Left Atrial Remodeling and Atrioventricular Coupling in a Canine Model of Early Heart Failure With Preserved Ejection Fraction

Abstract

Background

Left atrial (LA) compliance and contractility influence left ventricular (LV) stroke volume. We hypothesized that diminished LA compliance and contractile function occur early during development of heart failure with preserved ejection fraction (HFpEF) and impair overall cardiac performance.

Method and Results

Cardiac magnetic resonance imaging, echocardiography, LV and LA pressure-volume studies, and tissue analyses were performed in a model of early HFpEF (elderly dogs, renal wrap-induced hypertension, exogenous aldosterone; n=9) and young control dogs (sham surgery; n=13). Early HFpEF was associated with LA enlargement, cardiomyocyte hypertrophy and enhanced LA contractile function (median active emptying fraction 16% [95% CI 13–24] vs 12[10–14]%, p=0.008; end-systolic pressure-volume relationship slope 2.4[1.9–3.2]mmHg/mL HFpEF vs 1.5[1.2–2.2]mmHg/mL controls, p=0.01). However, atrioventricular coupling was impaired and the curvilinear LA end-reservoir pressure-volume relationship was shifted upward/leftward in HFpEF (LA stiffness constant, β_LA, 0.16[0.11–0.18]mmHg/mL vs 0.06[0.04–0.10]mmHg/mL controls, p=0.002) indicating reduced LA compliance. Impaired atrioventricular coupling and lower LA compliance correlated with lower LV stroke volume. Total fibrosis and titin isoform composition were similar between groups, however titin was hyperphosphorylated in HFpEF and correlated with β_LA. LA microvascular reactivity was diminished in HFpEF versus controls. LA microvascular density tended to be lower in HFpEF and inversely correlated with β_LA.

Conclusions

In early-stage hypertensive HFpEF, LA cardiomyocyte hypertrophy, titin hyperphosphorylation and microvascular dysfunction occur in association with increased systolic and diastolic LA chamber stiffness, impaired atrioventricular coupling and decreased LV stroke volume. These data indicate that maladaptive LA remodeling occurs early during HFpEF development, supporting a concept of global myocardial remodeling.

Left atrial (LA) compliance¹ and contractility² are important determinants of left ventricular (LV) stroke volume. During LV isovolumic contraction and systole, the compliant LA acts as a reservoir for pulmonary venous return and influences early LV diastolic filling pressure. Upon mitral valve opening, LA conduit and booster pump (contractile) functions enable passive and active LV filling respectively. In the setting of LV diastolic dysfunction, the ability to augment LA reservoir capacity and booster pump function becomes critical for preserving LV filling and cardiac output³.

Heart failure (HF) with preserved ejection fraction (HFpEF) commonly presents with LV diastolic dysfunction and LA chamber enlargement. Recent human studies have also proposed a role for LA dysfunction in HF development^{4, 5} and uniquely in HFpEF pathophysiology^6–9, though corroborative invasive and histological assessment of LA myocardial properties have been lacking. Moreover, while LA function is recognized to decline early in the course of some dilated cardiomyopathies, consistent with a primary atrial myopathy¹⁰, LA remodeling and dysfunction in HFpEF has traditionally been considered a late sequelae of hypertension, LV diastolic dysfunction, and chronic LA pressure overload^{6, 9}. By contrast, Paulus and Tschöpe have proposed a novel paradigm for HFpEF pathophysiology implicating coronary microvascular endothelial inflammation and altered intra-myocardial signaling as seminal mechanisms¹¹. These are inherently not chamber-specific.

Based on this discord, we proposed a hypothesis that maladaptive alterations in LA structure and function could, in fact, be identified at an early stage of HFpEF development and would impair overall cardiac performance. To test this hypothesis, LA and LV properties and left atrioventricular coupling were defined using echocardiography, cardiac magnetic resonance imaging (MRI) and invasive pressure-volume loop analyses in a canine model of early-stage hypertensive HFpEF induced by aging, renal wrap-induced hypertension and pro-inflammatory aldosterone excess. Chamber specific fibrosis, myocyte hypertrophy, titin phosphorylation, and microvascular structure and endothelial function were assessed to elucidate pathophysiological mechanisms.

