Escolar Documentos
Profissional Documentos
Cultura Documentos
Background—The significance of thrombocytopenia in patients experiencing an acute coronary syndrome (ACS) has not
been examined systematically. We evaluated this condition in a large non–ST-elevation ACS clinical trial, with
particular interest paid to its correlation with clinical outcomes.
Methods and Results—Patients presenting without persistent ST elevation during an ACS were randomized to receive a
double-blind infusion of the platelet glycoprotein (GP) IIb/IIIa inhibitor eptifibatide or placebo in addition to other
standard therapies including heparin and aspirin. The primary end point was death/nonfatal myocardial infarction (MI)
at 30 days, whereas bleeding and stroke were the main safety outcomes. Thrombocytopenia (nadir platelet count
,1003109/L or ,50% of baseline) occurred in 7.0% of enrolled patients. The time to onset was a median of 4 days
in both treatment arms. Patients with thrombocytopenia were older, weighed less, were more likely nonwhite, and had
more cardiac risk factors. These patients experienced significantly more bleeding events: they were more than twice as
likely to experience moderate/severe bleeding after adjustment for confounders. Univariably, ischemic events (stroke,
MI, and death) occurred significantly (P,0.001) more frequently in patients with thrombocytopenia; multivariable
regression modeling preserved this association with death/nonfatal MI at 30 days. Neither the use of heparin or
eptifibatide was found to independently increase thrombocytopenic risk.
Conclusions—Although causality between thrombocytopenia and adverse clinical events could not be established
definitively, thrombocytopenia was highly correlated with both bleeding and ischemic events, and the presence of this
condition identified a more-at-risk patient population. (Circulation. 1999;99:2892-2900.)
Key Words: platelets n eptifibatide n glycoproteins n angina n myocardial infarction n coronary artery disease
Received July 22, 1998; revision received March 2, 1999; accepted March 29, 1999.
From the Duke Clinical Research Institute (M.W.M., S.D.B., R.S., R.T., L.G.B., R.M.C., R.A.H.), Durham, NC; Methodist Hospital (N.S.K.), Houston,
Tex; Cleveland Clinic Foundation (A.M.L., E.J.T.), Cleveland, Ohio; Cardialysis (J.D.), Rotterdam, Netherlands; Estudios Cardiologicos Latinoamerica
(R.D.), Rosario, Argentina; Eberhard Karls University (K.R.K.), Tubingen, Germany; COR Therapeutics (D.G., M.K.), South San Francisco, Calif; and
Thoraxcenter University Hospital (M.S.), Rotterdam, Netherlands.
This study was supported by COR Therapeutics, Inc (South San Francisco, Calif) and Schering Plough Research Institute (Kenilworth, NJ).
Correspondence to Robert A. Harrington, MD, Duke Clinical Research Institute, 2400 Pratt St, Durham, NC 27705. E-mail Harri019@mc.duke.edu
© 1999 American Heart Association, Inc.
Circulation is available at http://www.circulationaha.org
2892
Downloaded from http://circ.ahajournals.org/ by guest on March 17, 2016
McClure et al June 8, 1999 2893
Platelet Glycoprotein IIb/IIIa in Unstable Angina: Recep- found to have severe and profound thrombocytopenia were then
tor Suppression Using Integrilin Therapy (PURSUIT) trial reviewed individually, when possible, to verify the occurrence and
degree of reported thrombocytopenia.
evaluated the clinical efficacy and safety of the platelet
Additionally, several different thrombocytopenic populations
glycoprotein (GP) IIb/IIIa inhibitor eptifibatide in this were initially analyzed as the PURSUIT thrombocytopenia popula-
important patient population. tion. Each of these 3 groups experienced a platelet nadir that fulfilled
We examined thrombocytopenia in PURSUIT and sought the PURSUIT definition of thrombocytopenia, but they were differ-
to determine its incidence in this patient population, its ent in that the timing with respect to coronary bypass grafting, a
procedure that was expected to be a significant confounder, was
prognostic significance, and finally, whether antiplatelet ther-
considered. The 3 groups analyzed were: (1) patients who developed
apy through GP IIb/IIIa inhibition predisposes this population thrombocytopenia and never underwent CABG; (2) those who
to develop thrombocytopenia.8 experienced thrombocytopenia before CABG or who never under-
went CABG; and (3) those who developed thrombocytopenia regard-
Methods less of the presence or timing of CABG. In each of these analyses,
although the actual numbers varied, the overall trends in outcome
Patient Population variables were similar. We therefore concluded that thrombocytope-
The international, multicenter PURSUIT trial9 evaluated 10 948 nia affected outcomes in a uniform direction regardless of the
patients presenting with a non–ST-elevation ACS. Eligible pa- presence or timing of bypass grafting, and we elected to the use the
tients had ischemic chest pain symptoms within 24 hours of broadly defined third definition for our final PURSUIT thrombocy-
enrollment and met either prespecified ECG or cardiac enzyme topenic population.
