Background and Purpose: Although antiplatelets are known to be effective for secondary prevention of cerebral infarction, the number needed to treat is rather large and the effects in stroke patients with complications such as hypertension or diabetes are inadequately defined. This study was conducted to examine the effect of such complications on recurrence of cerebral infarction, and to assess the effect of cilostazol, an antiplatelet agent, in these high-risk subjects.
Methods: A post hoc subgroup analysis of the already reported Cilostazol Stroke Prevention Study, which was a placebo-controlled double-blind trial, has been carried out to clarify the influence of various complications on recurrence in the placebo group and the effects of cilostazol in 1,095 patients with noncardioembolic ischemic cerebrovascular disease. Treatment continued for an average of 1.8 +/- 1.3 years (maximum 4.8 years).
Results: The recurrence rate of the diabetic stroke patients was significantly higher compared with the nondiabetics in the placebo group (9.4 vs. 4.7%/year, p = 0.01). Furthermore, our study showed that the relative risk reduction (RRR) for recurrence of infarction was 41.7% with cilostazol. This treatment provided a significant benefit in patients with lacunar infarction (RRR 43.4%, p = 0.04), with diabetes (RRR 64.4%, p = 0.008), or with hypertension (RRR 58.0%, p = 0.003).
Conclusions: Diabetic patients are particularly at risk for recurrence of cerebral infarction. Cilostazol is useful for the prevention of the recurrence of vascular events in patients with lacunar infarction, and is probably effective in high-risk patients with diabetes and/or hypertension.
A meta-analysis performed by the Antithrombotic Trialists' Collaboration  indicated that antiplatelet agents such as aspirin, ticlopidine, dipyridamole or clopidogrel are effective in preventing recurrence of vascular events in patients with stroke or transient ischemic attack. However, the calculated number needed to treat (NNT) was 28 for recurrence of vascular events during the 29-month observation period and 40 for recurrence of cerebral infarction over 3 years of observation . This means that 40 patients have to take these medications for more than 2 years in order to rescue 1 patient from recurrence of cerebral infarction. Moreover, the effects of antiplatelets on subtypes of nonembolic stroke, such as lacunar infarction, which may have a different pathogenesis from atherothrombotic infarction, remain an enigma. Numerous adverse reactions have been reported in association with the use of these antiplatelets, and new medications have been sought that would provide more effective action with fewer side effects.
The Cilostazol Stroke Prevention Study (CSPS) was a multicenter double-blind placebo-controlled, randomized clinical trial performed between 1992 and 1996 to study the long-term safety and effectiveness of the antiplatelet drug cilostazol [2, 3]. Cilostazol prevents platelet aggregation by increasing cyclic adenosine monophosphate (cAMP) levels in platelets via inhibition of cAMP phosphodiesterase, thus reducing the risk of recurrence of cerebral infarction . In addition, cilostazol is reported to have biological actions beyond antiplatelet activity, such as a vasodilatory action [4, 5], a vascular endothelial cell protection  and improvement of lipid metabolism . The use of aspirin for the prevention of recurrence of cerebral infarction was not approved by the Japanese Ministry of Health, Labour and Welfare at the time of the CSPS, which covered more than 1,000 CT- or MRI-proven ischemic stroke patients, so this trial was placebo controlled, which would no longer be possible for new studies, for ethical reasons. It was also the first prospective study classifying patients into subtypes of ischemic stroke, that is, atherothrombotic infarction, lacunar infarction and undetermined type.
Since we thought the CSPS data would be valuable for evaluating the efficacy of antiplatelet therapy for cerebral infarction, we reexamined the data and performed a post hoc subgroup analysis. The objectives of this report were to examine the effect of various complications on recurrence of cerebral infarction in the placebo group, and in particular, to assess the effect of cilostazol in patients at high risk.
The results of the general analysis of the CSPS have been reported elsewhere . In brief, the study was performed between 1992 and 1996, at 183 institutes throughout Japan, covering 1,095 patients with a prior history of symptomatic cerebral infarction (transient ischemic attack and cardioembolic infarction were excluded from this study) that had developed 1-6 months before randomization. Diagnosis was confirmed by clinical signs and symptoms, and CT or MRI in all subjects. All patients provided written informed consent and were randomized to cilostazol (547 patients) or placebo (548 patients). Cilostazol was administered as a 100-mg tablet twice daily and placebo tablets were matched for appearance and administration schedule. Treatment continued for an average of 1.8 +/- 1.3 years (maximum 4.8 years). Out of 1,095 patients, 1,067 were available for the intention-to-treat analysis.