Methods

Animal model

Nine elderly (8–13 years, 25.5 [19.5–27.6] kg, 2 male, 7 female) dogs underwent bilateral renal wrapping, which is an established model of experimental hypertension¹² (Supplemental methods i). Five weeks after surgery, intramuscular desoxycorticosterone pivalate (DOCP, 1.4mg/kg) was administered to accentuate inflammation and model transition to HFpEF^{13, 14}. Young control dogs (n=13; age ≈1 year, 23.9 [22.5–25.3] kg, 7 male, 6 female) underwent sham surgery alone. On judicious review of published data, young and elderly dogs without hypertension exhibit comparable LV dimensions, indexed mass, and fibrosis¹⁵. Based on the restricted availability of aged dogs, young dogs were therefore selected to undergo sham surgery, acknowledging the potential limitation. All animals were maintained in accordance with National Institutes of Health guidelines for the “Care and Use of Laboratory Animals”¹⁶. The study protocol was approved by the Mayo Clinic Institutional Animal Care and Use Committee.

Non-invasive imaging

All dogs underwent two-dimensional transthoracic echocardiography (Philips, 3.5MHz transducer) in the conscious (standing) state prior to surgery and after 8 weeks (Supplemental methods ii).

Magnetic resonance imaging (MRI; 1.5T scanner, GE Healthcare, USA) was used to measure LA and LV chamber volumes and function at week 8 (Supplemental methods iii). LA volume-time curves were constructed to determine LA maximum, minimum, mid-diastolic (end of rapid emptying), and pre-atrial contraction (pre-A) volumes¹⁷. LA reservoir, stroke, and conduit volumes, and LA function parameters were calculated according to standard equations (Supplemental Table 1).

In vivo hemodynamic study

Open chest, open pericardium hemodynamic studies were performed at week 8 under anesthesia (IV propofol, 2–6mg/kg, followed by inhaled isoflurane, 0.5–2.5%). Mid-ascending aortic pressure and LV and LA pressure-volume loops were measured using 7F manometer tipped pressure and 7F (LV) or 5F (LA) ADVantage admittance catheters (SciSense, Ontario, Canada). Hydraulic occluders were placed around the superior and inferior vena cava to allow acute preload reduction. Intravenous dextran was administered prior to haemodynamic measurements to achieve a LV end-diastolic pressure ≈20 mmHg. All data were acquired with ventilation suspended at end-expiration and atrial pacing at 10–15bpm above the sinus rate (Supplemental methods iv).

Pressure-volume analysis

LA end-systolic volume (LAV_ES) was taken as the minimum LA volume at end atrial contraction. Its corresponding pressure defined LA end-systolic pressure (LAP_ES). LA end-reservoir volume (LAV_ER) and pressure (LAP_ER) were taken as the maximum LA volume and pressure immediately before mitral valve opening.

LA myocardial contractility was evaluated using a time-varying elastance model^{18, 19}. LAP_ES and LAV_ES data from variably loaded pressure-volume loops were fit to the linear regression equation: LAP_ES=E_LA (LAV_ES -V_0s), where the slope of the end-systolic pressure-volume relation represents LA systolic elastance (E_LA) and V_0s is the extrapolated volume intercept of the relation. The coefficient of determination, R², was used to describe goodness-of-fit for pressure-volume data to the linear model. Curvilinearity over the measured range of LA pressure and volumes yielded lower R² values.

LA compliance was assessed from LA end-reservoir data during vena cava occlusion. LAP_ER and LAV_ER were fit to the mono-exponential equation LAP_ER=αe^β*LAV_ER where α reflects the y-intercept (curve-fitting constant), e is the base of the natural logarithm, and β represents the dynamic stiffness constant (modulus of chamber stiffness). Data goodness-of-fit were assessed using R². LV pressure-volume loops were analysed as previously described (Supplemental methods iv)²⁰.

At the end of the study, full thickness LV and LA biopsies were obtained from the beating heart and flash frozen in liquid nitrogen. Dogs were euthanized with intravenous potassium chloride under deep anesthesia, consistent with guidelines of the Panel on Euthanasia of the American Veterinary Medical Association. Harvested myocardial tissue was flash-frozen in liquid nitrogen and stored at minus 80°C. Additional sections were immediately fixed in 10% formalin solution and embedded in paraffin.

Fibrosis assessment

Transmural cross-sections of LA and LV tissue 4μm thick were stained with Picrosirius Red to demonstrate extracellular matrix (Supplemental methods v) and scanned by whole field digital microscopy. Interstitial fibrosis was expressed histomorphometrically as a percentage of total tissue area excluding endocardial, epicardial and perivascular regions (ImageJ, National Institutes of Health, Bethesda, MD) in the LA anterior free wall (LAFW), LA appendage (LAA), and LA posterior wall adjacent to the pulmonary veins (LAPW). Myocardial collagen content was determined using the hydroxyproline assay (Supplemental methods v).

