Hugh S. Markus, BM Bch, University of Cambridge, Cambridge, United Kingdom, presented the first results of the Cervical Arterial Dissection in Stroke Study (CADISS) trial [ISRCTN44555237]. The CADISS trial demonstrated that the recurrent stroke rate due to cervical artery dissection appears to be low, with no significant difference in outcome by using antiplatelet vs anticoagulant treatment for future stroke prevention. Investigators also determined that the number of patients needed to show a difference between treatment arms was high and could be challenging to obtain.

Carotid and vertebral dissection is an important cause of stroke in young and middle-aged patients, accounting for up to 25% of cases [Beletsky V et al. Stroke. 2003; Touzé E et al. Neurology. 2003]. It carries an associated risk of recurrent stroke, although the reported numbers vary widely, depending on the study reviewed. Physicians usually prescribe either anticoagulation or antiplatelet agents to prevent recurrent strokes; however, data from randomized controlled trials showing the superiority of one treatment over another are currently lacking, with some nonrandomized data providing inconclusive results [Kennedy F et al. Neurology. 2012].

CADISS was a randomized, open-label trial, with blinded adjudication of end points [Markus HS et al. Int J Stroke. 2007]. The primary aim was to determine whether antiplatelet vs anticoagulation treatment was more effective at preventing stroke in patients with acute cervical dissection. The secondary aim was to measure the frequency of recurrent stroke in this population. Initially, 250 patients were recruited in order to establish whether there would be a sufficient number of clinical end points to determine a treatment effect, and if an adequate number of patients can be feasibly recruited. The primary end point was ipsilateral stroke or death at 3 months. Eligible patients had to have symptoms of ipsilateral stroke or transient ischemic attack, or ipsilateral Horner syndrome, neck pain, or headache within 7 days of recruitment.

Out of 250 patients, 126 were randomized to antiplatelets and 124 to anticoagulants. End points included 4 ipsilateral strokes (all in patients presenting with stroke; 3 in antiplatelet group and 1 in anticoagulant group) and 1 subarachnoid hemorrhage (in anticoagulant group). No deaths were reported. The overall stroke rate was strikingly low at 1.6% (2.1% in those presenting with stroke or transient ischemic attack). Given the low number of events, the difference between treatment arms was not significant (Table 1).

A predefined per-protocol analysis was also performed after excluding patients whose diagnosis of dissection could not be confirmed by reviewing their imaging data. Overall, 53 patients were excluded. The per-protocol analysis included 197 patients. There were 3 per-protocol end point events (defined as stroke, major bleeding, or death) among 101 patients in the antiplatelet group and 2 among 96 patients in the anticoagulant group (Table 2), with insignificant difference between treatment arms.

The power calculation revealed that the sample size for a definitive study would have to be 9752 patients (power set to 0.8 and significance to 0.05). However, Prof Markus mentioned that although it would be very challenging to recruit such a large number of patients, this was a rather wide estimate because of the low number of end points.

In conclusion, Prof Markus highlighted that the recurrent stroke rate in the CADISS trial was much lower than that reported in some observational studies. He also cautioned that the diagnosis of dissection was not confirmed in 20% of cases, suggesting that diagnostic criteria may not always be properly applied in clinical practice.