Aspirin and Ticlopidine for Prevention of Recurrent Stroke in Black Patients a Randomized Trial

ORIGINAL CONTRIBUTION

Aspirin and Ticlopidine for Prevention of Recurrent Stroke in Black Patients A Randomized Trial

Philip B. Gorelick, MD, MPH Context Blacks are disproportionately affected by stroke, and they are about 2 times DeJuran Richardson, PhD more likely than most other individuals in the United States to die of or experience stroke. Michael Kelly, MD Objective To determine the efficacy and safety of aspirin and ticlopidine to prevent recurrent stroke in black patients. Sean Ruland, DO Design, Setting, and Patients Randomized, double-blind, investigator-initiated, Elena Hung, MS multicenter trial of 1809 black men and women who recently had a noncardioembolic Yvonne Harris, MPA, CCRA ischemic stroke and who were recruited between December 1992 and October 2001 from 62 academic and community hospitals in the United States and followed up for Steven Kittner, MD, MPH up to 2 years. Sue Leurgans, PhD Intervention A total of 902 patients received 500 mg/d of ticlopidine and 907 re- for the African American Antiplatelet ceived 650 mg/d of aspirin. Stroke Prevention Study (AAASPS) Main Outcome Measures Recurrent stroke, myocardial infarction, or vascular death Investigators was the composite primary end point (according to intention-to-treat analysis). The LACKS ARE DISPROPORTION- secondary outcome was fatal or nonfatal stroke. ately affected by stroke, yet Results The blinded phase of the study was halted after about 6.5 years when futility they have been underrepre- analyses revealed a less than 1% probability of ticlopidine being shown superior to as- sented in clinical trials.1-8 Rec- pirin in the prevention of the primary outcome end point. The primary outcome of re- ommendationsB for stroke prevention in current stroke, myocardial infarction, or vascular death was reached by 133 (14.7%) of this population have been based largely 902 patients assigned to ticlopidine and 112 (12.3%) of 907 patients assigned to aspirin (hazard ratio, 1.22; 95% confidence interval, 0.94-1.57). Kaplan-Meier curves for on trials that have included few black time to event for the primary outcome did not differ significantly (P=.12 by log-rank participants. This may not be an opti- test). Kaplan-Meier curves for time to the secondary outcome of fatal or nonfatal stroke mal practice because blacks are among approached a statistically significant reduction favoring aspirin over ticlopidine (P=.08 those with a higher prevalence of ma- by log-rank test). The frequency of laboratory-determined serious neutropenia was 3.4% jor cardiovascular risk factors, a differ- for patients receiving ticlopdine vs 2.2% for patients receiving aspirin (P=.12) and 0.3% ent distribution of atherosclerotic oc- vs 0.2% for thrombocytopenia, respectively (P=.69). One ticlopidine-treated patient clusive cerebral vascular lesions, developed thrombocytopenia, which was thought to be a case of possible thrombotic vascular biological differences such as thrombocytopenia purpura, and recovered after therapy with plasmapheresis. low renin hypertension, and a differ- Conclusions During a 2-year follow-up, we found no statistically significant different pattern of use of medical proce- ence between ticlopidine and aspirin in the prevention of recurrent stroke, myocardial dures and access to care8-13 that could infarction, or vascular death. However, there was a nonsignificant trend for reduction influence outcome. of fatal or nonfatal stroke among those in the aspirin group. Based on these data and the risk of serious adverse events with ticlopidine, we regard aspirin as a better treat- A subgroup analysis of the Ticlopi- ment for aspirin-tolerant black patients with noncardioembolic ischemic stroke. 14,15 dine Aspirin Stroke Study (TASS) JAMA. 2003;289:2947-2957 www.jama.com suggested a more favorable risk- benefit profile for nonwhites than Author Affiliations: Departments of Neurologic Sci- County Hospital, Chicago, Ill (Dr Kelly); and the De- whites. Specifically, among the 495 ences (Drs Gorelick, Ruland, and Leurgans, and Ms partment of Neurology, University of Maryland, Bal- black and 108 nonwhite and non- Harris) and Preventive Medicine (Dr Richardson and timore (Dr Kittner). Ms Hung), Rush Medical College, Chicago, Ill; De- Corresponding Author and Reprints: Philip B. Gorelick, partment of Mathematics and Computer Science, Lake MD, MPH, Center for Stroke Research, Rush Medi- For editorial comment see p 3005. Forest College, Lake Forest, Ill (Dr Richardson); De- cal College, 1645 W Jackson, Suite 400, Chicago, IL partment of Medicine, Division of Neurology, Cook 60612 (e-mail: [email protected]).

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

black study participants, there was a dose of 650 mg/d in accordance with Primary and Secondary 24.1% relative risk reduction (RRR) for the recommendations by Barnett et al.22 Hypotheses stroke and death at 2 years favoring Eligibility criteria included black (Af- The primary hypothesis of this ran- ticlopidine (500 mg/d) over aspirin rican American) race; 29-85 years of age domized, double-blind, investigator- (1300 mg/d), and 10% fewer serious ad- inclusive; noncardioembolic ischemic initiated clinical trial was that ticlopi- verse events (SAEs).15 Overall in TASS, stroke with onset at least 7 days but not dine (500 mg/d) was more effective than there was a 12% RRR for nonfatal stroke more than 90 days; cranial computed aspirin (650 mg/d) in preventing the or death from any cause (P=.05) favor- tomographic scan or magnetic reso- composite outcome of recurrent stroke, ing ticlopidine at 3 years. nance image of the brain consistent with myocardial infarction, or vascular death The current study was designed in occurrence of the entry cerebral in- (death due to ischemic or hemor- 1993, with the belief that a targeted re- farct; measurable neurological deficit rhagic stroke, myocardial infarction, current stroke prevention study for that correlates at onset with entry ce- sudden death, pulmonary embolism, blacks was justified given their dispro- rebral infarct; informed consent; and heart failure, visceral or limb infarc- portionate stroke burden, promising availablity of patient to be followed up tion, or a vascular procedure or opera- data for ticlopidine as a recurrent stroke in an outpatient treatment program.16 tion) among blacks with noncardioem- preventive treatment in nonwhites, and Exclusion criteria included tran- bolic ischemic stroke who were the lack of previous substantial repre- sient ischemic attack, subarachnoid followed up for up to 2 years. The pre- sentation of blacks in stroke clinical hemorrhage, cardiac source embo- specified secondary hypotheses were trials. The primary outcome of the Af- lism, iatrogenic stroke, nonatheroscle- that the incidence of the outcome end rican American Antiplatelet Stroke Pre- rotic stroke, postoperative stroke oc- points of recurrent stroke or death; non- vention Study (AAASPS) was the com- curring within 30 days of operation, or fatal or fatal stroke; recurrent stroke, posite end point of recurrent stroke, carotid endarterectomy as primary myocardial infarction, or death from all myocardial infarction, or vascular death. treatment measure for entry cerebral in- causes; vascular death; death from all farct; mean arterial blood pressure causes; or myocardial infarction would METHODS higher than 130 mm Hg on 3 consecu- be lower in ticlopidine-treated pa- A description of the design and meth- tive days; modified Barthel index of less tients compared with aspirin-treated pa- ods of AAASPS has been reported pre- than 10; history of dementia or neuro- tients followed up for up to 2 years. viously16 in accordance with criteria pro- degenerative disease; severe comorbid posed by the Consolidated Standards of condition (eg, cancer) judged to limit Randomization Reporting Trials.17,18 That article16 in- survival during 2-year follow-up; en- The ratio of those receiving ticlopi- cluded a discussion of barriers to black rollment in another clinical trial; al- dine and aspirin was 1:1, and the se- participation in clinical trials and how lergy or sensitivity to study drugs; quence was stratified by site to bal- they were overcome,12,19,20 the rationale woman of childbearing potential; gas- ance the treatment groups. All study for study drug selection, relationships trointestinal tract bleeding, bleeding personnel were masked (blinded) from established with primary care physi- diathesis, or platelet or other hemato- treatment assignment with the excep- cians, management of cardiovascular risk logic abnormality (judged to be a con- tion of 1 study statistician who devel- factors, and other major aspects of the traindication for administration of study oped the randomization algorithm. Af- study. The diagnosis of stroke and stroke drugs) currently active or clinically ac- ter written informed consent was subtype was determined after review of tive in the past year; hematuria or posi- obtained, local study site personnel source documents and case report forms tive stool guaiac test related to major called a dedicated automated tele- and by application of criteria from the bleeding source; and prolonged pro- phone registration system for AAASPS, Trial of ORG 10172 in Acute Stroke thrombin time or partial thromboplas- which was operated by the Moffitt Can- Treatment (TOAST)21 by local princi- tin time, blood urea nitrogen level cer Research Institute at the Univer- pal investigators for the entry stroke and higher than 40 mg/dL, serum creati- sity of South Florida in Tampa, to reg- by the AAASPS adjudication commit- nine level higher than 2.0 mg/dL (176.8 ister a study participant.23 tee for all outcome events. Entry and out- µmol/L), thrombocytopenia or neutro- come stroke cases received computed penia defined by the lower limit of nor- Medication Dosing Schedule tomography or magnetic resonance mal for the platelet count or white blood Ticlopidine was packaged as 250-mg imaging of the head. cell count (unless absolute neutrophil tablets and aspirin as 325-mg coated At the time our study design was de- count of at least 1800/mm3), or liver tablets. The placebo tablets had iden- veloped in the early to mid-1990s, we function tests 2 or more times the up- tical physical properties. Study medi- believed that there was uncertainty per limit of normal. All sites had to re- cations were dispensed from plastic about the preferred aspirin dose for receive formal approval from their insti- bottles. Medication compliance was de- current stroke prevention.22 Given this tutional review board before the study termined by pill count at each fol- uncertainty, we opted for an aspirin could commence at a local site. low-up visit. A dose of 250 mg of ticlo-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

