You are on page 1of 10

Research Center, University of British Columbia, S169-2211 Wesbrook Mall, Vancouver, BC V6T 2B5,

Canada, or at oscar.benavente@

*The SPS3 trial investigators are listed in the Supplementary Appendix, available at

N Engl J Med 2012;367:817-25.



Effects of Clopidogrel Added to Aspirin

in Patients with Recent Lacunar Stroke
The SPS3 Investigators*


Lacunar infarcts are a frequent type of stroke caused mainly by cerebral small- The members of the writing group (Oscar
R. Benavente, University of British
vessel disease. The effectiveness of antiplatelet therapy for secondary prevention Columbia, Vancouver, Canada; Robert G.
has not been defined. Hart, Population Health Research Insti-
tute, Hamilton, ON, Canada; Leslie A.
McClure and Jeffrey M. Szychowski, Uni-
METHODS versity of Alabama at Birmingham, Bir-
We conducted a double-blind, multicenter trial involving 3020 patients with recent mingham; Christopher S. Coffey, Univer-
sity of Iowa, Iowa City; and Lesly A.
symptomatic lacunar infarcts identified by magnetic resonance imaging. Patients Pearce, Minot, ND) of the Secondary
were randomly assigned to receive 75 mg of clopidogrel or placebo daily; patients Prevention of Small Subcortical Strokes
(SPS3) trial assume responsibility for the
in both groups received 325 mg of aspirin daily. The primary outcome was any re- overall content and integrity of the article.
current stroke, including ischemic stroke and intracranial hemorrhage. Address reprint requests to Dr.
Benavente at the Division of Neurology,
Department of Medicine, Brain

The participants had a mean age of 63 years, and 63% were men. After a mean follow- Copyright 2012 Massachusetts Medical Society.

up of 3.4 years, the risk of recurrent stroke was not significantly reduced with aspirin
and clopidogrel (dual antiplatelet therapy) (125 strokes; rate, 2.5% per year) as
compared with aspirin alone (138 strokes, 2.7% per year) (hazard ratio, 0.92; 95%
confidence interval [CI], 0.72 to 1.16), nor was the risk of recurrent ischemic stroke
(hazard ratio, 0.82; 95% CI, 0.63 to 1.09) or disabling or fatal stroke (hazard ratio, 1.06;
95% CI, 0.69 to 1.64). The risk of major hemorrhage was almost doubled with dual
antiplatelet therapy (105 hemorrhages, 2.1% per year) as compared with aspirin alone
(56, 1.1% per year) (hazard ratio, 1.97; 95% CI, 1.41 to 2.71; P<0.001). Among
classifiable recurrent ischemic strokes, 71% (133 of 187) were lacunar strokes. All-
cause mortality was increased among patients assigned to receive dual antiplatelet
therapy (77 deaths in the group receiving aspirin alone vs. 113 in the group receiv-ing
dual antiplatelet therapy) (hazard ratio, 1.52; 95% CI, 1.14 to 2.04; P = 0.004); this
difference was not accounted for by fatal hemorrhages (9 in the group receiving dual
antiplatelet therapy vs. 4 in the group receiving aspirin alone).

Among patients with recent lacunar strokes, the addition of clopidogrel to aspirin did
not significantly reduce the risk of recurrent stroke and did significantly increase the
risk of bleeding and death. (Funded by the National Institute of Neurological Dis-orders
and Stroke and others; SPS3 number, NCT00059306.)
N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012 817
The New England Journal of Medicine
Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.


