Electrocardiographic abnormalities and coronary artery calcium for coronary heart disease prediction and reclassification: The Multi-Ethnic Study of Atherosclerosis Chintan S. Desai MD, MS, Hongyan Ning MD, MS, Elsayed Z. Soliman MD, MSc, Gregory L. Burke MD, MSc, Steven Shea MD, MS, Saman Nazarian MD, M.D,. ScM Donald M. Lloyd-Jones, Philip Greenland MD PII: DOI: Reference:
S0002-8703(14)00358-5 doi: 10.1016/j.ahj.2014.06.009 YMHJ 4655
To appear in:
American Heart Journal
Received date: Accepted date:
30 May 2013 2 June 2014
Please cite this article as: Desai Chintan S., Ning Hongyan, Soliman Elsayed Z., Burke Gregory L., Shea Steven, Nazarian Saman, Donald M. Lloyd-Jones M.D,. ScM, Greenland Philip, Electrocardiographic abnormalities and coronary artery calcium for coronary heart disease prediction and reclassification: The Multi-Ethnic Study of Atherosclerosis, American Heart Journal (2014), doi: 10.1016/j.ahj.2014.06.009
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 Electrocardiographic abnormalities and coronary artery calcium for coronary heart
T
disease prediction and reclassification: The Multi-Ethnic Study of Atherosclerosis
RI P
Chintan S. Desai, MD, MS; Hongyan Ning, MD, MS; Elsayed Z. Soliman, MD, MSc; Gregory L. Burke, MD, MSc; Steven Shea, MD, MS; Saman Nazarian, MD;
SC
Donald M. Lloyd-Jones, MD, ScM; Philip Greenland, MD
NU
From: Johns Hopkins University School of Medicine, Department of Medicine (C.S.D., S.N.);
MA
Northwestern University Feinberg School of Medicine, Department of Preventive Medicine (H.N., D.M.L.-J., P.G.); Wake Forest School of Medicine, Division of Public Health Sciences
ED
(E.Z.S., G.L.B.); Columbia University Mailman School of Public Health (S.S.);
PT
Running title: ECG and CAC for coronary heart disease risk prediction Abstract word count: 250
Tables: 5
CE
Word count: 4957
Address for correspondence:
AC
Philip Greenland, MD
Harry W. Dingman Professor Departments of Preventive Medicine and Medicine 680 N. Lake Shore Drive, 14th Floor Chicago, IL 60611 E-mail: [email protected] Phone: 312-908-7914 Fax: 312-908-9588
1
Figures: 1
ACCEPTED MANUSCRIPT
ABSTRACT
RI P
T
BACKGROUND: Electrocardiographic (ECG) abnormalities and coronary artery calcium (CAC) identify different aspects of subclinical coronary heart disease (CHD). We sought to determine whether ECG abnormalities improve risk prediction for all CHD and fatal CHD events jointly with
SC
CAC measures.
NU
METHODS: We included 6406 men and women from the Multi-Ethnic Study of Atherosclerosis (MESA) aged 45-84 years who were free of cardiovascular disease at the time of enrollment
MA
(2000-02). We stratified participants by presence of ST-T and Q wave abnormalities: any major, any minor/no major, no major/minor using the Minnesota Code classifications. CAC was defined
ED
into one of the following strata: 0, 1-100, 101-300, > 300. We created risk prediction models using MESA-specific coefficients for traditional risk factors (RFs) and calculated categorical net
PT
reclassification improvement (NRI) for all and fatal CHD.
CE
RESULTS: Over a median follow up of 10 years, we observed that addition of ECG abnormalities to a risk prediction model for all CHD resulted in a categorical NRI of 0.05 (P =
AC
0.04). For fatal CHD alone, the addition of ECG abnormalities resulted in categorical NRI of 0.09 (P = 0.02). Addition of ECG abnormalities to a model containing RFs and CAC resulted in categorical NRI of 0.02 (P = 0.11) for all CHD events. We also observed differences in the association between ECG abnormalities and CHD when stratifying by CAC presence. CONCLUSION: ECG abnormalities improved risk prediction for CHD when added to RFs but not when added to CAC. ECG abnormalities particularly improved risk prediction for fatal CHD.
2
ACCEPTED MANUSCRIPT
INTRODUCTION
T
Clinic-based cardiovascular disease prevention strategies are targeted towards
RI P
individuals based on their estimated absolute risk for disease.1,2 Risk prediction algorithms such as the Framingham Risk Score and the American Heart Association/American College of
SC
Cardiology (AHA/ACC) risk calculator3 are recommended for use in clinical practice but have limitations in correctly stratifying the risk for coronary events.4-6 Prospective epidemiologic
NU
investigations have demonstrated that abnormalities on the resting 12-lead electrocardiogram (ECG) are independently associated with increased incidence of coronary heart disease (CHD)
MA
events7-11 and recent data in older adults suggest that ECG may have modest incremental value in risk prediction beyond traditional risk factors.12 However, there are limited data on the utility of
ED
ECG for risk re-classification in middle and older-aged adults.
PT
A consensus document from the American Heart Association/American College of Cardiology Foundation states that coronary artery calcium (CAC) measurement is reasonable
CE
for risk stratification in asymptomatic adults at intermediate risk for CHD.13 CAC testing has been shown to improve risk classification for CHD compared to risk factors alone.14 Recent
AC
evidence suggests that minor ECG abnormalities may be associated more strongly with fatal arrhythmic events, but not for non-fatal CHD events such as non-fatal MI.15 Therefore, it is possible that ECG abnormalities may identify different aspects of risk than CAC. The objective of our study is to determine if the addition of ECG abnormalities (ST-T abnormalities and Q waves) to a CHD risk prediction model based on traditional risk factors improves risk classification. We also sought to determine if ECG improves risk prediction when added to a model that also includes CAC. In addition, since CAC and ECG may detect divergent elements of the risk substrate, we sought to assess for risk prediction improvement separately for fatal and nonfatal CHD. METHODS 3
ACCEPTED MANUSCRIPT Cohort We studied participants from the Multi-Ethnic Study of Atherosclerosis (MESA); details of
T
the study design have been published.16 MESA recruited healthy participants without clinical
RI P
evidence or history of any cardiovascular disease at baseline in 2000-2002 from the following six communities in the United States: Baltimore, Maryland; Chicago, Illinois; Forsyth County,
SC
North Carolina; Los Angeles County, California; northern Manhattan, New York; St. Paul, Minnesota. Participants were aged 45-84 years at the time of the baseline examination and self-
NU
identified into one of four race-ethnic groups: white (38%), black (28%), Hispanic (22%), and
MA
Chinese (12%). We assessed age, sex, height, weight, body mass index, systolic and diastolic blood pressure, total and high-density lipoprotein (HDL) cholesterol, fasting blood glucose, selfreported diabetes, smoking status, and self-reported antihypertensive or lipid-lowering
ED
medication use at the baseline examination. We excluded participants who were missing these
PT
key covariates (N = 59). Electrocardiography
CE
A resting ECG was recorded at the baseline examination using a MAC 1200 ECG machine (Marquette Electronics, Milwaukee, WI, USA) at all field centers. All ECGs were read
AC
at a central laboratory (EPICARE, Wake Forest School of Medicine, Winston Salem, NC) and visually inspected for adequate quality and technical errors. ECG abnormalities were classified using the Minnesota ECG code (MC).17 Major ECG abnormalities were defined as follows: major Q wave abnormalities (MC 1.1.x, 1.2.x), minor Q/QS with major ST-T abnormalities (1.3.x, 4.1.x, 4.2, 5.1, 5.2), and major isolated ST-T abnormalities (MC 4.1.x, 4.2, 5.1, 5.2). Minor ECG abnormalities were defined as follows: minor isolated Q waves (MC 1.3.x) and minor isolated ST-T (MC 4.3, 4.4, 5.3, 5.4). We classified participants into three groups: 1) any major ECG abnormalities; 2) any minor but no major abnormalities; and 3) no major or minor abnormalities. We excluded participants who were missing ECG data (N = 49) and who had ECG diagnoses that suppressed the coding of ST-T or Q wave abnormalities (N = 300). Atrial fibrillation was an 4
ACCEPTED MANUSCRIPT exclusion criterion for MESA because gated cardiac computed tomography scanning was used to measure CAC; therefore, rhythm abnormalities were not included in our classification.
