Echocardiography is the standard method used to evaluate the severity and cause of mitral regurgitation (MR), and to assess left ventricular (LV) systolic function (1). However, it may be difficult to obtain an accurate quantitative assessment of the severity of regurgitation (2- 3). Quantitative estimates of severity based on measurement of the regurgitant fraction and the regurgitant orifice area (4) are technically demanding. Left ventricular function may also be difficult to evaluate in patients with severe MR because the ejection fraction (EF) can be maintained in the presence of LV dysfunction (5). The American College of Cardiology and American Heart Association recommend surgery for severe MR if symptoms occur or there is evidence of asymptomatic LV dysfunction, defined as an LV end-systolic dimension of >45 mm or an EF of <60% (6). However, difficulties in detecting early LV dysfunction, accurately assessing the severity of regurgitation, or recognizing early cardiac symptoms can make it difficult to determine the optimal timing of mitral valve surgery (7).

Brain natriuretic peptide (BNP), amino-terminal brain natriuretic peptide (N-BNP), and atrial natriuretic peptide (ANP) are synthesized and secreted by cardiac myocytes in response to increases in wall stress (8- 9). Plasma levels of the natriuretic peptides are elevated with LV systolic dysfunction after myocardial infarction (MI) and are independent prognosticators (10- 11). It is possible that natriuretic peptide testing in patients with MR will add to the information obtained from echocardiography, but to date there are only limited published data (12). In this study, natriuretic peptide levels were measured in patients with MR as potential markers of the severity of regurgitation, LV dysfunction, and the presence of symptoms.

Patients referred to a single cardiac center for echocardiography between January 2000 and October 2001 who met the entry criteria for the study were invited to participate. Forty-nine patients with isolated MR and LV EFs of >55% on the screening echocardiogram were enrolled in the study. The exclusion criteria included mitral stenosis (mitral valve area <1.5 cm²), aortic valve disease (peak velocity across aortic valve >2.5 m/s or more than mild aortic regurgitation), LV hypertrophy (LV wall thickness ≥12 mm), primary right heart disease, history of MI, previous cardiac surgery, uncontrolled hypertension (blood pressure >160/100 mm Hg), plasma creatinine level >0.18 mmol/l, and the presence of moderate or severe respiratory disease. One hundred normal subjects without clinical symptoms and without signs of cardiopulmonary or renal disease were used as controls. The study protocol was approved by the regional ethics committees, and all patients gave written informed consent.

Symptoms were assessed independently by two cardiologists blinded to the natriuretic peptide levels and echocardiographic findings. Patients were judged symptomatic if they had a history of dyspnea on exertion (New York Heart Association functional class ≥2) or required treatment with a loop diuretic for dyspnea. The heart rate, blood pressure, cardiac rhythm, height, weight, and current medications were recorded. Clinical outcomes including mitral valve surgery, death, and hospital admission for heart failure were determined for a median of 16 months following the baseline assessment.

All patients underwent echocardiographic and Doppler examination using a standard protocol on commercially available systems (Series 5 or Vivid 5 System, GE Vingmed Ultrasound, Horten, Norway). Analysis was performed at a later time by one of two experienced echocardiographers blinded to the patients’ symptom status and natriuretic peptide levels. The LV end-systolic and end-diastolic dimensions, LV wall thickness, and left atrial (LA) dimensions were measured according to American Society of Echocardiography guidelines (13). All measurements were averaged from three to five cardiac cycles. The LA volume was estimated using the biplane ellipsoid formula. The LV end-systolic and end-diastolic volumes and EF were measured from the apical four-chamber view using the modified Simpson’s single-plane method (13). Because the results of analyses were similar when the LA and LV sizes were estimated from a single parasternal dimension as compared with calculated volumes, the results for dimensions only are presented. The severity of MR was assessed from the regurgitant fraction (14) and vena contracta width (15), and the MR score described by Thomas et al. (2). The MR score is calculated from visual assessment of LA size, MR jet penetration, mitral continuous-wave Doppler characteristics, pulmonary venous flow pattern, tricuspid regurgitation velocity, and proximal isovelocity surface area radius.

