Utility of B-Type Natriuretic Peptide (BNP) as a Screen for Left Ventricular Dysfunction in Patients With Diabetes

This study was approved by the University of California, San Diego Institutional Review Board. The participants included 486 diabetic patients (88% type 2 diabetes, 12% type 1 diabetes) at the San Diego Veteran’s Healthcare System studied between June 1999 and February 2001. Patients were recruited from the diabetes clinic (n = 357) and the echocardiography suite (while waiting for their scheduled appointment, n = 129). Two-hundred sixty-three patients who underwent echocardiography comprised the database. Patients were divided into two groups depending on whether there was a clinical indication for echocardiography (CIE) (Fig. 1). Patients with a CIE met at least one of the following criteria: symptoms or signs of CHF (shortness of breath, lower extremity edema), history of CHF, or suspicion of left ventricular dysfunction by the physician. Of the 172 patients with CIE, 121 were recruited for the study at the time they were receiving their echo and 51 were in the diabetes clinic and had already either had an echo or were scheduled to receive one within 6 months of BNP draw. Eight patients from the echocardiography suite, whose referral was to assess valve disease, determine whether a vegetation was present, or rule out a cardiac cause of stroke, were excluded. Of the remaining 306 clinic patients with no CIE (no-CIE), a random sampling of ∼33% (91 patients) received echocardiography within 6 months of BNP draw. Patient referrals were made either by clinic physicians or attending or nurse practitioners.

Two-dimensional, M-mode, spectral, and color flow Doppler echocardiograms were obtained with commercially available instruments operating at 2.0–3.5 MHz. Two-dimensional imaging examinations were performed in the standard fashion in parasternal long- and short-axis views and apical four and two chamber views (12). Pulsed Doppler spectral recordings were obtained from a 4 × 4-mm sample volume placed at the tips of the mitral leaflets and in the pulmonary vein, adjusted to yield the maximal amplitude velocity signals. All data were hard copied to a 0.5-inch VHS videotape for subsequent playback, analysis, and measurement.

Two-dimensional echocardiograms were subjected to careful visual analysis to detect regional contractile abnormalities. Left ventricular systolic and diastolic volumes and ejection fractions were derived from biplane apical (two and four chamber) views using a modified Simpson’s rule algorithm (13). The transmitral pulsed Doppler velocity recordings from three consecutive cardiac cycles were used to derive measurements as follows: E and A velocities as the peak values reached in the early diastole and following atrial contraction, respectively, and deceleration time as the interval from the E wave to the decline of the velocity to baseline. In addition, pulmonary venous systolic and diastolic flow velocities were obtained as the maximal values reached during the respective phase of the cardiac cycle, and the pulmonary venous “A” reversal as the maximal velocity of retrograde flow into the vein following the P wave of the ECG. Finally, the left ventricular isovolumetric relaxation time (IVRT) was obtained from the apical five-chamber view with a continuous wave cursor or, if possible, a pulsed Doppler sample volume positioned to straddle the left ventricular outflow tract and mitral orifice so as to obtain signals from aortic valve closure, the termination of ejection and mitral valve opening, or the onset of transmitral flow. IVRT was taken as the time in milliseconds from the end of ejection to the onset of left ventricular filling. Experienced cardiologists who were blinded to the BNP levels interpreted all echocardiograms.

Normal ventricular function was defined as ejection fraction >50%, normal left ventricular end diastolic (3.5-5.7 cm) and end systolic dimension (2.5–3.6 cm), and no major wall motion abnormalities.

Systolic dysfunction was defined as ejection fraction <50% or global hypokinesia or discrete wall motion abnormalities.

Diastolic dysfunction was defined as impaired relaxation and restrictive and pseudonormal pattern, based on the following definition: 1) impaired relaxation: in patients <55 years of age E/A <1, deceleration time (DT) >240 ms. In patients >55 years of age E/A <0.8. Additional confirming evidence was isovolumic relaxation time (IVRT) >90 ms. 2) Restrictive pattern: E/A >1.5 and DT <150 ms. Additional confirming evidence was pulmonary vein diastolic flow more than pulmonary vein systolic flow and pulmonary A-wave duration more than mitral A-wave duration. Pulmonary diastolic flow reversal, IVRT <70 ms. 3) Pseudonormal: E/A >1, DT >240 ms. Confirmation by valsava when possible.

