This article compares the outcomes following spinal fusion and disk replacement for degenerative conditions of the lumbar and cervical spine. The short-term outcomes of lumbar and cervical total disk replacement are equivalent to that following spinal fusion. Long-term follow-up studies of total disk replacement are necessary to confirm its potential benefit in reducing or preventing adjacent level degeneration. Also discussed is the philosophy of the surgical management of degenerative conditions of the lumbar and cervical spine.
Quid est testimonium (in Latin, “what is the evidence”)? With 70% of the population likely to have at least one episode of neck and back pain in their lives it is not surprising that back pain is the second most common reason for visiting a family physician. The natural history of low back pain points toward a high rate of recurrence and ongoing disability. In addition to the suffering and disability, chronic spinal pain is associated with enormous direct and indirect costs. Axial spinal pain can originate from degenerated intervertebral disks; facet joint arthritis; and secondarily from the muscles, ligaments, and neural tissue. Pain perception involves a complex interaction at multiple levels of the neural pathway with central modulation and psychosocial factors influence the development and perpetuation of chronic pain. Nonoperative modalities including physiotherapy, pain pharmacotherapy, and spinal injections are the first-line treatment options for chronic axial spinal pain. Spinal fusion and total disk replacement (TDR) can be considered for recalcitrant symptoms.
Lumbar TDR is indicated in the management of discogenic back pain without facet arthritis; however, 95% of potential surgical patients are likely to have a contraindication for lumbar TDR. Lumbar fusion, however, has wider indications. In the cervical spine, anterior cervical diskectomy and fusion (ACDF) and TDR are advocated for radiculopathy and myelopathy with or without axial neck pain and nearly 50% of surgical patients with degenerative conditions qualify for a cervical TDR. This article compares the outcomes of spinal fusion and disk replacement for degenerative conditions of the lumbar and cervical spine. Also discussed is the philosophy of the surgical management of degenerative conditions of the lumbar and cervical spine.
Spinal fusion, first reported 100 years ago for posttuberculosis spinal deformity, is now being used to manage trauma, deformity, instability, and degenerative conditions. Fusion for degenerative conditions is based on the premise that obliteration of movement and off-loading the diseased motion segment relieves pain. Spinal fusion, with or without supplementary instrumentation, can be performed through a posterior (posterolateral fusion, posterior lumbar interbody fusion, and transforaminal lumbar interbody fusion), anterior (anterior lumbar interbody fusion [ALIF]), or combined approach (anteroposterior fusion). In the 5-year period from 1996 to 2001, the number of lumbar fusions performed in the United States increased by 113% with more procedures being performed for degenerative conditions and wide regional variations in the number and outcomes of lumbar fusion.
Despite the increasing number of lumbar fusions being performed, there are numerous unresolved issues including the efficacy of the procedure, potential for accelerated adjacent level degeneration (ALD), pseudoarthrosis, and bone graft donor site morbidity. Stiff fusion constructs in the lumbar spine lead to abnormal stresses on the adjacent levels potentially leading to accelerated ALD. Although the significance of radiologic ALD is unknown, the rate of symptomatic ALD requiring either decompression or fusion is predicted to be 17% at 5 years and 36% at 10 years following spinal fusion. Nevertheless, the debate continues whether the ALD is in reality a natural progression of the degenerative process.
Wide variations in the rate of pseudoarthrosis (nonunion) have been reported (2.3%–83.3%) following lumbar fusion, and can have a negative impact on the clinical outcome. Donor site morbidity at the bone graft harvest site is another cause for concern. Alternatives to autograft, such as allograft and bone morphogenetic protein (BMP), have shown equivalent outcomes; however, the use of BMP increases the cost of the procedure substantially. The efficacy of lumbar fusion for degenerative conditions is the major cause for concern, and in this context two issues need consideration.
What are the Outcomes of Lumbar Fusion for Degenerative Disease?
A review of literature analyzed the influence of subdiagnosis on the outcome of lumbar fusion and found significantly better clinical outcomes for spondylolisthesis compared with degenerative disk disease. Buttermann and colleagues reported satisfaction rates of 100%, 76%, and 69% in patients who underwent fusion for spondylolisthesis, disk degeneration, and postdiskectomy, respectively. Similarly, Glassman and colleagues found that a higher percentage of patients with spondylolisthesis attained minimum clinically important improvement for oswestry disability index (ODI) (71% vs 57%), back pain (60% vs 57%), and leg pain (63% vs 51%) compared with disk pathology following a posterolateral fusion. Slosar and colleagues also reported higher satisfaction following circumferential spinal fusion in patients with listhesis and stenosis (80%) when compared with painful disk degeneration (56%) and internal disk disruption (50%). Table 1 lists some of the studies relating to the outcomes following spinal fusion.
