Five-year Results of a Randomized Controlled Trial for Lumbar Artificial Discs in Single-level Degenerative Disc Disease

Abstract

Study Design. 

A prospective, multicenter, randomized, controlled, investigational device exemption (IDE) noninferiority trial.

Objective. 

The aim of this study was to compare the 5-year safety and effectiveness of the activL Artificial Disc with Control Total Disc Replacement (TDR) systems (ProDisc-L or Charité) in the treatment of patients with symptomatic single-level lumbar degenerative disc disease (DDD).

Summary of Background Data. 

The activL Artificial Disc received Food and Drug Administration approval in 2015 based on 2-year follow-up data.

Methods. 

Eligible patients presented with symptomatic, single-level, lumbar DDD who failed ≥6 months of nonsurgical management. At entry, 324 patients were randomly allocated (2 : 1) to treatment with activL (n = 218) or Control (n = 106, including n = 65 ProDisc-L and n = 41 Charité) TDR. At 5-year follow up, a total of 261 patients (176 activL patients and 85 Control patients) were available for analysis.

Results. 

The primary composite endpoint at 5 years for activL patients was noninferior to Control TDR. Relative to baseline, reductions in back pain severity and improvements in Oswestry Disability Index (ODI) were maintained for both the activL and Control TDR groups through 5 years. The activL group showed significantly better range of motion for flexion-extension rotation, flexion-extension translation, and disc angle, compared with Control TDR. Freedom from a serious adverse event through 5 years was 64% in activL patients, 47% in Control patients (log-rank P = 0.0068). Freedom from index-level and adjacent-level reoperation was high for TDR patients, ranging between 94% and 99%, respectively.

Conclusion. 

Long-term evidence supports lumbar total disc replacement as safe. The next-generation activL Artificial Disc is more effective at preserving range of motion than first-generation lumbar TDRs (ProDisc-L and Charité) and offers a higher safety profile. Other primary and secondary outcomes are similar between disc designs.

Level of Evidence: 2

In Brief

Five-year follow-up safety and effectiveness results were reported for the prospective, multicenter, randomized, controlled, investigational device exemption noninferiority trial for the activL Artificial Disc, compared with control total disc replacement (including ProDisc-L and Charité).

Over the past decade, lumbar total disc replacement (TDR) has been shown to be safe and effective in the treatment of lumbar discogenic low back pain (LBP) caused by degenerative disc disease (DDD). The long-term efficacy and safety of lumbar TDR devices, with 5-year data, is now available from United States Investigational Device Exemption (IDE) randomized trials.^1–4 In addition, level-1 meta-analyses of 5-year US IDE data support improved outcomes for lumbar TDR relative to fusion in DDD.^5,6

The mechanism of action of TDR is to preserve physiological motion by replicating the biomechanical performance of the intact intervertebral disc, which may ultimately reduce the risk of adjacent segment disease (ASD) compared with lumbar fusion.⁷ This premise is supported by a recent indirect comparison⁸ of TDR patients from the activL IDE study and lumbar fusion patients from Zigler and Delamarter,³ which reported a significantly lower rate of progression in radiographic adjacent-level degeneration for TDR than lumbar fusion over 5 years.⁸ Further, a 5-year meta-analysis demonstrated significant improvement of adjacent segment pathology and index-level range of motion (ROM), with lumbar TDR compared with fusion.⁶

In addition to ProDisc-L (Centinel Spine, West Chester, PA), the activL Artificial Disc (Aesculap Implant Systems, Center Valley, PA) has received US Food and Drug Administration (FDA) approval. A 2-year randomized trial that compared the performance of the activL Artificial Disc with other FDA-approved lumbar artificial discs, including Charité (DePuy Spine, Inc., Raynham, MA) and ProDisc-L, demonstrated that the primary composite endpoint was met whereby activL TDR was noninferior to control TDR (P < 0.001).⁹ A protocol-defined analysis of the primary composite endpoint also confirmed that the activL Artificial Disc was superior versus controls (59% vs. 43%; P < 0.01), due largely in part to the superior ROM outcomes.⁹ These patients have continued follow-up for postmarket surveillance, and their 5-year outcomes are reported in this paper.

