Long-term prognosis of diabetic patients with acute myocardial infarction in the era of acute revascularization

Full details of the HIJAMI registry have been described previously[4]. In brief, HIJAMI is a multicenter, prospective cohort of consecutive patients with AMI who were admitted within 48 hours after the onset of symptoms. Between January 1999 and June 2001, 3021 consecutive patients from 17 participating hospitals in Japan were registered. As HIJAMI was meant for observational purposes, treatment strategies, such as drug therapies and early reperfusion treatment, were used at the discretion of the physician responsible at each hospital. Clinical and angiographic data, including the patients' demographics, coronary risk factors, therapeutic modalities, complications, number of diseased vessels, infarct-related arteries, PCI strategies, laboratory data, and outcomes were prospectively collected using a standardized case report form. AMI was diagnosed on the basis of the following criteria: (1) typical chest pain; (2) a greater than two-fold elevation of cardiac muscle enzyme levels compared to normal levels; and (3) new appearance of abnormal Q waves, an elevation or reduction of ST segments, a typical change in T waves, or new appearance of left bundle branch block[5]. When at least two of the above three criteria were met, AMI was diagnosed. All patients discharged alive from hospital were followed to monitor their prognosis every year (in December) by review of their medical records or telephone survey. The study was conducted according to the Declaration of Helsinki and was approved by the ethical review board of each institution. At the time of enrollment, informed consent for participation in the study was obtained either in writing or orally from all subjects.

Diabetes mellitus was diagnosed according to the following criteria[6]: a prior diagnosis of diabetes in a patient receiving antidiabetic treatment, or a fasting blood glucose level ≥ 126 mg/dl, a blood glucose level higher than 200 mg/dL 2 hours after the 75 g OGTT, or a random blood glucose level ≥ 200 mg/dl on two or more occasions, or these standard values were exceeded once and any of the following criteria were met (i: typical symptoms (dry mouth, polydipsia, polyuria, and weight loss), ii: HbA_1c ≥ 6.5%, or iii: presence of diabetic retinopathy). Hypertension was diagnosed when the patient had a history of hypertension or had a systolic blood pressure ≥ 140 mmHg or a diastolic blood pressure ≥ 90 mmHg during index hospitalization. Hypercholesterolemia was diagnosed when the patient had a history of dyslipidemia or a fasting total cholesterol level ≥ 220 mg/dl or a fasting triglyceride level ≥ 150 mg/dl.

The primary endpoint of the study was the all-cause mortality rate, and the secondary endpoint was the occurrence of major cardiovascular events (including death from cardiovascular causes, recurrent myocardial infarction, unstable angina or heart failure requiring hospitalization, and coronary revascularization).

Two studies were performed. First, the long-term prognosis after myocardial infarction was compared between diabetic and non-diabetic patients. The patients were classified into a diabetic group with AMI and a non-diabetic group with AMI, and the primary and secondary endpoints were compared over the long term. Second, the effects of various therapies on life expectancy after myocardial infarction were assessed. To study the therapeutic modalities that affected life expectancy in diabetic patients with AMI, PCI in the acute phase and the drugs prescribed at discharge (aspirin, beta-blockers, calcium antagonists, nitrates, ACEIs, angiotensin receptor agonists (ARBs), and statins) were evaluated in diabetic patients with AMI. To balance observed covariates between the two groups, the propensity score method was used[7].

This study was funded by the Japan Research Promotion Society for Cardiovascular Diseases, which played no role in the conduct of the study.

Continuous variables showing a normal distribution are expressed as the mean and standard deviation. Data without a normal distribution are expressed as the median and interquartile range (IQR) (25%, 75%). Comparisons between the two groups were done using Student's t-test for data assumed to have a normal distribution, the Mann-Whitney U-test for data without a normal distribution or for ordered, categorical data, and the chi-square test for non-ordered, categorical data. All statistical tests were two-sided, and the level of significance was set at p = 0.05. Since this was a prospective observational study and not a comparative study using randomization, even the diabetic patients with AMI showed some differences in background factors. Therefore, the hazard ratio (HR) and its 95% confidence interval (CI) were calculated using multiple logistic regression after propensity scores were used to match subjects for the 20 factors mentioned below. Diabetic patients with AMI were divided into two groups according to use or non-use of each therapy, and patients with matched propensity scores were extracted to examine associations with life expectancy. For the calculation of propensity scores, age, sex, severity of heart failure at admission, history of hypertension, history of hypercholesterolemia, smoking habits, history of myocardial infarction, PCI at admission, left ventricular ejection fraction, number of stenosed coronary vessels, time from onset of AMI to revascularization, HbA_1c, C-reactive protein, serum creatinine, and medications prescribed at the time of discharge from hospital (aspirin, ACEIs, or angiotensin receptor blockers (ARBs), calcium antagonists, nitrates, and statins) were used, and then propensity scores were calculated using a multiple logistic regression model[7, 8]. The cumulative probabilities of event curves were estimated using the Kaplan-Meier method. To evaluate the effect of therapeutic modalities with respect to subsequent events, conventional Cox proportional hazards models were used. Statistical analysis was performed with SAS software version 9.1 (SAS Institute, Cary, NC, USA).

