An analytical and in silico strategy for estimating maximum stress and fatigue life of bone plates under in vivo loads: a rationale for regulatory testing

Abstract

Despite the innovations introduced by locking compression plates (LCP), implant failures still occur due to fatigue fractures caused by cyclic loads. The endurance of LCP, especially in lower limb plates subjected to ambulatory cyclic loads, is a critical factor that needs to be understood. Unfortunately, there is limited information available on the fatigue failure of LCP. The fatigue behavior is a crucial aspect of mandatory mechanical tests for regulatory purposes, which aim to determine the load at which the plate withstands under a specific number of cycles, known as the runout condition. The current test standards, such as ASTM F382, only provide the setup configuration without furnishing explicit guidelines regarding the required fatigue strength of the bone plate in the runout condition. The determination of the minimum level of in vivo performance that the plate must fulfill remains an open issue, which is frequently addressed by the direct comparison with predicate devices. To address this issue, this study proposes a rationale that combines analytical and in silico approaches to estimate the maximum stress and fatigue life of a bone plate under in vivo loads. Four-point bending tests were conducted on a diaphyseal femoral plate to determine the experimental runout load. Analytical and finite element (FE) models were first implemented to replicate the four-point bending setup and to calculate the maximum stress on the plate. The Goodman and Gerber criteria were exploited to determine the mean stress effect due to the four-point bending setup and to verify the predicted number of cycles. In addition, the force-displacement curves of the FE model were validated by means of experimental results. Analytical and FE models were then applied to calculate the maximum stress and assess the performance of the implanted plate under in vivo loading conditions. In the implanted plate condition, a mean number of cycles higher than 1.5 million was estimated. Analytical models showed good performance compared with in silico strategies, exhibiting errors below 6%. The comparison between the obtained results provides valuable insights for constructing a robust rationale to support the regulatory process in order to obtain CE marking.