For the efficacy analysis, a per protocol set of 526 patients in the cilostazol group (396 lacunar infarction, 73 atherothrombotic infarction, 57 undetermined type), and 526 patients in the placebo group (389, 68, and 69 patients, respectively) were available for subgroup analysis.
The criteria for hypertension were systolic blood pressure >=140 mm Hg and/or diastolic blood pressure >=90 mm Hg, or current treatment with antihypertensives. The diabetes mellitus group contained subjects who were diagnosed as diabetic by the investigators, based on standard criteria, or who were already receiving antidiabetics. Hypercholesterolemia was defined as total serum cholesterol >=220 mg/dl (5.67 mmol/l) or current treatment with cholesterol-lowering drugs. Prior ischemic heart disease was defined in terms of a history of angina pectoris or myocardial infarction.
Statistical results were calculated as follows. Annual recurrence rate was computed using the person-year method. Relative risk reduction (RRR) was determined from the incidence calculated by the person-year method. The standard error and 95% confidence intervals of these measures were estimated assuming an exponential distribution for the incidence of cerebral infarction. The log-rank test was used to compare the incidence of cerebral infarction in the two groups. Moreover, the heterogeneity of effects (RRR) among subgroups was investigated and statistically analyzed combining the results of the log-rank test for each subgroup. To investigate the effects of background factor and/or subgroup on the recurrence of cerebral infarction in the cilostazol group, univariate and multivariate analyses were performed using the Cox proportional hazards model, and the results were reported as hazard ratios (HRs). In the multivariate analysis, variables were selected by the backward elimination method.
Kaplan-Meier estimates were used for the calculation of NNT. Statistical tests were two-tailed, and a probability level of p < 0.05 was considered to indicate statistical significance, except for the investigation of interactions, where the level of significance was p < 0.20. Statistical analysis was performed using SAS version 8.02 software (SAS Institute Japan, Tokyo).
The annual recurrence rate of cerebral infarction and the number of patients available for analysis were determined in subgroups of patients having various complications in the placebo-treated group, and the results are shown in figure 1a. The difference in recurrence between the diabetes group and the nondiabetes group was statistically significant (9.4 vs. 4.7%/year, p = 0.01) (fig. 1a).
Fig. 1. Annual recurrence rate of cerebral infarction in placebo-treated patients with or without risk factors (a) and with or without other drug treatments for various underlying diseases (b). Bars show percent of patients with recurrence of cerebral infarction, with the number of patients in parentheses.
We next evaluated the annual recurrence rate of cerebral infarction in subgroups of patients in the placebo-treated group receiving other drug treatments for various underlying diseases; these results are shown in figure 1b. In patients being treated with so-called cerebral enhancers, which were widely used at the time in Japan and act by improving cerebral blood flow and metabolism, or vasodilators including ibudilast, ifenprodil tartrate and nicergoline, which are known to have a mild antiplatelet effect, the incidence of recurrence was significantly less than in patients not receiving such treatment (p = 0.04) (fig. 1b). Administration of statins, ACE inhibitors or Ca antagonists was also associated with reduced (albeit nonsignificant) recurrence rates, although the patients not receiving these drugs were mostly nonhyperlipidemic and nonhypertensive. However, diabetic patients who were receiving antidiabetic drugs showed a higher recurrence rate of cerebral infarction than patients not taking these drugs, who were mostly nondiabetic (fig. 1b).
In figure 2, we show the results of both univariate and multivariate analyses in the placebo group. The effects of diabetes (HR 2.05, 95% CI 1.19-3.53) and use of cerebral enhancers or vasodilators (HR 0.57, 95% CI 0.34-0.96) were significant in the multivariate analysis. The incidence of events was 9.0%/year in patients with diabetes but without hypertension, while 9.5%/year in patients with both diabetes and hypertension, which suggests that both diabetes and hypertension independently affect incidence of events.
Fig. 2. Factors influencing recurrence of cerebral infarction in placebo-treated patients, showing HR and 95% confidence interval obtained by univariate and multivariate analyses (Cox proportional hazards model).