Immunostaining and fluorescence microscopy

Cardiomyocyte cross sectional area was determined by manual planimetry (ImageJ, National Institutes of Health, Bethesda, MD) from 4μm thick sections of paraffin-embedded LA and LV myocardial tissue stained with fluorescein isothiocyanate-conjugated wheat germ agglutinin (FITC-WGA; Vector Laboratories, Burlingame, CA), to delineate cell membranes (Supplemental methods vi). Minimum dimension was calculated as perpendicular to the delineated contour of the cell. A minimum of 100 cardiomyocytes were measured per section.

Titin isoform expression and phosphorylation status

Titin isoform expression was determined by SDS-PAGE and densitometry. Total phosphoprotein staining (ProQ diamond) was used to assess total titin phosphorylation (Supplemental methods vii).

In vitro microvascular function

LA microvessels were isolated from freshly harvested tissue and subjected to physiological levels of shear stress, as described previously²¹, in the presence or absence of: intact endothelium, the endothelial nitric oxide synthase inhibitor N(ω)-nitro-L-arginine methyl ester (L-NAME), and iberiotoxin, an inhibitor of large conductance Ca²⁺-activated K⁺ (BK) channels involved in nitric oxide-mediated shear stress induced vasodilation (SSD; Supplemental methods viii).

Microvascular density

Paraffin-embedded myocardial sections were stained with the biotinylated endothelial cell-specific marker isolectin B4 (Vector Laboratories, Burlington, VT). Microvascular density was quantified on whole field digital micrographs using an automated color-detection algorithm (Definiens Tissue Studio 3.5, Definiens®, Germany, Supplemental methods ix). Microvessels (small pre-capillary arterioles) were defined by a cross sectional area of 78.5–314μm² and average luminal diameter of 10–20μm.

Statistics

Continuous variables are presented as median [25^th–75^th percentile]. Based on the sample size, the Wilcoxon Rank Sum (non-parametric) test was used to compare characteristics between HFpEF and control dogs. The Wilcoxon Signed Rank test was used for within-group comparisons of echocardiographic data at baseline (pre-surgery) versus week 8. Linear correlations between two variables were assessed using Pearson’s correlation coefficient, r. Goodness of fit/variance for pressure-volume data to regression models was assessed by the coefficient of determination, R². p<0.05 (2-sided) was considered statistically significant. Analyses were performed using JMP, version 9.0 (SAS Institute Inc, Cary, NC).

Results

Model development

Weekly post-operative blood pressures were higher in HFpEF dogs than controls (Figure 1). Compared with baseline echocardiography (pre-surgery), HFpEF dogs at week 8 maintained a normal ejection fraction (≥50%), however relative wall thickness and indexed LV mass increased and the LV diastolic dimension was diminished, indicating development of concentric LV hypertrophy (Supplemental Table 2). LA area also increased (baseline, 74 [61–87]mm² versus week 8, 90 [81–102]mm², p=0.008). Control dogs did not exhibit significant echocardiographic changes between baseline and week 8.

Figure 1. Mean arterial pressure over study duration.

HFpEF, heart failure with preserved ejection fraction; DOCP, desoxycorticosterone pivalate.

Left ventricular structure and function

Cardiac MRI (at week 8) revealed greater LV mass and LV mass to end-diastolic volume ratio in HFpEF compared with controls, while ejection fraction and cardiac output were similar between groups (Table 1). In vivo hemodynamic assessment revealed higher LV systolic (Ees) and LV diastolic (β_LV) elastance in HFpEF versus controls and a trend towards higher arterial elastance (Ea; p=0.057) (Supplemental results i, Supplemental Table 3).

Table 1.

Left ventricular (LV) and left atrial (LA) structure and function by magnetic resonance imaging (week 8).