T
Coronary artery calcium
RI P
The MESA coronary artery calcium (CAC) protocol for acquisition and interpretation has been described previously.18 Each participant was scanned twice and the scans were
SC
interpreted at a central reading laboratory. The Agatston method was used to calculate the CAC score, and the average of the two CAC scores was used.19 For the present study, we
MA
Ascertainment and adjudication of outcomes
NU
categorized CAC into one of the following strata: 0, 1-100, 101-300, or > 300.
At intervals of 9-12 months, interviewers contacted MESA participants or family members to inquire about interim hospital admissions, cardiovascular diagnoses, and deaths.
ED
To verify self-reported diagnoses, we requested copies of outpatient and inpatient medical
PT
records, as well as death certificates. Two trained physicians reviewed the records and verified diagnoses. If disagreements persisted after review, a full MESA morbidity and mortality
CE
committee reviewed the records for final adjudication. The primary outcome for our analysis was coronary heart disease (CHD). For the
AC
present analysis, CHD included nonfatal myocardial infarction (NFMI), death due to CHD, and hospitalized angina. NFMI was defined based on a combination of symptoms, ECG findings, and levels of cardiac biomarkers. Fatal CHD required myocardial infarction within 28 days of death, chest pain within 72 hours of death, or a known history of CHD in the absence of nonatherosclerotic and non-cardiac causes of death. Angina was defined as definite, probable, or absent based on documentation of clinical signs and symptoms. For our analyses, we included a composite outcome of definite or probable NFMI, CHD death, and hospitalized angina; we refer to the composite outcome as “all CHD.” We also analyzed definite/probable fatal CHD events and nonfatal CHD events separately. Additional details on MESA clinical outcomes are available online at http://www.mesa-nhlbi.org/. 5
ACCEPTED MANUSCRIPT Statistical analysis After exclusions, 6406 participants were eligible for our analyses. We compared
T
unadjusted baseline characteristics by ECG status; chi-square was used to compare differences
RI P
for categorical variables and ANOVA was used to compare means for continuous variables. We used Cox proportional hazards modeling to determine the association between ECG
SC
abnormalities and incident CHD. ECG abnormalities were modeled as a three-tiered variable. The presence of 1) any major abnormalities and 2) minor abnormalities only were each
NU
considered as separate variables in the model, and 3) no major or minor abnormalities was the
MA
reference group for the variable. We adjusted for age, race, sex, systolic blood pressure, diabetes, total cholesterol, HDL cholesterol, smoking status, and use of antihypertensive medication.
ED
We calculated the predicted 10-year risk for all CHD using MESA-specific coefficients for
PT
age, race, sex, systolic blood pressure, diabetes, total cholesterol, HDL cholesterol, smoking status, and use of antihypertensive medication. This was considered the baseline risk prediction
CE
model for CHD. We then determined the net reclassification improvement (NRI) with the addition of ECG abnormalities to the baseline model. The NRI was calculated using the
AC
following formula described by Pencina et al:20 (Number of events reclassified higher – Number of events reclassified lower)/total events + (Number of nonevents reclassified lower – Number of nonevents reclassified higher)/total nonevents
We used this calculation for the following 10-year risk categories for CHD: < 5%, 5-10%, and ≥ 10%. We also calculated a category-free NRI.21 We then used MESA-specific coefficients for traditional risk factors to calculate 10-year risk for fatal and nonfatal CHD separately. We used the same 10-year risk categories to calculate a categorical and a continuous NRI separately for fatal and nonfatal CHD. We calculated Harrell’s C-statistic for the baseline models containing 6
ACCEPTED MANUSCRIPT traditional risk factors and the C-statistic with the addition of ECG abnormalities.22 We also calculated the category-free NRI for CHD in only those participants who were in the
T
intermediate risk group (5-10%).
RI P
To assess for confounding and interaction between ECG abnormalities and CAC in the association with CHD, we calculated hazards ratios for incident CHD by ECG status separately
SC
among MESA participants by CAC strata. We then included CAC in the baseline risk prediction model and determined the categorical and continuous NRI for addition of ECG to the model.
NU
SAS version 9.3 was used for all analyses (Cary, NC). Two-sided P-values < 0.05 were
MA
considered statistically significant. Funding
This work was supported by the National Heart, Lung, and Blood Institute [N01-HC-
ED
95159 through N01-HC-95169] and by the National Center for Research Resources [UL1-RR-
RESULTS
CE
Baseline characteristics
PT
024156 and UL1-RR-025005]. No extramural funding was used to support this work.
Among the 6406 participants, 385 (6%) had any major ECG abnormalities and 897
AC
(14%) had any minor/no major ECG abnormalities; the remainder had no major or minor abnormalities. The baseline characteristics of the study cohort are shown in Table 1, stratified by ECG status. Participants with ECG abnormalities were older and more likely to be black, compared to participants with no major/minor abnormalities. Participants with ECG abnormalities also had higher systolic blood pressure, higher prevalence of diabetes, and were more likely to be taking antihypertensive and lipid-lowering medications. The prevalence of CAC > 0 among participants with no ECG abnormalities was 47%, compared to 56% and 63% in those with only minor or any major abnormalities, respectively (P < 0.001). Event rates and hazards ratios
7
ACCEPTED MANUSCRIPT Over a median of 10 years follow up, a total of 405 CHD events occurred. Table 2 shows the event rates and multivariable-adjusted hazards ratios for all CHD by ECG status. The
T
CHD event rates per 1000 person-years in participants with no major/minor, only minor, and any
RI P
major ECG abnormalities were 5.6, 11.0, and 15.4, respectively. After adjustment for clinical and demographic covariates, the hazards ratios for all CHD for participants with only minor and
SC
any major ECG abnormalities were 1.56 (95% CI, 1.22-1.99) and 1.81 (1.32-2.47), respectively, compared to those with no major/minor abnormalities. Table 2 also shows all CHD event rates
NU
and hazards ratios further stratified by CAC score. Among participants with a CAC score of 0,
MA
the presence of only minor ECG abnormalities did not increase the risk for CHD, compared to those with no ECG abnormalities. Participants with CAC score = 0 and major ECG abnormalities had a 3-fold increased risk for all CHD, compared to those with no ECG
ED
abnormalities. The magnitude of the hazards ratio for major ECG abnormalities and all CHD
PT
was lower in the other CAC groups.
Table 3 shows the event rates and hazards ratios for fatal and nonfatal CHD separately.
CE
The overall hazards ratios in all participants for major ECG abnormalities were similar for all CHD, fatal CHD, and nonfatal CHD. Compared to those with no major/minor abnormalities, the
AC
hazards ratio in those with only minor ECG abnormalities was 2.42 (1.44-4.09) for fatal CHD and 1.38 (1.04-1.83) for nonfatal CHD. Across all CAC strata, the hazards ratios for the association between fatal CHD and minor ECG abnormalities had a greater magnitude than those for nonfatal CHD. The Figure shows CHD event rates by ECG status and by presence or absence of CAC. Among participants with CAC = 0, the CHD event rates were similar for those with only minor ECG abnormalities and no major/minor abnormalities. Among participants with CAC > 0, the CHD event rates were substantially higher for those with minor ECG abnormalities, and were similar to rates in participants with any major ECG abnormalities. Discrimination and reclassification 8
ACCEPTED MANUSCRIPT Table 4 shows the reclassification table for predicted and observed risk for all CHD with addition of ECG abnormalities to the baseline model. The C-statistic for the baseline model was
T
0.727 and the C-statistic with the addition of ECG abnormalities was 0.736; the P-value for the
RI P
difference was 0.02. For the categorical NRI, 4% of events were correctly reclassified and 2% of nonevents were correctly reclassified; the resulting NRI was 0.053 (P = 0.02). The category-free
SC
NRI was 0.30 (P < 0.001).
We then determined the improvement in risk prediction for fatal and nonfatal CHD
NU
events separately. For fatal CHD events, the addition of ECG to the baseline risk prediction
MA
model resulted in a categorical NRI = 0.09 (P = 0.08) and continuous NRI = 0.47 (P < 0.001). For nonfatal CHD events, the addition of ECG to the baseline risk prediction model resulted in a categorical NRI = 0.02 (P = 0.35) and continuous NRI = 0.22 (P < 0.001).
