Abstract
Background
Bone morphogenetic protein-10 (BMP10), a protein predominantly secreted by atrial cardiomyocytes, has emerged as a promising biomarker in patients with atrial fibrillation (AF) and as a marker of atrial stress in patients with heart failure. Its clinical correlates in community-dwelling adults and its association with incident AF remain unexplored.
Methods and Results
BMP10 was measured in 5883 participants from the PREVEND community-based cohort (mean age 53.6 years; 51.5% females). The mean BMP10 concentration was 2.20±0.43 ng/ml. In multivariable linear regression analyses, the strongest positive correlates of BMP10 were female sex (standardized beta-[Sβ]: 0.34), high-density lipoprotein cholesterol (HDL-C; Sβ: 0.16), total cholesterol (Sβ: 0.13) and N-terminal pro-B-type natriuretic peptide (NT-proBNP; Sβ: 0.13). The strongest negative correlates were relative fat mass (Sβ: -0.21), C-reactive protein (CRP; Sβ: -0.16), and estimated glomerular filtration rate (eGFR; Sβ: -0.13); P for all<0.001. BMP10 was not significantly associated with prevalent hypertension, myocardial infarction, stroke, or heart failure. Over a median follow-up of 6.4 years, 154 participants (2.6%) developed incident AF, corresponding to 4.2 cases per 1000 person years. In a multivariable Cox-regression model, higher BMP10 levels were significantly associated with incident AF (Hazard ratio per 1SD increase: 1.58; 95%CI: 1.23-2.04).
Conclusions
In community-dwelling adults, BMP10 levels are higher in women, positively associated with HDL-C and NT-proBNP, and inversely associated with fat mass, inflammation, and kidney function. Although not linked to prevalent cardiovascular disease, higher BMP10 levels are independently associated with incident AF.
Abbreviations
- BMP10
Bone morphogenetic protein-10
- AF
Atrial Fibrillation
- PREVEND
The Prevention of Renal and Vascular End-stage Disease
- NT-proBNP
N-terminal pro-B-type natriuretic peptide
- CRP
C-reactive protein
Introduction
Atrial fibrillation (AF), the most common sustained arrhythmia worldwide, imposes a substantial health burden by increasing the risk of cardiovascular morbidity and mortality (1). A hallmark of AF pathogenesis is adverse atrial remodelling (2), often driven by overactivation of pro-fibrotic signalling pathways such as the transforming growth factor-beta (TGFβ) pathway – in response to haemodynamic, inflammatory and metabolic stress (3).
Bone morphogenetic protein 10 (BMP10) is a cardiac-specific member of the TGFβ superfamily (4) and is thought to be predominantly produced by atrial cardiomyocytes (5, 6). Its expression in atrial tissue is regulated by paired-like homeodomain transcription factor-2 (PITX2), a transcription factor essential for left atrial development (7). In murine models, reduced PITX2 expression in the left atrium correlated with higher BMP10 expression (8). In patients undergoing AF ablation, reduced PITX2 expression in left atrial tissue and elevated circulating BMP10 levels were associated with recurrent AF (8). Additional studies in independent AF cohorts have shown that higher circulating BMP10 levels are associated not only with AF recurrence (9) but also with other AF-related complications such as stroke, heart failure (HF), and mortality (10–12). More recently, in a cohort of HF patients, BMP10 has emerged as a marker of atrial stress and remodelling (13).
To our knowledge, there are no studies examining the distribution of BMP10 or its association with clinical characteristics in the general population. Characterizing these associations would provide insight into BMP10 biology and its potential relevance to AF pathogenesis. Therefore, we measured circulating BMP10 in a large, well-characterized community-based sample and evaluated its association with baseline clinical characteristics and also with the risk of incident AF.
Methods
Study Population
The Prevention of Renal and Vascular End-stage Disease (PREVEND) study was founded in 1997 as a prospective community-based study (14, 15). In brief, all inhabitants (between 28 and 75 years) of the city of Groningen, the Netherlands, were invited (n=85,421), and 47.8% (n=40,856) responded. Individuals with urinary albumin excretion (UAE) > 10 mg/L (n=7768) in their morning urine, as well as a randomly selected control group with UAE < 10 mg/L (n=3394), were selected to attend the PREVEND outpatient clinic (14, 15). After excluding individuals with insulin-dependent diabetes, pregnant women, and those unable or unwilling to participate, a total of 8592 individuals completed the initial PREVEND screening programme (1997-1998) (14, 15). This programme encompassed a comprehensive evaluation of demographic, anthropometric, and clinical characteristics, acquisition of a 12-lead ECG, and collection of fasting venous blood samples along with two 24-h urine specimens (Supplementary Fig. S1).
Participants were subsequently re-evaluated at approximately three-year intervals at the PREVEND outpatient clinic, following a similar standardized protocol. All biological samples were stored at –80°C for future analyses. Plasma BMP10 concentrations were determined from samples obtained during the second study visit (2001-2004), attended by 6894 participants (Supplemental Fig. S1) (16, 17). This visit was used as the baseline for the present analyses.
Of the 6894 participants, we excluded those with prevalent AF (n=85), unknown rhythm status (n=178) (16, 17), unavailable plasma samples (n=734), or an estimated glomerular filtration rate (eGFR) below 30 mL/min/1.73m2 (n=14), resulting in a final study cohort of 5883 participants (Supplemental Fig. S1). Baseline characteristics of the full second-visit cohort (n=6894) and the analytical sample (n=5883) are presented in Supplemental Table S1. Ethical approval was obtained from the local medical ethics committee of the University Medical Center Groningen (MEC96/01/022); all participants signed informed consent, and the study was conducted in accordance with the declaration of Helsinki.