Variable	Control (n=10)	HFpEF (n=7)	p-value
LV structure and function
Heart rate, bpm	100 (91–100)	101 (90–115)	0.88
LV end-diastolic volume, mL	46.6 (44.1–52.3)	34.3 (33.3–49.9)	0.077
LV end-systolic volume, mL	21.9 (18.2–24.4)	19.6 (11.5–24.3)	0.30
LV stroke volume, mL	25.5 (23.4–28.4)	21.1 (15.2–28.4)	0.08
LV mass, g	86.3 (78.8–92.3)	99.2 (89.5–129.6)	0.015
LV mass/end-diastolic volume, g/mL	1.9 (1.6–2.1)	2.3 (2.2–3.4)	0.0007
Cardiac output, L/min	2.6 (2.2–3.0)	1.9 (1.5–3.2)	0.22
LV ejection fraction, %	55.0 (52.1–58.8)	61.0 (40.3–66.0)	0.92
LA structure
Maximum volume, mL	22.9 (19.7–25.4)	27.0 (24.5–28.6)	0.019
Pre-A volume, mL	18.1 (15.8–20.7)	22.8 (21.3–23.7)	0.0018
Minimum volume, mL	15.9 (14.3–17.6)	18.5 (16.5–19.8)	0.015
Reservoir volume, mL	7.4 (5.7–7.9)	8.4 (5.9–8.7)	0.19
Passive emptying volume, mL	4.8 (3.9–5.6)	4.2 (2.9–4.9)	0.24
LA stroke volume, mL	2.1 (1.8–2.9)	3.9 (3.4–5.2)	0.0029
Conduit volume, mL	24.5 (21.5–30.9)	17.5 (14.5–30.1)	0.28
LA function
Total emptying fraction, %	31.3 (27.9–32.2)	30.8 (24.3–34.7)	0.92
Passive emptying fraction, %	21.4 (17.2–22.3)	15.4 (11.6–17.2)	0.019
Active emptying fraction, %	12.2 (9.8–13.8)	16.4 (13.4–24.2)	0.0084
Active ejection rate, mL/s	27.1 (16.5–41.8)	46.9 (42.3–51.3)	0.0047
HR corrected ejection rate, %/s	1.3 (0.7–1.5)	1.6 (1.3–2.1)	0.040
LA kinetic energy, kdynes·cm	0.56 (0.42–0.69)	1.46 (1.10–1.76)	0.0011
LA contribution to LV stroke volume
Reservoir, %	21.2 (16.7–24.4)	25.3 (19.4–37.2)	0.17
Active, %	6.8 (5.8–8.0)	13.1 (8.9–22.3)	0.0009
Conduit, %	78.8 (75.6–83.3)	74.7 (62.8–80.6)	0.17

Data presented as median (25^th–75^th percentile)

Left atrial geometry and phasic function

By MRI, at equivalent heart rate, LA volumes were greater in HFpEF than controls (Figure 2A–B) and positively correlated with the LV mass to end diastolic volume ratio (r=0.62 to 0.70, p<0.01 for all). LA reservoir volume was similar between HFpEF and controls (Table 1). LA stroke volume was significantly greater in HFpEF, while LA passive emptying volume was numerically, but not significantly, lower in HFpEF than controls. Accordingly, total LA emptying fraction was similar between groups due to increased LA active but reduced passive emptying fractions in HFpEF (Table 1). In both groups, LA conduit volume contributed the most to LV filling. Estimated LA kinetic energy expenditure²² and heart rate-corrected mean LA ejection rate²³, were higher in HFpEF, suggesting greater LA stroke work and inotropy (Table 1). These findings indicate augmented LA active (booster pump) function in early-stage HFpEF without a concomitant increase in LA reservoir function.

Figure 2. Left atrial (LA) volumes assessed by magnetic resonance imaging.

A, Example volume-time curves (a, maximum; b, mid-diastolic; c, pre-atrial contraction [pre-A]; d, minimum LA volume). B, LA volumes (median, 75^th percentile displayed). HFpEF, heart failure and preserved ejection fraction.

Left atrial booster pump function and systolic atrioventricular coupling

Representative LA pressure-volume loops are shown in Figures 3A–B. LA pressure-volume data during vena cava occlusion demonstrated a linear LA end systolic pressure-volume relationship (R²=0.96 to 0.99, Figure 3C). At matched heart rate and mean LA pressure, the LA end systolic pressure-volume relationship slope (E_LA) was steeper and the volume axis intercept (V₀) was greater in HFpEF dogs compared with controls (Figure 3D, Table 2). When normalized for preload (i.e. LA end reservoir volume), E_LA remained significantly greater in HFpEF, confirming increased LA systolic elastance (Table 2).

Figure 3. In vivo hemodynamic assessment.

A to D, representative left atrial (LA) pressure-volume loops obtained from control and HFpEF dogs at matched mean LA pressure: A and B, steady state; C and D, during acute preload reduction (see text for definition of terms). E, correlation between LA systolic elastance (E_LA), left ventricular systolic elastance (Ees) and effective arterial elastance (Ea). F, correlation between left ventricular stroke volume (LVSV), LA-LV physiological coupling (β_LV/E_LA, left graph), and LA-LV contractile coupling (Ees/E_LA, right graph). See text for further explanation.

Table 2.

Hemodynamic assessment of left atrial (LA) function and left atrioventricular coupling.

