Elevated body mass index and intermediate-term clinical outcomes after acute coronary syndromes

Study population 

Between August 1997 and August 1999, the SYMPHONY and 2nd SYMPHONY trials randomized 15904 patients from 931 clinical centers in 37 countries to evaluate sibrafiban, an oral platelet glycoprotein IIb/IIIa inhibitor, for secondary prevention after acute coronary syndromes. Each center’s ethics committee or institutional review board approved the protocols, and all patients gave written, informed consent. Methods and results of these trials have been published.8, 9, 10 Briefly, the SYMPHONY and 2nd SYMPHONY trials enrolled patients with both ST-segment elevation myocardial infarction and non-ST-segment elevation acute coronary syndrome who presented with ≥20 minutes of chest pain or anginal-equivalent symptoms and met previously described electrocardiographic or cardiac marker criteria.10 Patients were randomized within 7 days of their qualifying event and were stable for ≥12 hours (without recurrent ischemia or hemodynamic instability and Killip class <II).

SYMPHONY randomized 9233 patients a median of 3.6 days after their qualifying event to 90-day treatment with either aspirin 80 mg twice daily or 1 of 2 sibrafiban dosing strategies. Patients were followed for 1 year after study entry. Compared with aspirin alone, there was no benefit of either sibrafiban dose-regimen with regard to the primary endpoint of death, myocardial infarction, or severe recurrent ischemia at 90 days, and sibrafiban-treated patients had a dose-related increase in bleeding.

The 2nd SYMPHONY trial randomized 6671 patients a median of 3.7 days after their qualifying event to either aspirin 80 mg twice daily, aspirin 80 mg plus low-dose sibrafiban twice daily, or high-dose sibrafiban alone twice daily. Although 12-18 months treatment was planned, median treatment duration was only 90 days due to early study termination by the sponsor after the results of SYMPHONY were known. Patients were continued on assigned study treatment until the final visit, when they were switched to open-label aspirin (or other antiplatelet therapy as appropriate). There was no statistically significant difference between treatment groups in the primary endpoint of death, myocardial infarction, or severe recurrent ischemia, and bleeding was higher in the sibrafiban treatment arms.

Body mass index groups 

A total of 15187 patients had data available to calculate body mass index. Underweight patients (BMI <18.5 kg/m2, n = 116) were prospectively excluded to permit comparisons between normal-weight subjects and those with varying degrees of elevated BMI. We classified the remaining 15071 patients using a modification of the National Institutes of Health BMI-based categories: normal (18.5-24.9 kg/m2), overweight (25.0-29.9 kg/m2), and obese I, II, III (≥30.0 kg/m2).11 We classified patients with BMI between 30.0 and 34.9 kg/m2 as obese, but due to the small number of patients with BMI ≥35 kg/m2 in our study, we collapsed the obesity II and III categories into one category, which we called “very obese.”

Measurements 

The following candidate variables were used for stepwise selection in study models:

Demographics: age, sex, race/ethnicity (white, black, Asian, Hispanic, American Indian), height, weight, BMI, and geographic region of enrollment (Australia/New Zealand, Asia, Latin America, Eastern Europe, Western Europe, United States, Canada).

Clinical history before qualifying event: family history of coronary artery disease, congestive heart failure, hypertension, hypercholesterolemia, angina, diabetes mellitus, diabetes mellitus with end-organ damage, insulin-treated diabetes mellitus, stroke, transient ischemic attack, myocardial infarction, angiography, coronary artery bypass graft, severe chronic obstructive pulmonary disease, cancer, abnormal liver enzymes, chronic renal insufficiency, peripheral arterial vascular disease, atrial fibrillation, supraventricular tachycardia, ventricular dysrhythmias, heart block, smoking history (never, current, past), angina within 6 weeks before qualifying event, multiple episodes of angina before qualifying event.

Qualifying event type: myocardial infarction or unstable angina.

Vital signs at qualifying event: heart rate, systolic blood pressure, diastolic blood pressure, mitral regurgitation, S3 gallop, Killip class >I, >II, and >III.

Vital signs at randomization: heart rate, systolic blood pressure, diastolic blood pressure, mitral regurgitation, S3 gallop.