Clinical and laboratory measurements
All anthropometric measurements were performed in a standing position. Waist circumference (WC) was measured midway between the lowest rib and the iliac crest at the end of expiration. Relative fat mass (RFM) was calculated using height and WC with the following equation: 64 – (20 × height/WC) + (12 × sex), with sex = 0 (males) and sex = 1 (females) (18). Body mass index (BMI) was calculated as weight/height2 (kg/m2). Blood pressure (BP) was calculated as the average of two seated measurements. Diabetes was defined as fasting glucose ≥126 mg/dL (7.0 mmol/L), a non-fasting glucose ≥200 mg/dL (11.1 mmol/L), or the use of hypoglycaemic medication. Prevalent cardiovascular disease (CVD), defined as a history of myocardial infarction (MI) or stroke, was obtained from a structured questionnaire, which included criteria such as hospitalization lasting 3 days or more due to the specified condition (14). This data collection was supplemented by an examination of medical records. History of HF was obtained from hospital charts. Smoking behaviour was self-reported and categorized as current smoking (active smoking or cessation within the past year), past smoking (cessation over one year ago), or never smoking. Total cholesterol and plasma glucose were measured by a dry chemistry method (Eastman Kodak, Rochester, New York). High-density lipoprotein cholesterol (HDL-C) was measured by a homogeneous method (direct HDL, Aeroset System; Abbott Laboratories, Abbott Park, Illinois). Plasma N-terminal pro-B-type natriuretic peptide (NT-proBNP) was measured using an electrochemiluminescence sandwich immunoassay (Elecsys proBNP, Roche Diagnostics, Mannheim, Germany). Plasma C-reactive protein (CRP) was measured with a high-sensitivity assay (Dade Behring BNII nephelometer, Marburg, Germany). Plasma Galectin-3 was measured using an enzyme-linked immunosorbent assay (ELISA; BG Medicine, Waltham, MA, USA). Urinary albumin concentration was determined by nephelometry (BNII, Dade Behring Diagnostic, Marburg, Germany) (14). eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-cystatin C equation.
BMP10 measurements
Plasma BMP10 levels were quantified using a Cobas e601 analyzer and a non-commercial, robust prototype electrochemiluminescence immunoassay developed by Roche Diagnostics. The assay employs a quantitative sandwich principle, where the first monoclonal antibody specifically binds the BMP10 as a capture antibody, and a ruthenylated second monoclonal antibody binds to BMP10 as a detection antibody. Recombinant BMP10 is used to normalize the measurements across the runs with a high degree of accuracy. The coefficient of variation was 6.0% and 4.3% for BMP10 concentrations of 1.39 and 3.56 ng/mL, respectively. The lower limit of detection (LOD) was 0.003 ng/mL (3 ng/L), the functional sensitivity / lower limit of quantification (LLOQ) was 0.012 ng/mL (12 ng/L), and the upper limit of quantification (ULOQ) was 10.1 ng/mL (10100 ng/L). Within-run imprecision was 3% and dilutional linearity was 99% (10–13).
AF incidence
Ascertainment of incident AF has been described in detail previously (15–17). In brief, incident AF was diagnosed if present on a 12-lead ECG obtained during scheduled PREVEND visits (i.e., visit 3 or 4), or during an outpatient visit or hospital admission in either of the two hospitals of the city of Groningen. All ECGs were digitally stored and first electronically screened for atrial flutter, ectopic atrial rhythm, or absence of the PR interval. Suspected AF cases were then independently reviewed by two physicians experienced in ECG interpretation. Discrepancies or cases where both observers diagnosed atrial flutter or AF, were adjudicated by two independent cardiologists. The date of incident AF was defined as the date of the first ECG with a definite diagnosis of AF was used (15–17). Follow-up time was calculated from visit 2 to the first occurrence of AF, death, or 31 December 2008.
Statistical analyses
Continuous variables are presented as means (standard deviations) for normally distributed data or as medians (25th-75th percentiles) for skewed data. Categorical variables are presented as counts (percentages). Differences in continuous variables between groups were assessed using Student’s t-test or the Mann-Whitney U test for two groups, and analysis of variance (ANOVA) or the Kruskal-Wallis test for more than two groups. Categorical variables were compared using chi-square tests and Fisher’s exact tests. For subsequent analyses, BMP10 was standardized using sex-specific z-scores.
To examine associations between clinical covariates (independent variables) and BMP10 (dependent variable), we used linear regression models. The initial model was adjusted for age and sex. To identify independent clinical correlates of BMP10, we used a stepwise linear regression model, including baseline covariates with a P-value <0.10 in the initial model. Age and sex were forced in the selection model with a retention P-value equal to 0.05.
We examined associations between BMP10 and incident AF using Cox proportional hazards regression models. The first model was unadjusted; the second model was adjusted for age and sex; the third model was further adjusted for components of the CHARGE-AF model: height, smoking, weight, history of type-2 diabetes, history of hypertension, history of MI or stroke, and history of HF (19). Because there was a significant interaction between BMP10 and time, and the proportional hazards assumption was not met, BMP10 was treated as a time-varying covariate in all models. In sensitivity analyses, Model 3 was further adjusted separately for eGFR, CRP and NT-proBNP. In exploratory subgroup analyses, we assessed age- and/or sex-adjusted associations of BMP10 with incident AF across categories of age, sex, obesity, systemic inflammatory status, renal dysfunction and prevalent CVD. Finally, we examined the predictive performance of BMP10 and NT-proBNP relative to a clinical model, and the predictive performance of BMP10 when added to a clinical model with NT-proBNP, using area under the curve (AUC) and Akaike information criteria (AIC).
Results from Cox models are reported as hazard ratios (HRs) with 95% confidence intervals (CIs). All tests were two-tailed, and p values <0.05 were considered statistically significant. Statistical analysis was performed using STATA version 14.0 (Stata Corp. College Station, TX, USA).
Results
Plasma BMP10 levels
The study included 5883 individuals, with a mean (SD) age of 53.6 (12.1) years; 51.5% of participants were female. The distribution of BMP10 in the total PREVEND population is shown in Fig. 1 and Supplemental Table S2. The mean (SD) BMP10 concentration was 2.20 (0.43) ng/mL, with a range from 0.06 to 4.89 ng/mL. The distribution of BMP10 was broadly similar in both sexes, but mean (SD) BMP10 levels were significantly higher in females than in males [2.31 (0.44) vs 2.08 (0.38) ng/mL, P<0.001].
