Secretion of Prohormone of B-Type Natriutretic Peptide, proBNP1–108, is Increased in Heart Failure

Abstract

OBJECTIVES

Using a novel, specific assay for proBNP_1–108 we tested the hypotheses that proBNP_1–108 is secreted by both non-failing and failing human hearts and that proBNP_1–108 secretion is increased in failing hearts.

BACKGROUND

The prohormone of B-type natriuretic peptide (proBNP_1–108) is a 108-amino acid peptide produced primarily by the heart and cleaved into biologically active BNP_1–32 and the biologically inactive NT-proBNP_1–76. It is unknown to what extent increased cardiac proBNP_1–108 secretion as compared to reduced peripheral processing is responsible for elevated proBNP_1–108 levels in patients with heart failure (HF) as compared to subjects without HF.

METHODS

The transcardiac gradient of proBNP_1–108 was determined by collecting arterial blood and blood from the coronary sinus (CS). Samples from subjects without overt heart disease (n=9) were collected during cardiac catheterization after coronary artery disease had been excluded. Samples from HF patients (n=21) were collected during implantation of a biventricular pacemaker. ProBNP_1–108 was measured with a new assay (BioRad). Values are median (25^th/75^th percentile).

RESULTS

The gradient of proBNP_1–108 across the non-failing hearts was 8 (2/20) ng/L (aorta: 15 (1/25) ng/dL, CS: 24 (8/41) ng/dL; p=0.018). The transcardiac gradient of proBNP_1–108 in the failing hearts was 326 (96/482) ng/dL (arterial: 381 (201/586) ng/dL, CS: 709 (408/1087) ng/dL; p<0.001). The transcardiac gradient was greater in failing than non-failing hearts (p=0.001).

CONCLUSION

ProBNP_1–108 is secreted by non-failing and failing human hearts, but more so in the latter. It remains to be established where peripheral processing of proBNP_1–108 occurs and how this is affected by disease.

INTRODUCTION

Mature B-type natriuretic peptide (BNP) has vasodilating, natriuretic, antihypertrophic, antifibrotic, and metabolic properties. Human BNP is initially synthesized as the 134 amino acid precursor, preproBNP. A 26 amino acid signal peptide is cleaved and the 108 amino acid prohormone, proBNP_1–108, is formed. The proBNP_1–108, which is less biologically active than mature BNP, is then cleaved enzymatically, presumably by corin and furin, to the biologically inactive amino terminal fragment NT-proBNP_1–76 and the biologically active, mature peptide BNP_1–32.^1–3 BNP is secreted primarily by the heart in response to increased wall stress; thus, assays for BNP or NT-proBNP were developed as cardiovascular disease biomarkers.^4–13

It was initially assumed that the “mature”, most bioactive BNP form (BNP_1–32) was the only form secreted into the circulation.¹⁴ Now it is known that some of the circulating BNP immunoreactivity in normal humans is proBNP_1–108 and various proBNP_1–108 derivatives, including BNP_1–32.^15–22 Also, proBNP_1–108 is cleaved in human plasma to produce BNP_1–32.³ Conventional assays for BNP are not specific because antibodies directed against BNP_1–32 and NT-proBNP_1–76 crossreact with proBNP_1–108.²³ In some disease states, BNP_1–32 levels measured by mass spectrometry are dramatically lower than the BNP immunoreactivity measured by conventional BNP assays.^{24, 25} Indeed, proBNP_1–108, and not BNP_1–32, is the major BNP molecular form in heart failure (HF).^{19, 20, 22} This may explain the paradox seen in HF of high circulating BNP levels, but reduced BNP activity.

A specific assay for uncleaved, full length proBNP_1–108 was recently developed. It is directed against the proBNP_1–108 hinge region, which is only present in the uncleaved prohormone, proBNP_1–108, and not in BNP_1–32 or NT-proBNP_1–76.¹⁶ (Fig. 1) Using this assay, we and others reported that proBNP_1–108 circulates in healthy subjects’ plasma and is elevated in cardiovascular disease, including HF.^{10, 16, 26}

Figure 1.