requirements. Patients were excluded from study participation for
predisposition to bleeding, pregnancy, abnormal baseline labora- Clinical Outcomes
tory measurements (including platelet count ,1003109/L), or use Clinical events such as death and myocardial ischemia/MI were
of thrombolytics, other GP IIb/IIIa antagonists, or experimental documented on the case report form. Those interventions and clinical
drug therapies. events that occurred after the baseline hospitalization but within 30
days of enrollment were prospectively documented on a standard
Treatments 30-day form.
Patients were randomized in a double-blind fashion to receive either The primary efficacy outcome of the present study was the
intravenous eptifibatide (180 mg/kg bolus plus 2 mg z kg21 z min21 combined end point of death and nonfatal MI by 30 days after
infusion [high dose] or 180 mg/kg bolus plus 1.3 mg z kg21 z min21 randomization. This end point was adjudicated by an independent,
infusion [low dose]) or intravenous placebo control for 72 to 96 blinded Clinical Events Committee (CEC) using prespecified diag-
hours. Early in the trial, however, safety parameters for the high dose nostic criteria.9
of eptifibatide were within acceptable limits, and as specified by the
protocol, the low-dose arm was discontinued, with all subsequent Safety Outcomes
patients receiving the high-dose regimen or placebo. Safety outcomes for PURSUIT were CEC-adjudicated stroke; a
Concomitant use of aspirin, heparin, and other medications was calculated bleeding index, defined as [number of units of packed
left to the discretion of the individual investigators. For those red blood cells transfused1(observed drop in hematocrit/3)]; and
patients weighing ,70 kg and receiving intravenous unfractionated the incidence of bleeding based on the GUSTO definition of
heparin, the protocol suggested a weight-adjusted heparin-dosing bleeding severity.12 According to these criteria, severe bleeding
nomogram.10 It was recommended that patients who weighed .70 was defined as intracranial hemorrhage or a bleeding event that
kg receive a standard nonadjusted heparin dose (5000 U IV bolus– caused hemodynamic compromise and required intervention,
1000 U/h infusion). Transfusion guidelines were provided to all whereas moderate bleeding events were defined as those requiring
investigators to standardize transfusion usage.11 All decisions re- blood transfusion but not leading to hemodynamic compromise or
garding cardiac catheterization or revascularization (PTCA or need for an intervention.
CABG) were left to the individual investigators.
Statistics
Thrombocytopenia: Determination and Definition Continuous variables are presented as medians with 25th and 75th
Platelet count determinations were recommended at baseline, on a percentiles. Categorical data are displayed as an incidence (%).
daily basis during study-drug infusion, and at the investigator’s Differences in baseline characteristics and clinical/safety outcomes
discretion during the remainder of hospitalization. The case report between patients with and without thrombocytopenia were compared
form for all patients documented the baseline platelet count, #5 univariably with the median test for all reported medians and Pearson
platelet measurements during study-drug infusion, and 1 nadir x2 test for all reported incidences.
platelet count after study-drug infusion. Thrombocytopenia was Because the outcomes in this patient population were multifacto-
defined as a nadir platelet count of ,1003109/L or a relative rial, we performed multivariable regression to isolate certain vari-
reduction of the nadir platelet count to ,50% of baseline.3 Severe ables and assess their independent association with these outcomes.
