The Effect of Diabetes on B-Type Natriuretic Peptide Concentrations in Patients With Acute Dyspnea: An analysis from the Breathing Not Properly Multinational Study

The Breathing Not Properly Multicenter Study was an international, seven-center, prospective study (five U.S. centers and two European centers). The study design and main results of the Breathing Not Properly Multinational Study have been published elsewhere (10). Study investigators and centers are listed in the appendix. The study was conducted from April 1999 to December 2000. The study protocol was approved by the institutional review boards of all study centers, and written informed consent was obtained from all participants.

A total of 1,666 patients presenting to the study centers with a primary complaint of dyspnea were screened. Eighty patients were excluded from the study based on the protocol exclusion criteria, which included the presence of advanced renal failure (on dialysis or estimated glomerular filtration rate of <15 ml · min⁻¹ · 1.73 m⁻²) (11), acute myocardial infarction, or overt cause of dyspnea including chest wall trauma or penetrating lung injury. A final total of 1,586 participants were enrolled in the study. There were 367 patients (23.1%) who had diabetes listed in the final discharge diagnosis and 1,219 (76.9%) who did not.

Baseline demographics, clinical history, and objective assessment of clinical signs were gathered by trained research personnel in the emergency department, including age, sex, race, height, and weight (necessary for calculation of BMI). Findings from the electrocardiogram, chest X-ray, and blood tests, including serum creatinine, were recorded in a structured checklist completed by the emergency department attending physician. From the database of 1,586 patients, subgroups were prepared based on the presence of diabetes and heart failure status (active versus having history of dyspnea, but not due to heart failure). Because BNP concentrations in healthy individuals are dependent on age, sex, BMI, renal function (12), and New York Heart Association (NYHA) classification (for heart failure patients), a subset of data were prepared whereby these parameters were matched between the no heart failure, heart failure, and history of heart failure groups (but not active heart failure). The database was sorted according to BMI and sex, given that these parameters were the most different between the diabetic and nondiabetic groups. Lean nondiabetic female patients were consecutively removed from the nondiabetic groups until the BMI and sex matched the diabetic group. Retrospectively, no adjustments in the database were necessary for the estimated creatinine clearance. After all adjustments, the final subgroupings were: group 1 (n = 324) had neither diabetes nor heart failure, group 2 (n = 107) had diabetes and no heart failure, group 3 (n = 247) had heart failure but not diabetes, group 4 (n = 183) had both diabetes and heart failure, group 5 had a history of heart failure but current dyspnea not due to heart failure and no diabetes (n = 41), and group 6 (n = 21) had a history of heart failure and diabetes.

During initial evaluations, 5 ml of blood was collected into tubes containing K₃EDTA, and the BNP concentration was measured using the first generation Triage B-Type Natriuretic Peptide test (Biosite, San Diego, CA). Precision, analytical sensitivity, and stability characteristics of the system have been previously described (13). The measurable range of the BNP assay was 5.0–1,300 pg/ml. (After completion of this study, the manufacturer has released a second-generation BNP assay whereby the assay range has been extended to 5,000 pg/ml.) The test was run in a concealed fashion, with the results kept in separated data binders only linked by a separate study code, hence blinding both emergency department physicians and adjudicating cardiologists.

Approximately 30 days after the emergency department visit, the case report form (excluding the emergency department physician estimate of probability of heart failure), electrocardiogram, chest X-ray, echocardiogram, and all other clinical tests, consultations, and chart information were reviewed by two independent cardiologists. In addition, information was used from the case report form to calculate the Framingham (requiring two major or one major and two minor criteria for heart failure) and National Health and Nutrition Examination Survey (requiring three or more points for heart failure) scores for heart failure. After reviewing all information, if agreement was achieved, then the case was categorized as one of the following: 1) dyspnea due to heart failure, 2) history of heart failure but dyspnea due to noncardiac cause, or 3) dyspnea due to noncardiac cause. In the event of disagreement, the case was adjudicated by the study end points committee.

Baseline characteristics were reported in counts and proportions or means ± SDs as appropriate. Univariate comparisons of the nonparametric data were made with the Wilcoxon test using the median and 95% CIs. A P value of <0.05 was considered significant. Separate receiver-operating characteristic (ROC) curves were generated for BNP and the final diagnosis of heart failure in diabetic versus all subjects. The area under the ROC curve (AUC) and 95% CIs were reported. Multiple conditional logistic regression was used to evaluate the independent relationships between clinical factors including BNP and presence or absence of diabetes on the outcome of final adjudicated diagnosis of heart failure. Statistical calculations were performed using MedCalc for Windows version 7.3.0.1 (Berkeley, CA).

