Impaired Fasting Glucose and Cardiovascular Outcomes in Postmenopausal Women with Coronary Artery Disease
A. Kanaya,D. Herrington,E. Vittinghoff,F. Lin,V. Bittner,J. Cauley,S. Hulley,E. Barrett-Connor
DOI: https://doi.org/10.7326/0003-4819-142-10-200505170-00006
IF: 39.2
2005-05-17
Annals of Internal Medicine
Abstract:Context Impaired fasting glucose levels increase the risk for cardiovascular events, but the effect on recurrent cardiovascular events among people with known coronary heart disease (CHD) is unknown. Contribution This study followed 2763 postmenopausal women with known CHD for 6.8 years and found no association between fasting glucose levels of 5.6 to 6.9 mmol/L (100 to 125 mg/dL) (the 2003 definition of impaired fasting glucose) and new cardiovascular events. Women with fasting glucose levels of 6.1 to 6.9 mmol/L (110 to 125 mg/dL) (the 1997 definition of impaired fasting glucose) were at increased risk for new CHD events but not for congestive heart failure hospitalization, stroke, or transient ischemic attack. Implications Impaired fasting glucose is not a powerful risk factor for cardiovascular disease events among women who have already had a CHD event. The Editors An estimated 18 million U.S. adults have diabetes (1), and another 44 million have impaired fasting glucose, defined as a fasting glucose level between 5.6 mmol/L (100 mg/dL) and 6.9 mmol/L (125 mg/dL) (2). Diabetes is associated with increased risk for cardiovascular disease (CVD) in both men and women in most racial or ethnic groups and in almost every population studied (3, 4). Current national guidelines recommend that diabetes be considered a coronary heart disease (CHD) equivalent (5). Individuals with impaired fasting glucose have increased risk for future diabetes, and impaired fasting glucose is associated with other cardiovascular risk factors (6, 7). Moreover, individuals with impaired fasting glucose may also have increased risk for CVD (8-11). If impaired fasting glucose is associated with increased risk for CHD, then initiation of preventive interventions could be considered for this group of patients. This issue is complicated by the fact that the definition of impaired fasting glucose has recently changed. In 2003, the American Diabetes Association Expert Committee recommended that the glucose level for impaired fasting glucose be lowered from 6.1 mmol/L (110 mg/dL) to 5.6 mmol/L (100 mg/dL) (12). This decision was prompted partly by population-based studies showing that a cut-point of 5.6 mmol/L would increase the sensitivity of predicting future diabetes. No studies to date have examined the association between impaired fasting glucose according to this expanded definition and risk for future cardiovascular events. We examined data from 2763 postmenopausal women enrolled in the Heart and Estrogen/progestin Replacement Study (HERS) who were followed for an average of 6.8 years. We aimed to determine whether women with known coronary artery disease and impaired fasting glucose at baseline had an increased risk for a new CHD event, cerebrovascular event, hospitalization for congestive heart failure (CHF), or any cardiovascular event and to compare the magnitude of this increase with that observed in women with diabetes. We also evaluated the effect of lowering the cut-point for defining impaired fasting glucose on the association with cardiovascular outcomes. Methods Study Setting, Patients, and Design The design, methods, baseline characteristics (13), and main findings of the trial (14) and the cohort follow-up (15) have been published. Briefly, HERS was a randomized, blinded, controlled trial that evaluated daily hormone therapy for preventing coronary events in postmenopausal women with established CHD. Participants were randomly assigned to 0.625 mg of conjugated estrogens daily plus 2.5 mg of medroxyprogesterone acetate or placebo and were followed for a mean of 4.1 years. The trial enrolled 2763 women at 20 clinical centers in the United States between January 1993 and September 1994. The combined primary outcome of nonfatal myocardial infarction (MI) and CHD death did not significantly differ between women assigned to hormone therapy and those assigned to placebo (14). After completion of the trial, women enrolled in HERS were invited to enroll in a prospective cohort study for an additional 2.7 years. Of the 2510 women alive at the end of HERS, 2321 (92.4%) enrolled in HERS II. Contact information was available for more than 99% of the women in HERS II at the end of that trial (15). After 6.8 years of follow-up, hormone therapy did not reduce the risk for cardiovascular events in women with established CHD (15). For this analysis, we classified women as having diabetes if they reported a physician diagnosis of diabetes, were taking diabetes medication, or had a fasting serum glucose level of 7.0 mmol/L or more (126 mg/dL). We updated the diabetes category with incident cases of diabetes if the participant self-reported new diabetes or a complication directly related to diabetes, initiated therapy with a hypoglycemic medication at any point during follow-up, or