Effect of device constraint: a comparative network meta-analysis of ACDF and cervical disc arthroplasty.

Background Context: Clinical trials have demonstrated that cervical disc arthroplasty (CDA) is an effective and safe alternative treatment to anterior cervical discectomy and fusion (ACDF) for cervical degenerative disc disease in the appropriately indicated patient population. Various devices for CDA exist, differing in the level of device constraint.

Purpose: To investigate outcomes following Anterior Cervical Discectomy and Fusion (ACDF) versus CDA stratified based on the level of device constraint: Constrained, Semiconstrained, and Unconstrained.

Study Design: Systematic review and network meta-analysis.

Patient Sample: A total of 2,932 CDA patients (979 Constrained, 1,214 Semiconstrained, 739 Unconstrained) and 2,601 ACDF patients from 41 studies that compared outcomes of patients undergoing CDA or ACDF at a single level at a minimum of 2 years follow-up.

Outcome Measures: Outcomes of interest included the development of adjacent segment degeneration (ASD), index and adjacent segment reoperation rates, range of motion (ROM), high-grade heterotopic ossification (HO, McAfee Grades 3/4), and patient-reported outcomes (NDI/VAS).

Methods: CDA devices were grouped based on the degrees of freedom (DoF) allowed by the device, as either Constrained (3 DoF), Semiconstrained (4 or 5 DoF), or Unconstrained (6 DoF). A random effects network meta-analysis was conducted using standardized mean differences (SMD) and log relative risk (RR) were used to analyze continuous and categorical data, respectively.

Results: Semiconstrained (p=.03) and Unconstrained CDA (p=.01) demonstrated a significantly lower risk for ASD than ACDF. All levels of CDA constraint demonstrated a significantly lower risk for subsequent adjacent segment surgery than ACDF (p<.001). Semiconstrained CDA also demonstrated a significantly lower risk for index level reoperation than both ACDF and Constrained CDA (p<.001). Unconstrained devices retained significantly greater ROM than both Constrained and Semiconstrained CDA (p<.001). As expected, all levels of device constraint retained significantly greater ROM than ACDF (p<.001). Constrained and Unconstrained devices both demonstrated significantly lower levels of disability on NDI than ACDF (p=.02). All levels of device constraint demonstrated significantly less neck pain than ACDF (p<.05), while Unconstrained CDA had significantly less arm pain than ACDF (p=.02) at final follow-up greater than 2 years.

Conclusion: CDA, particularly the unconstrained and semiconstrained designs, appears to be more effective than ACDF in reducing the risk of adjacent segment degeneration and the need for further surgeries, while also allowing for greater range of motion and better patient-reported outcomes. Less constrained CDA conferred a lower risk for index level reoperation, while also retaining more range of motion than more constrained devices.