Lumbar spinal stenosis (LSS) with degenerative spondylolisthesis (DS) is the most common indication for lumbar fusion surgery in the United States1–4. Two Level-I studies have presented conflicting results regarding the relative efficacy of decompression alone versus decompression plus fusion5,6. The ongoing controversy regarding these competing surgical approaches is further fueled by national database studies and randomized prospective studies demonstrating high reoperation rates (10% to 30% at 3 years, 30% to 40% at 10 years) with both decompression alone and decompression plus fusion6–10. Given the substantial impact of revision spine surgery on both patient-reported outcome measures (PROMs) and health-care costs (estimated 2-year cost of revision lumbar spine surgery, >$100,000), there remains strong interest in identifying the optimal surgical approach to LSS with DS8,11,12.

Decompression plus lumbar facet arthroplasty is an alternative treatment for LSS with DS. Lumbar facet arthroplasty allows for thorough decompression of the neural elements, stabilization of the spondylolisthesis, and preservation of motion, thereby limiting both the persistent instability that leads to recurrent stenosis following isolated decompression and the pathologic redistribution of motion that leads to adjacent segment degeneration following fusion. At present, there are no U.S. Food and Drug Administration (FDA)-approved lumbar facet arthroplasty devices. The Total Posterior Spine System (TOPS; Premia Spine) is a lumbar facet arthroplasty device that has been the subject of a recently concluded FDA Investigational Device Exemption (IDE) trial comparing the relative efficacy of lumbar facet arthroplasty versus fusion to treat lumbar grade-I DS with stenosis. Lumbar facet arthroplasty with the TOPS device has shown promise in retrospective clinical studies and in a single-arm interim review of the TOPS IDE trial13–16.

The purpose of the present study is to report the results of the primary 2-year outcome of the TOPS IDE trial, as well as secondary outcomes including PROMs, radiographic parameters, and complications. We hypothesized that lumbar facet arthroplasty would outperform fusion in the primary composite outcome while successfully stabilizing the spondylolisthesis and maintaining motion at the surgical level.

Materials and Methods Study Design and Oversight

In this randomized controlled IDE trial (IDE number G160168, ClinicalTrials.gov NCT03012776), patients were assessed for eligibility during the study period from July 17, 2017, to June 20, 2022, at 37 medical centers. The majority of patients were enrolled at 1 of 10 sites (see Appendix A,1). The primary outcome was the overall rate of composite clinical success at 24 months postoperatively. Data from a prior FDA IDE trial (G060063) provided the basis for the determination that 303 patients (202 arthroplasty, 101 fusion) would need to be randomized to provide 80% power to reject the null hypothesis of equality. Patients were randomized in a 2:1 ratio to either decompression plus lumbar facet arthroplasty or decompression plus fusion. Randomization was by blocks, with randomly varying block sizes to minimize the likelihood that a site would be able to infer the next randomization assignment in the sequence. Randomization schedules were maintained centrally by ProSoft Clinical, which utilized a secure software application to perform the randomization and transmit the treatment allocation directly to center-specific study personnel within 24 hours prior to surgery. Patients were blinded prior to surgery but were informed of their assignment postoperatively due to the impracticality of ensuring that they did not view their radiographs during follow-up appointments. Physicians and radiologists were not blinded to the treatment assignment. Appendix A,2 provides details regarding third-party data management, trial surveillance, independent review of indeterminate clinical events, and data transmission. Institutional review board approval was obtained at all participating sites prior to enrolling patients. All enrolled patients were counseled extensively regarding the surgical options and the ongoing clinical trial, and provided written informed consent.