pidine was administered with a placebo materials about stroke prevention, and of 15%, 23%, 35%, 67%, and 76%, af- aspirin tablet twice a day with meals. a booklet about the AAASPS program ter which the blinded treatment phase A dose of 325 mg of aspirin was ad- that was developed in collaboration of the study was terminated. ministered with a placebo ticlopidine with the AAASPS community advi- All analyses were conducted using SAS tablet twice a day with meals. sory board. This booklet also listed (SAS Institute Inc, Cary, NC) and medications that were to be avoided or S-Plus (Mathsoft, Cambridge, MA) sta- Laboratory Monitoring were contraindicated. Study patients tistical software according to the inten- and Scheduling of Visits were educated about possible SAEs of tion-to-treat principle and were based Before entry into the study, at 12 months, the study medications and given con- on available patient data through March 24 months, and at any time a study par- tact information to use in the event 31, 2002. Categorical values were com- ticipant experienced an outcome event questions or problems arose. For study pared using ␹2 tests and the Fisher or terminated from the trial, the follow- patients who reached an outcome end exact test. Continuous data were com- ing laboratory studies were obtained: point, local principal investigators were pared using Wilcoxon and t tests. Time- complete blood count and platelet count; given the option to treat these patients to-event curves were calculated using the blood urea nitrogen; serum creatinine; with open-label aspirin or to transi- Kaplan-Meier method and were com- total cholesterol, low-, high- and very tion patients to community care for pared between treatment groups using low-density lipoprotein cholesterol, and stroke prevention according to their pri- the log-rank test. Cox proportional haz- triglycerides; total bilirubin, alanine ami- mary care physicians. ards modeling was also used to com- notransferase, aspartate aminotransfer- pare treatment groups while adjusting ase, lactate dehydrogenase, and alka- Maintenance of Safety for various covariates such as age, sex, line phosphatase; serum glucose and The AAASPS included several mecha- and entry stroke subtype and sever- electrolytes; and urinalysis. Complete nisms to ensure patient safety: a prede- ity.26 Secondary end points, as well as blood count and platelet count were per- termined laboratory “panic” value sys- prespecified subgroups (sex, entry stroke formed every 2 weeks during the first 3 tem whereby the main laboratory notified subtype, and stroke severity, and vari- months of the study or at any unsched- the local investigative team and the clini- ous categories of elapsed time between uled time the local investigative team cal safety monitor of a critical value; an onset of the index stroke and start of deemed that it was indicated. internal AAASPS inhouse safety com- treatment), were analyzed using simi- Study participants were examined in mittee; and an external data safety and lar methods. Study patients who person at baseline; every 2 weeks dur- monitoring board that was appointed by dropped out of the study were cen- ing the first 3 months; and at 6, 10, 12, the National Institutes of Health. sored at the time of their last visit. 16, 20, and 24 months; and at any un- Fultility analyses were performed and scheduled time the investigative team Sample Size Calculation presented to the data safety and moni- deemed that it was indicated for rea- and Statistical Analysis toring board (DSMB) appointed by the son of patient safety, medication com- Data from blacks and nonwhites in National Institutes of Health (NIH) as pliance, or the occurrence of outcome TASS14,15 were used to develop sample part of study interim analysis reports. events or SAEs. Telephone contact was size estimates for AAASPS. We antici- These analyses were conducted to as- made during study months for which pated that AAASPS patients would gen- sess the likelihood of observing a sta- patients did not have an in-person ex- erally be in poorer health than those in tistically significant result if the trial was amination to screen for medication TASS. A 2-year event rate of 25% was to continue to full information (the ob- compliance, outcome events, and SAEs. projected for the AAASPS aspirin treat- servance of 306 primary outcome Because blacks are more likely to ex- ment group. A total of 1410 patients events). The method of futility analy- perience cyclical neutropenia,15 and the would be required to achieve 80% sis was that of conditional power, as de- incidence of severe neutropenia with power to detect a 25% RRR with re- fined by the conditional probability of ticlopidine use is uncommon (Ͻ1%),15 spect to the 2-year primary event rate rejecting the hypothesis of no treat- we believe that study personnel re- based on use of a 2-tailed log-rank test ment difference at the end of the trial mained blinded in relation to labora- with an overall ␣ level of .05. Allow- given current interim data. This method tory data. ing for an increase of 15% for attrition was chosen before the trial began. The due to voluntary withdrawal and loss conditional probability was computed Therapy Phase to follow-up, and an additional 7% to by simulating the number and timing Before entry into the trial, acute hos- allow for interim analyses using the of future events based on the current pital care and stroke diagnostic evalu- Lan-DeMets strategy with O’Brien- event rates. Simulations were per- ations were at the discretion of the lo- Fleming boundaries, a total sample size formed using S-Plus. A conditional cal principal investigator or primary of 1800 was estimated to yield 306 pri- power close to zero indicated little care physician. Study patients re- mary outcome events.24,25 Interim analy- chance of crossing the statistical bound- ceived verbal instruction and written ses were performed at information times ary were the trial to go to completion.

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

(15.1% for ticlopidine vs 11.5% for as- Figure 1. Flow of Participants Through the African American Antiplatelet Stroke Prevention Study pirin; P =.02). The time-to-occurrence of lost to follow-up and volun- 1809 Participants Randomized tary withdrawal did not differ significantly between treatment groups 902 Assigned to Receive Ticlopidine 907 Assigned to Receive Aspirin (PϾ.25 for each comparison). At the time the blinded phase of the Status as of March 31, 2002 Status as of March 31, 2002 study was halted, excluding patients 477 Completed 2-Year Follow-up or Experienced 485 Completed 2-Year Follow-up or Experienced Outcome Event Outcome Event with outcome events, 47.1% of the ticlo- 276 Withdrawn 246 Withdrawn pidine group and 46.5% of the aspirin 81 Adverse Events 71 Adverse Events group had not completed the 2-year fol- 1 Used Contraindicated Medication 3 Used Contraindicated Medication 37 Lost to Follow-up 31 Lost to Follow-up low-up, and 149 participants re- 13 Misrandomization 16 Misrandomization mained in the ticlopidine group and 176 28 Nonadherent 25 Nonadherent in the aspirin group (P=.11). Overall, 10 Withdrawn by Primary Physician 6 Withdrawn by Primary Physician 6 Relocated 4 Relocated the study database contained an aver- 100 Voluntary Withdrawals 90 Voluntary Withdrawals age of 1.54 years of follow-up data per 149 Remained in Follow-up 176 Remained in Follow-up study patient. Median follow-up was 710 days for ticlopidine-treated pa- 902 Included in Analysis 907 Included in Analysis tients and 716 days for aspirin-treated patients. On average, study participants completed 88% of study visits. Study Organization Futility analyses were computed as The main AAASPS organizational com- part of the statistical analyses presented Patient Characteristics ponents are listed at the end of the ar- to the NIH-appointed DSMB as part of Baseline characteristics of patients in the ticle, with a listing of local sites, the in- study interim analysis reports. The lat- 2 treatment groups are balanced and are vestigative team at each local site, and est such report was based on available pa- summarized in TABLE 1. Blacks have a the number of study patients enrolled tient data through March 31, 2002, and higher risk of lacunar infarction than at each site. Also, we performed rou- suggested that there was less than a 1% whites.8 Furthermore, our eligibility cri- tine audits at 27 centers that were our chance that ticlopidine-treated patients teria excluded those with atrial fibrilla- larger enrollment centers, and at 4 cen- would have a superior outcome com- tion, cardiac sources of embolism that ters for cause (3 sites for possible dif- pared with aspirin-treated patients if the would require warfarin therapy, or large ficulty with data collection and fol- trial were to continue to completion. Af- artery carotid occlusive disease treated low-up procedures, and at 1 site for ter careful review of this report and con- by carotid endarterectomy, which would possibly entering a participant for sultation with study staff, the study sci- serve to increase the likelihood of en- whom several key data points did not entific advisory committee, and other rolling lacunar infarction patients into the match the source documentation). In external experts, the DSMB recom- study. Drug compliance was assessed by all audits, all issues were resolved, and mended that the blinded phase of the pill count. Overall, median compliance no major action was required to be study end on July 15, 2002. At that junc- was 91%; 90% in the ticlopidine group taken by the AAASPS clinical or data ture, study participants were given the and 92% in the aspirin group. management centers. option of remaining in the study taking study-sponsored open-label aspirin or Outcomes RESULTS transition into the community for stroke A total of 245 primary outcomes (the Recruitment and Follow-up prevention therapy according to their composite of recurrent stroke, myocar- Recruitment commenced on Decem- community physician. dial infarction, or vascular death) oc- ber 12, 1995, and was completed on Oc- FIGURE 1 summarizes the status of curred during the trial (TABLE 2). There tober 1, 2001. A total of 1809 patients patients in the trial by treatment group. were 133 primary outcomes among the were enrolled in the study; 902 in the Three hundred seventy participants ticlopidine-treated patients and 112 ticlopidine group and 907 in the aspi- (41.0%) in the ticlopidine group and among the aspirin-treated patients. As rin group. The number of patients en- 403 participants (44.4%) in the aspi- shown by Kaplan-Meier curves in rolled by study year was 354 in 1996, rin group completed the 24-month ex- FIGURE 2, there was no statistically sig- which included December 12, 1995 amination. In general, all types of with- nificant difference between treatment through December 31, 1996; 326 in drawal were slightly more common in groups in the occurrence of the pri- 1997; 315 in 1998; 303 in 1999; 314 patients receiving ticlopidine, but a sta- mary outcome end point (P=.12 by in 2000; and 197 from January 1 tistically significant difference was ob- log-rank test; the O’Brien-Fleming through October 1, 2001. served for withdrawal due to SAEs boundaries were not crossed). Cox

