Circulating growth factors and cardiac remodeling in the community: The Framingham Heart Study

Abstract

Background and aims:

Cardiac and vascular growth factors (GF) may influence myocardial remodeling through cardiac growth and angiogenic effects. We hypothesized that concentrations of circulating GF are associated with cardiac remodeling traits.

Methods:

We related blood concentrations of vascular endothelial GF (VEGF), VEGFR-1 (sFlt1), angiopoietin 2 (Ang-2), soluble angiopoietin type-2 receptor (sTie2), hepatocyte GF (HGF), insulin-like GF (IGF)-1, IGF binding protein (IGFBP)-3, and growth differentiation factor-15 (GDF-15) to echocardiographic traits in 3151 Framingham Study participants (mean age 40 years, 55% women). We evaluated the following measures: left ventricular (LV) mass index (LVMi), LV ejection fraction (LVEF), global longitudinal strain (GLS), mitral E/e’, and aortic root diameter (AoR). All biomarker values were sex-standardized.

Results:

In multivariable-adjusted analyses, higher GDF-15 concentrations were associated with higher log-LVMi (β=0.009 per SD, P =0.01). Similarly, sTie2 concentrations were positively associated with log-E/e’ (β=0.011 per SD, P =0.04). IGF-1 and Ang-2 concentrations were positively and negatively associated with GLS, respectively (β_IGF-1 =0.16 per SD and β_Ang-2 =−0.15 per SD, both P <0.05), whereas higher sFlt1 and Ang-2 levels were associated with smaller log-AoR (β_sFlt1 =−0.004 per SD and β _Ang-2 =−0.005 per SD, respectively; P <0.05).

Conclusion:

In our large community-based sample, we observed patterns of associations between several circulating vascular GF and cardiac remodeling indices that are consistent with the known biological effects of these pro- and anti-angiogenic factors on the myocardium and conduit arteries. Additional studies are warranted to replicate our findings and assess their prognostic significance.

Introduction

Cardiac remodeling is the result of hemodynamic and neurohormonal stressors,¹ leading to physiological reversible responses of myocardial adaptation or pathological and progressive myocardial damage. Adverse ventricular remodeling may result in subsequent decline of the left ventricular ejection fraction (LVEF) leading to incident heart failure (HF)^2,3,4,5 In spite of the important advances made in understanding pathophysiological mechanisms underlying HF, many gaps in the knowledge of this condition remain. Therefore, a central focus of research is the role of biologically active molecules on the myocardium and its remodeling.²

An important mechanism underlying cardiac remodeling is impaired angiogenesis that leads to cardiomyocyte loss and myocardial contractile dysfunction.^4,6 Vascular endothelial growth factor (VEGF) is a molecule with angiogenic effects.⁶ It binds to its transmembrane tyrosine kinase receptor VEGFR-1 (sFlt1),⁷ which is expressed on cardiomyocytes, endothelial cells and macrophages, and can be found in the circulation as soluble VEGFR-1. VEGFR-1, an anti-VEGF compound at embryogenesis, is associated with decreased angiogenesis, and promotes inflammation and atherosclerosis,⁸ and stimulates endothelial dysfunction.⁹ Likewise, angiopoietin-2 (Ang-2) has known angiogenic effects acting predominantly on vascular endothelial cells in the presence of VEGF.⁶ Ang-2 binds to a tyrosine kinase receptor Tie-2 leading to an increased vascular permeability and vascular destabilization.¹⁰ Furthermore, hepatocyte growth factor (HGF), which is detected in circulation after endothelial injury, exerts angiogenic effects on endothelial cells and protective effects on cardiomyocytes contributing to the process of myocardial remodeling.^{11, 12} Finally, insulin growth factor (IGF)-1 modulates myocardial growth as well as cardiac contractility.^13,14 IGF-1 binds to several IGF binding proteins, the major one being IGFBP-3, which regulates the bioactivity of IGF-1.¹⁵

The selected cardiac and vascular biomarkers have been related to risk of cardiovascular disease in several clinical and epidemiological studies.^{7,16,17,18,19,20} and may be in the pathway between CVD risk factors and cardiac remodeling. We hypothesized that circulating growth factor concentrations are associated with cardiac and aortic root remodeling consistent with their known biological effects in experimental settings. Accordingly, we related concentrations of the aforementioned circulating growth factors and their soluble receptors to several echocardiographic indices of cardiac and aortic root remodeling.

