Effect of Endurance Exercise Training on Endothelial function and Arterial Stiffness in Older Patients with Heart Failure and Preserved Ejection Fraction: A Randomized, Controlled, Single-Blind Trial

Abstract

Objectives

To evaluate the effects of endurance exercise training (ET) on endothelial dependent flow-mediated arterial dilation (FMD) and carotid artery stiffness and their potential contributions to the training-related increase in peak exercise oxygen consumption (VO₂) in older patients with heart failure with preserved ejection fraction (HFPEF).

Background

Elderly HFFEF patients have severely reduced peak VO₂ which improves with ET, however the mechanisms of this improvement are unclear. FMD and arterial distensibility are critical components of the exercise response and are reduced with aging. However, it's unknown whether these improve with ET in elderly HFPEF or contribute to the training-related improvement in peak VO₂.

Methods

63 HFPEF patients (70±7 years) were randomized to 16 weeks of ET (walking, arm and leg ergometry, n=32) or attention control (CT; n=31). Peak VO₂, brachial artery FMD in response to cuff ischemia, carotid artery distensibility by high-resolution ultrasound, LV function, and QOL were measured at baseline and follow-up.

Results

ET increased peak VO₂ (ET: 15.8±3.3 vs. CT: 13.8±3.1 ml/kg/min, p=0.0001) and QOL. However, brachial artery FMD (ET: 3.8±3.0% vs. CT: 4.3±3.5%, p=0.88), and carotid arterial distensibility (ET: 0.97±0.56 vs. CT: 1.07±0.34 × 10^-3mm × mmHg^-1 p=0.65) were unchanged. Resting LV systolic and diastolic function were unchanged by ET.

Conclusions

In elderly HFPEF patients, 16 weeks of ET improved peak VO₂ without altering endothelial function or arterial stiffness. This suggests that other mechanisms, such as enhanced skeletal muscle perfusion and / or oxygen utilization, may be responsible for the ET-mediated increase in peak VO₂ in older HFPEF patients.

INTRODUCTION

Approximately 50% of older heart failure (HF) patients have preserved left ventricular ejection fraction (HFPEF)^1,2 and their primary symptom is severe exercise intolerance, measured objectively as reduced peak exercise oxygen uptake (VO₂peak)^3-9. Reduced peak and reserve arteriovenous oxygen difference (A-VO₂Diff) are important contributors to the reduced VO₂peak in HFPEF^5,8. Furthermore, improved A-VO₂Diff accounts for most of the increase in VO₂peak following endurance exercise training (ET)¹⁰. This suggests roles for impaired arterial and/or skeletal muscle function in the severe exercise intolerance in HFPEF and its improvement with ET¹¹.

In HFPEF, arterial stiffness is increased and endothelial function may be decreased, and both may contribute to exercise intolerance^6,7,12-15. In HF with reduced ejection fraction (HFREF), ET improves endothelial function and arterial compliance^16-22 and the former is correlated with the ET-related improvement in VO₂peak^18,23. Therefore, we performed a randomized, controlled, single-blind trial to test the hypothesis that ET would enhance endothelial function and arterial distensibility in HFPEF and that these would be important contributors to the improvement in VO₂peak.

METHODS

Study Protocol

The study was approved by the Wake Forest School of Medicine Institutional Review Board for Protection of Human Subjects and registered (NCT01113840). During the screening visit, informed consent was obtained and participants were familiarized with the testing environment and procedures. During a single subsequent visit, all outcome measures were obtained. Subjects were then randomized to 16 weeks of ET or attention control (CT). Exercise performance, echocardiography, and quality-of-life (QOL) were assessed at baseline and following the 16-week intervention. Brachial artery flow mediated arterial dilation (FMD) and carotid arterial distensibility were measured at baseline, 8 and 16 weeks. Testing was performed and results were analyzed by individuals blinded to patient group. Baseline characteristics have been reported^15,24.

