Left atrial volume index in patients with heart failure and severely impaired left ventricular systolic function: the role of established echocardiographic parameters, circulating cystatin C and galectin-3

Abstract

Backround:

Left atrial (LA) enlargement plays an important role in the development of heart failure (HF) and is a robust prognostic factor. Fibrotic processes have also been advocated to evoke HF through finite signalling proteins.

Methods:

We examined the association of two such proteins, cystatin C (CysC) and galectin-3 (Gal-3), and other clinical, echocardiographic and biochemical parameters with LA volume index (LAVi) in patients with HF with severely impaired left ventricular ejection fraction (LVEF). Severe renal, liver, autoimmune disease and cancer were exclusion criteria.

Results:

A total of 40 patients with HF (31 men, age 66.6 ± 1.7) with LVEF = 25.4 ± 0.9% were divided into two groups according to the mean LAVi (51.03 ± 2.9 ml/m²) calculated by two-dimensional transthoracic echocardiography. Greater LAVi was positively associated with LV end-diastolic volume (p = 0.017), LV end-systolic volume (p = 0.025), mitral regurgitant volume (MRV) (p = 0.001), right ventricular systolic pressure (RVSP) (p < 0.001), restrictive diastolic filling pattern (p = 0.003) and atrial fibrillation (p = 0.005). Plasma CysC was positively correlated with LAVi (R² = 0.135, p = 0.019) and log-transformed plasma Gal-3 (R² = 0.109, p = 0.042) by simple linear regression analysis. Stepwise multiple linear regression analysis showed that only MRV (t = 2.236, p = 0.032), CysC (t = 2.467, p = 0.019) and RVSP (t = 2.155, p = 0.038) were significant predictors of LAVi.

Conclusions:

Apart from known determinants of LAVi, circulating CysC and Gal-3 were associated with greater LA dilatation in patients with HF with reduced LVEF. Interestingly, the correlation between these two fibrotic proteins was positive.

Introduction

It is proven that the left atrium (LA) is not just a muscular chamber between the left ventricle (LV) and the pulmonary circulation, but it also plays a significant role in LV filling modulation, has endocrine function (atrial natriuretic peptide synthesis and secretion), and contributes to sympathetic activity and vasopressin release.¹ In fact, LA size has been shown to be related to cardiovascular mortality, myocardial infarction (MI), stroke and atrial fibrillation (AF) in a majority of the population.^2–4 As far as heart failure (HF) with reduced LV ejection fraction (LVEF) is concerned, LA enlargement remained a robust and independent prognostic factor for mortality in patients with HF, even after adjustment for LVEF, E-wave deceleration time (EWDT), New York Heart Association (NYHA) class or age in a relatively recent large meta-analysis.⁵

Nowadays biomarkers (BMRs) are very useful and important tools for the diagnosis and prognosis of patients with HF. Natriuretic peptides are the cornerstone and their measurement is recommended in all patients with acute dyspnoea and suspected HF.⁶ What is more, the prognostic value of changes over time of N-terminal pro-brain natriuretic peptide (NT-proBNP) was well demonstrated in the Valsartan Heart Failure Trial.⁷ An emerging BMR, cystatin C (CysC), is a cysteine protease inhibitor produced at a constant rate by all nucleated cells in mammals.⁸ It has not only been advocated to be a more accurate estimate of glomerular filtration rate (GFR) than serum creatinine, but has been also shown to be associated with LV structural and functional parameters, due to its potential role in extracellular matrix (ECM) remodelling.^9–11 Galectin-3 (Gal-3) is a β-galactoside-binding protein, which is involved in myocardial fibrosis and ventricular remodelling, and is proposed to play a crucial role in the development of HF.^12,13 It has also proven to be a robust prognostic marker of HF mortality and hospitalization.¹⁴

In this single centre observational study we searched for clinical, echocardiographic and biochemical determinants of the LA volume index (LAVi) in patients with HF with severely impaired LV systolic function, with a special focus on CysC and Gal-3, which are both involved in ECM modulation and fibrosis. We further investigated the correlation between these two proteins.

