Re-evaluation of categorial outcomes using common clinically relevant improvement thresholds following bilateral L2 medial branch restorative neurostimulation verses sham.

Abstract

When chronic low back pain (cLBP) persists after conservative management there are limited treatment options. Increasingly, researchers and clinicians are defining distinct phenotypes within the cLBP population to develop and provide more targeted and effective interventions. One such phenotype involves people with mechanical low back pain and impaired motor control identified by physical exam or MRI evidence of multifidus dysfunction. An implantable neurostimulation device that delivers an electrical stimulation to the L2 medial branch has been recently FDA-approved with initially encouraging preliminary results. The pivotal trial demonstrated statistically-significant benefits over sham treatment in all secondary outcomes but not the primary outcome at the 120-day primary endpoint. However, recently published 1- and 5-year studies have shown better outcomes compared to optimal medical management. It was our purpose to expand the comparison of clinical outcomes between this neurostimulation treatment and sham by re-evaluating the data from the aforementioned pivotal trial using common clinically-relevant thresholds of treatment success that were not included in the pivotal trial's uncommon composite definition of treatment success or in its analysis of secondary outcomes for ODI and EQ-5D that compared mean improvements instead of categorical responder rates. Post-hoc reanalysis of previously published data from a prospective, randomized, double-blind, multi-center sham-controlled clinical trial. From 26 centers, 204 participants age 46.53 ± 9.38 years (range 22-71) and 53.9% (110/204) female were recruited with mechanical cLBP despite physical therapy and medical management, and a positive prone instability test on physical exam. Participants completed several patient-reported-outcomes at baseline and at the primary endpoint of 120 days following activation of the implanted stimulator or sham, including: VAS, ODI, EQ-5D, and subject global impression of change (SGIC). Using a complete-case analysis we compared differences between the neurostimulation treatment group and the sham-control group. Demographics are expressed using descriptive statistics. Patient-reported outcomes are compared between groups with a categorical analysis, setting responder thresholds at commonly used benchmarks and established minimum clinically important change (MCIC) thresholds. Statistical significance is determined by a two-sided p-value from the Fisher's exact test with significance set at p<0.05. Potential conflicts of interest directly related to this manuscript include research funding to support MS (level e) and DL (level b). The sponsor funded the original sham-controlled trial and funded this reanalysis but had no role in the reanalysis design, its prespecified analysis plan, its statistical analysis, or manuscript preparation. The authors had full access to the data and final responsibility for the decision to submit. Three patients (2 active, 1 sham) were lost to follow-up leaving 201 participants in the complete-case analysis. From baseline to the 120-day primary endpoint, statistically-significant differences were observed in favor of active treatment at the MCIC thresholds of ODI: ≥15-point improvement (59% vs 40.6%; p=0.0111) and ≥30% improvement (65.0% vs 47.5%; p=0.0156); the MCIC of EQ-5D: ≥0.15 improvement (57.0% vs 36.0%; p=0.0045); and SGIC of ''better'' or more (54.0% vs 33.7%; p=0.0045). Pain outcomes were mixed; categorical responder rates did not differ significantly (MCIC ≥2-point VAS improvement and ≥50% improvement), although mean changes favored active treatment (-3.3 ± 2.7 vs -2.4 ± 2.9; p=0.0209). Comparing bilateral L2 medial branch restorative neurostimulation to sham at the primary endpoint of 120 days, employing common clinically-relevant thresholds for treatment success, this sponsor-funded reanalysis of the sham-controlled pivotal trial demonstrated statistically-significant and clinically meaningful benefits in favor of treatment in disability, quality of life and patient satisfaction, while changes in pain bore mixed results.