Subjects

HFPEF patients were identified from fortified search lists^{4,5,10,15,25,26} and had symptoms and signs of HF defined by NHANES HF score ≥3 and the criteria of Rich et al²⁷, LV ejection fraction ≥50%, no segmental wall motion abnormalities, and no significant ischemic or valvular heart disease, pulmonary disease, anemia, or other disorder that could explain the patients’ symptoms^4,5,26. Because of the profound impact of atherosclerosis on arterial function²⁸, patients with hyperlipidemia, cigarette smoking, coronary, cerebrovascular, or peripheral arterial disease were excluded by history, examination, exercise echocardiography, and vascular ultrasound^15,24.

Echocardiography

Doppler-echocardiograms were performed and analyzed as previously described^4,5,10,24,26. Doppler LV filling patterns were categorized using E/A ratio, early deceleration time, and normal reference ranges for age^29,30.

Exercise testing

Exercise testing with expired gas analysis was performed as previously described in the upright position on an electronically braked bicycle using a staged protocol to exhaustion^{4,5,10,15,24-26,31}. Peak values were averaged from the last 2 15-second intervals during peak exercise^{4,5,10,24-26,31}. Ventilatory anaerobic threshold (VAT) and ventilation per carbon dioxide (Ve/VCO₂) slope were assessed as previously described^25,26,32. A six-minute walk test was performed by the method of Guyatt^25,26,32,33.

Brachial artery FMD

Brachial FMD was measured as previously described in our laboratory and in accordance with international standards^24,35-36. Participants were examined in the post-prandial state after 15 minutes of quiet, supine rest using a Biosound Phase II ultrasound system with a 9 mHz transducer to record images of the right brachial artery. A forearm cuff was then inflated 50 mm Hg above systolic pressure for 4 minutes. Images were recorded during the final 30 seconds prior to and 3 minutes following rapid cuff deflation^32,35.

Video frames were automatically digitized and analyzed using previously described techniques ^24,35. The maximum diameter from the dilation phase was automatically determined, and the change in diameter and percent change were calculated as follows: absolute change = maximum minus baseline; percent change = 100 × absolute change/baseline. Brachial FMD by these techniques is highly reproducible in our laboratory³⁵.

Arterial stiffness

As previously described, ten-second images of the left common carotid artery were recorded using a Sequoia ultrasound instrument (Accuson, Inc.) fitted with a 10 MHz linear probe^15,32,34 while optimizing the four boundaries defining the media-adventitia and blood-intima interfaces on the near and far walls. Later, these boundaries were traced from the digital images on a dedicated workstation to generate the mean, maximum, and minimum values of the arterial diameter, the lumen diameter, the far wall thickness, and the near wall thickness. Carotid stiffness indexes were calculated as previously described and according to accepted formulae^15,32,37,38. Overall systemic arterial stiffness was measured as pulse pressure/stroke volume.

Quality-of-life

As previously described^26,32 health-related quality-of-life was assessed with the short-form healthy survey (SF-36)³⁹ and the Minnesota Living with HF (MLHF)⁴⁰ questionnaire.

B-type natriuretic peptide-32 measurement (BNP)

As previously described, a commercially available radioimmunoassay (Phoenix Pharmaceuticals Inc; Mountain View, Calif) was used for BNP^4,26,32.

Endurance ET

The ET group exercised for one hour, three times per week for 16-weeks. Each session had three phases: 10-minute warm-up, stimulus, and 10-minute recovery. The stimulus phase consisted of walking on a track and cycle ergometry (Schwinn Airdyne). Isolated arm ergometry (operating the Airdyne with arms only) was performed ≥ 10 minutes each session in order to ensure upper extremity training. Patients exercised initially at 40-50% of heart rate reserve (HRR) for 5-10 minutes each of walking and ergometry. The intensity increased gradually until 70% HRR was maintained for at least 20 minutes each of walking and ergometry.

Attention Control

The CT group received telephone calls every 2 weeks for 16-weeks. These focused on retention, reminders and encouragement to keep study visits, and capture of medical events and did not address exercise behaviors.

Statistical analysis

The sample size was derived from a power analysis using data from a pilot study of 16 patients indicating that a sample size of 44 evaluable patients (22 per group) would provide 80% power to detect a 2% absolute change in brachial FMD, the primary outcome. The secondary outcome was carotid distensibility. VO₂peak (ml/kg/min) was the main exercise outcome.