Venous blood samples were taken on usual medications with the patient resting quietly while semi-recumbent. The samples were taken into chilled ethylene-diamine-tetra-acetic acid vacutainers, placed immediately on ice, and centrifuged within 20 min at +4°C. The plasma was stored at −80°C before being assayed for BNP, N-BNP, and ANP using established radioimmunoassays (16- 18). The within-assay coefficients of variation for the natriuretic peptides were 5.2% for BNP, 3.5% for N-BNP >200 pmol/l, 3.9% for N-BNP 50 to 200 pmol/l, 11.6% for N-BNP <50 pmol/l and 2% for ANP. To convert natriuretic peptide levels expressed in pmol/l to pg/ml, multiply by 3.47 for BNP, 8.46 for N-BNP, and 3.08 for ANP (16- 18).

The Student t test was used to compare continuous variables between groups. The Fisher exact test was used to compare categorical variables, as more than 25% of the classes had frequencies <5. Normality was checked for each variable using box and whisker and probability plots and the Shapiro-Wilk normality test. Distribution of the peptides were skewed but were normal after natural logarithmic (Ln) transformation. The Pearson correlation coefficient was used to assess the association between the Ln of natriuretic peptide levels and echocardiographic variables. General linear models were used to assess natriuretic peptide levels in symptomatic and asymptomatic patients with adjustment for measures of the severity of MR, LV function, or age and gender (where stated). For the ratio of the geometric means between the groups (symptomatic/asymptomatic), 95% confidence intervals (CIs) are reported for each peptide (19). The areas under the receiver-operator characteristic (ROC) curves were used to evaluate the diagnostic performance of natriuretic peptide and echocardiographic measurements for the presence of symptoms. A Stata 6.0 statistical software package (Stata Corporation, College Station, Texas) was used for ROC curve analysis and SAS release 8.0 software (SAS Institute, Cary, North Carolina) was used for the remaining analyses.

Of the 49 patients with MR, 25 were men and 24 women. The mean age was 55 ± 19 years. There were 100 normal controls, of whom 41 were men and 59 women (mean age 55 ± 11 years). The etiology of MR was rheumatic in 15 patients, 4 of whom had mild mitral stenosis (mitral valve area >1.5 cm²), mitral valve prolapse in 33 patients, and postendocarditis in one patient. Thirty-three patients were asymptomatic and 16 were symptomatic. The clinical characteristics and echocardiographic measurements of asymptomatic and symptomatic patients are compared in (Table le1). Symptomatic patients were more likely to be in atrial fibrillation and to be treated with a loop diuretic or an angiotensin-converting enzyme inhibitor. On average, symptomatic patients had more severe MR as measured by the regurgitant fraction, vena contracta width, and MR score. They also had larger LA dimensions and a trend toward larger LV end-diastolic and end-systolic dimensions that did not reach statistical significance. The LV EFs were similar in asymptomatic and symptomatic patients, and no significant differences existed in age, gender, smoking, history of hypertension, measured blood pressure, diabetes mellitus, or serum creatinine levels.

Correlation coefficients for the associations between each of the natriuretic peptides and echocardiographic measures of the severity of MR, LA dimension, and LV size and function are presented in (Table le2). The plasma levels of BNP, N-BNP, and ANP rose with increasing severity of MR and with increasing LA dimensions. The correlation coefficients were similar for BNP, N-BNP, and ANP. No statistically significant correlation existed between any of the natriuretic peptides and the LV dimensions or EF.

In normal subjects the plasma levels of natriuretic peptides rose with increasing age. For an increase in age of 10 years, BNP on average increased 1.25 (95% CI 1.16 to 1.36) times, N-BNP increased 1.40 (95% CI 1.25 to 1.57) times, and ANP increased 1.30 (95% CI 1.21 to 1.38) times. Plasma levels of natriuretic peptides were higher in women than in men after adjustment for age (BNP 1.33 [95% CI 1.13 to 1.57] times higher, N-BNP 1.70 [95% CI 1.33 to 2.18] times higher, and ANP 1.31 [95% CI 1.13 to 1.50] times higher, p < 0.001 for all). No statistically significant interaction existed between age and gender.