ED
Table 5 shows the cross-tabulations for predicted and observed CHD risk with addition
PT
of ECG to the baseline model, which includes risk factors and CAC strata. The C-statistic for the baseline risk prediction model was 0.783; with addition of ECG, the C-statistic was 0.788 (P-
CE
value for difference = 0.03). The overall categorical NRI was 0.02 (P = 0.11) and the continuous NRI was 0.26 (P < 0.001). For fatal CHD, the addition of ECG abnormalities to the baseline
AC
model including risk factors and CAC strata resulted in a categorical NRI of 0.08 (P = 0.15) and category-free NRI of 0.48 (P < 0.001). For nonfatal CHD, the addition of ECG abnormalities to a baseline model including risk factors and CAC resulted in a categorical NRI of 0.01 (P = 0.66) and category-free NRI of 0.16 (P = 0.04). Among the 1853 participants with predicted 10-year all CHD risk 5-10%, the addition of ECG abnormalities to the baseline model including only traditional risk factors resulted in a category-free NRI = 0.23 (P = 0.01). The magnitude of the category-free NRI was similar with addition of ECG to a baseline model containing risk factors and CAC (category-free NRI = 0.23, P = 0.01). DISCUSSION 9
ACCEPTED MANUSCRIPT Findings In MESA participants, the addition of ECG abnormalities to a risk prediction model based
T
on traditional risk factors resulted in modestly improved reclassification for all CHD. Addition of
RI P
ECG abnormalities to a model including traditional risk factors and CAC did not result in improved risk reclassification. We observed differences in hazards ratios for CHD when
SC
stratifying by CAC presence, suggesting the possibility of interaction. The observation of a
for fatal CHD, rather than nonfatal CHD.
MA
Implications
NU
modestly improved classification for all CHD with ECG was driven by improved reclassification
In the present study, we analyzed the improvement in CHD risk prediction with the use of easily identified ST-T and Q wave abnormalities. In comparison with other biomarkers, the
ED
magnitude of the NRI (categorical NRI = 0.053, continuous NRI = 0.29) is similar to that
PT
achieved by ankle-brachial index and high-sensitivity C-reactive protein.23 To the best of our knowledge, our study is the first to examine CHD risk prediction jointly with ECG and CAC. The
CE
association between ECG abnormalities and CHD was not attenuated after adjustment for the presence of CAC; thus, CAC is not a confounder in this association. When we stratified by CAC
AC
presence, we observed differences in the hazards ratios for CHD. In participants with no detectable CAC, those with only minor ECG abnormalities are not at increased risk for CHD compared to those with no major/minor ECG abnormalities. Major ECG abnormalities, however, were associated with a 3-fold increased risk for CHD even in the absence of CAC. Among participants with CAC score > 0, those with minor and major ECG abnormalities had a similar risk for incident CHD. These findings suggest that the association between ECG abnormalities and CHD is modified by CAC presence. At each category of ECG abnormality (no major/minor, only minor, any major), the absolute magnitude of risk was greater in participants with CAC than in those without CAC.
10
ACCEPTED MANUSCRIPT The distinguishing feature of ECG abnormalities in our study is the superior performance in risk prediction for fatal CHD (categorical NRI = 0.09, continuous NRI = 0.48) compared with
T
nonfatal CHD (categorical NRI = 0.02, continuous NRI = 0.16). These findings are consistent
RI P
with a previous analysis of the Cardiovascular Health Study, which found that minor isolated STT abnormalities were associated with a nearly 2-fold increase in risk for incident fatal CHD, but
SC
were not associated with increased risk for nonfatal CHD.15 The potential association between ECG abnormalities and arrhythmic events could be further tested in clinical trials involving
NU
agents known to modify the risk of arrhythmia, such as beta-adrenergic receptor antagonists.
MA
The clinical implications of moving to a higher risk category primarily involve the use of statin therapy. The recent AHA/ACC guidelines for primary prevention recommend initiation of statin therapy for primary prevention when the 10-year risk of atherosclerotic cardiovascular
ED
disease exceeds 7.5% in individuals 40 years or older.2 Many of the individuals in the
PT
intermediate group in our study (10-year risk of CHD only 5-10%) may already meet that threshold. However, for patients who are borderline, or when the clinician desires additional
CE
information, ECG may be a reasonable test for additional risk stratification in these selected individuals.
AC
One limitation of our study is that it is a secondary analysis of observational data. However, MESA is a well-phenotyped cohort comprised of individuals with validated ECG and outcome measures. Furthermore, our results are concordant with previous observations from the Cardiovascular Health Study.15 A strength of this study is that we used MESA-specific coefficients to calculate our baseline risk scores, thus reducing the possibility of observing a falsely high NRI. Conclusions In a middle and older-aged, multi-ethnic cohort, we observed that major and minor ST-T and Q wave abnormalities offered modestly improved CHD risk prediction beyond traditional risk factors, particularly for fatal CHD. We also observed differences in the association between 11
ACCEPTED MANUSCRIPT ECG and CHD when stratifying by CAC presence, suggesting the possibility of interaction
AC
CE
PT
ED
MA
NU
SC
RI P
T
between ECG and CAC.
12
ACCEPTED MANUSCRIPT
CONTRIBUTIONS
T
The authors thank the other investigators, the staff, and the participants of the MESA study for
RI P
their valuable contributions. A full list of participating MESA investigators and institutions can be
SC
found at http://www.mesa-nhlbi.org.
The authors are solely responsible for the design and conduct of this study, all study analyses,
AC
CE
PT
ED
MA
NU
the drafting and editing of the paper, and its final contents.
13
ACCEPTED MANUSCRIPT
REFERENCES
AC
CE
PT
ED
MA
NU
SC
RI P
T
1. Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III guidelines. Circulation 2004;110:227-39. 2. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013. 3. Goff DC, Jr., Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013. 4. Berry JD, Lloyd-Jones DM, Garside DB, Greenland P. Framingham risk score and prediction of coronary heart disease death in young men. Am Heart J 2007;154:80-6. 5. Cavanaugh-Hussey MW, Berry JD, Lloyd-Jones DM. Who exceeds ATP-III risk thresholds? Systematic examination of the effect of varying age and risk factor levels in the ATP-III risk assessment tool. Prev Med 2008;47:619-23. 6. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013;382:1762-5. 7. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: final report of the pooling project. The pooling project research group. J Chronic Dis 1978;31:201-306. 8. Kannel WB. Common electrocardiographic markers for subsequent clinical coronary events. Circulation 1987;75:II25-7. 9. Liao YL, Liu KA, Dyer A, et al. Major and minor electrocardiographic abnormalities and risk of death from coronary heart disease, cardiovascular diseases and all causes in men and women. J Am Coll Cardiol 1988;12:1494-500. 10. Dekker JM, Schouten EG, Klootwijk P, Pool J, Kromhout D. ST segment and T wave characteristics as indicators of coronary heart disease risk: the Zutphen Study. J Am Coll Cardiol 1995;25:1321-6. 11. Sutherland SE, Gazes PC, Keil JE, Gilbert GE, Knapp RG. Electrocardiographic abnormalities and 30-year mortality among white and black men of the Charleston Heart Study. Circulation 1993;88:268592. 12. Auer R, Bauer DC, Marques-Vidal P, et al. Association of major and minor ECG abnormalities with coronary heart disease events. JAMA 2012;307:1497-505. 13. Greenland P, Bonow RO, Brundage BH, et al. ACCF/AHA 2007 clinical expert consensus document on coronary artery calcium scoring by computed tomography in global cardiovascular risk assessment and in evaluation of patients with chest pain: a report of the American College of Cardiology Foundation Clinical Expert Consensus Task Force (ACCF/AHA Writing Committee to Update the 2000 Expert Consensus Document on Electron Beam Computed Tomography) developed in collaboration with the Society of Atherosclerosis Imaging and Prevention and the Society of Cardiovascular Computed Tomography. Journal of the American College of Cardiology 2007;49:378-402. 14. Polonsky TS, McClelland RL, Jorgensen NW, et al. Coronary artery calcium score and risk classification for coronary heart disease prediction. JAMA 2010;303:1610-6. 15. Kumar A, Prineas RJ, Arnold AM, et al. Prevalence, prognosis, and implications of isolated minor nonspecific ST-segment and T-wave abnormalities in older adults: Cardiovascular Health Study. Circulation 2008;118:2790-6. 16. Bild DE, Bluemke DA, Burke GL, et al. Multi-ethnic study of atherosclerosis: objectives and design. American journal of epidemiology 2002;156:871-81. 14
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
17. Tuinstra CL, Rautaharju PM, Prineas RJ, Duisterhout JS. The performance of three visual coding procedures and three computer programs in classification of electrocardiograms according to the Minnesota Code. Journal of electrocardiology 1982;15:345-50. 18. Carr JJ, Nelson JC, Wong ND, et al. Calcified coronary artery plaque measurement with cardiac CT in population-based studies: standardized protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) study. Radiology 2005;234:35-43. 19. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Jr., Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990;15:827-32. 20. Pencina MJ, D'Agostino RB, Sr., D'Agostino RB, Jr., Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157-72; discussion 207-12. 21. 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. 22. Harrell FE, Jr., Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests. JAMA 1982;247:2543-6. 23. Yeboah J, McClelland RL, Polonsky TS, et al. Comparison of novel risk markers for improvement in cardiovascular risk assessment in intermediate-risk individuals. JAMA 2012;308:788-95.
