Author + information
- Received September 11, 2001
- Revision received March 28, 2002
- Accepted April 5, 2002
- Published online July 3, 2002.
- Martin K Rutter, MB ChB, MRCP*,* (, )
- Shahid T Wahid, MB BS, MRCP*,
- Janet M McComb, MD, FRCP† and
- Sally M Marshall, MD, FRCP*
- ↵*Reprint requests and correspondence:
Dr. Martin K. Rutter, Cardiology Department, The Lahey Clinic, 41 Mall Road, Burlington, Massachusetts 01805, USA.
Objectives The aim of this study was to investigate the relationships between future coronary heart disease (CHD) events and baseline silent myocardial ischemia (SMI) and microalbuminuria (MA) in subjects with type 2 diabetes (T2D) free from known CHD.
Background Coronary heart disease is often asymptomatic in subjects with diabetes. There is limited information on the prognostic value of SMI and MA in this group.
Methods Eighty-six patients with T2D and no history of CHD were studied (43 with MA individually matched with 43 normoalbuminuric patients; mean [SD] age 62 [±7] years, 62 men). Metabolic assessment, three timed overnight urine collections for albumin excretion rate, a treadmill exercise test and ankle brachial index (ABI) were performed at baseline. Patients were followed for 2.8 years.
Results Forty-five (52%) patients had SMI during treadmill testing. At review, there had been 23 coronary (CHD) events in 15 patients. Univariate Cox regression analysis showed that CHD events were significantly related to baseline ABI (p = 0.014), SMI (p = 0.020), MA (p = 0.046), 10-year Framingham CHD risk >30% (p = 0.035) and fibrinogen (p = 0.026). In multivariate analysis, SMI was the strongest independent predictor of CHD events (p = 0.008); risk ratio (95% confidence interval) for SMI: 21 (2 to 204). In the prediction of CHD events, SMI showed higher sensitivity and positive predictive value than MA or Framingham calculated CHD risk.
Conclusions The presence of baseline SMI and MA are associated with future CHD events in asymptomatic patients with T2D and may be of practical use in risk stratification.
Coronary heart disease (CHD) is the leading cause of death in patients with type 2 diabetes (T2D), is often asymptomatic (1)and may present without warning as acute myocardial infarction (AMI), heart failure, arrhythmia or sudden death. In AMI (2)and heart failure (3), mortality is increased in the presence of T2D, thus emphasizing the potential value of identifying high-risk asymptomatic individuals with diabetes.
Microalbuminuria (MA) is present in approximately 25% of patients with T2D (4)and is associated with a doubling of the risk of early death, mainly from CHD (5). Microalbuminuria has been defined by consensus, based on risk of renal disease, as a urinary albumin excretion rate between 20 and 200 μg/min, though rates of >10.6 μg/min have been linked to increased macrovascular events in T2D (6).
Silent myocardial ischemia (SMI) can be detected by various methods (7,8). Using treadmill exercise testing, SMI has been defined as exercise-induced ST-segment depression in the absence of CHD symptoms (9), and, in men free from known CHD, this finding has been associated with increased mortality (10,11). There is very little data on the prognostic value of SMI, detected by any method, in asymptomatic patients with T2D (7,12–14).
We have previously shown that T2D is associated with a high prevalence of SMI, especially in those with MA (15). This cohort, free from symptoms of CHD at baseline, has been followed to determine the influence of SMI and MA on prognosis.
Baseline clinical characteristics and methods have been described previously (15). Briefly, 86 patients with T2D and no history of cardiac disease were studied: 43 with MA (albumin excretion rate [AER] > 10.5 to 200 μg/min) individually matched with 43 normoalbuminuric patients (AER < 10.5 μg/min) for age (±2 years), gender, diabetes duration (±3 years) and smoking status (never/previous/ex).
All patients performed three timed overnight urine collections to assess AER and treadmill exercise testing for SMI. Ankle brachial index (ABI), echocardiography, ambulatory blood pressure (BP) monitoring, autonomic function testing and fasting blood testing for metabolic parameters were performed. Exercise tests were reported blind, and SMI was defined as >1 mm down sloping or horizontal ST-segment depression from baseline occurring 80 ms after the J point for three consecutive beats. A test was defined as suboptimal if the patient failed to achieve 85% of their age-predicted maximum heart rate. Ten-year probability of coronary events (the Framingham 10-year CHD risk) were calculated using the software package supplied with the Joint British Recommendations on the Prevention of Coronary Heart Disease (16). For technical reasons, accurate echocardiographic measurements of the left ventricle were possible in only 67 subjects.
