Accepted Manuscript A risk assessment tool incorporating new biomarkers for cardiovascular events in acute coronary syndromes: the Organization to Assess Strategies in Ischemic Syndromes (OASIS) risk score Shamir R. Mehta, MD, MSc, John W. Eikelboom, MD, MSc, Purnima Rao-Melacini, MSc, Jeffrey I. Weitz, MD, Sonia S. Anand, MD, PhD, Guillaume Pare, MD, MSc, Andrezj Budaj, MD, PhD, Janice Pogue, PhD, Keith A.A. Fox, MBBS, FRCP, Salim Yusuf, MBBS, DPhil PII:
S0828-282X(16)00069-6
DOI:
10.1016/j.cjca.2016.01.029
Reference:
CJCA 2015
To appear in:
Canadian Journal of Cardiology
Received Date: 25 September 2015 Revised Date:
7 January 2016
Accepted Date: 27 January 2016
Please cite this article as: Mehta SR, Eikelboom JW, Rao-Melacini P, Weitz JI, Anand SS, Pare G, Budaj A, Pogue J, Fox KAA, Yusuf S, A risk assessment tool incorporating new biomarkers for cardiovascular events in acute coronary syndromes: the Organization to Assess Strategies in Ischemic Syndromes (OASIS) risk score, Canadian Journal of Cardiology (2016), doi: 10.1016/j.cjca.2016.01.029. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
A risk assessment tool incorporating new biomarkers for cardiovascular events in acute coronary syndromes: the Organization to Assess Strategies in Ischemic
RI PT
Syndromes (OASIS) risk score Shamir R. Mehta MD, MSc; John W. Eikelboom MD, MSc; Purnima Rao-Melacini MSc; Jeffrey I. Weitz MD; Sonia S. Anand MD, PhD; Guillaume Pare MD, MSc; Andrezj
SC
Budaj MD, PhD; Janice Pogue PhD; Keith A. A. Fox MBBS, FRCP and Salim Yusuf
M AN U
MBBS, DPhil
Population Health Research Institute, McMaster University and Hamilton Health Sciences (SRM, JWE, PRM, SSA, GP, AB, JLW, JP, SY); Thrombosis and Atherosclerosis Research Institute, McMaster University and Hamilton Health Sciences, Hamilton, Canada (JIW); Department of Cardiology, Postgraduate Medical School,
TE D
Grochowski Hospital, Warsaw, Poland (AB); Royal Infirmary, University of Edinburgh, Edinburgh, Scotland (KAAF)
EP
Correspondence to:
Shamir R. Mehta MD, MSc
AC C
Population Health Research Institute 237 Barton St East, Hamilton, ON L8L 2X2 Phone: 905-521-2631
Email:
[email protected] Word count: 3780 (includes abstract, text, tables and figure legends) Short Title: OASIS risk score for ACS
1
ACCEPTED MANUSCRIPT
Brief summary Several new biomarkers improve risk stratification in non-ST-segment elevation ACS, however they are not integrated into risk prediction tools. A risk score to predict CV
RI PT
death/MI/stroke was developed by incorporating new biomarkers with standard variables. The addition of NT-proBNP and hemoglobin A1C to 5 standard variables (age, prior MI/stroke, sex, ST-segment deviation, troponin-T) created a simple risk score that
SC
improved prediction at 1 year and aided in risk-based selection of patients for dual
AC C
EP
TE D
M AN U
antiplatelet therapy.
2
ACCEPTED MANUSCRIPT
Abstract Background: Several biomarkers have been shown to improve risk stratification in
they have not been integrated into risk prediction tools.
RI PT
patients with non-ST-segment elevation acute coronary syndrome (NSTEACS), however
Methods: C-reactive-protein (CRP), NT-pro-brain-natriuretic peptide (NT-proBNP) and hemoglobin A1C were measured in 6,447 patients with NSTEACS who were enrolled in
SC
the Clopidogrel to prevent Recurrent Events (CURE) trial. A risk score to predict CV death, MI or stroke at 1-year was developed by incorporating biomarkers that were
M AN U
independently predictive of events with traditional variables, ECG and troponin T. Model discrimination was evaluated using C-statistic, integrated discrimination improvement (IDI) and net reclassification index (NRI) and validated using bootstrap methods. Results: During 1 year of follow-up, 686 patients experienced a CV event. Each
TE D
biomarker predicted CV death, MI or stroke, however only NT-proBNP and hemoglobin A1C improved model discrimination, increasing the C-statistic (0.66 to 0.71), IDI to 3.4%, and NRI to 17.5% (P<0.0001 for all measures). A risk score ranging from 0-20
EP
points including variables for age, prior MI/stroke, sex, ST-segment deviation, troponinT, NT-proBNP and hemoglobinA1C classified individuals into low, intermediate and
AC C
high risk groups with rates of CV death/MI/stroke of 3.7%, 9.1% 17.8%, respectively. The absolute benefit of dual antiplatelet therapy versus aspirin alone was 1.0%, 4.7% and 3.0% in low, intermediate and high-risk groups. Conclusions: The addition of NT-proBNP and hemoglobin A1C to 5 standard variables creates a 7-variable risk score that improves prediction of cardiovascular events at 1 year and aids in risk-based selection of NSTEACS patients for dual antiplatelet therapy.