Methods

Study population

A total of 40 patients with symptoms of decompensated HF who were hospitalized in the General Hospital of Nikea in Athens, Greece were included in this study. All investigations, including clinical parameters, biochemical tests and comprehensive two-dimensional (2D) transthoracic echocardiogram (TTE) were performed when patients were in stable condition, before discharge. Participants should have a LVEF < 35%, while organic mitral regurgitation (MR), severe renal, liver, autoimmune disease, cancer or any other condition that was thought to affect the results of the study were exclusion criteria. Subjects with congenital heart disease or severe valvular heart disease requiring surgery, apart from severe functional MR and tricuspid regurgitation, were also excluded. The study was approved by the local Ethics Committee and it conformed to the principles outlined in the Declaration of Helsinki. All participants gave written informed consent to clinical examinations, laboratory analyses and the use of data records for research purposes.

Data analysis

2D TTE, including continuous and pulse-wave Doppler measurements (Sonos 7500 echocardiography system, S3-ultraband sector transducer, Philips Medical Systems, Andover, MA, USA) was performed according to standard operating procedures by a trained and certified echocardiographer. LVEF and LA end-systolic volume were calculated using the biplane Simpson method. Mitral regurgitant volume (MRV) was calculated using the proximal isovelocity surface area method.¹⁵ Effective stroke volume (SV) was estimated by subtracting MRV from SV. LV mass was calculated using the formula described by Devereaux and colleagues.¹⁶ Right ventricular systolic pressure (RVSP) was calculated from the maximal tricuspid regurgitant jet velocity using the systolic transtricuspid pressure gradient calculated by the modified Bernoulli equation (ΔP = 4V², where V is the maximal tricuspid regurgitant jet velocity), plus the right atrial pressure according to inferior vena cava size and its respiratory variation.¹⁷ Restrictive diastolic filling pattern was defined if E/A ratio was > 2 and EWDT ⩽ 160 ms.¹⁸ Measurements were indexed to body surface area when appropriate. An electrocardiogram (ECG) was obtained from each patient, using a multi-channel ECG machine (Nihon Kohden Cardiofax Q 9130K, Tokyo, Japan), which automatically calculated the QRS duration. The principal investigator reviewed each ECG and confirmed the findings. The estimated GFR (eGFR) was calculated by the Cockcroft–Gault equation. All participants performed a 6-min walk test (6MWT) and completed the ‘Minnesota living with heart failure’ quality-of-life questionnaire (MLWHFQ).

Measurement of BMRs

Venous blood was obtained from each patient at 8:00 a.m. on the day of discharge. In addition to the full blood count and regular biochemical screening, five 2.7 ml BD Vacutainer plus plastic citrate blood collection tubes were filled from each subject and immediately centrifuged at 1550 × g for 10 min. The supernatant plasma was transferred with the use of sterile plastic Pasteur-like pipettes (GPP-3.0G-S1, Fisher Scientific, Hampton, NH, USA) in 2 ml sterile, DNase and RNase-free Eppendorf-like tubes (G053, Fisher Scientific), and subsequently stored at −80°C. When the collection of tubes from all participants was completed, the tubes were transferred in dry ice in expanded polystyrene insulated lab boxes from the Nikea General Hospital to the Biomedical Research Foundation, Academy of Athens, where they were immediately restored at −80°C. After samples were thawed, further assays for the following BMRs of HF were conducted: NT-proBNP (Human Enzyme-linked Immunosorbent Assay [ELISA] Kit, Abnova Corporation, Taipei, Taiwan), CysC (Quantikine Human Cystatin C Immunoassay, R&D Systems, Minneapolis,MN, USA) and Gal-3 (Human Galectin-3 ELISA Kit, Aviscera Bioscience, Santa Clara, CA, USA). The tests were performed according to the manufacturers’ protocols.