Variable	Control (n=12)	HFpEF (n=8)	p-value
Heart rate, bpm	127 (110–132)	126 (119–147)	0.59
Mean LA pressure, mmHg	13.4 (11.9–18.5)	13.7 (9.0–19.7)	0.64
LA active function
Peak A, mmHg	15.9 (13.2–22.2)	19.2 (10.4–25.3)	1.00
ELA, mmHg/mL	1.53 (1.17–2.19)	2.40 (1.86–3.20)	0.011
ELA R value	0.97 (0.95–0.98)	0.98 (0.95–0.99)	0.54
V0, mL	7.5 (6.1–10.4)	14.0 (10.0–19.7)	0.0055
Normalized ELA*, mmHg	35.8 (26.9–47.9)	66.6 (57.1–85.5)	0.002
A wave relaxation slope, mmHg/mL	0.04 (0.03–0.07)	0.07 (0.03–0.17)	0.25
A wave deceleration time, ms	103 (81, 125)	94 (79, 136)	0.82
LA reservoir function
Peak V, mmHg	17.2 (14.0–22.0)	17.9 (11.5–23.9)	0.88
Atrial pulsatility, mmHg	5.7 (4.6–6.7)	6.4 (5.2–10.1)	0.25
βLA, mmHg/mL	0.063 (0.044–0.101)	0.164 (0.108–0.183)	0.0020
βLA R value	0.96 (0.95, 0.98)	0.96 (0.94–0.98)	0.88
LA-LV coupling
Ees/ELA	0.66 (0.41–1.35)	1.34 (0.99–1.63)	0.0069
βLV/ELA	0.010 (0.066–0.013)	0.022 (0.0084–0.028)	0.076

Data presented as median (25^th–75^th percentile)

^*Normalized for preload (LA end reservoir volume)

E_LA was positively correlated with both LV systolic (Ees) and arterial (Ea) elastances (Figure 3E). LV Ees and Ea were also correlated (r=0.6, p=0.038) consistent with in series left-sided coupling. The ratio of LV to LA systolic elastances (Ees/E_LA) was 0.66 [0.41–1.01] in control dogs, indicating greater LA than LV systolic elastance in the normal state (Table 2). In HFpEF, the Ees/E_LA ratio was higher, 1.34 [0.99–1.63], p=0.0069) indicating that atrial systolic stiffening does not match LV systolic stiffening in early-stage HFpEF. Impaired systolic atrioventricular coupling (higher Ees/E_LA) was associated with lower LV stroke volume (Figure 3F).

Left atrial compliance and atrioventricular coupling

End reservoir LA pressure-volume data displayed a curvilinear relationship over the measured range (Figure 3C). Compared with controls, the LA end reservoir pressure-volume relationship was shifted leftward and upward in HFpEF (Figure 3D), i.e. the LA chamber stiffness constant (β_LA) was greater (Table 2), consistent with reduced LA compliance. Furthermore, β_LA was inversely correlated with LV stroke volume (r=−0.46, p=0.04) suggesting that increased LA stiffness limits LV performance in HFpEF. Notably, β_LA was not significantly correlated with maximum (p=0.24), pre-A wave (p=0.10), or minimum (p=0.09) LA volume by MRI.

As the LA faces LV diastolic stiffness during atrial contraction, physiological left atrioventricular coupling may be depicted as the ratio of LV diastolic stiffness (β_LV) to E_LA. β _LV/E_LA was numerically higher (p=0.076) in HFpEF than controls (Table 2). A higher β _LV/E_LA ratio was significantly associated with lower LV stroke volume (Figure 3F).

Left atrial and ventricular hypertrophy and fibrosis

Autopsy LV mass index was numerically greater but not significantly different in HFpEF dogs compared with controls (Figure 4A). However, the ratio of LV mass (autopsy) to LV end diastolic volume (MRI; Figure 4B) and LV cardiomyocyte cross-sectional area (Figure 4C–D) were significantly increased indicating concentric LV hypertrophy.

Figure 4. Left atrial and left ventricular remodeling.

A. Autopsy mass indexed to body weight. B. Autopsy mass indexed to MRI-assessed diastolic volume. C. Representative sections stained with wheat germ agglutinin. Scale bar represents 50μm. D, Cardiomyocyte dimensions. CSA, cross sectional area.

Autopsy LA mass index (Figure 4A) and LA cardiomyocyte cross-sectional area (Figure 4C–D) were greater in HFpEF dogs compared with controls. The ratio of LA mass (autopsy) to LA end-reservoir volume (MRI; Figure 4B) was conserved between groups, indicating eccentric LA hypertrophy.