ED
FIGURE LEGEND
AC
CE
PT
Figure: Ten-Year Coronary heart disease event rates by ECG and CAC status in the MultiEthnic Study of Atherosclerosis
15
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 1: Baseline characteristics by ECG status Covariate No major/minor Any minor/No major Any major P-value (N = 5124) (N = 897) (N = 385) Age (yrs) 61 ± 10 65 ± 10 67 ± 10 < 0.001 Male sex (N,%) 2370 (46%) 385 (43%) 204 (53%) 0.004 Race (N,%) < 0.001 2015 (39%) 291 (32%) 117 (30%) • Caucasian 1327 (26%) 294 (33%) 142 (37%) • Black 646 (13%) 86 (10%) 50 (13%) • Chinese 1136 (22%) 226 (25%) 76 (20%) • Hispanic Systolic blood 124 ± 20 133 ± 23 138 ± 24 < 0.001 pressure (mm Hg) Body mass index 28 ± 5 29 ± 6 29 ± 5 < 0.001 (kg/m2) Total cholesterol 5.02 ± 0.93 5.04 ± 0.93 4.97 ± 0.96 0.27 (mmol/L) HDL cholesterol 1.32 ± 0.39 1.32 ± 0.39 1.27 ± 0.36 0.04 (mmol/L) Diabetes (N,%) 560 (11%) 160 (18%) 72 (19%) < 0.001 Current smoker (N,%) 664 (13%) 128 (14%) 48 (12%) 0.52 Antihypertensive 1502 (29%) 387 (43%) 207 (54%) < 0.001 medication (N,%) Lipid-lowering 796 (16%) 150 (17%) 83 (22%) 0.007 therapy (N,%) CAC > 0 (N,%) 2407 (47%) 503 (56%) 243 (63%) < 0.001 ECG = electrocardiographic; HDL = high-density lipoprotein; CAC = coronary artery calcium
16
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
Table 2: Event rates and hazards ratios for all CHD events by ECG abnormality No major/minor Only minor Any major (N = 5124) (N = 897) (N = 385) Event rate among 5.57 11.0 15.42 overall cohort (per 1000 P-Y) Stratified by CAC 1.62 1.87 7.49 • CAC = 0 (N = 3253) 5.76 13.40 12.02 • CAC 1-100 (N = 1709) 13.10 13.89 21.65 • CAC 101-300 (N = 687) 21.29 35.02 38.93 • CAC > 300 (N = 757)
AC
CE
PT
ED
Hazards ratios# for 1.0 (ref) 1.56 (1.22-1.99)* 1.81 (1.32-2.47)* all CHD among overall cohort Stratified by CAC 1.0 (ref) 0.97 (0.43-2.29) 3.32 (1.62-6.83)* • CAC = 0 (N = 3253) 1.0 (ref) 2.20 (1.41-3.42)* 1.82 (0.97-3.40) • CAC 1-100 (N = 1709) 1.0 (ref) 0.94 (0.52-1.71) 1.39 (0.67-2.89) • CAC 101-300 (N = 687) 1.0 (ref) 1.55 (1.05-2.29)* 1.81 (1.09-3.02)* • CAC > 300 (N = 757) *denotes P < 0.05 # denotes adjustment for age, race, sex, systolic blood pressure, total and HDL cholesterol, diabetes, smoking, antihypertensive medication CHD = coronary heart disease; ECG = electrocardiographic; CAC = coronary artery calcium; P-Y = person-years
17
ACCEPTED MANUSCRIPT
NU
SC
RI P
T
Table 3: Fatal and nonfatal CHD event rates and hazards ratios by ECG abnormalities and CAC No major/minor Only minor Any major (N = 5124) (N = 897) (N = 385) Fatal CHD event rate among 0.79 2.67 2.29 overall cohort (per 1000 P-Y) Stratified by CAC score 0.29 0.77 1.43 • CAC = 0 (N = 3253) 0.90 2.58 0.91 • CAC 1-100 (N = 1709) 1.40 5.57 4.12 • CAC 101-300 (N = 687) 2.58 6.14 4.88 • CAC > 300 (N = 757) Hazards ratios# for fatal CHD among all MESA participants Stratified by CAC score • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709) • CAC 101-300 (N = 687)
CAC > 300 (N = 757)
1.55 (0.72-3.37)
2.53 (0.49-13.00) 0.58 (0.07-4.62) 2.19 (0.42-11.47)
1.0 (ref)
2.15 (0.53-8.65) 2.51 (0.91-6.96) 3.77 (1.2111.75)* 2.32 (0.93-5.78)
4.74
8.12
12.95
1.32 4.80 11.57 18.28
1.07 10.62 7.94 27.65
5.99 11.02 16.84 32.33
1.0 (ref)
1.38 (1.041.83)*
1.83 (1.30-2.58)*
1.0 (ref) 1.0 (ref)
0.68 (0.24-1.94) 2.13 (1.303.49)* 0.63 (0.30-1.34) 1.42 (0.92-2.19)
3.50 (1.57-7.82)* 2.14 (1.10-4.17)*
MA
CE
PT
Nonfatal CHD event rate/1000 P-Y Stratified by CAC score • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709) • CAC 101-300 (N = 687) • CAC > 300 (N = 757)
AC
2.42 (1.444.09)*
1.0 (ref) 1.0 (ref) 1.0 (ref)
ED
•
1.0 (ref)
Hazards ratios# for nonfatal CHD among all MESA participants Stratified by CAC score: • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709)
1.85 (0.51-6.73)
1.0 (ref) 1.26 (0.55-2.87)* • CAC 101-300 (N = 687) 1.0 (ref) 1.79 (1.03-3.12)* • CAC > 300 (N = 757) *denotes P < 0.05 # denotes adjustment for age, race, sex, systolic blood pressure, total and HDL cholesterol, diabetes, smoking, antihypertensive medication CHD = coronary heart disease; ECG = electrocardiographic; CAC = coronary artery calcium; P-Y = person-years
18
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 4: Reclassification table for CHD events in models with and without ECG 10-year risk in model with ECG 10-year risk in 0-5% 5-10% ≥ 10% Total model without ECG 0-5% 3058 173 0 3231 Cases (N, %) 62 (2%) 20 (12%) 0 82 (3%) 5-10% 246 1394 213 1853 Cases (N, %) 14 (6%) 93 (7%) 30 (14%) 137 (7%) ≥ 10% 0 221 1101 1322 Cases (N, %) 0 21 (10%) 165 (15%) 186 (14%) 6406 Categorical NRI = 0.05, P = 0.02 Continuous NRI = 0.30, P < 0.001 CHD = coronary heart disease; CAC = coronary artery calcium; ECG = electrocardiographic; NRI = net reclassification improvement
19
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 5: Reclassification table for all CHD events in models with and without ECG (both models include CAC strata) 10-year risk in model with ECG 10-year risk in 0-5% 5-10% ≥ 10% Total model without ECG 0-5% 3619 99 0 3718 Cases (N, %) 67 (2%) 9 (11%) 0 76 (2%) 5-10% 110 1026 121 1257 Cases (N, %) 4 (4%) 67 (7%) 12 (10%) 83 (7%) ≥ 10% 0 126 1305 1431 Cases 0 9 (7%) 237 (18%) 246 (17%) 6406 Categorical NRI = 0.024, P = 0.11 Continuous NRI = 0.26, P < 0.001 CHD = coronary heart disease; CAC strata = coronary artery calcium score stratified as 0, 1-100, 101-300, > 300; ECG = electrocardiographic; NRI = net reclassification improvement
20
PT
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
Figure 1
21
S0002-8703(14)00358-5 doi: 10.1016/j.ahj.2014.06.009 YMHJ 4655
To appear in:
American Heart Journal
Received date: Accepted date:
30 May 2013 2 June 2014
Please cite this article as: Desai Chintan S., Ning Hongyan, Soliman Elsayed Z., Burke Gregory L., Shea Steven, Nazarian Saman, Donald M. Lloyd-Jones M.D,. ScM, Greenland Philip, Electrocardiographic abnormalities and coronary artery calcium for coronary heart disease prediction and reclassification: The Multi-Ethnic Study of Atherosclerosis, American Heart Journal (2014), doi: 10.1016/j.ahj.2014.06.009
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 Electrocardiographic abnormalities and coronary artery calcium for coronary heart
T
disease prediction and reclassification: The Multi-Ethnic Study of Atherosclerosis
RI P
Chintan S. Desai, MD, MS; Hongyan Ning, MD, MS; Elsayed Z. Soliman, MD, MSc; Gregory L. Burke, MD, MSc; Steven Shea, MD, MS; Saman Nazarian, MD;
SC
Donald M. Lloyd-Jones, MD, ScM; Philip Greenland, MD
NU
From: Johns Hopkins University School of Medicine, Department of Medicine (C.S.D., S.N.);
MA
Northwestern University Feinberg School of Medicine, Department of Preventive Medicine (H.N., D.M.L.-J., P.G.); Wake Forest School of Medicine, Division of Public Health Sciences
ED
(E.Z.S., G.L.B.); Columbia University Mailman School of Public Health (S.S.);
PT
Running title: ECG and CAC for coronary heart disease risk prediction Abstract word count: 250
