Background: flow-mediated arterial dilation (FMAD), an indicator of endothelial function, is reduced in patients with heart failure and reduced left ventricular ejection fraction (HFREF). Many elderly patients with heart failure exhibit a normal left ventricular ejection fraction (HFNEF). It is unknown whether FMAD is severely reduced in the elderly with HFNEF. Methods and Results: 30 participants >60 yr of age, 11 healthy, 9 with HFNEF, and 10 with HFREF, underwent a cardiovascular magnetic resonance (CMR) assessment of FMAD in the superficial femoral artery followed within 48 h by symptom-limited exercise with expired gas analysis. Elderly patients with HFREF and HFNEF had severely reduced peak oxygen consumption (V̇o2 peak; 12 ± 2 and 13 ± 1 ml·kg−1·min−1, respectively) vs. their healthy age-matched contemporaries (20 ± 3 ml·kg−1·min−1). FMAD was 3.8 ± 1.3% (0.85 ± 0.22 mm2) in patients with HFREF; it was 12.1 ± 3.6% (3.1 ± 1.2 mm2) and 13.7 ± 5.9% (3.9 ± 1.7 mm2), respectively, in patients with HFNEF and age-matched healthy older individuals. After adjustment for age and gender, the association of FMAD with V̇o2 was high in healthy and HFREF subjects (P = 0.05 and 0.02, respectively) but less so in HFNEF participants (P = 0.58). Conclusions: elderly patients with HFNEF do not exhibit marked reduction in leg FMAD. These data suggest that mechanisms other than impaired femoral arterial endothelial function contribute to the severe exercise intolerance experienced by these individuals.
- endothelial function
- magnetic resonance imaging
impaired peripheral arterial endothelial function and reduced lower extremity blood flow during exercise contribute to exercise intolerance in middle-aged patients with heart failure associated with reduced left ventricular (LV) ejection fraction (HFREF) (10, 18, 35). Previously, we demonstrated that leg flow-mediated arterial dilation (FMAD), a noninvasive measure of peripheral arterial endothelial function, is reduced in heart failure patients compared with normal subjects (1). Importantly, however, it is unknown how the aging process influenced our results, or whether our findings would be present in those with heart failure and normal LV ejection fraction (HFNEF). This question is particularly relevant in older persons, who comprise the majority of persons with heart failure, have particularly severe exercise intolerance (21), and, unlike their middle-aged counterparts, more commonly exhibit a normal ejection fraction (14, 37).
Accordingly, we hypothesized that, similar to middle-aged individuals, FMAD would be reduced in older patients with HFNEF or HFREF. To test our hypothesis, we measured in older healthy patients as well as those with heart failure 1) FMAD in the lower extremity by cardiovascular magnetic resonance (CMR) (1, 33, 34); 2) stimuli for FMAD, such as longitudinal vessel wall sheer stress; and 3) exercise performance by expired gas analysis.
MATERIALS AND METHODS
The Institutional Review Board of the Wake Forest University School of Medicine approved the study protocol, and all participants gave written informed consent. The study population consisted of 11 healthy community-dwelling volunteers aged >60 yr who took no medication, had no illness, had a normal physical examination, and had a systolic and diastolic blood pressure below 140 and 90 mmHg, respectively, and a normal exercise echocardiogram (NO); 9 individuals aged >60 yr with established New York Heart Association Class II or III heart failure and a LVEF ≥55% without evidence of ischemic or valvular heart disease, or chronic pulmonary disease, and a mitral inflow E/A ratio of 0.83 ± 0.28 (HFNEF); and 10 older individuals aged >60 yr with New York Heart Association Class II or III heart failure associated with reduced LVEF (HFREF). As previously described (21), the diagnosis of heart failure was based on clinical criteria that included a heart failure clinical score from the National Health and Nutrition Examination Survey-I of '4 (32) and those utilized by Rich et al. (31) that included a history of acute pulmonary edema or the occurrence of at least two of the following that improved with diuretic therapy without another identifiable cause: dyspnea on exertion, paroxysmal nocturnal dyspnea, orthopnea, bilateral lower extremity edema, or exertional fatigue.
Each subject underwent a CMR study of the superficial femoral artery (SFA) followed within 48 h by maximal cycle ergometry with expired gas analysis (14, 21). All tests were performed between 0800 and 1100 (24-h clock) with the patient in a postabsorptive baseline compensated state in which medications were withheld at least 12 h before testing. Participants ineligible for study included those with a contraindication to CMR (implantable pacemakers or defibrillator, intracranial metal, or claustrophobia), chronic renal insufficiency (creatinine ≥2.5 mg/dl), anemia (hemoglobin ≤11 mg/dl), or symptoms or signs of peripheral arterial disease including claudication, arterial bruits, or a luminal narrowing of >20% in the iliac or SFA segments as determined by our gradient-echo magnetic resonance angiography techniques. Strenuous physical activity was not performed 24 h before testing, and substances such as theophylline, caffeine, and nicotine that could affect vascular reactivity were withheld 24 h before CMR and exercise testing.