Follow-up data and definition of events
All patients were under regular review at diabetes clinics and received detailed structured hospital annual review. The following information was recorded from the most recent annual review proforma: diabetes treatment, medication, claudication, BP, body mass index (BMI), glycated hemoglobin (HbA1c), lipids, creatinine, urinary albumin-creatinine ratio, smoking status, alcohol intake and information regarding CHD events. When a CHD event had occurred, hospital casenotes were reviewed for verification and determination of the precise timing of the event. Predetermined primary end points were as follows: 1) sudden cardiac death was defined as sudden unexplained death; 2) AMI was defined as chest pain with characteristic electrocardiogram (ECG) or creatine kinase changes; and 3) new onset angina was defined as exercise-related chest discomfort diagnosed clinically or by investigation. If a patient experienced more than one event during follow-up, then only the first was recorded, and no additional data was used for survival analysis. If two events occurred simultaneously, then the more severe event was recorded.
Calculation of test performance
Patients were recruited as matched pairs of microalbuminuric and normoalbuminuric patients. Events occurring within the shorter duration of follow-up (median [SD] 2.5 [0.9] years) for individuals within each pair were used to calculate sensitivity, specificity and positive and negative predictive value.
Glycated hemoglobin, lipids and fibrinogen were measured using standard assays. Urinary albumin concentration was determined using radioimmunoassay (interassay coefficient of variation of 3.6% at a urine concentration of 4.95 mg/l) (17).
Results are presented as means (SD) or medians (range). Chi-squared or Fisher’s exact test were used to compare categorical data. For continuous variables, differences between groups were compared using either the two sample Students ttest or the Mann-Whitney Utest. Results were considered statistically significant if p < 0.05 unless otherwise stated. Kaplan-Meier survival curves were computed for categorical risk factors, and comparison of survival curves was performed using the single variable log-rank test. Univariate Cox regression analyses were performed with coronary event as the outcome variable. The following univariate predictors were tested: age, gender, diabetes duration, smoking status, baseline claudication, BMI, clinic and 24-h ambulatory BP, total cholesterol, high-density lipoprotein cholesterol, HbA1c, fibrinogen, AER, MA, positive exercise test, ABI, echocardiographically determined left ventricular mass index, number of abnormal autonomic function tests, heart rate variability and calculated Framingham coronary risk (incorporating ECG evidence of left ventricular hypertrophy). Univariate predictors with p < 0.10 were entered into multivariate Cox regression analysis. Calculations were performed using the Minitab (Minitab Inc., State College, Pennsylvania) and SPSS (SPSS Inc., Surrey, United Kingdom) statistical packages. The study was approved by our local ethics committee, and all patients gave informed consent.
Baseline clinical characteristics
Baseline characteristics have been presented previously (15). Briefly, microalbuminuric patients were more likely to have symptoms of claudication, peripheral neuropathy and have evidence of diabetic retinopathy compared with those with normoalbuminuria (Table 1). Those with MA also had increased BMI, increased clinic and ambulatory systolic BP, increased fibrinogen and HbA1c, reduced ABI and reduced heart rate variability.
Treadmill exercise testing
Compared with normoalbuminuric patients, those with MA had a higher prevalence of positive tests (28 vs. 17, p < 0.05), exercised for shorter periods (5.0 vs. 7.1 min, p < 0.01), performed less work (6 vs. 8 METs, p < 0.01), achieved lower heart rates, had more suboptimal tests and a higher number of negative suboptimal tests (15).
Medication at baseline and review
Patients with either MA or SMI were more actively treated than patients without. Those with MA were more likely to be prescribed beta-blockers (number of patients: 8 vs. 1) at baseline and more likely to be prescribed aspirin (28 vs. 13), angiotensin-converting enzyme inhibitors (22 vs. 10), beta-blockers (14 vs. 3), nitrates (15 vs. 2) and statins (22 vs. 4) at final review. Those with SMI were more likely to be prescribed aspirin (24 vs. 14), beta-blockers (15 vs. 2) and nitrates (16 vs. 1) at final review.