MATERIALS AND METHODS

Overview of Study Design

This prospective, multicenter, randomized controlled IDE trial was approved by the U.S. FDA and the institutional review board at each participating site. The trial was prospectively registered at ClinicalTrials.gov (NCT00589797). In brief, eligible patients reported lumbar pain and back dysfunction due to a radiographically confirmed diagnosis of DDD at a single symptomatic level (L4-L5 or L5-S1) following at least 6 months of nonsurgical management (see Table, Supplemental Digital Content 1, https://links.lww.com/BRS/B449, for key study inclusion and exclusion criteria). Patients were randomly allocated (2 : 1) to implant with activL or Control discs. In patients randomized to the Control group, choice of TDR (ProDisc-L or Charité) was at the discretion of the investigator. Detailed descriptions of the activL,^9,10 ProDisc-L,³ and Charité^11,12 TDRs have been previously reported.⁹ Patients remained blinded to treatment assignment through 2 years post-treatment. Patients returned for follow-up visits at 6 weeks, 3 months, 6 months, and annually thereafter for 5 years. A physical examination, neurological assessment, and six-view x-rays were performed at follow-up visits. Additional details regarding study design including sample size, study hypotheses, data quality, and procedures are reported elsewhere.⁹

Outcomes

The primary endpoint of this trial was a composite treatment success outcome at 2 years. Treatment success required patients to meet all the following criteria: (1) ≥15-point improvement in ODI, (2) maintenance or improvement in neurological status, (3) maintenance or improvement in ROM at index level, (4) freedom from revision, reoperation, removal, or supplemental fixation at index level, and (5) freedom from serious device-related adverse events (AEs). Secondary outcomes included back pain severity and leg pain severity on a 0 to 100 Visual Analogue Scale (VAS; both back and leg pain were measured at rest, and patients were considered responders if they achieved at least 20 mm VAS improvement compared with baseline; for leg pain severity, no worsening in other leg was permitted), ODI, health-related quality of life assessed with the SF-36 questionnaire, including Physical Component Summary (PCS) and Mental Component Summary (MCS), patient satisfaction, return to work, narcotic usage, ROM at the index level, radiographic evaluations of device status, AEs, and reoperations (defined as any surgical procedure at the level of the original implant that does not include removal, modification, or addition of any components of the system). A serious AE was defined as any event that was fatal, life-threatening, required prolonged hospitalization, resulted in permanent anatomic or physiological impairment, caused a malignant tumor, or resulted in distress, congenital anomaly, or death of a fetus. Serious device-related AEs were those serious AEs attributable to TDR.

Statistical Methods

Continuous data were reported as mean ± standard deviation or median (min-max) depending on normality assumptions. Categorical data were reported as frequencies and percentages. Comparisons of continuous data were made using a two-sample t test, while comparisons of categorical data were performed using a Fisher exact test. Time to event data were analyzed using Kaplan-Meier methods with a log-rank test for group differences. Statistical significance was set at P < 0.05. Imputation methods used to account for missing patient data for the primary composite endpoint included multiple imputation = missing data replaced using multiple imputation procedures (SAS method),¹³ complete case = missing data excluded from denominator, LOCF = last observation carried forward, missing = failure –missing data counted as failure in each group, best case = missing data counted as success for activL and failure for control, worst case = missing data counted as failure for activL and success for control, and missing = success –missing data counted as success in both groups. Multiple imputation analyses were presented for all other results. Complete case analyses were presented as supplemental material. In addition to comparing activL to both control TDRs (ProDisc-L + Charité), activL was also compared with ProDisc-L alone to assess only TDRs that are currently commercially available.

