Most patients who have had a stroke are given aspirin; however, aspirin-related cerebral haemorrhage is a complication that is currently of concern, particularly in China where there is a high incidence of cerebral haemorrhage in secondary prevention programmes and within the community. Cilostazol, a phosphodiesterase 3 (PDE3) inhibitor, is an alternative to aspirin that works through a different mechanism. This trial aimed to compare the efficacy and safety of cilostazol with that of aspirin for the long-term prevention of the recurrence of ischaemic stroke.

720 patients (mean age 60·2 years, SD 9·86) who had had an ischaemic stroke within the previous 1–6 months were enrolled consecutively in a prospective, multicentre, double-blind, randomised trial. 360 patients were randomly assigned to receive cilostazol and 360 patients to receive aspirin. Analysis was by intention to treat. Patients in both groups took the medication for 12–18 months. The primary endpoint was any recurrence of stroke (ischaemic stroke, haemorrhagic stroke, or subarachnoid haemorrhage) during the trial period. All patients had MRI with T1 MRI, T2 MRI, diffusion-weighted imaging (DWI), T2 fluid-attenuated inversion recovery (FLAIR), and T2 gradient echo imaging (T2*) at the beginning and the end of the study. This trial is registered with ClinicalTrials.gov, number NCT00202020.

The average duration of treatment was 740 person-years, and 719 patients were analysed (360 in the cilostazol group and 359 in the aspirin group). The primary endpoint was reported in 12 patients in the cilostazol group and in 20 patients in the aspirin group. The estimated hazard ratio, calculated with Kaplan–Meier curves (risk of primary endpoint in cilostazol group vs aspirin group), was 0·62 (95% CI 0·30–1·26; p=0·185). Symptomatic cerebral haemorrhage was reported in six patients: one in the cilostazol group and five in the aspirin group. Asymptomatic cerebral haematoma was found in four patients in the aspirin group and one patient in the cilostazol group. Brain bleeding events were significantly more common in the aspirin group than in the cilostazol group (7 vs 1, p=0·034). All of the six patients with symptomatic haemorrhage had previous cerebral microbleeds in the area where the haematoma was located.