Statistical analysis

Categorical variables are presented as percentages, while continuous variables are presented as mean ± 1 standard deviation. Data that did not follow a Gaussian distribution were logarithmically transformed when needed. Patients were divided into two groups, according to the mean LAVi (51.03 ± 2.85 ml/m²). Baseline characteristics of the two groups were compared using the chi-square test for categorical variables; continuous variables were compared by the Student’s t-test or nonparametric Mann–Whitney U test with Monte Carlo simulation (confidence level 99% for 10,000 number of samples) in non-normally distributed data. Linear bivariate analysis by Pearson’s correlation coefficient was used to detect associations of LAVi with 6MWT, MLWHFQ, LVEF, eGFR and log-transformed CysC. Linear bivariate analysis by Spearman’s rank correlation was used to detect associations of LAVi with other variables without normal distribution. Furthermore, simple linear regression analysis was used to detect associations between LAVi (dependent variable) and the following variables: LV mass, LV end-systolic volume (LVESV), LV end-diastolic volume (LVEDV), MRV, RVSP, NT-proBNP, CysC and Gal-3. Stepwise multivariable linear regression analysis was used to evaluate the association of LAVi (dependent variable) with LVESV, LVEDV, LV mass, EWDT, MRV, RVSP, CysC and Gal-3 in a predictive model. Finally, the association between CysC and log-transformed Gal-3 was investigated using simple linear regression analysis. Variables with more than 5% of missing data were excluded from analysis. For variables with less than 5% of missing data, mean imputations were used, when necessary. P value < 0.05 was considered statistically significant. SPSS 22 (IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY, USA: IBM Corp.) was used for statistical analysis.

Results

Baseline characteristics of all patients are shown in Table 1. Coronary artery disease (as assessed by coronary angiography or a history of MI), AF and subsequent treatment with vitamin K antagonists were more prevalent in patients with greater LAVi (p = 0.001, p = 0.005 and p = 0.0001, respectively). Regarding echocardiographic parameters, patients with greater LAVi also had more dilated LV, as assessed by LVEDV and LVESV, worse MR (greater MRV), restrictive physiology as assessed by E/A-wave and EWDT, and higher RVSP. Interestingly, greater LAVi was correlated with a lower level of total cholesterol and triglycerides (TG). Although there was a trend in association between greater LAVi and higher levels of NT-proBNP, CysC and Gal-3, nonparametric Mann–Whitney U test revealed that this was statistically nonsignificant. The two groups of patients did not differ with regard to their LVEF, SV, NYHA class, QRS duration, 6MWT or MLWHFQ score.

Table 1.

Baseline clinical and echocardiographic characteristics.