commonly known as lacunar strokes, con-stitute about
25% of ischemic strokes1-3 and are particularly frequent
and practitioners aware of the group assign-ments).
Randomized assignments, stratified ac-cording to
among Hispanics.4-9 Al- clinical center and baseline hyperten-sive status,
though lacunar strokes occasionally result from
mechanisms of brain ischemia such as cardiogen-ic were generated with the use of a permuted-block
embolism or carotid-artery stenosis, most result design (with a variable block size) and protected
from intrinsic disease of the small penetrating ar- from previewing. All partici-pants were given 325
teries. This underlying disorder is the most fre- mg of enteric-coated aspirin daily and were
quent cause of covert brain infarcts and vascular randomly assigned to receive 75 mg of clopidogrel
10 -12 daily or a matching placebo, with adherence
cognitive impairment. To our knowledge, the
measured by means of pill counts performed at
secondary prevention of lacunar stroke as detected
quarterly follow-up visits.
by the use of magnetic resonance imaging (MRI)
The study was conducted and reported in ac-
has not been the focus of a randomized trial.
cordance with the protocol and statistical analysis
Aspirin is accepted as standard antiplatelet
13 plan, which are available with the full text of this
therapy in patients with lacunar infarcts. The article at The trial was designed and
addition of clopidogrel to aspirin has been shown executed by the SPS3 investigators. The mem-bers
to reduce the risk of stroke among pa-tients with
14 of the writing group vouch for the data and wrote
atrial fibrillation and those with acute coronary this report without professional editorial assistance.
syndromes, but dual antiplate-let therapy has Participation required written informed consent and
16 approval by the human research sub-jects
been associated with increased bleeding.
The Secondary Prevention of Small Subcortical committee at each study center.
Strokes (SPS3) trial tested two randomized inter- SPS3 was an investigator-initiated trial funded
ventions, in a 2-by-2 factorial design, in patients by a cooperative agreement with the National In-
with recent symptomatic, MRI-confirmed lacu-nar stitute of Neurological Disorders and Stroke
stroke: clopidogrel and aspirin versus aspirin alone (NINDS). Clopidogrel and the matching placebo
and two target levels of systolic blood pressure. The were donated by Sanofi-Aventis and Bristol-Myers
antiplatelet component of the trial was terminated Squibb, but neither company had any involve-ment
at the recommendation of the data and safety in the design or execution of the trial or in the
monitoring committee because of lack of efficacy analysis or reporting of the data. There were no
combined with evidence of harm. The final results confidentiality agreements between the study
of the antiplatelet com-ponent of the trial are sponsor (NINDS) and investigators.
presented here.
ME THODS Patients were eligible for participation in the study
if they were 30 years of age or older, had
STUDY DESIGN undergone a symptomatic lacunar stroke within the
In brief, SPS3 was a randomized, multicenter preceding 180 days, and did not have surgi-cally
clinical trial conducted in 82 clinical centers in amenable ipsilateral carotid artery disease or major
North America, Latin America, and Spain. (De-tails risk factors for cardioembolic sources of stroke. To
of the rationale for the study, the design, and avoid a lowering of blood pressure after acute
characteristics of the participants have been stroke, randomization did not take place for at least
17,18 2 weeks after the qualifying stroke. Participants
described elsewhere. ) In accordance with the
with a clinical lacunar syndrome were required to
2-by-2 factorial design of the study, eligible pa-
meet MRI criteria that included a lesion measuring
tients underwent simultaneous randomization to the
2.0 cm or less in diameter on diffusion-weighted
antiplatelet intervention (in which both pa-tients
imaging that corresponded to a positive apparent-
and practitioners were unaware of group
diffusion-coefficient image or a lesion with a well-
assignments) and to one of the two groups de-fined
delineated area of focal hy-perintensity that was 2.0
by target levels for systolic blood pressure (<130
mm Hg vs. 130 to 149 mm Hg) (with patients cm or less in diameter on fluid-attenuated inversion
recovery imaging or T2-weighted imaging that
corresponded to the

818 N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012

The New England Journal of Medicine

Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

clinical syndrome. Patients with transient lacunar rent stroke took place 3 to 6 months after the
ischemic attacks were included only if there was initial stroke. Recurrent stroke was considered to
evidence of the attacks on diffusion-weighted MRI. be disabling if the modified Rankin score was 4
MRI scans were assessed by the investigators at or higher. In the 25% of participants for whom a
each study site and then submitted for inter- Rankin score obtained within the 3-to-6-month
pretation by a neuroradiologist located at the SPS3 interval after the initial stroke was not available,
Coordinating Center. Patients with MRI evidence the last available Rankin score after stroke re-
of a recent or remote cortical infarct, a large sub- currence was used. Strokes were counted as fatal
cortical infarct (measuring more than 1.5 cm in if death occurred within 30 days or if it occurred
diameter), or a history of intracerebral hemor-rhage after 30 days and was attributable to the stroke.
were excluded (but those with microbleed-ing were Secondary outcomes included acute myocardial
infarction and death, classified as having a vascu-
not). Additional exclusion criteria were disabling
lar, nonvascular, or unknown cause. The primary
stroke (defined by a modified Rankin score of 4 or
more on a scale of 0 to 6, with higher scores safety outcome was major extracranial hemor-
indicating more severe disability) and previous rhage, defined as serious or life-threatening bleed-
intracranial hemorrhage (with the exception of ing requiring transfusion of red cells or surgery or
resulting in permanent functional sequelae or death.
traumatic hemorrhage) or cortical ischemic stroke.
All reported efficacy and safety outcomes were
confirmed by a central adjudication commit-tee that
OUTCOMES was unaware of the treatment assignments and that
The primary hypothesis was that clopidogrel classified ischemic strokes according to the
add-ed to aspirin would be superior to aspirin presumed mechanism on the basis of available
alone in reducing the primary outcome of stroke diagnostic studies.
recur-rence (any ischemic stroke or intracranial
hemor-rhage, including subdural hematomas). STATISTICAL ANALYSIS
Prespeci-fied subgroup analyses assessed the The initial sample size of 2500 patients was
primary outcome according to ethnic group calcu-lated on the basis of an assumed average
(Hispanic vs. non-Hispanic white), use of aspirin follow-up of 3 years, an estimated 3-year rate of
at the time of the qualifying event (yes vs. no), recur-rent stroke of 21%, and a 25% relative
and status with respect to diabetes. Ethnic group reduction in the risk of stroke in the group
was self-report-ed; participants who identified receiving dual antiplatelet therapy (clopidogrel
themselves as Spanish or Latino were classified and aspirin), with a type I error of 0.05 and a
as Hispanic ac-cording to the criteria used in the type II error of 0.10. The sample size was
19 reestimated midway through the trial to assess
San Antonio Heart Study.
Patients taking aspirin (at any dose) at the power on the basis of the ob-served overall event
time of their qualifying event were included in rate at that time, with the re-sult that the sample
size was increased from 2500 to 3000 patients,
the subgroup analyses; patients taking aspirin in 20
combination with clopidogrel (45 patients), clo with a 1-year extension of follow-up.
pidogrel alone (35 patients), or other antiplatelet The main analyses were based on standard
agents, alone or in combination (34 patients), time-to-event methods, with each treatment
were not included in these analyses. Additional group assessed with the use of the log-rank test;
subgroup analyses, which were not prespecified, Cox proportional-hazards models were used to
were based on age, sex, and region; the results of calculate hazard ratios. The time to an event was
these analyses are reported here. calculated as the time to the first event in the
Ischemic stroke was clinically defined as a focal case of multiple events of the same type and for
neurologic deficit of sudden onset persisting for the composite end point of stroke, myocardial
more than 24 hours, and without evidence of infarction, or death from vascular causes. Data
hemorrhage on neuroimaging. Intracranial hem- for patients without events were censored at the
orrhages included those in intracerebral, subdural, time of termination of study participation or at
epidural, and subarachnoid locations as docu- death. The interaction between the antiplatelet
mented on neuroimaging. Assessment for recur- therapy and blood-pressure therapy was assessed

