Patients with type 2 diabetes and stable coronary artery disease (CAD) may present with typical angina, angina equivalents such as effort dyspnea, or no cardiac symptoms. Such diversity raises the question whether symptom status affects prognosis in these patients. Silent ischemia may be the first manifestation of ischemic heart disease and may occur following coronary revascularization in previously symptomatic patients. Exertional dyspnea as an angina equivalent could be a manifestation of ischemia-induced systolic and/or diastolic dysfunction and might signal severe coronary disease portending a worse prognosis. On the other hand, patients with typical angina might come to earlier medical attention and treatment and have a better prognosis than patients without “warning” symptoms. Prognostic studies of patients with these clinical presentations are few, have disparate findings, and are limited by small sample size, lack of documentation of CAD, and short follow-up ((1),(2),3). Given the limited and inconclusive evidence base and the need to optimize risk stratification in patients with type 2 diabetes and stable CAD, it is important to determine whether symptom status impacts prognosis in these patients.

The BARI 2D (Bypass Angioplasty Revascularization Investigation 2 Diabetes) trial was carried out in 2,368 patients with type 2 diabetes and stable CAD with documented myocardial ischemia who were followed for 5.3 years. At entry into the study, BARI 2D patients had angina, angina equivalents, or no cardiac symptoms (4). Data from this trial constitute an important source of information to assess the prognostic relevance of symptom status in these patients. We thus tested the hypothesis that BARI 2D patients without angina or angina equivalent symptoms at entry have a risk of all-cause mortality and cardiovascular events similar to that of patients who have these symptoms.

The BARI 2D trial was a randomized trial with a 2 × 2 factorial design to determine whether a treatment strategy of prompt coronary revascularization and optimal medical therapy was superior to a strategy of initial optimal medical treatment alone with the option of subsequent revascularization and whether a strategy using insulin-sensitizing agents was more efficient than a strategy using insulin-providing agents to reduce total mortality and the composite of death, nonfatal myocardial infarction (MI) and nonfatal stroke. Details of the BARI 2D trial have been published ((4),(5),6) and are briefly outlined.

The 2,368 participants in the BARI 2D trial were recruited between January 1, 2001, and March 31, 2005, at 49 clinical sites: 1,499 participants were from the United States, 356 from Brazil, 353 from Canada, 85 from Mexico, and 75 from Austria and the Czech Republic. Participants had to be age ≥25 years and have type 2 diabetes, stable CAD (≥50% stenosis in a major epicardial coronary artery), and myocardial ischemia defined as a positive electrocardiographic, myocardial perfusion scan or echocardiographic test induced by exercise or pharmacological agents or the presence of typical angina with coronary stenosis ≥70%. Patients with unstable angina who became stable on medical therapy before entry into the study were included. All patients had to have suitable coronary anatomy for randomization into either percutaneous coronary intervention (PCI) strata or coronary artery bypass graft surgery (CABG) strata. The main exclusion criteria were a clinical indication for immediate revascularization, coronary revascularization within the 12 months before randomization, left main coronary disease, class III or IV heart failure, serum creatinine >2.0 mg/dl (177 μmol/l), and glycosylated hemoglobin (HbA_1c) >13.0%. The protocol was approved by each institution's review board or ethics committee. All participants provided written informed consent.

The principal investigator at each BARI 2D center had to classify the cardiac symptoms according to a standardized questionnaire into angina, angina equivalents, and no symptoms based on the medical history and findings of the stress test. Angina was characterized by ischemic chest pain occurring on exertion or stress relieved by rest and/or nitroglycerine. The participants with angina were also characterized according to their Canadian Cardiovascular Society (CCS) grades of angina (7) including stabilized unstable angina in grade IV CCS angina. Angina equivalents had to be manifested by dyspnea, fatigue, or diaphoresis on exertion evaluated by the principal investigator as abnormal and attributable to myocardial ischemia. Finally, participants without symptoms were classified as asymptomatic. Patients with both angina and angina equivalents were included in the angina group. Among the 2,368 participants enrolled in the BARI 2D study, there was satisfactory information on baseline angina status in 2,364, and they constituted the sample of the present study.