3
ACCEPTED MANUSCRIPT
Introduction Although there have been substantial improvements in the management and outcome of patients with non-ST-segment elevation acute coronary syndromes
RI PT
(NSTEACS), there is still a significant risk of early and long term cardiovascular (CV)
events.1 Practice guidelines for management of NSTEACS recommend that all patients be risk stratified for their likelihood of developing recurrent CV events or death.2-4
SC
Established risk stratification tools such as the Global Registry of Acute Coronary Events (GRACE) and the Thrombolysis in Myocardial Infarction (TIMI) risk scores are derived
M AN U
from demographic, clinical, laboratory and ECG-related variables. 5-9 They do not incorporate the use of newer biomarkers, particularly those that have shown incremental value in the prediction of CVevents.10-17 Combining biomarkers representing different pathophysiologic processes could provide complementary prognostic information,
TE D
thereby improving risk stratification beyond traditionally used variables.13-16 To further explore this possibility, we measured baseline levels of C-reactive protein (CRP), a marker of inflammation, N-terminal -pro-brain natriuretic peptide (NT-
EP
proBNP), a marker of hemodynamic stress, and hemoglobin A1C, a marker of dysglycemia, in 6,447 patients with NSTEACS. Our goal was to first determine their
AC C
independent prognostic value for prediction of major CV events at 1 year when incorporated into multivariable models that include traditional variables, including those found in the TIMI and GRACE risk scores.5-9 We then incorporated the biomarkers found to offer independent prognostic value into a risk score that predicted the occurrence of CV death, myocardial infarction (MI) or stroke 1 year after NSTEACS. Finally, we validated this new Organization to Assess Strategies in Ischemic Syndromes
4
ACCEPTED MANUSCRIPT
(OASIS) risk score using bootstrapping methods and compared its performance with
AC C
EP
TE D
M AN U
SC
RI PT
established risk scores.
5
ACCEPTED MANUSCRIPT
METHODS Patients Baseline blood samples were collected from 6,447 of the 12,500 patients enrolled in the
RI PT
Clopidogrel in Unstable angina to prevent Recurrent Events (CURE) trial, which
randomized individuals with NSTEACS to receive either clopidogrel or placebo, in
addition to aspirin and other standard therapies.18 Patients were enrolled if they presented
SC
within 24 hours of the last episode of ischemic symptoms and had either ST segment deviation or elevated levels of troponin or creatine kinase MB fraction. The primary
M AN U
outcome was the composite of death from CV causes, new MI or stroke at the end of follow-up (which was a mean of 9 months or maximum of 1 year). The CURE study and the biomarker substudy protocols were approved by the institutional review board of each participating hospital and all patients provided informed consent to participate in the
TE D
main study and in the substudy.
Blood Collection and Assay Methods
EP
Each study site was provided with standardized supplies for the collection, processing, storage and shipment of blood specimens, as well as detailed instructions in both written
AC C
and video format. All analyses were performed on serum with the exception of the whole blood that had been applied to Roche filter papers for hemoglobin A1C determination. CRP was quantified with the high sensitivity DADE Behring immunotubidimetric assay performed on a DADE BNII analyzer (Dade Behring Inc., Newark, DE). NT-proBNP and troponin T were quantified using the Roche Elecsys pro-BNP and Troponin T immunoassays on an Elecsys 1010 (F. Hoffman-LaRoche Ltd., Basel, Switzerland).
6
ACCEPTED MANUSCRIPT
Hemoglobin A1C was quantified with the Tina-Quant II assay using a Hitachi 917 instrument.
RI PT
Statistical Analyses
Continuous variables were summarized by means and standard deviations where a
normal distribution could reasonably be assumed or by medians and inter-quartile ranges
SC
otherwise. Biomarkers were presented as 1 standard deviation above the mean on a log transformed measurement. Baseline variables were compared between individuals that
M AN U
experienced CV death, MI or stroke and those who did not, using a Student’s t-test or Wilcoxon rank sum test for continuous variables, and a Chi-square test for dichotomous variables. The association between biomarkers and CV death, MI or stroke was explored using backward elimination in a Cox proportional hazards model. The potential variables
TE D
incorporated were those already incorporated into the TIMI and GRACE risk scores, including: age, creatinine, gender, history of diabetes, hypertension, high cholesterol, current smoking, history of heart failure, prior vascular event (either MI or stroke),
EP
previous coronary artery bypass graft surgery or percutaneous coronary intervention, troponin positivity (2X ULN) and ST-segment depression or elevation on the
AC C
electrocardiogram. Troponin and ST segment shift were forced into the final model, given that they are standard tests used to risk stratify patients with NSTEACS. After development of the multivariable model, the risk score was developed using variables in addition to troponin and ST segment shift, that were statistically significant predictors of events by assigning weighted points to the model’s variable coefficients using the method of Sullivan et al.18 We also investigated whether the addition of different combinations
7
ACCEPTED MANUSCRIPT
of the biomarkers improved the discrimination of the model with the method described by Pencina et al., by calculating net reclassification improvement (NRI) and integrated discrimination improvement (IDI).20 For the NRI, we estimated the risk categories from
RI PT
the full model (incorporating clinical variables, ECG changes, and levels of troponin T, NT-proBNP and hemoglobin A1C), stratified into 3 groups of risk using tertiles of the risk score (≤4% per year, 5-9% per year, and ≥10% per year). For calculation of the
SC
GRACE risk score, we used the updated GRACE 2.0 score described by Fox et al.8
For mulitivariable model validation, we used bootstrap re-sampling to assess the
M AN U
predictors of CV death, MI or stroke.21 One thousand bootstrap samples were drawn from a derivation data set (random selection of two thirds of the original data set ) and backwards elimination was used to develop a predictive model from each sample. The proportion of bootstrap samples in which that variable was identified as an independent
TE D
predictor of the outcome was determined. The candidate model that included those variables that were identified as significant predictors of the primary outcome in at least 60% of the bootstrap samples performed the best. Each candidate model was fitted to the
EP
validation data set (remaining one third of the original data set), assessed for its predictive accuracy (Hosmer-Lemeshow g-statistic and c-index) and a final model was
AC C
selected. P values less than 0.05 from two-sided tests were considered to indicate statistical significance. The statistical software packages SAS (version 9.2 for unix, SAS Institute) and R (version 2.14.1, The R Foundation for Statistical Computing) were used.