RESULTS

Patient Characteristics

Between January 2007 and December 2009, 324 patients were treated with activL (n = 218) or Control (n = 106; Charite = 41, ProDisc-L = 64) artificial discs at 14 sites. A total of 81% patients returned for a 5-year follow-up visit (see Figure, Supplemental Digital Content 2, https://links.lww.com/BRS/B449, for the CONSORT flow diagram). For patients with 5-year follow-up data, baseline variables for patient demographics, medical history, symptoms, and ROM were well-matched between groups (Table 1). The PCS for health-related quality of life was significantly different between patients with 5-year follow up data for activL and Control groups (P = 0.02). Mean patient age was 40 years and 49% of patients were female. The total number of patients with L5-S1 and L4-L5 disease was 190 and 71, respectively. More than 90% of the trial patients had leg pain, in addition to their primary complaint of back pain (92% in activL and 93% in controls).

TABLE 1: Baseline Patient Characteristics (ITT and at 5 yr)

Primary Composite Endpoint

Using margins of 10% and 15%, the activL group was noninferior to the Control group for the primary composite outcome at 5 years (Figure 1). All imputation analyses supported this finding, except for worst case. Individual components of the primary composite endpoint were mostly numerically higher for patients with the activL Artificial Disc compared with the Control disc for multiple imputation (Table 2) and complete case (see Table, Supplemental Digital Content 3, https://links.lww.com/BRS/B449) analyses, although there were no statistically significant differences between groups.

Figure 1: Primary and sensitivity analyses for composite endpoint at 5 years for activL versus Control artificial discs. Noninferiority with the activL Artificial Disc is demonstrated if the entire 95% confidence interval is greater than the prespecified noninferiority margin. Superiority with the activL Artificial Disc is demonstrated if the entire 95% confidence interval is greater than 0.

TABLE 2: Components of the Composite Treatment Success Endpoint at 5 Years for activL versus Control Artificial Discs

Secondary Clinical Outcomes

At 5-year follow-up, the activL Artificial Disc was effective in reducing back pain, improving back function, and increasing health-related quality of life. Mean back pain severity decreased from 79 mm at baseline to 15 mm for activL and from 79 mm to 17 mm for Control at 5 years (P = 0.00 for change from baseline for each group) (Figure 2). Mean ODI values decreased from 57 to 14 with activL patients and from 59 to 16 with Control patients (P = 0.00 for change from baseline for each group) (Figure 3). Similar results were produced using complete case imputation analyses for back pain severity (See Figure, Supplemental Digital Content 4, https://links.lww.com/BRS/B449) and ODI (see Figure, Supplemental Digital Content 5, https://links.lww.com/BRS/B449). Significant improvements in health-related quality of life were realized in both groups over 5 years. The proportion of patients who maintained a PCS improvement ≥15% at 5 years compared with baseline was high for activL and Control patients (87% and 82%, P = 0.24). Approximately half of the patients maintained an MCS improvement at 5 years compared with baseline for activL and Control patients (53% and 54%, P = 0.92). Further, >90% of patients stated they would definitely or probably undergo TDR again.

Figure 2: Back pain severity Visual Analogue Scale score (mm) through 5 years post-treatment for activL versus Control artificial discs. Multiple imputation was used for missing patient data. Values are mean ± 95% confidence interval.

Figure 3: Oswestry Disability Index scores through 5 years post-treatment for activL versus Control artificial discs. Multiple imputation was used for missing patient data. Values are mean ± 95% confidence interval.

Radiographic Findings

At 5 years, patients with the activL Artificial Disc showed significantly greater ROM for flexion-extension rotation (P = 0.02) and flexion-extension translation (P = 0.03) compared with Control, as well as significantly greater disc angle (P = 0.01) (Table 3). Disc height was similar between activL and Control at 5 years (P = 0.70). Decreased back and leg pain VAS scores and ODI scores were significantly correlated with preservation of ROM (flexion-extension rotation) after TDR surgery (Table 4). Complete case imputation analyses generally aligned with multiple imputation results for radiographic findings (see Table, Supplemental Digital Content 6, https://links.lww.com/BRS/B449) and correlation of ROM with pain and function scores (see Table, Supplemental Digital Content 7, https://links.lww.com/BRS/B449).