N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012 819

The New England Journal of Medicine
Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

subgroups. All analyses were based on the inten-

Table 1. Baseline Characteristics of the Participants.*
tion-to-treat principle.
Aspirin plus Aspirin plus The trial was monitored by an independent
Placebo Clopidogrel data and safety monitoring committee selected
Characteristic (N=1503) (N=1517)
by the study sponsor. Two planned interim
Mean age (yr) 63 63
analyses were performed after one third and two
Male sex (%) 64 62 thirds of the primary events had transpired, with
Race or ethnic group (%) the use of HaybittlePeto bounds and with
White 52 52 futility analyses based on conditional power. At
Hispanic 31 31 the recommendation of the data and safety
monitor-ing committee, the antiplatelet
Black 17 17
component of the trial was stopped by the
Region or country (%)
sponsor 10 months before the planned end date,
North America 65 65 after completion of the second planned interim
Latin America 23 23 analysis, because of futility with respect to the
Spain 12 12 primary outcome coupled with evidence of harm.
History of hypertension (%) 74 76 The component of the trial involving blood-
Mean blood pressure at screening visit 143/78 143/78
pressure targets is ongoing, and no significant
(mm Hg) interactions be-tween the two interventions have
Diabetes (%) 38 35 been found with regard to the primary outcome.
Ischemic heart disease (%) 11 10
Previous clinical stroke or transient ischemic 15 15
attack (%)
Current tobacco smoker (%) 21 20
Between 2003 and 2011, a total of 3020 patients
Qualifying event (%) were enrolled in the study: 1503 in the group
Ischemic stroke (%) 97 97 treated with aspirin plus placebo and 1517 in the
Transient ischemic attack (%) 3 3 group treated with aspirin plus clopidogrel. A to-
Type of lacunar syndrome (%) tal of 1960 of the participants (65%) were from
Pure motor hemiparesis 35 31
North America, 694 (23%) from Latin America,
and 366 (12%) from Spain. Participants had been
Pure sensory stroke 10 10
followed for a mean of 3.4 years (range, 0 to 8.2)
Sensorimotor stroke 30 32
at the time of termination of the antiplatelet
Other 25 27 component of the study in August 2011 (see Fig.
Use of aspirin at time of qualifying event (%) 28 28 S1 in the Supplementary Appendix, available at
Use of statin at any follow-up visit (%) 85 84 The mean (SD) age of the partici-
pants was 6311 years, and 63% were men; 75%
There were no significant differences between the treatment groups for any of the participants had a history of hypertension,
variable. Additional baseline characteristics have been reported elsewhere. 37% had diabetes, and 20% were current tobacco

Data on race or ethnic group are self-reported. Participants who reported smokers (Table 1). The median time from the
that they were Spanish or Latino were classified as Hispanic according to
19 date of the qualifying stroke to randomization
the crite-ria used in the San Antonio Heart Study.
was 62 days. Among all participants, the mean
systolic blood pressure was 14319 mm Hg at
to ensure that it was not significant before the study entry and declined to 13116 mm Hg by
effect of the antiplatelet therapy was assessed. the time of the last follow-up visit.
For each model, we assessed the interaction be- During active participation, the estimated aver-
tween study treatment and study time, which was age rate of adherence to the assigned antiplatelet
defined as a continuous variable to test whether regimen was 94%. Permanent discontinuation of
the proportional-hazards assumption was assigned antiplatelet therapy occurred in 30% of the
violated. Interactions between planned co- patients in the group receiving dual antiplatelet
variates and antiplatelet assignment were evalu- therapy and in 27% of the patients receiving aspi-
ated with the use of Cox models to determine rin alone (P=0.02). Among participants who did
whether the treatment effect differed in specific