(Table le3)shows the natriuretic peptide levels of the symptomatic and asymptomatic patients with MR and the control subjects. After adjustment for age, gender, and body surface area, the levels of each of the natriuretic peptides were on average higher in asymptomatic patients with MR than in normal controls (p < 0.001 for all), and higher in symptomatic than in asymptomatic patients with MR (p < 0.001 for all). The differences between symptomatic and asymptomatic patients with MR remained after adjustment was also made for the LV end-systolic dimension (BNP 2.06 [95% CI 1.32 to 3.23] times higher, N-BNP 2.63 [95% CI 1.46 to 4.75] times higher, and ANP 2.11 [95% CI 1.38 to 3.21] times higher), LA dimension (BNP 1.45 [95% CI 0.95 to 2.22] times higher, N-BNP 1.57 [95% CI 1.09 to 2.70] times higher, and ANP 1.79 [95% CI 1.15 to 2.81] times higher), MR score (BNP 1.9 [95% CI 1.1 to 3.1] times higher, N-BNP 2.5 [95% CI 1.3 to 4.9] times higher, and ANP 1.8 [95% CI 1.1 to 2.9] times higher), and vena contracta width (BNP 1.5 [95% CI 1.0 to 2.4] times higher, N-BNP 2.0 [95% CI 1.1 to 3.7] times higher, and ANP 2.1 [95% CI 1.3 to 3.3] times higher). The associations between BNP and the LV end-systolic dimension, vena contracta width, and LA dimension in asymptomatic and symptomatic patients are shown in (Figure 1).

The sensitivity, specificity, and area under the ROC curve for symptoms by natriuretic peptide levels and echocardiographic measures are shown in Table 4. An ROC curve showing the sensitivity and specificity of BNP, LA dimension, vena contracta width, and LV end-systolic dimension for the presence of symptoms is depicted in (Figure 2). The area under the ROC curve was higher for each of the natriuretic peptides than for the echocardiographic measures of the severity of regurgitation, LA size, and LV dimensions.

After a median follow-up of 16 (range 6.2 to 23) months, seven (44%) symptomatic patients had mitral valve surgery (BNP values at baseline: 11 pmol/l, 12 pmol/l, 21 pmol/l, 22 pmol/l, 23 pmol/l, 33 pmol/l, and 52 pmol/l). One symptomatic patient had a normal exercise stress echocardiogram and was not referred for surgery (BNP 12 pmol/l). The other eight symptomatic patients declined surgery or were not referred because of co-morbidity; two died from noncardiac causes. Four (12%) asymptomatic patients underwent mitral valve surgery; in three there was no clinical or echocardiographic change before surgery performed for a flail leaflet (n = 2, BNP 6 pmol/l and 7 pmol/l, respectively) or coexistent coronary artery disease (n = 1, BNP 7 pmol/l). One asymptomatic patient developed worsening LV function and symptoms and was referred for surgery after 20 months’ follow-up (BNP 4 pmol/l). Serial measurements of natriuretic peptides were not obtained during follow-up.

Obtaining an accurate quantitative assessment of the severity of MR is technically demanding (2- 4). For this reason a qualitative assessment of severity based on several echocardiographic measures is often used in clinical practice. In this study the natriuretic peptides BNP, N-BNP, and ANP increased with increasing MR assessed by three echocardiographic measures, the width of the vena contracta (15), the regurgitant fraction (14), and the MR score (2). Conversely, no statistically significant correlation existed between the LV end-systolic dimension, LV end-diastolic dimension or EF, and natriuretic peptide levels. The echocardiographic measure most strongly associated with both natriuretic peptide levels and the presence of symptoms was the LA dimension.

In addition to associations with echocardiographic measures of the severity of MR, natriuretic peptide levels were higher in symptomatic than in asymptomatic patients. Symptomatic patients were more likely to have more severe MR, but the association between natriuretic peptide levels and symptoms remained statistically significant, although weaker, after adjustment for echocardiographic measures of the severity of regurgitation and LV function. These observations suggest that natriuretic peptide levels provide an additional method for assessing the severity and symptoms of MR when the LV EF is normal.