		Mean ± 1 SD	LAVi (ml/m2)		p value
			⩽ 51.03 ± 2.85 (n = 22)	> 51.03 ± 2.85 (n = 18)
Age, years		66.6 ± 1.7	64.59 ± 2.44	69.06 ± 2.39	0.204
Female, n (%)		9 (22.5)	6 (27.3)	3 (16.7)	0.476
Body mass index (kg/m2)		27.74 ± 0.83	28.15 ± 1.36	27.24 ± 0.83	0.519
NYHA class, n (%)	II	6 (15)	4 (18.2)	2 (11.1)	0.690
	III	29 (72.5)	16 (72.7)	13 (72.2)
	IV	5 (12.5)	2 (9.1)	3 (16.7)
Coronary artery disease, n (%)		20 (50)	6 (27.3)	14 (77.8)	0.001*
Atrial fibrillation, n (%)		13 (32.5)	3 (13.6)	10 (55.6)	0.005*
Hypertension, n (%)		23 (57.5)	12 (54.5)	11 (61.1)	0.676
Diabetes, n (%)		11 (27.5)	8 (36.4)	3 (16.7)	0.286
eGFR, ml/min		66.78 ± 3.7	70 ± 5.9	63 ± 4.3	0.368
Electronic device, n (%)	PPM (single or dual lead)	5 (12.5)	3 (13.6)	2 (11.1)	1.0
	CRT-P or CRT-D	3 (7.5)	2 (9.1)	1 (5.6)	1.0
	ICD	1 (2.5)	1 (4.5)	0 (0)	1.0
QRS duration, ms		134.63 ± 6.33	130.95 ± 8.29	139.11 ± 9.93	0.529
QRS morphology, n (%)	LBBB	15 (37.5)	6 (27.3)	9 (50)	0.14
	RBBB	5 (12.5)	4 (18.2)	1 (5.6)	0.355
	LAH	14 (35)	8 (36.4)	6 (33.3)	0.842
Medication, n (%)	ACE-inhibitors or ARBs	29 (72.5)	16 (72.7)	13 (72.2)	1.0
	Loop diuretic	39 (97.5)	22 (100)	17 (94.4)	0.45
	Mineralocorticoid receptor blockers	22 (55)	10 (45.5)	12 (66.7)	0.18
	Aspirin	21 (52.5)	13 (59.1)	8 (44.4)	0.356
	Clopidogrel	5 (12.5)	4 (18.2)	1 (5.6)	0.355
	Vitamin K antagonists	14 (35)	2 (9.1)	12 (66.7)	0.0001*
	Beta- blockers	35 (87.5)	20 (90.9)	15 (83.3)	0.642
	Ivabradine	4 (10)	4 (18.2)	0 (0)	0.114
	Statins	20 (50)	10 (45.5)	10 (55.6)	0.525
	Amiodarone	7 (17.5)	2 (9.1)	5 (27.8)	0.211
	Digoxin	9 (22.5)	3 (13.6)	6 (33.3)	0.253
6MWT, m		304.21 ± 22.56	303 ± 32.13	306 ± 32.17	0.954
MLWHFQ, score		42.48 ± 3.34	44.59 ± 3.87	39.89 ± 5.78	0.49
Echocardiographic characteristics	LVEDV, ml	206.52 ± 13.77	181.73 ± 11.23	236.82 ± 26.05	0.017*
	LVESV, ml	155.10 ± 11.86	134.92 ± 9.21	179.78 ± 22.89	0.025*
	LVEF, %	25.39 ± 0.88	26.12 ± 1.28	24.51 ± 1.20	0.375
	LV mass, g	259.7 ± 12.52	240 ± 15.16	283.76 ± 19.77	0.082
	Eff SV, ml	35.51 ± 2.30	36.51 ± 2.94	34.29 ± 3.70	0.638
	Eff SV index, ml/m2	18.57 ± 1.19	19.10 ± 1.54	17.93 ± 1.90	0.631
	MRV, ml	15.94 ± 2.06	10.51 ± 2.41	22.57 ± 2.87	0.001*
	E-wave, m/s	0.85 ± 0.05	0.715 ± 0.068	1.018 ± 0.050	0.001*
	A-wave, m/s	0.69 ± 0.05	0.759 ± 0.039	0.507 ± 0.114	0.072
	E/A- wave	1.39 ± 0.22	0.93 ± 0.12	2.58 ± 0.52	0.018*
	E-wave deceleration time, ms	166.43 ± 9.63	188.6 ± 15.4	139.3 ± 6.0	0.006*
	RVSP, mmHg	30.61 ± 3.05	20.5 ± 3	42.3 ± 4	< 0.001*
	Restrictive pattern, n (%)	23 (57.5)	8 (36.4)	15 (83.3)	0.003*
Total cholesterol, mmol/L		4.13 ± 0.16	4.48 ± 0.22	3.74 ± 0.19	0.021*
Triglycerides, mmol/L		1.39 ± 0.34	1.85 ± 0.64	0.88 ± 0.05	0.011*
NT-proBNP, pg/ml		807.79 ± 73.45	716.77 ± 84.32	913.97 ± 123.05	0.321
Cystatin C, μg/ml		1.10 ± 0.06	0.998 ± 0.0586	1.226 ± 0.1075	0.069
Galectin-3, ng/ml		9.15 ± 1.03	9.09 ± 1.23	9.22 ± 1.73	0.951