As previously reported , the CSPS analysis showed an annual rate of cerebral infarction recurrence of 5.78% in the placebo group and 3.37% in the cilostazol group (RRR 41.7%, p = 0.02) (fig. 3a). The calculated NNT was 18.7 over 3 years of observation. Although the heterogeneity of effects (RRR) among subtypes of cerebral infarction did not reach significant difference, subgroup analysis with stratification by subtypes showed a recurrence rate of 5.25% in the placebo group and 2.97% in the cilostazol group for lacunar infarction (RRR 43.4%, p = 0.04) (fig. 3b). CSPS is the first evidence showing the effect of antiplatelet agents on lacunar infarction using a prospective design. The RRR with cilostazol treatment was 39.8% for atherothrombotic infarction and 44.5% for infarction of undetermined type.
Fig. 3. Subgroup analysis of stroke recurrence stratified by clinical disease type. a Annual recurrence rate of all subjects. b Annual recurrence rate stratified by clinical subtype. Figures in parentheses indicate number of patients.
A significant difference in the heterogeneity of effects (RRR) was seen only in subgroups with diabetes and hypertension. RRR determined in the cilostazol group showed statistically significant differences in the subgroups of patients with diabetes (64.4%, 95% CI 22.3-83.7) and hypertension (58.0%, 95% CI 23.8-76.9), as well as in the subgroup with no findings of ischemic heart disease (46.2%, 95% CI 13.2-66.7) and in the subjects without hypercholesterolemia (41.5%, 95% CI 0.6-65.6) (fig. 4). Therefore, it is concluded that cilostazol is effective for the prevention of recurrence of cerebral infarction, particularly in stroke patients with diabetes or hypertension. Cilostazol treatment was nearly always associated with positive values for RRR, both in groups of patients receiving concomitant treatment for various types of underlying disease and in groups not receiving such treatment.
Fig. 4. RRR effects of cilostazol with 95% confidence intervals. The figures in parentheses indicate the number of patients treated with placebo (n) and the number of patients treated with cilostazol (m). a Heterogeneity of effects (RRR) among factors: significant.
As shown in figure 5a, cilostazol administration was associated with a significant (RRR 64.4%, p = 0.008) reduction in cerebral infarction recurrence among diabetic patients, and indeed the administration of cilostazol reduced the recurrence rate of ischemic stroke in diabetic patients to the same level as in nondiabetic subjects.
Fig. 5. Annual recurrence rate of cerebral infarction in patients with diabetes and hypertension treated with or without cilostazol. a Effects in the presence/absence of diabetes. b Effects in the presence/absence of hypertension. Figures in parentheses indicate number of patients.
Cilostazol treatment was also associated with a significant reduction in cerebral infarction recurrence in ischemic stroke patients with hypertension (RRR 58.0%, p = 0.003) (fig. 5b). In nonhypertensive patients, the RRR was only 5.2%.
The incidence of bleeding by cilostazol was not significantly different from that of placebo.
When we studied recurrence of cerebral infarction in the placebo-treated patients stratified according to underlying complications, we found a higher incidence of recurrence for hypertension in comparison with patients lacking this risk factor, although the difference was not statistically significant. The lack of significance may be explained by the fact that antihypertensive treatment itself is effective in reducing the recurrence of stroke [8,9,10] (fig. 1b).
On the other hand, cerebral infarction patients with diabetes in the placebo group were at significantly greater risk for recurrence than those without diabetes. Earlier studies have shown that diabetic patients have a two- to fourfold greater risk of first-ever cerebral infarction than nondiabetic patients [11,12,13,14], and the present report also demonstrates that diabetes increases the risk of cerebral infarction recurrence, which is consistent with previous results [15,16,17].
However, findings from the UK Prospective Diabetes Study showed no reduction in the incidence of cerebral infarction in patients with type 2 diabetes, even when drug therapy reduced HbA1c by 0.9% for at least 10 years . Another study found no association between glucose control and risk of recurrence of stroke among diabetic patients with a history of stroke . This suggests that blood glucose control alone may be insufficient for decreasing the recurrence of cerebral infarction, which is consistent with our results (fig. 1b).