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

proportional hazards modeling that in- not find a statistically significant differ- Serious Adverse Events cluded key covariates, such as treat- ence based on intention-to-treat or on- Data on patients reporting SAEs appear ment group assignment, age, and base- treatment analyses. in TABLE 4. Overall, 532 patients line risk factors, showed a similar treatment difference (P=.11). There were slightly more fatal (4 vs Table 1. Baseline Characteristics 2) or nonfatal (102 vs 84) recurrent Ticlopidine Aspirin strokes among ticlopidine-treated pa- (n = 902) (n = 907) P Value* tients, but this was not statistically sig- Characteristic† nificant (Table 2). However, the Age, mean (SD), y 60.9 (10.7) 61.6 (10.4) .13 Body mass index, mean (SD) 29.9 (7.1) 30.0 (6.8) .23 Kaplan-Meier curves indicate time to Women 492 (54.5) 475 (52.4) .35 fatal or nonfatal stroke (Figure 2) ap- Education proached a statistically significant dif- ՅHigh school 666 (73.8) 676 (74.5) .74 ference (P=.08 by log-rank test) in fa- ϾHigh school 216 (24.0) 209 (23.0) .67 vor of the aspirin-treated patients. Unknown 20 (2.2) 22 (2.4) .77 Secondary outcomes including any Household income Ͻ recurrent stroke, death (all cause), vas- $14 999 397 (44.0) 403 (44.4) .86 Ͼ$14 999 269 (29.8) 258 (28.4) .52 cular death, recurrent stroke or death Unknown 236 (26.2) 246 (27.1) .64 (all cause), and the composite end point of recurrent stroke, myocardial infarc- Risk Factor‡ tion, or death (all cause) were not sta- Hypertension 764/899 (85.0) 775/898 (86.3) .43 Diabetes mellitus 359/896 (40.1) 379/901 (42.1) .39 tistically significantly different be- Cardiovascular tween treatment groups (Table 2). Hypercholesterolemia 331/815 (40.6) 366/840 (43.6) .22 Overall, we observed a 2-year primary Angina pectoris 98/893 (11.0) 93/899 (10.3) .67 event rate of 19.7% among ticlopidine- Myocardial infarction 81/891 (9.1) 89/896 (9.9) .54 treated patients and 16.3% among those Congestive heart failure 44/891 (4.9) 51/896 (5.7) .48 treated with aspirin. Valvular heart disease 16/892 (1.8) 16/899 (1.8) .98 We compared recurrent stroke sub- Atrial fibrillation 10/895 (1.1) 14/902 (1.6) .42 type and severity between treatment Surgery Cardiac 31/899 (3.4) 37/902 (4.1) .47 groups and did not find a statistically sig- Thoracic or abdominal aortic 17/899 (1.9) 15/902 (1.7) .71 nificant difference (TABLE 3). Also, we Peripheral arterial vascular 18/900 (2.0) 13/901 (1.4) .36 extended the follow-up data for the pri- Carotid endarterectomy 4/900 (0.4) 5/903 (0.6) .99§ mary outcome cluster by adding infor- Other mation from April to December 2002. Chronic lung disease 47/899 (5.2) 61/901 (6.8) .17 There were 5 additional events: 3 strokes Epilepsy or seizures 41/897 (4.6) 44/902 (4.9) .76 and 1 myocardial infarction in the ticlo- Leg claudication 38/893 (4.3) 35/897 (3.9) .71 Family history pidine group and 1 stroke in the aspi- Stroke 378/809 (46.7) 382/814 (46.9) .93 rin group. The primary outcome find- Myocardial infarction 307/801 (38.3) 303/815 (37.2) .64 ings, however, were not substantially Hypercholesterolemia 132/545 (24.2) 121/550 (22.0) .38 changed (P=.08 by log-rank test). Lifestyle In addition to the intention-to-treat Cigarette smoking (past or current) 556/898 (61.9) 574/906 (63.4) .53 analysis, we performed an on-treat- No exercise 550/882 (62.4) 559/899 (62.2) .94 ment analysis for the primary event out- Alcohol use within 24 hours of stroke 86/880 (9.8) 86/897 (9.6) .90 come cluster. For each patient, we com- Entry Stroke† puted the percentage of time on blinded Large vessel atherothromboembolic 185 (20.5) 171 (18.9) .38 study medication prior to the occur- Small vessel (lacunar) 600 (66.5) 621 (68.5) .36 rence of an outcome event or comple- Other etiology or unknown cause 117 (13.0) 114 (12.6) .80 tion of the study (whichever came first). Barthel score, mean (SD) 18.6 (2.5) 18.6 (2.5) .94 Glasgow Outcome Assessment scale We then performed log-rank tests on No or minimal disability 569 (63.2) 570 (62.8) .87 subgroups of patients based on this per- Moderate or severe disability 331 (36.8) 337 (37.2) .87 centage (ie, those having a percentage on NIH Stroke Scale score, mean (SD) 3.03 (3.00) 3.05 (3.32) .89 blinded study medication (Ն60% or Abbreviation: NIH, National Institutes of Health. Ն *Based on the ␹2 test and t test. 80%). We observed P=.10 for the 60% †Values expressed as number (percentage) unless otherwise indicated. cutoff and P=.11 for the 80% cutoff com- ‡Values expressed as No. of patients reporting/No. of patients for whom a response was given (percentage). §Based on Fisher exact test. pared with P=.12 overall. Thus, we did