820 N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012

The New England Journal of Medicine

Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

Table 2. Primary Efficacy Outcomes.*

Aspirin plus Placebo Aspirin plus Clopidogrel Hazard Ratio

Outcome (N=1503) (N=1517) (95% CI) P Value

no. rate (%/yr) no. rate (%/yr)

All strokes (ischemic and hemorrhagic) 138 2.7 125 2.5 0.92 (0.721.16) 0.48
Ischemic stroke 124 2.4 100 2.0 0.82 (0.631.09) 0.13
Intracranial hemorrhage 13 0.25 21 0.42 1.65 (0.833.31) 0.15
Unknown 1 0.02 4 0.08 3.97 (0.4435.47) 0.22
Disabling or fatal stroke 40 0.78 42 0.84 1.06 (0.691.64) 0.79
Transient ischemic attack without 39 0.78 28 0.57 0.73 (0.451.18) 0.19
Myocardial infarction 38 0.71 31 0.59 0.84 (0.521.35) 0.47
Other thromboembolic events 12 0.22 21 0.40 1.81 (0.893.68) 0.10
Major vascular event 174 3.4 153 3.1 0.89 (0.721.11) 0.29
All deaths 77 1.4 113 2.1 1.52 (1.142.04) 0.004
Vascular causes 19 0.35 27 0.51 1.46 (0.812.64 0.20
Cerebral 9 0.17 10 0.19 1.13 (0.462.78) 0.79
Noncerebral 10 0.18 17 0.32 1.77 (0.813.87) 0.15
Probable vascular causes 6 0.11 18 0.34 3.09 (1.237.80) 0.02
Nonvascular causes 31 0.57 39 0.73 1.31 (0.822.10) 0.26
Uncertain 21 0.39 29 0.55 1.41 (0.822.52) 0.21

* A time-to-first-event model was used for each outcome category; rates are annualized. The total number of
patient-years of exposure for the primary outcome (all strokes) was 5026 for patients assigned to aspirin
plus placebo and 5114 for those assigned to aspirin plus clopidogrel. CI denotes confidence interval.
The five patients with strokes classified as unknown were those who did not undergo neuroimaging; three of the
stroke events were defined as probable ischemic events on central adjudication, and two were defined as
probable ischemic events on local adjudication.
Of 82 strokes that were classified as disabling or fatal (including ischemic, hemorrhagic, and unknown), 16 were fatal
strokes and 65 were defined as disabling on the basis of a modified Rankin score of 4 or more (on a scale of 0 to 6,
with higher scores indicating more severe disability). Another 14 strokes could not be classified and were excluded
from these analyses. Data for patients with nondisabling strokes were censored at the time of the primary event.
Other thromboembolic events included venous thromboembolism (18 events with dual antiplatelet therapy and
10 with aspirin alone) and peripheral arterial embolism (2 and 1, respectively).
Major vascular events were defined as stroke, myocardial infarction, or death resulting from a vascular event.

not complete the study, 2% were lost to follow-up, apies (P=0.46 for the interaction); the hazard ratio
7% withdrew consent, 5% left because of site for recurrent stroke in the group with a blood-
closure, 1% withdrew at the physicians request, pressure target of 130 to 149 mm Hg was 0.84
and 1% withdrew for other reasons (Fig. S1 in the (95% confidence interval [CI], 0.61 to 1.17) and the
Supplementary Appendix). Statin therapy was hazard ratio in the group with a target of less than
prescribed for 84% of patients during follow-up. 130 mm Hg was 1.01 (95% CI, 0.71 to 1.45).
The risk of recurrent stroke among patients
RECURRENT STROKE assigned to receive aspirin alone was 2.7% per year
A total of 263 participants had a recurrent stroke: and was not significantly reduced among those
224 (85%) had an ischemic stroke and 34 (13%) receiving dual antiplatelet therapy (2.5% per year;
had an intracranial hemorrhage; the type of hazard ratio, 0.92; 95% CI, 0.72 to 1.16) (Table 2
stroke was unknown for the remaining 5 pa- and Fig. 1). A nonsignificant decrease of 18% in the
tients (2%) because they did not undergo neuro- relative risk of recurrent ischemic stroke associated
imaging. There was no significant interaction with dual antiplatelet therapy was observed and
between the antiplatelet and blood-pressure ther- was offset by a nonsignificant

N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012 821

The New England Journal of Medicine
Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