Patients

Patients with LSS and grade-I DS were eligible for inclusion. Participants were required to be between 35 and 80 years of age and to have undergone >6 months of nonsurgical therapy that was unsuccessful. Patients were also required to have an Oswestry Disability Index (ODI) score of at least 40 of 100 and a visual analog scale (VAS) score for leg pain of at least 40 of 100 for at least 1 leg at baseline. Patients were excluded if >1 motion segment required a surgical procedure, radiographs revealed substantial disc collapse (disc height < 4 mm at the index level), a prior surgery had been performed at the index level or an adjacent level (unless it involved only the posterior elements), or instrumented lumbar spine surgery had been performed at any level. A complete overview of inclusion and exclusion criteria is provided in Appendix A,3.

Trial coordinators at each site screened and enrolled patients. Radiographs and magnetic resonance imaging scans were reviewed centrally for each enrolled patient to verify the presence of LSS with grade-I DS without disc herniation.

Interventions

Patients underwent either decompression plus transforaminal interbody fusion (fusion group) or decompression plus lumbar facet arthroplasty (arthroplasty group) at the single level of spondylolisthesis (Fig. 1). Appendix A,4 provides additional details regarding these 2 interventions.

Fig. 1:

Schematic (Fig. 1-A), anteroposterior radiograph (Fig. 1-B), and lateral radiograph (Fig. 1-C) demonstrating the TOPS device affixed to pedicle screws.

Outcomes

The primary outcome was a study-specific composite clinical success measure determined at 2 years postoperatively. We defined clinical success as meeting all 5 of the following criteria: (1) the absence of reoperation, lumbar injection, or spinal cord stimulator implantation; (2) the absence of a major device adverse event; (3) a reduction in the ODI of ≥15 points; (4) the absence of a new or progressive neurologic deficit; and (5) the absence of fusion failure (pseudarthrosis in the fusion group or spontaneous fusion in the arthroplasty group)17,18. A major device adverse event was defined as device component degradation, breakage, separation, disassembly, or loosening (including screw loosening) or an increase in spondylolisthesis at the index level by at least 1 grade at any postoperative time point. These major adverse events were specifically device- or spine-related and did not include medical adverse events. An improvement in the ODI of ≥15 points has been determined to be the minimum clinically important difference (MCID) in multiple previous randomized controlled trials17–19. Appendix A,5 describes the definition of a neurologic deficit utilized in this study and limitations inherent to the neurologic examination. Appendix A,6 details the methodology used to assess fusion status with radiographs.

Patients were required to have reached 24 months of postoperative follow-up to be deemed a clinical success; however, a patient could be deemed an early clinical failure due to reoperation, lumbar injection, a major device adverse event, or development of a new neurologic deficit, or by the presence of a fusion in the arthroplasty treatment group, prior to 24 months. Consequently, the number of patients analyzed for the primary outcome of composite clinical success was the sum of those patients who reached 24 months of clinical follow-up and those patients who were deemed early clinical failures.

Additional prespecified outcomes included mean ODI scores, VAS back and leg pain scores, Zurich Claudication Questionnaire (ZCQ) scores, Short Form (SF)-12 scores, surgical variables (estimated blood loss, length of stay, time in surgery), postoperative pain medication use, complications, reoperations, and radiographic alignment and motion parameters. Motion in both flexion-extension and lateral bending at the surgical level was measured in degrees on dynamic radiographs. Spondylolisthesis was graded according to the Meyerding classification and was determined to be stabilized if there was no postoperative increase in grade20. Reoperation was performed at the discretion of the treating surgeon. The proportion of patients achieving the MCID in the ODI (≥15-point improvement), VAS back pain (>10.5-point improvement), VAS leg pain (>10.5-point improvement), and ZCQ (>0.75-point improvement) was determined on the basis of established MCID values in the literature21,22.