|Christensen, 2004 |
|9 studies (2 retrospective, 1 prospective, 5 RCT’s, and 1 animal study)||Spondylolisthesis served better with noninstrumented posterolateral fusion; degenerative conditions better with anteroposterior fusion|
|Gibson and Waddell, 2005 |
|31 RCT’s||Limited evidence to support surgery|
|Resnick et al, 2005 |
|2 RCT’s and numerous case series and control studies||Insufficient evidence to support a treatment guideline for intractable low back pain without stenosis or spondylolisthesis|
|Mirza and Deyo, 2007 |
|4 RCT’s||Surgery may be more efficacious than unstructured nonsurgical care but may not be more than structured cognitive behavioral therapy; limitations of the RCT’s prevent firm conclusions|
|Ibrahim et al, 2008 |
|3 RCT’s for primary analysis and 1 for sensitivity analysis||Evidence at present does not support routine surgical fusion for treatment of chronic low back pain|
|Carreon et al, 2008 |
|25 studies||Spondylolisthesis did better than DDD following fusion; no difference in outcomes between different fusion procedures|
|Chou et al, 2009 |
(American Pain Society Clinical Practice guidelines)
|84 trials and 24 reviews||Fusion is no more effective than intensive rehabilitation but has small to moderate benefits compared with standard nonsurgical therapy; clinical benefits of instrumented versus noninstrumented fusion is unclear|
|Han et al, 2009 |
|4 RCT’s||No significant difference in clinical outcome between instrumented posterolateral fusion and circumferential fusion|
Several studies have reported better clinical outcomes following a circumferential fusion for patients with degenerative disk disease, whereas others did not find any difference between various methods of fusion. A meta-analysis of randomized controlled trials (RCT) concluded that although circumferential fusion enhanced the fusion rate and reduced the reoperation rate, there was no significant difference in the global assessment of clinical outcomes compared with instrumented posterolateral fusion.
Are the Outcomes of Spinal Fusion Superior to Nonoperative Treatment?
Three RCTs found no significant difference when lumbar fusion was compared with a structured rehabilitation program using cognitive principles. Fritzell and colleagues, however, found lumbar fusion to be more efficacious than unstructured nonoperative therapy. Two reviews of RCTs comparing fusion with nonoperative treatment concluded that limitations in the RCTs prevented a definitive conclusion of comparative efficacy, whereas another systematic review concluded that fusion is no more effective than intensive rehabilitation but associated with small to moderate benefits compared with standard nonsurgical therapy. A meta-analysis of randomized trials concluded that cumulative evidence did not support surgical fusion for the treatment for chronic low back pain. The details of studies comparing fusion with nonoperative treatment are given in Table 2 .
|Authors and Type of Study||Sample Size and Follow-up||Outcomes||Comments|
|Fritzell et al, 2001 |
|73 noninstrumented posterolateral fusion; 74 instrumented posterolateral fusion; 75 anteroposterior fusion (circumferential fusion); 72 nonoperative treatment |
Age: 25–65 y
F/U : 2 y
|Greater improvement in pain, disability, return to work, and satisfaction in surgical group; 63% of surgical patients improved compared with 29% nonoperative; 75% of surgery patients would go through procedure again compared with 53% of nonoperative||Nonoperative group had unstructured (usual care) therapy |
Listhesis, postdiskectomy, fractures, and infection excluded
Various surgical procedures compared with nonoperative
Data excluded crossover patients
|Brox et al, 2003 |
|37 instrumented posterolateral fusion versus 27 cognitive/exercises |
Age: 25–60 y
F/U: 1 y
|No difference in ODI, back pain, and general function between groups; leg pain improved in surgical group; 71% surgical and 63% nonoperative patients rated their outcomes as successful (not significant)||Nonoperative group had exercises and cognitive therapy |
Excluded: patients with stenosis, radiculopathy, and previous surgery
|Fairbank et al, 2005 |
|176 fusion versus 173 nonoperative treatment |
Age: 18–55 y
F/U: 2 y
|ODI: minor difference in favor of surgery |
SF-36: no difference
Shuttle walk test: no difference between groups
|Nonoperative group had exercises and cognitive therapy |
High drop-out and crossover rate
Multiple diagnosis (listhesis, postlaminectomy, CLBP)
Multiple surgical procedures including Graf stabilization
|Brox et al, 2006 |
|28 instrumented posterolateral fusion versus 29 nonoperative treatment for postdiskectomy back pain |
Age: 25–60 y
F/U: 1 y
|No significant difference in ODI scores |
Overall success rate: 50% (surgical) versus 48%
|Nonoperative group had exercises and cognitive therapy |
Five patients did not attend treatment and two crossed over
The clinical outcomes following fusion for spondylolisthesis are favorable, whereas for lumbar disk degeneration, it is unclear whether fusion is more efficacious than a structured nonoperative program.