Cardiac markers at qualifying event: elevated cardiac enzymes (CK, CKMB or troponin T or I).

Electrocardiogram at qualifying event: left bundle branch block, paced rhythm, T-wave pseudonormalization, Q wave, ST-segment depression, ST-segment elevation, T-wave inversion, location of electrocardiogram changes (inferior, lateral, posterior), abnormal qualifying event electrocardiogram.

Clinical events between qualifying event and randomization: acute mitral regurgitation, atrial fibrillation, congestive heart failure, shock, pulmonary edema, recurrent coronary ischemia, heart block, supraventricular tachycardia, ventricular fibrillation.

Study treatment characteristics: time from qualifying event to treatment, treatment assignment (control group, low-dose sibrafiban, high-dose sibrafiban).

Laboratory measures at randomization: serum creatinine, creatinine clearance ({[140 − age] × weight in kilograms/serum creatinine × 72} × 0.85 if female).

Medication use: baseline medications (angiotensin-converting enzyme inhibitor, anti-arrhythmic, aspirin, beta-blocker, calcium channel blocker, coumadin, digitalis, diuretic, heparin, intravenous glycoprotein IIb/IIIa inhibitor, lipid-lowering therapy, low-molecular-weight heparin, nitrates, nonsteroidal anti-inflammatory agents, oral anti-platelet therapy, statins, vitamins), prerandomization thrombolysis.

Medication use during qualifying event: thrombolysis, streptokinase, rPA, tPA, APSAC.

Procedure use: Any prerandomization angiography, angiography between qualifying event and randomization, emergency percutaneous coronary intervention, any prerandomization stent, stent between qualifying event and randomization, any prerandomization percutaneous coronary intervention, percutaneous coronary intervention between qualifying event and randomization, any prerandomization coronary artery bypass graft, coronary artery bypass graft between qualifying event and randomization.

The following variables were not used because there were too many missing: angiography results (qualifying event angiography was not required), ejection fraction, left ventricular dysfunction, number of diseased vessels, stenosis, TIMI grade, arterial diseases: left anterior descending, left circumflex, right coronary artery, left main, graft.

Endpoints 

We assessed the relationships of BMI group with 30-day, 90-day, and 1-year mortality and 30-day and 90-day death or myocardial infarction. Myocardial infarction occurring within the 90-day SYMPHONY treatment period and in 2nd SYMPHONY, before trial termination, was centrally adjudicated.10 These myocardial infarction endpoints were used for the current analysis.

Data analysis 

Baseline characteristics are presented as percentages for discrete variables and medians (25th and 75th percentiles) for continuous variables. Kaplan-Meier estimates were generated for each endpoint. Using the normal weight BMI group as the reference, adjusted hazard ratios (with 95% confidence intervals [CI]) were determined using Cox proportional-hazards models previously developed in this population.12 For each model, we first assessed univariable associations between outcomes and candidate covariates for which there were <200 patients missing information. Nonlinear continuous variables were transformed using restricted cubic spline techniques. After developing an initial model using stepwise selection, variables for which there were >200 patients missing information, but which were considered clinically important, were tested one at a time for subsequent inclusion. Backwards elimination was then performed to establish the final model. A similar process was used to develop all study models. The proportional hazards assumption was verified for each retained variable.

Because BMI groups were not randomly assigned, we performed additional analyses to test whether other variables might explain or modify, body mass index’s association with this study’s endpoints. Specifically, we repeated our previously described analyses stratifying on 3 baseline variables (ie, diabetes, hypertension, and smoking status) that have been associated with BMI in other studies. We also included variables denoting postrandomization therapy use (ie, percutaneous coronary intervention, statins, angiotensin-converting enzyme inhibitors, and beta-blockers). Therapies were modeled both as time-dependent covariates and as propensity scores representing their likelihood of use.

Baseline patient information 

Overall, 73% of study patients were overweight (BMI ≥25 kg/m2) and 28% were obese or very obese (BMI ≥30 kg/m2). However, results varied among geographic regions (Table 1). Very obese patients were most likely to be enrolled in the United States (13% of patients) and least likely to be enrolled in Asia (1% of patients). Normal-weight patients were least likely to be enrolled in the United States (22% of patients) and most likely to be enrolled in Asia (58% of patients). The distribution of BMI categories also varied within racial/ethnic subgroups. Only 1% of Asian patients were very obese and 54% were normal weight compared with 16% of black patients who were very obese and only 21% normal weight.