8
ACCEPTED MANUSCRIPT
RESULTS Patients enrolled in the biomarker sub-study (N=6,447) were representative of the overall CURE trial population (N=12,562), with identical rates of CV death, MI or stroke
RI PT
(primary outcome) in both groups (10.3%). The mean age was 64 years in both groups, with similar proportions of diabetics (21% in biomarker substudy vs 23% overall), those with hypertension (60% vs 59%) and current (21 vs 23%) and former smokers (40% vs
SC
38%, respectively).
A total of 686 patients (10.6%) had a major CV event (CV death, MI or stroke) during 1
M AN U
year of follow-up. Of these, 361 (5.6%) had CV death, 39 (0.6%) had non-CV death, 387 (6%) had an MI and 80 (1.2%) had a stroke. Baseline characteristics of patients with or without a CV event are shown in Table 1. Significantly more patients with a CV event had a prior history of diabetes, hypertension, MI, congestive heart failure and ST segment
TE D
depression on the baseline ECG. Median concentrations of all biomarkers, including Troponin T, CRP, hemoglobin A1C and NT-proBNP, were higher in patients who
EP
experienced a CV event than in those without an event (Table 1).
Inflammatory Markers and Risk
AC C
C-reactive protein (HR 1.22, P<0.001), NT-proBNP (HR 1.98 ,P<0.001) and hemoglobin A1C (HR 1.28, P<0.001) all significantly predicted the risk of CV events in univariable analyses. In multivariable analyses, after correcting for baseline factors, troponin T and ST segment shift, NT-proBNP (HR 1.73, P<0.001), hemoglobin A1C (HR 1.28, P<0.001) and CRP (HR 1.11, P=0.015) all predicted the composite risk of CV death, MI or stroke, as well as CV death alone (Table 2). NT-proBNP and hemoglobin A1C, but not CRP,
9
ACCEPTED MANUSCRIPT
significantly predicted the risk of non-fatal MI and non-fatal stroke. A similar pattern of results was observed in patients who were troponin T positive and in those who were troponin T negative. In patients with an established diagnosis of diabetes at baseline,
RI PT
hemoglobin A1C and NT-proBNP, but not CRP, significantly predicted the composite outcome, as they did in those without a history of diabetes. Similarly, when stratified
death, MI and stroke in both men and women (table 2).
M AN U
Risk Prediction Model and Validation
SC
according to sex, both NT-proBNP and HBA1C significantly predicted the risk of CV
A risk prediction tool ranging from 0 to 20 points and incorporating hemoglobin A1C and NT-proBNP was developed (Table 4). Variables included in the model were: age, sex, prior vascular event (MI or stroke), ST segment shift on the baseline ECG, troponin T,
TE D
hemoglobin A1C and NT-proBNP. Chi square values for hemoglobin A1C (72.25) and NT-proBNP (97.32) accounted for 78% of the total and exceeded the chi square for all other variables, including age (chi square 19.05). Event rates for CV death, MI or stroke
EP
at 1 year were 3.7% for low risk patients (score 0-7), 9.1% for moderate risk patients (score 8-11), and 17.8% for high risk patients (score 12-20, p<0.0001 for trend) (Figure
AC C
1). Similarly, the model also predicted each component of this of this outcome (Figure 1). In patients who underwent early PCI within 72 hours of enrollment (N=580), event rates for CV death, MI or stroke were 7.3% for low risk patients (score 0-7), 11.1% for intermediate risk patients (score 8-11) and 18.1% for high risk patients (score 12-20). Within all 3 risk groups, approximately one-third of the CV death, MI or stroke events occurred during the initial hospitalization and approximately two-thirds occurred after
10
ACCEPTED MANUSCRIPT
discharge: Low risk group 31 vs 58 events, intermediate risk group 58 vs 11 and high risk group 100 vs 213, respectively. The annualized event rates were higher in the post discharge period in all 3 risk groups (Supplementary Table S1).
RI PT
Regarding the comparison of clopidogrel versus placebo, the absolute benefit of clopidogrel was substantially larger in the intermediate (absolute risk reduction [ARR] 4.7%, number needed to treat [NNT] 21) and high risk (ARR 3.0%, NNT 33) groups,
SC
compared with the low risk group (ARR 1.0%, NNT 103) (figure 2). There was no statistical heterogeneity in the relative risk according to risk group.