Schematic of proBNP_1–108 and two of its derivatives, NTproBNP_1–76 and BNP_1–32 (which is proBNP_77–108). Two putative enzymes involved in proBNP cleavage are corin and furin. The specific proBNP_1–108 assay (Bio-Rad, Hercules, CA, USA) uses a capture antibody directed against the hinge region, i.e. region in which proBNP_1–108 is cleaved into bioactive BNP_1–32 and inactive NTproBNP_1–76. There is a monoclonal detection antibody directed against BNP_1–32 epitopes. Thus, the proBNP_1–108 assay only detects uncleaved proBNP forms.

While the presence of proBNP_1–108 in the peripheral circulation is established, it remains undefined if the normal heart secretes proBNP_1–108 and if secretion increases in HF. Net cardiac secretion of BNP_1–32 immunoreactivity has previously been reported; however, the assays used were not BNP_1–32 specific and probably also detected proBNP_1–108.²⁷

This study had two objectives. The first was to confirm that this proBNP_1–108 assay is proBNP_1–108 specific and does not detect degraded BNP forms that circulate in HF. The second objective using this novel assay was to assess invasively for the first time the actual cardiac secretion of proBNP_1–108 in humans without cardiac disease and with HF. We hypothesized that proBNP_1–108 is secreted by the normal and failing human heart and that secretion increases in HF.

METHODS

proBNP1–108 assay specificity

We confirmed and extended Giuliani et al. by measuring proBNP_1–108 immunoreactivity in 1 mL samples of normal human plasma with the BioRad assay¹⁶ spiked with either 1,000 pg or 10,000 pg of the following BNP molecular forms: BNP_1–32, BNP_8–32, proBNP_44–76, NT-proBNP_8–29, nesiritide (Scios, Mountain View, CA, USA), and proBNP_1–108 (Hytest, Finland). A wider range of proBNP_1–108 was used for spiking; 500 to 10,000 pg. Measurements were repeated three times.

Study population

Samples from subjects without heart disease were collected at the Mayo Clinic (Rochester, MN, USA) during cardiac catheterization after exclusion of coronary artery disease. Samples from subjects with left ventricular (LV) systolic dysfunction were collected at the Helios Clinic (Wuppertal, Germany) during implantation of a biventricular pacemaker for cardiac resynchronization therapy. All HF patients had transthoracic echocardiography performed at the Helios Clinic, Wuppertal. All patients without heart disease had ejection fractions (EF) determined by transthoracic echocardiography or nuclear medicine scan at the Mayo Clinic. All subjects gave informed consent and had their medical records reviewed. The study was approved by the respective institutions’ institutional review board or equivalent.

Sample processing and assays

The proBNP_1–108 transcardiac gradient was determined by collection of arterial and coronary sinus (CS) blood. Blood samples were placed into ethylenediaminetetraacetic acid-tubes on ice, chilled to 4°C, centrifuged at 2,500 rpm for 10 min, and the plasma stored at −80°C until used. Samples from Wuppertal were shipped to the Mayo Clinic on dry ice. ProBNP_1–108 was measured at the Mayo Clinic with the Bio-Rad assay (Bio-Rad, Hercules, California) on a commercially unavailable automated analyzer. The assay was developed by Giuliani et al.¹⁶ The lower limit of detection (LOD) is 2 ng/L; levels below this were set as 1 ng/L (i.e. half-way between 0 and LOD). The interassay and intra-assay variabilities are 10.3% and 11.6%, respectively. All other laboratory values were measured at the respective institutions.

Statistical methods

Values are expressed as mean±SEM for normally distributed data and as median (25^th/75^th percentile) for not normally distributed data. ProBNP_1–108 secretion was assessed separately in subjects with non-failing and failing hearts by comparing arterial levels to CS levels with paired Wilcoxon signed-rank test. The transcardiac gradients, i.e. difference between levels in CS vs. arterial blood, were compared between the non-failing and failing hearts by Mann-Whitney U-test. The range of transcardiac proBNP gradients was wide in HF, so we looked at clinical variables associated with the gradient; these analyses are hypothesis-generating and p-values nominal only. For categorical variables, the Mann-Whitney U-test was used. For continuous variables, correlations with proBNP_1–108 were calculated according to Spearman.