Table 1. Distribution of body mass index category by region and race⁎

		Normal weight	Overweight	Obese	Very obese
	Region
	Australia/New Zealand (n = 592)	26%	47%	20%	7%
	Asia (n = 485)	58%	36%	5%	1%
	Latin America (n = 548)	28%	46%	20%	6%
	Eastern Europe (n = 2584)	29%	48%	18%	4%
	Western Europe (n = 3232)	31%	48%	17%	4%
	United States (n = 6801)	22%	41%	24%	13%
	Canada (n = 829)	23%	47%	23%	7%
	Race
	American Indian	33%	27%	30%	10%
	Asian	54%	38%	7%	1%
	Black	21%	37%	26%	16%
	Caucasian	26%	45%	21%	8%
	Hispanic	22%	48%	21%	9%
	Other	36%	43%	16%	5%

⁎ Totals may not equal 100% due to rounding.

Increasing BMI was associated with younger age, female sex, and a worse cardiovascular risk profile (Table 2). Normal-weight patients were the most likely to report being a current smoker and least likely to report having successfully quit smoking, and very obese patients were most likely to have had prior coronary angiography and a prior percutaneous coronary intervention. At the qualifying event, increasing BMI was associated with higher blood pressure and heart rate (Table 2).

Table 2. Baseline characteristics by body mass index category

		Normal weight (n = 4071)	Overweight (n = 6711)	Obese (n = 3072)	Very obese (n = 1217)
	Demographics
	Age	62 (52, 71)	60 (51, 68)	57 (50, 66)	55 (48, 63)
	Men	69%	77%	73%	61%
	Cardiac risk factors
	Current smoker	42%	35%	35%	35%
	Past smoker	28%	34%	35%	35%
	Family history of coronary artery disease	42%	45%	48%	55%
	Hypertension	41%	49%	58%	65%
	Hyperlipidemia	42%	48%	54%	53%
	Total cholesterol⁎	193 (168, 219)	196 (171, 225)	199 (172, 226)	196 (167, 223)
	Low density lipoprotein cholesterol	113 (93, 136)	116 (94, 139)	115 (95, 139)	109 (91, 139)
	High density lipoprotein cholesterol	44 (37, 52)	41 (34, 49)	38 (32, 45)	38 (31, 44)
	Triglycerides	146 (111, 198)	170 (129, 233)	193 (143, 267)	191 (135, 259)
	Diabetes	12%	17%	23%	33%
	Chronic obstructive pulmonary disease	4%	3%	3%	5%
	Peripheral vascular disease	6%	5%	4%	4%
	Cardiac history
	History of congestive heart failure	5%	4%	4%	7%
	History of angina	44%	45%	45%	45%
	Prior myocardial infarction	19%	21%	20%	21%
	Prior diagnostic angiography	20%	22%	23%	27%
	Prior percutaneous coronary intervention	9%	11%	12%	15%
	Prior coronary bypass surgery	9%	9%	10%	9%
	Clinical presentation
	Qualifying event type
	Myocardial infarction	73%	73%	74%	74%
	Unstable angina	27%	27%	26%	26%
	At qualifying event
	Systolic blood pressure	140 (120, 160)	140 (124, 160)	145 (129, 163)	147 (130, 170)
	Diastolic blood pressure	80 (70, 90)	81 (72, 94)	84 (74, 96)	85 (73, 97)
	Heart rate	75 (64, 87)	75 (64, 86)	77 (68, 89)	80 (70, 92)
	Killip class >II	12%	10%	10%	11%
	At randomization
	Systolic blood pressure	120 (110, 130)	120 (110, 132)	122 (110, 136)	123 (112, 140)
	Diastolic blood pressure	70 (60, 80)	70 (64, 80)	72 (65, 80)	71 (64, 80)
	Heart rate	69 (60, 78)	69 (60, 76)	70 (62, 79)	72 (64, 80)
	Creatinine clearance	71 (56, 88)	86 (69, 105)	104 (83, 126)	125 (100, 156)

⁎ Cholesterol data represents SYMPHONY patients only.