In the LAFW, LAA and LV, there were no significant differences in interstitial fibrosis or collagen content, between HFpEF and controls (Figure 5A–C). However, greater regional fibrosis was evident in the LAPW of HFpEF dogs versus controls (Figure 5A, C). LA fibrosis exceeded LV fibrosis in all LA regions assessed. There was no correlation between LA or LV percent fibrosis and LA or LV chamber stiffness constants (p>0.35 for all).

Figure 5. Fibrosis.

A. Percent interstitial fibrosis. B. Collagen content. C, Representative sections stained with Picrosirius Red. LAA, left atrial appendage; LAFW, left atrial free wall (anterior); LAPW, left atrial posterior wall; LV, left ventricle.

Titin isoform expression and phosphorylation status

Titin isoform expression was similar between groups in the LAFW, LAA and LV (Figure 6A–B). In the LAFW, both N2B (stiffer) and N2BA (compliant) titin isoforms were hyperphosphorylated in HFpEF versus controls (Figure 6C) and the extent of phosphorylation correlated with LA chamber stiffness (β_LA vs. N2B %phosphorylation, r=0.54, p=0.03; β_LA vs. N2BA %phosphorylation, r=0.49, p=0.05). Conversely, in the LAA, isoform N2B was hypophosphorylated in HFpEF and N2BA similar between groups (Figure 6C). Neither status correlated with LA chamber stiffness.

Figure 6. Titin status.

A, Representative gels comparing titin isoform expression and phosphorylation status for control and HFpEF left ventricular myocardium. B, Titin isoform expression ratio (box plot shows median, 25^th to 75^th percentile). C, Titin phosphorylation status for N2B and N2BA titin isoforms. LAA, left atrial appendage; LAFW, left atrial free wall; LV, left ventricle.

In the LV, both N2B and N2BA titin isoforms were hyperphosphorylated in HFpEF versus controls (Figure 6C) and the degree of phosphorylation correlated with LV chamber stiffness (β_LV vs. N2B %phosphorylation, r=0.48, p=0.04; β_LV vs. N2BA %phosphorylation, r=0.52, p=0.03).

Left atrial microvessel endothelial function

Physiological shear stress produced graded dilatation of LA microvessels from both groups, however SSD was significantly attenuated in microvessels from HFpEF dogs compared with controls (Figure 7A), suggesting reduced shear stress-mediated vasodilator generation in HFpEF. Removal of the endothelium markedly attenuated SSD in both groups (Figure 7A). Inhibition of endothelial nitric oxide synthase with L-NAME virtually abolished SSD in control dogs but only partially inhibited the vasodilator response in HFpEF vessels (Figure 7B), indicating impaired NO signaling in HFpEF. Blockade of BK channels with iberiotoxin, almost completely abolished SSD in control vessels but had only a marginal effect in microvessels from HFpEF dogs (Figure 7C), suggesting that SSD in HFpEF microvessels is not BK channel dependent.

Figure 7. Microvessel in vitro reactivity and rarefaction.

A–C, Shear stress-induced vasodilation (SSD) of LA microvessels in the presence and absence of: A, intact endothelium, expressed as a percentage of maximal vessel dimension in the relaxed state. (-Endo signifies endothelium denuded, *p<0.05 Control vs. Control-Endo, # p<0.05 Control vs. HFpEF, +p<0.05 HFpEF vs. HFpEF-Endo [n=5 vessels assessed for each group]); B, endothelial nitric oxide synthase inhibition with N(w)-nitro-L-arginine methyl ester (L-NAME; *p<0.05 Control vs. Control+L-NAME, # p<0.05 Control vs. HFpEF, +p<0.05 HFpEF vs. HFpEF+L-NAME [n=13, control and HFpEF; n=9, control+L-NAME and HFPEF+L-NAME]); C, Iberiotoxin (IBTX, an inhibitor of large conductance Ca²⁺-activated K⁺ channels; *p<0.05 Control vs. Control+IBTX. # p<0.05 Control vs. HFpEF [control and control+IBTX, n=8 vessels each; HFpEF and HFpEF+IBTX, n=9 vessels each]). D, Correlation between LA microvessel density (MVD) and LA diastolic stiffness (β_LA).

Microvessel density

LA microvessel density was lower in HFpEF dogs than controls (2294 [1590–2712] vessels/mm² HFpEF vs. 2694 [2245–2919] vessels/mm² controls; p=0.068) and inversely correlated with the LA chamber stiffness constant, β_LA (Figure 7D). LV microvessel density was also lower in HFpEF than controls (2678 [2485–2957] vessels/mm² HFpEF vs. 3220 [3027–3399] vessels/mm² controls; p=0.0094) but did not significantly correlate with the LV chamber stiffness constant, β_LV.