TABLE 3: Radiographic Findings: 5-year Endpoint for activL versus Control and versus ProDisc-L Artificial Discs^∗

TABLE 4: Correlation of Total Disc Replacement Flexion-Extension Rotation with Pain and Function Scores at 5 Years

Complications

Patients with the activL Artificial Disc had a significantly lower risk of serious AEs when compared with control through 5 years; freedom from a serious AE was 64% with activL and 47% with Control artificial discs (log-rank P = 0.0068) (Figure 4). The activL patients experienced significantly lower lumbar/leg pain at 5 years compared with Controls (P = 0.05), while other device-related serious AEs were similar between groups (Table 5). When comparing activL to Control artificial discs, freedom from lumbar reoperation was 95% versus 90% (log-rank P = 0.07), freedom from index level reoperation was 95% versus 94% (log-rank P = 0.81), and freedom from adjacent level reoperation was 99% versus 94% (log-rank P = 0.01).

Figure 4: Freedom from serious adverse events through 5 years for activL versus Control artificial discs. Kaplan-Meier estimate is 64% with activL and 47% with Control. Log-rank P = 0.0068.

TABLE 5: Device-related Serious Adverse Events at 5 Years for activL versus Control Artificial Discs

Narcotic Usage

At the time of TDR surgery, 64.7% of activL and 61.3% of Control patients were using narcotics (Figure 5). Narcotic use was significantly reduced at 1 through 5 years for both activL and Control patients compared with baseline (P < 0.001). At 5 years, the proportion of patients taking narcotics decreased to <2% for both activL and Control patients. Similar results were achieved with complete case analysis (see Table, Supplemental Digital Content 8, https://links.lww.com/BRS/B449).

Figure 5: Narcotic usage through 5 years post-treatment for activL versus Control artificial discs. At each follow-up time point post-op, a significantly lower proportion of patients with activL and Control TDR devices used narcotics compared with baseline (P < 0.001 at each time point). At 12 months follow-up, there was a trend for fewer activL patients using narcotics compared with control (P = 0.07). Multiple imputation was used for missing patient data.

ProDisc-L Subanalyses

When ProDisc-L data were separated from the Control group, the activL Artificial Disc maintained noninferiority to ProDisc-L for the primary composite endpoint with the LOCF and missing = success imputation analyses (Figure 6). At the 5-year endpoint, radiographic findings were significantly better for activL versus ProDisc-L patients across several parameters, including disc height (P = 0.0025), disc angle (P = 0.0022), flexion-extension rotation (P = 0.018), and flexion-extension translation (P = 0.033) (Table 3). These results were mostly replicated by complete case analysis, except there was no significant difference in flexion-extension translation between activL and ProDisc-L patients (see Table, Supplemental Digital Content 6, https://links.lww.com/BRS/B449). Freedom from serious AEs through 5 years was also significantly higher for activL patients (64%) compared with ProDisc-L patients (49%), P = 0.038 (Figure 7); however, there were no significant differences between activL and ProDisc-L patients for device-related serious AEs through 5 years. Risk of reoperation was similar for activL versus ProDisc-L patients, with very high proportions of patients with freedom from lumbar reoperation (95% vs. 89%, P = 0.09), freedom from index-level reoperation (95% vs. 95%, P = 0.89), and freedom from adjacent-level reoperation (99% vs. 94%, P = 0.01) through 5 years.

Figure 6: Primary and sensitivity analyses for composite endpoint at 5 years for activL versus ProDisc-L artificial discs. Noninferiority with the activL Artificial Disc is demonstrated if the entire 95% confidence interval is greater than the prespecified noninferiority margin. Superiority with the activL Artificial Disc is demonstrated if the entire 95% confidence interval is greater than 0.

Figure 7: Freedom from serious adverse events through 5 years for activL versus ProDisc-L artificial discs. Kaplan-Meier estimate is 64% with activL and 49% with ProDisc-L discs. Log-rank P = 0.038.