Following the qualifying event, increasing BMI was associated with a greater likelihood of diagnostic catheterization, percutaneous coronary intervention, and stent procedures despite similar degrees of left ventricular dysfunction and coronary artery disease (Table 3).

Table 3. Angiography, revascularization and concomitant medication use by body mass index

	Category	Normal weight (n = 4071)	Overweight (n = 6711)	Obese (n = 3072)	Very obese (n = 1217)
	Diagnostic angiography	52%	55%	60%	69%
	Left ventricular dysfunction
	Normal	53%	54%	53%	51%
	Mild	26%	28%	27%	28%
	Moderate	17%	15%	16%	17%
	Severe	4%	3%	3%	3%
	Number of diseased vessels
	0	5%	3%	4%	7%
	1	41%	42%	39%	41%
	2	31%	30%	31%	30%
	≥3	23%	25%	27%	23%
	Percutaneous coronary intervention	30%	34%	37%	43%
	Stent placement	20%	23%	25%	31%
	Coronary artery bypass surgery	4%	5%	6%	5%
	Medication⁎	(n = 2498)	(n = 4061)	(n = 1807)	(n = 689)
	Before qualifying event
	Aspirin	68%	70%	69%	69%
	Beta-blockers	45%	51%	52%	53%
	Angiotensin-converting enzyme inhibitors	22%	23%	26%	30%
	Calcium channel blockers	16%	18%	20%	24%
	Diuretics	11%	11%	16%	21%
	Any lipid-lowering agent	15%	18%	22%	23%
	Statin	12%	14%	18%	17%
	Intravenous glycoprotein	3%	3%	4%	5%
	IIb/IIIa inhibitors
	Oral platelet inhibitor	3%	4%	5%	6%
	At randomization
	Aspirin	83%	83%	84%	84%
	Beta-blockers	70%	72%	75%	73%
	Angiotensin-converting enzyme inhibitors	35%	36%	39%	42%
	Calcium channel blockers	18%	19%	21%	24%
	Diuretics	15%	16%	19%	23%
	Any lipid-lowering agent	27%	31%	37%	37%
	Statin	24%	27%	33%	32%
	Intravenous glycoprotein	7%	9%	12%	15%
	IIb/IIIa inhibitors
	Oral platelet inhibitor	15%	16%	19%	25%

⁎ For patients with complete medication data by medication type available at both time points.

Use of concomitant medications 

As body mass index increased, patients were more likely to be treated with beta-blockers, angiotensin-converting enzyme inhibitors, calcium channel blockers, diuretics, statins, intravenous glycoprotein IIb/IIIa platelet inhibitors, and oral platelet aggregation inhibitors (Table 3). Although the use of these medications increased in all BMI groups between the qualifying event and randomization, differences among groups persisted.

Unadjusted clinical outcomes 

At 30 days, mortality was greatest among normal-weight patients (Table 4). Rates of 30-day death or myocardial infarction were similar in the overweight and obese groups and were lower than those in the normal-weight and very obese groups, which were similar. Revascularization and rehospitalization rates were highest for the very obese.

Table 4. Clinical outcomes by body mass index category

		Normal weight (n = 4071)	Overweight (n = 6711)	Adjusted hazard ratio (95% CI)	Obese (n = 3072)	Adjusted hazard ratio (95% CI)	Very obese (n = 1217)	Adjusted hazard ratio (95% CI)
	30-day
	Death	1.7%	1.0%	0.66 (0.47, 0.94)	0.9%	0.61 (0.39, 0.97)	1.2%	0.89 (0.48, 1.64)
	Death or myocardial infarction	5.1%	4.7%	0.96 (0.79, 1.15)	4.8%	0.95 (0.76, 1.19)	5.2%	0.87 (0.63, 1.20)
	Revascularization	18.5%	19.4%	—	18.5%	—	20.4%	—
	Rehospitalization	14.0%	13.8%	—	13.3%	—	15.3%	—
	90-day
	Death	2.6%	1.6%	0.68 (0.51, 0.90)	1.3%	0.60 (0.41, 0.88)	1.4%	0.70 (0.41, 1.21)
	Death or myocardial infarction	7.5%	6.7%	0.92 (0.79, 1.07)	7.0%	0.96 (0.79, 1.16)	6.7%	0.83 (0.63, 1.09)
	Revascularization	22.1%	23.6%	—	23.4%	—	24.2%	—
	Rehospitalization	24.3%	23.4%	—	23.6%	—	25.6%	—
	1-year
	Death	4.3%	2.7%	0.74 (0.60, 0.93)	2.2%	0.68 (0.51, 0.92)	2.6%	0.97 (0.64, 1.45)
	Revascularization	26.7%	28.1%	—	27.2%	—	28.3%	—
	Rehospitalization	33.0%	31.9%	—	31.9%	—	33.9%	—

