18 What Have We Learned from Mechanical Total Disc Replacement?
Timothy T. Roberts, Colin M. Haines, and Edward C. Benzel
This chapter reviews the major lessons derived from the collected experience with cervical and lumbar mechanical disc replacements. Lessons are in three interrelated topics: device design, patient selection, and outcome analysis. Virtually all mechanical discs share identical goals: (1) to eliminate the painful degenerative/dysplastic elements of the joint, (2) to preserve or restore the natural range of spinal motion, and (3) to mitigate stresses on adjacent spinal segments, thereby theoretically limiting adjacent segment disease. Healthy cervical, thoracic, and lumbar vertebrae move with 6 degrees of freedom (DOF), meaning they have the ability to rotate in three planes and translate in three planes. Most physiological movements require combinations of rotation and translation—motions that often occur simultaneously in multiple planes. This is referred to as coupled motion. The complexity of coupled spinal motion is difficult to replicate mechanically. This is one of several reasons why intervertebral disc (IVD) replacement has yet to achieve the same clinical success as total hip arthroplasty, a comparatively simple ball-and-socket joint. Mechanical disc designs, both modern and historical, vary greatly in function, material composition, and performance in vivo. Current designs range from relatively constrained (i.e., rotation only, 3 DOF) ball-and-socket prostheses to completely unconstrained elastomeric rubberlike discs that can rotate and translate freely (6 DOF). Each design has its respective advantages and disadvantages. Both clinical and laboratory experience demonstrate that highly mobile unconstrained devices can lead to facet joint overload, whereas highly constrained devices are subject to greater sheer forces that are associated with premature wear and potential failure. Although the optimal device design remains disputed, evidence shows that the careful matching of physiological and mechanical range of motion (ROM) correlates with clinical success. Mismatched motion following arthroplasty procedures is associated with worse outcomes than no motion obtained with fusion procedures. Indications for disc arthroplasty have expanded over the past 2 decades. Today, cervical disc replacements are U.S. Food and Drug Administration (FDA) approved for both one- and two-level myelopathy and/or radiculopathy. Indications for lumbar disc replacements remain comparatively narrow; although studies supporting their safety and effectiveness in both multilevel arthroplasty and arthroplasty-fusion hybrids are emerging. Successful outcomes in disc replacement are highly controversial and are dependent on a range of factors. The most concrete and widely accepted conclusion is that both cervical and lumbar disc arthroplasty can effectively reduce or eliminate preoperative pain and neurological symptoms in both short- and long-term periods. In the cervical spine, the majority of current data seem to support preservation—but not necessarily improvement—of motion at the index level. Spinal segments immediately adjacent to one-and two-level cervical disc replacements may exhibit less postoperative motion than levels adjacent to one- and two-level fusions. Motion data following lumbar disc replacement are comparatively limited and inconsistent. Several short- to midterm studies suggest that disc arthroplasty reduces the radio-graphic incidence of adjacent segment deterioration (ASD) in both the cervical and lumbar spines. The degree to which this is clinically significant, however, is intensely controversial. At this time, disc replacement cannot be said to reduce symptomatic adjacent segment disease.
Keywords: degenerative disc disease, mechanical disc replacement, total disc arthroplasty, total disc replacementpinal biomechanics
In its relatively brief history, intervertebral total disc arthroplasty (TDA) has undergone considerable evolution. From its origin as an interpositional ball-bearing spacer to the latest in elastomeric hybrid designs, successes and failures of a diversity of devices continue to direct us toward the so-called Holy Grail of spine surgery: motion preservation in the absence of discomfort and dysfunction.
This chapter reviews the major lessons derived from the collected experience with cervical and lumbar mechanical disc replacements. Lessons are divided into three interrelated topics: device design, patient selection, and outcome analysis. With device design, we review the basics of healthy spinal motion, the biomechanics of various disc designs, and the successes and failures of several notable devices. With patient selection, we review the evolving indications and contraindications for disc arthroplasty, as well as the factors that influence operative successes and failures. Finally, with outcomes, we discuss the short- and long-term effects on symptoms, motion, and adjacent segment degeneration (ASD), and we compare the efficacy of disc arthroplasty with the alternative gold standard, intervertebral fusion.