Materials and methods

Study Sample

The study design of the Third Generation Framingham cohort has been previously described.²¹ In the present investigation, we included Third Generation participants who attended their first examination cycle (2002–2005) when routine echocardiography was performed. Of 4095 eligible participants, 944 participants were excluded for the following reasons: history of cardiovascular disease (n=64), body mass index (BMI) <18.5 kg/m² (n=56), triglycerides >200 mg/100mL (n=283), and inadequate echocardiographic data (n=541), resulting in a relatively healthy sample of 3151 participants eligible for the present investigation. All participants provided written informed consent. The research protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Institutional Review Board of the Boston University Medical Center.

Biomarker measurements

Blood was drawn in the morning on participants after a 12-hour overnight fast and was then immediately centrifuged and stored at −80° C until biomarkers were assayed. For the present investigation, we used data for concentrations of the following blood biomarkers: VEGF, VEGFR-1 (sFlt1), Ang-2, soluble Tie-2 (sTie2), HGF, IGF-1, IGFBP-3, and GDF-15.

Serum concentrations of VEGF, sFlt1, Ang-2, sTie2, HGF, IGF-1 and IGFBP-3 were measured using commercial assay kits (R&D Systems Inc; Minneapolis, MN). The inter-assay coefficients of variation were 2.1% for VEGF, 6.4% for sFlt1, 5.7% for Ang-2, 3.2% for sTie2, 1.6% for HGF, and 9.1% for IGFBP3. GDF-15 was measured using a Roche Elecsys analyzer (Roche Diagnostics, Mannheim, Germany) as described previously.²² The analytical measuring range for the Roche GDF-15 assay is 400–20,000 ng/L, and has a detection limit below 200 ng/L.

Echocardiographic measurements

Standardized two-dimensional echocardiography with color flow and tissue Doppler imaging was performed by experienced sonographers using a Hewlett-Packard Sonos 5500 Ultrasound machine (Phillips Healthcare, Andover, MA). Experienced readers analyzed the digitized echocardiographic images (DICOM format) using an offline system (Digiview System Software ver. 3.7.9.3, Digisonic Inc., Houston, Texas, USA). Linear dimensions, displacement, and tissue velocities were measured. Speckle tracking echocardiographic strain analysis of the left ventricle (LV) was also performed using an offline image analysis program to measure myocardial deformation based on 2-dimensional images, (2D Cardiac Performance Analysis v1.1, TomTec Imaging Systems, Unterschleissheim, Germany).²³

Primary Echocardiographic indices

For the current investigation, we used the following primary echocardiographic traits capturing select aspects of cardiac structure and function: left ventricular (LV) mass index (LVMi), LV ejection fraction (LVEF), global longitudinal strain (GLS), E/e’, and aortic root diameter (AoR). LV mass (LVM) was measured using linear methods according to American Society of Echocardiography guidelines, taking into consideration the LV end-diastolic diameter (LVDD), and the thickness of the LV walls: end-diastolic LV septal wall thickness (SWT) and posterior wall thickness (PWT).²⁴ LVM was then indexed to body surface area (LVMi). LVEF was derived from the LV volumes calculated from M-mode linear measurements. GLS, the myocardial shortening in the long axis plane, was calculated as the average peak longitudinal strain measured in 12 regions in the apical two- and four chamber views.²⁵ Higher (less negative) signed values of GLS indicate worse cardiac function. Early transmittal velocity (E wave) was measured by pulse-wave Doppler from the apical 4-chamber view.²⁶ Peak lateral mitral annular relaxation velocities (e’) were measured by tissue Doppler imaging.²⁷ The E/e’ ratio was used as an indicator of LV diastolic function. AoR was measured using the leading-edge technique as recommended by the American Society of Echocardiography.²⁸