Patients randomized to the prompt revascularization strategy had to have PCI or CABG within 4 weeks of randomization. Optimal medical therapy in both strategies consisted of lifestyle management targeting smoking cessation, diet, weight loss, regular physical exercise, and pharmacological therapy to maintain HbA_1c <7.0%, low-density lipoprotein cholesterol <100 mg/dl (2.6 mmol/l), and blood pressure ≤130/80 mm Hg. Participants received individualized antiangina medications, including beta-blockers, calcium channel blockers, and long-acting nitrates. If during the study patients developed incapacitating angina or angina equivalent symptoms, if patients had documented worsening ischemia despite optimal medical treatment, or if an acute coronary syndrome occurred, the treating cardiologist could recommend subsequent coronary revascularization, which was defined in the present analysis as the first revascularization procedure for patients randomized to optimal medical strategy or the repeated revascularization procedure for patients randomized to prompt revascularization strategy. Participants were evaluated on a monthly basis for 6 months and every 3 months thereafter for a mean follow-up of 5.3 years.

The outcomes were all-cause death and the composite of all-cause death, nonfatal MI, and nonfatal stroke and have been previously described (4). Events were adjudicated by an independent committee unaware of the study group allocation. Cardiac death included sudden death, fatal MI, congestive heart failure death, death within 30 days of a cardiac procedure, cardiogenic shock, unwitnessed death, and presumed cardiac death. The clinical diagnosis of nonprocedural MI was based on cardiac markers, either troponin or creatine kinase MB, with evidence of ischemia on the basis of symptoms, electrocardiography, or imaging. Stroke was defined as a rapid neurological deficit attributed to an obstruction or rupture of cerebral artery lasting >24 h or <24 h if there was a lesion on computerized tomography or magnetic resonance imaging compatible with an acute stroke.

Major comparisons were made among the 3 groups based on baseline symptom status. Analysis of variance was used for comparison of continuous variables and chi-square test for categorical variables. If the 3-group comparison was statistically significant, then post hoc pair-wise comparisons between each pair of symptom status were performed to demonstrate which pair differed significantly. The pairwise comparisons used the Tukey-Kramer test for continuous variables and the Wald test for categorical variables. Furthermore, comparisons among the severity degree (CCS grades) of baseline angina were also conducted for all outcomes. Paralleling the aggregate analysis on all BARI 2D patients, the stratified analyses were performed on the 2 subgroups stratified by the randomization strategy (i.e., the prompt revascularization strategy vs. the optimal medical strategy) to address the possible confounding effect of the initial revascularization procedures on one-half of the BARI 2D patients within 4 weeks of enrollment. The cumulative incidence rate of outcomes was estimated with the Kaplan-Meier method, and rate curves were compared using the log-rank test. Multivariable Cox regression models were built to estimate hazard ratios (HRs). After adjustment for randomized strategies and prespecified revascularization strata (PCI or CABG), the stepwise selection algorithm chose significant confounders from the baseline characteristics, including demographics, clinical history, blood and urine measures, physical status, angiographic characteristics, and relevant medications, as listed in (Table 1) and (6). The interaction effect between baseline angina and randomized treatment strategies was tested and dropped when there was no significant interaction effect. For correction of the multiple comparison issue, the statistical significance level was set at 0.01 by Bonferroni correction. Analyses were performed using SAS version 9.2 (SAS Institute Inc., Cary, North Carolina), and plots were drawn using R (R Development Core Team, Wien, Austria).

Among the 2,364 patients at entry into the study, there were 1,434 (61%) with angina, 506 (21%) with angina equivalents, and 424 (18%) with neither angina nor angina equivalents. Among the 1,434 patients with angina, 1,005 had CCS grades I and II and 429 had CCS grades III and IV. Baseline characteristics by angina status are summarized in (Table 1). The significant results from the pairwise comparisons are presented for each with p < 0.01. Compared with patients in the other groups, patients with angina were younger, were more often women, and were more frequently on beta-blockers and nitrates. Patients with angina had more previous PCIs than asymptomatic patients and a slightly higher left ventricular ejection fraction compared with patients with equivalent angina. Compared with the other groups, the asymptomatic group had a higher rate of men and a lower rate of hypertension; the asymptomatic patients had a higher index of myocardium at risk compared with patients with equivalent angina. Patients with equivalent angina were more frequently on diuretics compared with the others. Body mass index (BMI) differed among the 3 groups. There was no significant difference among the 3 groups with respect to the duration of diabetes, neuropathy score, lipid and glucose profiles, blood pressure, smoking status, renal function, previous MI, stroke, or use of aspirin, statins, and renin-angiotensin modulators.