Enrollment, Randomization, and Statistical Analysis

Figure 2 shows the enrollment, randomization, and follow-up for the TOPS IDE trial. Overall, 541 patients were assessed for eligibility, of whom 196 were excluded on the basis of the trial inclusion and exclusion criteria and 24 had not yet been randomized to a treatment group at the time of this analysis. The remaining 321 patients were randomized, with 219 patients assigned to undergo facet arthroplasty and 102 patients assigned to undergo fusion. Of the 219 patients randomized to the arthroplasty group, 8 withdrew consent, 2 received a non-study treatment, 1 had not yet undergone surgery, and 2 were excluded by the independent review panel, leaving 206 patients eligible for analysis. Of the 102 patients assigned to the fusion group, 4 withdrew consent, 2 received a non-study treatment, 2 had not yet undergone surgery, and 1 was excluded by the independent review panel, leaving 93 patients eligible for analysis. No crossovers occurred from either treatment group.

Fig. 2:

Eligibility, randomization, and follow-up.

In accordance with the predetermined statistical plan, an interim analysis of the primary outcome was performed when >300 patients were randomized, even though not all patients had reached 24 months of follow-up. Of the 206 arthroplasty group patients and 93 fusion group patients eligible for analysis, 105 patients (51.0%) in the arthroplasty group and 43 (46.2%) in the fusion group had reached 24 months of follow-up. An additional 8 patients (3.9%) in the arthroplasty group and 4 (4.3%) in the fusion group were deemed early clinical failures. Thus, a total of 160 patients were included in the primary outcome analysis, including 113 patients (54.9% of randomized patients) in the arthroplasty group and 47 patients (50.5% of randomized patients) in the fusion group. Although only approximately one-half of the randomized patients were analyzed for the primary outcome, the between-group difference in the primary outcome was sufficient to satisfy the predetermined criteria for concluding the clinical trial (see Appendix A,7). Although false-negative results can occur when a clinical trial is concluded or analyzed prematurely, a larger-than-expected between-group difference is an established precedent to conclude a clinical trial comparing 2 surgical interventions23,24.

Of the patients not assessed for the primary outcome, 4 patients in each group were lost to follow-up, and the remaining patients are being followed prospectively but have not yet reached 24 months of postoperative follow-up. Patients who underwent reoperation and thus had treatment failure according to the composite end point were included in the primary analyses, but their PROMs were censored at reoperation to avoid confounding the results of the primary treatment by including those of successful secondary treatment following a failed primary treatment25.

Baseline demographics and operative characteristics were compared between the treatment groups using mean differences and 95% confidence intervals (CIs) for continuous variables, differences in percentages and 95% CIs for binary variables, and Fisher exact tests for categorical variables (see Appendix B).

Results Patients

Baseline characteristics of the patients assigned to the arthroplasty and fusion treatment groups are shown in Table I. No differences between treatment groups, including in baseline PROMs, were identified.

TABLE I - Baseline Characteristics of the Patients* Arthroplasty (N = 206) Fusion (N = 93) P Value Demographics  Age (yr) 63.3 ± 8.2 63.9 ± 8.6 0.53  BMI (kg/m 2 ) 29.4 ± 4.9 29.9 ± 5.3 0.43  Female sex 116 (56.3%) 50 (53.8%) 0.71 Race 0.63  White 191 (92.7%) 86 (92.5%)  Black 3 (1.5%) 3 (3.2%)  Asian 3 (1.5%) 2 (2.2%)  Other 9 (4.4%) 2 (2.2%) Medical comorbidities  Congestive heart failure 0 (0.0%) 1 (1.1%) 0.74  Diabetes mellitus 6 (2.9%) 5 (5.4%) 0.11  Osteopenia 13 (6.3%) 6 (6.5%) 0.92  Depression 29 (14.1%) 13 (14.0%) 0.89  Chronic kidney disease 9 (4.4%) 2 (2.2%) 0.43  COPD 3 (1.5%) 2 (2.2%) 0.61 Use of nicotine products 0.97  No, never smoked 127 (61.7%) 59 (63.4%)  No, but prior history 73 (35.4%) 32 (34.4%)  Current smoker 6 (2.9%) 2 (2.2%) Prior lumbar surgery 0.80  Yes 12 (5.8%) 6 (6.5%)  No 194 (94.2%) 87 (93.5%) Level implanted 0.59  L1-L2 0 0  L2-L3 0 0  L3-L4 10 (4.9%) 6 (6.5%)  L4-L5 196 (95.1%) 87 (93.5%) *Group means of baseline patient-reported outcome measures are reported in Table III. Continuous variables are given as the mean ± standard deviation, and categorical variables are given as the number with the percentage in parentheses. BMI = body mass index, COPD = chronic obstructive pulmonary disease.
Primary Outcome