^*p value < 0.05. 6MWT, 6-min walk test; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; CRT-D, cardiac resynchronization therapy-defibrillator; CRT-P, cardiac resynchronization therapy- pacemaker; Eff SV, effective stroke volume; eGFR, estimated glomerular filtration rate; ICD, implantable cardioverter defibrillator; LAH, left anterior hemiblock; LAVi, left atrial volume index; LBBB, left bundle branch block; LV, left ventricle; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end-systolic volume; MLWHFQ, Minnesota living with heart failure quality-of-life questionnaire; MRV, mitral regurgitant volume; NT-proBNP, N-terminal pro-brain natriuretic peptide; NYHA, New York Heart Association; PPM, permanent pacemaker; RBBB, right bundle branch block; RVSP, right ventricular systolic pressure; SD, standard deviation.

Linear bivariate analysis by Pearson’s correlation coefficient revealed positive correlation between LAVi and log-transformed CysC, (r Pearson’s = 0.323, p = 0.042), while there was no significant correlation between LAVi and 6MWT, MLWHFQ, LVEF, LV mass and eGFR (p = 0.475, p = 0.951, p = 0.669, p = 0.076 and p = 0.221, respectively). Linear bivariate analysis by Spearman’s correlation coefficient did not reveal any significant association between LAVi and SV (p = 0.328), SV index (p = 0.395), NT-proBNP (p = 0.172) or Gal-3 (p = 0.772).

Simple linear regression analysis, using LAVi as the dependent variable and biochemical and echocardiographic parameters as predictors in an univariate model (Table 2), showed that LAVi was linearly correlated with LVEDV (p = 0.018), LVESV (p = 0.028), EWDT (p = 0.012), MRV (p < 0.0001) (Figure 1), RVSP (p < 0.0001) (Figure 2) and CysC (p = 0.019) (Figure 3).

Table 2.

Simple linear regression analysis for left atrial volume index (dependent variable).

	R	R 2	Adjusted R2	B	p
LVEDV, ml	0.373	0.139	0.116	0.077	0.018*
LVESV, ml	0.348	0.121	0.098	0.084	0.028*
LV mass, g	0.284	0.080	0.056	0.065	0.076
E-wave deceleration time, ms	0.392	0.153	0.131	–0.116	0.012*
MRV, ml	0.590	0.348	0.331	0.816	< 0.0001*
RVSP, mmHg	0.581	0.338	0.320	0.557	< 0.0001*
NT-proBNP, pg/ml	0.244	0.059	0.034	0.010	0.135
Cystatin C, μg/ml	0.368	0.135	0.113	17.469	0.019*
Galectin-3, ng/ml	0.138	0.019	–0.008	0.402	0.41

^*p value < 0.05 for the F-test on the regression analysis. LV, left ventricle; LVEDV, left ventricular end-diastolic volume; LVESV, left ventricular end-systolic volume; MRV, mitral regurgitant volume; NT-proBNP, N-terminal pro-brain natriuretic peptide; RVSP, right ventricular systolic pressure.

Figure 1.

Simple scatterplot with regression line of mitral regurgitant volume (MRV) and left atrial volume index (LAVi). Positive correlation is depicted between the two variables (R² linear = 0.348, p < 0.0001). Curved lines represent 95% confidence intervals for the mean predicted LAVi.

Figure 2.

Simple scatterplot with regression line of right ventricular systolic pressure (RVSP) and left atrial volume index (LAVi). Positive correlation is depicted between the two variables (R² linear = 0.338, p < 0.0001). Curved lines represent 95% confidence intervals for the mean predicted LAVi.

Figure 3.

Simple scatterplot with regression line of serum levels of cystatin C and left atrial volume index (LAVi). Positive correlation is depicted between the two variables (R² linear = 0.135, p = 0.019). Curved lines represent 95% confidence intervals for the mean predicted LAVi.