1.0 The rate of all major hemorrhages was 1.1% per
year among patients receiving aspirin alone and

0.8 was almost doubled among those assigned to dual
0.7 antiplatelet therapy (hazard ratio, 1.97; 95% CI,
0.6 1.41 to 2.71; P<0.001) (Table 3, and Fig. S2 in the
0.5 Supplementary Appendix) Although the increase in
central nervous system bleeding in the dual
antiplatelet therapy group was not significant
Aspirin plus placebo (hazard ratio, 1.52; 95% CI, 0.79 to 2.93), the
yearly rate of major extracranial hemorrhage was
Aspirin plus clopidogrel
more than doubled in that group (hazard ratio, 2.15;
0 1 2 3 4 5 6 7 8 95% CI, 1.49 to 3.11; P<0.001) (Table 3).
Years since Randomization
Aspirin plus 1517 1272 1027 788 574 355 189 83 3 Among the patients with recurrent acute ische mic
Aspirin plus 1503 1288 1030 802 589 371 205 90 5
strokes who underwent neuroimaging, 58% (126 of
clopidogrel 219) had evidence of an acute small sub-cortical
infarct on neuroimaging, and 71% of clas-sifiable
Figure 1. Probability of the Primary Outcome. recurrent ischemic strokes (133 of 187) were
The hazard ratio for the primary outcome, recurrent stroke, was 0.92 deemed to be recurrent lacunar strokes on the basis
(95% CI, 0.72 to 1.2). of both clinical assessment and imaging review.
The rate of recurrent lacunar infarcts was not
reduced among patients receiving dual anti-platelet
increase of 70% in the relative risk of intracere-bral therapy (infarcts occurred in 67 of the patients
hemorrhage (Table 2, and Fig. S2B in the receiving aspirin alone and 66 of the pa-tients
Supplementary Appendix) There was no hetero- receiving dual antiplatelet therapy) (Table S1 in the
geneity of treatment effect on the primary outcome Supplementary Appendix).
according to age or sex or among the prespecified
subgroups (Fig. 2). In the group receiving dual DIS CUS SION
antiplatelet therapy, there was no significant re-
duction in the risk of disabling or fatal stroke In this cohort of patients with recent lacunar stroke
(hazard ratio, 1.06; 95% CI, 0.69 to 1.64) or in the identified on MRI, the addition of clopido-grel to
composite outcome of stroke, myocardial infarc- aspirin did not reduce stroke recurrence. The
tion, or death from vascular causes (hazard ratio, extreme of the 95% confidence interval around the
0.89; 95% CI, 0.72 to 1.11) (Table 2). observed point estimate was bounded by a relative
risk reduction of 28% with dual an-tiplatelet
DEATHS therapy, arguably excluding a clinically meaningful
All-cause mortality was increased among patients reduction, considering the increased bleeding
assigned to dual antiplatelet therapy as compared associated with dual antiplatelet thera-py. After the
with those assigned to aspirin alone (hazard ratio, SPS3 trial had been designed and initiated, two
1.52; 95% CI, 1.14 to 2.04; P =0.004), with simi- large, randomized trials involving patients with a
lar trends toward an increase in death with dual heterogeneous spectrum of vas-cular disease or
antiplatelet therapy observed for all categories of vascular risk factors assessed dual antiplatelet
cause of death (Table 2, and Fig. S2C in the Sup- therapy with clopidogrel and as-pirin for the
plementary Appendix). Fatal stroke occurred in 16 prevention of vascular events, in-cluding stroke, as
patients, with death occurring in 13 of these compared with either aspirin alone or clopidogrel
patients within 30 days of the stroke. Fatal hem- alone, and the overall results were negative.
orrhages occurred in 9 patients assigned to dual However, the importance of assessing the effect of
antiplatelet therapy and 4 assigned to aspirin alone antiplatelet agents on well-defined subtypes of
(hazard ratio, 2.29; P=0.17); 85% of fatal ischemic stroke is illus-trated by the strongly
hemorrhages (11 of 13) were intracranial (Table 3). positive reduction in the

822 N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012

The New England Journal of Medicine

Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

Aspirin plus Aspirin plus P Value for

Subgroup No. Placebo Clopidogrel Hazard Ratio (95% CI) Interaction
no. of events (annualized rate)
Age 0.40
<65 yr 1757 79 (2.7) 66 (2.2) 0.84 (0.601.20)
65 yr 1263 59 (2.7) 59 (2.8) 1.00 (0.721.50)
Sex 0.91
Male 1902 96 (3.0) 86 (2.8) 0.93 (0.691.20)
Female 1118 42 (2.2) 39 (2.0) 0.90 (0.581.40)
History of diabetes 0.98
Diabetic 1106 70 (3.8) 60 (3.6) 0.93 (0.661.30)
Nondiabetic 1914 68 (2.1) 65 (1.9) 0.93 (0.661.30)
Aspirin at index stroke 0.97
No aspirin 2180 89 (2.5) 80 (2.3) 0.91 (0.671.20)
Aspirin 838 49 (3.3) 45 (3.1) 0.92 (0.621.40)
Race or ethnic group 0.94
Hispanic 916 35 (2.4) 28 (2.0) 0.81 (0.491.30)
White 1538 65 (2.5) 62 (2.4) 0.96 (0.681.40)
Black 492 34 (3.7) 31 (3.6) 0.97 (0.601.60)
Other or mixed 74 4 (3.4) 4 (2.6) 0.97 (0.193.10)
Region of residence 0.83
North America 1960 104 (2.9) 96 (2.7) 0.93 (0.711.20)
Latin America 694 22 (2.2) 21 (2.1) 0.95 (0.521.70)
Spain 366 12 (2.5) 8 (1.8) 0.70 (0.291.70)
0.2 0.5 1.0 2.0 5.0