Systolic and diastolic dysfunction was defined as ejection fraction <50% with global hypokinesia or discrete wall motion abnormalities along with diastolic dysfunction as described above.

History of CHF was obtained from patients’ medical records with one or more of the following objective findings: previous abnormal left ventricular function by echocardiography or nuclear scan, hospitalization for CHF, diagnosis of CHF made by cardiologist, or regular follow-up in the outpatient cardiomyopathy clinic. Patients with no history of CHF were additionally characterized by either lack of any prior study of left ventricular function or prior echocardiography with negative findings. The evidence of symptoms and suspicion of left ventricular dysfunction was obtained from the echocardiography consult form, where the provider requesting an echo detailed reasons for the echo request.

During initial evaluations, a small sample (5 ml) was collected into tubes containing potassium EDTA (1 mg/ml blood). BNP was measured using the Triage B- Type Natriuretic Peptide test (Biosite, San Diego, CA). The Triage BNP test is a fluorescence immunoassay for the quantitative determination of BNP in whole-blood and plasma specimens and has been recently characterized with regards to precision, analytical sensitivity, stability, and utility (14–16). BNP values were determined on site utilizing the point of care method with either whole-blood or plasma samples.

Group comparisons of BNP values were made using t tests for independent samples and ANOVA with post hoc Tukey’s tests where indicated. In all cases, these were computed with raw BNP values and repeated with log-transformed BNP values since the BNP distribution was positively skewed. Both versions yielded the same conclusions. Results are expressed as mean ± SE for raw values. Sensitivity, specificity, and accuracy were computed for BNP using a selection of possible cut points. The diagnostic utility of BNP alone was compared with the echocardiographic probability of left ventricular dysfunction using receiver-operating characteristic (ROC) curves. The results are expressed in terms of the area under the curve and the 95% CI for this area. Logistic regression was used in a multivariate approach for evaluating the ability of BNP to identify left ventricular dysfunction over and above the information provided by other indicators.

The characteristics of all patients who underwent echocardiography are shown in Table 1. Fifty-five percent of patients had abnormal left ventricular function by echo. Nearly all patients had risk factors for heart disease, including 85% with hypertension and 48% with coronary artery disease.

The 91 patients with no-CIE who were randomly selected to receive echocardiography had similar BNP levels (83 ± 16 pg/ml) to the 215 patients with no-CIE who did not receive echocardiography (BNP 63 ± 10, P = 0.10). Table 2 shows that patients with CIE had higher rates of hypertension, coronary artery disease, myocardial infarction, and higher BNP levels (301 ± 31 pg/ml) than those with no-CIE (P < 0.001).

Figure 2 presents mean BNP values for those patients with normal and abnormal left ventricular function. In patients with normal left ventricular function, BNP levels were low, regardless of whether a CIE was present (51 ± 17 and 41 ± 7 pg/ml, respectively). Patients who had a CIE and were found to have subsequent abnormal left ventricular function (n = 112) had a mean BNP concentration of 435 ± 41 pg/ml. Patients who had no-CIE, yet who were found to have abnormal left ventricular function on echo (n = 32), had BNP levels of 161 ± 40 pg/ml. Differences between patients with abnormal and normal left ventricular function in both groups were significant (P < 0.001 in both cases).

With regards to treatment modalities, there were no differences in raw and log BNP levels to those taking or not taking ACE inhibitors or calcium channel blockers. However, patients taking diuretics, digitalis, and β-blockers had higher BNP levels than those not taking medication (diuretics vs. no diuretics: 171 ± 26 vs. 55 ± 8 pg/ml, P < 0.001; digitalis vs. no digitalis: 301 ± 73 vs. 70 ± 8 pg/ml, P < 0.001; β-blockers vs. no β-blockers: 158 ± 27 vs. 71 ± 9 pg/ml, P = 0.001).

Figure 3 presents BNP mean values for those patients with abnormal left ventricular function divided into purely systolic, purely diastolic, and the combination of systolic and diastolic. BNP levels for all abnormal left ventricular function groups are significantly higher than for all the normal left ventricular function groups. Of the 32 patients who had no-CIE, yet who had abnormal left ventricular function, the majority (n = 21) had diastolic dysfunction, with a mean BNP value of 190 ± 60, significantly different from those with normal left ventricular function (41 ± 7 pg/ml, P < 0.001).