Despite the considerable range of arthroplasty designs, virtually all mechanical discs share identical goals: (1) to eliminate the painful degenerative/dysplastic elements of the joint, (2) to preserve or restore, to one degree or another, the natural range of spinal motion, and (3) to mitigate stresses on adjacent spinal segments, thereby theoretically limiting ASD.
Total hip replacement is arguably the most successful mechanical arthroplasty procedure performed today. By effectively eliminating the diseased elements of the joint, restoring near-perfect natural joint motion, and relieving stresses on its respective adjacent joints, hip replacements have achieved remarkable success and popularity throughout the world. By many measures, intervertebral disc (IVD) replacement has yet to duplicate the success of hip, knee, and shoulder arthroplasty. The comparatively complex and interrelated motions of the spine make it more difficult to mechanically replicate than the ball-and-socket design of total hip replacement.1 To an extent, the historical successes and failures of several designs are attributable to an ability and inability to accurately reproduce these natural motions.
18.2 Device Designs
18.2.1 Normal Spine Biomechanics
A basic comprehension of the anatomy and physiology of the IVD is essential to understanding prosthetic disc designs. In the healthy adult spine, the nucleus pulposis (NP) behaves as a gelatinous cushion that resists and transfers compressive forces to the end plates and encompassing annulus. The annulus, by contrast, resists tensile, torsional, and radial forces. Posteriorly, two facet joints govern motion and protect the neural elements. Historically, facet joints are often overlooked as a source of pain and dysfunction. However, isolated facet degeneration may account for up to 20% of all axial back pain.2
The functional spinal unit (FSU) is a term for the collective motion between two vertebrae. Normal FSU function requires 6 degrees of freedom (DOF): the ability to rotate in three planes (flexion-extension/axial turning/lateral bending) as well as the ability to translate or slide in three planes (sagittal displacement [antero- and retrolisthesis], coronal or lateral displacement, and compression/distraction) (▶ Fig. 18.1). The healthy IVD exhibits all six motions and is thereby termed unconstrained. By contrast, the ball-and-socket hip joint is constrained to 3 DOF (flexion/extension, adduction/abduction, and internal/external rotation); a finger interphalangeal joint is highly constrained to 1 DOF (flexion/extension).
Fig. 18.1 Schematic detailing the 6 degrees of freedom of spinal motion. Each (a) sagittal, (b) coronal, and (c) axial plane has its own rotational and translational motion. Spinal motions are usually coupled, meaning an intended translational or rotational motion along one axis is accompanied by motion in a second axis.
Virtually all physiological spinal motions require simultaneous rotatory and translational movement. This means that the center of rotation for any given motion is not a fixed point but one that moves along an axis. When the spine is flexed, the rostral vertebral body translates anteriorly, whereas the NP is displaced posteriorly. The rostral body thus rotates in the sagittal plane around a dynamic fulcrum that shifts posteriorly with the NP. When the spine extends, the rostral vertebral body translates posteriorly, whereas the NP moves anteriorly. These dynamic or instantaneous centers of rotation (ICOR) vary greatly and depend on the level, plane, direction, and forces associated with motion.3 During sagittal and coronal plane motions in the cervical spine, for example, the axis of rotation is generally located in the anterior portion of the subjacent vertebral body. During axial rotation, however, it shifts closer to the posterior annulus.3 With lateral bending of the cervical and lumbar spine, the ICOR shifts to the convexity of motion, that is, leaning to the left rotates the rostral vertebrae around a point on the right side of the disc. During flexion and extension in the lumbar spine, ICORs for L2 through L5 are fairly consistently located around the posterior one third of the disc space, just below the superior end plate of the inferior vertebrae. The ICOR of L5–S1 is more variable but is often located in the posterior one third of the disc halfway between the end plates (▶ Fig. 18.2).4 Degenerative processes can dramatically alter spinal kinematics in ways that are both difficult to analyze and predict.3
Fig. 18.2 Average instantaneous centers of rotation (ICOR) for flexion/extension in the lumbar spine. ICORs for L2–5 are fairly consistently located around the posterior one third of the disc space, just below the superior end plate of the inferior vertebrae. The ICOR of L5–S1 is more variable but is often located in the posterior one third of the disc halfway between end plates.