Table 1 illustrates the baseline characteristics of the entire 1,586-patient dataset, comparing the 367 patients with diabetes with the 1,219 patients without diabetes. Diabetic patients were slightly older (65.6 vs. 63.5 years), had a more frequent history of heart disease (stable angina, prior myocardial infarction, prior coronary artery bypass graft surgery, heart failure, higher diastolic blood pressure, elevated jugular venous pressure, orthopnea, and paroxysmal nocturnal dyspnea), and therefore a higher incidence of cardiac medications (ACE inhibitors, β-blockers, calcium channel blockers, antiarrhythmics, diuretics, and digoxin). They also had a higher mean BMI (31.6 vs. 27.9 m/kg²). Estimated creatinine clearance was lower for the diabetic group (65.8 vs. 63.0 ml · min⁻¹ · 1.73m⁻²), but results were not statistically different.

Table 2 illustrates the results of the subgroup analysis that were matched for median age, sex, BMI, estimated creatinine clearance, and NYHA classification (for heart failure patients). With regard to average BNP concentrations, there was no significant difference between diabetic and nondiabetic subjects among the pairs of groups, no heart failure patients (groups 1 and 2), heart failure patients (groups 3 and 4), and patients with a history of heart failure but presenting with dyspnea not due to heart failure (groups 5 and 6) (Fig. 1). Figure 2 illustrates the ROC curve analysis for the subset groups, also showing no significant difference in the AUC (0.87–0.88). When considering median BNP concentrations, the heart failure group with diabetes had slightly lower BNP concentrations than the heart failure group without diabetes (P < 0.05).

A previous publication (11) from the Breathing Not Properly Multinational Study had tabulated the variables that were most useful as an independent predictor of heart failure using multiple logistic regression. Table 3 shows that the presence of diabetes is an independent factor for heart failure (odds ratio [OR] 1.61, 95% CI 1.1–2.3). The OR for the subgroup matched for age, sex, BMI, and estimated glomerular filtration rate was unchanged. The log BNP concentration was the most powerful predictor of heart failure (11.32, 8.24–15.02).

The identification of asymptomatic diabetic patients with left ventricular dysfunction can be potentially significant because therapeutic agents such as ACE inhibitors or angiotensin II receptor antagonists may be useful to reduce morbidity and mortality. In a subgroup consisting of diabetic patients in the Losartan Intervention For Endpoint (LIFE) trial, there was a significant reduction in composite cardiovascular morbidity and mortality (OR 0.87, 95% CI 0.77–0.98) for Losartan compared to atenolol in patients with hypertension and left ventricular hypertrophy (14). These findings have led some to question if tests should be implemented to screen asymptomatic subjects for heart failure (15). As such, there have been several studies that have examined the potential use of BNP as such a test. The North Glasgow MONICA (Monitoring Trends and Determinants of Cardiovascular Disease) study produced an AUC of 0.88, of patients with left ventricular systolic dysfunction, when BNP was targeted at high-risk subjects aged >55 years with a history of prior ischemic heart disease (16). These AUCs were within the range of other widely used and accepted laboratory screening tests, such as prostate-specific antigen, Papanicolaou smear, and mammography (17). In contrast to the MONICA study, results from the Framingham Heart Study for BNP screening produced a different conclusion (18). The ROC AUC from blood tested for BNP on 3,177 subjects was ≤0.75, leading these investigators to suggest that mass screening for left ventricular dysfunction was not appropriate, even when high-risk subgroups were targeted (i.e., subjects who were aged ≥60 years, hypertensive, diabetic, and with known cardiovascular disease).

In this study, we attempted to determine whether diabetic patients symptomatic for heart disease produced higher BNP concentrations than nondiabetic patients, with and without confirmed heart failure. Our data showed that the mean BNP concentrations were identical in diabetic versus nondiabetic subjects who were matched in parameters known to affect BNP concentrations, i.e., age, sex, BMI, and renal function. Although diabetes was determined to be an independent predictor of heart failure (Table 3), the OR was very modest.