Figure 1.
BMP10 distribution in the total PREVEND population (black), males (blue) and females (red). Range of BMP10 in the total population was 0.06 to 4.89; in men was 0.73 to 4.52; and in women was 0.06 to 4.89. Mean (SD) BMP10 was 2.20 (0.43) in the total population; 2.08 (0.38) in men; and 2.31 (0.44) in women. Units of BMP10 are in ng/mL.
Participant characteristics according to sex-pooled BMP10 quartiles are presented in Table 1. Age, prevalence of hypertension, and NT-proBNP levels increased progressively across BMP10 quartiles whereas CRP levels decreased in a stepwise manner. Prevalence of diabetes, MI or stroke, and HF did not significantly vary across BMP10 quartiles.
Table 1. PREVEND characteristics according to sex-pooled quartiles of BMP10.
| Characteristics | Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | P-value |
|---|---|---|---|---|---|
| n = 1471 | n = 1471 | n = 1471 | n = 1470 | ||
| BMP10 (mean, SD), ng/mL | 1.7 (0.2) | 2.0 (0.1) | 2.3 (0.1) | 2.7 (0.3) | <0.001 |
| Age (mean, SD), years | 51.0 (11.4) | 52.5 (11.5) | 54.3 (12.1) | 56.6 (12.5) | <0.001 |
| Female sex, n (%) | 758 (51.5) | 758 (51.5) | 758 (51.5) | 757 (51.5) | 1.00 |
| Smoking status | <0.001 | ||||
| Current smokers, n (%) | 487 (33.1) | 440 (29.9) | 356 (24.2) | 318 (21.6) | |
| Past smokers, n (%) | 566 (38.5) | 584 (39.7) | 663 (45.1) | 675 (45.9) | |
| Height (mean, SD), cm | 173.2 (9.3) | 172.6 (9.6) | 172.8 (9.4) | 171.8 (9.7) | <0.001 |
| Weight (mean, SD), kg | 82.7 (14.6) | 80.4 (14.0) | 78.7 (13.7) | 76.1 (14.2) | <0.001 |
| Body mass index, kg/m2 | 27.6 (4.5) | 27.0 (4.3) | 26.3 (4.0) | 25.8 (4.2) | <0.001 |
| Waist circumference, cm | 94.2 (12.4) | 92.7 (12.3) | 91.3 (12.4) | 89.8 (13.1) | <0.001 |
| Relative fat mass, % | 32.8 (7.2) | 32.3 (7.4) | 31.7 (7.2) | 31.2 (7.2) | <0.001 |
| Total cholesterol (mean, SD), mmol/L | 5.3 (1.0) | 5.4 (1.1) | 5.5 (1.1) | 5.6 (1.0) | <0.001 |
| HDL-C (mean, SD), mmol/L | 1.2 (0.3) | 1.2 (0.3) | 1.3 (0.3) | 1.4 (0.3) | <0.001 |
| CRP (median, P25, P75), mg/L | 1.7 (0.8, 4.0) | 1.5 (0.6, 3.1) | 1.2 (0.6, 2.7) | 1.1 (0.5, 2.3) | <0.001 |
| Diabetes, n (%) | 91 (6.3) | 97 (6.6) | 96 (6.6) | 97 (6.6) | 0.97 |
| Antidiabetic medication, n (%) | 48 (3.3) | 46 (3.1) | 49 (3.3) | 49 (3.3) | 0.99 |
| Hypertension, n (%) | 430 (29.3) | 481 (32.8) | 486 (33.1) | 571 (38.9) | <0.001 |
| Antihypertensive medication, n (%) | 299 (20.4) | 315 (21.5) | 312 (21.3) | 371 (25.3) | 0.007 |
| MI or stroke, n (%) | 117 (8.0) | 92 (6.3) | 109 (7.4) | 111 (7.6) | 0.32 |
| Heart failure, n (%) | 12 (0.8) | 12 (0.8) | 14 (1.0) | 12 (0.8) | 0.97 |
| NT-proBNP (median, P25, P75), ng/L | 35.4 (18.6, 66.3) | 39.9 (21.2, 72.7) | 43.8 (23.5, 84.0) | 52.5 (28.9, 102.8) | <0.001 |
| Galectin 3 (median, P25, P75), | 11.3 (8.8, 14.0) | 11.2 (8.8, 14.0) | 11.6 (9.2, 14.4) | 12.0 (9.3, 14.9) | <0.001 |
| UAE (median, P25, P75), mg/24h | 8.2 (5.9, 13.8) | 8.1 (6.0, 13.2) | 8.0 (6.0, 12.6) | 8.6 (6.0, 15.6) | 0.020 |
| eGFR (mean, SD), mL/min/1.73m2 | 95.8 (15.3) | 93.5 (15.8) | 91.4 (16.5) | 88.1 (17.9) | <0.001 |
Abbreviations: HDL-C, high-density lipoprotein cholesterol; CRP, C-reactive protein; MI, myocardial infarction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; UAE, 24h urinary albumin excretion; eGFR, estimated glomerular filtration rate
Associations of plasma BMP10 with clinical characteristics
In linear regression models, BMP10 showed significant associations with several baseline clinical characteristics (Table 2). In a multivariable model including age, sex, smoking, RFM, total cholesterol, HDL-C, NT-proBNP, CRP, galectin-3, UAE and eGFR, all variables remained statistically significant. The strongest positive correlates of BMP10 were female sex (standardized beta [Sβ]: 0.34), HDL-C (Sβ: 0.16), total cholesterol (Sβ: 0.13), and NT-proBNP (Sβ: 0.13). The strongest negative correlates were RFM (Sβ: -0.21), CRP (Sβ: -0.16), and eGFR (Sβ: -0.13); P for all <0.001 (Table 2).
