Design and characterization of additively manufactured compliant features for spinal implants

Abstract

Spinal fusion is the most common surgical reconstruction method to treat spinal disorders. In this procedure, the degenerated disc is removed, with an interbody device inserted between two adjacent vertebral bodies to restore disc height. Eventually, bone will grow over the implant providing long-term stabilization and immobilization of the joint. Spinal fusion implants are often metallic to provide the necessary structural support; however, most devices are much stiffer than the underlying bone, which can cause stress shielding and bone failure leading to poor patient outcomes. Furthermore, fusion implants do not reflect the compliant and mobile nature of the spine. An alternative to fusion is total disc replacement which uses an artificial spinal disc as a replacement to the natural disc, however, there are few on the market and most have high revision rates.

Another approach to the current rigid spinal implants is to consider compliant implants, with flexible or energy absorbing features to more naturally respond to spinal loading similar to the intervertebral disc. While these designs are largely unproven to date, the use of additive manufacturing (AM) enables the realization of new metamaterials and compliant mechanisms to evaluate the potential of these features in spinal implants. Therefore, the overall goal of this thesis was to design and manufacture AM compliant features in interbody devices for spinal reconstructive surgeries and characterize them using novel test methods. To achieve this goal, two studies were designed: (1) to explore the design of flexible elements in a cervical interbody device, and (2) to evaluate the performance of a lumbar dynamic surface conforming device to improve vertebral endplate contact area under asymmetric loading.

In the first study, a series of compliant features were developed for cervical interbody devices (CID) using engineered surfaces, optimization simulation, and lattices. The CIDs were manufactured in a polyether-ether ketone (PEEK)-like resin and in a titanium alloy (Ti64) using electron beam melting. The CAD model was compared to both types of print; however, a sub-optimal melt theme was used for the metal parts leading to larger discrepancies and print failures. The successful samples were speckled with paint to measure full-field surface strain using the Aramis 3D motion capture system. The AMTI VIVO joint motion simulator was used to apply compressive loading to the CIDs placed on Sawbones specimens. One set of CIDs was tested until failure resulting in an average ultimate compressive strength of 281 N in resin and 2569 N in Ti64 which varied considerably based on device geometry. Another set of implants was tested in repeated compression (5 cycles at 0.05Hz) to 50N in resin and 500N in Ti64, with three repetitions of each device. The results for Von Mises strain and load-displacement curves were averaged per device geometry. Each geometry exhibited similar levels of strain in both the resin and titanium devices. The resin devices exhibited non-linear stiffness displaying initial laxity with a steep increase in stiffness while the Ti64 devices were effectively linear. The most compliant device in both materials was an engineered flexible element relying on bending for energy storage.

In the second study, an iterative design process was used to develop a lateral lumbar interbody fusion (LLIF) device with articulating endplates for contact surface optimization, establishing a precedent for designing additively manufactured devices. The developed dynamic surface conforming (DSC) implant and the replicated commercial implants with flat and anatomic endplates were manufactured in PEEK-like resin and in Ti64 using laser powder bed fusion. The design for AM of the DSC implant was successful and printed well in both materials. A custom fixture using machined Sawbones samples was developed to characterize bone-implant contact area in a challenging vertebral endplate geometry associated with partial implant pseudoarthrosis. Compression, flexion-extension, lateral bending, and gait physiological waveforms were derived from the OrthoLoad database and applied in six degrees-of-freedom on the VIVO for 5 cycles at 0.05Hz. Force and displacement data and range of motion data were collected from the VIVO while contact area and pressure were recorded on a Tekscan sensor. Each experiment was repeated three times, and the data was aligned and averaged to one loading cycle. The DSC and anatomically shaped implants had the highest normalized area with the resin implants having an overall higher contact area than the Ti64 implants. The DSC minimized the geometric effect of the challenging scenario while providing as much of a range of motion in other axes as the other implants. Contact pressure was measured to be the lowest for the DSC and anatomically shaped implants, consistent with the contact area measurement. Furthermore, this work has shown differences in the performance of implants in compression and physiologically relevant loading waveforms.

The devices and methods developed in this thesis yields new insights into the strain behavior of engineered and computationally generated flexures and the application of these flexures towards a surgical problem. Despite the preliminary nature, this work demonstrates a functional implant following design for AM principles. Future work is needed to further evaluate the robustness of these devices under fatigue loading and the reliability of these tests compared to established standards.