Secondary echocardiographic indices

The following secondary echocardiographic indices were also evaluated: mitral annular plane systolic excursion (MAPSE, a measure of long axis function), global circumferential strain (GCS), LV longitudinal segment synchrony (LSS), LVDD and LV wall thickness (LVWT).²⁹ MAPSE was calculated at the lateral side of the mitral valve annulus in the apical four-chamber view by measuring the systolic excursion of the annulus from its lowest point at end-diastole to its highest point at the time of aortic valve closure, as recommended.²⁹ GCS, an indicator of the myocardial shortening in the short axis, and LSS,³⁰ calculated as the standard deviation (SD) of time-to-peak systolic longitudinal strains,³¹ were generated from speckle-tracking echocardiography.

Covariates

For the present investigation, we used the following covariates: age, sex, height, weight, smoking status, systolic blood pressure, use of hypertensive medications, heart rate, diabetes status (defined as fasting glucose ≥126 mg/dL or treatment with hypoglycemic medications), serum total cholesterol / high density-lipoprotein ratio, and serum creatinine.

Statistical Analysis

All biomarker values were sex-standardized. LVMi, LVEF, E/e’ and AoR values were natural logarithmically transformed to normalize their skewed distributions. Since GLS values are on the negative scale, they were not logarithmically transformed. We estimated age- and sex-adjusted Spearman correlations among the non-standardized growth factors.

We used generalized linear models (GLM) to relate the biomarkers (independent variables, separate model for each) to the echocardiographic indices (dependent variables, separate model for each). We adjusted all models for age and sex first, and then additionally adjusted for height, weight, current smoking, systolic blood pressure, current use of hypertensive medications, heart rate, diabetes, total cholesterol/HDL ratio, and serum creatinine (multivariable-adjusted model). In addition, we accounted for relatedness among participants, and also accounted for multiple comparisons by estimating false discovery rates (FDR) for all associations evaluated.³²

To visualize the distribution of the echocardiographic traits in relation to concentrations of biomarkers that were significantly associated, we created least square means (LSM) plots for each echocardiographic trait by tertile of the associated biomarker. LSMs were adjusted for the same covariates we used for the GLM models. Furthermore, we also tested for nonlinearity of the associations between biomarkers and echocardiographic indices (only for biomarkers significantly associated with echocardiographic indices) by plotting multivariable-adjusted splines, truncating biomarker values at the 95^th percentile with knots placed at 25^th, 50^th, and 75^th percentile values of the biomarkers. Finally, we tested for effect modification of the associations between the growth factors and the echocardiographic traits by sex.

Results

Our analytical sample consisted of 3151 participants, with a mean age of 40 ± 9 years (age range: 19–72 years), 55% women, relatively obese, with a low prevalence of smoking and hypertension, and free of cardiovascular disease (Table 1).

Table 1.