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

reported SAEs: 270 (29.9%) among patients, but this did not reach statis- treatment with plasmapheresis. Gas- ticlopidine-treated patients and 262 tical significance. One case of throm- trointestinal tract hemorrhage was (28.9%) among aspirin-treated patients. bocytopenia, which occurred early after slightly more common in aspirin- Diarrhea (0.3% vs 0.2%), neutropenia the enrollment phase of the study was treated patients (0.9% vs 0.4%), but this (3.4% vs 2.2%), and thrombocytope- initiated, was diagnosed as a possible did not reach statistical significance. The nia (0.3% vs 0.2%) were slightly more case of thrombotic thrombocytopenic time-to-occurrence of a SAE did not dif- frequent among ticlopidine-treated purpura. The patient recovered after fer significantly between the treatment groups (P=.39), but the time-to- Table 2. Outcome Events According to Treatment Assignment occurrence of a SAE did (P=.003) (not No. (%) of Patients shown in Table 4). For study patients who had to per- Ticlopidine Aspirin P manently discontinue blind phase (n = 902) (n = 907) HR (95% CI)* Value† Primary Outcome Event‡ medication due to a SAE (not shown Recurrent stroke in Table 4), thrombocytopenia (0.2% Fatal 4 (0.4) 2 (0.2) NA .45§ vs 0.0%), neutropenia (1.8% vs 1.2%), Nonfatal 102 (11.3) 84 (9.3) 1.25 (0.94-1.67) .15 and rash (1.7% vs 0.6%) were more Myocardial infarction common among ticlopidine-treated pa- Fatal 1 (0.1) 0 (0) NA .50§ tients, but only the occurrence of rash Nonfatal 8 (0.9) 8 (0.9) 1.02 (0.38-2.71) .99 differed significantly between the treat- Vascular death࿣ Major 7 (0.8) 4 (0.4) NA .36 ment groups (P=.02). Gastrointesti- Other 11 (1.2) 14 (1.5) 0.80 (0.36-1.76) .56 nal tract hemorrhage requiring premature discontinuation of study Secondary Outcome Event¶ medication (not shown in Table 4) was Any recurrent stroke 107 (11.9)# 86 (9.5) 1.28 (0.96-1.72) .10 more common among aspirin-treated Death (all causes) 45 (5.0) 40 (4.4) 1.15 (0.74-1.77) .56 patients but did not reach statistical sig- Vascular death 23 (2.5) 19 (2.1) 1.23 (0.66-2.29) .52 nificance (0.6% vs 0.2%; P=.45). Recurrent stroke or death (all causes) 138 (15.3) 117 (12.9) 1.22 (0.95-1.57) .14 Recurrent stroke, myocardial infarction, 145 (16.1) 125 (13.8) 1.21 (0.95-1.55) .17 or death (all causes) COMMENT Abbreviations: CI, confidence interval; HR, hazard ratio; NA, not available due to insufficient data for adequate estimation. Ticlopidine, a unique inhibitor of plate- *An HR higher than 1 indicates a higher rate of outcome events among those randomized to the ticlopidine treatment 27,28 group. let aggregation, was approved for †Based on the ␹2 test unless otherwise indicated. clinical use in the early 1990s by the US ‡Composite HR is 1.22 (95% CI, 0.94-1.57). §Based on Fisher exact test. Food and Drug Administration to re- ࿣Major refers to fatal ischemic or hemorrhagic stroke, fatal myocardial infarction, or sudden death. Other refers to pulmonary embolism and death ascribed to heart failure, visceral or limb infarction, or a vascular procedure or opera- duce the risk of fatal or nonfatal throm- tion. botic stroke in patients with stroke pre- ¶Values reflect any occurrence of the indicated event, even if the patient had already experienced a different outcome event. Therefore, these values will be greater than or equal to the corresponding primary outcome events. cursors and in patients who have had #Includes 1 patient who had a stroke after first experiencing a primary outcome of myocardial infarction during study. a completed thrombotic stroke. Two large clinical trials, TASS14 and the Ca- Figure 2. Time to Recurrent Stroke, Myocardial Infarction, or Vascular Death, and Time to nadian American Ticlopidine Study Recurrent Fatal or Nonfatal Stroke (CATS),29 showed efficacy for ticlopidine in stroke prevention. The CATS Recurrent Stroke, Myocardial Infarction, Recurrent Fatal or or Vascular Death Nonfatal Stroke intention-to-treat analysis showed a 0.20 RRR of 23.3% (P=.02) for the compos-

0.16 ite outcome of stroke, myocardial infarction, or vascular death favoring 0.12 Ticlopidine Ticlopidine ticlopidine over placebo. Reversible severe neutropenia and reversible se- 0.08 Aspirin Proportion Aspirin vere skin rash occurred in ticlopidine- 0.04 treated patients in about 1% of cases and Log Rank P = .12 Log Rank P = .08 diarrhea in 2% of cases. In TASS, the 0 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 RRR favoring ticlopidine over aspirin Days Days at 3 years for the primary end point of No. at Risk nonfatal stroke or death from any cause Ticlopidine 902 801735666 600 553 475 419 902 805740671 602 557 484 428 Aspirin 907 828764684 623 567 509 442 907 809765688 630 571 513 450 was 12% (95% confidence interval [CI], −2% to 26%) with a marginally signifi- Median follow-up was 710 days in the ticlopidine group and 716 days in the aspirin group. cant difference in primary events

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

(P=.048). In addition, there was an RRR ease risk profile and that aspirin might voring ticlopidine over aspirin.14 Fur- of 21% (95% CI, 4%-38%; P=.02) for prove to be an inferior platelet inhibi- thermore, clopidogrel, a compound in fatal or nonfatal stroke favoring ticlo- tor in these high-risk patients. In ret- the same drug class as ticlopidine, has pidine at 3 years, and ticlopidine was rospect, we are somewhat surprised by also been studied. The Clopidogrel ver- more effective than aspirin in both men our results because TASS showed a sus Aspirin in Patients at Risk of Ische- and women. Diarrhea (20%), skin rash modest reduction in the primary out- mic Events (CAPRIE) study showed a (14%), and severe but reversible neu- come of stroke or all-cause death fa- modest benefit with 75 mg/d of clopi- tropenia (Ͻ1%) occurred in patients receiving ticlopidine. Efficacy data from our trial of black Table 3. Recurrent Stroke Subtype and Severity by Treatment Group* Ticlopidine Aspirin patients with noncardioembolic ische- (n = 105) (n = 85) P Value mic stroke did not show a reduction of Stroke Subtype† the composite outcome of recurrent Large vessel atherosclerosis 19 (18.1) 12 (14.1) .48 stroke, myocardial infarction, or vas- Small vessel (lacunar) 38 (36.2) 40 (47.1) .12 cular death among ticlopidine-treated Other etiology or unknown 48 (45.7) 33 (38.8) .34 patients (P=.12 by log-rank test). The Stroke Severity‡ outcome of fatal or nonfatal stroke ap- (n = 97) (n = 77) proached a statistically significant dif- Barthel score, mean (SD) 14.60 (6.55) 15.35 (5.81) .53 ference (P=.08 by log-rank test) favor- Glasgow Outcome Assessment ing the aspirin treatment group. No or minimal disability 31 (32.0) 30 (39.0) .34 The blinded phase of our study was Moderate or severe disability 64 (66.0) 46 (59.7) .40 halted by the data and safety monitor- NIH Stroke Scale score, mean (SD)§ 9.2 (16.2) 7.5 (16.6) .10 ing board appointed by the National In- Abbreviation: NIH, National Institutes of Health. *Values expressed as number (percentage) unless otherwise indicated. stitutes of Health after the recruit- †Information on ischemic stroke subtype missing for 1 patient in each treatment group. ment and follow-up phases of the study ‡Data are missing for some patients due to fatal stroke or patient noncompliance for outcome examination, although source documents were available for adjudication of the outcome event. had been ongoing for about 6.5 years §For this comparison, there were 96 patients in the ticlopidine group and 74 in the aspirin group. because futility analyses indicated a less than 1% chance of ticlopidine being sig- Table 4. Patients Reporting a Serious Adverse Event* nificantly better than aspirin therapy in No. (%) of Patients the prevention of our primary outcome if the trial were to continue to Ticlopidine Aspirin P completion. These analyses also indi- Adverse Event (n = 902) (n = 907) RR (95% CI)† Value‡ cated a 40% to 50% likelihood of aspi- Body as a whole 30 (3.3) 42 (4.6) 1.41 (0.88-2.28) .16 Cardiovascular system 66 (7.3) 76 (8.4) 1.16 (0.82-1.63) .40 rin being significantly better than ticlo- Digestive system 38 (4.2) 43 (4.7) 1.10 (0.70-1.73) .59 pidine in reducing the risk of recurrent Diarrhea 3 (0.3) 2 (0.2) 0.66 (0.11-3.97) .69§ fatal or nonfatal stroke if the trial were Major gastrointestinal tract 4 (0.4) 8 (0.9) 2.00 (0.60-6.66) .39§ to continue to completion. The deci- hemorrhage sion of the data and safety monitoring Endocrine system 11 (1.2) 10 (1.1) 0.90 (0.38-2.14) .82 board to stop the study was based on Hemic and lymphatic system 38 (4.2) 29 (3.2) 0.75 (0.46-1.23) .25 the potential futility of ticlopidine use Neutropenia 31 (3.4) 20 (2.2) 0.66 (0.37-1.16) .12 for the primary study outcome end Thrombocytopenia 3 (0.3) 2 (0.2) 0.66 (0.11-3.97) .69§ point and the small likelihood, but po- Musculoskeletal system 17 (1.9) 11 (1.2) 0.64 (0.30-1.37) .25 tential for SAEs among ticlopidine- Nervous system 66 (7.3) 60 (6.6) 0.90 (0.62-1.29) .56 treated patients. Psychiatric system 10 (1.1) 5 (0.6) 0.49 (0.17-1.45) .19 Based on data from a subanalysis of Respiratory system 38 (4.2) 37 (4.1) 0.97 (0.61-1.54) .89 nonwhite patients in TASS,15 we be- Skin and appendages 15 (1.7) 15 (1.7) 0.99 (0.48-2.05) .99 lieved that there was a substantial like- Urogenital system 24 (2.7) 17 (1.9) 0.70 (0.37-1.31) .26 lihood that ticlopidine would be more Special senses 3 (0.3) 6 (0.7) 2.00 (0.50-8.00) .51§ effective than aspirin in reducing vas- Other bleeding 6 (0.7) 11 (1.2) 1.83 (0.68-4.98) .23 cular events among a high-risk stroke Total with Ն1 serious adverse event 270 (29.9) 262 (28.9) 0.95 (0.78-1.16) .63 population. We hypothesized that ticlo- Abbreviations: CI, confidence interval; RR, relative risk. *Defined as events that were fatal, life-threatening, permanently disabling, requiring hospitalization, prolonging hospi- pidine, a more global platelet inhibi- talization, or associated with congenital anomaly, cancer, or overdose, and were classified according to the scheme of the Trial of ORG 10172 in Acute Stroke Treatment (TOAST).21 tor, would be more effective in black †Relative risk of aspirin to ticlopidine. ischemic stroke patients with a sub- ‡Based on the ␹2 test. §Based on the Fisher exact test. stantial stroke and cardiovascular dis-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