M AN U
Model validation was undertaken using bootstrapping methods. The significant predictors of the candidate model with 60% of the bootstrap samples included age, NTproBNP, haemoglobin A1C, prior vascular event and gender. We included Troponin T≥ 2x ULN and ST depression/elevation as well, which accounted for 40% of the bootstrap
TE D
samples because of their clinical significance. The Hosmer-Lemeshow goodness of fit statistic from the final model had a p-value = 0.196.
EP
Model Discrimination and Reclassification Both hemoglobin A1C (c-statistic 0.69, NRI 0.075, P<0.001) and NT-proBNP (c-statistic
AC C
0.69, NRI 0.121, P<0.001), but not CRP, improved the c-statistic, IDI and the NRI (Table 5). When based on log transformed standard deviation data, combining the NT-proBNP and hemoglobin A1c levels resulted in greater improvement in model discrimination (cstatistic 0.708; NRI 17.5%, P<0.0001; IDI 3.4%, P<0.001) (Supplementary Table S2, Figure 3). Similar results were observed when the data were categorized using the
11
ACCEPTED MANUSCRIPT
optimal cut-point for the biomarker and when the biomarker data were divided into equal tertiles. Of patients who had a CV event at follow-up, reclassification after adding NT-
RI PT
proBNP and hemoglobin A1C occurred in 38.0% of patients at intermediate risk (6.7% to low risk and 31.3% to high risk) and 10.4% at high risk (0.5% to low risk and 9.9% to intermediate risk, Supplementary Table S2). Of patients not experiencing a CV event,
high-risk categories (Supplementary Table S3).
SC
45.2% of patients at intermediate risk were reclassified: 29.4% to low risk and 15.8% to
M AN U
Figure 3 shows the receiver operating curves for the new OASIS model compared with the GRACE 2.0 and TIMI risk scores. In this population of NSTEACS patients, the new OASIS score performed better than the TIMI or GRACE risk scores (c-statistics 0.708, 95% CI 0.687-0.729; 0.638, 95% CI 0.616-0.661, and 0.653, 95% CI 0.630-0.676,
TE D
respectively). (Figure 3) The C-statistics for mortality alone were also better with the OASIS score (0.78, 95% CI 0.76 -0.81) compared with the GRACE (0.74, 95% CI 0.71 -
AC C
EP
0.77) and TIMI scores (0.68, 95% CI 0.65 -0.71).
12
ACCEPTED MANUSCRIPT
DISCUSSION In this population of patients with NSTEACS, we created a 7 variable risk stratification tool, the OASIS risk score, that predicts the likelihood of CV death, MI or stroke at 1
RI PT
year. This new risk prediction tool built on variables used in prior risk scores, by
incorporating NT-proBNP and hemoglobin A1C, two biomarkers that offer substantial
incremental prognostic value. Reclassification methods demonstrated that the new score
SC
improved risk prediction mainly in those with an intermediate pre-test probability, where the greatest uncertainty exists in risk-based decision making. The risk score predicted
M AN U
events that occurred both during the initial hospitalization as well as in the post discharge period and demonstrated a similar ability to predict events in those treated with an initial non-invasive strategy, as well as in those undergoing early PCI. The new risk score predicted a larger absolute benefit of dual antiplatelet therapy in intermediate and high
TE D
risk patients compared with low risk patients. Finally, the new risk score compared favorably with established risk scores.
Because NSTEACS represents a heterogeneous group of patients, risk
EP
stratification is an important part of the management of these patients. For example, use of the GRACE risk score predicts which individuals with NSTEMI benefit preferentially
AC C
from an early versus delayed invasive strategy.22 Novel biomarkers, such as NT-proBNP and CRP, measured alone or in combination at baseline in ACS patients have been reported to predict mortality and to aid in the selection of therapies.17 In this study, our primary goal was to evaluate the utility of these novel biomarkers in the clinical context and to incorporate them into a risk prediction tool. The validated risk prediction tool, consisting of 5 clinical variables and 2 biomarkers, performed well, with a better ability
13
ACCEPTED MANUSCRIPT
to predict events compared with established risk scores that do not incorporate these biomarkers. NT-proBNP was a strong and robust predictor not only of mortality, but also of
RI PT
MI and stroke and the composite of these events. NT-proBNP has not been previously
demonstrated to predict these latter events. Similarly, hemoglobin A1C robustly predicted both fatal and non-fatal outcomes in the overall population, as well as in diabetics and
SC
non-diabetics. The value of hemoglobin A1C as a prognostic marker was similar in both diabetics and non-diabetics. The increased predictive value of NT proBNP and
M AN U
hemoglobin A1C was in addition to the clinical variables included in the TIMI and GRACE risk scores, which were incorporated into the multivariable models used to develop the score. In the final multivariable model, the chi square for NT-proBNP and hemoglobin A1C accounted for a remarkable 78% of the total variance and exceeded the
TE D
chi square for age (19.05) and troponin T (3.45). Thus, incorporation of NT-proBNP and hemoglobin A1C into routine risk stratification adds important new information to clinical variables alone. Similar to other recent studies of patients with NSTEACS15,22,
EP
CRP did not predict events as robustly as the other biomarkers and was omitted from the final model and risk prediction tool.