The results of this subgroup analysis of the CSPS also showed that cilostazol significantly decreased the recurrence of stroke in patients with lacunar infarction. Although the heterogeneity of effects (RRR) among the subtypes did not reach significant difference, the RRR for inhibition of annual recurrence following lacunar infarction was 43.4% (p = 0.04). CSPS is the first evidence of the efficacy of antiplatelet drugs in lacunar infarction, whose findings are consistent with the results from a later reported ESPS-2 post hoc analysis . On the other hand, the effect of cilostazol on atherothrombotic stroke was not statistically significant because of the small number of patients available for analysis in the relevant subgroups, although the RRR was 39.8%.
Cilostazol acts as an antiplatelet by specially inhibiting phosphodiesterase III, blocking cAMP degradation and raising cAMP levels within the cytoplasm [2, 21,22]. In addition, a vasodilatory action due to vascular relaxation has been reported [4, 5], as well as a protective effect on vascular endothelial cells , and beneficial effects on lipid metabolism . Such actions would be expected to contribute to a decrease in cerebral infarction recurrence [23,24,25].
Although the etiology of lacunar infarction is still poorly understood, it appears to be elicited by thrombus formation, encouraged by abnormal lipid metabolism and hypertension-induced damage to endothelial cells. Also, flow cytometry using anti-CD62P antibodies and PAC-1 to quantify platelet activation has shown that patients with lacunar infarction, as well as atherothrombotic infarction, had a significantly higher activated platelet count than age-matched MRI-proven control subjects . Although cilostazol is positioned as an antiplatelet drug, it has pleiotropic effects, including vasodilating action, vascular endothelial cell protection and improvement of the lipid pattern, and this may be the reason for the efficacy of cilostazol in reducing the recurrence of lacunar infarction.
Aspirin is currently recommended as a grade A treatment in the American Diabetes Association guidelines for management of diabetes . However, a meta-analysis by the Antithrombotic Trialists' Collaboration  showed no significant inhibition of recurrence of vascular events as a result of antiplatelet therapy in diabetic stroke patients. Data on cardiovascular disease in diabetic patients have also been reported from a multicenter, centrally randomized, open-label clinical trial study (the Primary Prevention Project)[28, 29]. That study assessed the primary preventative effects of low-dose aspirin in subjects with no history of cardiovascular disease, but having at least one risk factor, such as hypertension, diabetes, or hyperlipidemia. During the mean follow-up of 3.6 years, aspirin use significantly inhibited cardiovascular events (RRR = 23%). When a subgroup analysis was performed (diabetic vs. nondiabetic), the incidence of the primary endpoint in the diabetic group was 3.9% among patients taking aspirin and 4.3% in the control group, which was a nonsignificant difference; in the nondiabetic group, aspirin use significantly inhibited cardiovascular events (RRR = 41%). It was concluded that the use of aspirin did not significantly inhibit cardiovascular events in diabetic patients, and that mortality from cardiovascular disease actually tended to be higher in the aspirin-treated group . Moreover, no data were presented on the prevention of recurrence of stroke in this paper.
Our data showed firstly that the recurrence rate of cerebral infarction was higher in diabetic stroke patients receiving antidiabetic drugs (fig. 1b). This may be attributable to the fact that such patients usually have severer diabetes than nontreated diabetic patients. However, our study also showed that cilostazol was effective for the secondary prevention of cerebral infarction in patients with diabetes (fig. 5a). The reason why a statistically significant reduction was found is mainly because the incidence of recurrence is much higher in diabetic patients and the administration of cilostazol reduced the recurrence rates in diabetic patients to as low a level as in nondiabetic patients (fig. 5a).
Cilostazol administration was also associated with a significant reduction in cerebral infarction recurrence in hypertensive patients (fig. 5b). This result also reflects the higher incidence of recurrence in the high-risk group. Reports indicate that cilostazol has a circulation-improving action, as well as an antiplatelet action, and this may be the reason for the lower NNT of cilostazol (18.7 over 3 years, recalculated in the present study from previously reported data ) for recurrence of cerebral infarction, compared with other antiplatelets.
In summary, our findings indicate that cilostazol can be expected to be particularly effective in lacunar infarction and, although this is a post hoc analysis, probably effective in patients with diabetes or hypertension, who are at high risk of ischemic stroke recurrence. These findings need to be supported by further research comparing different antiplatelet regimens.
Antiplatelet; Diabetes; Hypertension; Lacunar infarction; Stroke prevention