and profound thrombocytopenia were defined as platelet nadirs
These variables are presented as ORs with 95% CIs. Four regression
,503109/L and ,203109/L, respectively.5
models were performed, 3 of which examined the contribution of
thrombocytopenia to moderate/severe bleeding, the primary end
Additional Platelet Information point, and death alone, respectively. To assess the independent
Periodic review of the blinded PURSUIT database indicated which contribution of thrombocytopenia to these various outcomes, the
study patients were experiencing thrombocytopenia. A standard model was adjusted for variables found to be important prognostic
thrombocytopenia form was then used to collect additional informa- factors in previous ACS trials,13 all statistically significant differ-
tion such as hematology consultations, heparin-induced antibody ences in baseline characteristics, the presence of a moderate/severe
determinations, and a complete listing of all platelet counts with bleeding event (except in the moderate/severe bleeding model),
regard to those patients who were experiencing severe or profound procedural interventions, and factors presumed to be important
thrombocytopenia. predictors of these outcomes. The fourth multivariable regression
model, which attempted to predict risk factors for thrombocytopenia,
PURSUIT Thrombocytopenia Population was created with disproportionately represented baseline character-
For purposes of this analysis, only data from the high-dose eptifi- istics, procedural interventions, and known or presumed thrombocy-
batide and placebo arms of PURSUIT were analyzed. Those patients topenia predictors used as covariates.
TABLE 1. Incidence of Thrombocytopenia The Figure depicts the time to onset of thrombocytopenia for
Placebo Eptifibatide
the overall thrombocytopenia definition and the combined se-
Degree of Thrombocytopenia (n54603) (n54614) P vere/profound thrombocytopenia groups, respectively. In the
overall thrombocytopenia population, the timing of thrombocy-
,1003109/L or ,50% of baseline 319 (7.0%) 314 (6.9%) 0.794
topenia was nearly identical in the 2 treatment arms (4 [2, 7]
,1003109/L 227 (4.9%) 227 (4.9%) 0.979
days for placebo versus 4 [2, 8] days for eptifibatide). In the
,503109/L (severe thrombocytopenia) 18 (0.4%) 23 (0.5%) 0.438
combined severe/profound thrombocytopenia group, the curves
,203109/L (profound thrombocytopenia) 2 (,0.1%) 5 (0.1%) 0.258 diverged early, with eptifibatide-treated patients developing
thrombocytopenia at a median of 52 (32, 81) hours after
Results randomization and placebo-treated patients developing throm-
bocytopenia at a median of 126 (99, 293) hours.
The overall incidence of thrombocytopenia was 7.0%. There
was an even distribution of thrombocytopenia between the Baseline Characteristics
placebo- and eptifibatide-treated arms across all definitions of Patients with thrombocytopenia were older, weighed less, and
thrombocytopenia (Table 1). were more likely to not be white (Table 2). Additionally, they
Time to onset of thrombocytopenia for overall thrombocytopenia definition (top) and combined severe/profound thrombocytopenia
groups (bottom) for patients treated with eptifibatide (E) or placebo (P).
had a greater number of cardiac risk factors, were more mic events (stroke, MI, death, and recurrent ischemia leading
frequently diabetic, and had a greater incidence of previous to cardiac procedure) at any point after enrollment was higher
MI, prior angioplasty, and history of peripheral vascular in patients with thrombocytopenia. The median length of stay
disease. in both intensive care units and during baseline hospitaliza-
tion was also significantly (P,0.001) longer for patients
Platelet Measures experiencing thrombocytopenia.
Although there was a significant (P,0.001) difference in
baseline platelet count between the thrombocytopenia Use of Antiplatelet Therapies
(2123109/L [1633109/L, 2603109/L]) and nonthrombocy- The use of abciximab (0.3% versus 0.2%; P50.435),
topenia groups (2263109/L [1923109/L, 2683109/L]), the ticlopidine (7.1% versus 9.6%; P50.041), aspirin (91.3%
total decrease in platelet count from baseline, as expected, versus 93.6%; P50.025), and thrombolytic therapy (2.1%
was more marked in the thrombocytopenia group versus 2.3%; P50.656)—all drugs with known or potential
(1273109/L [833109/L, 1643109/L] versus 293109/L antiplatelet effects—was similarly distributed between the
[133109/L, 513109/L] for nonthrombocytopenia). The thrombocytopenia and nonthrombocytopenia groups, re-
median absolute nadir platelet counts were 903109/L spectively. Heparin, also a drug with a known thrombocy-
(75310 9/L, 103310 9/L) and 193310 9/L (161310 9/L, topenic effect,14 was used with more prevalence (94.3%
2313109/L) for the thrombocytopenia and nonthrombocy- versus 89.6%; P,0.001) and was infused for a longer
topenia groups, respectively. period (87.0 versus 76.5 hours; P50.021) in the thrombo-
cytopenic group.