Tables: 5
CE
Word count: 4957
Address for correspondence:
AC
Philip Greenland, MD
Harry W. Dingman Professor Departments of Preventive Medicine and Medicine 680 N. Lake Shore Drive, 14th Floor Chicago, IL 60611 E-mail: [email protected] Phone: 312-908-7914 Fax: 312-908-9588
1
Figures: 1
ACCEPTED MANUSCRIPT
ABSTRACT
RI P
T
BACKGROUND: Electrocardiographic (ECG) abnormalities and coronary artery calcium (CAC) identify different aspects of subclinical coronary heart disease (CHD). We sought to determine whether ECG abnormalities improve risk prediction for all CHD and fatal CHD events jointly with
SC
CAC measures.
NU
METHODS: We included 6406 men and women from the Multi-Ethnic Study of Atherosclerosis (MESA) aged 45-84 years who were free of cardiovascular disease at the time of enrollment
MA
(2000-02). We stratified participants by presence of ST-T and Q wave abnormalities: any major, any minor/no major, no major/minor using the Minnesota Code classifications. CAC was defined
ED
into one of the following strata: 0, 1-100, 101-300, > 300. We created risk prediction models using MESA-specific coefficients for traditional risk factors (RFs) and calculated categorical net
PT
reclassification improvement (NRI) for all and fatal CHD.
CE
RESULTS: Over a median follow up of 10 years, we observed that addition of ECG abnormalities to a risk prediction model for all CHD resulted in a categorical NRI of 0.05 (P =
AC
0.04). For fatal CHD alone, the addition of ECG abnormalities resulted in categorical NRI of 0.09 (P = 0.02). Addition of ECG abnormalities to a model containing RFs and CAC resulted in categorical NRI of 0.02 (P = 0.11) for all CHD events. We also observed differences in the association between ECG abnormalities and CHD when stratifying by CAC presence. CONCLUSION: ECG abnormalities improved risk prediction for CHD when added to RFs but not when added to CAC. ECG abnormalities particularly improved risk prediction for fatal CHD.
2
ACCEPTED MANUSCRIPT
INTRODUCTION
T
Clinic-based cardiovascular disease prevention strategies are targeted towards
RI P
individuals based on their estimated absolute risk for disease.1,2 Risk prediction algorithms such as the Framingham Risk Score and the American Heart Association/American College of
SC
Cardiology (AHA/ACC) risk calculator3 are recommended for use in clinical practice but have limitations in correctly stratifying the risk for coronary events.4-6 Prospective epidemiologic
NU
investigations have demonstrated that abnormalities on the resting 12-lead electrocardiogram (ECG) are independently associated with increased incidence of coronary heart disease (CHD)
MA
events7-11 and recent data in older adults suggest that ECG may have modest incremental value in risk prediction beyond traditional risk factors.12 However, there are limited data on the utility of
ED
ECG for risk re-classification in middle and older-aged adults.
PT
A consensus document from the American Heart Association/American College of Cardiology Foundation states that coronary artery calcium (CAC) measurement is reasonable
CE
for risk stratification in asymptomatic adults at intermediate risk for CHD.13 CAC testing has been shown to improve risk classification for CHD compared to risk factors alone.14 Recent
AC
evidence suggests that minor ECG abnormalities may be associated more strongly with fatal arrhythmic events, but not for non-fatal CHD events such as non-fatal MI.15 Therefore, it is possible that ECG abnormalities may identify different aspects of risk than CAC. The objective of our study is to determine if the addition of ECG abnormalities (ST-T abnormalities and Q waves) to a CHD risk prediction model based on traditional risk factors improves risk classification. We also sought to determine if ECG improves risk prediction when added to a model that also includes CAC. In addition, since CAC and ECG may detect divergent elements of the risk substrate, we sought to assess for risk prediction improvement separately for fatal and nonfatal CHD. METHODS 3
ACCEPTED MANUSCRIPT Cohort We studied participants from the Multi-Ethnic Study of Atherosclerosis (MESA); details of
T
the study design have been published.16 MESA recruited healthy participants without clinical
RI P
evidence or history of any cardiovascular disease at baseline in 2000-2002 from the following six communities in the United States: Baltimore, Maryland; Chicago, Illinois; Forsyth County,
SC
North Carolina; Los Angeles County, California; northern Manhattan, New York; St. Paul, Minnesota. Participants were aged 45-84 years at the time of the baseline examination and self-
NU
identified into one of four race-ethnic groups: white (38%), black (28%), Hispanic (22%), and
MA
Chinese (12%). We assessed age, sex, height, weight, body mass index, systolic and diastolic blood pressure, total and high-density lipoprotein (HDL) cholesterol, fasting blood glucose, selfreported diabetes, smoking status, and self-reported antihypertensive or lipid-lowering
ED
medication use at the baseline examination. We excluded participants who were missing these
PT
key covariates (N = 59). Electrocardiography
CE
A resting ECG was recorded at the baseline examination using a MAC 1200 ECG machine (Marquette Electronics, Milwaukee, WI, USA) at all field centers. All ECGs were read
AC
at a central laboratory (EPICARE, Wake Forest School of Medicine, Winston Salem, NC) and visually inspected for adequate quality and technical errors. ECG abnormalities were classified using the Minnesota ECG code (MC).17 Major ECG abnormalities were defined as follows: major Q wave abnormalities (MC 1.1.x, 1.2.x), minor Q/QS with major ST-T abnormalities (1.3.x, 4.1.x, 4.2, 5.1, 5.2), and major isolated ST-T abnormalities (MC 4.1.x, 4.2, 5.1, 5.2). Minor ECG abnormalities were defined as follows: minor isolated Q waves (MC 1.3.x) and minor isolated ST-T (MC 4.3, 4.4, 5.3, 5.4). We classified participants into three groups: 1) any major ECG abnormalities; 2) any minor but no major abnormalities; and 3) no major or minor abnormalities. We excluded participants who were missing ECG data (N = 49) and who had ECG diagnoses that suppressed the coding of ST-T or Q wave abnormalities (N = 300). Atrial fibrillation was an 4
ACCEPTED MANUSCRIPT exclusion criterion for MESA because gated cardiac computed tomography scanning was used to measure CAC; therefore, rhythm abnormalities were not included in our classification.
T
Coronary artery calcium
RI P
The MESA coronary artery calcium (CAC) protocol for acquisition and interpretation has been described previously.18 Each participant was scanned twice and the scans were
SC
interpreted at a central reading laboratory. The Agatston method was used to calculate the CAC score, and the average of the two CAC scores was used.19 For the present study, we
MA
Ascertainment and adjudication of outcomes
NU
categorized CAC into one of the following strata: 0, 1-100, 101-300, or > 300.