Few published studies have evaluated natriuretic peptide levels in MR, Brookes et al. (12) measured BNP levels in 22 patients with MR, but that study included patients with impaired LV function and did not assess the relation between the severity of MR and natriuretic peptide levels. As in the current study, BNP levels were higher in symptomatic patients, and no significant association was seen between BNP levels and LV dimensions.

In the current study the increase in plasma N-BNP levels seen with increasing severity of MR and with increasing symptoms was greater than for BNP, but the strength of the association was similar. The N-terminal fragment of BNP does not have specific clearance receptors and, therefore, has a longer half-life and a higher plasma concentration than BNP (20). Previous studies suggest that BNP is primarily synthesized by ventricular myocytes and reflects changes in ventricular function, whereas ANP is predominantly stored in atrial myocytes and released in response to increases in atrial wall “stress” (8,21). In the current study, natriuretic peptide secretion in patients with MR appeared to be related to increases in LA rather than LV wall “stress.” The relative increases in the plasma levels of BNP and N-BNP were as large as for ANP, and the correlations with the LA dimension, severity of MR, and symptoms were similar for the three peptides. A possible explanation is that atrial myocytes synthesize BNP as well as ANP in response to the chronic increase in LA pressure. This hypothesis is supported by the demonstration of synthesis of BNP by atrial cardiomyocytes and co-storage of ANP and BNP in atrial granules (22- 23).

Results of this study suggest that natriuretic peptide testing may add to the information obtained by echocardiography in the assessment of MR in clinical practice. When echocardiographic assessment is technically difficult, low N-BNP, BNP, or ANP levels would suggest that MR is not severe. Measurement of natriuretic peptides may also be useful when it is not clear whether symptoms of dyspnea or fatigue are due to cardiac disease (24- 25). Prospective follow-up studies are needed to determine whether natriuretic peptides can be used to monitor asymptomatic patients with moderate to severe MR in the primary care setting, or to guide timing of surgery. Limited clinical follow-up in the current study suggests a single baseline measurement does not reliably predict disease progression, but serial measurements may be useful.

In normal subjects, natriuretic peptides increased with age, were higher in women than in men, and were inversely related to body surface area, as observed in other studies (26). Patients with MR were compared with the normal subjects, but further studies are needed to determine whether and how the impact of age, gender, and body surface area on natriuretic peptide levels should influence the interpretation of plasma levels. To allow assessment of the effects of MR on natriuretic peptide levels, patients with other causes of elevated plasma natriuretic peptides were excluded from this study. Further investigations are needed to determine whether the association between natriuretic peptide levels and symptoms is similar in persons with additional cardiac conditions that are known to increase natriuretic peptide levels. As in clinical practice, assessment of symptoms may be difficult. It is likely that in some patients the symptoms were not a consequence of MR, while others were classified as asymptomatic because they undertook little activity or ignored subtle symptoms. Previous studies have demonstrated a clear inverse relationship between the LV EF and BNP and N-BNP levels after MI or with cardiac failure (9- 11). In the current study, no clear association was seen between natriuretic peptide levels and the EF, but patients with an EF of <55% on the screening echocardiogram were excluded. Additional studies are needed to confirm that natriuretic peptide levels increase further when LV function declines in patients with MR. Another consideration is the possible presence of LV dysfunction when the EF is maintained in MR. Additional investigations not performed in the current study, such as exercise stress echocardiography, can identify LV dysfunction with a maintained resting EF, and these may improve the assessment of symptoms in some patients (27).

Plasma levels of BNP, N-BNP, and ANP rise with increasing severity of MR and are higher in symptomatic than in asymptomatic patients, even after adjustment for echocardiographic measures of the severity of regurgitation. Prospective follow-up studies are needed to determine whether natriuretic peptides will provide a simple low-cost method for monitoring asymptomatic patients and whether repeated measurements are useful for deciding on timing of surgery.

We wish to thank Charlene Nell for her secretarial assistance and Anna Breckon for her editorial assistance.