Table 2. Associations of BMP10 with clinical risk factors and biomarkers.
| Characteristics | Age- and sex-adjusted | Multivariable | ||
|---|---|---|---|---|
| Sβ | P-value | Sβ | P-value | |
| Age | 0.15 | <0.001 | 0.08 | <0.001 |
| Female sex | 0.27 | <0.001 | 0.34 | <0.001 |
| Smoking | -0.08 | <0.001 | -0.06 | <0.001 |
| Height | -0.02 | 0.261 | - | - |
| Weight | -0.21 | <0.001 | - | - |
| BMI | -0.20 | <0.001 | - | - |
| WC | -0.24 | <0.001 | - | - |
| RFM | -0.34 | <0.001 | -0.21 | <0.001 |
| Total cholesterol | 0.09 | <0.001 | 0.13 | <0.001 |
| HDL-C | 0.25 | <0.001 | 0.16 | <0.001 |
| CRP | -0.20 | <0.001 | -0.16 | <0.001 |
| Diabetes | -0.03 | 0.013 | - | - |
| Hypertension | -0.02 | 0.165 | - | - |
| MI or stroke | -0.01 | 0.309 | - | - |
| Heart failure | -0.02 | 0.140 | - | - |
| NT-proBNP | 0.14 | <0.001 | 0.13 | <0.001 |
| Galectin-3 | 0.03 | 0.012 | 0.03 | 0.012 |
| UAE | 0.02 | 0.095 | 0.04 | 0.004 |
| eGFR | -0.11 | <0.001 | -0.13 | <0.001 |
Abbreviations: Sβ, standardized beta; BMI, body mass index; WC, waist circumference; RFM, relative fat mass; HDL-C, high-density lipoprotein cholesterol; MI, myocardial infarction; NT-proBNP, N-terminal pro-B-type natriuretic peptide; CRP, C-reactive protein; UAE, 24h urinary albumin excretion; eGFR, estimated glomerular filtration rate. Although all anthropometric measures (weight, BMI, WC, and RFM) were significantly associated with BMP10 in age- and sex-adjusted models, RFM, which showed the strongest association, was retained in the multivariable model to avoid multicollinearity.
For comparison, multivariable associations of NT-proBNP with clinical characteristics are presented in Supplemental Table S3, and clinical correlates of NT-proBNP versus BMP10 are illustrated in Supplemental Fig. S2.
Associations of plasma BMP10 with incident AF
During a median (P25-P75) follow-up of 6.4 (5.9-6.9) years, 154 participants (2.6%) developed AF, corresponding to an incidence rate of 4.2 (95%CI: 3.6-5.0) new AF events per 1000 person-years. Participant characteristics according to incident AF status are provided in Supplemental Table S4. The mean (SD) plasma BMP10 level in participants who developed AF was not significantly higher than in those who did not develop AF [2.3 (0.42) vs 2.2 (0.43) ng/mL; P=0.059]. When stratified by sex, men who developed AF had significantly higher BMP10 levels than those who did not develop AF [2.20 (0.40) vs 2.04 (0.38); P<0.001], whereas women who developed AF did not have significantly higher BMP 10 levels compared with those who did not develop AF [2.43 (0.43) vs 2.31 (0.44); P=0.076].
The cumulative incidence of AF across sex-pooled BMP10 quartiles is shown in Supplemental Fig. S3. Overall, individuals in the lowest quartile had the lowest risk, whereas those in the highest quartile had the greatest risk but the risk of AF over time overlapped across quartiles. Incidence rates of AF per 1000-person years were 2.5 (1.7-3.8) in the first quartile; 4.0 (2.9-5.5) in the second quartile; 5.3 (4.0-7.0) in the third quartile, and 5.2 (3.9-7.0) in the fourth quartile.
In unadjusted Cox regression models, higher BMP10 levels were significantly associated with an increased risk of incident AF (HR per 1-standard deviation increase: 1.74; 95% CI: 1.38-2.21). This association remained statistically significant after adjusting for age and sex (HR: 1.52; 95%CI: 1.18-1.96) and persisted with further adjustment for height, smoking, weight, and history of type-2 diabetes, hypertension, MI or stroke, and HF (HR: 1.58; 95%CI: 1.23-2.04). In all models, the strength of the association attenuated over time, indicating that BMP10 was more strongly associated with incident AF in the short term (Table 3). In sensitivity analyses, Model 3 was further adjusted separately for eGFR, CRP, and NT-proBNP, with results remaining materially unchanged (Supplemental Table S5).
Table 3. Associations of BMP10 with Incident Atrial Fibrillation.
| Hazard Ratio (95% CI) | P-value | |
|---|---|---|
| Model 1 | 1.74 (1.38, 2.21) | <0.001 |
| Model 2 | 1.52 (1.18, 1.96) | 0.001 |
| Model 3 | 1.58 (1.23, 2.04) | <0.001 |
Model 1 was unadjusted. Model 2 was adjusted for age and sex. Model 3 was adjusted for age and sex, and further adjusted for components of the CHARGE-AF model: height, smoking, weight, history of type-2 diabetes, history of hypertension, history of myocardial infarction or stroke, and history of heart failure. Due to a significant interaction with time, BMP10 and sex were treated as time-varying covariates in these analyses. Hazard ratios are presented per 1SD increase in BMP10. Time interaction variable (BMP10 × time interaction) was used to account for violoation of the proportionality hazards assumption, and the hazard ratio for this interaction term was 0.91 (0.85, 0.98) in Model 1, 0.91 (0.84, 0.97) in Model 2, and 0.91 (0.85, 0.98) in Model 3, indicating the strength of association attenuated over time.
In exploratory subgroup analyses, we found a significant interaction between BMP10 levels and systemic inflammatory status – with regard to AF incidence. No significant interactions were observed for age, sex, obesity, prevalent CVD, or renal dysfunction (Supplemental Table S6).
Finally, we evaluated whether BMP10 improved risk prediction for incident AF. Adding BMP10 to a clinical model with age, sex and CHARGE-AF components modestly improved model fit (Supplemental Table S7), but provided negligible improvement when NT-proBNP was additionally included (Supplemental Table S8).
Discussion
In this study of 5884 community-dwelling adults, we report the first population-based assessment of circulating BMP10, describing its physiological and pathophysiological correlates as well as its association with incident AF.