dogrel compared with 325 mg/d of as- In an AAASPS exploratory study, we tiplying this estimate by a factor of 2, was pirin for the reduction of recurrent found that voluntary withdrawal oc- 16.1%; a figure quite similar to that of stroke, myocardial infarction, or vas- curred for such reasons as distrust of the WARSS. Our study was not designed to cular death among patients who en- medical establishment, fear of continu- determine if the lower-than-expected tered with either recent ischemic stroke, ation in a blinded trial, and pressure from rates for stroke were due to baseline recent myocardial infarction, or symp- family members, primary caregivers, or characteristics in the sample, better con- tomatic peripheral arterial disease.30 In primary care physicians.20 trol of risk factors, or more widespread a CAPRIE subanalysis of 6431 pa- We believe that our drop-out rate did use of concomitant therapies. We specu- tients with stroke, there was only a non- not limit our ability to detect a treat- late, however, that this lower-than- statistically significant reduction of is- ment effect because we anticipated this expected rate could be related to better chemic stroke, myocardial infarction, level of attrition and accounted for it control of risk factors as we noted in- or vascular death favoring clopidogrel in our sample size calculation. In ad- creasing use of cholesterol-lowering (RRR, 7.3%; 95% CI, −5.7% to 18.7%; dition, we conducted both intention- agents during the study period. P=.26). However, the CAPRIE sub- to-treat and on-treatment analyses and The SAEs with ticlopidine have been analysis of stroke patients30 and the did not find a statistically significant dif- well described.14,29,33-36 Our SAE data TASS subanalysis of nonwhites15 were ference between treatment groups for show that slightly more occurred not adequately powered statistically to the primary event outcome cluster. among ticlopidine-treated patients be conclusive. Therefore, it is unlikely that our find- (29.9%) overall than aspirin-treated pa- We are uncertain why ticlopidine was ings are due to a switch to open-label tients (28.9%). In addition, there were not superior to aspirin in reducing ma- aspirin therapy during the course of the slightly more clinically reported SAEs jor vascular events in our study pa- trial thereby diluting the ticlopidine of diarrhea in the ticlopidine-treated tients. We used a lower aspirin dose treatment group. We note that the num- patients (0.3% vs 0.2%), but more clini- than in TASS,14,15 but we can only ber of primary outcomes was less than cally reported SAEs of major gastroin- speculate whether this or some other the projected number of 306, which was testinal tract bleeding among aspirin- biological difference in response to due to premature stoppage of the study. treated patients (0.9% vs 0.4%). therapy might explain our findings. However, had the study continued to We defined serious neutropenia as a Overall, there is a relative paucity of sci- completion, we estimated the occur- laboratory-determined absolute neutro- entific information about biological dif- rence of at least 310 outcome events. phil count of less than 1000/mm3 and ferences in drug response by race or eth- Thus, the nonborderline nature of the serious thrombocytopenia as a labora- nic group.31 We are not aware of such study results, a futility analysis that tory-determined platelet count of less published information for aspirin and showed a less than 1% chance of ticlo- than 100000/mm3. In either case, once ticlopidine in blacks. pidine performing better than aspirin, the absolute neutrophil count or plate- The drop-out rate in our study was and loss of information that did not oc- let count reached the critical prespeci- generally higher than most recurrent cur substantially more in one treat- fied level, the blinded phase study medi- stroke prevention trials. In the CAPRIE ment group than the other, makes it cation was discontinued. We observed study, for example, the number of study unlikely that loss of information sig- a higher percentage of laboratory- participants withdrawn or lost to fol- nificantly affected our results. determined serious neutropenia (3.4% low-up was 22 (0.2%) of 9553 among Our recurrent primary event rates vs 2.2%) among ticlopidine-treated pa- those receiving clopidogrel and 20 were lower than expected. For ex- tients and about the same percentage of (0.2%) of 9546 among those receiving ample, the expected 2-year event rate laboratory-determined serious throm- aspirin.30 In our study, the correspond- was projected to be 25% in the aspirin bocytopenia (0.3% vs 0.2%) among the ing figures for lost to follow-up or vol- group but was observed to be 16.3%. treatment groups. All cases of neutro- untary withdrawal were 15.2% in the Our rate, however, for stroke and death penia were reversible, but one case of ticlopidine treatment group and 13.3% is similar to that of the recently com- thrombocytopenia resulted in possible for those receiving aspirin. Most of these pleted Warfarin Aspirin Recurrent thrombotic thrombocytopenic pur- were due to voluntary withdrawal, Stroke Study (WARSS).32 WARSS had a pura in a patient who recovered after which we defined as the patient decid- preponderance of lacunar infarctions at treatment with plasmapheresis. ing to discontinue participation unre- entry (56%), similar to AAASPS (67.5%), Our data support the prior TASS sub- lated to an AE but rather to personal and the rate of recurrent stroke and analysis findings15 suggesting that some preference. We believe that our higher death in WARSS was 16.0% at 2 years key AEs might be less frequent in ticlo- drop-out rate can be attributed to study for the aspirin group. The correspond- pidine-treated nonwhites. We ob- participants of generally lower socio- ing rate for our aspirin group, derived served lower percentages of diarrhea economic status, who have tradition- by taking the 1-year rate from our aspi- (8.4% in ticlopidine-treated patients vs ally had less access to medical care, and rin study participants because most had 9.0% in TASS subanalysis vs 20.4% in less involvement in clinical trials.8,9,12,20 completed 1 year of follow-up, and mul- TASS overall14) and rash (5.9% vs 9.0%