Characteristics of Study Sample

	Men (n=1413)	Women (n=1738)
Cardiovascular Risk Factors
Age, years (SD)	40 (9)	40 ( 9)
Height, cm (SD)	178 ( 6)	164 (6)
Weight, kg (SD)	87 (15)	69 (15)
Body mass index, kg/m2 (SD)	27.3 (4.2)	25.5 (5.3)
Smoking, %	15	15
Systolic Blood Pressure, mm Hg (SD)	120 (12)	112 (13)
Diastolic Blood Pressure, mm Hg (SD)	77 (9)	72 (9)
Hypertension, %	17	10
Hypertension treatment, %	7	6
Heart Rate, bpm (SD)	58 (9)	61 (9)
Diabetes, %	2.3	1.3
Serum creatinine, mg/100ml (SD)	0.9 (0.1)	0.7 (0.1)
High Density Lipoprotein, mg/100ml (SD)	49 (12)	62 (16)
Low Density Lipoprotein, mg/100ml (SD)	120 (31)	105 (29)
Triglycerides, mg/100ml	96 (68–130)	78 (59–108)
Cardiac and Vascular Growth Factors
Vascular endothelial GF (VEGF), pg/mL	272 (156, 435)	282 (157, 466)
VEGF receptor type 1, pg/mL	148 (110, 191)	136 (100, 187)
Angiopoietin-2, ng/ml	1.7 (1.3, 2.2)	2.0 (1.5, 2.7)
sTie2 , ng/ml	14.9 (12.7, 18.1)	14.2 (11.9, 17.2)
Hepatocyte GF, pg/mL	799 (687, 943)	815 (692, 963)
Insulin-like GF (IGF) 1, ng/ml	127 (104, 152)	125 (101, 155)
IGF binding protein (IGFBP)-3, ng/ml	2813 (2151, 3577)	3053 (2359, 3975)
GDF-15, ng/l	454 (375, 585)	501 (394, 646)
Primary Ecocardiographic Indices
LV Mass Index, g/m2	90.4 (81.8, 100.3)	72.9 (65.9, 80.9)
LV Ejection Fraction, %	63.7 (61.1, 66.6)	65.4 (62.6, 68.4)
GLS, %	−18.7 (−20.6, −17.2)	−21.0 (−22.9, −19.2)
E/E’	5.4 (4.7, 6.2)	5.6 (4.8, 6.6)
Aortic Root, cm	3.4 (3.1, 3.6)	2.9 (2.7, 3.1)
Secondary Ecocardiographic Indices
MAPSE, cm	1.6 (1.5, 1.8)	1.6 (1.5, 1.8)
GCS, %	−27.6 (−30.3, −25.0)	−28.7 (−31.5, −25.8)
LSS, msec	105.6 (91.9, 119.4)	106.2 (93.1, 120)
LV Diastolic Diameter, cm	5.2 (5.0, 5.4)	4.8 (4.5, 5.0)
LV Wall thickness, cm	1.9 (1.8, 2.1)	1.6 (1.5, 1.7)

Values are mean (SD) or median (Q1, Q3), unless otherwise noted.

GDF indicates growth differentiation factor; LV, left ventricular; GLS, global longitudinal strain; MAPSE, mitral annular plane systolic excursion; GCS, global circumference strain; LSS, longitudinal segmental synchrony.

Age- and sex-adjusted Spearman correlations among the non-standardized growth factors were low to moderate, with the highest pair-wise correlation observed between IGF-1 and IGFBP3 (r= 0.30, P<0.0001).

In multivariable-adjusted analyses, higher blood concentrations of GDF-15 were associated with higher LVMi, and higher sTie2 blood concentrations were associated with higher E/e’ values. IGF-1 and Ang2 concentrations were associated with GLS but in opposite directions (positively and negatively, respectively). The highest tertile of IGF-1 was related to less negative GLS compared to the lowest tertile (Figure 1; (less negative values of GLS indicate worse cardiac function)) consistent with the findings when using IGF-1 as a continuous variable. Furthermore, there was an inverse association of sFlt1 and Ang2 concentrations with AoR. VEFG, HGF and IGFBP3 were not significantly associated with any of the echocardiographic indices evaluated (Table 2).

Figure 1. Association between echocardiographic variables and tertiles of growth factors*.

* Using Least-Square Means, adjusted for age, sex, height, weight, current smoking, diabetes, systolic blood pressure, current use of hypertensive medications, heart rate, diabetes, total cholesterol/HDL ratio, and serum creatinine

Table 2.