DISCUSSION

The results of this trial demonstrate that lumbar TDR is safe and effective for the treatment of single-level DDD through 5 years follow-up, with a remarkably high freedom from reoperation. The activL TDR was shown to be safer and more efficacious for select radiographic outcomes versus Control TDRs and when ProDisc-L outcomes were separated out from the Control.

Few randomized trials have directly compared the performance of different lumbar TDRs. Guyer et al¹⁴ reported outcomes from a randomized trial comparing the Kineflex-L Disc and the Charité artificial disc. Through 2 years follow-up, safety and effectiveness outcomes were comparable with a 9% reoperation rate. A network meta-analysis of 2-year data showed activL TDR had the highest probability of being the best treatment across ODI success, back pain, patient satisfaction, and employment status, compared with other TDRs, including ProDisc-L, Charité, Maverick, and Kineflex.¹⁵ In the present 5-year study, we identified no statistically significant difference in the primary composite endpoint at 5 years between activL and Control artificial discs or when ProDisc-L was separated out, but individual efficacy and safety components of the composite endpoint trended in favor of the activL Artificial Disc. Furthermore, ROM parameters were significantly better at 5 years for activL patients compared with the Control group (and ProDisc-L alone), which may reflect the differences in device design that relate to ROM and device insertion. Patients with a greater improvement in degrees of ROM at the TDR level were less likely to have radiographic adjacent level degeneration, as described in a post-hoc analysis of this 5 year activL IDE study.⁸ In the present study, improvement in flexion-extension rotation at 5 years post-TDR surgery was found to significantly correlate with decreased VAS and ODI scores. Both activL and Control maintained significant improvement in pain and function compared with baseline.

Through 5 years, activL patients had a significantly lower risk of serious AEs compared with the Control group and compared with the ProDisc-L patients alone. A significantly lower proportion of serious back pain events in activL versus Control patients contributed to this reduced risk. As the activL TDR is available with lower overall construct heights and has the ability for semi-constrained translational movement, both of which are attributes not available with other TDRs, the TDR may better mimic the anatomy and function of a healthy intervertebral disc with less resulting pain.

Another major finding from this study was that reoperation rates compared favorably to previous studies. In the FDA-IDE study of the Charité TDR,¹ index-level reoperations through 5 years were performed in 8% of patients treated with TDR and 16% of patients who underwent fusion. Similarly, in the FDA-IDE study of ProDisc-L,³ index-level reoperation rates were 8% and 12%, respectively. The reoperation rates in these lumbar fusion groups are comparable to the 18% rate through 5 years reported in population-based studies.¹⁶ Index-level reoperation rates at 5 years were 6% for TDR and 8% for lumbar fusion in a randomized trial utilizing several different TDRs.¹⁷ Furthermore, recent meta-analyses including IDE trial data demonstrated that reoperation rates at 5 years are significantly lower with lumbar TDR compared with fusion.^5,6 In the current study, index-level reoperations were performed in ∼5% of patients. On the basis of the collective results of these studies, it can reasonably be concluded that there are no important differences in index-level reoperation rates among TDRs, and these rates are improving with next-generation TDRs and experience.

While the choice between lumbar fusion or TDR remains controversial, and may be highly dependent on patient characteristics, TDR with the activL device is associated with a lower risk of serious complications relative to other commercially available TDRs and may represent an attractive motion-preserving treatment option in appropriate patients.

In this study, patients in the activL and Control TDR groups showed a reduction in narcotic use, from 61% to 64% at baseline to approximately 2% at 5 years. This is in contrast to the study by Zigler et al,³ in which 76% to 84% of patients used narcotics before surgery, while 40% of fusion patients and 38% of TDR patients were using narcotics at the 5-year follow-up visit. Differences in populations and protocols may explain the differences between these studies. For our study population, it may be inferred that after TDR surgery, patients have a lower opioid dependence, have a better chance to participate in the work force, and have a lower opioid medication cost burden. Some data suggest that narcotic use may still be high after lumbar fusion^18–20; however, long-term meta-analyses are not available that compare narcotic use between lumbar TDR and fusion.

