Prognostic Value of NH₂-Terminal Pro B-Type Natriuretic Peptide in Patients With Diabetes and Stable Coronary Heart Disease

A total of 1,034 consenting patients with angina pectoris, or evidence of ischemia on exercise electrocardiography or myocardial radionuclide imaging, were included in this analysis. A detailed description of the study population has been reported previously (3). The Danish health authorities and regional ethics committee approved the study, and all participating subjects provided informed consent.

At baseline, selective coronary angiography and left ventriculography were performed in all patients. A thorough medical history of previous cardiovascular disease, diabetes, and smoking habits was recorded. Diabetes was defined as a history of diabetes, on antidiabetic treatment, or fasting plasma glucose (FPG) ≥7 mmol/l. Fasting blood samples were drawn for measurement of plasma glucose, lipids, creatinine, and NT-proBNP (Elecsys proBNP; Roche Diagnostics, Mannheim, Germany). Information on vital status was obtained from the Danish Central Person Registry by means of a computerized search performed on 1 August 2001. No patient was lost to follow-up.

Baseline characteristics, grouped according to NT-proBNP concentration above or below the median and diabetes status, were compared using the χ² test for discrete variables and the Wilcoxon or Kruskal-Wallis rank-sum test for continuous variables. Relative risks and 95% CIs were calculated as hazard ratios (HRs) derived from the Cox proportional-hazards regression analysis. The following covariates were considered potential confounders: age, sex, family history of ischemic heart disease, previous myocardial infarction, angina, hypertension, smoking, BMI, estimated glomerular filtration rate, plasma lipids, left-ventricular ejection fraction (LVEF), and severity of coronary disease at angiography. There was no interaction between NT-proBNP and diabetes status. Version 8.2 of the Statistical Analysis System (SAS Institute, Cary, NC) was used for all analyses.

A total of 197 patients (19%) had diabetes. Median (interquartile range) NT-proBNP level was increased in patients with diabetes (230 ng/l [78–632]) compared with patients without diabetes (161 ng/l [61–423]) (P = 0.015). Patients with supramedian NT-proBNP (>169 ng/l) were older, had a higher frequency of prior myocardial infarction, higher left-ventricular end-diastolic pressure, and lower LVEF and glomerular filtration rate, irrespective of diabetes status. Diabetic patients had higher frequency of history of hypertension, BMI, triglycerides, FPG, and insulin, and lower HDL cholesterol, irrespective of NT-proBNP level. Patients with both supramedian NT-proBNP and diabetes had more three-vessel disease and disease of the left main coronary artery. After a median follow up of 9.2 years, 288 (28%) patients had died, including 206 (25%) nondiabetic patients and 82 (43%) diabetic patients. Kaplan-Meier estimates of survival were made after dividing the subjects into four groups, as shown in Fig. 1. By multivariable Cox regression analysis, the HR was 1.8 (95% CI 1.1–3.0, P = 0.02) for the group of patients with low NT-proBNP levels and diabetes, 1.6 (1.2–2.3, P = 0.004) for the group of patients with high NT-proBNP without diabetes, and 2.8 (1.9–4.2, P < 0.0001) for the group of patients with high NT-proBNP and diabetes, as compared with patients with low NT-proBNP levels without diabetes. Other covariates independently associated with mortality were increasing age, smoking, and decreasing LVEF.

This study demonstrates that the combination of NT-proBNP and diabetes in patients with stable coronary heart disease provides independent prognostic information on all-cause mortality. Our results extend currently available information by showing that patients with diabetes and low NT-proBNP are at similar risk of death as patients with high NT-proBNP and no diabetes, and the combination of the two nearly tripled the risk.

Results from the EUROASPIRE surveys showed that patients with diabetes are currently not treated according to present guidelines (7). Identification of patients with diabetes at particularly high risk is clinically important because therapeutic agents to reduce morbidity and mortality in these patients are readily available. Previous studies investigating natriuretic peptide levels in patients with diabetes have shown conflicting results (5,8–15). Data from the Breathing Not Properly Multinational Study, investigating patients with acute dyspnea, failed to show any difference in levels of BNP between patients with and without diabetes, regardless of the presence of heart failure (16). However, this study may have excluded patients with more severe diabetes, as patients with advanced renal dysfunction were excluded.

Several different explanations for our findings are possible. Patients with diabetes may have a higher prevalence of diastolic dysfunction or have more peripheral and distal atherosclerotic changes in the coronary tree, which may be clinically important. Adjusting for these confounders in the multivariable models did not change the results. Nevertheless, the possibility remains that the increased risk of death in patients with high NT-proBNP and diabetes may be due to higher left-ventricular mass, left-ventricular hypertrophy, and fibrosis, as suggested by other investigators (17). The combination of diabetes and high NT-proBNP did not result in a superadditive effect on risk, even though patients with this combination may have had a longer duration of diabetes and be in a more dysmetabolic, insulin-resistant state. However, regardless of diabetes status, there was no difference in levels of FPG or insulin between patients with high versus low levels of NT-proBNP.

We conclude that NT-proBNP and diabetes are complementary independent risk factors of long-term mortality in patients with stable coronary disease. Combined assessment of these two readily obtainable risk factors provides risk stratification superior to that provided by either alone.