AC C
The absolute benefit of clopidogrel was greatest in intermediate and high-risk
patients (ARR 4.7% and 3.0%, respectively) compared with low risk patients (1.0%). The number needed to treat to prevent 1 death, MI or stroke was only 21 and 33 for moderate and high risk patients, respectively. By contrast, the number need to treat to prevent one of these events for low risk patients was 100. This suggests that the risk score may have particular value in identifying a lower risk cohort where the reduced absolute
14
ACCEPTED MANUSCRIPT
benefit of DAPT in preventing ischemic events should be more carefully weighed with the risk of bleeding. Our new risk prediction tool may also help to select patients who are likely to derive greater benefit from newer adenosine receptor antagonists such as
RI PT
ticagrelor23 or prasugrel.24 These agents have been shown to be superior to clopidogrel for prevention of cardiovascular events, but with increased bleeding. Therefore, risk
stratification is particularly important in optimizing the benefit-risk balance when using
SC
these newer agents.
A particular strength of our risk prediction tool is that it considered all of the
M AN U
variables incorporated in the GRACE and TIMI risk scores, in addition to several biomarkers. We were able to substantially simplify the score to only 7 variables and this approach produced a comparable, if not a better ability to predict events compared with older scores that do not incorporate newer biomarkers. Secondly, it is based on a large-
TE D
scale randomized comparison of dual antiplatelet therapy versus aspirin alone in patients with NSTEACS, which allowed us to evaluate the absolute benefit of dual antiplatelet therapy. Thus, the score will have clinical value in risk-based decision-making, as it not
EP
only predicts risk, but also helps with the selection of therapy. Limitations of our study also merit consideration. First, our risk assessment tool
AC C
was not tested in a separate validation cohort. Therefore, it is possible that the results in a second cohort may not be as effective in the prediction of risk as in the current database. Because the CURE trial required elevated biomarkers or ischemic ECG changes for entry, it is possible that the risk score was derived from a higher risk ACS population than what might be observed in a broader population presenting to hospital with ischemic symptoms. Second, although patients at low risk according to our score derived little
15
ACCEPTED MANUSCRIPT
benefit from DAPT (ARR 1.0%), high-risk patients (ARR 3.0%) did not seem to derive greater benefit compared with intermediate risk patients (ARR 4.7%). While the reason for this is uncertain, the greatest utility of our risk score might be in identifying low risk
RI PT
patients, in whom the risk-benefit profile of DAPT should be more carefully evaluated. Third, lack of a strong association between CRP and risk could be due to the fact that diabetes, statin therapy and gender influence CRP levels and the fact that CRP levels
SC
were not measured at least twice with 10-14 days between measurements. Fourth, the
CURE trial enrolled patients undergoing a non-invasive strategy. However, 2600 patients
M AN U
underwent PCI and there was no heterogeneity in this cohort. Further, the risk score was highly predictive of events in those undergoing early PCI (7.3%, 11.1% and 18.1% in those at low, intermediate and high risk). Like the TIMI9 and GRACE5 scores, the CURE trial was performed some years ago. However, the risk score would be expected to apply
TE D
equally to patients with NSTEACS today because the variables used to predict risk are independent of treatment and remain clinically relevant. In conclusion, we evaluated the incremental prognostic value of measuring 3
EP
biomarkers (CRP, NT-proBNP and hemoglobin A1C) that reflect different pathophysiological processes involved in acute coronary syndrome. Using conventional
AC C
and reclassification methods, we demonstrated that NT-proBNP and hemoglobin A1C, but not CRP, robustly improved the prediction of risk, mainly in patients at intermediate risk. The absolute risk of dual antiplatelet therapy was greater in patients at intermediate and high risk, compared with low risk patients. A simple tool including these two biomarkers, in addition to 5 clinical variables, enhances risk prediction in this population of patients.
16
ACCEPTED MANUSCRIPT
Disclosures This study was funded by a peer-review grant from the Canadian Institutes of Health
AC C
EP
TE D
M AN U
SC
Population Health Research Institute, McMaster University.
RI PT
Research (99781). The CURE trial was funded by Sanofi-Aventis and coordinated by the
17
ACCEPTED MANUSCRIPT
References
1. Fox KA, Steg PG, Eagle KA, Goodman SG, Anderson FA Jr, Granger CB, Flather
RI PT
MD, Budaj A, Quill A, Gore JM; GRACE Investigators. Decline in rates of death and heart failure in acute coronary syndromes, 1999-2006. JAMA 2007;297:1892-900.
SC
2. Amsterdam EA, Wenger NK, Brindis RG, et al. 2014 AHA/ACC guideline for the
management of patients with non–ST-elevation acute coronary syndromes: a report of the
M AN U
American College of Cardiology/ American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2014;64:e139–228.
3. Roffi M, Patrono C, Collet J-P et al. 2015 ESC guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation
TE D
Task Force for the Management of Acute Coronary Syndromes in Patients Presenting without Persistent ST-Segment Elevation of the European Society of Cardiology (ESC).