Discussion

The objective of this study was to determine if maladaptive LA structural and functional remodeling was present in a canine model reflecting early-stage HFpEF comprising advanced age, renal-induced hypertension, and pro-inflammatory aldosterone excess. Compared with controls, dogs with early hypertensive HFpEF demonstrated LA chamber and cardiomyocyte hypertrophy, increased LA booster pump function and greater LA contractility. However, there was no attendant increase in LA reservoir function in HFpEF, as may be expected to compensate for the degree of LV concentric remodeling and diastolic dysfunction. Furthermore, LA compliance was reduced in HFpEF versus controls and atrioventricular coupling was impaired, in association with lower LV stroke volume. Thus, we have demonstrated that LA remodeling occurs early in HFpEF development and is maladaptive. Moreover, we provide novel evidence of LA myocyte hypertrophy, titin hyperphosphorylation and endothelium-dependent LA microvascular dysfunction occurring in early HFpEF, concomitant with LV remodeling. Collectively, these data support the hypothesis that common pathophysiological processes drive global (all-chamber) myocardial remodeling in HFpEF¹¹.

Presence of an atrial myopathy and impaired left atrioventricular coupling in early-stage HFpEF

Left atrial enlargement is common in HFpEF and correlates with the severity and duration of LV diastolic dysfunction²⁴. Accordingly, LA remodeling in HFpEF has been considered a late and adaptive response to increased LA afterload^{6, 9}, facilitating greater LA preload and augmented LA active emptying fraction, as per an atrial Frank-Starling mechanism²⁵. In overt HFpEF⁶, as in some dilated¹⁰ and hypertrophic²⁶ cardiomyopathies, this compensatory LA contractile response is blunted. In the present study of early-stage HFpEF, however, we did not find evidence of LA contractile failure. Rather, LA hypertrophy was accompanied by augmented booster pump function and inotropy relative to controls, including when normalised for LA volume. Even so, the enhanced LA inotropy was insufficient to match increases in LV end-systolic elastance or adequately compensate for increased LV diastolic stiffness in HFpEF, thereby giving rise to impaired systolic (higher Ees/E_LA) and physiological (higher β_LV/E_LA) atrioventricular coupling. That both parameters of atrioventricular coupling inversely correlated with LV stroke volume suggests a key role for efficient mechanical matching between the LA and LV in early-stage HFpEF. Similarly abnormal atrioventricular coupling has been observed in patients with a recent myocardial infarction and in patients with HF and reduced ejection fraction (HFrEF)²⁷; this study represents the first invasive assessment in HFpEF.

Recent human HFpEF studies have also reported a reduction in LA compliance elicited on strain echocardiography^{7, 9} or, in a single study, from steady-state LA pressure-volume co-ordinates (maximal v wave height minus a wave nadir divided by echocardiography-derived LA reservoir volume)²⁸. Herein we provide verification of an upward and leftward shift in the LA end-reservoir pressure-volume relationship, elicited during preload reduction in experimental HFpEF (i.e. greater chamber stiffness or reduced compliance) versus controls, which negatively affected overall cardiac performance. This finding contrasts previously reports of LA adaptation in mild diastolic dysfunction³, essential hypertension without HF²⁹, and LA volume overload in chronic mitral regurgitation³⁰, where LA compliance and reservoir function increase to preserve LV filling and buffer elevated LA pressures.

Potential mechanisms for LA stiffening in HFpEF include greater fibrosis, increased cardiomyocyte stiffness (principally related to the sarcomeric protein titin), or LA myocardial ischemia. We did not observe an increase in LAFW or LV fibrosis in this model, consistent with other reports of LA fibrosis as a late finding^{31, 32}. The observed increase in localized peri-pulmonary vein (LAPW) fibrosis may signify the onset of a fibrotic response and substrate for AF genesis, which is frequent in HFpEF³³. Similarly, LA and LV titin isoform composition were similar between HFpEF and controls, however LA and LV total-titin phosphorylation was increased in HFpEF, and N2B phosphorylation (the shorter stiffer isoform) modestly correlated with LA and LV chamber stiffness coefficients. Although site-specific titin phosphorylation and passive cardiomyocyte tension were not directly assessed, protein kinase Cα (PKCα)-mediated hyperphosphorylation of the titin proline, glutamate, valine and lysine (PEVK) site is recognized to increase passive cardiomyocyte stiffness^34–36 and is a possible mechanism for this finding. Our corroborative demonstration of impaired NO signaling in HFpEF, renders titin hyperphosphorylation less likely to result from cGMP-mediated mechanisms. Since LA and LV myocardium exhibited the same pattern, a chamber-specific response is also excluded and supports a process of global (all-chamber) adverse myocardial remodeling in HFpEF.