Baseline comparisons were made using two-sample t-tests for continuous data, Fisher's exact tests for dichotomous data, or chi-square tests for categorical data. Comparisons of exercise testing, LV function, and qualify-of-life measures were made by ANCOVA, with the baseline value of the measure as the covariate. Brachial FMD and carotid distensibility were analyzed using repeated measure ANCOVA models fitted with the week 8 and 16 values as outcomes and the baseline value as the covariate. A two-sample t-test was used to compare baseline characteristics between participants who dropped out and those who completed the 16-week study. Logarithmic transformation was performed for highly skewed data. A 5% two-sided significance level was used.

RESULTS

Participants

Fortified search lists from hospital discharges, clinic visits, and echocardiogram reports were reviewed from Wake Forest Medical Center. From 827 records reviewed, 543 patients potentially met inclusion criteria; 156 responded to an invitation, passed telephone screening, and were scheduled for a screening visit. Most common reasons for exclusion were patient unwillingness or did not meet criteria.

Sixty-three HFPEF patients (age 70±7 years) were enrolled; 32 were randomized to ET and 31 to CT (Table 1). All had NYHA class II-III symptoms. There were no significant baseline intergroup differences in key characteristics except for more beta-blockers among CT patients.

Table 1.

Baseline participant characteristics

	ET(N=32)	CT (N= 31)	p-value
Female, %	72	80	0.56
Caucasian, %	66	71	0.79
African-American, %	28	23	0.77
Age, years	70 ± 7	70 ± 7	0.63
Body mass index, kg/m2	32.2 ± 6.7	32.0 ± 6.6	0.92
Body surface area, m2	1.95 ± 0.22	1.94 ±0.25	0.84
History of hypertension, %	94	84	0.25
Diabetes mellitus, %	28	19	0.56
Systolic blood pressure, mm Hg	146 ± 17	147 ± 17	0.81
Diastolic blood pressure, mm Hg	82 ± 11	83 ± 10	0.83
Left ventricular mass, g	247 ± 95	266 ± 94	0.47
Posterior wall thickness, cm	1.3 ± 0.2	1.2 ± 0.2	0.41
Left atrial diameter, cm	3.6 ± 0.7	3.4 ± 0.7	0.27
Smoking history
Never, %	41	61
Former, %	50	35
Current, %	9	3	0.26
New York Heart Association class
II, %	47	55	0.62
III, %	53	45
Diastolic Filling
Normal, %	0	0	1.0
Impaired relaxation, %	44	52	0.61
Pseudonormal, %	53	41	0.44
Restrictive, %	3	7	0.59
B-type natriuretic peptide, pg/ml	64.8 ± 74.5	72.2 ± 60.9	0.62
Medications
Diuretics, %	66	55	0.45
Angiotensin converting enzyme inhibitors, %	44	39	0.80
Beta blockers, %	9	35	0.02
Calcium channel blockers, %	44	19	0.06
Angiotensin receptor blockers, %	6	3	1.00
Estrogen (♀), %	34	36	0.92
Nitrates, %	13	10	1.00

Values are mean ± sd or %. For B-type natriutetic peptide, medians (25^th percentile, 75^th percentile) are 39.1 (18.8, 78.5) vs. 55.4 (22.6, 109.7) and the p-value shown is following logarithmic transformation of this highly skewed variable.

ET Safety, Compliance, and Adherence

One patient developed transient hypoglycemia during an exercise session; there were no other protocol-related events. Fifty-four patients (86%) completed final testing (24 ET, 30 CT). Reasons for not completing follow-up testing were: patient unwillingness (2 ET and 1 CT), elective surgery (2 ET), non-HF illness (3 ET), and HF hospitalization (1 ET). ET patients attended 88% of ET sessions. Among participants who completed follow-up testing, there were no significant intergroup differences in key baseline variables including: VO₂peak (14.2±2.2 vs.14.0±3.2 ml/kg/min, p=0.54); brachial FMD (4.0±2.2 vs. 4.7±3.5, p=0.18); and carotid distensibility (1.07±0.50 vs. 0.89±0.35, p=0.19).