Compared with the baseline observations, each of the 3 groups at the 3-year follow-up had a decrease in the rates of current smokers, HbA_1c, low-density lipoprotein cholesterol, and blood pressure and an increase in high-density lipoprotein cholesterol. However, there was a small but significant increase in BMI in the 3 groups. Use of long-acting nitrates decreased, and use of beta-blockers and calcium channel blockers increased; there was significantly higher use of nitrates and calcium blockers in the angina group. During the 3-year follow-up, the asymptomatic patients had a lower rate of subsequent PCI than the patients in the other 2 groups, but there was no difference in the rates of subsequent CABG among the 3 groups (6). At the 5-year follow-up, the cumulative Kaplan-Meier rates of the subsequent revascularization (PCI or CABG) were 25% (98 of 424) in the asymptomatic group, 32% (144 of 506) in the equivalent angina group, and 35% (450 of 1,434) in the angina group (p < 0.001). The 4 main independent reasons for revascularization in the asymptomatic group at entry were acute coronary syndrome in 28.6%, worsened ischemia in 22.4%, severe angina in 22.4%, and CAD progression in 12%. For the angina equivalent and angina groups, it was severe angina in 45% for both groups, acute coronary syndrome in 22.2% and 23.3%, worsened ischemia in 14.6% and 17.8%, and CAD progression in 11.8% and 6.4%, respectively.

At the 5-year follow-up, there were 316 deaths and 548 patients with a first occurrence of the composite of all-cause death, nonfatal MI, or nonfatal stroke. The cumulative 5-year death rate was 12% in patients with angina, 14% in patients with angina equivalents, and 10% in asymptomatic patients (log-rank p = 0.3). The cumulative 5-year composite outcome rate was 24% in patients with angina, 24% in patients with angina equivalents, and 21% in asymptomatic patients (Figure 36_gr1). As shown in (Table 2), there was no difference in the unadjusted HR as well as the HR adjusted for all covariates between the asymptomatic group and each of the symptomatic groups for all-cause mortality, composite outcome, cardiovascular disease (CVD) death, nonfatal MI, and nonfatal stroke.

Stratified analyses were performed on patients randomized to the prompt revascularization strategy as well as patients randomized to the optimal medical strategy to address the possible confounding effect of the initial revascularization procedures. In the patients randomized to optimal medical group strategy, there were no significant differences among the asymptomatic, angina equivalent, and angina groups for the cumulative 5-year death rate, death, MI or stroke, CVD death, nonfatal MI, and nonfatal stroke; there was also no difference in the unadjusted and adjusted HR between the 3 groups and the different outcomes ((Table 2), 6). Similar findings were observed for the same outcomes in the 3 groups of patients randomized to the prompt revascularization strategy ((Table 2), 6).

There was also no difference in the adjusted HR for each outcome between the asymptomatic group and the symptomatic groups when the data were restricted to the 842 patients with either triple-vessel disease or double-vessel disease including left proximal anterior descending coronary artery stenosis (6).

The higher rates of outcomes among patients with CCS grades III and IV compared with patients with CCS grades I and II were not statistically different and not different when compared with asymptomatic patients and patients with angina equivalents (Table 2). However, patients with CCS grades III and IV had higher rates of revascularization during the study (Figure 36_gr2).

In this study, patients with stable CAD and type 2 diabetes had a similar long-term prognosis across a variety of cardiovascular outcomes irrespective of their symptom status. Indeed, asymptomatic patients at entry into the study were at the same relatively high risk for all-cause mortality, the composite outcome, CVD death, nonfatal MI, and nonfatal stroke as patients with angina and patients with angina equivalent symptoms. These observations were independent of clinical baseline characteristics, severity of CAD, and coronary revascularization. Furthermore, the results remained unchanged when the optimal medical therapy arm and the revascularization arm were analyzed separately.

