In vitro and in silico simulations of femoral heads with avascular necrosis

Avascular necrosis of the femoral head (AVN) is a complex disease that is linked to multiple aetiologies including steroid use and alcohol abuse. Its pathology is characterised by localised ischemia leading to cell death followed by a period of partial repair during which a sclerotic boundary forms around the necrotic lesion. In many patients the pathology progresses to include involvement of the articular surface leading to arthritic degeneration of the joint. Current classification systems for AVN evaluate lesion size and location and cannot accurately predict whether fracture will occur. There is a need for a prognostic tool to differentiate patients who would benefit from conservative therapies from those for whom arthroplasty is indicated. Previous research has indicated that lesion size and location affects disease progression and it was hypothesised that a morphology-based assessment would better quantify risk of progression. In vitro experimental disease models were constructed by substituting a plug of bone from the central portion of porcine femoral heads with less stiff, weaker, bone from bovine lateral epicondyles. These models and control femoral heads were compressed to incrementally increasing displacements between flat platens. A parametric study using finite element analysis was used to demonstrate the sensitivity of the disease model to geometry and material properties and to develop a risk score that quantified stress discontinuities at the lesion boundary. A more physiologically representative method of load application through a compliant and conforming surface was developed in silico. This was used to evaluate cone-shaped simulated lesions with varying size and orientation and a series of 47 organic lesion geometries derived directly from the proximal femoral anatomy of eight patients suffering from AVN, two of whom had subchondral fractures. The experimental disease models were significantly less stiff than the control femoral heads and both were shown to behave linear-elastically at displacements below 1mm. This was beneficial as it allowed linear-elastic material properties to be used in the in silico simulations. These simulations confirmed that lesion properties and morphology significantly affected the stress distribution and highlighted that a physiologically representative method of load application is essential for future in vitro studies. The risk score allowed patients to be ranked according to their fracture risk. The rank matched that obtained using the current gold standard grading system but with improved granularity as well as the opportunity to apply a threshold value for categorizing risk. This proved the potential for using a morphology-based risk analysis to better differentiate patients for whom early surgical intervention may be beneficial from patients who would benefit more from total hip arthroplasty.