RESULTS

proBNP_1–108 assay specificity

Figure 2 reports average proBNP_1–108 immunoreactivity as a function of BNP molecular form added to normal human plasma (mean proBNP_1–108 immunoreactivity = 14.3 ng/L). No significant changes from baseline immunoreactivity occurred with any BNP form addition other than proBNP_1–108.

Figure 2.

ProBNP_1–108 immunoreactivity of normal human plasma spiked with various proBNP_1–108 derivatives and proBNP_1–108. Except for when proBNP_1–08 was added (far right), none of the added proBNP_1–108 derivatives had proBNP_1–108 immunoreactivity different from unspiked plasma (far left).

Subject Characteristics

Table 1 gives subject characteristics, with and without HF. The majority of HF subjects were older males with an ischemic (62%).etiology.

Table 1.

Characteristics of subjects with non-failing and failing hearts.

Characteristic	Non-Heart Failure	Heart Failure
N	9	21
Female, %	78	14
Age, years	46 ± 12	69 ± 9
Creatinine (mg/dL)*	0.9±0.2	1.4±0.3
NYHA class III,/IV	NA	17/3
Ischemic etiology, %	NA	62
Coronary artery disease, %	0	62
Ejection fraction, %	(>50%)	24 ± 7
Diabetes mellitus, %	0	33
Atrial fibrillation, %	0	29

Values are % or mean ± SD.

^*Creatinine values for the two groups were measured at different institutions and therefore not directly comparable.

Transcardiac gradient of BNP

The transcardiac gradient of proBNP_1–108 across the non-failing hearts was 8 (2/20 ng/L (arterial: 15 (1/25) ng/L, CS: 24 (8/41) ng/L; p=0.018; Fig. 3A). In two subjects (both male) proBNP_1–108 levels were below the LOD. The proBNP_1–108 transcardiac gradient in HF was 326 (96/482) ng/L (arterial: 381 (201/586) ng/L, CS: 709 (408/1087) ng/L; p<0.001; Fig. 3B). The transcardiac gradient was greater in HF (p=0.001; Fig. 4A). The shape and variance of the two distributions of transcardiac gradients are not the same, therefore, Mann-Whitney U test assumptions are not met. This is an analysis limitation. Figure 4B shows transcardiac proBNP_1–108 gradients in relation to age for the non-failing and failing hearts.

Figure 3.

ProBNP_1–108 secretion in subjects without HF. Left graph: lines show individual subjects. Right graph: bars show median, 25^th and 75^th percentile (A). ProBNP_1–108 secretion in patients with HF. Left graph: lines show individual subjects. Right graph: bars show median, 25^th and 75^th percentile (B).

Figure 4.

Transcardiac gradients of proBNP_1–108 secretion in subjects with non-failing and failing hearts. Left graph: Scatter plot shows individual subjects. Right graph: bars show median, 25^th and 75^th percentile (A). Scatter plot shows individual transcardiac proBNP_1–108 gradients in subjects with and without HF in relation to age. Age was not the major determinant of the transcardiac gradient, but rather HF was (B).

Given the wide range of proBNP_1–108 transcardiac gradients in the HF group, we assessed within this group whether any clinical variables were associated with the gradient. The proBNP_1–108 transcardiac gradient was greater in diabetics (no diabetes vs. diabetes: 215 (56/363) vs. 491 (448/1119), p=0.01; levels were not different between diabetics and non-diabetics in the arterial blood (p=0.80) and in the CS (p=0.17)). No associations were observed for age (r=0.15, p=0.52), sex (p=0.89), body mass index (r=−0.20, p=0.38), creatinine (r=−0.11, p=0.64), NYHA class (class 3 vs. 4: p=0.84), HF etiology (ischemic vs. non-ischemic, p=0.92), rhythm (sinus rhythm vs. atrial fibrillation: p=0.97), hypertension (p=0.91), echocardiographic LV EF (r=−0.23, p=0.34), LV end diastolic diameter (r=0.36, p=0.19), left atrial diameter (r=−0.22, p=0.38), or medication (spironolactone (p=0.49), statin (p=0.85)).