To further complicate matters, many motions in the spine are coupled, meaning that for an intended primary motion in one plane, simultaneous motion will occur in another. Coupled motion is perhaps most evident with lateral bending in the cervical and lumbar spines. Laterally bending one’s head to the left, for example, is accompanied by ipsilateral vertebral body rotation. The swiveled spinous processes are then palpable to the right of the midline. Bending one’s torso to the side, however, couples lumbar lateral bending with contralateral vertebral rotation: Laterally bending the torso to the left rotates the vertebral bodies to the right—the spinous processes are then usually palpable to the left of the midline.3 Virtually all motions of the lumbar spine are coupled. These complex kinematic relationships vary significantly with spinal level, posture, and the presence of joint degeneration or pathologies such as scoliosis.
Coupled motion is guided primarily by the orientation and morphology of the facet joints. Isolated facet degeneration has been shown to alter motion of the FSU.5 Additionally, the annulus fibers, the anterior and posterior spinal ligaments, the ligamentum flavum, and to a lesser extent the remaining surrounding soft tissues play a role in guiding and restraining motion. The uncovertebral joints of the cervical spine may also guide motion; however, their exact role is disputed.5
Comprehending the basics of spinal motion is essential for the appropriate utilization of prostheses in clinical practice. It is also important from a surgically technical standpoint, as motion restoration requires both appropriate design selection as well as positioning of prostheses. Surgically mismatched motion is associated with worse outcomes than no motion at all (fusion).5 Poorly positioned prostheses accelerate degeneration at the index-level facets by up to 2.5 times, as well as at adjacent segments.5,6 Pars and pedicle stress fractures have also been associated with implant malpositioning.5 In the lumbar spine, positioning of Mobidisc (LDR Medical, Warsaw, IN) prostheses within 5 mm of individual patients’ natural COR were associated with significantly greater improvements in symptoms than those positioned farther than 5 mm from CORs as determined by preoperative flexion/extension radiographs.6 This finding is supported by in vitro models in which 5-mm anterior or posterior malpositioning of the disc resulted in marked changes to posterior muscular tension as well as forces across the facets and other supporting ligaments.7
The coupled and interdependent motions of the FSU are deceptively difficult to replicate mechanically. Further, implantable devices must be sturdy, biocompatible, and durable enough to withstand decades of continuous use. Next, we explore several key prosthetic designs, both historical and modern, and examine the degree to which they foster physiological motion, eliminate pathology, and remain functional in vivo.
18.2.2 Noteworthy Mechanical Disc Designs
The first human disc replacement was performed by Fernström in the late 1950s. His device consisted of a single stainless-steel ball bearing, which he inserted into diseased lumbar disc spaces.8 The 10- to 16-mm diameter sphere allowed the superior vertebrae to rotate with 3 DOF, but it did not effectively permit translation. The bearings acted as spacers, allowing indirect decompression of the neural elements, opening of the neural foramen, and appropriate tensioning of the paraspinous musculature.5
Fernström’s initial success was short lived. In vivo, the steel bearings transferred extreme contact stresses onto adjacent end plates. Many implants failed due to subsidence of the stiff steel into the comparatively soft vertebral bodies. Fernström bearings also resulted in hypermobility at the index level as the devices lacked internal restraints to motion.9 These early failures emphasized the importance of broader and equal force distribution between vertebrae and were early indicators that supraphysiological ROMs were detrimental.