Stepwise multivariable linear regression analysis to evaluate the association of the dependent variable LAVi with LVESV, LVEDV, EWDT, MRV, RVSP, LV mass, CysC and Gal-3 in a predictive model revealed that only MRV (t = 2.236, p = 0.032), CysC (t = 2.467, p = 0.019) and RVSP (t = 2.155, p = 0.038) were significant predictors of LAVi (Table 3). The model was not affected after adjustment for age, sex, body mass index, history of hypertension and eGFR. Interestingly, linear bivariate analysis by Spearman’s rank showed statistically significant correlation between plasma levels of CysC and Gal-3 (p = 0.019). This was further confirmed by simple linear regression analysis between CysC (independent variable) and log-transformed Gal-3 (R² = 0.109, p = 0.042) as shown in Figure 4.

Table 3.

Stepwise multiple linear regression analysis evaluating the association of the left atrial volume index (dependent variable) with mitral regurgitant volume, serum cystatin C and right ventricular systolic pressure in a predictive model.

Model	Predictors	R 2	Adjusted R2	R2 change	B (95% confidence interval)	SE of B	t	p
1	MRV	0.348	0.331	0.348	0.816 (0.449–1.183)	0.181	4.501	< 0.0001*
2	MRV	0.438	0.407	0.090	0.766 (0.418–1.114)	0.172	4.460	< 0.0001*
	CysC				14.340 (2.399–26.281)	5.894	2.433	0.020*
3	MRV	0.502	0.460	0.064	0.475 (0.044–0.906)	0.212	2.236	0.032*
	CysC				13.880 (2.468–25.292)	5.627	2.467	0.019*
	RVSP				0.317 (0.019–0.615)	0.147	2.155	0.038*

^*p values < 0.05 for the t-tests. Other factors such as left ventricular end-systolic and end-diastolic volume, E-wave deceleration time, left ventricular mass and galectin-3 were excluded from the model as nonsignificant predictors. CysC, cystatin C; MRV, mitral regurgitant volume; RVSP, right ventricular systolic pressure; SE, standard error.

Figure 4.

Simple scatterplot with regression line of serum levels of cystatin C and log-transformed galectin-3. Positive correlation is depicted between the two variables (R² linear = 0.109, p = 0.042). Curved lines represent 95% confidence intervals for the mean predicted log-transformed galectin-3.

We further examined the association of Gal-3 with renal function, as assessed by eGFR. Linear bivariate analysis by Spearman’s correlation coefficient revealed negative correlation between serum Gal-3 and eGFR (ρ Spearman’s = -0.385, p = 0.017). Indeed, this was also confirmed by simple linear regression; increased plasma levels of Gal-3 were correlated with worse renal function (B = −1.184, p = 0.022) as shown in Figure 5.

Figure 5.

Simple scatterplot with regression line of serum levels of galectin-3 and estimated glomerular filtration rate. Negative correlation is depicted between the two variables (R² linear = 0.137, B = −1.184, p = 0.022). Curved lines represent 95% confidence intervals for the mean predicted galectin-3.

Finally, because of the finding that coronary artery disease was more prevalent in the group of patients with greater than mean LAVi, we investigated whether coronary artery disease was associated with levels of circulating CysC or Gal-3. This was not confirmed by nonparametric Mann–Whitney U test (CysC = 1.107 ± 0.095 μg/ml versus 1.095 ± 0.076 μg/ml, p = 0.487 and Gal-3 = 9.369 ± 1.222 ng/ml versus 8.902 ± 1.731 ng/ml, p = 0.279, for ischaemic versus non-ischaemic HF, respectively).