Multivariable-adjusted Associations of Sex-standardized Growth Factors with Echocardiographic Indices of Structure and Function

Growth Factor	LVMi (log[g/m2])		LVEF (log[%])		GLS (%)		E/e‘ (log[E/e‘ ratio])		Aortic root (log[cm])
	β(SE)	FDR q-value	β(SE)	FDR q-value	β(SE)	FDR q-value	β(SE)	FDR q-value	β(SE)	FDR q-value
VEGF	0.001	0.94	0.0005	0.82	0.015	0.91	−0.003	0.74	−0.0004	0.93
	(0.003)		(0.0012)		(0.049)		(0.004)		(0.002)
sFlt1	0.002	0.77	0.0013	0.62	0.009	0.93	0.002	0.81	−0.004	0.04
	(0.003)		(0.0011)		(0.046)		(0.004)		(0.001)
Ang-2	0.004	0.55	0.0006	0.82	−0.154	0.03	0.009	0.15	−0.005	0.04
	(0.003)		(0.0011)		(0.045)		(0.004)		(0.002)
sTie2	−0.004	0.53	0.0023	0.30	0.021	0.854	0.011	0.04	0.0003	0.93
	(0.003)		(0.0013)		(0.045)		(0.004)		(0.002)
HGF	−0.004	0.55	0.0025	0.19	0.008	0.93	0.003	0.74	−0.0013	0.73
	(0.003)		(0.0012)		(0.047)		(0.004)		(0.001)
IGF-1	0.006	0.22	0.0008	0.80	0.156	0.04	−0.003	0.74	−0.0001	0.98
	(0.003)		(0.0014)		(0.051)		(0.004)		(0.002)
IGFBP-3	0.005	0.34	−0.0019	0.44	0.107	0.15	−0.004	0.72	0.0001	0.98
	(0.003)		(0.0012)		(0.046)		(0.004)		(0.002)
GDF-15	0.009	0.01	0.0009	0.70	−0.060	0.53	0.002	0.85	0.0011	0.78
	(0.002)		(0.0011)		(0.042)		(0.004)		(0.002)

Values are expressed per 1 SD increase in the growth factors

VEGF indicates vascular endothelial growth factor (GF); sFlt1, VEGF receptor type 1; Ang-2, angiopoietin 2; sTie2, soluble angiopoietin Tie-2 receptor; HGF, hepatocyte GF; IGF-1, insulin-like GF-1; IGFBP-3, IGF binding protein type 3; GDF-15, growth differentiation factor-15

FDR= False Discovery Rates, q-values

Models adjusted for age, sex, height, weight, current smoking, diabetes, systolic blood pressure, current use of hypertensive medications, heart rate, total cholesterol/HDL ratio, and serum creatinine.

We created LSM plots relating tertiles of the significant biomarkers to the echocardiographic traits (Figure 1). We also tested for nonlinearity and confirmed that all significant growth factors showed evidence of a linear association with the echocardiographic indices evaluated (Figure 2). We did not observe significant effect modification of the association between the growth factors and the echocardiographic traits by sex.

Figure 2. Multivariable-adjusted splines for testing non-linearity.

Multivariable-adjusted Splines of GDF-15 by log LVMi (log(g/m²)), Tie-2 by log E/e’ (log e/e’), Ang-2 by GLS (%), IGF-1 by GLS (%), Ang-2 by log AoR (log(cm)), and SFLT1 by log AoR (log(cm)) with knoxts placed at 25^h, 50^th and 75^th percentile values of the biomarkers

In our investigation, Ang-2 concentrations were positively associated with MAPSE values. We did not observe any other statistically significant associations.

Discussion

We related a panel of eight circulating growth factors/receptors to several echocardiographic indicators of cardiac remodeling in a community-based sample. We observed that the concentrations of the cardiac and vascular growth factors studied were associated with cardiac structure and function in a manner that was mostly consistent with their known biological effects. The associations were modest in terms of both statistical significance and effect size.