Aspirin plus Aspirin plus

Clopidogrel Better Placebo Better

Figure 2. Hazard Ratios for the Primary Outcome in Subgroups of Participants.

risk of stroke observed in a study comparing lacunar strokes, a finding that supports the hy-
clopidogrel plus aspirin with aspirin alone in pa- pothesis that the role of platelets is different in
tients with atrial fibrillation in whom the primary different types of ischemic cerebrovascular dis-
mechanism of stroke was cardioembolic. The
14 ease (Table S1 in the Supplementary Appendix).
It has been speculated that thrombosis may have
SPS3 trial documents the lack of benefit of dual
a minimal role in precipitating occlusions of
antiplatelet therapy in a specific, well-defined 25
sub-type of ischemic stroke that results primarily small, penetrating cerebral arteries.
from cerebral small-artery disease. The risk of major hemorrhage was increased
These results must be interpreted in the con- among patients assigned to dual antiplatelet thera-
text of the other component of the trial, which py rather than aspirin (hazard ratio, 1.97; P<0.001),
involved vigorous management of blood pres- and this increase was observed in cases of both
17 extracranial bleeding (hazard ratio, 2.15; P<0.001)
sure. The rate of recurrent stroke among pa-
tients taking aspirin alone (2.7% per year) was and intracranial bleeding (hazard ratio, 1.52; P =
substantially lower than anticipated. The use of 0.21), although the latter increase was not
statins by the majority of the study participants significant. An absolute increase in the rate of
and the good blood pressure control achieved major extracranial bleeding from about 1% per year
probably contributed to the low observed rate in with aspirin to 1.5 to 1.7% per year with dual
our trial, which was similar to the rates in other antiplatelet therapy was anticipated on the basis of
recent trials testing antiplatelet therapies for the data from previous studies.
In the SPS3
prevention of recurrent stroke. 14,15
trial, as in previous trials, the ma-jority of
Dual antiplatelet therapy was associated with
cases of excess bleeding with dual anti-platelet
a trend toward a reduction in recurrent strokes therapy were gastrointestinal. In the large
attributed to atherosclerosis but not recurrent CHARISMA (Clopidogrel for High Atherothrom-

N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012 823

The New England Journal of Medicine
Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.

Table 3. Safety Outcomes.*

Aspirin plus Placebo Aspirin plus Clopidogrel Hazard Ratio

Outcome (N=1503) (N=1517) (95% CI) P Value

no. rate (%/yr) no. rate (%/yr)

All major hemorrhages 56 1.1 105 2.1 1.97 (1.412.71) <0.001
Intracranial hemorrhages 15* 0.28 22 0.42 1.52 (0.792.93) 0.21
Intracerebral 8 0.15 15 0.28 1.92 (0.824.54) 0.14
Subdural or epidural 6 0.11 7 0.13 1.23 (0.413.64) 0.72
Other 4 0.07 2 0.04 0.53 (0.102.89) 0.46
Extracranial bleeding 42 0.79 87 1.7 2.15 (1.493.11) <0.001
Gastrointestinal 28 0.52 58 1.1 2.14 (1.363.36) <0.001
Fatal hemorrhages 4 0.07 9 0.17 2.29 (0.707.42) 0.17
Intracranial 4 0.07 7 0.13 1.78 (0.526.07) 0.36
Extracranial 0 0 2 0.04

* A time-to-first-event model was used for each outcome category; rates are annualized. All adjudications were
performed centrally by the SPS3 adjudication committee. CI denotes confidence interval.
In the group taking aspirin plus placebo, two events were adjudicated as both intracerebral and other, and one event
was adjudicated as both intracerebral and subdural. In the group taking aspirin plus clopidogrel, one event was adjudi-
cated as both intracerebral and other, and one event was adjudicated as both intracerebral and subdural.
The site of bleeding was determined by an investigator at the local study center.