At intervals of 9-12 months, interviewers contacted MESA participants or family members to inquire about interim hospital admissions, cardiovascular diagnoses, and deaths.
ED
To verify self-reported diagnoses, we requested copies of outpatient and inpatient medical
PT
records, as well as death certificates. Two trained physicians reviewed the records and verified diagnoses. If disagreements persisted after review, a full MESA morbidity and mortality
CE
committee reviewed the records for final adjudication. The primary outcome for our analysis was coronary heart disease (CHD). For the
AC
present analysis, CHD included nonfatal myocardial infarction (NFMI), death due to CHD, and hospitalized angina. NFMI was defined based on a combination of symptoms, ECG findings, and levels of cardiac biomarkers. Fatal CHD required myocardial infarction within 28 days of death, chest pain within 72 hours of death, or a known history of CHD in the absence of nonatherosclerotic and non-cardiac causes of death. Angina was defined as definite, probable, or absent based on documentation of clinical signs and symptoms. For our analyses, we included a composite outcome of definite or probable NFMI, CHD death, and hospitalized angina; we refer to the composite outcome as “all CHD.” We also analyzed definite/probable fatal CHD events and nonfatal CHD events separately. Additional details on MESA clinical outcomes are available online at http://www.mesa-nhlbi.org/. 5
ACCEPTED MANUSCRIPT Statistical analysis After exclusions, 6406 participants were eligible for our analyses. We compared
T
unadjusted baseline characteristics by ECG status; chi-square was used to compare differences
RI P
for categorical variables and ANOVA was used to compare means for continuous variables. We used Cox proportional hazards modeling to determine the association between ECG
SC
abnormalities and incident CHD. ECG abnormalities were modeled as a three-tiered variable. The presence of 1) any major abnormalities and 2) minor abnormalities only were each
NU
considered as separate variables in the model, and 3) no major or minor abnormalities was the
MA
reference group for the variable. We adjusted for age, race, sex, systolic blood pressure, diabetes, total cholesterol, HDL cholesterol, smoking status, and use of antihypertensive medication.
ED
We calculated the predicted 10-year risk for all CHD using MESA-specific coefficients for
PT
age, race, sex, systolic blood pressure, diabetes, total cholesterol, HDL cholesterol, smoking status, and use of antihypertensive medication. This was considered the baseline risk prediction
CE
model for CHD. We then determined the net reclassification improvement (NRI) with the addition of ECG abnormalities to the baseline model. The NRI was calculated using the
AC
following formula described by Pencina et al:20 (Number of events reclassified higher – Number of events reclassified lower)/total events + (Number of nonevents reclassified lower – Number of nonevents reclassified higher)/total nonevents
We used this calculation for the following 10-year risk categories for CHD: < 5%, 5-10%, and ≥ 10%. We also calculated a category-free NRI.21 We then used MESA-specific coefficients for traditional risk factors to calculate 10-year risk for fatal and nonfatal CHD separately. We used the same 10-year risk categories to calculate a categorical and a continuous NRI separately for fatal and nonfatal CHD. We calculated Harrell’s C-statistic for the baseline models containing 6
ACCEPTED MANUSCRIPT traditional risk factors and the C-statistic with the addition of ECG abnormalities.22 We also calculated the category-free NRI for CHD in only those participants who were in the
T
intermediate risk group (5-10%).
RI P
To assess for confounding and interaction between ECG abnormalities and CAC in the association with CHD, we calculated hazards ratios for incident CHD by ECG status separately
SC
among MESA participants by CAC strata. We then included CAC in the baseline risk prediction model and determined the categorical and continuous NRI for addition of ECG to the model.
NU
SAS version 9.3 was used for all analyses (Cary, NC). Two-sided P-values < 0.05 were
MA
considered statistically significant. Funding
This work was supported by the National Heart, Lung, and Blood Institute [N01-HC-
ED
95159 through N01-HC-95169] and by the National Center for Research Resources [UL1-RR-
RESULTS
CE
Baseline characteristics
PT
024156 and UL1-RR-025005]. No extramural funding was used to support this work.
Among the 6406 participants, 385 (6%) had any major ECG abnormalities and 897
AC
(14%) had any minor/no major ECG abnormalities; the remainder had no major or minor abnormalities. The baseline characteristics of the study cohort are shown in Table 1, stratified by ECG status. Participants with ECG abnormalities were older and more likely to be black, compared to participants with no major/minor abnormalities. Participants with ECG abnormalities also had higher systolic blood pressure, higher prevalence of diabetes, and were more likely to be taking antihypertensive and lipid-lowering medications. The prevalence of CAC > 0 among participants with no ECG abnormalities was 47%, compared to 56% and 63% in those with only minor or any major abnormalities, respectively (P < 0.001). Event rates and hazards ratios
7
ACCEPTED MANUSCRIPT Over a median of 10 years follow up, a total of 405 CHD events occurred. Table 2 shows the event rates and multivariable-adjusted hazards ratios for all CHD by ECG status. The
T
CHD event rates per 1000 person-years in participants with no major/minor, only minor, and any
RI P
major ECG abnormalities were 5.6, 11.0, and 15.4, respectively. After adjustment for clinical and demographic covariates, the hazards ratios for all CHD for participants with only minor and
SC
any major ECG abnormalities were 1.56 (95% CI, 1.22-1.99) and 1.81 (1.32-2.47), respectively, compared to those with no major/minor abnormalities. Table 2 also shows all CHD event rates
NU
and hazards ratios further stratified by CAC score. Among participants with a CAC score of 0,
MA
the presence of only minor ECG abnormalities did not increase the risk for CHD, compared to those with no ECG abnormalities. Participants with CAC score = 0 and major ECG abnormalities had a 3-fold increased risk for all CHD, compared to those with no ECG
ED
abnormalities. The magnitude of the hazards ratio for major ECG abnormalities and all CHD
PT
was lower in the other CAC groups.
Table 3 shows the event rates and hazards ratios for fatal and nonfatal CHD separately.
CE
The overall hazards ratios in all participants for major ECG abnormalities were similar for all CHD, fatal CHD, and nonfatal CHD. Compared to those with no major/minor abnormalities, the
AC
hazards ratio in those with only minor ECG abnormalities was 2.42 (1.44-4.09) for fatal CHD and 1.38 (1.04-1.83) for nonfatal CHD. Across all CAC strata, the hazards ratios for the association between fatal CHD and minor ECG abnormalities had a greater magnitude than those for nonfatal CHD. The Figure shows CHD event rates by ECG status and by presence or absence of CAC. Among participants with CAC = 0, the CHD event rates were similar for those with only minor ECG abnormalities and no major/minor abnormalities. Among participants with CAC > 0, the CHD event rates were substantially higher for those with minor ECG abnormalities, and were similar to rates in participants with any major ECG abnormalities. Discrimination and reclassification 8
ACCEPTED MANUSCRIPT Table 4 shows the reclassification table for predicted and observed risk for all CHD with addition of ECG abnormalities to the baseline model. The C-statistic for the baseline model was
T
0.727 and the C-statistic with the addition of ECG abnormalities was 0.736; the P-value for the
RI P
difference was 0.02. For the categorical NRI, 4% of events were correctly reclassified and 2% of nonevents were correctly reclassified; the resulting NRI was 0.053 (P = 0.02). The category-free
SC
NRI was 0.30 (P < 0.001).
We then determined the improvement in risk prediction for fatal and nonfatal CHD
NU
events separately. For fatal CHD events, the addition of ECG to the baseline risk prediction
MA
model resulted in a categorical NRI = 0.09 (P = 0.08) and continuous NRI = 0.47 (P < 0.001). For nonfatal CHD events, the addition of ECG to the baseline risk prediction model resulted in a categorical NRI = 0.02 (P = 0.35) and continuous NRI = 0.22 (P < 0.001).
