Meta-analysis of pre-clinical studies of early decompression in acute spinal cord injury: a battle of time and pressure

Abstract

Background: The use of early decompression in the management of acute spinal cord injury (SCI) remains contentious despite many pre-clinical studies demonstrating benefits and a small number of supportive clinical studies. Although the pre-clinical literature favours the concept of early decompression, translation is hindered by uncertainties regarding overall treatment efficacy and timing of decompression.

Methods: We performed meta-analysis to examine the pre-clinical literature on acute decompression of the injured spinal cord. Three databases were utilised; PubMed, ISI Web of Science and Embase. Our inclusion criteria consisted of (i) the reporting of efficacy of decompression at various time intervals (ii) number of animals and (iii) the mean outcome and variance in each group. Random effects meta-analysis was used and the impact of study design characteristics assessed with meta-regression.

Results: Overall, decompression improved behavioural outcome by 35.1% (95%CI 27.4-42.8; I(2)=94%, p<0.001). Measures to minimise bias were not routinely reported with blinding associated with a smaller but still significant benefit. Publication bias likely also contributed to an overestimation of efficacy. Meta-regression demonstrated a number of factors affecting outcome, notably compressive pressure and duration (adjusted r(2)=0.204, p<0.002), with increased pressure and longer durations of compression associated with smaller treatment effects. Plotting the compressive pressure against the duration of compression resulting in paraplegia in individual studies revealed a power law relationship; high compressive forces quickly resulted in paraplegia, while low compressive forces accompanying canal narrowing resulted in paresis over many hours.

Conclusion: These data suggest early decompression improves neurobehavioural deficits in animal models of SCI. Although much of the literature had limited internal validity, benefit was maintained across high quality studies. The close relationship of compressive pressure to the rate of development of severe neurological injury suggests that pressure local to the site of injury might be a useful parameter determining the urgency of decompression.

Figure 1. Flow diagram depicting the number of publications initially identified, number of records following removal of duplicates and exclusions, and the final number of publications included for analysis.

Image adapted from: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(6): e1000097. doi: 10.1371/journal. pmed1000097.

Figure 2. Effect size and 95% confidence intervals of the neurobehavioural assessments in the 79 experiments included in meta-analysis.

The reference line represents the overall effect size of 35.1% with the gray shaded bar the 95% confidence intervals (27.5-42.8) of the global estimate.

Figure 3. Meta-regression of functional (neurobehavioral) improvement versus compressive pressure (p=0.004).

The size of each point reflects the precision of each comparison.

Figure 4. Line graph demonstrating the relationship between the duration of compression producing severe neurological injury and the compressive pressure in studies included in the meta-analysis.

The association obeys a power law distribution (y = 743.17x^-0.443), evidenced by a linear relationship on a log-log plot of the variables (inset).

Figure 5. Line graph exploring the relationship between compressive duration and compressive pressure.

(A) The association between the duration of compression producing severe neurological injury and the compressive pressure in those studies in which there was an initial injury to the spinal cord followed by compression. The data demonstrates a close correlation and again obeys a power law relationship (y = 829.06x^-0.459) with a linear distribution on a log-log plot of the variables (upper inset). (B) Power law (y = 144.62x^-0.248) relationship between compressive pressure and duration in studies employing an initial injury to the spinal cord followed by narrowing of the spinal canal to induce compression. These models had lower estimated pressures and longer durations of compression were necessary to produce paraplegia.

Figure 6. The change in effect size with (A) Region of injury, (B) Method of compression (Clip = aneurysm clip), (C) Neurobehavioural score (NDS = Neurologic deficit score; Olby = Olby score; Tarlov = Tarlov scale; Multiple = ≥2 behavioural tests; Motor = Motor test; BBB = Basso Beattie Bresnahan scale).

The shaded gray bar represents the 95% confidence limits of the global estimate. The vertical error bars represent the 95% confidence intervals for the individual estimates. The width of each bar reflects the log of the number of animals contributing to that comparison. Each stratification accounts for a significant proportion of the heterogeneity observed between studies. (D) Meta-regression of functional neurobehavioural improvement versus the time of final assessment (p=0.046). The size of each point reflects the precision of each comparison.

Figure 7. Effect of reported study blinding on effect size.

The shaded gray bar represents the 95% confidence limits of the global estimate. The vertical error bars represent the 95% confidence intervals for the individual estimates. The width of each bar reflects the log of the number of animals contributing to that comparison.

Figure 8. Evidence of publication bias demonstrated by (A) Egger regression analysis of early decompression experiments.

The 95% confidence intervals of the regression line do not include the origin, suggesting the presence of a significant publication bias. (B) Funnel plot showing the data in black and the additional missing studies suggested by trim and fill in red. The red vertical line indicates the possible global estimate in the absence of publication bias.