A total of 3021 consecutive patients with AMI was enrolled between January 1999 and June 2001. There were 2136 men (70.7%) and 885 women (29.3%); their average age was 69 years. Of these patients, 1102 (36.5%) had diabetes mellitus. Their average age was 66 years, 73% of them were men, and they more often had a history of hypertension, hypercholesterolemia, or myocardial infarction than the non-diabetic patients. Also, congestive heart failure was more common in the acute phase, and a higher percentage of patients had a reduced left ventricular ejection fraction and multi-vessel coronary artery disease in the chronic phase. There were no significant differences between the diabetic and non-diabetic patients with respect to CRP and renal function. During the index hospitalization, successful revascularization was achieved in 71.5% of the AMI patients with diabetes and in 77.5% of the AMI patients without diabetes. The overall death rate was 9.3% for the diabetic group and 9.2% for the non-diabetic group during index hospitalization (Table 1).

A total of 2,736 patients was discharged alive and followed regularly. The average duration of follow-up for assessment of the prognosis was 4.2 years (0.0027 to 6.4 years), and the follow-up rate was 96%. In the HIJAMI study, the follow-up assessment was conducted by a telephone survey in 31.8% (with a response rate of 94.3%).

Of the patients who were discharged alive, 1,000 were diabetic (Table 2). Although the long-term survival rate was worse in the diabetic group than in the non-diabetic group, the difference in rates was not significant (HR = 1.20, 95% C. I.:0.97-1.49, p = 0.09, Figure 1). The causes of death in the diabetic group were cardiac (7.2%), including heart failure (3.0%), myocardial infarction (2.1%), and other cardiac deaths (2.1%), cerebrovascular disease (1.2%), malignancy (1.8%), and other conditions (5.2%). In the non-diabetic group, the causes of death were cardiac (4.6%), including heart failure (2.0%), myocardial infarction (0.8%), and other cardiac deaths (1.8%), cerebrovascular disease (0.9%), malignancy (1.8%), and other conditions (4.3%). The diabetic group showed a significantly higher incidence of cardiovascular events than the non-diabetic group (1.40 [1.20-1.64], p < 0.0001, Figure 2).

Aspirin was prescribed at discharge to 1,550 (89%) of the 1,736 non-diabetic AMI patients, and 881 (88%) of the 1,000 diabetic AMI patients. ACEIs and ARBs were also prescribed at relatively high rates (63%) in both groups. However, β-blockers, calcium antagonists, and nitrates were prescribed at higher rates in the diabetic group (Table 2).

To elucidate factors with an effect on the long-term prognosis of the diabetic patients with AMI, propensity scores were calculated. Diabetic patients with AMI were divided into two groups according to use or non-use of each therapy, and patients with matched propensity scores were extracted to examine associations with life expectancy. Use of an ACEI or ARB at discharge was significantly related to a better long-term prognosis for diabetic patients with AMI (0.53 [0.36-0.76], p = 0.001). Similarly, use of aspirin at discharge was significantly associated with a good long-term prognosis (0.27 [0.15-0.50], p < 0.0001). In contrast, use of beta-blockers (0.97 [0.67-1.42], p = 0.889), calcium antagonists (1.27 [0.88-1.85], p = 0.205), and nitrates (0.96 [0.61-1.53], p = 0.874) was not associated with prognosis. Doses of statins based on the standard, approved Japanese doses were associated with improved prognosis in AMI patients with diabetes, but not significantly (0.67 [0.39-1.06], p = 0.11). On the other hand, a significant association was found between successful PCI during the acute phase and a favorable long-term prognosis (0.52 [0.36-0.74], p = 0.02; Figure 3).

There are certain limitations in this retrospective analysis of data from a prospective cohort study. The major limitation of the current study is that it is based on a data-driven post hoc analysis of a cohort study. Therefore, therapeutic modalities were not allocated randomly. Although the analysis of the effects of each therapeutic modality on adverse event rates was performed with the powerful propensity score-matching technique, this control was limited to variables for which data were available. In the present study, the percentage of patients who were treated with ACEIs or ARBs was higher than the percentage who were treated with beta-blockers or statins. For this reason, no significant differences were obtained by comparison, which is inevitable in an observational study. Furthermore, this study is not a clinical trial that targeted lipid-lowering. Consequently, participants were treated with standard, Japanese-approved doses of statins. Although statin therapy showed a tendency to improve the prognosis of diabetic patients, a significant improvement was not shown in the present study. We investigated the prognosis of diabetic patients with AMI but did not assess other coronary risk factors that overlap with the components of the metabolic syndrome in this analysis. As the HIJAMI was a database constructed from cardiologists' perspective, it includes no details on diabetic complications, diabetes duration, and type of diabetes. As the epidemiology of cardiovascular disease in Japan is substantially different from that of non-Japanese, one cannot simply assume that the results of the present study can be extrapolated to non-Japanese populations. In the future, randomized, controlled studies should be conducted to elucidate the association of each risk factor with the prognosis of AMI patients with diabetes.