Lastly, attenuation of microcirculatory reserve has been associated with reduced LA compliance in experimental HFrEF³⁷. LA microvascular dysfunction, with impaired NO signaling, and microvascular rarefaction, as seen in this study, may contribute to LA diastolic dysfunction in early-stage HFpEF. Importantly, the inverse correlation between LA microvessel density and LA chamber stiffness, which was not observed in the LV, suggests that the LA is both vulnerable, and potentially more sensitive than the LV, to the effects of microvascular ischemia.

Inflammation and endothelial dysfunction in early-stage HFpEF

It is hypothesized that comorbidity-driven systemic and coronary microvascular inflammation results in impaired NO-cGMP signaling as the impetus to myocardial fibrosis and cardiomyocyte stiffening in HFpEF¹¹. Our experimental model incorporates renal-wrap induced perinephritis and tubulointerstitial inflammation³⁸, along with mineralocorticoid excess which is pro-inflammatory in cardiac tissue¹³. In this setting, we demonstrated impaired endothelium-dependent LA microvascular vasodilator capacity. As L-NAME did not completely block SSD in HFpEF dogs, we cannot rule out concomitant up-regulation of an alternative endothelium-derived vasodilating factor, and therefore further studies are required to elucidate the precise mechanism of SSD in HFpEF, to include age-matched controls. The additional presence of microvascular rarefaction, however, supports a role for microvascular inflammation in the development of HFpEF.

Limitations

Our experimental model incorporates well-established features of human HFpEF including advanced age, hypertension, and pro-inflammatory mineralocorticoid excess, thus our findings are relevant for a ‘hypertensive’ HFpEF phenotype. Additional studies are required to evaluate LA parameters in HFpEF subjects without LV concentric remodeling. We recognize the limitation of not studying an elderly control group, however our previous examination of this model exhibited similar characteristics between young and elderly control dogs, with respect to LV and vascular remodeling and diastolic function, and distinct from dogs with experimental HF¹⁵. Thus we believe our findings represent a significant effect of experimental HF, over and above that of aging alone. Importantly, elderly dogs available for research are frequently retired breeders and therefore our sample is predominantly female. Published data demonstrate that gender differences in LA size are almost completely accounted for by body size³⁹, however further examination of sex differences in LA function is warranted.

Conclusions

In the setting of hypertensive HFpEF, LA remodeling occurs early and exhibits maladaptive alterations in LA compliance and left atrioventricular coupling which compromise overall cardiac performance and may exacerbate increases in LA and pulmonary pressures^{40, 41}. Early development of LA non-compliance and atrioventricular mismatch may underlie reported associations between LA dysfunction and future HF in populations with preserved LV ejection fraction^{4, 5}. Pathophysiological mechanisms, including myocardial hypertrophy, titin hyperphosphorylation and microvascular rarefaction affect LA and LV myocardium concomitantly, even in the absence of prolonged atrial hypertension or marked fibrosis, thus supporting a paradigm of global (all-chamber) adverse myocardial remodeling in HFpEF. Novel therapeutic strategies targeting these mechanisms may be able to lessen or reverse LA and LV remodeling in HFpEF and warrant investigation in human studies.

Clinical Perspective.

Left atrial (LA) compliance and contractility are important determinants of left ventricular (LV) stroke volume. Preliminary human studies have proposed LA dysfunction in heart failure with preserved ejection fraction (HFpEF); however, corroborative invasive and histological data have been lacking. Furthermore, it is not known whether LA dysfunction represents a late sequela of LV diastolic dysfunction or early feature of global myocardial remodeling in HFpEF. In a canine model reflecting early-stage HFpEF comprising advanced age, renal-induced hypertension, and aldosterone excess, we observed LA chamber and myocyte hypertrophy and enhanced LA contractile function compared with controls. However, left atrioventricular coupling was impaired (in sinus rhythm) and LA compliance was reduced, in association with lower LV stroke volume. Pathophysiological alterations including myocyte hypertrophy, titin hyperphosphorylation, and microvascular rarefaction were congruous between LA and LV myocardium, suggesting early and global (i.e. non-chamber specific) maladaptive remodeling. In HFpEF, development of LA non-compliance may contribute to increases in LA and pulmonary pressures, predisposing to dyspnoea, pulmonary hypertension and right ventricular failure. Novel therapeutic strategies targeting early global remodeling mechanisms may modify LA and LV properties and warrant investigation in human studies.