EP
Eur Heart J. 2015 Aug 29. pii: ehv320. [Epub ahead of print]
AC C
4. Tanguay J-F, Bell AD, Ackman ML et al. Focused 2012 Update of the Canadian Cardiovascular Society Guidelines for the Use of Antiplatelet Therapy. Can J Cardiol 2013;29:1334–1345
5. Eagle KA, Lim MJ, Dabbous OH, et al. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004; 291:2727-33. 18
ACCEPTED MANUSCRIPT
6. Fox KA, Dabbous OH, Goldberg RJ, et al. Prediction of risk of death and myocardial infarction in the six months after presentation with acute coronary syndrome: prospective
RI PT
multinational observational study (GRACE). BMJ 2006;333:1091.
7. Granger CB, Goldberg RJ, Dabbous O, et al. Predictors of hospital mortality in the
SC
Global Registry of Acute Coronary Events. Arch Intern Med 2003;63:2345–53.
M AN U
8. Fox KAA, FitzGerald G, Puymirat E, et al. Should patients with acute coronary disease be stratified for management according to their risk? Derivation, external validation and outcomes using the updated GRACE risk score. BMJ Open 2014;4:e004425.
TE D
doi:10.1136/bmjopen-2013-004425.
9. Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/nonST elevation MI: A method for prognostication and therapeutic decision making. JAMA
EP
2000;284:835-42.
AC C
10. Lindahl B, Toss H, Siegbahn A, Venge P, Wallentin L. Markers of myocardial damage and inflammation in relation to long-term mortality in unstable coronary artery disease. FRISC Study Group. Fragmin during Instability in Coronary Artery Disease. N Engl J Med 2000;343:1139-47.
19
ACCEPTED MANUSCRIPT
11. Liuzzo G, Biasucci LM, Gallimore JR, Grillo RL, Rebuzzi AG, Pepys MB, Maseri A. The prognostic value of C-reactive protein and serum amyloid a protein in severe
RI PT
unstable angina. N Engl J Med 1994;331:417-24.
12. James SK, Lindahl B, Siegbahn A, et al. N-terminal pro-brain natriuretic peptide and other risk markers for the separate prediction of mortality and subsequent myocardial
SC
infarction in patients with unstable coronary artery disease: a Global Utilization of
Strategies To Open occluded arteries (GUSTO)-IV substudy. Circulation 2003;108:275-
M AN U
81.
13. Giraldez RR, Clare RM, Lopes RD, et al. Prevalence and clinical outcomes of undiagnosed diabetes mellitus and prediabetes among patients with high-risk non-ST-
TE D
segment elevation acute coronary syndrome. Am Heart J 2013;165:918-925.
14. Sabatine MS, Morrow DA, de Lemos JA, Omland T, Sloan S, Jarolim P, Solomon
EP
SD, Pfeffer MA, Braunwald E. Evaluation of multiple biomarkers of cardiovascular stress for risk prediction and guiding medical therapy in patients with stable coronary
AC C
disease. Circulation 2012;125:233-40.
15. Zethelius B, Berglund L, Sundström J, Ingelsson E, Basu S, Larsson A, Venge P, Arnlöv J. Use of multiple biomarkers to improve the prediction of death from cardiovascular causes. N Engl J Med 2008;358:2107-16.
20
ACCEPTED MANUSCRIPT
16. Scirica BM, Sabatine MS, Jarolim P, Murphy SA, de Lemos JL, Braunwald E, Morrow DA. Assessment of multiple cardiac biomarkers in non-ST-segment elevation acute coronary syndromes: observations from the MERLIN-TIMI 36 trial. Eur Heart J
RI PT
2011;32:697-705.
17. Morrow DA, Cannon CP, Jesse RL, et al. National Academy of Clinical
SC
Biochemistry Laboratory Medicine Practice Guidelines: clinical characteristics and utilization of biochemical markers in acute coronary syndromes. Circulation
M AN U
2007;115:e356–e375.
18. Yusuf S, Zhao F, Mehta SR, Chrolavicius S, Tognoni G, Fox KK. Effects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without ST-
TE D
segment elevation. N Engl J Med 2001;345:494-502.
19. Sullivan LM, Massaro JM, D'Agostino RB, Sr.: Presentation of multivariate data for
EP
clinical use: The Framingham Study risk score functions. Stat Med 2004;23:1631-60.
AC C
20. Pencina MJ, D'Agostino RB, Sr., Steyerberg EW. Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers. Stat Med 2011;30:11-21.
21. Austin PC, Tu JV. Bootstrapping methods for developing predictive models. The American Statistician. 2004; 58: 131-137
21
ACCEPTED MANUSCRIPT
22. Mehta SR, Granger CB, Boden WE et al. Early versus delayed invasive intervention in acute coronary syndromes. N Engl J Med 2009; 360:2165-2175.
RI PT
23. Eggers KM, Lagerqvist B, Venge P, Wallentin L, Lindahl B. Prognostic value of
biomarkers during and after non-ST-segment elevation acute coronary syndrome. J Am
SC
Coll Cardiol 2009;54:357–364.
24. Wallentin L, Becker RC, Budaj A, et al. Ticagrelor versus Clopidogrel in Patients
M AN U
with Acute Coronary Syndromes, N Engl J Med 2009; 361:1045-1057.
25. Wiviott SD, Braunwald E, McCabe CH, et al. Prasugrel versus Clopidogrel in
AC C
EP
TE D
Patients with Acute Coronary Syndromes. N Engl J Med 2007; 357:2001-2015.