Exercise testing

Following the 16-week intervention, VO₂peak, exercise time, peak power output (all p<0.0001), and VAT (p=0.01) were significantly greater in ET than CT (Table 2, Figure 1). Although peak heart rate was mildly higher in ET than CT, there was no difference in respiratory exchange ratio (RER; p=0.68), suggesting similar, exhaustive levels of effort (Table 2). Further, 43 participants (80%) achieved RER ≥1.05 at the final visit. Of 11 participants with RER <1.05, 7 (23%) were CT and 4 (17%) were ET (p=0.74), indicating similar effort. Six-minute walk distance was greater after 16-weeks in ET than CT (p=0.009).

Table 2.

Effect of endurance exercise training on exercise performance

	ET		CT		P-Value
	Baseline	Final	Baseline	Final
N	24	24	30	30
Peak Exercise (Bike)
Exercise time, min	9.5 ± 3.2	11.4 ± 3.1	9.6 ± 3.1	9.5 ± 3.3	<0.0001
Power output, watts	72 ± 31	89 ± 30	72 ± 26	67 ± 27	<0.0001
VO2, ml/kg/min	14.2 ± 2.8	15.8 ± 3.3	14.0 ± 3.2	13.8 ± 3.1	0.0001
VO2, ml/min	1260 ± 329	1388 ± 378	1233 ± 398	1217 ± 370	0.0004
Heart rate, bpm	128 ± 18	132 ± 16	131 ± 21	127 ± 17	0.01
Systolic blood pressure, mmHg	197 ± 19	199 ± 21	192 ± 27	192 ± 29	0.49
Diastolic blood pressure, mmHg	91 ± 14	91 ± 11	91 ± 12	91 ± 14	0.99
Pulse pressure, mmHg	106 ± 21	107 ± 21	102 ± 22	100 ± 27	0.47
Respiratory exchange ratio	1.13 ± 0.07	1.11 ± 0.08	1.11 ± 0.10	1.09 ± 0.08	0.68
6-minute walk distance, m	447 ± 107	486 ± 89	438 ± 79	448 ± 70	0.009
Ventilatory anaerobic threshold, ml/min	699 ± 178	796 ± 163	734 ± 189	702 ± 186	0.01
Ve/VCO2 slope	31.5 ± 4.4	32.2 ± 4.5	30.6 ± 3.6	30.2 ± 3.3	0.07

Data represent means ± SD; p-value represents comparison of least square means at final visit following adjustment for baseline values. VO₂: oxygen consumption; Ve: minute ventilation; VCO₂: carbon dioxide production.

Figure 1. Effects of endurance training on peak VO₂ (A), brachial artery flow mediated dilation (B) and carotid arterial distensibility (C) in older HFPEF patients.

Data are shown as raw mean ± SE at each time point. *=p<0.05 between groups.

Following ET, there were no intergroup differences in resting or exercise systolic, diastolic, and pulse pressure (Table 2). At baseline, 17 (27%) patients met criteria for chronotropic incompetence. Despite the ET-related improvement in peak heart rate, ET did not reduce CI significantly at follow-up (p=0.22)

Brachial FMD

Brachial FMD (the primary study outcome) was not different between groups after 8 or 16 weeks (Table 3, Figure 1). Within the ET group, FMD actually decreased slightly (4.0 ± 2.0% to 3.8 ± 3.0%). The estimated treatment effect size on absolute brachial FMD was only 0.2%; thus we can be 94% confident that the true effect is <1.0%, indicating that it is highly unlikely that the trial missed a clinically meaningful improvement in FMD.

Table 3.