Discussion

We showed that in patients with HF with severely impaired LV function, greater than mean LAVi was associated with greater LV remodelling, larger MRV and AF. Similar to our findings, Rossi and colleagues demonstrated in 337 patients with dilated cardiomyopathy, that LV dimension, severity of MR, degree of diastolic dysfunction and AF were significant determinants of left atrial volume (LAV), calculated by 2D TTE.¹⁹

Interestingly, in our group of patients, greater than mean LAVi was associated with higher prevalence of coronary artery disease. Nevertheless, older studies have shown that LA fibrosis was more prominent in idiopathic dilated cardiomyopathy than in ischaemic dilated cardiomyopathy, and that LA booster pump function was compromised in the former and preserved in the latter.^20,21 These findings indicated that the LA wall was affected by the myopathic process in idiopathic dilated cardiomyopathy, irrespective of the LA mechanical overload. However, patients included in these small volume studies were younger, with less severe symptoms (lower NYHA class) and milder MR in comparison with our cohort, while ventriculography was used to assess the LVEF. More recently, a study using speckle-tracking 2D strain echocardiography has demonstrated that LA contractile and reservoir function is more compromised in idiopathic dilated cardiomyopathy than in ischaemic dilated cardiomyopathy, even after cardiac resynchronization therapy.²² Again, patients in this study were younger and had a higher LVEF than our cohort. Overall, there might be a more pronounced LA involvement in the myopathic process in idiopathic dilated cardiomyopathy than in ischaemic dilated cardiomyopathy, especially in the early stages of the disease. However, these differences might be attenuated as HF progresses and then fibrotic processes (e.g. through CysC and Gal-3) become similar. Further studies are mandated to prove this hypothesis.

The relation between LA dimension and AF has been well known for some decades.²³ The electrophysiological and molecular mechanisms beneath the vicious circle where AF causes atrial dilatation and vice versa have been demonstrated: atrial stretching alters the effective refractory period in different atrial regions creating the perfect substrate for AF, while downregulation or altered function of the L-type Ca channel and possibly an upregulation of the Na/Ca-exchanger are responsible for the depressed atrial contractility in AF.^24,25

We found that indexes of LV diastolic dysfunction, such as E/A-wave or restrictive physiology were also correlated with greater LAVi. Notwithstanding, LAVi is a more robust indicator of diastolic dysfunction that is not affected by loading conditions of LV, in contrast to E/A- wave or EWDT.²⁶

Interestingly, lower cholesterol and TG levels were correlated with higher LAVi. We have previously demonstrated that lower total cholesterol levels were associated with worse NYHA class, as well as high-density lipoprotein (HDL) well predicted functional status, as assessed by the 6MWT, in a similar group of patients as in the current study.²⁷ In the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA), higher baseline low-density lipoprotein, HDL, apolipoprotein A-I (ApoA-1), ApoB and TG were associated with a better prognosis.²⁸ This finding is probably due to several factors. Lower total cholesterol may represent a more advanced stage of HF (and therefore larger LAVi), where a shift from anabolic to catabolic processes happens.²⁹ Moreover, it has been proposed that lipoproteins act as buffers for endotoxin in patients with HF, thus diminishing inflammation and immune activation.³⁰

MRV was the most significant determinant of LAVi, in both simple linear and stepwise multiple regression analysis. LA enlargement is a well-known pathophysiological adaptation to volume overload in MR, leading to better LA pressure compliance.³¹ In experimental models, MR induces interstitial fibrosis and inflammation to the LA wall, thus causing dilatation.³² In fact, LAVi has been shown to be an independent predictor of survival in patients with organic MR and preserved EF, treated conservatively.³³ Our study demonstrates that MR still plays a crucial role in LA remodelling in HF with severely impaired LV function. This might indicate that these patients can still be favoured by mitral valve replacement or repair (even with the MitraClip), which can withhold further LA dilatation and even promote LA reverse remodelling and improve contraction.³⁴

CysC, apart from being a robust estimator of GFR, is also known to affect ECM remodelling by inhibiting cathepsin B.^9,11 The association between higher levels of CysC, LV hypertrophy and increased LV mass assessed by magnetic resonance imaging (MRI) was shown in the Dallas heart study.¹⁰ In our group of patients, plasma CysC levels were positively associated with LA dimension, denoting that CysC could play a significant role in cardiac remodelling in patients with severe HF. Specifically, our findings may indicate an ongoing interstitial fibrotic process in established HF with severely impaired LV function, promoted by increased levels of plasma as well as myocardial CysC, which antagonizes ECM degradation caused by cathepsins.