The ability of BNP to differentiate patients with normal versus abnormal left ventricular function was assessed with an ROC curve analysis (Fig. 4). The area under the ROC curve, using BNP to separate normal versus abnormal left ventricular function in patients with a CIE, was 0.91 (95% CI 0.86–0.96, P < 0.001). A BNP cutoff value of 79 pg/ml had a sensitivity of 86%, a specificity of 92%, and an accuracy of 88%. Levels of ≥79 pg/ml had a positive predictive value of 95% (Table 3).

The area under the ROC curve using BNP to separate normal versus abnormal left ventricular function in no-CIE patients was less than that for those with an indication, but still highly significant (AUC = 0.81, 95% CI 0.71–0.91, P < 0.001). In addition, a BNP cutoff value of 60 pg/ml had a sensitivity of 84%, a specificity of 76%, and an overall accuracy of 79%. Levels ≤60 pg/ml had a negative predictive value of 90% (Table 3).

Logistic regression was used in a multivariate approach for evaluating the ability of BNP to identify left ventricular dysfunction over and above the information provided by other indicators (Table 4). Historical variables and BNP were used as independent variables in this analysis and the presence of left ventricular dysfunction was the dependent variable. History of CHF, coronary artery bypass graft, and atrial fibrillation each contributed to the identification of left ventricular dysfunction, with BNP being the strongest predictor (P < 0.001).

The Framingham Study (17) firmly established the epidemiological link between diabetes and heart failure. The poor prognosis of heart failure in diabetic patients was demonstrated in a report from SOLVD (Studies of Left Ventricular Dysfunction) (18) where, compared with nondiabetic subjects, diabetic subjects were more likely to be admitted for heart failure and had mortality related to pump failure. In the presence of coronary disease, diabetes is an independent factor for progression of heart failure (19).

Early detection of left ventricular dysfunction enables administration of treatment that can improve survival and increased well-being (20,21). Treatment can delay progression from the precursor state, asymptomatic left ventricular dysfunction (22). However, ventricular dysfunction may be difficult to diagnose because patients may be asymptomatic or have nonspecific symptoms and abnormal findings on physical examination are often absent (23,24). Echocardiography, one of the fastest growing procedures in cardiology (25), is not feasible as a widespread screening test in the outpatient setting due to both its limited availability and expense.

BNP is a 32-amino acid polypeptide containing a 17-amino acid ring structure common to all natriuretic peptides (26). It is found mainly in the cardiac ventricles, and its release appears to be directly proportional to ventricular volume expansion and pressure overload (7,27,28). BNP is an independent predictor of high left ventricular end-diastolic pressure (7) and correlates to both NYHA and prognosis (29). A rapid assay for BNP has been approved for the diagnosis of CHF, and its use in this capacity has already been delineated in recently published heart failure guidelines (30).

To the limited extent that BNP has been used as a screening procedure in primary care settings, it has been shown to be a useful addition to the evaluation of possible CHF (31–33). In a study of 1,252 randomly selected subjects, a plasma BNP concentration ≥17.9 pg/ml had a sensitivity and specificity for left ventricular dysfunction of 77 and 87% for all participants and 92 and 72% for those ≥55 years of age, respectively (31). In a study of 122 consecutive patients with suspected new heart failure referred by general practitioners to a rapid-access heart failure clinic for diagnostic confirmation, a BNP level of 76 pg/ml, chosen for its negative predictive value of 98% for heart failure and similar to the cutoff in the present study, had a sensitivity of 97%, a specificity of 84%, and a positive predictive value of 70% (33). Maisel et al. (9) characterized patients who had both echocardiography and BNP levels and found that of the patients with no documented history of CHF and no past determination of left ventricular function, 51% had abnormal echocardiographic findings. In this group, BNP levels were significantly higher (328 ± 29 pg/ml) than the 49% of patients with no history of CHF and a normal echocardiogram (30 ± 3, P < 0.001). In patients with a known history of CHF, with preciously documented left ventricular dysfunction, all had abnormal findings (n = 102) with elevated BNP levels (545 ± 45 pg/ml).