botic Risk and Ischemic Stabilization, Manage- effect of clopidogrel added to aspirin have not
ment, and Avoidance) trial, the risk of intracranial 15,32
bleeding was not increased when clopidogrel was shown an overall increase in mortality. Con-
added to aspirin, but this result is at odds with thesequently, the increased mortality associated with
14,28 dual antiplatelet therapy observed in the SPS3 trial
findings in other randomized trials. can probably be explained by the spe-cific patient
The dose of enteric-coated aspirin (325 mg population or by chance. The SPS3 trial cohort had
daily) in the SPS3 trial was higher than that used in
a low frequency of recognized coronary artery
several other trials testing the effects of clo pidogrel
14,22 disease at entry (10%), a low inci-dence of
combined with aspirin. Exploratory analyses in myocardial infarction (0.7% per year), and a lower
one study suggested that when com-bined with observed mortality than that in other recent stroke-
clopidogrel, higher doses of aspirin could be less 21,23,24
efficacious than lower doses for the prevention of prevention trials, suggesting that vascular
29 disease associated with lacunar stroke has a unique
vascular events. However, in two trials of aspirin profile.
plus clopidogrel in patients with acute coronary In conclusion, in this clinical trial of clopido-
syndromes, the aspirin dose had no observed effect
30,31 grel and aspirin, as compared with aspirin alone, in
on the prevention of is chemic events, nor did patients with a recent lacunar stroke identified on
it have an effect on stroke outcomes in a trial MRI, we found that the anticipated increase in the
involving patients with vascular disease or vascular
29 risk of major hemorrhage with dual anti-platelet
risk factors, mak-ing it unlikely that the aspirin therapy was not offset by a reduction in the risk of
dose tested in the SPS3 trial accounted for the stroke recurrence, and there was an unexpected
absence of an in-crease in efficacy with dual
increase in mortality. Additional re-sults from the
antiplatelet therapy.
component of the SPS3 trial involv-ing blood
The increase in the rate of death from any
pressure control are anticipated in 2012.
cause among the patients assigned to clopidogrel
plus aspirin was unexpected, and the observed Supported by a grant from the National Institute of Neuro-
logical Disorders and Stroke (U01 NS38529-04A1) and by
increase was not accounted for by fatal hemor- Sanofi-Aventis and Bristol-Myers Squibb, which donated the
rhages, the most biologically plausible explana- clopidogrel and matching placebo used in the study.
tion. Previous randomized trials assessing the Disclosure forms provided by the authors are available with
the full text of this article at

824 N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012

The New England Journal of Medicine

Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.