Adjustment models: 30-day mortality, 90-day mortality: Age, heart failure between qualifying event and randomization, heart rate at randomization, history of heart failure, prerandomization percutaneous coronary intervention, prerandomization angiotensin-converting enzyme inhibitor, prior transient ischemic attack, streptokinase for qualifying event, qualifying event myocardial infarction, prior myocardial infarction, Latin American region of enrollment, randomized to control, prior severe chronic obstructive pulmonary disease, angina within 6 weeks before qualifying event, prior hypertension, prior angiography, serum creatinine at randomization. 30-day death/MI, 90-day death/MI: Age, prerandomization diuretic use, hours from qualifying event to treatment start, percutaneous coronary intervention between qualifying event and randomization, congestive heart failure between qualifying event and randomization, history of heart failure, elevated enzymes at qualifying event, angina within 6 weeks before qualifying event, Asian race, streptokinase for qualifying event, prior hypertension, heart rate at randomization, lateral electrocardiogram abnormality at qualifying event, randomized to high dose sibrafiban, prior myocardial infarction, Eastern European region of enrollment, diabetes mellitus with end organ damage, qualifying event Killip class >I, recurrent coronary ischemia between qualifying event and randomization, prerandomization heparin use, angiography between qualifying event and randomization, serum creatinine at randomization. 1-year mortality: Age, heart rate at qualifying event, heart rate at randomization, randomized to control group, white race, Hispanic ethnicity, Eastern Europe region of enrollment, Latin America region of enrollment, qualifying event myocardial infarction, qualifying event Killip class >I, prerandomization angiotensin-converting enzyme inhibitor use, prerandomization diuretic use, prior myocardial infarction, prior atrial fibrillation, past smoker, lateral electrocardiogram abnormality at qualifying event, prerandomization angiography, prerandomization thrombolysis, systolic blood pressure at qualifying event, congestive heart failure between qualifying event and randomization, prior congestive heart failure, prior hypertension.

At 90 days, mortality rates were higher among normal-weight patients who also had the highest rate of death or myocardial infarction and the lowest 90-day rate of revascularization. Similar patterns were observed at 1 year. As shown in Figure 1, 90-day survival for normal-weight patients was lower than that for other BMI groups. However, event-free survival (ie, freedom from death or myocardial infarction) was similar among BMI groups (Figure 2).

View Large Image Download to PowerPoint

Figure 1. 90-day unadjusted mortality by body mass index category. BMI = 18.5-24.9 kg/m2, heavy solid line; BMI = 25-29.9 kg/m2, dotted line; BMI = 30-34.9 kg/m2, thin solid line; BMI ≥35 kg/m2, dashes.

View Large Image Download to PowerPoint

Figure 2. 90-day unadjusted death or myocardial infarction by body mass index category. BMI = 18.5-24.9 kg/m2, heavy solid line; BMI = 25-29.9 kg/m2, dotted line; BMI = 30-34.9 kg/m2, thin solid line; BMI ≥35 kg/m2, dashes.

Adjusted clinical outcomes 

After multivariable adjustment, survival for very obese and normal-weight patients was similar (Table 4). However, the adjusted hazard ratios for the overweight and obese groups reflect a lower mortality risk at 30 days, 90 days, and at 1 year relative to normal-weight patients. Similar to the unadjusted analyses for death or myocardial infarction at 30 and 90 days, there were no statistically significant differences in the adjusted hazard ratios for overweight, obese, and very obese individuals, and the adjusted hazards were not significantly different from that of the normal weight reference group (Table 4).

