The diagnostic accuracy of magnetic resonance angiography for blunt vertebral artery injury detection in trauma patients: A systematic review and meta-analysis

Abstract

The role of magnetic resonance angiography (MRA) in the evaluation of patients with blunt vertebral artery has not been fully established. Our aim is to define the diagnostic accuracy of MRA in comparison to digital subtraction angiography (DSA) for the detection of blunt vertebral artery injury in trauma patients. A computer-assisted literature search of the PubMed, Scopus, Highwire, Web of Science, and LILACS was conducted, in order to identify studies reporting on the sensitivity and specificity of MRA in comparison to DSA for the detection of blunt vertebral artery injury in trauma patients. The Database search retrieved 91 studies. Five studies fulfilled our eligibility criteria. Two authors assessed the risk of bias and applicability concerns using QUADAS-2. Two-by-two contingency tables were constructed on a per-vessel level. Heterogeneity was tested by the statistical significance of Cochran's Q, and was quantified by the Higgins's I² metric. The pooled estimates of sensitivity and specificity for blunt vertebral artery injury detection with MRA in comparison to DSA were calculated based on the bivariate model. The meta-analysis was supplemented by subgroup and sensitivity analysis, as well as analysis for publication bias. There was significant clinical heterogeneity in the targeted population, inclusion criteria, and MRA related parameters. The reporting bias and applicability concerns were moderate and low, respectively. In the overall analysis, the sensitivity ranged from 25% to 85%, while the specificity varied from 65% to 99%, across studies. According to the bivariate model, the pooled sensitivity and specificity of MRA in the evaluation of patients with blunt vertebral artery was as high as 55% (95% CI 32.1%-76.7%), and 91% (95% CI 66.3%-98.2%), respectively. Subgroup analysis in terms of MRA sequence sensitivity of phase, the contrasted MRA (75% [95% CI 43%-92%]) seemed to be superior to the TOF MRA (46% [95%CI 20%-74%]). The addition of contrast enhancement did not seem to improve the diagnostic yield of MRA. The Egger's test did not identify any significant publication bias (p=0.2). An important limitation of the current meta-analysis is the small number of eligible studies, as well as the lack of studies on newer, high-field MR scanners. We concluded that MRA has a moderate diagnostic accuracy in the diagnosis of blunt vertebral artery injuries. Further studies on high-field magnetic resonance scanners are recommended.

Keywords: Bivariate model; Blunt vertebral artery injury; Diagnostic odds ratio; Digital subtraction angiography; Magnetic resonance angiography.