In cross-sectional analyses, BMP10 displayed some similarities to natriuretic peptides, with higher levels in women (ie, lower levels in men) and lower levels with increasing fat mass (Supplemental Fig. S2) (20, 21). For natriuretic peptides, sex-related differences in plasma levels have been attributed, in part, to testosterone, as suppression of testosterone production in men led to increases in NT-proBNP, whereas testosterone replacement restored NT-proBNP toward baseline levels (22). Whether similar hormonal regulation explains sex-related differences in BMP10 levels remains unclear and warrants further investigation. Similarly, the mechanisms underlying adiposity-related lowering of BMP10, and whether these differ by sex, also warrant further study.
Next, in contrast to natriuretic peptides, which are strongly associated with prevalent CVD, BMP10 showed no such association, suggesting it may not reflect generalized cardiovascular burden. Moreover, whereas natriuretic peptides are positively related to systemic inflammation (23), BMP10 showed a strong inverse association with CRP (Supplemental Fig. S2); this finding is partly supported by experimental data showing that extracellular BMP10 reduces chemokine (C-C motif) ligand 2 (CCL2) production, thereby limiting inflammatory cell recruitment to the vessel wall (24), and by a study of patients with idiopathic pulmonary arterial hypertension reporting an inverse correlation between circulating levels of BMP10 and CRP (25). Finally, while natriuretic peptides are known to be inversely associated with total cholesterol (26), BMP10 showed a positive association.
In longitudinal analyses, higher BMP10 levels were independently associated with an increased risk of incident AF (Table 3). Although the predictive value of BMP10 in the general population appears modest compared with NT-proBNP (Supplemental Tables S7 and S8) (27), the consistency of its association with incident AF suggests its potential biological relevance. Specifically, elevated BMP10 may represent an adaptive or compensatory response to atrial stress, a hypothesis supported by experimental studies in which BMP10 overexpression in mice reduced cardiomyocyte apoptosis and fibrosis following prolonged isoproterenol exposure (28). By contrast, a recent preprint reported that sustained exposure of engineered ventricular heart tissue to very high concentrations of recombinant human BMP10 impaired its contractile function (6); these observations, however, warrant cautious interpretation, as exposure to supraphysiological BMP10 levels may not reflect endogenous physiological conditions, and the biological activity of recombinant BMP10 could differ substantially from that of native BMP10 (29).
In summary, our study presents novel data on circulating BMP10, its physiological and pathophysiological correlates, and its association with incident AF in a community-based population. The inverse relationship of BMP10 with inflammation, together with its prospective association with incident AF, raises the possibility that elevated BMP10 may reflect a cardioprotective response to atrial stress. Taken together with existing preclinical data showing both protective (28) and potentially harmful effects (6) of BMP10 (depending on context and concentration), our findings underscore the need for further mechanistic research. Specifically, future studies should investigate how BMP10 interacts with TGFβ signalling and inflammatory pathways (30) to determine whether it plays an adaptive or maladaptive role in the development and progression of AF. Additionally, although BMP10 did not emerge as a strong predictor of incident AF in the current analysis, future studies should explore its potential to improve short-term AF risk prediction in selected subpopulations, such as pregnant women (31), where natriuretic peptide testing may be less informative.
Strengths and limitations
This study is the first to examine the clinical correlates of BMP10 and its association with incident AF in community-dwelling adults. A key strength of our study is the measurement of circulating BMP10 using a highly specific electrochemiluminescence immunoassay, ensuring reliable quantification across the population. Other strengths include comprehensive clinical phenotyping, a balanced sex distribution, head-to-head comparison of clinical correlates of BMP10 with NT-proBNP, and adjudicated AF outcomes (15). Nevertheless, several limitations warrant consideration. First, the PREVEND study, by design, enrolled a higher proportion of individuals with mildly elevated UAE. However, this is unlikely to have biased our findings, as prior work has shown that results from the PREVEND cohort are largely consistent with those observed in broader community-based cohorts (32). Second, although the associations of NT-proBNP with clinical risk factors were consistent with previous literature (33), independent external validation of our findings related to BMP10 is warranted; specifically, given the predominantly Caucasian study population, their generalizability should be confirmed in more ethnically diverse cohorts. Third, BMP9 levels were not measured in parallel with BMP10, which might have provided additional biological context regarding their complementary roles within the BMP subgroup of the TGFβ superfamily (34). Fourth, the observational nature of the study precludes causal inference, and the possibility of residual confounding cannot be excluded. Finally, AF cases in PREVEND were identified through screening ECGs at study visits and from hospital records. While this approach captures clinically relevant AF, it may miss subclinical or brief AF episodes, such as atrial high-rate episodes (AHREs). This limitation provides a rationale for future studies to examine whether BMP10 could help identify recent or near-term AF episodes missed by a single ECG screening.
Conclusions
In community-dwelling adults, BMP10 levels were higher in women, positively associated with HDL-C, and inversely associated with fat mass, inflammation, and kidney function. Although not linked to prevalent cardiovascular disease, higher BMP10 levels were independently associated with incident AF. These findings provide novel, population-based insights into BMP10 biology and lay the groundwork for future mechanistic studies in atrial physiology and disease.
Supplementary Material
Supplemental Material: Supplemental Figs.S1-S3 and Supplemental Tables S1-S8. https://doi.org/10.5281/zenodo.17245912
New & Newsworthy.
This is the first population-based study of circulating bone morphogenetic protein 10 (BMP10), a heart-specific protein secreted by atrial cardiomyocytes. In over 5800 community-dwelling adults, BMP10 showed distinct physiological and pathophysiological correlates, including higher levels in women, an inverse association with fat mass and inflammation (CRP), and a positive association with incident atrial fibrillation. These findings represent an important first step toward understanding the role of BMP10 in atrial biology and disease.