ED
Table 5 shows the cross-tabulations for predicted and observed CHD risk with addition
PT
of ECG to the baseline model, which includes risk factors and CAC strata. The C-statistic for the baseline risk prediction model was 0.783; with addition of ECG, the C-statistic was 0.788 (P-
CE
value for difference = 0.03). The overall categorical NRI was 0.02 (P = 0.11) and the continuous NRI was 0.26 (P < 0.001). For fatal CHD, the addition of ECG abnormalities to the baseline
AC
model including risk factors and CAC strata resulted in a categorical NRI of 0.08 (P = 0.15) and category-free NRI of 0.48 (P < 0.001). For nonfatal CHD, the addition of ECG abnormalities to a baseline model including risk factors and CAC resulted in a categorical NRI of 0.01 (P = 0.66) and category-free NRI of 0.16 (P = 0.04). Among the 1853 participants with predicted 10-year all CHD risk 5-10%, the addition of ECG abnormalities to the baseline model including only traditional risk factors resulted in a category-free NRI = 0.23 (P = 0.01). The magnitude of the category-free NRI was similar with addition of ECG to a baseline model containing risk factors and CAC (category-free NRI = 0.23, P = 0.01). DISCUSSION 9
ACCEPTED MANUSCRIPT Findings In MESA participants, the addition of ECG abnormalities to a risk prediction model based
T
on traditional risk factors resulted in modestly improved reclassification for all CHD. Addition of
RI P
ECG abnormalities to a model including traditional risk factors and CAC did not result in improved risk reclassification. We observed differences in hazards ratios for CHD when
SC
stratifying by CAC presence, suggesting the possibility of interaction. The observation of a
for fatal CHD, rather than nonfatal CHD.
MA
Implications
NU
modestly improved classification for all CHD with ECG was driven by improved reclassification
In the present study, we analyzed the improvement in CHD risk prediction with the use of easily identified ST-T and Q wave abnormalities. In comparison with other biomarkers, the
ED
magnitude of the NRI (categorical NRI = 0.053, continuous NRI = 0.29) is similar to that
PT
achieved by ankle-brachial index and high-sensitivity C-reactive protein.23 To the best of our knowledge, our study is the first to examine CHD risk prediction jointly with ECG and CAC. The
CE
association between ECG abnormalities and CHD was not attenuated after adjustment for the presence of CAC; thus, CAC is not a confounder in this association. When we stratified by CAC
AC
presence, we observed differences in the hazards ratios for CHD. In participants with no detectable CAC, those with only minor ECG abnormalities are not at increased risk for CHD compared to those with no major/minor ECG abnormalities. Major ECG abnormalities, however, were associated with a 3-fold increased risk for CHD even in the absence of CAC. Among participants with CAC score > 0, those with minor and major ECG abnormalities had a similar risk for incident CHD. These findings suggest that the association between ECG abnormalities and CHD is modified by CAC presence. At each category of ECG abnormality (no major/minor, only minor, any major), the absolute magnitude of risk was greater in participants with CAC than in those without CAC.
10
ACCEPTED MANUSCRIPT The distinguishing feature of ECG abnormalities in our study is the superior performance in risk prediction for fatal CHD (categorical NRI = 0.09, continuous NRI = 0.48) compared with
T
nonfatal CHD (categorical NRI = 0.02, continuous NRI = 0.16). These findings are consistent
RI P
with a previous analysis of the Cardiovascular Health Study, which found that minor isolated STT abnormalities were associated with a nearly 2-fold increase in risk for incident fatal CHD, but
SC
were not associated with increased risk for nonfatal CHD.15 The potential association between ECG abnormalities and arrhythmic events could be further tested in clinical trials involving
NU
agents known to modify the risk of arrhythmia, such as beta-adrenergic receptor antagonists.
MA
The clinical implications of moving to a higher risk category primarily involve the use of statin therapy. The recent AHA/ACC guidelines for primary prevention recommend initiation of statin therapy for primary prevention when the 10-year risk of atherosclerotic cardiovascular
ED
disease exceeds 7.5% in individuals 40 years or older.2 Many of the individuals in the
PT
intermediate group in our study (10-year risk of CHD only 5-10%) may already meet that threshold. However, for patients who are borderline, or when the clinician desires additional
CE
information, ECG may be a reasonable test for additional risk stratification in these selected individuals.
AC
One limitation of our study is that it is a secondary analysis of observational data. However, MESA is a well-phenotyped cohort comprised of individuals with validated ECG and outcome measures. Furthermore, our results are concordant with previous observations from the Cardiovascular Health Study.15 A strength of this study is that we used MESA-specific coefficients to calculate our baseline risk scores, thus reducing the possibility of observing a falsely high NRI. Conclusions In a middle and older-aged, multi-ethnic cohort, we observed that major and minor ST-T and Q wave abnormalities offered modestly improved CHD risk prediction beyond traditional risk factors, particularly for fatal CHD. We also observed differences in the association between 11
ACCEPTED MANUSCRIPT ECG and CHD when stratifying by CAC presence, suggesting the possibility of interaction
AC
CE
PT
ED
MA
NU
SC
RI P
T
between ECG and CAC.
12
ACCEPTED MANUSCRIPT
CONTRIBUTIONS
T
The authors thank the other investigators, the staff, and the participants of the MESA study for
RI P
their valuable contributions. A full list of participating MESA investigators and institutions can be
SC
found at http://www.mesa-nhlbi.org.
The authors are solely responsible for the design and conduct of this study, all study analyses,
AC
CE
PT
ED
MA
NU
the drafting and editing of the paper, and its final contents.
13
ACCEPTED MANUSCRIPT
REFERENCES
AC
CE
PT
ED
MA
NU
SC
RI P
T
1. Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III guidelines. Circulation 2004;110:227-39. 2. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013. 3. Goff DC, Jr., Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013. 4. Berry JD, Lloyd-Jones DM, Garside DB, Greenland P. Framingham risk score and prediction of coronary heart disease death in young men. Am Heart J 2007;154:80-6. 5. Cavanaugh-Hussey MW, Berry JD, Lloyd-Jones DM. Who exceeds ATP-III risk thresholds? Systematic examination of the effect of varying age and risk factor levels in the ATP-III risk assessment tool. Prev Med 2008;47:619-23. 6. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013;382:1762-5. 7. Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: final report of the pooling project. The pooling project research group. J Chronic Dis 1978;31:201-306. 8. Kannel WB. Common electrocardiographic markers for subsequent clinical coronary events. Circulation 1987;75:II25-7. 9. Liao YL, Liu KA, Dyer A, et al. Major and minor electrocardiographic abnormalities and risk of death from coronary heart disease, cardiovascular diseases and all causes in men and women. J Am Coll Cardiol 1988;12:1494-500. 10. Dekker JM, Schouten EG, Klootwijk P, Pool J, Kromhout D. ST segment and T wave characteristics as indicators of coronary heart disease risk: the Zutphen Study. J Am Coll Cardiol 1995;25:1321-6. 11. Sutherland SE, Gazes PC, Keil JE, Gilbert GE, Knapp RG. Electrocardiographic abnormalities and 30-year mortality among white and black men of the Charleston Heart Study. Circulation 1993;88:268592. 12. Auer R, Bauer DC, Marques-Vidal P, et al. Association of major and minor ECG abnormalities with coronary heart disease events. JAMA 2012;307:1497-505. 13. Greenland P, Bonow RO, Brundage BH, et al. ACCF/AHA 2007 clinical expert consensus document on coronary artery calcium scoring by computed tomography in global cardiovascular risk assessment and in evaluation of patients with chest pain: a report of the American College of Cardiology Foundation Clinical Expert Consensus Task Force (ACCF/AHA Writing Committee to Update the 2000 Expert Consensus Document on Electron Beam Computed Tomography) developed in collaboration with the Society of Atherosclerosis Imaging and Prevention and the Society of Cardiovascular Computed Tomography. Journal of the American College of Cardiology 2007;49:378-402. 14. Polonsky TS, McClelland RL, Jorgensen NW, et al. Coronary artery calcium score and risk classification for coronary heart disease prediction. JAMA 2010;303:1610-6. 15. Kumar A, Prineas RJ, Arnold AM, et al. Prevalence, prognosis, and implications of isolated minor nonspecific ST-segment and T-wave abnormalities in older adults: Cardiovascular Health Study. Circulation 2008;118:2790-6. 16. Bild DE, Bluemke DA, Burke GL, et al. Multi-ethnic study of atherosclerosis: objectives and design. American journal of epidemiology 2002;156:871-81. 14
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
17. Tuinstra CL, Rautaharju PM, Prineas RJ, Duisterhout JS. The performance of three visual coding procedures and three computer programs in classification of electrocardiograms according to the Minnesota Code. Journal of electrocardiology 1982;15:345-50. 18. Carr JJ, Nelson JC, Wong ND, et al. Calcified coronary artery plaque measurement with cardiac CT in population-based studies: standardized protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) study. Radiology 2005;234:35-43. 19. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Jr., Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990;15:827-32. 20. Pencina MJ, D'Agostino RB, Sr., D'Agostino RB, Jr., Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157-72; discussion 207-12. 21. 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. 22. Harrell FE, Jr., Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests. JAMA 1982;247:2543-6. 23. Yeboah J, McClelland RL, Polonsky TS, et al. Comparison of novel risk markers for improvement in cardiovascular risk assessment in intermediate-risk individuals. JAMA 2012;308:788-95.