Short- and long-term evidence shows the clinical benefit of lumbar TDR compared with conservative nonsurgical care. A recent network meta-analysis of 2-year data demonstrated activL TDR has the highest probability of being the best treatment for DDD.¹⁵ When compared with conservative care, TDR shows greater improvements in disability, pain, and quality of life.¹⁵ In a multicenter, randomized controlled trial with 8-year follow up, Furunes et al²¹ reported a difference in ODI of 6.1 points that favored TDR to multidisciplinary rehabilitation (P = 0.02), as well as a significantly greater proportion of patients who reported a full recovery following surgery compared with rehabilitation (18% vs. 6%, P = 0.002). At 8 years, more patients in the TDR group had clinically important improvement from baseline compared with the rehabilitation group (70% vs. 50%, respectively, P = 0.03).²¹ This study demonstrated significant long-term improvement after both rehabilitation and TDR, and statistically significant long-term results in favor of TDR compared with rehabilitation for functional improvement and pain relief.²¹

The strengths of this study included a randomized design, use of validated outcome measures, multiple imputation methods, and long-term follow-up. This study also had several limitations. First, although this study reports comparative outcomes among lumbar TDRs, direct inference to alternative treatments such as lumbar fusion or structured rehabilitation cannot be made. A 2-year network meta-analysis demonstrated favorable findings for the activL Artificial Disc relative to fusion and conservative care¹⁵; however, these analyses will need to be updated to reflect 5-year outcomes. A 1-year retrospective study demonstrated that DDD patients receiving the activL Artificial Disc present quicker return to work, less back pain, and lower disability scores compared with anterior lumbar fusion patients.²² Second, the Control group consisted of patients who received ProDisc-L or Charité at the discretion of the implanting physician. Third, while the use of rigorous inclusion criteria may be viewed as a strength in terms of mitigating potential confounding variables, the ability to generalize these findings to the population is unknown. Real-world, long-term data are increasingly available for lumbar TDR.^23–28 For example, a Swiss nation-wide registry demonstrated that lumbar TDR (including the activL Artificial Disc) remained effective over 5 years for pain relief, reduction in pain medication, and improvement in quality of life.²⁸ After the first year of a 10-year postmarket surveillance study, surgeons were satisfied with the overall performance of the activL Artificial Disc and did not report any safety concerns.²⁹ Additional long-term results from real-world activL usage would be a useful adjunct to these clinical trial findings to optimize generalizability. Last, as with any device trial, unblinding is always a potential concern. Aside from maintenance of single blind status being an FDA requirement, several factors helped to mitigate this risk (e.g., high efforts to prevent knowledge of disc type, similar disc shapes).

CONCLUSION

This study demonstrates that the activL Artificial Disc is safe and effective for the treatment of symptomatic lumbar DDD through 5 years. The activL TDR offers significantly better results for several ROM parameters, as well as a lower risk of serious AEs, compared with the Control TDR group and ProDisc-L Control patients alone. The likelihood for reoperation was also low for TDR patients through 5 years. This study is aligned with the body of evidence that supports TDR as a safe and effective alternative to fusion for lumbar DDD over 5 years, as well as a network meta-analysis that supports the activL Artificial Disc as the best treatment among other TDR devices, fusion, and conservative care.^{1–3,5,6,8,15}

Key Points

• A total of 261 patients with single-level lumber DDD unresponsive to at least 6 months of nonsurgical management provided 5-year follow up data after randomization to activL (n = 176) or Control (n = 85) TDR.
• The primary composite endpoint at 5 years was statistically noninferior for the activL Artificial Disc compared with Control.
• Patients implanted with the activL Artificial Disc had significantly better range of motion for flexion-extension rotation, flexion-extension translation, and disc angle, compared with Control.
• Through 5 years, freedom from a serious adverse event was significantly higher for activL patients (64%) versus Control patients (47%, log-rank P = 0.0068), and versus ProDisc-L patients alone (49%, log-rank P = 0.038).
• The study demonstrated a remarkably high freedom from reoperation for TDR.