Additional analyses containing stratified baseline variables (ie, diabetes, hypertension, and smoking status) as well as variables denoting postrandomization therapies produced results similar to those described above, demonstrating that these variables did not alter previously described associations between BMI group and this study’s endpoints.

Obesity and cardiovascular risk 

Between 75% and 80% of the US population is estimated to be overweight or obese, with one third being obese.3, 13 With the exception of Asia, overweight and obese patients represented the majority of individuals in every region participating in these trials. The highest prevalence of overweight patients was observed in the United States (78%), including 37% who had a BMI ≥30 kg/m2. These statistics underscore the importance of developing a better understanding of obesity as a disease, its influence on outcomes, and the effects of therapies.

The association of body weight with clinical outcomes has been assessed in multiple population-based studies. Although several studies demonstrated a linear association between increasing BMI and clinical risk,4, 6, 14, 15, 16, 17 others suggested a U-shaped relationship with increased risk among cohorts with the lowest and highest BMI.4, 18, 19, 20 Fewer studies have evaluated the association of BMI with prognosis once coronary artery disease is clinically manifest. Two studies reported lower short-term morbidity and mortality associated with greater BMI among patients undergoing percutaneous coronary intervention,21, 22 but higher BMI was associated with worse outcomes following coronary artery bypass graft.23 In contrast, 2 studies reported a U-shaped relationship between BMI and survival after percutaneous coronary intervention.24, 25, 26

Previous studies of the association of BMI with clinical outcomes following acute coronary ischemic events failed to demonstrate a linear association between overweight/obesity and clinical outcomes.7, 27, 28 Hoit and colleagues reported a U-shaped relationship between BMI and 1-year survival following myocardial infarction, and Kaplan et al found the same relationship at 3.4 years mean follow-up.27, 28 Similarly, we found that after multivariable adjustment, the most favorable outcomes were observed among overweight and obese individuals, with the very obese and the normal weight groups having the highest risk-adjusted mortality. The explanation for this U-shaped relationship remains unclear, but it persisted even with less extensive adjustment using age alone or age, sex, and race.

We observed that the use of evidence-based therapies was associated with increasing BMI, despite minimal differences in cardiovascular disease severity or left ventricular dysfunction. However, this is unlikely to be the sole explanation for better clinical outcomes among overweight and obese patients as adjusted outcomes for the very obese (the BMI group with the highest use of procedures and evidence-based medicines) were similar to those of normal-weight patients and worse than for the overweight and obese BMI groups.

There was a 7-year mean age difference between the normal-weight and very obese patients in our study. Previous studies reported premature occurrence of acute myocardial infarction in overweight and obese patients and that older patients receive fewer evidence-based therapies.29, 30, 31, 32, 33, 34 Thus, some of the differences among BMI groups in treatments received might be due to the tendency among clinicians to provide fewer evidence-based therapies to older patients. However, this would not explain the similarity between outcomes in the youngest group (very obese) and oldest group (normal weight).

Obesity is associated with multiple risk factors for the development of atherosclerotic vascular disease, such as sedentary lifestyle, type 2 diabetes, hypertension, and dyslipidemia.35 As in previous studies, we found that increasing BMI was associated with younger age but a progressively worsening cardiac risk profile.7 In previous studies, even after adjusting for other risk factors, obesity remained an independent predictor for coronary artery disease.36 Therefore, it is probable that other, obesity-associated factors (eg, hemodynamic burden,37 metabolic influences,38 and inflammation39) contribute to the development and progression of coronary artery disease. However, factors contributing to the genesis and progression of atherosclerotic disease are quite different from those that contribute to subsequent clinical outcomes.17, 40, 41 Thus, obesity may directly influence the prevalence of coronary artery disease but not the incidence of acute complications of the disease, including myocardial infarction, stroke, and cardiovascular death. Alternatively, some obese patients may have a more aggressive form of disease. In that situation, only younger obese patients who are more likely to survive despite a worse cardiac risk profile would be considered for analysis. In this regard, the findings with obesity may be similar to the smoker’s paradox, in which acute coronary syndrome patients who smoke are typically younger and have lower short-term mortality even after risk adjustment.42, 43, 44, 45, 46