The ability of BNP to detect abnormal cardiac function (systolic or diastolic) has been assessed with ROC analysis (10). The area under the ROC curve using BNP to detect any abnormal echocardiographic finding was 0.952. A BNP value of 75 pg/ml had a sensitivity of 85%, a specificity of 97%, and an accuracy of 90% for predicting left ventricular dysfunction. In these as well as previous studies, it appeared that patients with diabetes often had high BNP levels, even when there was no previously documented history of heart failure (6,7,9–11). The current study specifically examined two groups of diabetic patients: those with high probability for left ventricular dysfunction (those with a CIE) and those with low probability for left ventricular dysfunction (no-CIE). Since BNP levels rise in proportion to the severity of cardiac dysfunction (34), it is not surprising that the group with CIE had higher BNP levels than those with no-CIE. The accuracy of BNP in detecting left ventricular dysfunction in diabetic patients (ROC curve analysis) is similar to that of the prostate-specific antigen for prostate cancer detection, Papanicolaou smears for cervical cancer, and mammography for the detection of breast cancer (35–37). In addition, for those with no symptoms of CHF, BNP levels showed a high negative predictive value in ruling out left ventricular dysfunction (91% for BNP values <39 pg/ml), while in those 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).

Despite the association between diabetes and increased cardiovascular morbidity and mortality rates, the prevalence of myocardial systolic and diastolic functional abnormalities in asymptomatic diabetic patients is not well defined. A recent echocardiographic study of 66 normotensive asymptomatic subjects with type 2 diabetes showed reduced left ventricular systolic and diastolic function compared with healthy subjects (38). Another study of 86 patients with type 2 diabetes free of cardiovascular disease found diastolic dysfunction in 41 of 86 patients (47%) (39). Diastolic dysfunction in type 2 diabetic patients is often found despite adequate metabolic control and freedom from clinically detectable disease (39). The National Institute of Health Sponsored Studies of left ventricular dysfunction in patients with asymptomatic left ventricular dysfunction demonstrated humoral activation characterized by increases in the natriuretic peptides without activation of the circulating renin-angiotensin system (18). In the present study, 21 of the 31 asymptomatic patients who had an abnormal echo had diastolic dysfunction, with BNP levels of 190 ± 60 pg/ml. While BNP levels cannot be used by itself to differentiate between systolic and diastolic dysfunction, a low BNP level in the setting of normal systolic function by echocardiography can likely rule out clinically significant diastolic dysfunction (11).

While the optimal treatment of diastolic heart failure has not been identified (40), occult coronary disease is a potentially reversible cause of diastolic dysfunction. It has been shown that myocardial ischemia augments the synthesis and release of BNP even in the absence of myocardial necrosis or preexisting left ventricular dysfunction. Reversible ischemia may transiently increase left ventricular wall stress, which may be sufficient to cause an elevation in BNP levels (41). Thus, elevated BNP levels may not only indicate left ventricular dysfunction but also be a marker for myocardial ischemia.

This was an observational study done at a single Veteran’s hospital, so one must be careful about generalizing the results to the entire population. Both the area under ROC curve and the negative predictive values are dependent on the patient population studied. Our population represents generally older, predominantly male veterans with a high prevalence of cardiac disease.

Echocardiographic recordings form the basis of the diagnosis of systolic and diastolic dysfunction in the current study. Numerous previous reports have validated the ability of cardiac ultrasonography to detect abnormalities of contractile function and to quantitate left ventricular volumes and ejection fraction (12,13). All patients in this study so designated had clear-cut evidence of left ventricular systolic dysfunction. Although diastolic dysfunction implies an abnormal relationship between left ventricular volume and pressure, echocardiography is capable of assessing only parameters related to volume. Therefore transmitral and pulmonary venous flow velocities provide only indirect measurements of diastolic performance. Nevertheless, these parameters have been shown to provide reliable markers of impaired diastolic function and are applied for this purpose in clinical practice.

Finally, in this study, BNP is being used to identify “any” impairment of ventricular function rather than “significant” impairment. Thus symptoms in such patients are not necessarily of cardiac origin and could challenge the value of labeling those patients as “abnormal” with BNP.

In summary, a rapid whole-blood test for BNP, which can be performed in the laboratory or the clinic, can reliably detect the presence or absence of left ventricular dysfunction in patients with diabetes, regardless of whether they have symptoms of left ventricular dysfunction. BNP levels should not replace imaging techniques in the diagnosis of CHF because these methods provide complementary information. However, a normal plasma concentration of BNP makes significant ventricular dysfunction unlikely. Elevated plasma concentrations warrant further cardiological evaluation. We believe that BNP may be a useful screening tool for left ventricular dysfunction, especially in the community of diabetic patients where the greatest risk of cardiac disease exists and where there is a limited access to echocardiography. In this setting, it is likely that BNP analysis would greatly assist in the appropriateness of patient referral and in optimization of drug therapy.