Transfusions/Bleeding Events
The development of thrombocytopenia was associated with a Eptifibatide Versus Placebo in
substantially higher incidence of any bleeding event, a Thrombocytopenic Patients
moderate/severe bleeding event, and the need for red blood A comparison of clinical events in the thrombocytopenic
cell or platelet transfusion at any point during baseline population of PURSUIT, grouped by study drug, revealed
hospitalization (Table 3). The bleeding index was also sig- a consistent but not statistically significant (except CABG,
nificantly higher (5.3 for thrombocytopenia versus 1.4 for P50.027) trend toward better clinical outcomes in patients
nonthrombocytopenia; P,0.001). with thrombocytopenia who were treated with eptifibatide
(Table 4). This relationship occurred in spite of an excess
Clinical Outcomes incidence of any bleeding (P50.017) and severe bleeding
Data at 30 days after randomization documented worse (P50.046) in this group. A formal statistical test for
outcomes for patients with thrombocytopenia than for those interaction in our primary end-point regression model
without thrombocytopenia (Table 3). The incidence of ische- confirmed no significant interaction between treatment
Downloaded from http://circ.ahajournals.org/ by guest on March 17, 2016
2896 Thrombocytopenia in Non–ST-Elevation ACS
group and thrombocytopenia, implying that the benefit (OR 12.2 [9.1 to 16.2]), as were patients with moderate to
seen in the overall PURSUIT trial occurred in the same severe bleeding events and those treated with an intra-aortic
direction and with similar magnitude for the thrombocy- balloon pump. However, the use of heparin, a therapy well
topenia subpopulation. known to cause thrombocytopenia,14 did not significantly
predispose patients to increased thrombocytopenic risk in our
Regression Models model when evaluated dichotomously (heparin use: yes/no;
The multivariable regression model for predictors of moder- exposure time: ,5 days/$5 days) and continuously (infusion
ate/severe bleeding (Table 5) demonstrated that even after duration). Similarly, the use of other antiplatelet therapies (ie,
adjustment for confounders, thrombocytopenia was indepen- aspirin, ticlopidine, abciximab, thrombolytics, and eptifi-
dently correlated with an increased bleeding risk (OR 2.0 [1.6 batide) was not associated with a significantly increased risk
to 2.6]). Similarly, the model for predictors of the PURSUIT of thrombocytopenia.
primary end point (Table 5) revealed that thrombocytopenia
was independently associated with an increased rate (OR 1.3 Discussion
[1.0 to 1.6]) of death/nonfatal MI within 30 days of random- The data (Table 3) suggest a clear correlation between
ization, even after adjustment for a large number of con- thrombocytopenia and adverse clinical outcomes including
founding variables. Thrombocytopenia did not significantly serious bleeding and ischemic events. Although baseline
increase the risk of death alone, however, in our third characteristics of the thrombocytopenia population appear
regression model (not shown). to predispose this population to poorer outcomes, our
Finally, a number of variables were found to correlate analysis shows that even after controlling for a number of
independently with increased risk of thrombocytopenia (Ta- these differences and other confounders (Table 5), the
ble 6). Patients who underwent CABG were highly signifi- isolated occurrence of thrombocytopenia remains signifi-
cantly predisposed to the development of thrombocytopenia cantly correlated with an increased incidence of moderate/
Downloaded from http://circ.ahajournals.org/ by guest on March 17, 2016
McClure et al June 8, 1999 2897
possible, however, that the total numbers involved (especially platelet counts during the initial 72 to 96 hours of enrollment,
with abciximab, ticlopidine, and thrombolytics) were inade- only 1 platelet count determination was recorded after infu-
quate to provide power to detect the potential contribution of sion of the study drug. Consequently, it is not possible to
these medications to risk of thrombocytopenia. address several important questions, such as total thrombo-
A recent meta-analysis of thrombocytopenic risk in the cytopenia duration and the time course to complete platelet
context of GP IIb/IIIa inhibition supports the concern that this recovery.
