (Figure 4A). Correspondingly, the pressure at the implant–bone interface was found to decrease proximally under the loading platform and increase at the stem, particularly adjacent to the stem-head junction (Figure 4B). The FE results support the hypothesis that the presence of fibrous tissue decreases load transfer to the proximal bone, which further exacerbates stress shielding and resorption. To better understand the cause of implant failures in hip resurfacing arthroplasty, additional retrieval and FE studies are necessary.

Taking greater control

The clinical performance of an implant is affected by many patient, surgical and design factors. A FE model, just like a physical experiment, is unable to account for all these factors. However, FEA allows better control of the study design to help quantify the discrete contributions from specific parameter changes to the performance measures. FEA is an invaluable tool in guiding the design process when the models are developed and interpreted by well trained researchers. It should be considered as another tool, alongside physical testing and clinical trials, during the rigorous development of an implant.

Kevin Ong Ph.D is Managing Engineer and Steven Kurtz Ph.D is Corporate Vice President at Exponent, Inc, 3401 Market Street, Suite 300, Philadelphia, Pennsylvania, 19104, USA, tel. +1 215 594 8800, e-mail: kong@exponent.com, www.exponent.com.

The authors also have research and academic appointments/affiliations with the School of Biomedical Engineering Science and Health Systems, Drexel University, Philadelphia, Pennsylvania, USA.