The data from the Breathing Not Properly Multinational Trial would suggest that diabetes status is not a confounding variable to be considered when interpreting BNP concentrations in patients who present acutely with dyspnea. Our results are in contrast to other studies where diabetic patients had higher BNP concentrations than nondiabetic ones. Given that symptomatic diabetic patients did not produce higher BNP results than matched nondiabetic subjects in our study, one might be tempted to suggest that screening asymptomatic diabetic patients would not be useful. However, Epshteyn et al. (9) screened diabetic patients with and without a clinical indication for echocardiography (CIE). Patients with CIE and subsequent abnormal left ventricular function (n = 112) had a mean BNP concentration of 435 ± 41 pg/ml compared with those with no CIE but who had abnormal left ventricular function on echo (161 ± 40 pg/ml, n = 32). Twenty-one of 32 patients with no CIE but with abnormal left ventricular function had diastolic dysfunction (BNP 190 ± 60 pg/ml). A ROC curve revealed that the AUC was 0.91 for patients with CIE and 0.81 for patients without CIE (P < 0.001). In those with no heart failure symptoms, BNP levels showed a high negative predictive value (91% for BNP values <39 pg/ml), whereas in patients who had a CIE, BNP levels showed a high positive predictive value for the detection of left ventricular dysfunction (96% with BNP levels >90 pg/ml).

In the present study, our excluded subjects included those with advanced renal dysfunction (estimated glomerular filtration rate <15 ml · min⁻¹ · 1.73 m⁻²). As renal failure is a major complication of diabetes, it may be possible that the patients tested in this study had milder forms of diabetes and did not have the initial stages of heart failure. In the study of Yano et al. (19), patients with diabetes and microalbuminuria had higher BNP concentrations than diabetic patients who had normal microalbuminuria. Nagai et al. (20) showed a difference in diabetic patients with nephropathy and/or retinopathy, but no difference in diabetic patients without albuminuria. Siebenhofer et al. (21) found that patients with mild or significant diabetic nephropathy had higher NH₂-terminal proBNP, the inactive circulating fragment of the BNP prohormone, than diabetic patients without albuminuria. In a commentary, Chan et al. (22) suggest a relationship between microangiopathy, as indicated by microalbuminuria or retinopathy, and BNP concentrations. Their notion was suggested following the report of Poirier et al. (23), who found a trend toward a higher incidence of microalbuminuric patients with abnormal left ventricular diastolic dysfunction (54%) than diabetic patients who had normal diastolic function, although their results had not reached statistical significance. Both BNP and atrial natriuretic peptide have been shown to increase tubular protein excretion of albumin and α1-microglobulin (24). This study would support the relationship of BNP and microalbuminuria. Unfortunately, in the Breathing Not Properly Multinational Study, fasting blood glucose, HbA_1c, and urine microalbumin levels were not measured in the database, thus no estimate of diabetes disease severity could be made.

In the Breathing Not Properly Trial, diabetic patients without heart failure had the same concentration of BNP than non–heart failure and nondiabetic patients who were matched by age, sex, BMI, and renal function. There was also no difference in BNP concentrations among patients with heart failure who were also matched for these parameters and for heart failure disease severity (NYHA classification). The presence of diabetes was an independent predictor of heart failure, but was not as powerful as the log of BNP concentration (OR 1.61 vs. 11.31, respectively). However, no conclusions can be rendered as to the role of BNP testing for screening asymptomatic diabetic patients for left ventricular dysfunction because the degree of disease severity among the diabetic patients could not be assessed.

Alan S. Maisel, MD.

Peter A. McCullough, MD, MPH.

Alan S. Maisel (Chair), Peter A. McCullough, Richard M. Nowak, James McCord, Judd E. Hollander, Philippe Duc, Torbjørn Omland, Alan B. Storrow, William T. Abraham, and Alan H.B. Wu.

James McCord (Chair), Alan S. Maisel, Howard C. Herrmann, Philippe Duc, Torbjørn Omland, William T. Abraham, and Alberto Perez.

Peter A. McCullough (Chair), Alan S. Maisel, Richard M. Nowak, and William T. Abraham.

Alan S. Maisel, Peter A. McCullough, Paul Clopton, Richard M. Nowak, James McCord, Judd E. Hollander, Philippe Duc, Torbjørn Omland, Alan B. Storrow, William T. Abraham, and Alan H.B. Wu.

Alan S. Maisel, Peter A. McCullough, Radmila Kazanegra, Padma Krishnaswamy, Richard M. Nowak, James McCord, Judd E. Hollander, Howard C. Herrmann, Philippe Duc, Philippe G. Steg, Torbjørn Omland, Westheim, Cathrine Wold Knudsen, Alan B. Storrow, William T. Abraham, Sumant Lamba, Alan H.B. Wu, and Alberto Perez.