Fernstrom’s early attempts to retain segmental motion (1950s) using metallic balls led to subsidence because of the small area of contact and disparity in the modulus of elasticity between the spherical prosthesis and flat bony end plates. Structural failure of the elastomeric-viscoelastic polyolefin rubber and silicon core attached to two titanium plates in the Acroflex prosthesis (Depuy, Acromed Corporation, Cleveland, OH, USA) was noticed during clinical trials. Among the numerous patented artificial intervertebral disk designs very few have undergone clinical trials. The clinical outcomes following the implantation of two Food and Drug Administration (FDA) approved disk prosthesis are discussed next and are also listed in Table 3 .
|Blumenthal et al, 2005 |
|205 Charite TDR versus 99 ALIF |
Mean age: 39 y
F/U: 2 y
|No significant difference in VAS /ODI |
73% (TDR) and 53% satisfaction
70% (TDR) and 50% would have procedure again ( P <.05)
57% (TDR) and 46% clinical success (FDA criteria)
|Industry sponsored |
11 TDR (5.4%) and 9 (9.1%) fusion patients underwent additional surgery at index level
|Guyer et al, 2009 |
|90 Charite TDR versus 43 ALIF |
Mean age: 39 y
F/U: 5 y
|No significant difference in VAS /ODI /SF-36 (PCS) |
58% (TDR) and 51% clinical success (FDA criteria)
78% (TDR) and 72% of patients satisfied
|Industry sponsored |
43% lost to follow-up
|David, 2007 |
|108 one-level Charite TDR patients |
Mean age: 36 y
F/U: 10 y
|82% good-excellent outcome |
89% return to work with 77% return to previous hard labor
Mean flex-ext ROM = 10.1 degrees;
7.5% revision fusion; 8 patients (7.5%) underwent a posterior fusion at index level and 3 patients (2.8%) underwent adjacent level surgery (two disk herniation and one spinal canal stenosis)
|106 patients available for follow-up|
|Lemaire et al, 2005 |
|107 Charite prosthesis (1–3 levels) |
Mean age: 39 y
F/U: 10 y
|90% good-excellent outcome, 91% of those eligible returned to work, mean flex-ext ROM = 10.3 degrees |
Minor posttraumatic subsidence (2), periprosthetic ossification (3), adjacent level degeneration (2) and requiring secondary fusion (5)
|100 patients available for follow-up|
|Tropiano et al, 2005 |
|64 ProDisc patients (1–3 level) |
Mean age: 46 y
F/U: 8.7 y
|75% good-excellent results; significant improvements in back and leg pain and disability||55 available for follow-up|
|Zigler et al, 2007 |
|161 ProDisc TDR versus 75 APF (1–2 level) |
Mean age: 40 y
F/U: 2 y
|53% (TDR) and 40% success (FDA criteria) |
81%(TDR) and 69% would have same procedure again
No significant difference in ODI, pain, SF-36 scores
|Industry sponsored |
6 failures in TDR group: prosthesis migration (1), core migration (3), improper insertion of core (1), persistent pain requiring fusion (1)
|Berg et al, 2009 |
|80 TDR versus 72 fusion |
Age: 20–55 y
F/U: 2 y
|No significant difference in ODI, 84% (TDR) and 86%(fusion) improved, Minor difference in back and leg pain in favor of TDR||Three different prosthesis used and two fusion procedures (PLF, PLIF) |
Small numbers in each group
TDR, by restoring disk height, segmental lordosis, and segmental motion, can potentially reduce the risk of accelerated ALD. Biomechanical analysis has shown that the disk prosthesis preserved physiologic loading at adjacent nonoperated levels compared with spinal fusion. A retrospective radiographic study of 42 patients followed for 8 years following lumbar TDR showed that implanted prosthesis exhibiting greater than 5 degrees of motion revealed 0% and those with less than 5 degrees of motion showed a 34% incidence of ALD. TDR obviates the need for bone graft thereby avoiding donor site morbidity and the need for the expensive BMP to augment the fusion.