Follow-up data were obtained in all patients. Median (range) follow-up was 2.8 (1.3 to 4.9) years during which time patients were attending diabetes clinics and were receiving standard clinical advice on appropriate risk factor management. At the time of follow-up, there had been 23 CHD events (cardiac death, 3; AMI, 7; new-onset angina, 13) in 15 patients (annual patient event rate = 15/[86 × 2.8] = 6.2%).
Relationship of SMI and MA to CHD events
Figures 1 and 2⇓⇓show Kaplan-Meier CHD event-free survival plots for patients with and without SMI and for patients with and without MA. Comparing single risk factors, SMI was more strongly related to the presence of CHD events than the presence of MA (log-rank test: 8.7, p = 0.003 vs. 4.8, p = 0.029). Figure 3shows a comparison of coronary event-free survival for the following subgroups: 1) patients with neither SMI nor MA; 2) patients with MA but no SMI; 3) patients with SMI but no MA; and 4) patients with both SMI and MA (n = 28). Patients with neither risk factor (n = 26) were less likely to experience CHD events compared with the remaining patients (log-rank test 4.6, p = 0.03). Patients with both SMI and MA were much more likely to experience CHD events compared with other patients (log-rank test 11.3, p = 0.0008).
Cox regression: predictors of CHD events
When considered as a continuous variable, baseline ABI was the variable most strongly related to the presence of future CHD events (Table 2). However, when considered as a discrete variable (ABI > or < 0.9) (18), the relationship was not significant (p = 0.08). The presence of SMI, MA and Framingham 10-year CHD risk >30% were all significantly related to future CHD events. When considered as a continuous variable, baseline fibrinogen was also linked to future CHD events. In multivariate analysis, the only independent predictors of CHD events were SMI (p = 0.008) and ABI (p = 0.032). Microalbuminuria was not independently related to CHD events after correcting for the effects of SMI, ABI, BMI and fibrinogen. Hazard ratios (95% confidence interval) for SMI were 21 (2 to 204) and for ABI were 17 (1.3 to 213) for a difference in ABI of one unit. When both MA and SMI were entered together as predictor variables in a multivariate Cox regression model, only SMI (beta [se], 2.2 [1.0]; p = 0.03) was independently related to CHD events (beta [se], 1.2 [0.8]; p = 0.11 for MA).
Individual and combined test performance
Silent myocardial ischemia was the single most sensitive test for the identification of patients developing CHD events (Table 3). Ankle brachial index showed low sensitivity (50%) for predicting coronary events when considered as a discrete variable (threshold < 0.9). When different ABI threshold values were used, improved sensitivity occurred at the expense of reduced specificity and positive predictive value. Combining SMI and MA results identified patient groups at particularly high and low risk of future CHD events.
This study has shown that the presence of SMI, as assessed by exercise ECG, is strongly and independently related to subsequent CHD events in asymptomatic patients with T2D.
Prevalence of SMI and CHD events
The high prevalence of SMI and the high event rate in this study is probably explained by the selection of only T2D patients, the age of the patients, the high proportion (50%) with MA and the geographical location of the study population (northeast England) where there is a high background prevalence and incidence of CHD (19).
SMI and prognosis
To our knowledge there has been only one previous study showing that ST-segment change during treadmill exercise testing has prognostic value in asymptomatic patients with T2D (12). However, this study was limited by male selection, poor characterization of diabetes type, incomplete follow-up, wide variation of follow-up duration and failure to assess albuminuria status. Two recent studies of the relationship between ST-segment change during exercise and coronary events have yielded negative results (7,13). The relatively young age of some patients studied, the grouping of subjects with type 1 and type 2 diabetes and incomplete follow-up may have influenced the results of these studies. In the study presented here, angina was the first manifestation of CHD in about half of the subjects, the others presenting with AMI or cardiac death. Thus, the presence of SMI is not simply a prelude to symptomatic (loud) ischemia (20). Similar findings were noted in the Framingham study and other studies (21), and the proportions of patients presenting with hard events were also similar in population-based studies of asymptomatic subjects with SMI (22,23).
SMI and test performance
Rubler et al. (12)found that ST-segment change during treadmill exercise testing in diabetes had a low sensitivity (50%) and high specificity (83%) in the prediction of CHD events. The higher sensitivity of treadmill exercise testing in our study is probably explained by patient selection. Our patients were older (62 vs. 53 years) and a high proportion had MA. These factors are likely to have increased the proportion of patients with more severe CHD, which would increase the sensitivity of treadmill testing. The high incidence of CHD events in our study population would have increased the positive predictive value compared with testing in lower risk groups.