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

in TASS subanalysis vs 11.9% in TASS tration of these other agents in black of enrollees): Cook County Hospital, Chicago, Ill (n=252); principal investigator: Michael Kelly, MD; overall). patients will vary based on patient and coprincipal investigator: Lafayette Singleton, MD; study Based on data from the IMS Health’s physician preference. coordinators: Ruby Freeman, RN, Elana Tabb. Rush- Presbyterian St Luke’s Medical Center, Chicago, Ill National Prescription Audit Plus and Prior systematic reviews and guide- (n=153); principal investigators: Michael A. Sloan, MD, National Disease and Therapeutic In- line statements suggest that aspirin is Michael Schneck, MD; coprincipal investigators: Gwen- dex of Dispensed Prescriptions for non- of benefit in a wide range of patients dolyn Ford-Lynch, MD, Timothy Lukovits, MD, Sean Ruland, MD; study coordinator: Patricia Samuels, Ly- aspirin antiplatelet agents used specifi- with suspected acute ischemic stroke esther Walker. Wayne State University, Detroit, Mich cally or “customized” for stroke or to reduce the risk of early stroke recur- (n=105); principal investigator: Seemant Chatur- 38 vedi, MD; coprincipal investigators: Ann Guyot, MD, transient ischemic attack patients in rence and as an initial choice of Nishith Joshi, MD; study coordinators: Ellen St. Pierre, 2002, generic and branded ticlopidine therapy for recurrent stroke preven- RN, Bryan Bertasio, RN, Lorraine Femino, RN, Felicia was third in market share with an es- tion.39-41 Before our study was com- Mada, RN. Washington University, St Louis, Mo (n=91); principal investigators: Chung Hsu, MD, PhD, timated 3.2% of dispensed oral anti- pleted, thienopyridine derivatives (ticlo- Abdullah M. Nassief, MD; coprincipal investigator: John platelet prescriptions; dipyridamole was pidine and clopidogrel) appeared to be Choi, MD, PhD; study coordinators: Gerry Banet, RN, Jackie Epps-Wilbanks. Medical College of Virginia/ second at 17.9%; and clopidogrel was modestly more effective than aspirin in Virginia Commonwealth University, Richmond first at 78.9% (personal communica- the prevention of serious vascular (n=82); principal investigator: Warren L. Felton III, MD; 42 coprincipal investigator: John R. Taylor, MD; study co- tion from David Milbauer, Marketing events in high-risk patients. Aspirin ordinators: Tammy Anderson,CCRC, Kim Gitter, RN. Research, Boehringer Ingelheim Phar- is much less expensive than other ma- Henry Ford Hospital, Detroit, Mich (n=62); principal maceuticals Inc. Data extracted April jor antiplatelet agents, is readily avail- investigator: Panayiotis Mitsias, MD; coprincipal investigators: Steven Levine, MD, Nikolaos Papamit- 2003). Aspirin use is not captured in able, easy to use, and relatively safe. sakis, MD, Brian Silver, MD; study coordinators: Jac- the IMS Health Index of Dispensed Pre- Head-to-head comparison with other queline Reuther, RN, Joyce Hargrow. Indiana University, Indianapolis (n=53); principal investiga- scriptions because it is an over-the- agents indicates that it may be diffi- tor: Jose Biller, MD; coprincipal investigators: Askiel counter medication. With an esti- cult to outperform aspirin as a stroke Bruno, MD, Linda Williams, MD, J. D. Fleck, MD, Al- fredo Lopez, MD, William Jones, MD; study coordi- mated 4 million stroke survivors and prevention therapy in some noncardio- nators: Carol Kempf, RN, Alison Sears, RN, Linda Chad- 31,32,43,44 between 600000 and 750000 strokes embolic ischemic stroke patients. wick, RN. Michael Reese Hospital, Chicago, Ill (n=50); occurring in the United States annu- Our data call into question the su- principal investigator: Lawrence Scibilia, MD; study coordinator: Robynn Whitehead, BS. St Louis Univer- ally, of which about 85% are ischemic, periority of the thienopyridine ticlopi- sity, St Louis, Mo (n=50); principal investigators: Sal- up to 108800 persons in the United dine in black noncardioembolic ische- vador Cruz-Flores, MD, Daryl Thompson, MD; coprincipal investigators: Enrique Leira, MD, Rodney States could be taking ticlopidine and mic stroke patients, and suggest that Leacock, MD, John Selhorst, MD; study coordinator: up to 20400 persons could receive ticlo- ticlopidine is unlikely to be superior to Eve Holzemer, RN, MSN. Hines Veterans Affairs Hos- pidine as an initial stroke prevention aspirin for prevention of recurrent pital, Hines, Ill (n=49); principal investigator: Sudha Gupta, MD; study coordinators: Beth Scott, RN, Dawn therapy each year. Our findings do not stroke and major vascular events in Wetherald, RN. University of Maryland, Baltimore support the use of ticlopidine as a first- these patients. Furthermore, ticlopi- (n=45); principal investigator: Marcella A. Wozniak, MD, PhD; coprincipal investigator: Steven J. Kittner, line agent for blacks for recurrent stroke dine may have a less favorable and po- MD, MPH; study coordinators: Mary J. Sparks, RN, prevention, provide substantial data to tentially SAE profile. Therefore, aspi- BSN, Mary Jo Bankard, RN, BSN, Nancy K. Zappala, RN, CDE. University of Chicago, Chicago, Ill (n=42); challenge the subanalysis of TASS that rin is a reasonable first choice principal investigator: James Brorson, MD; coprinci- suggested benefits of ticlopidine in re- prevention agent in aspirin-tolerant pal investigator: Avertano Noronha, MD; study co- current stroke prevention in non- black patients with noncardioembolic ordinator: Christi Kordeck, RN. Humana (Advocate) 15 Health System, Chicago, Ill (n=40); principal inves- white patients, and provide new data ischemic stroke. tigator: Jan Clark, MD; coprincipal investigator: Sha- to challenge current guidelines on sec- hida Ahmad, MD; study coordinators: Debra Hollins, ondary stroke prevention. Author Contributions: As director of data manage- RN, Robynn Whitehead, BS. Mount Sinai Hospital, ment operations for the AAASPS, Dr Richardson had Chicago, Ill (n=40); principal investigator: Bindu De- The implications and relevance of our access to all of the data in the study and takes re- sai, MD, Aurora Pajeau, MD, MPH; study coordina- data as they affect other antiplatelet sponsibility for the integrity of the data and the ac- tor: Robynn Whitehead, BS. University of Illinois, Chi- curacy of the data analyses. cago (n=40); principal investigator: Cathy Helgason, agents for recurrent stroke prevention Study concept and design: Gorelick, Richardson, Kelly, MD, MLA; study coordinator: Elizabeth Brennan, RN. in blacks remain speculative as we did Harris, Leurgans. Louisiana State University, Shreveport (n=34); prin- Acquisition of data: Gorelick, Richardson, Kelly, Ruland, cipal investigator: Alireza Minagar, MD, Aurora Pajeau, not directly compare our study agents Hung, Harris, Kittner. MD, MPH; coprincipal investigator: Roger E. Kelley; with clopidogrel or the combination as- Analysis and interpretation of data: Gorelick, study coordinator: Yan Wang, MD. Mercy Hospital, Richardson, Kelly, Hung, Leurgans. Chicago, Ill (n=33); principal investigator: Jeffery Kra- pirin plus extended-release dipyridam- Drafting of the manuscript: Gorelick, Richardson, Kelly, mer, MD [previous principal investigator: Alan Hir- ole. Furthermore, the clinical trials that Harris. sch, MD]; study coordinator: Robynn Whitehead, BS tested these other agents had rela- Critical revision of the manuscript for important in- [previous study coordinator: Charlene Bermele, RN]. Drew University Los Angeles Calif (n 31) 30,37 tellectual content: Gorelick, Richardson, Kelly, Ruland, = ; principal tively few or no black enrollees. For Hung, Kittner, Leurgans. investigator: George Locke, MD; coprincipal investi- clinicians who remain unconvinced Statistical expertise: Richardson, Hung, Leurgans. gator: Lowell Nelson, MD, PhD; study coordinators: Obtained funding: Gorelick, Richardson. Ardientes Sisson, RN, Marcia Montenegro, RN. Howard about the efficacy or safety of these Administrative, technical, or material support: Gorelick, University, Washington, DC (n=31); principal inves- other agents, aspirin could be the pre- Richardson, Kelly, Ruland, Harris. tigator: Annapurni Jayam Trouth, MD; coprincipal in- Study supervision: Gorelick, Richardson, Harris, Kittner. vestigator: Chitra Chari, MD; study coordinators: Shub- ferred agent for recurrent stroke pre- AAASPS Sites (listed by recruitment for sites that ran- hangi Chumble, Kolla Ratnavalli, Sonia Thomas, vention in blacks. However, adminis- domized Ն1 enrollee, in descending order of number Sudhamayi Mokalapalli, Reena Bommasani, Eliza-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