22
ACCEPTED MANUSCRIPT
Figure Legends Figure 1. One year incidence of CV death, MI or stroke, stratified by new OASIS
RI PT
(Organization to Assess Strategies in acute Ischemic Syndromes) risk score
Figure 2. Effect of clopidogrel versus placebo stratified by risk using the OASIS risk
SC
score.
Figure 3. Receiver operating curves for CV death, MI or stroke at 1 year for new OASIS
M AN U
(Organization to Assess Strategies in acute Ischemic Syndromes), GRACE (Global Registry of Acute Coronary Events) and TIMI (Thrombolysis In Myocardial Infarction)
AC C
EP
TE D
risk scores.
23
ACCEPTED MANUSCRIPT
Table 1. Baseline clinical characteristics, electrocardiographic findings, medications and biomarker levels in patients with and without a cardiovascular (CV) event. With CV Death, MI or Stroke
No CV Death, MI or Stroke
(N=686)
(N=5761)
67.94 (10.73)
63.09 (10.98)
<0.0001
Female Sex—no. (%)
259 (37.8)
2403 (41.7)
0.0466
Current Smoker—no. (%)
124 (18.1)
1325 (23.0)
0.0035
Former Smoker—no. (%)
303 (44.2)
2154 (37.4)
0.0005
1217 (21.1)
<0.0001
464 (67.6)
3512 (61.0)
0.0007
46 (6.7)
198 (3.4)
<0.0001
304 (44.3)
1792 (31.1)
<0.0001
80 (11.7)
374 (6.5)
<0.0001
367 (53.5)
2423 (42.1)
<0.0001
23 (3.4)
162 (2.8)
0.4226
5 (0.7)
36 (0.6)
0.7461
140 (20.4)
1559 (27.1)
0.0002
535 (78.0)
4180 (72.6)
0.0024
C reactive protein (IR)
6.4 (14.2)
4.5 (9.2)
<0.0001
Troponin T (IR)
0.14 (0.59)
0.01 (0.29)
<0.0001
BNP (IR)
165.3 (430.0)
53.6 (142.0)
<0.0001
Hemoglobin A1c (IR)
0.06 (0.012)
0.058 (0.008)
<0.0001
History of: 221 (32.2)
Hypertension—no. (%) Stroke—no. (%) Myocardial Infarction—no. (%)
ECG:
TE D
Heart Failure—no. (%)
ST depression≥1 mm
EP
ST elevation≤1 mm
Transient ST elevation > 2 mm
AC C
T wave inversion>2 mm Any of the above
RI PT
M AN U
Diabetes—no. (%)
SC
Age—year (SD)
P value
Biomarkers
24
ACCEPTED MANUSCRIPT
Table 2. Adjusted relative hazards of CV death, MI or stroke in the overall population and according to troponin and diabetes status. Biomarker
CV Death/MI/Stroke Hazard Ratio (CI)
P value
CV Death Hazard Ratio (CI)
MI P value
Hazard Ratio (CI)
Stroke P value
Hazard Ratio (CI)
P value
All Patients 1.73(1.55-1.92)
HbA1C HS CRP
<0.001
2.26(1.95-2.62)
<0.001
1.36(1.18-1.56)
1.28(1.21-1.35)
<0.001
1.3(1.2-1.4)
<0.001
1.22(1.12-1.31)
1.11(1.02-1.2)
0.016
1.34(1.2-1.5)
<0.001
0.94(0.84-1.06)
1.67(1.25-2.23)
0.001
<0.001
1.36(1.22-1.51)
<0.001
0.305
1.02(0.80-1.28)
0.898
NT-proBNP
1.66(1.42-1.93)
<0.001
2.29(1.85-2.83)
<0.001
1.26(1.03-1.56)
0.025
1.82(1.16-2.87)
0.009
HbA1C
1.24(1.12-1.37)
<0.001
1.28(1.12-1.46)
<0.001
1.14(0.99-1.31)
0.066
1.21(0.9-1.62)
0.212
HS CRP
1.12(1-1.25)
0.049
1.44(1.25-1.67)
<0.001
0.9(0.78-1.05)
0.195
1.07(0.77-1.47)
0.694
NT-proBNP
1.79(1.55-2.07)
<0.001
2.23(1.8-2.75)
<0.001
1.43(1.18-1.73)
<0.001
1.59(1.08-2.33)
0.018
HbA1C
1.29(1.22-1.37)
<0.001
1.32(1.20-1.45)
<0.001
1.24(1.14-1.35)
<0.001
1.39(1.24-1.56)
<0.001
HS CRP
1.10(0.97-1.24)
0.128
1.22(1.03-1.46)
0.021
0.99(0.84-1.17)
0.944
0.96(0.69-1.35)
0.831
NT-proBNP
1.53(1.28-1.83)
<0.001
1.95(1.50-2.53)
<0.001
HbA1C
1.25(0.98-1.58)
0.069
1.28(0.81-2.05)
0.294
1.26(1.11-1.43)
<0.001
1.34(1.12-1.6)
HS CRP
1.03(0.89-1.20)
0.662
1.34(1.09-1.65)
NT-proBNP
1.79(1.57-2.04)
<0.001
2.38(1.99-2.86)
HbA1C
1.28(1.20-1.38)
<0.001
1.31(1.20-1.44)
HS CRP
1.12(1.01-1.23)
0.030
1.34(1.17-1.52)
NT-proBNP
1.57(1.38-1.8)
<0.001
HbA1C
1.22(1.14-1.31)
<0.001
HS CRP
1.1(1-1.22)
0.058
NT-proBNP
2.01(1.68-2.41)
HbA1C HS CRP
Diabetes +ve
M AN U
Troponin Negative