Over the following decades, researchers investigated a variety of silicon, rubber, and other polymeric-based designs. These devices feature elastic cores that might reproduce the natural cushioning effects of the disc, thereby theoretically preventing subsidence. Several designs claimed to replicate the true 6 DOF of the natural disc. The AcroFlex disc (DePuy-AcroMed; Raynham, MA) consisted of a polyalkene rubber cylinder capped with two titanium end plates. Emerging in the 1980s, it was the first major commercial one-piece or self-contained implant. Its elastic body added 3 degrees of translation to the 3 rotational DOF afforded by ballbearing or ball-and-socket designs. Although laboratory performance of the AcroFlex was impressive, the few early discs that were implanted garnered lackluster results. Possible carcinogenic properties of the vulcanized rubber led to its withdrawal from the market.9 Further, Computed tomography (CT) scans at 1- to 2-year follow-up demonstrated premature development of defects in several of the polyalkene rubber cores.9
Perhaps the simplest variation on Fernström’s steel-bearing, ball-and-socket devices pair two metallic end plates with concave and convex hemispherical bearings of various materials (metal-on-metal [MOM], metal-on-polyethylene [MOP], etc.). One of the earliest examples of this design, the ProDisc (Synthes Spine; West Chester, PA), was introduced in Europe in the 1990s. The ProDisc was the first widely used ball-and-socket implant for both the lumbar and cervical spine. It featured 3 degrees of constrained rotational motion (3 DOF). Two large-footprint cobalt-chromium-molybdenum (CoCrMo) plates at each bony interface dispersed axial and sheer loads, mitigating risk of subsidence. Modern versions of the ProDisc feature metal-alloy plates with extensive porous backside-coating to promote bony ongrowth.10 On the articular side, the superior, concave partial-spherical cup is highly polished CoCrMo that contacts an ultra-high molecular weight polyethylene (UHMWP) partial-spherical convex post.
The major drawback to pure ball-and-socket devices are that their constrained design does not permit translational motion. These 3-DOF devices cannot fully replicate natural coupled spinal motions. In situations where the ICOR is small, such as with the L2–3 and L3–4 IVDs, both the adjacent soft tissues and facets may experience the same excursion and forces with TDA as would the normal spine.11 However, in segments with larger ICORs, ball-and-socket implants must resist translational motions or catastrophic dislocation (at either the bone-implant or the implant articular interface) will occur. Even in the absence of catastrophic dislocation, large ICORs generate sheer forces at the bone-implant interface that may lead to gradual implant loosening. Further, repetitive and simultaneous axial and sheer forces generate high contact pressures across the artificial articulation, leading to accelerated wear. As discussed, potential for early failure is exacerbated when implants are malpositioned.12
Over 15,000 ProDisc devices have been implanted worldwide since 1990 with generally positive results.10 Short-term clinical success has been reported in more than 90% of patients at an average of 1-year follow-up without major complications.13 Five-year follow-up from large, randomized control trials (RCTs) of ProDisc-L versus arthrodesis demonstrated that the implant was no less efficacious than circumferential (combined anteroposterior [AP]) fusion in the mid-term.14 In the United States, the ProDisc-L received FDA approval in 2006 for implantation in single-level degenerative disc disease (DDD) from L3 to S1.15
Another early ball-and-socket design, the Cummins-Bristol disc, utilized two metal-on-metal articulating stainless-steel plates that were anteriorly fixed to each vertebral body with a single AP screw.16 Early complications involved screw breakage, implant migration, and frequent reports of dysphagia that were attributed to the relatively thick anterior profile of the initial design. Several iterations later, the device was rebranded with a slimmer anterior profile and endowed with multiple points of locking-screw fixation. More significantly, the inferior hemispherical “cup” articulation was elongated to a “trough,” thereby introducing a groove in which the superior convex “ball” can translate in sagittal plane.16 This 4-DOF device was renamed Prestige (Medtronic; Memphis, TN) and has changed relatively little since its redesign. The Prestige is currently approved by the U.S. FDA for investigational device exemption (IDE) use in the cervical spine.