ED
FIGURE LEGEND
AC
CE
PT
Figure: Ten-Year Coronary heart disease event rates by ECG and CAC status in the MultiEthnic Study of Atherosclerosis
15
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 1: Baseline characteristics by ECG status Covariate No major/minor Any minor/No major Any major P-value (N = 5124) (N = 897) (N = 385) Age (yrs) 61 ± 10 65 ± 10 67 ± 10 < 0.001 Male sex (N,%) 2370 (46%) 385 (43%) 204 (53%) 0.004 Race (N,%) < 0.001 2015 (39%) 291 (32%) 117 (30%) • Caucasian 1327 (26%) 294 (33%) 142 (37%) • Black 646 (13%) 86 (10%) 50 (13%) • Chinese 1136 (22%) 226 (25%) 76 (20%) • Hispanic Systolic blood 124 ± 20 133 ± 23 138 ± 24 < 0.001 pressure (mm Hg) Body mass index 28 ± 5 29 ± 6 29 ± 5 < 0.001 (kg/m2) Total cholesterol 5.02 ± 0.93 5.04 ± 0.93 4.97 ± 0.96 0.27 (mmol/L) HDL cholesterol 1.32 ± 0.39 1.32 ± 0.39 1.27 ± 0.36 0.04 (mmol/L) Diabetes (N,%) 560 (11%) 160 (18%) 72 (19%) < 0.001 Current smoker (N,%) 664 (13%) 128 (14%) 48 (12%) 0.52 Antihypertensive 1502 (29%) 387 (43%) 207 (54%) < 0.001 medication (N,%) Lipid-lowering 796 (16%) 150 (17%) 83 (22%) 0.007 therapy (N,%) CAC > 0 (N,%) 2407 (47%) 503 (56%) 243 (63%) < 0.001 ECG = electrocardiographic; HDL = high-density lipoprotein; CAC = coronary artery calcium
16
ACCEPTED MANUSCRIPT
MA
NU
SC
RI P
T
Table 2: Event rates and hazards ratios for all CHD events by ECG abnormality No major/minor Only minor Any major (N = 5124) (N = 897) (N = 385) Event rate among 5.57 11.0 15.42 overall cohort (per 1000 P-Y) Stratified by CAC 1.62 1.87 7.49 • CAC = 0 (N = 3253) 5.76 13.40 12.02 • CAC 1-100 (N = 1709) 13.10 13.89 21.65 • CAC 101-300 (N = 687) 21.29 35.02 38.93 • CAC > 300 (N = 757)
AC
CE
PT
ED
Hazards ratios# for 1.0 (ref) 1.56 (1.22-1.99)* 1.81 (1.32-2.47)* all CHD among overall cohort Stratified by CAC 1.0 (ref) 0.97 (0.43-2.29) 3.32 (1.62-6.83)* • CAC = 0 (N = 3253) 1.0 (ref) 2.20 (1.41-3.42)* 1.82 (0.97-3.40) • CAC 1-100 (N = 1709) 1.0 (ref) 0.94 (0.52-1.71) 1.39 (0.67-2.89) • CAC 101-300 (N = 687) 1.0 (ref) 1.55 (1.05-2.29)* 1.81 (1.09-3.02)* • CAC > 300 (N = 757) *denotes P < 0.05 # denotes adjustment for age, race, sex, systolic blood pressure, total and HDL cholesterol, diabetes, smoking, antihypertensive medication CHD = coronary heart disease; ECG = electrocardiographic; CAC = coronary artery calcium; P-Y = person-years
17
ACCEPTED MANUSCRIPT
NU
SC
RI P
T
Table 3: Fatal and nonfatal CHD event rates and hazards ratios by ECG abnormalities and CAC No major/minor Only minor Any major (N = 5124) (N = 897) (N = 385) Fatal CHD event rate among 0.79 2.67 2.29 overall cohort (per 1000 P-Y) Stratified by CAC score 0.29 0.77 1.43 • CAC = 0 (N = 3253) 0.90 2.58 0.91 • CAC 1-100 (N = 1709) 1.40 5.57 4.12 • CAC 101-300 (N = 687) 2.58 6.14 4.88 • CAC > 300 (N = 757) Hazards ratios# for fatal CHD among all MESA participants Stratified by CAC score • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709) • CAC 101-300 (N = 687)
CAC > 300 (N = 757)
1.55 (0.72-3.37)
2.53 (0.49-13.00) 0.58 (0.07-4.62) 2.19 (0.42-11.47)
1.0 (ref)
2.15 (0.53-8.65) 2.51 (0.91-6.96) 3.77 (1.2111.75)* 2.32 (0.93-5.78)
4.74
8.12
12.95
1.32 4.80 11.57 18.28
1.07 10.62 7.94 27.65
5.99 11.02 16.84 32.33
1.0 (ref)
1.38 (1.041.83)*
1.83 (1.30-2.58)*
1.0 (ref) 1.0 (ref)
0.68 (0.24-1.94) 2.13 (1.303.49)* 0.63 (0.30-1.34) 1.42 (0.92-2.19)
3.50 (1.57-7.82)* 2.14 (1.10-4.17)*
MA
CE
PT
Nonfatal CHD event rate/1000 P-Y Stratified by CAC score • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709) • CAC 101-300 (N = 687) • CAC > 300 (N = 757)
AC
2.42 (1.444.09)*
1.0 (ref) 1.0 (ref) 1.0 (ref)
ED
•
1.0 (ref)
Hazards ratios# for nonfatal CHD among all MESA participants Stratified by CAC score: • CAC = 0 (N = 3253) • CAC 1-100 (N = 1709)
1.85 (0.51-6.73)
1.0 (ref) 1.26 (0.55-2.87)* • CAC 101-300 (N = 687) 1.0 (ref) 1.79 (1.03-3.12)* • CAC > 300 (N = 757) *denotes P < 0.05 # denotes adjustment for age, race, sex, systolic blood pressure, total and HDL cholesterol, diabetes, smoking, antihypertensive medication CHD = coronary heart disease; ECG = electrocardiographic; CAC = coronary artery calcium; P-Y = person-years
18
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 4: Reclassification table for CHD events in models with and without ECG 10-year risk in model with ECG 10-year risk in 0-5% 5-10% ≥ 10% Total model without ECG 0-5% 3058 173 0 3231 Cases (N, %) 62 (2%) 20 (12%) 0 82 (3%) 5-10% 246 1394 213 1853 Cases (N, %) 14 (6%) 93 (7%) 30 (14%) 137 (7%) ≥ 10% 0 221 1101 1322 Cases (N, %) 0 21 (10%) 165 (15%) 186 (14%) 6406 Categorical NRI = 0.05, P = 0.02 Continuous NRI = 0.30, P < 0.001 CHD = coronary heart disease; CAC = coronary artery calcium; ECG = electrocardiographic; NRI = net reclassification improvement
19
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Table 5: Reclassification table for all CHD events in models with and without ECG (both models include CAC strata) 10-year risk in model with ECG 10-year risk in 0-5% 5-10% ≥ 10% Total model without ECG 0-5% 3619 99 0 3718 Cases (N, %) 67 (2%) 9 (11%) 0 76 (2%) 5-10% 110 1026 121 1257 Cases (N, %) 4 (4%) 67 (7%) 12 (10%) 83 (7%) ≥ 10% 0 126 1305 1431 Cases 0 9 (7%) 237 (18%) 246 (17%) 6406 Categorical NRI = 0.024, P = 0.11 Continuous NRI = 0.26, P < 0.001 CHD = coronary heart disease; CAC strata = coronary artery calcium score stratified as 0, 1-100, 101-300, > 300; ECG = electrocardiographic; NRI = net reclassification improvement
20
PT
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
Figure 1
21