Acknowledgements
Dr. de Boer and Suthahar are supported by the Netherlands Heart Foundation (Nederlandse Hartstichting) through grants 2020B005 (DOUBLE DOSE) and 01-003-2022-0358 (CARMA), by the Netherlands Organization for Scientific Research (Nederlandse Organisatie voor Wetenschappelijk Onderzoek), co-funded by ERA4Health through the CARDINNOV 2023 call, as part of the EnerLIGHT project (Grant Agreement N. 101095426 of the EU Horizon Europe Research and Innovation Programme) and by the European Research Council (ERC CoG 818715; SECRETE-HF). Dr. Rienstra is supported by unrestricted research grants from the Dutch Heart Foundation and is conducted in collaboration with and supported by the Dutch CardioVascular Alliance, 01-002-2022-0118 EmbRACE, and from ZonMW and the Dutch Heart Foundation; DECISION project 848090001, and by the Netherlands Cardiovascular Research Initiative: an initiative with support of the Dutch Heart Foundation; RACE V (CVON 2014–9), RED-CVD (CVON2017-11), and by Top Sector Life Sciences & Health to the Dutch Heart Foundation (PPP Allowance; CVON-AI (2018B017)), and by the European Union’s Horizon 2020 research and innovation programme under grant agreement; EHRA-PATHS (945260). PK is partially supported by European Union MAESTRIA (grant agreement 965286), British Heart Foundation (AA/18/2/34218), German Center for Cardiovascular Research supported by the German Ministry of Education and Research (DZHK, grant numbers DZHK FKZ 81X2800182, 81Z0710116, and 81Z0710110), German Research Foundation (Ki 509167694), Dutch Heart Foundation (DHF), the Accelerating Clinical Trials funding stream in Canada, and the Else-Kröner-Fresenius Foundation. LF is supported by EU Horizon 2020 MAESTRIA (grant agreement 965286). We thank Dr. Peter Kastner (Roche Diagnostics) for his assistance with assay-specific details, and to Roche Diagnostics for their contribution to BMP10 measurements.
Footnotes
Ethical approval for the PREVEND study was obtained from the local medical ethics committee of the University Medical Center Groningen (MEC96/01/022); all participants signed informed consent and the study was conducted in accordance with the declaration of Helsinki.
Author contributions: N.S. conceptualized the study, developed the research question, performed main data analysis, prepared the table, interpreted the data and wrote the paper. M.R. and S.C.Y. assisted with the development of the research question and methodology. A.A.F.d.V., L.F., W.P., R.T.G., S.J.L.B. and K.D. were involved in critical review of the project proposal and methodology. R.T.G., S.J.L.B. and K.D. were involved in approval on the project proposal on behalf of the PREVEND board. R.A.d.B was involved in conceptualization, supervision, resources and funding acquisition. All authors interpreted data, reviewed and critically revised the manuscript draft, and approved the final manuscript.
Competing interests: N.S. is listed as an inventor on a pending patent application related to BMP10 (WO2025061752), titled “Circulating BMP10 in the assessment of congestion and pulmonary hypertension.” S.C.Y. has received honoraria (speaker or consultancy fees) from Boston Scientific, Medtronic, Biotronik, Acutus Medical and Sanofi. In addition, he has received research grants from Medtronic, Biotronik, Boston Scientific and Biosense Webster. K.D. reports speaker/consultancy fees to his institution by Abbott, Astra Zeneca, Boehringer Ingelheim, Novartis, Echosense, Fire1 and Reprieve. L.F. has received institutional research grants and non-financial support from European Union (EU Horizon 2020 MAESTRIA (grant agreement number 965286)), DFG, DZHK, British Heart Foundation, Medical Research Council (UK), NIHR, and several biomedical companies. L.F. is listed as inventor of two patents held by the academic institution (Atrial Fibrillation Therapy WO 2015140571, Markers for Atrial Fibrillation WO 2016012783). P.K. received research support for basic, translational, and clinical research projects from German Research Foundation (DFG), European Union, British Heart Foundation, Leducq Foundation, Else-Kröner-Fresenius Foundation, Dutch Heart Foundation (DHF), the Accelerating Clinical Trials funding stream in Canada, Medical Research Council (UK), and German Center for Cardiovascular Research, from several drug and device companies active in atrial fibrillation, and has received honoraria from several such companies in the past, but not in the last five years. P.K. is listed as inventor on two issued patents held by University of Hamburg (Atrial Fibrillation Therapy WO 2015140571, Markers for Atrial Fibrillation WO 2016012783). M.R. reports receiving consultancy fees from Bayer (OCEANIC-AF national PI), InCarda Therapeutics (RESTORE-SR national PI) and Novartis paid to the institution and speaker fee from Daiichi-Sankyo and Pfizer also paid to the institution. R.A.d.B. is listed as an inventor on a pending patent application related to BMP10 (WO2025061752), titled “Circulating BMP10 in the assessment of congestion and pulmonary hypertension.” Outside the current work, R.A.d.B. has had speaker engagements with and/or received fees from and/or served on an advisory board for Abbott, AstraZeneca, Bristol Myers Squibb, NovoNordisk, Roche, and Zoll. R.A.d.B. received travel support from Abbott and NovoNordisk. The institution where R.A.d.B. is employed has received research grants and/or fees from Alnylam, AstraZeneca, Abbott, Bristol-Myers Squibb, NovoNordisk and Roche. The remaining authors do not have anything to disclose.
Data availability
The dataset analysed during the current study are available in the PREVEND repository https://umcgresearch.org/w/prevend
Code availability
Codes of Stata version 14 (Stata Corp., College Station, TX, USA) for statistical analysis of this study are available upon request from Navin Suthahar.