Effect of endurance exercise training on brachial artery diameter in response to cuff ischemia (FMD)

	ET			CT			P-Value
	Baseline	8 Week	Final	Baseline	8 Week	Final
N	25	23	24	30	22	29
Baseline Diameter, mm	4.17 ± 0.91	4.41 ± 0.89	4.30 ± 0.83	4.13 ± 0.87	4.11 ± 1.02	4.23 ± 0.98	0.78
Maximum Diameter, mm	4.33 ± 0.89	4.56 ± 0.88	4.45 ± 0.79	4.31 ± 0.85	4.28 ± 1.01	4.40 ± 0.95	0.41
Absolute Change, mm	0.16 ± 0.07	0.16 ± 0.08	0.15 ± 0.09	0.18 ± 0.12	0.16 ± 0.09	0.16 ± 0.10	0.94*
Brachial Artery FMD, %	4.0 ± 2.0	3.7 ± 2.2	3.8 ± 3.0	4.7 ± 3.5	4.3 ± 2.8	4.3 ± 3.5	0.88*

Data represent means ± SD; p-value represents comparison of least square means of the main effect across the follow-up visits following adjustment for baseline values. Abbreviations: FMD: flow-mediated dilation (the primary study outcome).

^*p-value is shown following logarithmic transformation of this highly skewed variable.

Arterial stiffness

Carotid distensibility, the main arterial stiffness outcome, was not different after 8 or 16 weeks (Table 4). There was also no intergroup difference in any other measure of arterial stiffness (Table 4).

Table 4.

Carotid Arterial Stiffness Measurements

	ET			CT			P-Value
	Baseline	8 Week	Final	Baseline	8 Week	Final
N	24	20	23	24	21	20
Main measure
AD (×10-3 mm*mmHg-1)	1.07 ± 0.50	1.14 ± 0.70	0.97 ± 0.56	0.89 ± 0.35	0.97 ± 0.51	1.07 ± 0.34	0.65
Other measures
AC (×10-3 mm*mmHg-1)	65.1 ± 20.9	76.8 ± 64.7	58.8 ± 32.8	51.6 ± 23.4	51.3 ± 17.4	61.7 ± 21.1	0.64*
PEM (kPa)	232 ± 150	223 ± 181	224 ± 198	242 ± 109	235 ± 147	242 ± 111	0.41*
YEM (kPa)	1504 ± 1160	1302 ± 1089	1427 ± 1385	1209 ± 642	1195 ± 864	1128 ± 648	0.82*
Beta Index	16.11 ± 9.36	16.45 ± 12.46	15.77 ± 12.66	16.91 ± 7.61	16.82 ± 10.46	16.58 ± 7.16	0.50*
Pulse pressure
PP (mmHg)	70 ± 16	65 ± 12	62 ± 13	73 ± 17	67 ± 12	68 ± 12	0.24
Arterial dimensions
IMT (mm)	0.74 ± 0.08	0.75 ± 0.09	0.76 ± 0.11	0.84 ± 0.14	0.82 ± 0.10	0.84 ± 0.10	0.30
SAD (mm)	8.33 ± 1.12	8.31 ± 1.18	8.45 ± 1.20	8.26 ± 0.92	8.11 ± 0.89	8.35 ± 0.97	0.47
DAD (mm)	7.93 ± 1.15	7.90 ± 1.17	8.01 ± 1.19	7.89 ± 0.90	7.72 ± 0.89	8.00 ± 0.96	0.52
SLD (mm)	6.98 ± 1.19	7.02 ± 1.29	6.83 ± 1.02	6.63 ± 1.02	6.46 ± 0.88	6.53 ± 0.87	0.90
DLD (mm)	6.53 ± 1.22	6.57 ± 1.17	6.43 ± 1.02	6.21 ± 0.94	6.08 ± 0.91	6.09 ± 0.80	0.72
ADC (mm)	0.40 ± 0.17	0.41 ± 0.20	0.44 ± 0.24	0.38 ± 0.16	0.39 ± 0.27	0.35 ± 0.14	0.49*

Data represent means ± SD; p-value represents comparison of least square means of the main effect across the follow-up visits following adjustment for baseline values. Abbreviations: AD: arterial distensibility; AC: arterial compliance; PEM: Peterson's elastic modulus; YEM: Young's elastic modulus; kPa: kilopascals; PP: pulse pressure; IMT: mean wall intimal medial thickness; SAD: systolic arterial diameter; DAD: diastolic arterial diameter; SLD: systolic lumen diameter; DLD: diastolic lumen diameter; ADC: arterial diameter change during cardiac cycle.

^*p-value is shown following logarithmic transformation of this highly skewed variable.