Charité Artificial Disk Prosthesis
Ninety percent of 100 patients implanted with the Charité III prosthesis (Depuy Spine, Raynham, MA, USA) reported good to excellent results (Modified Stauffer Coventry Scale) at 10 years with 80% returning to their previous employment. Although no prosthesis failure was reported, five patients underwent a secondary fusion. David reported the 10-year outcomes (mean follow-up, 13.2 years) following one-level disk arthroplasty with the Charité prosthesis in 106 patients. A total of 82% of the patients reported excellent or good clinical outcomes and 90% of implanted prostheses were mobile at the last follow-up. Eight patients (7.5%) underwent a posterior fusion at index level and three (2.8%) underwent adjacent-level surgery.
In the FDA investigational device exemption (FDA-IDE) study, patients with single-level disk degeneration between L4 and S1 were randomized to undergo either a Charité prosthesis implantation (N = 205) or a stand-alone anterior fusion with cages and autograft (N = 99). Clinical success at 2 years, defined as absence device failure, major complications, or neurologic deterioration and greater than 25% improvement in ODI score, was achieved in 57% of TDR patients and 46% of the fusion group ( P <.05). The study concluded that the 2-year clinical outcome was equivalent to anterior fusion and subsequently the Charité III device was approved by the FDA for implantation in patients with one-level disk degeneration from L4 to S1 who had failed at least 6 months of conservative treatment. Five-year follow-up of 90 Charité and 43 BAK fusion patients from the IDE study (44% of initial cohort) showed an overall success rate of 57% and 51%, respectively ( P = .0359). The improvement in the visual analogue scale (VAS) and ODI was similar in both groups at 2 and 5 years. The results at 5 years were similar to the 2-year results demonstrating noninferiority of the Charité group versus anterior fusion.
ProDisc-L Artificial Disk Prosthesis
Seven- to 11-year outcomes (mean, 8.7 years) following the ProDisc I implantation, in 55 out of 64 patients available for follow-up, showed 75% good to excellent results (relative improvement score >60%) and 64% were entirely satisfied with the results, whereas 5% were not satisfied. No device failure or migration was noted. As part of the randomized FDA-IDE study, patients with single-level disk degeneration between L3 and S1 who underwent TDR with the ProDisc-L (Synthes Spine, West Chester, Pennsylvania) prosthesis (N = 161) were compared with those who underwent a circumferential spinal fusion (N = 75) and followed-up for 2 years. Clinical success as defined by the FDA includes success for each individual case in the domains of ODI, SF-36, device-related, radiologic, and neurologic parameters, and was achieved in 41% of the fusion patients and 53% of the TDR patients. The study concluded that implantation of the ProDisc prosthesis was safe and superior to circumferential fusion using multiple clinical criteria. Subsequently, ProDisc-L was approved by the FDA in August 2006 for use in single-level disk degeneration between L3 and S1 following failure of 6 months of nonoperative management. A 3-year comparative analysis of the clinical outcomes (ODI, VAS, mean subjective improvement rate, and clinical success rate) following implantation of the ProDisc prosthesis and Charité prosthesis revealed similar results.
Financial Implications: Lumbar Fusion Versus TDR
With burgeoning health care expenditure, cost-benefit analysis of new technology needs careful consideration. Costs vary between hospitals, regions, states, and countries. Guyer and coworkers found that the hospital costs of ALIF plus autograft, ALIF plus BMP, and posterior lumbar interbody fusion were 12%, 36.5%, and 36.5% higher, respectively, compared with one-level TDR (Charité prosthesis) and concluded that the cost of a one-level TDR is less than or at worst equal to a one-level fusion. Another analysis of total hospital costs (excluding BMP) for transforaminal lumbar interbody fusion ($29,260), ALIF ($26,767), and TDR $27,972 was roughly similar and lower than anteroposterior fusion ($39,233). With inclusion of the cost of BMP, however, the cost of TDR was significantly lower than the three fusion procedures (transforaminal lumbar interbody fusion, $34,660; ALIF, $32,167; anteroposterior fusion, $44,633). Levin and coworkers also studied the hospital costs and found similar implant cost for one-level TDR and circumferential fusion but higher operating room and surgeon charges for fusion resulted in a 23% higher cost for fusion surgery. In the two-level surgery group, the implant costs were lower in the fusion group ($18,460 vs $23,000) but there was no significant difference in the total cost. Similarly, an Australian federal government report did not find a significant difference in costs (hospitalization plus prosthesis plus medical fees) between fusion ($26,854 Australian dollars) and lumbar TDR ($24,139 Australian dollars).