At 24 months after surgery, a significantly greater proportion of patients achieved composite clinical success in the arthroplasty group than in the fusion group (83 patients [73.5%] versus 12 patients [25.5%]; p < 0.001), equating to a between-group difference of 47.9% (95% CI, 33.0% to 62.8%) (Table II). Consequently, the null hypothesis was rejected, and it can be concluded that facet arthroplasty is superior to fusion in terms of the primary outcome. The fusion group demonstrated higher rates of developing a new or progressive neurologic deficit (11.4% versus 2.8%; p = 0.047) and fusion failure (43.9% versus 1.0%; p < 0.001) at 24 months postoperatively. Appendix A,8 provides additional details regarding between-group differences in the various components of the primary outcome.

TABLE II - Primary End Point: Composite Clinical Success Arthroplasty (N = 113*) Fusion (N = 47*) P Value Rate† % Rate† % Composite clinical success  Yes 83/113 73.5% 12/47 25.5% <0.001  No‡ 30/113 26.5% 35/47 74.5% <0.001   Reoperation, lumbar injection, or stimulator implantation 13/113 11.5% 5/47 10.6% 0.28   Major device adverse event 7/105 6.7% 2/41 4.9% 1.00   ODI reduction < 15 points§ 6/96 6.3% 8/35 22.8% 0.01   Fusion failure# 1/104 1.0% 18/41 43.9% <0.001   New or progressive neurologic deficit 3/107 2.8% 5/44 11.4% 0.047

*The number of patients in each treatment group represents all patients who reached 24 months of postoperative follow-up (105 arthroplasty, 43 fusion) plus those patients who had early treatment failure (8 arthroplasty, 4 fusion), defined as failure in ≥1 measure of the composite clinical success criteria prior to 24 months.

†Number of patients who met the criterion in the row/number of patients available to be evaluated for the criterion in the row.

‡Some patients failed to achieve the primary end point for >1 reason.

§The end point was censored if the patient underwent reoperation or lumbar injection prior to outcome measurement.

#Defined as lack of fusion at 24 months postoperatively in the fusion group or the presence of spontaneous/unintended fusion in the arthroplasty group. The outcome was censored in patients who underwent reoperation prior to fusion assessment.

Secondary Outcomes

The arthroplasty group demonstrated significantly lower VAS back pain at all postoperative time points, and the magnitude of the difference in treatment effect grew from 6 to 24 months postoperatively (Fig. 3,Table III). A greater proportion of patients in the arthroplasty group achieved the MCID in the ODI (93.8% versus 77.1%, p = 0.01) and in VAS back pain (84.4 versus 61.8%, p = 0.01) at 24 months postoperatively. The arthroplasty group demonstrated superior SF-12 physical (mean and standard deviation, 48.3 ± 11.2 versus 42.8 ± 13.6; p = 0.02) and mental (55.1 ± 8.2 versus 50.1 ± 10.2; p = 0.005) component summary scores at 24 months postoperatively (Fig. 4; see also Appendix A,9).

Fig. 3:

Mean scores on the Oswestry Disability Index (scores range from 0 to 100, with higher scores indicating more disability related to back pain) (Fig. 3-A), visual analog scale for low back pain (Fig. 3-B) and worst leg pain (Fig. 3-C) (scores range from 0 to 100, with higher scores indicating more pain in the specified anatomic region), and each component of the Zurich Claudication Questionnaire (ZCQ; scores range from 1 to 5 on the symptom score and from 1 to 4 on the physical and satisfaction scores, with lower numbers representing less severe symptoms related to neurogenic claudication, greater physical function, and greater satisfaction, respectively) (Figs. 3-D, 3-E, and 3-F), before and after surgery, among patients who were randomly assigned to undergo decompression plus lumbar facet arthroplasty or decompression plus fusion. *A significant difference (p < 0.05) between groups at the indicated time point.