DISCUSSION

We describe for the first time net cardiac secretion of proBNP_1–108 in subjects with non-failing and failing hearts. The proBNP_1–108 transcardiac gradient was greater in HF. We also showed that the proBNP_1–108 assay used here does not recognize mature BNP_1–32, degraded forms of mature BNP that are increased in HF, or degraded forms of the NT-proBNP region. Thus, we present the first evidence that the increased proBNP_1–108 immunoreactivity seen in the plasma of HF patients is at least partially due to increased secretion.

In the non-failing group, most subjects showed net proBNP_1–108 secretion. This is consistent with our report that healthy subjects have circulating proBNP_1–108.¹⁰ Thus, the previously reported proBNP_1–108 peripheral processing is a normal physiologic process.³ The two normal EF subjects with proBNP_1–108 levels below the LOD were the two males in this group. This fits with prior reports that circulating proBNP_1–108, BNP_1–32 and NTproBNP_1–76, are higher in women than men.^{8, 10, 28} In our previous study examining venous plasma proBNP_1–108 levels in a random sample of the community aged 45 and older, proBNP_1–108 levels were below the LOD in 5.5% of males and 2.2% of females. Median levels (5^th/95^th percentiles) of proBNP_1–108 in venous blood samples defined in a normal subgroup of this random sample of the general community were 17 (3–48) ng/L for women aged 45–54 years and 21 (3–81) ng/L for women 55–64 years. These values are similar to those found here in the arterial blood of the women in the non-failing group 16 (1–151) ng/L; the patient with the 151 ng/L was an outlier, with the next highest value being 27 ng/mL. Thus, most subjects in the non-failing group were likely normal despite undergoing cardiac catheterization.

Within the HF group, all patients had detectable proBNP_1–108 in arterial and CS blood. While the proBNP_1–108 transcardiac gradient was higher in HF, the range of gradients was wide, with some gradients being similar to the non-failing group. To evaluate factors potentially affecting proBNP_1–108 secretion, we looked for associations with clinical parameters. The only variable associated with proBNP_1–108 secretion was diabetes mellitus, with non-diabetics having lower proBNP_1–108 transcardiac gradients than diabetics. No difference in proBNP_1–108 levels between diabetics and non-diabetics existed in the peripheral circulation. This is interesting since higher BNP/ANP plasma levels are reported to be associated with lower plasma glucose levels and protective against development of diabetes.^{29, 30} The question arises if diabetes affects the expression or activity of proBNP cleaving enzymes or proBNP_1–108 glycosylation, a modification that changes the prohormone’s susceptibility to enzymatic cleavage.³¹ This study was not designed to address this, and the diabetes association is only hypothesis-generating.

There was an age and gender difference between the non-failing and HF groups. The normal EF group was predominantly female and younger. As female sex is associated with higher BNP_1–32 and higher proBNP_1–108 plasma concentrations, the sex mismatch makes it more difficult to show a difference between groups.^{8, 10, 28} Given that our predominantly female normal EF group still had lower CS proBNP_1–108 concentrations and lower transcardiac proBNP_1–108 gradients than the predominantly male HF group, this mismatch actually strengthened our conclusion. Regarding age, younger age is associated with lower plasma BNP_1–32 and proBNP_1–108 levels; however, the relative contribution of age to BNP concentrations is unlikely to explain the differences between the non-failing and failing hearts.^{9, 10, 28} Indeed, Fig. 3B suggests that age matching would not have changed our results substantially. Also, in the normal subgroup of a general population sample, venous plasma levels of proBNP_1–108 in males aged 65–74 years were 16 (2/53) ng/L, median (5^th/95^th percentiles).¹⁰ This again suggests that age is only a minor contributor to the elevated levels seen in the male HF subjects. Finally, we measured plasma concentrations and not flow. Therefore, although a step-up in concentration from arterial to CS blood suggests net secretion, lower coronary blood flow in HF theoretically might make proBNP_1–108 secretion appear higher. However, the median proBNP_1–108 gradient in HF was 40x higher than in non-failing hearts, so a difference in coronary blood flow is unlikely to explain our results.

In summary, with a novel, specific proBNP_1–108 assay we demonstrated for the first time cardiac secretion of proBNP_1–108 and that this secretion increases in HF. This does not rule out decreased peripheral proBNP_1–108 processing in HF, but suggests that it is not the sole reason for increased proBNP_1–108 levels in HF.