Statistics from Altmetric.com
In a recent report, Auer et aldescribed the clinical and imaging findings in 19 cases of extracranial vertebral artery dissection retrospectively.1 We make the following comments.
Firstly, the authors described the “sensitivity” and “specificity” of digital subtraction angiography (DSA), magnetic resonance imaging/angiography (MRI/A), and duplex sonography for diagnosing extracranial vertebral artery dissection.1These figures were based on the percentage of probable and definite features among the 19 patients. Nevertheless, sensitivity of a test is the number of cases with true positive results divided by the total number of positive results (including both true and false positives), and specificity is the number of cases with true negative results over the sum of true and false negatives. The authors misquoted the terms “sensitivity” and “specificity” in their report, as the diagnostic criteria of the various tests have not been applied to a control group to disclose the false positive cases and true negative cases. Secondly, the criteria for case inclusion were not defined. Apparently, extracranial vertebral artery dissection was diagnosed by either radiological features on MRI/A (which may be “pathognomonic” or “suggestive”) in the appropriate clinical context or confirmatory radiological features on DSA (which may be “specific” or “indirect”). The accuracy and usefulness of DSA, MRI/A, and duplex sonography cannot be compared directly, as no single “gold standard” diagnostic method was used and because results of the present study simply reflected the proportion of cases diagnosed by the authors.
Dissection of neck arteries was thought to be an uncommon cause of ischaemic stroke. The true incidence of this condition remains unknown as angiography is not performed in every patient during the acute or subacute phase. Younger patients are more likely to undergo early angiography when there is a history of recent neck trauma2or pain, or when no other causes of stroke are apparent. This selection bias may underestimate the incidence of stroke due to arterial dissection in older patients and those without neck trauma or pain.
Auer and Felber reply:
Cheung et al state in their comment, that dissections of the neck arteries are an underestimated cause of stroke, because angiography is not performed in every patient during the acute and subacute phase. This bias is even more important in the case of vertebral artery dissection if the initial symptoms are non-specific. Non-invasive diagnostic methods are likely to be performed earlier in these patients and this was our motivation to report on the magnetic resonance angiography of vertebral artery dissection.
The diagnosis of vertebral artery dissection is often based on the consensus of clinical and neuroradiological features. We agree with Cheung et al that no single “gold standard” test for a dissection exists. Imaging precedures more often show indirect signs which have to be interpreted in the appropriate clinical context. Therefore, the “inclusion criterion” we used for this retrospective analysis was the clinical and neuroradiological consensus on the diagnosis of vertebral artery dissection.1-1 The sensitivity of DSA, ultrasound, and MRI/angiography was determined from the findings of the affected and the contralateral normal vertebral arteries, there were no false positive results. The term specificity could have been misleading, because it did not refer to the overall specificity of a test but to the frequency of findings that reached a level of specificity sufficient to establish the diagnosis of vertebral artery dissection.1-1
Further prospective studies on the sensitivity and specificity of magnetic resonance for the diagnosis of vertebral artery dissections are certainly necessary, but our retrospective evaluation already showed that MRI and MR angiography will have a major contribution in future. As a non-invasive means, magnetic resonance can be employed without risk in patients with non-specific symptoms and may provide specific findings that are not accessible with other methods. This will lead to a better estimation of the true incidence of dissections and will improve the early diagnosis and management of dissections in individual patients.