Study Question: Is magnetic resonance angiography (MRA) better than duplex ultrasonography in detecting significant carotid stenosis?

Setting: Various (meta-analysis)

Study Design: Meta-analysis (other)

Synopsis: This group reviewed more than 900 papers from PubMed (supplemented by searches of bibliographies to find an additional 62 studies) that compared the diagnostic performance of duplex ultrasonography and MRA with digital subtraction angiography (DSA). The authors state that they excluded studies published before 1994 because the technology has evolved considerably since that time, making comparison between new and old studies less meaningful. They limited their search to English-language literature and do not report searching for unpublished data, so important data may be missing. They also excluded studies with fewer than 15 patients and contacted the authors to clarify data and to assist in situations in which studies may have included the same or overlapping subjects.

Two authors independently extracted the data and resolved discrepancies by consensus. They did not report how many patients were included in the studies or the prevalence of the various degrees of stenosis. The investigators made two comparisons: patients with severe stenosis (70 to 99 percent) versus those with less than 70 percent stenosis, and patients with total occlusion (i.e., 100 percent) versus those with less than 100 percent occlusion.

For severe stenosis, MRA was 95 percent sensitive and 90 percent specific (positive likelihood ratio [LR+] = 9.5; negative likelihood ratio [LR−] = 0.06) compared with duplex ultrasonography, which was 86 percent sensitive and 87 percent specific (LR+ = 6.6; LR− = 0.2). For total occlusion, both tests were nearly perfect. MRA was 98 percent sensitive and 100 percent specific (LR+ at infinity; LR− = 0.02) and duplex ultrasonography was 96 percent sensitive and 100 percent specific (LR+ at infinity; LR− = 0.04). The confidence intervals on all values were narrow.

Heterogeneity is the degree of variability between studies; if there is too much, it is inappropriate to combine results across studies. Verification bias occurs when the results of the diagnostic test (such as duplex ultra-sonography) influence the likelihood that the reference standard test is performed and tends to inflate sensitivity and reduce specificity. For the outcome of total occlusion, the authors found no heterogeneity among the MRA studies; the duplex ultrasonography studies had some heterogeneity attributed to verification bias and type of equipment. For severe (70 to 99 percent) versus nonsevere (less than 70 percent) stenosis, the duplex ultrasonography studies had heterogeneity because of verification bias and cutoff for severe stenosis of 70 percent; the MRA studies demonstrated heterogeneity because of equipment.

Bottom Line: MRA, although more costly, is more effective than duplex ultrasonography in ruling in and ruling out severe or total carotid stenosis. Duplex ultrasonography is not a bad second choice, however, if you do not have access to MRA. Although not addressed in this paper, some researchers have argued that MRA should replace DSA as the gold standard. (Level of Evidence: 1a–)