A recent analysis of the symptoms at entry into the BARI 2D patients showed that the absence of angina in the asymptomatic or angina equivalent patients may be explained in part by a previous revascularization (8), but nearly half of them presented with angina during the follow-up (9). Nevertheless, it remains intriguing why patients without angina but with documented myocardial ischemia had no symptoms or an angina equivalent. A study using positron emission tomography to evaluate regional cerebral blood flow suggested that abnormal central nervous system processing of painful stimuli is associated with silent myocardial ischemia (10). Autonomic neuropathy has also been thought to explain the lower frequency of angina in patients with diabetes and evidence of myocardial ischemia compared with patients without diabetes ((11),(12),(13),14). Regardless of the mechanisms responsible for symptom status in patients with diabetes and CAD, our findings suggest that these very different presentations are not explained by differences in the angiographic severity of CAD nor in left ventricular function.

Patients with and without diabetes and silent myocardial ischemia have better survival than patients with symptomatic ischemia 6 months after PCI (3). However, similar survival rates were reported between 45 asymptomatic and 37 angina patients with diabetes, documented angiographic CAD, and myocardial ischemia during a 6-year follow-up (1). Similar composite outcomes of death and nonfatal MI were found in 286 asymptomatic and 321 angina patients with diabetes and induced myocardial ischemia but without documented CAD during a median follow-up of 2 years (2). Our results confirmed these 2 studies in a much larger population followed for 5.3 years and included more outcomes and showed no difference in survival of patients according to their symptoms even after coronary revascularization. Considering the whole spectrum of silent ischemia in patients with type 2 diabetes, these findings are consistent with the similar survival rates between patients with diabetes with unrecognized MI and those with recognized MI (15).

It might have been expected that the predominance of exertion dyspnea as an angina equivalent would be associated with poorer left ventricular function and outcomes, as reported by Zellweger et al. (2), but this was not the case in our study in which left ventricular ejection fractions were 55%, 58%, and 57% in patients with angina equivalent symptoms, angina, and no symptoms, respectively. However, it is possible that exertion dyspnea may be related to diastolic ventricular dysfunction during exercise-induced ischemia, a variable not assessed in most of the patients. We cannot exclude the possibility that exertion dyspnea might have been related to other noncardiac factors such as increased BMI and physical deconditioning. It is interesting to note that myocardial jeopardy score was actually significantly highest in the asymptomatic subset and not in the angina equivalent subset. In contrast to the findings of Zellweger et al. (2), we found no significant difference in the outcomes between the patients who had only dyspnea compared with the others who did not (6).

Recruitment in the BARI 2D study was completed among patients with stable CAD and myocardial ischemia with mainly mild or no symptoms. However, a third of the patients with angina were classified as CCS grades III and IV, and half of the latter had a stable status before randomization after a recent history of unstable angina. These patients did not have significantly higher rates of events compared with patients with CCS grades I and II, as previously reported in a cohort of consecutive patients undergoing coronary angiographic studies (16). However, participants with CCS grades III and IV angina did have a higher rate of revascularization during the study than those with angina grades I and II. Although this may be one of the factors explaining the nonsignificant differences, the CCS gradation is known to have limitations, particularly for CCS grade IV (17).

Our study was a post-hoc analysis of the BARI 2D trial with its inherent limitations and should be considered as a hypothesis-generating study. The BARI 2D cohort was a selected population with stable CAD anatomy suitable for revascularization without standardized testing to determine the myocardial ischemia threshold. Nevertheless, our study has several strengths: it is based on a large sample of patients with well-documented and detailed baseline characteristics including systematic coronary angiography, myocardial ischemia, intensive risk factor management during a long follow-up, close monitoring, and blinded adjudication of fatal and nonfatal outcomes. Our findings have clinical relevance considering that these relatively high-risk patients with different stable symptomatology have similar cardiovascular prognosis.

Asymptomatic patients with type 2 diabetes, stable CAD, and documented myocardial ischemia are at high risk for all-cause mortality and major cardiovascular outcomes, and their risk is similar to the risk of patients with angina and angina equivalent symptoms. Our findings suggest that all of these patients should be similarly managed in terms of risk stratification and preventive therapies.