References
- 1.Lip GYH, Fauchier L, Freedman SB, Van Gelder I, Natale A, Gianni C, Nattel S, Potpara T, Rienstra M, Tse H-F, Lane DA. Atrial fibrillation. Nat Rev Dis Primers. 2016;2:16016. doi: 10.1038/nrdp.2016.16. [DOI] [PubMed] [Google Scholar]
- 2.Iwasaki Y, Nishida K, Kato T, Nattel S. Atrial fibrillation pathophysiology: implications for management. Circulation. 2011;124:2264–2274. doi: 10.1161/CIRCULATIONAHA.111.019893. [DOI] [PubMed] [Google Scholar]
- 3.Chang S-H, Yeh Y-H, Lee J-L, Hsu Y-J, Kuo C-T, Chen W-J. Transforming growth factor-β-mediated CD44/STAT3 signaling contributes to the development of atrial fibrosis and fibrillation. Basic Res Cardiol. 2017;112:58. doi: 10.1007/s00395-017-0647-9. [DOI] [PubMed] [Google Scholar]
- 4.Neuhaus H, Rosen V, Thies RS. Heart specific expression of mouse BMP-10 a novel member of the TGF-beta superfamily. Mech Dev. 1999;80:181–184. doi: 10.1016/s0925-4773(98)00221-4. [DOI] [PubMed] [Google Scholar]
- 5.E Molina C, Sommerfeld L, Zeller T, Obergassel J, Wieboldt H, Conradi L, Reichenspurner H, Nikolaev VO, Kirchhof P, Fabritz L. From cells to circulating biomarker: BMP10 is a myocyte-secreted peptide with potential to detect atrial fibrillation. European Heart Journal. 2023;44:ehad655.3140. doi: 10.1093/eurheartj/ehad655.3140. [DOI] [Google Scholar]
- 6.Sommerfeld L, Schrapers J, Müller K, Bravo L, Siebels B, Vermeer-Stoter A, Pan B, Höppner G, O’Shea C, Ridder J, Wieboldt H, et al. High rate triggers increased atrial release of BMP10, a biomarker for atrial fibrillation and stroke, which affects ventricular cardiomyocytes. 2025 doi: 10.1161/CIRCEP.125.013834. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kahr PC, Piccini I, Fabritz L, Greber B, Schöler H, Scheld HH, Hoffmeier A, Brown NA, Kirchhof P. Systematic Analysis of Gene Expression Differences between Left and Right Atria in Different Mouse Strains and in Human Atrial Tissue. PLoS ONE. 2011;6:e26389. doi: 10.1371/journal.pone.0026389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Reyat JS, Chua W, Cardoso VR, Witten A, Kastner PM, Kabir SN, Sinner MF, Wesselink R, Holmes AP, Pavlovic D, Stoll M, et al. Reduced left atrial cardiomyocyte PITX2 and elevated circulating BMP10 predict atrial fibrillation after ablation. JCI Insight. 2020;5:e139179. doi: 10.1172/jci.insight.139179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hennings E, Aeschbacher S, Coslovsky M, Paladini RE, Meyre PB, Voellmin G, Blum L, Kastner P, Ziegler A, Conen D, Zuern CS, et al. Association of bone morphogenetic protein 10 and recurrent atrial fibrillation after catheter ablation. Europace. 2023;25:euad149. doi: 10.1093/europace/euad149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hijazi Z, Benz AP, Lindbäck J, Alexander JH, Connolly SJ, Eikelboom JW, Granger CB, Kastner P, Lopes RD, Ziegler A, Oldgren J, et al. Bone morphogenetic protein 10: a novel risk marker of ischaemic stroke in patients with atrial fibrillation. European Heart Journal. 2023;44:208–218. doi: 10.1093/eurheartj/ehac632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hennings E, Blum S, Aeschbacher S, Coslovsky M, Knecht S, Eken C, Lischer M, Paladini RE, Krisai P, Reichlin T, Rodondi N, et al. Bone Morphogenetic Protein 10-A Novel Biomarker to Predict Adverse Outcomes in Patients With Atrial Fibrillation. J Am Heart Assoc. 2023;12:e028255. doi: 10.1161/JAHA.122.028255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gkarmiris KI, Lindbäck J, Alexander JH, Granger CB, Kastner P, Lopes RD, Ziegler A, Oldgren J, Siegbahn A, Wallentin L, Hijazi Z. Repeated Measurement of the Novel Atrial Biomarker BMP10 (Bone Morphogenetic Protein 10) Refines Risk Stratification in Anticoagulated Patients With Atrial Fibrillation: Insights From the ARISTOTLE Trial. JAHA. 2024;13:e033720. doi: 10.1161/JAHA.123.033720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ceelen DCH, Bracun V, Van Essen BJ, Voors AA, De Boer RA, Ter Maaten JM, Masson S, Kastner P, Lang CC, Suthahar N. Circulating bone morphogenetic protein 10 as a novel marker of atrial stress and remodelling in heart failure. Heart. 2025;111:172–179. doi: 10.1136/heartjnl-2024-324486. [DOI] [PubMed] [Google Scholar]
- 14.Hillege HL, Fidler V, Diercks GFH, van Gilst WH, de Zeeuw D, van Veldhuisen DJ, Gans ROB, Janssen WMT, Grobbee DE, de Jong PE. Urinary Albumin Excretion Predicts Cardiovascular and Noncardiovascular Mortality in General Population. Circulation. 2002;106:1777–1782. doi: 10.1161/01.cir.0000031732.78052.81. [DOI] [PubMed] [Google Scholar]
- 15.Vermond RA, Geelhoed B, Verweij N, Tieleman RG, Van der Harst P, Hillege HL, Van Gilst WH, Van Gelder IC, Rienstra M. Incidence of Atrial Fibrillation and Relationship With Cardiovascular Events, Heart Failure, and Mortality: A Community-Based Study From the Netherlands. J Am Coll Cardiol. 2015;66:1000–1007. doi: 10.1016/j.jacc.2015.06.1314. [DOI] [PubMed] [Google Scholar]