SC
Troponin Positive
<0.001
RI PT
NT-proBNP
1.12(0.94-1.33)
0.207
1.50(1.1-2.04)
0.011
0.83(0.67-1.02)
0.080
1.02(0.69-1.52)
0.915
<0.001
1.38(1.16-1.64)
<0.001
1.86(1.28-2.71)
0.001
<0.001
1.22(1.10-1.36)
<0.001
1.35(1.16-1.58)
<0.001
<0.001
0.97(0.85-1.12)
0.711
0.97(0.73-1.3)
0.865
2.05(1.71-2.47)
<0.001
1.21(1.01-1.44)
0.034
1.67(1.13-2.48)
0.011
1.24(1.13-1.37)
<0.001
1.14(1.02-1.28)
0.023
1.31(1.13-1.52)
<0.001
1.36(1.18-1.56)
<0.001
0.93(0.81-1.07)
0.300
1.04(0.76-1.41)
0.820
<0.001
2.69(2.07-3.51)
<0.001
1.63(1.28-2.06)
<0.001
1.72(1.1-2.69)
0.017
1.44(1.3-1.6)
<0.001
1.44(1.24-1.67)
<0.001
1.38(1.2-1.58)
<0.001
1.55(1.22-1.97)
<0.001
1.11(0.97-1.27)
0.139
1.3(1.08-1.57)
0.005
0.98(0.81-1.18)
0.800
1(0.71-1.42)
0.988
Female
AC C
Male
EP
Diabetes –ve
TE D
0.001
0.006
25
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Adjusted for treatment, age, sex and troponin T ≥2X upper limit of normal, history of vascular event (stroke or MI), ST segment depression or elevation BNP=brain natriuretic peptide; HbA1C=hemoglobin A1c; HS CRP=high sensitivity C-reactive protein; HR= hazard ratio, CI refers to 95% confidence intervals
26
ACCEPTED MANUSCRIPT
Table 3. Multivariable Cox proportional hazards model for prediction of CV death, MI or stroke at 1 year Hazard Ratio
95% Confidence Interval
P Value
Age (per year)
19.05
1.02
1.01-1.03
<0.0001
ST segment deviation
7.05
1.26
Prior vascular event (prior stroke or MI)
9.23
1.30
Male
7.90
Troponin T
3.45
Hemoglobin A1C** NT-proBNP**
SC
RI PT
Chi-Square
0.008
1.10-1.54
0.002
1.29
1.08-1.53
0.005
1.20
0.99-1.44
0.063
72.48
1.27
1.20-1.34
<0.0001
97.22
1.70
1.53-1.88
<0.0001
M AN U
1.06-1.50
TE D
Variable
AC C
EP
MI-myocardial infarction; ULN=upper limit of normal; BNP=Brain Natriuretic Peptide **Biomarkers were presented as 1 standard deviation above the mean on a log transformed measurement, Hazard Ratio confidence interval limits are the profile likelihood limits.
27
ACCEPTED MANUSCRIPT
Table 4. Final Risk Score Variable
Points
+1
60-69
+2
70-79
+3
80+
+4
2. Male
SC
50-59
RI PT
1. Age
+1
4. ST segment deviation 5. Troponin T 6. Hemoglobin 1C
+1
+1
0
TE D
<0.055
+1
M AN U
3. Prior Vascular Event (MI or stroke)
0.055-0.063
+1
>0.064
+3
EP
5. NT-proBNP
0
20.41 – 59.59
+3
59.6 – 177.71
+5
>177.72
+9
AC C
<20.41
Score Range
0-20
28
ACCEPTED MANUSCRIPT
Figure 1. One year incidence of CV death, MI or stroke, stratified by OASIS risk score
RI PT SC
10
CV Death at 1 year (%)
11.1
5
3.6
1.1
0
3.7
M AN U
9.1
12-20 High Risk
5
Risk Score
4 Stroke at 1 year (%)
10
8.8
MI at 1 year (%)
8-11 Intermediate Risk
Stroke
AC C
Myocardial Infarction
0-7 Low Risk
5
5.7
3
15
Risk Score
12-20 High Risk
TE D
8-11 Intermediate Risk
EP
0-7 Low Risk
2.1 2
0
5
10
15
17.8
1.4
1
2.8
0.4
0
29 0-7 Low Risk
8-11 Intermediate Risk Risk Score
12-20 High Risk
0
CV Death,MI or Stroke at 1 year (%)
CV Death
15
20
CV Death, MI or Stroke
0-7
8-11
12-20
Low Risk
Intermediate Risk Risk Score
High Risk
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Figure 2. Effect of clopidogrel versus placebo stratified by risk using the OASIS risk score.
30
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Figure 3. Receiver operating curves for CV death, MI or stroke at 1 year for new OASIS risk score, as well as the GRACE and TIMI risk scores.
OASIS=Organization to Assess Strategies in acute Ischemic Syndromes GRACE= Global Registry of Acute Coronary Events TIMI=Thrombolysis In Myocardial Infarction
31