Dual-bearing TDAs, such as the Charité III (DePuy-AcroMed), feature variations on a mobile biconvex core that is interposed between two concave plates. The prosthesis is essentially two ball-and-socket joints that face one another and share an oblate spheroid-shaped core. Working together, the two rotational joints function in equal but opposite motions to allow the superior plate to translate a small distance. Anterior translation, for example, is achieved with rotational flexion at the inferior bearing and rotational extension at the superior bearing (▶ Fig. 18.3). Several dual-bearing TDAs allow translational motion in the sagittal and coronal planes and therefore possess 5 DOF. To an extent, they can tolerate larger ICORs than fixed-bearing devices.
Fig. 18.3 A single-bearing ball-and-socket joint (top) exhibiting rotation in a single plane. Ball-and-socket bearings exhibit 3 degrees of freedom (DOF), meaning they can rotate in the coronal, sagittal, and axial planes, but cannot translate. By contrast, this dual-bearing joint (bottom) consisting of opposing ball-and-socket joints, can perform biplanar translational motion in addition to rotation with 3 DOF.
The relatively unconstrained motion of dual-bearing TDAs is theorized to promote natural motion of the FSU and thus mitigate sheer stresses associated with single-bearing ball-and-socket implants. Recent evidence suggests, however, that motion is not as symmetric or fluid as theorized. In a notable cadaver motion study, the superior bearing was found to undergo more than twice the excursion of the inferior bearing.12 This correlates with findings from in vivo retrieval studies of Charité implants in which superior bearing surfaces were disproportionately worn.12 For equal wear to occur, the ICOR must theoretically reside within the center of the biconvex core. Because the natural ICOR for lumbar flexion/extension is typically located in the inferior and posterior aspect of the subjacent lumbar body—a point substantially closer to the rotational center of the superior prosthetic bearing than the inferior—it is reasoned that the superior bearing experiences less resistance and therefore more motion.
Another drawback to early dual-bearing designs is that they lack internal restraint to extremes of motion that would normally be governed by the annulus. Without internal restraint, the facet’s joints must serve as the primary restraints to excess translation, resulting in facet loads up to 2.5 times greater those seen in healthy FSUs.13 Early dual-bearing designs have been associated with failures due to disproportionate rates of facet-generated pain.
In data derived from one FDA noninferiority RCT comparing the Charité (dual-bearing), Kineflex (ball-and-socket; Spinal-Motion, Sunnyvale, CA), and Maverick (ball-and-socket; Metronic Sofamor Danek, Memphis, TN) the authors found the number-one cause of clinical failure at 24 months was facet pain identified through diagnostic injections. Although all pros-theses experienced multiple causes of failure, 23% of failures in the Charité group were due to facet pain versus 8.6% and 8% in the single-bearing Kineflex and Maverick groups, respectively.17 This clinically relevant finding is supported by cadaveric studies in which implantation of L5–S1 Charité dual-bearing discs resulted in greater facet forces, particularly during lateral bending. Its contender, the ball-and-socket ProDisc, alternatively resulted in greater variations in pre- versus postimplantation ICOR.18
The Charité III was the first lumbar TDA approved in the United States in 2004, and is the most extensively implanted and studied artificial disc over the past 20 years.12 It was approved for single-level implantation for degenerative lumbar disc disease.15 Over 5,000 have been implanted worldwide.
The Bryan disc (Medtronic Sofamor Danek) is a more recent variation on the one-piece elastomeric design. The Bryan disc consists of a saline-bathed polyurethane central disc surrounded by two titanium end plates. The saline solution is purported to both lubricate articulating surfaces, as well as to provide a hydraulic dampening effect.19