beth Reena Francis, Mohan Kishore Garikaparti, Ad- Yousef Mohammad, MD; study coordinators: Greta (chair), DeJuran Richardson, PhD (cochair), Jose Biller, egboyega Abdulkadir. St Bernard Hospital, Chicago, Keys, RN, BSN, Cynthia Roach, RN, BSN. Albert Ein- MD, Seemant Chaturvedi, MD, Sudha Gupta, MD, Ill (n=30); principal investigator: Elsie Walker, MD; stein College of Medicine, New York, NY (n=7); prin- Steven Kittner, MD, MPH, Sue Leurgans, PhD. study coordinator: Deborah Ramsey, RN. Vanderbilt cipal investigator: Steven Sparr, MD; coprincipal in- Adjudication Panel: Michael A. Kelly, MD (chair), University, Nashville, Tenn (n=30); principal inves- vestigators: Daniel Rosenbaum, MD, Paul M. Katz, MD; Gwendolyn Ford-Lynch, MD, Cathy Helgason, MD, tigator: Howard Kirshner, MD; coprincipal investiga- study coordinators: J. P. Noonan, Quinn Alfinios– MLA, Philip Liebson, MD, Sean Ruland, DO, Michael tors: Jeff Harris, MD; study coordinator: Diane Klein, Renia, RN. Mercy Hospital, Chicago, Ill (n=7); prin- Schneck, MD, Michael Swiontoniowski, MD. RN. Lankenau Hospital, Lankenau, Pa (n=29); principal investigator: Alan Hirsch, MD; study coordina- Data Safety and Monitoring Board: Lloyd Chamb- cipal investigator: Edgar J. Kenton III, MD; coprinci- tor: Charlene Bermele, RN. Medical College of Georgia, less, PhD (chair), Edward Cooper, MD, Edward Feld- pal investigators: Gary H. Friday, MD, MPH, Milton Augusta (n=6); principal investigator: Fenwick T. Ni- mann, MD, Percy Karanjia, MD, Ann Lowe, MD, John Alter, MD, PhD; study coordinators: Angela V. Whit- chols, III, MD; study coordinator: Mary T. Sahm, BS. Marler, MD (ex-officio, National Institutes of Health tington-Smith, Charlotte A. Baker, RN. Doctor’s Hos- Evanston Hospital, Evanston, Ill (n=6); principal in- [NIH] Project Officer), Peter Gilbert, ScM (ex-officio, pital of Hyde Park, Chicago, Ill (n=29); principal investigator: Daniel Homer, MD; study coordinator: Bar- NIH), Barbara Radziszewska, PhD, MPH (ex-officio, vestigator: Elsie Walker, MD; study coordinators: bara Small, MBA, RN. Morehouse School of Medi- NIH). Minnie Al-Mosawi, Arzola Winters. University of Vir- cine, Atlanta, Ga (n=5); principal investigator: Patrick Central Laboratory: LabCorp Inc, Raritan, NJ. ginia, Charlottesville (n=28); principal investigator: Griffith, MD; study coordinator: Gail Moss, MA. Tu- Financial Management Center: Rush Medical Cen- Karen Johnston, MD; coprincipal investigator: Brad lane University, New Orleans, La (n=5); principal in- ter, Chicago, Ill. Worrall, MD; study coordinators: Gail Kongable, RN, vestigator: Leon Weisberg, MD; coprincipal investi- Clinical Management Center: Philip B. Gorelick, MD, MSN, Martha Davis, RN. Washington VA, Washing- gator: Dan Rodriguez, MD; study coordinators: Bee MPH (program director), Yvonne Harris, MPA, CCRA ton, DC (n=27); principal investigator: Mohammad Pollock, Dawn Alonzo, RN. University of Medicine and (project manager), Onyinye Erondu, RN, Eugenia Yaseen, MD; study coordinator: Ellyce Green. Uni- Dentistry, Newark, NJ (n=5); principal investigator: Gencheva, MD, Karen Keyes, Kelia Leach, RN, Par- versity of Texas, Dallas (n=24); principal investiga- Andrea Hidalgo, MD; study coordinator: Doreen rish Moore, Greg Podraza, RN, Dorothy Robinson, LPN, tor: Mark Johnson, MD; coprincipal investigators: Hal Monks, RN. Neurocenter, Gary, Ind (n=5); principal Sean Ruland, DO, Patricia Samuels, Michael J. Schneck, Unwin, MD, Dion Graybeal, MD; study coordinator: investigator: Jackie Carter, MD; study coordinator: Ver- MD, Michael A. Sloan, MD, Sherri Volious. Jennifer Stanford, RN, MSN, CCRC. West Suburban nita Reddix, RN. East Bay Neurology, Berkeley, Calif Data Management Center: DeJuran Richardson, PhD Hospital, Oak Park, Ill (n=24); principal investigator: (n=4); principal investigator: Brian Richardson, MD; (director), Merryl Billingsley (project manager), Shande John Jacobson, MD; coprincipal investigators: Madhu study coordinator: Dimitri Salkovski, MD. Rochester Chen, PhD, Lisa Crim, John Cursio, MS, Elena Hung, Soni, MD, Yvonne Curran, MD; study coordinator: General Hospital, Rochester, NY (n=4); principal in- MS, Glenda Kravitz, MS, Sue Leurgans, PhD (codi- Anna Treinkman, RN. St Francis Hospital, Blue Is- vestigator: Joshua Hollander, MD; coprincipal inves- rector), Crystal Newkirk, Rema Raman, PhD, Vidas land, Ill (n=23); principal investigator: Barry J. Riskin, tigators: Gerald W. Honch, MD; Cheryl Weber, MS, Simkus. MD; study coordinator: Katie Kindt, DDS. Georgia Neu- RNC, Ann M. Sass, RN. Pennsylvania Hospital, Phila- Funding/Support: This work was supported by grant rology Associates, Atlanta (n=22); principal investi- delphia (n=4); principal investigators: Dara Jamie- RO1 NS33430 from the National Institute of Neuro- gator: Mark Harris, MD, study coordinators: Nancy son, MD, Brett Skolnick, PhD; study coordinator: Jen- logical Disorders and Stroke. Medications and place- Green, RN, Demaree Trammel, MD. University of Cin- nifer Nisivoccia, BS. Trinity Hospital,Chicago, Ill (n=4); bos were supplied by Roche Laboratories and Bayer. cinnati, Cincinnati, Ohio (n=21); principal investiga- principal investigator: Shahida Ahmad, MD; study co- Roche Laboratories and Bayer had no role beyond sup- tor: Joseph P. Broderick, MD; coprincipal investiga- ordinator: Angela Britton. Albert Einstein College of plying study medications and placebos. Acknowledgment: We acknowledge technical assis- tors: Daniel Woo, MD, Brett Kissela, MD; study Medicine, Philadelphia, Pa (n=3); principal investi- tance and helpful suggestions from Rema Raman, PhD, coordinators: Colleen Rogge, RN, Kirainpal Sangha, gator: Anne Vigderman, MD; study coordinator: Con- Shande Chen, PhD, and John Cursio, MS, and help- PharmD. West Los Angeles Veterans Affairs Hospi- nie Borschell, RN. GV Montgomery VA Medical Cen- ful suggestions from Lloyd Chambless, PhD, Ann Lowe, tal, Los Angeles, Calif (n=20); principal investiga- ter, Jackson, Miss (n=2); principal investigators: Ethel MD, Edward Feldmann, MD, Peter Gilbert, ScD, Ed- tors: Stanley Cohen, MD, Roi Ann Wallis, MD; coprin- Rose, MD, Robert Herndon, MD; coprincipal inves- ward Cooper, MD, and Percy Karanjia, MD. cipal investigator: Hamid Salari, MD; study tigator: Eric Undesser, MD, PhD; study coordinator: coordinators: Nicole Birmingham, Dana Carmody, RN, Jo Ann Chinn, RN, MSN. Olympia Fields Hospital, Jonathan Wong, Kimberly Pannizon, JD. Medical Col- Olympia Fields, Ill (n=2); principal investigator: Mi- lege of Ohio, Toledo (n=18); principal investigators: chael Shaenboen, DO; study coordinator: Mary Cof- REFERENCES Gretchen Tietjen, MD, Nancy Futrell, MD; coprinci- fey, RN. St Elizabeth Hospital, Chicago, Ill (n=2); prin- 1. Sacco RL, Boden-Albala B, Gan R, et al. Stroke in- pal investigator: Rima Dafer, MD; study coordinator: cipal investigator: David Shenker, MD; study cidence among white, black and Hispanic residents of Andrea Korsnack, RN, CNRN. University of Texas, coordinator: Jana Dillon, RN. Northwestern Memo- an urban community: the Northern Manhattan Stroke Houston (n=18); principal investigators: Lewis B. Mor- rial Hospital, Chicago, Ill (n=1); principal investiga- Study. Am J Epidemiol. 1998;147:259-268. genstern, MD, James Grotta, MD; study coordinator: Jeff Curtin, DO; study coordinators: Laura Prz- 2. Broderick J, Brott T, Kothari R, et al. The Greater tor: Sandi Shaw, RN. Georgetown University, Wash- epiorka, RN. Medical University of South Carolina, Cincinnati/Northern Kentucky Stroke Study: prelimi- ington, DC (n=17); principal investigators: Stanley Charleston (n=1); principal investigator: Timothy nary first-ever and total incidence rates of stroke among Cohan, MD, Manuel Yepes, MD; coprincipal investi- Carter, MD; study coordinator: Elaine Tateman, RN. blacks. Stroke. 1998;29:415-421. gator: Heather Fitter, MD; study coordinators: Bar- Neurology Associates, Harvey, Ill (n=1); principal in- 3. Broderick JP, Brott T, Tomsick T, Huster G, Miller iatu Mustapha, BSN, Kathleen Johnson, NP, Brigid Rey- vestigator: Tonya Fuller, MD. South Suburban Neu- R. The risk of subarachnoid and intracerebral hemor- nolds, NP. Roseland Hospital, Chicago, Ill (n=14); rology Associates, Harvey, Ill (n=1); principal inves- rhage in blacks as compared with whites. N Engl J Med. principal investigator: Elsie Walker, MD; study coor- tigator: Armita Bijari, MD. 1992;326:733-736. dinator: Robynn Whitehead, BS. Buffalo General, Buf- Operations Committee: Philip B. Gorelick, MD, MPH 4. Morgenstern LB, Spears WD. A triethnic eompari- falo, NY (n=12); principal investigators: Richard K. T. (chair), DeJuran Richardson, PhD (cochair), James Bror- son of intracerbral hemorrhage mortality in Texas. Ann Chan, MD, Patrick Pullicino, MD, PhD; study coordi- son, MD, Sudha Gupta, MD, Cathy Helgason, MD, Neurol. 1997;42:919-923. nator: Sandra Starr, RN. North General, Harlem, NY Chung Hsu, MD, PhD, Michael A. Kelly, MD, Mi- 5. Gillum RF. Stroke mortality in blacks: disturbing (n=12); principal investigator: Jesse Weinberger, MD; chael Sloan, MD, Michael Swiontoniowski, MD, trends. Stroke. 1999;30:1711-1715. study coordinators: Josephine Britanico, RN, MSN, Yvonne Harris, MPA (ex-officio). 6. Giles WH, Kittner SJ, Hebel JR, Losconczy KG, Sher- Jolene Connors, RN, Lafayette Trawick, BS. Little Com- Scientific Advisory Committee: Toni Miles, MD, PhD win RW. Determinants of black-white differences in the pany of Mary Hospital, Evergreen Park, Ill (n=12); prin- (chair), Milton Alter, MD, PhD, Laurel Beckett, PhD, risk of cerebral infarction: the National Health and Nutri- cipal investigator: Michael Schwartz, MD; study co- Louis R. Caplan, MD, Mark Dyken, MD, J. Donald Eas- tion Examination Survey Epidemiologic Follow-up Study. ordinator: Linda Howell, RN. Jewish Hospital, ton, MD, Denis Evans, MD, Robert Woolson, PhD. Arch Intern Med. 1995;155:1319-1324. Louisville, Ky (n=11); principal investigators: Roy J. Community Advisory Committee: Ethel Alexander, 7. Friday G, Lai SM, Alter MA, et al. Stroke in the Le- Meckler, MD, Lynn T. Simon, MD; study coordina- Minnie Al-Mosawi, Edsel Hudson, MD, Helen Hud- high Valley: racial/ethnic differences. Neurology. 1989; tor: Deborah Lockridge, RN, CNRN, CCRC. Emory Uni- son, Ron O’Neal, Esq, Geneva Scott, Lawrence Tay- 39:1165-1168. versity, Atlanta, Ga (n=11); principal investigator: Mi- lor, Arnold Turner, MD, Wayne Williamson, MD. 8. Gorelick PB. Cerebrovascular disease in African chael Frankel, MD; study coordinator: Janet Braimah, Inhouse Safety Committee: Maynard Cohen, MD, Americans. Stroke. 1998;29:2656-2664. RN. Oklahoma City, Okla (n=10); principal investi- PhD (cochair), DeJuran Richardson, PhD (cochair), 9. Gornick ME, Eggers PW, Reilly TW, et al. Effects gators: James Couch, MD, PhD, Mark Fisher, MD; Henri Frischer, MD, PhD, Philip Liebson, MD, Greg of race and income on mortality and use of services study coordinator: Christina Leviston. University of Podraza, RN, Michael Sloan, MD, Philip B. Gorelick, among Medicare beneficiaries. N Engl J Med. 1996; Mississippi, Jackson (n=10); principal investigators: MD, MPH (ex-officio), Yvonne Harris, MPA 335:791-799. David Lee Gordon, MD, Nathaniel Lawson, MD; (ex-officio). 10. Carlise DM, Leake BD, Shaparo MF. Racial and coprincipal investigators: Patrick S. Reynolds, MD, Publications Committee: Philip B. Gorelick, MD, MPH ethnic disparities in the use of cardiovascular proce-