new class of drug may predispose patients to develop throm-
bocytopenia.8 Our data (Table 1) show a nearly identical Conclusions
distribution of thrombocytopenia between eptifibatide- and Nearly 1 in 14 patients with a non–ST-elevation ACS
placebo-treated arms in all severities of thrombocytopenia enrolled in the PURSUIT trial experienced thrombocytope-
except profound thrombocytopenia, and even in this group nia, defined as a platelet nadir ,1003109/L or ,50% of
the difference did not reach statistical significance baseline. Our analysis demonstrates that this large subset of
(P50.258). However, 5 of the 7 patients in this group were patients with unstable angina or non–Q-wave MI is at a
treated with eptifibatide, and time to onset of thrombocyto- significantly increased risk of experiencing not only serious
penia in eptifibatide-treated patients with severe/profound bleeding events (moderate/severe bleeding, need for red
thrombocytopenia diverges from placebo-treated patients blood cell/platelet transfusion) but clinically significant is-
quite early (Figure), leaving room for speculation that in this chemic events (death, nonfatal MI, stroke, recurrent ischemia
small subset of patients with thrombocytopenia, eptifibatide leading to a cardiac procedure) as well. Because causation
exposure could contribute to thrombocytopenic risk. Even if could not be established between thrombocytopenia and
this were true, however, only a very small subset of patients outcomes, it would be inappropriate to infer from this
(0.1% of all patients treated) would be at risk of developing analysis that prophylactic platelet transfusions should be
this condition, whereas the vast majority of non–ST-elevation administered to patients experiencing thrombocytopenia in an
ACS patients appear to be at an equal risk of developing attempt to decrease their risk of morbidity and mortality.
thrombocytopenia regardless of eptifibatide exposure. This Further study is warranted, however, to refine the estimate of
statement is substantiated by our regression model (Table 6), magnitude of the effect of thrombocytopenia on outcomes
which does not indicate any significant contribution of and to better examine its causes. In the meantime, daily
eptifibatide to thrombocytopenia after controlling for con- platelet count determinations should be collected in all
founders. More importantly, in general, except perhaps for patients experiencing an ACS to aid in early identification of
this group of patients with profound thrombocytopenia, this important at-risk patient population.
patients with thrombocytopenia who receive eptifibatide ap-
pear to benefit from its use, just as they do in the overall Acknowledgments
PURSUIT population.9 This conclusion is implied by the This study was supported by COR Therapeutics, Inc (South San
Francisco, Calif) and Schering Plough Research Institute (Ken-
consistent but small trend toward fewer adverse events ilworth, NJ). The authors wish to thank Gail Tudor, PhD, for her
(except bleeding) in the eptifibatide-treated thrombocytope- thoughtful review and insights into the contents of this manuscript.
nic population (Table 4) and the absence of a significant Additionally, we are grateful to Penny Hodgson and John Daniel for
treatment-by-thrombocytopenia interaction term in our pri- their editorial expertise in drafting this manuscript.
mary end-point regression model.
References
1. Murray CJL, Lopez AD. Mortality by cause for eight regions of the
Limitations world: Global Burden of Disease Study. Lancet. 1997;349:1269 –1276.
Our analysis clearly points toward a correlation between 2. American Heart Association. 1997 Heart and Stroke Statistical Update.
thrombocytopenia and adverse events but cannot be inter- Dallas, Tex: American Heart Association; 1997:2–29.
preted to establish causation. Some events preceded the 3. Harrington RA, Sane DC, Califf RM, Sigmon KN, Abbottsmith CW,
Candela RJ, Lee KL, Topol EJ, for the TAMI Study Group. Clinical
development of thrombocytopenia (eg, MI in Table 3) and importance of thrombocytopenia occurring in the hospital phase after
therefore cannot be included in a model to establish the administration of thrombolytic therapy for acute myocardial infarction.
predictive (causal) power of thrombocytopenia on negative J Am Coll Cardiol. 1994;23:891– 898.