1. Bogousslavsky J, Van Melle G, Regli 13. Furie KL, Kasner SE, Adams RJ, et al. botic events. N Engl J Med
F. The Lausanne Stroke Registry: analysis Guidelines for the prevention of stroke in 2006;354:1706-17.
of 1,000 consecutive patients with first patients with stroke or transient ischemic 23. Sacco RL, Diener H-C, Yusuf S, et al.
stroke. Stroke 1988;19:1083-92. attack: a guideline for healthcare profes- Aspirin and extended-release dipyrida
2. Kolominsky-Rabas PL, Weber M, Ge- sionals from the American Heart Associa- mole versus clopidogrel for recurrent
feller O, Neundoerfer B, Heuschmann PU. tion/American Stroke Association. Stroke stroke. N Engl J Med 2008;359:1238-51.
Epidemiology of ischemic stroke subtypes 2011;42:227-76. 24. Bousser M-G, Amarenco P, Chamorro A,
according to TOAST criteria: incidence, 14. The ACTIVE Investigators. Effect of et al. Terutoban versus aspirin in pa-tients
recurrence, and long-term survival in clopidogrel added to aspirin in patients with cerebral ischaemic events (PERFORM):
ischemic stroke subtypes: a population- with atrial fibrillation. N Engl J Med a randomised, double-blind, parallel-group
based study. Stroke 2001;32:2735-40. 2009;360:2066-78. trial. Lancet 2011;377:2013-
3. Wolfe CD, Rudd AG, Howard R, et al. 15. The Clopidogrel in Unstable Angina to 22. [Erratum, Lancet 2011;378:402.]
Incidence and case fatality rates of stroke Prevent Recurrent Events Trial Investi- 25. Wardlaw JM. What causes lacunar
subtypes in a multiethnic population: the gators. Effects of clopidogrel in addition stroke? J Neurol Neurosurg Psychiatry
South London Stroke Register. J Neurol to aspirin in patients with acute coronary 2005;76:617-9.
Neurosurg Psychiatry 2002;72:211-6. syndromes without ST-segment elevation. 26. Berger PB, Bhatt DL, Fuster V, et al.
4. Del Brutto OH, Mosquera A, Snchez X, N Engl J Med 2001;345:494-502. [Errata, Bleeding complications with dual anti-
Santos J, Noboa CA. Stroke subtypes among N Engl J Med 2001;345:1506, 1716.] platelet therapy among patients with sta-
Hispanics living in Guayaquil, Ec-uador: 16. Usman MHU, Notaro LA, ble vascular disease or risk factors for
results from the Luis Vernaza Hospi-tal Nagarakanti R, et al. Combination vascular disease: results from the Clo
Stroke Registry. Stroke 1993;24:1833-6. antiplatelet therapy for secondary stroke pidogrel for High Atherothrombotic Risk
5. Lavados PM, Sacks C, Prina L, et al. prevention: en-hanced efficacy or double and Ischemic Stabilization, Management,
Incidence, 30-day case-fatality rate, and trouble? Am J Cardiol 2009;103:1107-12. and Avoidance (CHARISMA) trial. Circu-
prognosis of stroke in Iquique, Chile: a 2- 17. Benavente OR, White C, Pearce LA, lation 2010;121:2575-83.
year community-based prospective study et al. The Secondary Prevention of Small 27. Delaney JA, Opatrny L, Brophy JM,
(PISCIS project). Lancet 2005;365:2206-15. Subcortical Strokes (SPS3) study. Int J Su-issa S. Drug-drug interactions between
6. Sacco RL, Kargman DE, Zamanillo Stroke 2011;6:164-75. antithrombotic medications and the risk of
MC. Race-ethnic differences in stroke risk 18. White CL, Szychowski JM, Roldan A, gastrointestinal bleeding. CMAJ 2007;
factors among hospitalized patients with et al. Clinical features and racial/ethnic 177:347-51.
cerebral infarction: the Northern differences among 3020 participants in the 28. Hart RG, Tonarelli SB, Pearce LA.
Manhattan Stroke Study. Neurology 1995; Secondary Prevention of Small Sub- Avoiding central nervous system bleeding
45:659-63. cortical Strokes (SPS3) trial. J Stroke during antithrombotic therapy: recent data
7. Saposnik G, Del Brutto OH. Stroke in Cere-brovasc Dis 2012 April 17 (Epub and ideas. Stroke 2005;36:1588-93.
South America: a systematic review of in- ahead of print). 29. Steinhubl SR, Bhatt DL, Brennan DM,
cidence, prevalence, and stroke subtypes. 19. Hazuda HP, Comeaux PJ, Stern MP, et al. Aspirin to prevent cardiovascular
Stroke 2003;34:2103-7. Haffner SM, Eifler CW, Rosenthal M. A disease: the association of aspirin dose and
8. Worley KL, Lalonde DR, Kerr DR, Be- comparison of three indicators for identi- clopidogrel with thrombosis and bleed-
navente O, Hart RG. Survey of the causes of fying Mexican Americans in epidemio- ing. Ann Intern Med 2009;150:379-86.
stroke among Mexican Americans in South logic research: methodological findings 30. The CURRENTOASIS 7 Investigators.
Texas. Tex Med 1998;94:62-7. from the San Antonio Heart Study. Am J Dose comparisons of clopidogrel and as-pirin
9. Zweifler RM, Lyden PD, Taft B, Kelly Epidemiol 1986;123:96-112. in acute coronary syndromes. N Engl J Med
N, Rothrock JF. Impact of race and ethnic- 20. McClure LA, Szychowski J, 2010;363:930-42. [Erratum, N Engl J Med
ity on ischemic stroke: the University of Benavente O, Coffey CS. Sample size re- 2010;363:1585.]
California at San Diego Stroke Data Bank. estimation in an on-going NIH-sponsored 31. Peters RJ, Mehta SR, Fox KA, et al.
Stroke 1995;26:245-8. clinical trial: the Secondary Prevention of Effects of aspirin dose when used alone or
10. Boiten J, Lodder J. Prognosis for sur- Small Subcortical Stroke trial experience. in combination with clopidogrel in pa-
vival, handicap and recurrence of stroke in Con-temp Clin Trials 2012;33:1088-93. tients with acute coronary syndromes:
lacunar and superficial infarction. 21. Diener H-C, Bogousslavsky J, Brass observations from the Clopidogrel in
Cerebrovasc Dis 1993;3:221-6. LM, et al. Aspirin and clopidogrel com- Unstable angina to prevent Recurrent
11. Ross GW, Petrovitch H, White LR, et pared with clopidogrel alone after recent Events (CURE) Study. Circulation
al. Characterization of risk factors for ischaemic stroke or transient ischaemic 2003;108:1682-7.
vascular dementia: the Honolulu-Asia Ag- attack in high-risk patients (MATCH): 32. Palacio S, Hart RG, Pearce LA, Bena-
ing Study. Neurology 1999;53:337-43. randomised, double-blind, placebo-con- vente OR. Effect of addition of clopidogrel to
12. Tatemichi TK, Desmond D, Paik M, et trolled trial. Lancet 2004;364:331-7. aspirin on mortality: systematic review of
al. Clinical determinants of dementia re- 22. Bhatt DL, Fox KA, Hacke E, et al. randomized trials. Stroke 2012;43:2157-
lated to stroke. Ann Neurol 1993;33:568- Clopidogrel plus aspirin versus aspirin 62.
75. alone for the prevention of atherothrom- Copyright 2012 Massachusetts Medical Society.

N ENGL J MED 367;9 NEJM.ORG AUGUST 30, 2012 825

The New England Journal of Medicine
Downloaded from on February 9, 2017. For personal use only. No other uses without permission.
Copyright 2012 Massachusetts Medical Society. All rights reserved.