Downloaded From: https://jamanetwork.com/ on 10/01/2021 PREVENTION OF RECURRENT STROKE IN BLACK PATIENTS

dures: associations with type of health insurance. Am 10172 in Acute Stroke Treatment (TOAST) Investi- 34. Bennett CL, Weinberg PD, Rozenberg-Ben-Dror J Public Health. 1997;87:263-267. gators. Low molecular weight heparinoid, ORG 10172 K, et al. Thrombotic thrombocytopenia purpura as- 11. Horner RD, Oddone EZ, Matchar DB. Theories ex- (danaparoid), and outcome after acute ischemic stroke: sociated with ticlopidine: a review of 60 cases. Ann plaining racial differences in the utilization of diag- a randomized controlled trial. JAMA. 1998;279:1265- Intern Med. 1998;128:541-544. nostic and therapeutic procedures for cerebrovascu- 1272. 35. Bennett CL, Kiss JE, Weinberg PD, et al. Throm- lar disease. Milbank Q. 1995;73:443-462. 22. Barnett HJM, Eliasziw M, Meldrum HE. Drugs and botic thrombocytopenia purpura after stenting and 12. Harris Y, Gorelick PB, Samuels P, Bempong I. Why surgery in the prevention of ischemic stroke. N Engl J ticlopidine. Lancet. 1998;352:1036-1037. African Americans may not be participating in clinical Med. 1995;332:238-248. 36. Steinhubl SR, Tan WA, Foody JM, Topol EJ, for trials. J Natl Med Assoc. 1996;88:630-634. 23. Krischer JP, Hurley C, Mohan P, et al. An auto- the EPISTENT Investigators. Incidence and clinical 13. Ford Lynch G, Leurgans SL, Raman R, Barboi A, mated patient registration and treatment randomiza- course of thrombotic thrombocytopenia purpura due Gorelick PB. A comparison of stroke risk factors in pa- tion system for multi-center trials. Control Clin Trials. to ticlopidine following coronary stenting. JAMA. 1999; tients enrolled in stroke prevention trials. J Natl Med 1991;12:367-377. 281:806-810. Assoc. 2001;93:79-86. 24. Kim K, DeMets D. Group sequential monitoring 37. Diener HC, Cunha L, Forbes C, Sivenius J, Smets 14. Hass WK, Easton JD, Adams HP, et al. A random- in clinical trials. JAmStatAssoc.1987;74:149-154. P, Lowenthal A. European Stroke Prevention Study 2: ized trial comparing ticlopidine hydrochloride with as- 25. Richardson D, Leurgans S. Sample size justifica- dipyridamole and acetylsalicylic acid in the second- pirin for the prevention of stroke in high-risk pa- tion in phase III/IV clinical trials. Neuroepidemiol- ary prevention of stroke. J Neurol Sci. 1996;143:1- tients. N Engl J Med. 1989;321:501-507. ogy. 1998;17:63-66. 13. 15. Weisberg LA, for the Ticlopidine Aspirin Stroke 26. Cox DR. Regression models and life tables. JR 38. Chen ZM, Sandercock P, Pan HC, et al. Indica- Study Group. The efficacy and safety of ticlopidine and Stat Soc. 1972;34:187-220. tions for early aspirin use in acute ischemic stroke: a aspirin in non-whites: analysis of a patient subgroup 27. Teitelbaum P. Pharmacodynamics and pharma- combined analysis of 40,000 randomized patients from from the Ticlopidine Aspirin Stroke Study. Neurol- cokinetics of ticlopidine. In: Hass WK, Easton JD, eds. the Chinese Acute Stroke Trial and the International ogy. 1993;43:27-31. Ticlopidine, Platelets and Vascular Disease. New York, Stroke Trial. Stroke. 2000;31:1240-1249. 16. Gorelick PB, Leurgans S, Richardson D, Harris Y, NY: Springer-Verlag; 1993:27-40. 39. Antithrombotic Trialists’ Collaboration. Collabo- Billingsley M, and the AAASPS Investigators. African 28. Sharis PJ, Cannon CP, Loscalzo J. The antiplate- rative meta-analysis of randomized trials of antiplate- American Antiplatelet Stroke Prevention Study: clini- let effects of ticlopidine and clopidogrel. Ann Intern let therapy for prevention of death, myocardial in- cal trial design. J Stroke Cerebrovascular Dis. 1998; Med. 1998;129:394-405. farction, and stroke in high risk patients. BMJ. 2002; 7:426-434. 29. Gent M, Blakely JA, Easton JD, et al. The Cana- 324:71-86. 17. Begg C, Cho M, Eastwood S, et al. Improving the dian American Ticlopidine Study (CATS) in thrombo- 40. Albers GW, Amarenco P, Easton JD, Sacco RL, Teal quality of reporting of randomized controlled trials: embolic stroke. Lancet. 1989;1:1215-1220. P. Antithrombotic and thrombolytic therapy for is- the CONSORT statement. JAMA. 1996;276:637- 30. CAPRIE Steering Committee. A randomized, chemic stroke. Chest. 2001;119:300S-320S. 639. blinded, trial of Clopidogrel versus Aspirin in Patients 41. Wolf PA, Clagett GP, Easton JD, et al. Prevent- 18. Rennie D. How to report randomized controlled at Risk of Ischemic Events (CAPRIE). Lancet. 1996; ing ischemic stroke in patients with prior stroke and trials: the CONSORT statement. JAMA. 1996;276: 348:1329-1339. transient ischemic attack: a statement for healthcare 649. 31. Burroughs VJ, Maxey RW, Levy RA. Racial and professionals from the Stroke Council of the Ameri- 19. Gorelick PB, Richardson D, Hudson E, et al. Es- ethnic differences in response to medicines: towards can Heart Association. Stroke. 1999;30:1991-1994. tablishing a community network for recruitment of Af- individualized pharmaceutical treatment. J Natl Med 42. Hankey GJ, Sudlow CLM, Dunbabin DW. Thieno- rican Americans into a clinical trial: the African Ameri- Assoc. 2002;94(suppl):1-26. pyridines or aspirin to prevent stroke and other seri- can Antiplatelet Stroke Prevention Study (AAASPS) 32. Mohr JP, Thompson JLP, Lazar RM, et al. A com- ous vascular events in patients at high risk of vascular experience. J Natl Med Assoc. 1996;88:701-704. parison of warfarin and aspirin for the prevention of disease? a systematic review of the evidence from ran- 20. Gorelick PB, Harris Y, Burnett B, et al. The recruit- recurrent ischemic stroke. N Engl J Med. 2001;345: domized trials. Stroke. 2000;31:1779-1784. ment triangle: reasons why African Americans enroll, 1444-1451. 43. Warlow C. Aspirin should be the first-line anti- refuse to enroll, or voluntarily withdraw from a clini- 33. Antiplatelet Trialists’ Collaboration. Collaborative platelet therapy in the secondary prevention of stroke. cal trial: an interim report from the African American overview of randomized trials of antiplatelet therapy, I: Stroke. 2002;33:2137-2138. Antiplatelet Stroke Prevention Study (AAASPS). J Natl prevention of death, myocardial infarction, and stroke 44. Powers WJ. Oral anticoagulation therapy for the Med Assoc. 1998;90:141-145. by prolonged antiplatelet therapy in various categories prevention of stroke. N Engl J Med. 2001;345:1493- 21. The Publications Committee for the Trial of ORG of patients. BMJ. 1994;308:81-106. 1495.

I have taken all knowledge to be my province. —Francis Bacon (1561-1626)

Downloaded From: https://jamanetwork.com/ on 10/01/2021