Influence of medication
Data on prescribed medication and the risk factor changes during follow-up (data available on request) strongly suggest that high event rates in patients with SMI and MA have not been due to differences in management.
MA and ABI as risk factors
To our knowledge, this is the first study to show that ABI is significantly related to CHD events in asymptomatic patients with T2D. However, the data suggests that use of threshold ABI values lack the high sensitivity and specificity required for clinical use. Studies in the general population have shown the prognostic value of ABI (24), and further studies in diabetes are needed to determine if the test can be useful clinically. We have confirmed the results of previous studies in patients with T2D showing that the presence of MA is a significant coronary risk factor (5,14).
Influence of exercise capacity
Inclusion in the present study required patients to be ambulant and free from severe chronic disease. However, 45% of the study patients exercised for <6 min, and 13% had suboptimal tests. Although exercise duration was less in patients experiencing a coronary event (4.9 [3.4 to 10.0] vs. 6.3 [1.3 to 13.1] min), it was not significantly related to CHD events in Cox regression analysis. Vanzetto et al. (14)recently studied high-risk subjects with T2D, a third of whom had evidence of baseline CHD. In patients predicted by questionnaire to be unable to perform treadmill exercise testing, large perfusion defects on dipyridamole thallium-201 single-photon emission computed tomography predicted CHD events. Patients predicted to be able to exercise were a lower risk group, and, in these patients, treadmill testing was of no prognostic value. It seems likely that exclusion of higher-risk subjects from treadmill testing in the Vanzetto et al. (14)study reduced the power to show the prognostic value of SMI.
The sample size and number of cardiovascular events in this study are relatively small, and confidence intervals around estimates of test performance are likely to be wide. Patients in the study were of European race and were selected from a geographical area with a high background prevalence of CHD. Event rates in this population were higher than in many previous studies, and results may not be applicable in lower-risk populations. Data on prognosis and SMI refer to exercise-induced ST depression; other detection modes were not assessed.
This study has shown that baseline SMI and MA are significantly related to future CHD events in asymptomatic patients with T2D. The relationship of SMI to CHD events is independent of the presence of MA. This study suggests that SMI, MA and ABI could be of practical value in risk stratification.
Further studies are required to clarify the natural history of CHD in optimally managed medically treated asymptomatic patients with diabetes and, in particular, to accurately quantify the risk of hard and soft CHD events, arrhythmic death and heart failure. Future studies should aim to determine appropriate risk thresholds for initiation of anti-ischemia therapy and for CHD screening to identify those who are likely to benefit from revascularization.
The authors thank Dr. R. W. Nesto for his helpful comments in the preparation of this manuscript.
☆ Supported by fellowships and grants from the Northern Regional Health Authority, the Freeman Hospital Board of Trustees, Novo Nordisk Ltd., Eli Lilly, Bayer and GlaxoSmithKlein.
- ankle brachial index
- albumin excretion rate
- acute myocardial infarction
- body mass index
- blood pressure
- coronary heart disease
- glycated hemoglobin
- silent myocardial ischemia
- type 2 diabetes
- Received September 11, 2001.
- Revision received March 28, 2002.
- Accepted April 5, 2002.
- American College of Cardiology Foundation
- Miettinen H.,
- Lehto S.,
- Salomaa V.,
- et al.
- Marshall S.M.
- Valensi P.,
- Sachs R.N.,
- Harfouche B.,
- et al.
- Laukkanen J.A.,
- Kurl S.,
- Lakka T.A.,
- et al.
- Vanzetto G.,
- Halimi S.,
- Hammoud T.,
- et al.
- Joint British Recommendations on Prevention of Coronary Heart Disease in Clinical Practice
- Christansen C.,
- Orskov C.
- Fabsitz R.R.,
- Sidawy A.N.,
- Go O.,
- et al.
- Roper N.A.,
- Bilous R.W.,
- Kelly W.F.,
- Unwin N.C.,
- Connolly V.M.
- Fleg J.L.,
- Gerstenblith G.,
- Zonderman A.B.,
- et al.
- Thaulow E.,
- Erikssen J.,
- Sandvik L.,
- Erikssen G.,
- Jorgensen L.,
- Cohn P.F.
- Hooi J.D.,
- Stoffers H.E.,
- Knottnerus J.A.,
- van Ree J.W.