had a fasting glucose level of 7.0 mmol/L or more (126 mg/dL) at the first annual visit or trial closeout visit. We classified impaired fasting glucose by using either the 1997 definition (fasting glucose level, 6.1 to 6.9 mmol/L [110 to 125 mg/dL]) or the 2003 definition (fasting glucose level, 5.6 to 6.9 mmol/L [100 to 125 mg/dL]) at baseline and updated it at the first annual visit and the trial closeout visit. Data Collection At baseline, participants completed a questionnaire that asked about their age, race or ethnicity, education, chronic medical illnesses, and previous percutaneous transluminal coronary angioplasty or coronary artery bypass grafting. We also assessed history of MI, percutaneous transluminal coronary angioplasty, or coronary artery bypass grafting by reviewing medical records at the end of the study period. At baseline, participants reported smoking habits (current, former, or never), alcohol consumption (drinks per week), regular exercise or walking activity, and overall general health. At baseline and at each annual visit, we performed a cardiac examination and assessed the following physical examination variables: body weight, height, waist and hip circumference, and systolic and diastolic blood pressure. We calculated body mass index (BMI) and performed standard 12-lead electrocardiography at baseline and at each annual visit. At the baseline, first annual, and end-of-trial visits, participants had fasting blood tests for measurement of total cholesterol, high-density lipoprotein (HDL) cholesterol, triglyceride, and lipoprotein(a) levels by the Lipoprotein Analytical Laboratory at Johns Hopkins Hospital, Baltimore, Maryland. Fasting serum glucose levels were measured at these 3 clinical visits by SmithKline Beecham Clinical Laboratory, Van Nuys, California, using the hexokinase enzymatic method. Outcome Ascertainment The primary outcome of HERS was any new CHD event, defined as a fatal or nonfatal MI, sudden CHD death, or other CHD death. We based the diagnosis of nonfatal MI on an algorithm that included ischemic symptoms, electrocardiogram abnormalities, and elevated cardiac enzyme levels (13). Independent physicians adjudicated CHD death if the participant had a fatal documented MI, sudden death within 1 hour of symptom onset, unobserved out-of-hospital death without another known cause, or death due to CHF or coronary revascularization procedure. The secondary cardiovascular outcomes we examined included stroke and transient ischemic attack (TIA), hospitalizations for CHF, and any CVD event (an aggregate of all cardiovascular outcomes examined in our analysis). Statistical Analysis We assessed the effects of glucose status on various study end points by using multivariable Cox proportional hazards models. We adjusted our models for potential confounders that we determined a priori: age, race or ethnicity, education, current smoking, physical activity, alcoholic drinks per week, BMI, self-reported health (poor/fair vs. good/very good/excellent), previous percutaneous transluminal coronary angioplasty, previous coronary artery bypass grafting, signs of CHF on physical examination, more than 1 previous MI, statin use, diuretic use, angiotensin-converting enzyme inhibitor use, and hormone therapy assignment. We also stratified our models by clinical center. We updated information on glucose status, current smoking, exercise, alcoholic drinks per week, BMI, overall health status, statin use, diuretic use, and angiotensin-converting enzyme inhibitor use by using time-dependent covariates, and we modeled the effect of hormone therapy assignment by year. In supplementary analyses, we further adjusted for variables that may be in the causal pathway between hyperglycemia and cardiovascular outcomes, such as waist circumference, HDL cholesterol level, triglyceride level, systolic blood pressure, and lipoprotein(a) level, all as time-dependent covariates. We qualitatively compared the difference in cardiovascular outcomes for the impaired fasting glucose group by using the 1997 definition (fasting glucose level, 6.1 to 6.9 mmol/L [110 to 125 mg/dL] [16]) and the 2003 definition (fasting glucose level, 5.6 to 6.9 mmol/L [100 to 125 mg/dL] [12]). We also evaluated the effect of a baseline fasting glucose level of 5.6 to 6.0 mmol/L (100 to 109 mg/dL) on each cardiovascular outcome, adjusting for the potential confounders listed. We also obtained an approximate test of the equivalence of the 1997 and 2003 definitions of impaired fasting glucose by comparing the Cox model coefficients for the subgroups with fasting glucose levels of 5.6 to 6.0 mmol/L (100 to 109 mg/dL) and 6.1 to 6.9 mmol/L (110 to 125 mg/dL). In model checking, we examined potential interactions of impaired fasting glucose and diabetes with age, race or ethnicity, and hormone therapy use, as well as interactions of diabetes and oral hypoglycemic medication or insulin use, by using product terms. We assessed the proportional hazards assumption by fitting models that allowed for variation in the relative hazards for diabetes and impaired fasting glucose either smoothly, by using prod