Principal findings

Our principal findings are 5-fold: First, concentrations of circulating GDF-15 were positively associated with LVMi values. Second, higher concentrations of circulating Ang-2 were associated with better (more negative) GLS values whereas higher concentrations of circulating IGF-1 were associated with worse (less negative) GLS values. The relations of Ang-2 and IGF-1 with GLS are intriguing as they are in the opposite direction of their known biological effects, given that a more negative value of GLS is considered to reflect better myocardial systolic function. Third, higher concentrations of circulating sTie2 were associated with higher E/e’ values. Fourth, higher concentrations of sFlt1 and Ang-2 were associated with smaller AoR; both sFlt1 and Ang-2 have angiogenic effects and have been linked to arteriosclerosis leading to aortic stiffness, which may be accompanied by a smaller AoR. Fifth, VEGF, IGFBP3 and HGF were not associated with any of the echocardiographic indices evaluated.

Comparison with the published literature

Association of GDF-15 with LVM

The positive association between GDF-15 and LVM has been reported previously in human studies.^33,17 GDF-15 has also been associated with HF¹⁷ in other population-based studies and has been reported to be a strong predictor of HF after acute myocardial infarction.³⁴ In the elderly Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) Swedish cohort³³ higher concentrations of GDF-15 were associated with higher LVMi. These aforementioned investigations included older participants or those with clinically evident disease; it is noteworthy that we observed a significant positive association between GDF-15 and LVM among much younger participants (mean age 40 years as compared to mean age 70 years in the PIVUS study) without any prior evidence of disease, suggesting that GDF-15 could serve as a marker for subclinical disease.

Association of Ang-2 and IGF-1 with GLS

HF is a process that involves ongoing angiogenesis where VEGF and angiopoietins respond to systemic hypoxemia. VEGF initiates angiogenesis while Ang-2 promotes vascular remodeling,³⁵ so it is expected that Ang-2 increases due to acute HF in which there is a state of hypoxia; however, higher concentrations of Ang-2 and its receptor, sTie2, have been reported in patients with both acute and chronic HF compared to healthy controls.³⁵ In the latter investigation, concentrations of sTie2 and VEGF in both acute and chronic HF patients were as high as the values we observed in our sample; however, concentrations of Ang-2 were about 3 times higher in the healthy controls and even much higher in both acute and chronic HF groups than in our sample. In our current investigation, higher values of Ang-2 were associated with better cardiac function as determined by lower (more negative) GLS values. However, in a German cohort,³⁶ higher concentrations of Ang-2 were associated with lower LV systolic function as assessed by lower values of fractional shortening (FS) measured by echocardiography. The latter sample was older and participants had a higher prevalence of smoking, hypertension and diabetes compared to our participants who were at least 15 years younger on average and with lower prevalence of traditional CVD risk factors. A previous study, using the same sample, observed that Ang-2 was inversely related with indices of central hemodynamic load such as the mean arterial pressure and carotid-femoral pulse wave velocity (CFPWV), an index of aortic wall stiffness.³⁷ In our analysis, Ang-2 and IGF-1 had an opposite relation with GLS. Worse cardiac function (higher GLS) was related with lower levels of Ang-2 and higher levels of IGF-1.

On the other hand, the potentially favorable effect of IGF-1 on cardiac remodeling and HF remains controversial. Investigators have reported both positive associations of IGF-1 with LV systolic function^38,39 as well as null associations between IGF-1 and cardiac contractility in HF patients.⁴⁰ Others have reported a positive relation between IGF-1 concentrations and subclinical CVD as assessed by carotid intima-media thickness.⁴¹ In the current investigation, higher IGF-1 concentrations were associated with worse GLS after adjustment for CVD risk factors. It is conceivable that this association indicates a complex feedback loop wherein IGF-1 levels increase and provide a partially offsetting cardioprotective effect in the face of progressive LV systolic dysfunction that is similar to the pattern of increasing levels of B-type natriuretic peptide, with its myocardial protective effects, in parallel with progressively worse LV systolic function.