Alan H.B. Wu, Alan S. Maisel, Peter A. McCullough, Paul Clopton, Richard M. Nowak, James McCord, Torbjørn Omland, Judd E. Hollander, William T. Abraham, and Philippe Duc.

Alan H.B. Wu, Alan S. Maisel, Peter A. McCullough, Richard M. Nowak, James McCord, Torbjørn Omland, Judd E. Hollander, and William T. Abraham.

Alan H.B. Wu, Alan S. Maisel, Peter A. McCullough, Richard M. Nowak, James McCord, Torbjørn Omland, Judd E. Hollander, William T. Abraham, Philippe Duc, Philippe G. Steg, Alan H.B. Wu, Howard C. Herrmann, Radmila Kazanegra, Alberto Perez, Alan B. Storrow, Paul Clopton, Padma Krishnaswamy, Sumant Lamba, Arne Westheim, and Cathrine Wold Knudsen.

Alan H.B. Wu, Peter A. McCullough, Paul Clopton, and William T. Abraham.

Alan S. Maisel, Peter A. McCullough, Richard M. Nowak, James McCord, Philippe Duc, Judd E. Hollander, Alan H.B. Wu, Alan B. Storrow, and Torbjørn Omland.

Alan S. Maisel, Radmila Kazanegra, Padma Krishnaswamy, Peter A. McCullough, Paul Clopton, Richard M. Nowak, James McCord, Judd E. Hollander, Philippe Duc, Torbjørn Omland, Alan B. Storrow, William T. Abraham, Alan H.B. Wu, and Alberto Perez.

Alan S. Maisel, Peter A. McCullough, James McCord, Richard M. Nowak, Judd E. Hollander, Alan B. Storrow, William T. Abraham, Alan H.B. Wu, Torbjørn Omland, Arne Westheim, Philippe Duc, and Philippe G. Steg.

University of California, San Diego, Veteran’s Affairs Medical Center, San Diego, CA: Study Principal Investigator, Alan S. Maisel, MD; Site Principal Investigators: Radmila Kazanegra, MD; Padma Krishnaswamy, MD; Patricia Hlavin, MD; Leslie A. Lenert, MD; Padma Krishnaswamy, MD; Biostatistician: Paul Clopton, MS. Henry Ford Hospital, Detroit, MI: Site Principal Investigators: Richard M. Nowak, MD, MBA; James McCord, MD; Study Coordinators: Michele Whitican, RN; James Babiarz, RN. University of Pennsylvania, Philadelphia, PA: Site Principal Investigators: Judd E. Hollander, MD; Howard C. Herrmann, MD; Evan Loh, MD; Study Coordinator: Frank D. Sites, RN, BSN. Hospital Bichat, Paris, France: Site Principal Investigators: Philippe Duc, MD;Philippe G. Steg, MD; Co-Investigators:Marie Claude Aumont, MD; Valerie Beaumesnil, MD; Lamia Hafi, MD; Armelle Desplanques, MD; Joelle Benessiano, MD. Ullevål University Hospital, Oslo, Norway: Site Principal Investigators: Arne Westheim, MD, PhD; Torbjørn Omland, MD, PhD, MPH; Cathrine Wold Knudsen, MD; Co-Investigators: Alexandra Finsen, MD; Jon Sigurd Riis, MD; Tor Ole Klemsdal, MD, PhD. University of Cincinnati College of Medicine, Cincinnati, OH: Site Principal Investigators: Alan B. Storrow, MD. University of Kentucky College of Medicine, Lexington, KY: Co-Investigators: Sumant Lamba, MD; William T. Abraham, MD. Hartford Hospital, Hartford, CT: Site Principal Investigators: Alan H.B. Wu, PhD; Alberto Perez, MD; Richard Kamin, MD. University of Missouri-Kansas City School of Medicine, Truman Medical Center, Kansas City, MO, and William Beaumont Hospital, Royal Oak, MI: Study Co-Principal Investigator: Peter A. McCullough, MD, MPH.

Triage BNP devices and meters and some financial support were provided by Biosite (San Diego, CA).

The authors are indebted to the efforts of the emergency department staff at the following Breathing Not Properly Multinational Study Centers: San Diego Veteran’s Affairs Medical Center, San Diego, CA; Henry Ford Hospital, Detroit, MI; Hospital of the University of Pennsylvania, Philadelphia, PA; Hospital Bichat, Paris, France; Ullevål University Hospital, Oslo, Norway; University of Cincinnati Medical Center, Cincinnati, OH; and Hartford Hospital, Hartford, CT.