TABLE III - Patient-Reported Outcome Measures* Arthroplasty (N = 206) Fusion (N = 93) P Value MCID Achievement P Value N Value† N Value† Arthroplasty Fusion ODI  Preop. 206 56.5 ± 12.1 93 55.8 ± 13.1 0.75  Week 6 194 23.5 ± 16.4 84 30.2 ± 17.0 0.001 84.0% 73.8% 0.07  Month 3 183 15.7 ± 16.5 82 22.1 ± 17.8 0.02 89.1% 84.1% 0.31  Month 6 171 13.4 ± 15.5 74 16.9 ± 15.9 0.01 91.8% 90.5% 0.81  Month 12 143 11.6 ± 13.7 65 16.9 ± 17.2 <0.001 94.4% 89.2% 0.25  Month 24 105 9.4 ± 14.5 43 21.1 ± 22.3 <0.001 93.7% 77.2% 0.01  P value‡ <0.001 <0.001 VAS, low back pain  Preop. 206 68.6 ± 23.3 93 69.5 ± 22.2 0.75  Week 6 194 18.5 ± 18.0 83 27.7 ± 25.3 0.001 83.5% 68.7% 0.009  Month 3 183 16.2 ± 21.3 82 23.1 ± 24.5 0.02 83.6% 79.3% 0.39  Month 6 171 14.7 ± 21.1 74 22.7 ± 24.8 0.01 86.0% 79.7% 0.26  Month 12 143 12.4 ± 19.6 65 24.5 ± 27.6 <0.001 86.0% 76.9% 0.11  Month 24 105 11.1 ± 18.1 43 30.9 ± 33.1 <0.001 84.4% 61.8% 0.01  P value‡ <0.001 <0.001 VAS, worst leg pain  Preop. 206 82.7 ± 13.5 93 85.1 ± 10.8 0.13  Week 6 194 12.9 ± 20.5 84 17.9 ± 25.2 0.09 92.8% 92.8% 1.00  Month 3 182 13.3 ± 22.5 82 15.9 ± 23.7 0.38 94.5% 92.7% 0.58  Month 6 171 12.9 ± 22.7 74 17.0 ± 24.9 0.21 92.4% 91.9% 1.00  Month 12 143 12.8 ± 22.0 65 18.7 ± 27.8 0.10 94.4% 90.8% 0.38  Month 24 105 13.7 ± 24.2 43 23.3 ± 33.8 0.08 90.6% 88.2% 0.74  P value‡ <0.001 <0.001 ZCQ, symptom  Preop. 206 3.7 ± 0.6 93 3.7 ± 0.6 0.63  Week 6 194 2.0 ± 0.7 84 2.1 ± 0.8 0.04 93.8% 90.5% 0.32  Month 3 180 1.8 ± 0.7 81 2.0 ± 0.7 0.09 95.6% 96.3% 1.00  Month 6 169 1.8 ± 0.7 74 1.9 ± 0.7 0.22 95.9% 93.2% 0.36  Month 12 141 1.7 ± 0.7 63 2.0 ± 0.8 0.04 96.5% 92.3% 0.29  Month 24 103 1.7 ± 0.8 41 2.2 ± 1.0 0.01 93.8% 85.7% 0.16  P value‡ <0.001 <0.001 ZCQ, physical  Preop. 206 2.9 ± 0.4 93 2.9 ± 0.4 0.50  Week 6 194 1.7 ± 0.6 84 1.9 ± 0.6 0.01 86.7% 76.2% 0.04  Month 3 180 1.5 ± 0.6 81 1.6 ± 0.6 0.11 91.2% 91.5% 1.00  Month 6