Not surprisingly, we also found a positive correlation between LAVi and RVSP. It is well known that passive backward transmission of filling pressures due to LV dysfunction, MR and loss of atrial compliance is the main factor causing pulmonary hypertension in HF.³⁵

Finally, the fact that levels of plasma CysC were positively associated with levels of plasma Gal-3, which is also a marker of myocardial fibrosis, may reflect common pathophysiological mechanisms of ECM alteration in HF.¹³ Sharma and colleagues in their pioneering study have shown that Gal-3 induces cardiac fibroblast proliferation through stimulation of cyclin-D1, and promotes the production of collagen I, thus increasing collagen I/III ratio.³⁶ Interestingly, these effects were evident before overt development of HF. Moreover, in the HF-ACTION study, the largest study to date for Gal-3 in ambulatory patients with systolic HF, Gal-3 was strongly associated with poor functional status and poor outcome, as an independent predictor.³⁷ Similarly to our study, Tang and colleagues demonstrated a strong association between circulating Gal-3 and poor renal function, evaluated by eGFR and CysC, in patients with chronic systolic HF.³⁸ Gal-3 apart from myocardial, can also promote renal fibrosis, as confirmed in a recent study in an animal model.³⁹ Of note, Gal-3 is primarily cleared by the liver and therefore, the positive association of Gal-3 with CysC in our study further elucidates the cross-talk between the heart and kidneys in cardiorenal syndrome.⁴⁰ In our cohort, even though severe renal dysfunction was an exclusion criterion, results may suggest that in HF with reduced EF, Gal-3 is responsible for initiation and progression of myocardial and renal interstitial fibrosis, which is preserved by increased levels of circulating CysC, in a vicious circle. In other words, fibrosis may beget further fibrosis. However, the direct pathophysiological interactions of CysC and Gal-3 warrant further studies. Most importantly, we showed that levels of CysC and Gal-3 were independent of the presence of coronary artery disease, denoting common fibrotic mechanisms between ischaemic and non-ischaemic HF. This is in agreement with the study of Van Kimmenade and colleagues who demonstrated that the aetiology of HF, that is, ischaemic or non-ischaemic, did not play a role in the plasma levels of Gal-3 in patients with dyspnoea at the emergency department.⁴¹ It has been also demonstrated that CysC can strongly predict mortality in acute HF, irrespective of the presence of ischaemic heart disease.⁴²

Surprisingly, NT-proBNP was not correlated with LAVi in this study. This was probably due to the fact that natriuretic peptides become elevated in response to acute myocardial wall stress.⁴³ Our patients were stable regarding HF symptoms, and presumably lower myocardial wall stress was present to evoke differences in natriuretic peptide release between patients with different LA dimensions.

Limitations

The small number of patients is a major limitation of our study. This was due to the exclusion of patients with other comorbidities, mainly renal impairment and autoimmune diseases or cancer, which might have affected the levels of BMRs. Moreover, a definite causal relation between CysC, Gal-3 and structural alterations in failing hearts was not proven. Lastly, it is known that TTE underestimates LA volumes compared with cardiac MRI.⁴⁴ Intra-observer variability in TTE findings should also be mentioned.

Conclusion

In conclusion, we found that in patients with HF with severely impaired LV systolic function, greater LAVi was associated with worse LV remodelling, as assessed by LVEDV and LVESV, worse diastolic dysfunction, higher incidence of AF, more severe MR, and higher RVSP and plasma levels of CysC. However, only MRV, CysC and RVSP were significant predictors of LAVi in multiple linear regression analysis. Further analysis revealed a positive correlation between circulating CysC and plasma Gal-3, indicating that these proteins may be responsible for the initiation and perpetuation of fibrotic processes accountable for cardiorenal syndrome and HF progression. Further studies are warranted to confirm these findings. Finally, inhibition of this vicious circle through suppression of CysC and Gal-3 seems ultimately to be a promising new therapeutic target for these patients.