LV function

There were no significant differences after 16 weeks for resting LV volumes, ejection fraction, Doppler LV filling, or pulse pressure/stroke volume ratio (Table 5).

Table 5.

Effect of endurance exercise training on resting left ventricular volumes and Doppler filling

	ET		CT		P-Value
	Baseline	Final	Baseline	Final
N	24	24	28	28
End-diastolic volume, ml	103 ± 32	101 ± 34	88 ± 26	85 ± 22	0.41
End-systolic volume, ml	43 ± 13	43 ± 16	39 ± 12	38 ± 11	0.64
Stroke volume, ml	60 ± 22	59 ± 21	50 ± 15	47 ± 12	0.29
Pulse pressure/stroke volume, mmHg/ml	1.42 ± 0.47	1.31 ± 0.51	1.52 ± 0.44	1.38 ± 0.31	0.98
Ejection fraction, %	58 ± 6	58 ± 6	56 ± 5	56 ± 5	0.58
Early filling velocity, cm/s	77 ± 21	75 ± 19	72 ± 23	68 ± 15	0.24
Atrial filling velocity, cm/s	87 ± 25	85 ± 24	78 ± 24	78 ± 21	0.95
Early/Atrial filling velocity ratio	1.03 ± 0.94	0.99 ± 0.59	1.07 ± 0.87	0.99 ± 0.64	0.81*
Early deceleration time, ms	225 ± 48	219 ± 51	216 ± 63	233 ± 57	0.24
Isovolumic relaxation time, ms	80 ± 24	80 ± 23	81 ± 22	81 ± 24	0.95

Data represent means ± SD; p-value represents comparison of least square means at final visit following adjustment for baseline values.

^*p-value is shown following logarithmic transformation of this highly skewed variable.

Quality-of-Life

After the 16-week intervention, the emotional and physical SF-36 scores were higher in ET than CT, but the MLHFQ scores were not significantly different (Table 6).

Table 6.

Effect of endurance exercise training on health-related quality of life

	ET		CT		P-Value
	Baseline	Final	Baseline	Final
N	20	20	27	27
SF-36
Physical	48 ± 18	63 ± 20	50 ± 25	53 ± 27	0.03
Emotional	63 ± 30	83 ± 31	66 ± 37	62 ± 35	0.04
MLHFQ
Physical	17 ± 8	12 ± 8	13 ± 11	11 ± 10	0.50
Emotional	6 ± 5	5 ± 4	4 ± 5	3 ± 4	0.19*
Total	36 ± 19	26 ± 19	28 ± 23	25 ± 22	0.50

Data represent means ± SD; p-value represents comparison of least square means at final visit following adjustment for baseline values. Abbreviations: MLHFQ: Minnesota living with heart failure questionnaire

^*p-value is shown following logarithmic transformation of this highly skewed variable

Analysis based on medications and dropouts

Although beta-blockers were somewhat more prevalent in CT, the overall results were unchanged after adjusting for this. Baseline brachial FMD was not different between those who dropped out (4.4±2.9%, N=9) and those who completed the study (3.2±2.0, N=54, p=0.3). Baseline carotid distensibility was not different between those who dropped out (0.67±0.59×10^-3mm*mmHg^-1, N=5) compared to those who completed (0.97±0.44×10^-3mm*mmHg^-1, N=48, p=0.29). Baseline VO₂peak was lower in participants who dropped versus completed (11.9±2.8, N=9, and 14.1±3.2, N=54, respectively, p=0.03).

Responders analyses

A responder was defined as having an absolute increase of 1.0% for brachial FMD and a 10% relative increase in carotid distensiblity³⁶ at follow-up. There were 5 (21%) FMD responders in ET and 6 (21%) in CT (p=1.0). There were 4 (17%) carotid distensibility responders in ET and 7 (35%) in CT (p=0.29). Within ET, an additional analysis was performed to determine whether any key baseline variable differed between responders and non-responders. For FMD, only E deceleration time was significantly different (174±28 vs. 239±43 ms, respectively, p=0.004). For carotid distensibility, no variable was significantly different.