outcomes. To establish such a model, one must include only 4. Bovill EG, Terrin ML, Stump DC, Berke AD, Frederick M, Collen D,
Feit F, Gore JM, Hillis D, Lambrew CT, Leiboff R, Mann KG, Markis JE,
those events that can be shown to have occurred after the Pratt CM, Sharkey SW, Sopko G, Tracy RP, Chesebro JH, for the TIMI
onset of thrombocytopenia. Because of the nature of our Investigators. Hemorrhagic events during therapy with recombinant
timing data, only death, which by definition cannot be tissue-type plasminogen activator, heparin, and aspirin for acute myo-
cardial infarction. Ann Intern Med. 1991;115:256 –265.
followed by events such as thrombocytopenia, is eligible for 5. Berkowitz SD, Sane DC, Sigmon KN, Shavender JH, Harrington RA,
such a predictive model. Our regression model for predicting Tcheng JE, Topol EJ, Califf RM, for the EPIC Study Group. Occurrence
death (not shown) failed to demonstrate an increased inci- and clinical significance of thrombocytopenia in a population undergoing
dence of death in patients with thrombocytopenia, however. high-risk percutaneous coronary revascularization. J Am Coll Cardiol.
1998;32:387–392.
Whether this observation reflects a lack of a causal role of 6. Berkowitz SD, Sparapani RA, Califf RM, Cohen M. Occurrence and
thrombocytopenia on death or simply the low incidence of significance of thrombocytopenia in unstable angina and non-Q-wave
death in the trial is difficult to establish. infarction in the ESSENCE trial. Blood. 1997;90:81b. Abstract.
7. Berkowitz SD, Granger CB, Thompson T, Kleiman NS, Topol EJ, Califf
A second limitation of this study is the amount of platelet
RM. Thrombocytopenia was associated with greater hemorrhage and
count data collected during baseline hospitalization. Although higher mortality in GUSTO IIb trial. Thromb Haemost. 1997;236:PS-962.
the case report form allowed documentation of up to 5 Abstract.
8. Giugliano RP, Hyatt RR Jr. Thrombocytopenia with GP IIb/IIIa inhibi- determined by baseline clinical factors. J Am Coll Cardiol. 1999;
tors: a meta-analysis. J Am Coll Cardiol. 1998;31:185A. Abstract. 33:359A. Abstract.
9. PURSUIT Trial Investigators. Inhibition of platelet glycoprotein IIb/IIIa 14. Brieger DB, Mak KH, Kottke-Marchant C, Topol EJ. Heparin-induced
with eptifibatide in patients with acute coronary syndromes. N Engl thrombocytopenia. J Am Coll Cardiol. 1998;31:1449 –1459.
J Med. 1998;339:436 – 443. 15. Rao AK, Pratt C, Berke A, Jaffe A, Ockene I, Schreiber TL, Bell WR,
10. Wali AU, Gavrila D, Sim MJ, Malhotra S, Palazzo AM, Hochman JS. An Knatterud G, Robertson TL, Terrin ML, for the TIMI Investigators.
improved regimen for heparin use in acute coronary syndromes. Circu- Hemorrhagic manifestations and changes in plasma fibrinogen and the
lation. 1997;96(suppl I):I-597. Abstract. fibrinolytic system in patients treated with recombinant tissue plas-
11. American College of Physicians. Practice strategies for elective red blood minogen activator and streptokinase. J Am Coll Cardiol. 1988;
cell transfusion. Ann Intern Med. 1992;116:403– 406. 11:1–11.
12. The GUSTO Investigators. An international randomized trial comparing 16. Robblee JA. Platelet utilization in cardiopulmonary bypass surgery. Curr
four thrombolytic strategies for acute myocardial infarction. N Engl Stud Hematol Blood Transfus. 1988;54:101–106.
J Med. 1993;329:673– 682. 17. Warkentin TE, Levine MN, Hirsh J, Horsewood P, Roberts RS, Gent M,
13. Boersma E, Pieper K, Wilcox RG, Steyerberg E, Lee KL, Akkerhuis M, Kelton JG. Heparin-induced thrombocytopenia in patients treated with
Harrington RA, Califf RM, Simoons ML. The risk of adverse outcome in low-molecular-weight heparin or unfractionated heparin. N Engl J Med.
patients with acute coronary syndromes without ST-elevation is mainly 1995;332:1330 –1335.
Circulation. 1999;99:2892-2900
doi: 10.1161/01.CIR.99.22.2892
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1999 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/99/22/2892
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.