Summary: Lumbar Fusion Versus TDR
The short-term outcomes of lumbar TDR are equivalent to that following lumbar fusion. The cost of TDR in United States hospitals seems to be less than or equal to fusion. Lack of adequate long-term follow-up data prevents an analysis of the potential benefit of TDR in preventing or reducing the incidence of ALD.
ACDF was described by Bailey and Badgley in 1939 and later modified by Cloward (1953) and Smith and Robinson (1955) to treat cervical radiculopathy or myelopathy with and without axial neck pain. Unlike the outcomes of fusion for lumbar degenerative disk disease, the outcomes following ACDF are generally favorable and range from 90% to 95% and are equivalent to hip arthroplasty. Nevertheless, there are a few shortcomings including ALD, pseudoarthrosis, and donor site morbidity.
In vitro studies have shown increased segmental motion, intradiscal pressure, and facet forces at levels adjacent to a cervical fusion construct. Goffin and coworkers reported a 92% incidence of ALD 5 years after an ACDF with a 6% incidence of adjacent-level surgery following a failure to respond to nonoperative treatment. Hilibrand and coworkers predicted a 25.6% incidence of symptomatic ALD within 10 years (2.9% per year) after an ACDF. A higher rate of ALD following one-level fusion compared with multilevel fusion, however, led the authors to believe that the changes were a natural progression of spondylosis rather than a consequence of surgery. Further, the 3.9% per year incidence of ALD, following noninstrumented posterior foraminolaminotomy, is roughly equal to the incidence following ACDF. A prospective study revealed significantly higher rates of radiologic ALD following an ACDF, however, compared with TDR followed-up for 24 months. Additionally, a serial MRI study showed increased disk degeneration adjacent to an ACDF compared with a posterior laminoplasty. The debate continues, however, on this subject.
In vitro studies showed that one- and two-level TDR, in addition to restoring motion at the index level, do not increase the motion at the adjacent levels, thereby potentially reducing the incidence of ALD. TDR also precludes the need for autologous bone graft, avoiding bone graft donor site morbidity, which is a shortcoming of ACDF.
ACDF Versus Cervical TDR
Two-year follow-up of a randomized controlled trial of 99 patients with one-level disk disease who underwent either ACDF or cervical TDR (Bryan artificial disk prosthesis, Medtronic Sofamor Danek, Memphis, TN, USA) from three participating centers of an FDA-IDE trial were reported. Significantly better outcomes in the domains of neck disability index, arm and neck pain, and the SF-36 physical component score were found in the TDR group. Revision surgery was performed in four patients from the fusion group for nonunion (N = 2) and ALD (N = 2) and two patients from the TDR group for ALD. The same authors reported on the outcomes following an RCT with 2-year follow-up of 115 patients showing statistically better outcomes relating to neck disability index (NDI), neck pain, and SF-36 physical component score in the TDR group. Another participating center of the FDA-IDE trial showed similar clinical outcomes in 33 patients with single-level disk disease randomized to have either ACDF (N = 16) or TDR (N = 17).
Heller and coworkers reported the results of a multicenter RCT comparing TDR (Bryan disk prosthesis) (242 patients) with ACDF (221 patients) followed for 2 years. The patients who underwent TDR showed significantly better outcomes in the domains of NDI and neck pain but there was no difference between the two groups with regards to arm pain, SF-36 scores, neurologic success, and number of secondary surgical procedures. Overall success, as defined as greater than 15-point improvement in NDI scores, maintenance or improvement in neurologic status, absence of implant- or implant-surgical procedure–related serious adverse events, and absence of subsequent surgery or intervention, was statistically higher in the TDR group. An RCT comparing the outcomes of TDR (Bryan disk prosthesis) with ACDF in 65 patients with two-level disk disease revealed significantly better outcomes in the TDR group with respect to NDI and arm and neck pain. In the TDR group, 29 patients reported good to excellent outcomes and one reported a fair outcome compared with 27 patients with good to excellent outcomes, fair outcomes in four patients, and poor outcomes in one patient.
Six-month and 1-year follow-up of an RCT comparing outcomes following single-level TDR (ProDisc-C, Synthes Spine, West Chester, PA, USA) with ACDF found similar improvements in arm and neck pain in both groups. Murrey and coworkers reported the outcome of an RCT comparing the outcomes of a single-level TDR (ProDisc-C) and ACDF. At the 2-year mark, there was no statistical difference between the two groups with relation to NDI, neurologic success, VAS score for neck and arm pain, SF-36, and patient satisfaction. A statistically higher proportion of the TDR patients reported overall success according to the FDA criteria and the minimally clinically important difference criteria. A few studies comparing the outcomes of TDR with ACDF are listed in Table 4 .