DISCUSSION

To our knowledge, this is the first report on the effect of ET on endothelial function and arterial stiffness in HFPEF. The effect size of ET on peak VO₂ was 1.8 ml/kg/min (13%, 15.7±0.3 vs. 13.9±0.3 ml/kg/min; least square means from ANCOVA). This supports findings from three previous trials,^26,41,42 increasing the number of HFPEF patients reported in randomized ET trials nearly 50%. Despite the significant improvement in VO₂peak, and in contrast to our hypothesis, ET did not improve brachial FMD (the primary outcome) or arterial stiffness. This indicates that improvements in large arterial function were not responsible for the ET-related improvement in VO₂peak. There were also no changes in resting LV function. Combined with our previously reported trial and that recently reported by Fujimoto et al^10,43, this suggests that the ET-related improvement in VO₂peak in HFPEF may be related to microvascular and/or skeletal muscle adaptations that increase diffusive oxygen transport and/or utilization by the active muscles^19,44-47.

This study was driven by evidence suggesting that impaired arterial function (vasodilation and/or arterial stiffness) contributes to exercise intolerance in HFPEF. Borlaug et al showed that in HFPEF vasodilator reserve was reduced and correlated with reduced exercise capacity.⁶ Puntawangkoon et al⁴⁸ found that submaximal exercise leg blood flow was reduced. Borlaug et al also reported reduced digital arterial function in HFPEF⁷. However, the finding of abnormal vasodilation in HFPEF has not been universal, since Haykowsky et al, using high resolution brachial ultrasound, and Hundley et al, using phase contrast femoral MRI, found FMD was not reduced in older HFPEF patients screened for atherosclerosis^13,24.

Arterial stiffness has been uniformly reported to be increased in HFPEF. We and others have shown that aortic and carotid distensibility are significantly reduced in elderly HFPEF and are independent predictors of reduced VO₂peak ^13,15,49,50. Borlaug et al reported that arterial stiffness increases during exercise and is inversely related to exercise performance^7,14.

This study was also driven by several reports that ET in HFREF patients enhances arterial function^{16,17,19-21,23,51}. Hornig et al¹⁶ and others^17,21 showed that ET-mediated improvements in endothelial arterial function may be region-specific (to the locally active muscles), whereas Hambrecht et al found that effects of ET extended beyond regional effects^21,23 since lower-limb ET increased upper limb FMD.^21,23 Moreover, Parnell et al demonstrated that ET improves arterial stiffness in HFREF²².

In our study, 16-weeks of ET did not change endothelial function or arterial stiffness. This suggests that the vascular adaptations to ET may differ between HFREF and HFPEF patients⁵². If so, that may be partly explained by the fact that the effects of ET on vascular function appear to be sex and age-dependent. This is relevant since the majority of the HFPEF patients in our study were older women, as in the general population¹, whereas most HFREF patients are men. Specifically, Pierce et al recently found that brachial FMD was not different between older women undergoing high-frequency, long-term ET (≥5 years) versus age-matched healthy sedentary controls, whereas it was 49% greater in older men with long-term ET compared to controls⁵³. Also, two-months of ET improved brachial FMD in previously sedentary healthy middle-aged men but not in older post-menopausal women⁵³. Of note, the HFREF patients in prior studies of the effect of ET on endothelial function were predominantly men¹. Furthermore among master ET athletes, carotid artery compliance is lower in older women compared to men⁵⁴. Thus, sex and age-dependent effects may explain, at least partly, differences in the effect of ET on arterial function in HFPEF patients in the present study compared to previous studies of HFREF.

Another potential explanation for differing results of ET on arterial function in HFPEF compared with HFREF is the profound effect of atherosclerosis on vascular function²⁸. A strength of the present study is that patients were screened to exclude those with coronary, cerebrovascular, and peripheral vascular disease to avoid the strong, confounding influence of atherosclerosis^28,55. However, this has not been accounted for in most prior studies of FMD and arterial stiffness, and atherosclerosis is more common in HFREF than HFPEF¹. However, studies using animal models of HFREF and carefully selected HFREF patient groups indicate that abnormal endothelial function is present in HFREF even in the absence of atherosclerosis^56,57. Our finding that ET did not improve brachial FMD and that FMD did not contribute to the ET-related improvement in VO₂peak, taken together with our two prior reports that found no significant difference in brachial or femoral FMD at baseline in older patients with HFPEF versus controls^13,24, suggests that in contrast to HFREF, abnormal FMD may not be a fundamental component of the pathophysiology of exercise intolerance in older HFPEF patients.