- 16.Al-Mubarak AA, Grote Beverborg N, Zwartkruis V, van Deutekom C, de Borst MH, Gansevoort RT, Bakker SJL, Touw DJ, de Boer RA, van der Meer P, Rienstra M, et al. Micronutrient deficiencies and new-onset atrial fibrillation in a community-based cohort: data from PREVEND. Clin Res Cardiol. 2025;114:41–52. doi: 10.1007/s00392-023-02276-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Suthahar N, Gansevoort RT, Kok TF, El-Qendouci M, Petersen TB, Bakker SJL, Boersma E, Damman K, Kardys I, Rienstra M, de Boer RA. The senescence-inducing factor IGFBP7 and risk of atrial fibrillation: findings from the PREVEND study. NPJ Cardiovasc Health. 2025;2:42. doi: 10.1038/s44325-025-00079-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Suthahar N, Bergman RN, de Boer RA. Replacing body mass index with relative fat mass to accurately estimate adiposity. Nat Rev Endocrinol. 2025;21:393–394. doi: 10.1038/s41574-025-01120-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Alonso A, Krijthe BP, Aspelund T, Stepas KA, Pencina MJ, Moser CB, Sinner MF, Sotoodehnia N, Fontes JD, Janssens ACJW, Kronmal RA, et al. Simple risk model predicts incidence of atrial fibrillation in a racially and geographically diverse population: the CHARGE-AF consortium. J Am Heart Assoc. 2013;2:e000102. doi: 10.1161/JAHA.112.000102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Clark BA, Elahi D, Epstein FH. The influence of gender, age, and the menstrual cycle on plasma atrial natriuretic peptide. J Clin Endocrinol Metab. 1990;70:349–352. doi: 10.1210/jcem-70-2-349. [DOI] [PubMed] [Google Scholar]
- 21.Suthahar N, Meems LMG, Ho JE, Boer RA. Sex-related differences in contemporary biomarkers for heart failure: a review. Eur J Heart Fail. 2020;22:775–788. doi: 10.1002/ejhf.1771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Bachmann KN, Huang S, Lee H, Dichtel LE, Gupta DK, Burnett JC, Miller KK, Wang TJ, Finkelstein JS. Effect of Testosterone on Natriuretic Peptide Levels. J Am Coll Cardiol. 2019;73:1288–1296. doi: 10.1016/j.jacc.2018.12.062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fish-Trotter H, Ferguson JF, Patel N, Arora P, Allen NB, Bachmann KN, Daniels LB, Reilly MP, Lima JAC, Wang TJ, Gupta DK. Inflammation and Circulating Natriuretic Peptide Levels. Circ Heart Fail. 2020;13:e006570. doi: 10.1161/CIRCHEARTFAILURE.119.006570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Upton PD, Park JES, De Souza PM, Davies RJ, Griffiths MJD, Wort SJ, Morrell NW. Endothelial protective factors BMP9 and BMP10 inhibit CCL2 release by human vascular endothelial cells. J Cell Sci. 2020;133:jcs239715. doi: 10.1242/jcs.239715. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hodgson J, Swietlik EM, Salmon RM, Hadinnapola C, Nikolic I, Wharton J, Guo J, Liley J, Haimel M, Bleda M, Southgate L, et al. Characterization of GDF2 Mutations and Levels of BMP9 and BMP10 in Pulmonary Arterial Hypertension. Am J Respir Crit Care Med. 2020;201:575–585. doi: 10.1164/rccm.201906-1141OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Spannella F, Giulietti F, Bordicchia M, Burnett JC, Sarzani R. Association Between Cardiac Natriuretic Peptides and Lipid Profile: a Systematic Review and Meta-Analysis. Sci Rep. 2019;9:19178. doi: 10.1038/s41598-019-55680-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wang W, Zhou T, Li J, Yuan C, Li C, Chen S, Shen C, Gu D, Lu X, Liu F. Association between NT-proBNP levels and risk of atrial fibrillation: a systematic review and meta-analysis of cohort studies. Heart. 2025;111:109–116. doi: 10.1136/heartjnl-2024-324685. [DOI] [PubMed] [Google Scholar]
- 28.Qu X, Liu Y, Cao D, Chen J, Liu Z, Ji H, Chen Y, Zhang W, Zhu P, Xiao D, Li X, et al. BMP10 preserves cardiac function through its dual activation of SMAD-mediated and STAT3-mediated pathways. J Biol Chem. 2019;294:19877–19888. doi: 10.1074/jbc.RA119.010943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Spanou CES, Yang C, Godwin ARF, Morosky S, Anbalagan A, Lütke S, Mörgelin M, Marcous F, Aziz U, Wohl AP, Jabeen I, et al. Prodomain processing controls BMP-10 bioactivity and targeting to fibrillin-1 in latent conformation. FASEB J. 2025;39:e70373. doi: 10.1096/fj.202401694R. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Brennan E, Kantharidis P, Cooper ME, Godson C. Pro-resolving lipid mediators: regulators of inflammation, metabolism and kidney function. Nat Rev Nephrol. 2021;17:725–739. doi: 10.1038/s41581-021-00454-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Minhas AS, Rooney MR, Fang M, Zhang S, Ndumele CE, Tang O, Schulman SP, Michos ED, McEvoy JW, Echouffo-Tcheugui J, Christenson R, et al. Prevalence and Correlates of Elevated NT-proBNP in Pregnant Women in the General U.S. Population. JACC Adv. 2023;2:100265. doi: 10.1016/j.jacadv.2023.100265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.de Boer RA, Nayor M, deFilippi CR, Enserro D, Bhambhani V, Kizer JR, Blaha MJ, Brouwers FP, Cushman M, Lima JAC, Bahrami H, et al. Association of Cardiovascular Biomarkers With Incident Heart Failure With Preserved and Reduced Ejection Fraction. JAMA Cardiol. 2018;3:215. doi: 10.1001/jamacardio.2017.4987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Welsh P, Campbell RT, Mooney L, Kimenai DM, Hayward C, Campbell A, Porteous D, Mills NL, Lang NN, Petrie MC, Januzzi JL, et al. Reference Ranges for NT-proBNP (N-Terminal Pro-B-Type Natriuretic Peptide) and Risk Factors for Higher NT-proBNP Concentrations in a Large General Population Cohort. Circ: Heart Failure. 2022;15 doi: 10.1161/CIRCHEARTFAILURE.121.009427. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Desroches-Castan A, Tillet E, Bouvard C, Bailly S. BMP9 and BMP10: Two close vascular quiescence partners that stand out. Dev Dyn. 2022;251:178–197. doi: 10.1002/dvdy.395. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The dataset analysed during the current study are available in the PREVEND repository https://umcgresearch.org/w/prevend
Codes of Stata version 14 (Stata Corp., College Station, TX, USA) for statistical analysis of this study are available upon request from Navin Suthahar.