Association of sTie2 with E/e’

Higher sTie2 levels have been reported in a small sample of HF patients compared to healthy controls.³⁵ In another large population-based study, concentrations of sTie2 were inversely associated with left ventricular diastolic diameter (LVDD), a component of LVM.³⁶ In two different studies using participants from the FHS Third Generation cohort, higher levels of sTie2 were associated with higher levels of CFPWV³⁷ after adjustment for age and sex, and with obesity and diabetes⁴² after adjustment for age, sex and other CVD risk factors; our findings confirm the aforementioned association. Moreover, to our knowledge, our investigation appears to be the first to report that higher concentrations of sTie2 are associated with higher E/e’ values, an indicator of LV diastolic function, which may suggest that sTie2 levels may be elevated in the presence of diastolic impairment, a premise that warrants further study.

Association of sFlt1 and Ang-2 with AoR

VEGF, sFlt1 and Ang-2 act in a highly coordinated way during the process of angiogenesis, with Ang-2 considered to be the initiator of neovascularization.⁴³ Whereas VEGF is a regulator of angiogenesis, sFlt1 modulates vascular sprout formation by the recruitment of hematopoietic precursors and migration of monocytes and macrophages.^44,45 In two previous reports from FHS investigators using data from the same FHS Third Generation cohort, Ang-2 levels were positively associated with major vascular risk factors like hypertension, smoking and diabetes, and with pulse pressure⁴² and inversely associated with several measures of vascular stiffness and hemodynamic load, such as MAP and CFPWV.³⁷ Pulse pressure, an indicator of conduit vessel stiffness was inversely related with AoR in several investigations using data from participants comprising the first and second FHS generations.⁴⁶ Although this sample was older than the one we used and had higher prevalence of hypertension, participants were also free of CVD. Additionally, it is plausible that the inverse association between Ang-2 and AoR we observed in our sample may be mediated by pulse pressure. Similarly, in our sample we observed that higher VEGF-1 levels were associated with shorter AoR. Ang-2 as well as VEGF-1 may be promoting angiogenesis at the vascular walls, but at early stages of atherosclerosis the size of the aorta root is not increased yet as it has been observed in more advance atherosclerosis disease where Ang-2 is associated with the developing of abdominal aorta aneurism.¹⁹ Our observation of an association of Ang2 and sFlt1 with smaller AoR dimensions are consistent with prior studies and with the overall physiological function of these growth factors.⁸ We did not observe associations between VEGF, HGF and IGFBP3 with any of the echocardiographic measures.

Strengths and Limitations

We evaluated a large community-based sample with a wide age range, as well as a comprehensive panel of circulating growth factors and several measures of cardiac structure and function that offer complementary information. However, our investigation has several limitations. We evaluated cross-sectional associations; therefore, we can’t determine directionality precluding any causal inference. We also did not have available data on serial measurements of growth factors or echocardiographic variables used for this investigation; therefore, we could not evaluate variability of the measurements over time. Our sample is predominantly comprised of white individuals of European ancestry, which limits the generalizability of our findings. We adjusted for a broad set of covariates but cannot exclude residual confounding (i.e., inflammatory biomarkers). Our analysis involved multiple comparisons, but we accounted for multiple testing by using false discovery rates. Our sample is relatively healthy and young, with a low prevalence of cardiovascular risk factors and no overt CVD. On one hand, a healthier sample may explain the small effect sizes we observed. On the other hand, our observations are consistent with a role of these growth factors in dynamic myocardial remodeling in a state of relative health.

Conclusion

In our large young-to-middle-aged community-based sample, we observed very modest associations between select circulating growth factors and echocardiographic traits. Some of our findings were concordant with the known biological effects of these factors whereas other findings were opposite of what we expected, likely reflecting complex biological feedback loops in the context of the cross-sectional nature of our investigation. The potential clinical and prognostic significance of the observed associations warrant further studies.

Highlights.

• There are patterns of associations between several circulating vascular GF and cardiac remodeling

• The observed associations are consistent with the known biological effects of these pro- and anti-angiogenic factors on the myocardium and conduit arteries