If neither increased exercise cardiac output¹⁰ or improved large arterial function primarily account for the improved VO₂peak following ET in older HFPEF, this suggests a potential role for skeletal muscle adaptations, such as either increased diffusive oxygen transport of O₂ from red blood cell to muscle mitochondria and / or improved O₂ utilization by the active muscles, primarily via mitochondrial function. Some independent data support this possibility. Bhella et al recently showed that skeletal muscle oxidative metabolism was abnormal in a small number of older patients with HFPEF and appeared related to their severely reduced VO₂peak⁸. Furthermore, several studies have shown that abnormal skeletal muscle perfusion and metabolism contribute significantly to exercise intolerance in HFREF and to improvements in VO₂peak following ET^44,58-61.

Our study also confirmed prior reports that ET improves peak heart rate response, quality-of-life, and 6-minute walk distance in HFPEF^10,26,41.

Limitations

Although our study duration was long enough to produce a significant improvement in VO₂peak and included upper as well as lower extremity training, we cannot exclude that training that was longer and more intense could produce detectable improvements in vascular or ventricular function. Molmen-Hansen et al compared high-intensity aerobic interval exercise with moderate intensity continuous ET and usual care in patients with essential hypertension⁶². They reported that change in VO₂peak after high-intensity interval training was 3 and 5-fold greater than moderate continuous training or usual care groups, respectively. LV systolic and diastolic function and brachial FMD improved significantly after high-intensity training but were unchanged with moderate continuous training or usual care⁶². Wisloff et al showed that high-intensity aerobic interval training was superior to continuous moderate intensity ET for improving VO₂peak, resting ejection fraction, and brachial FMD in older (75 years) men with HFREF²⁰.

We did not measure FMD or arterial stiffness (other than pulse pressure which was unchanged, Table 2) during peak exercise, which would provide the most definitive evidence of relation to changes in VO₂peak, but would be challenging. However, in studies of ET in HFPEF patients where hemodynamics were assessed at peak exercise, neither 16-weeks or 1 year of ET produced a change in peak exercise or reserve pulse pressure/stroke volume ratio, a measure of arterial stiffness^10,43.

There were more dropouts in the ET than the CT group, however, the number of evaluable patients at the end of the study was within our sample size estimates. Furthermore, among those who completed the study, key baseline variables were similar in ET and CT participants. Also, brachial FMD and carotid distensibility were not different at baseline between those who completed compared to drop-outs.

Tissue Doppler was not measured. Thus, there may have been improvements in resting LV diastolic function following ET that were not detected.

By design, participants were stable, well compensated outpatients who were able to participate in exhaustive exercise testing and ET. As a result, the mean BNP level was less than in patients with acute, decompensated HF. BNP levels are also known to be lower in HFPEF than in HFREF, likely due to smaller LV cavity size⁶³. The BNP levels are similar to other studies of stable HFPEF patients able to undergo maximal exercise testing^7,12,32,64 and are several-fold increased compared to healthy, age-matched, normal subjects.⁴ However, our results may not apply to patients who are sicker, poorly compensated, or less clinically stable. Further, we enrolled HFPEF patients without known or suspected coronary, cerebrovascular, and peripheral arterial disease. Thus, we cannot exclude the possibility that ET improves arterial stiffness or endothelial function in a broader population of HFPEF patients with these comorbidities.

Conclusion

ET significantly improves VO₂peak without altering brachial FMD or carotid arterial stiffness in older HFPEF patients. Combined with our previous finding that improved peripheral function (A-VO₂ Diff) is an important contributor to the ET-related increase in VO₂peak ¹⁰, this suggests that microvascular and/or skeletal muscle adaptations may contribute to the ET-related increase in VO₂peak in older HFPEF patients.