The CMR procedure was performed using previously described validated techniques that have the advantage compared with ultrasound of appreciating very small changes in arterial area and blood flow velocity in virtually any vascular territory with high precision and low variance (1, 33, 34). In addition, unlike intravascular infusions of acetylcholine and guide wire or angiographic determinations of vessel velocity and area, they do not require an interventional procedure that is associated with a risk of bleeding or infection.
On arrival for study, participants rested in a dimly lit room for 30–45 min and then were placed supine, feet first, with electrocardiographic monitoring leads, a brachial blood pressure cuff, and a large thigh cuff and a phased-array surface coil placed around the left thigh. Before and throughout the course of testing, heart rate and systemic pressure were monitored. Axial, coronal, and oblique sagittal gradient-echo images of the left iliac and SFA were obtained, and a position of roughly 20 cm distal to the femoral head along the SFA was isolated (Fig. 1). The purpose of these acquisitions was to obtain a cross-sectional view of the area within the vessel lumen of the artery along a straight vessel segment perpendicular to its course that would minimize partial volume effects during scanning. After scanning of the cross-sectional area of the lumen of the artery three times, the thigh cuff was inflated to 50 mmHg above the participant's systolic blood pressure measured in the forearm. A CMR acquisition was obtained to confirm absence of flow in the artery; if flow was present, the cuff was inflated an additional 20 mmHg, and repeat images were acquired to confirm vessel occlusion. After occluding the artery for 5 min, pressure within the thigh cuff was released, and CMR scans were repeated every 30 s for 3 min.
CMR was performed using a 1.5-Tesla Horizon whole body imaging system (GE Medical Systems, Waukesha, WI). Scan parameters included a 7-mm-thick slice, a 256 × 256 matrix, a 12- to 15-cm field of view, a through-plane velocity encoding of 200 cm/s, a 40° flip angle, an 18-ms repetition time, and a 6.7-ms echo time. Segmented k-space (views per segment) and view sharing were used to obtain four actual and three view-shared frames per cardiac cycle (temporal resolution = 72 ms with view sharing).
Exercise testing protocol.
Upright stationary bicycle exercise began at an initial workload of 12.5 W, was advanced to 25 W 2 min into exercise, and was then increased by 25-W increments every 3 min thereafter to the point of exhaustion (14, 21). Continuous expired gas analysis was performed with a Medical Graphics CardiO2 system (Medigraphics, Minneapolis, MN) that was calibrated before each study with gases of known volume and concentration, as previously described (14, 21). Data were collected during the fifth minute of the rest stage and during the last minute of each stage of exercise. Peak oxygen consumption (V̇o2 peak) was considered to be the highest oxygen consumption achieved during exercise.
All data were analyzed in a blinded fashion by individuals who did not perform the scanning procedure and had no knowledge of patient subgroup or other clinical information. Vessel area was determined by tracing a region of interest that delineated the lumen of the artery on the magnitude image of each frame. The image was magnified with pixel replication by a factor of eight, and then the lumen boundary was defined as the point at which the intensity of the blood dropped to 50% of the peak intensity within the lumen as judged visually (36). The lumen area was defined as the flow lumen during systole (the average lumen area of the third and fourth frames of each scan sequence). The baseline area was defined as the average area from the three baseline acquisitions measured before cuff inflation. The vasodilatory response was expressed as an absolute and percent change as relative to the baseline area of the blood vessel.
With the use of phase contrast image data, blood flow velocity was measured at rest and during peak blood flow after cuff release. This information was used to assess longitudinal sheer stress (a stimulus for FMAD) in the SFA. To calculate longitudinal shear stress, we utilized a technique similar to the 3-dimensional (3-D) parabolic fitting method described by Oshinski et al. (26) and Oyre et al (27), based on two assumptions: the blood velocity at the vessel wall is zero, and the velocity distribution within the lumen is parabolic except for the boundary and center pixels. Parabolic fitting was performed on an annulus, and for each parabola being fit, a sector was defined whose arc extended two pixels to either side of the parabola's lower endpoint. Pixels inside this sector were included in a linearly weighted fashion to improve the fit. Shear stress was then calculated as the product of the velocity gradient at the wall and the blood viscosity as determined from the patient's serum hemoglobin level measured within 6 mo of the performance of the CMR exam. With the use of this technique, assumptions were not needed regarding the shape of the vessel. We perceived this feature to be important because of the potential for distortion of the vessel geometry after cuff inflation.
Statistical comparisons of measurements of vessel area and percent change, shear stress, exercise testing, and demographics were evaluated using one-way three-group analysis of variance. If the global test of group differences was significant at the 5% level, then pairwise group comparisons were performed at the 5% two-sided level of significance. In addition, comparisons were performed using analysis of co-variance to adjust for baseline characteristics in the study population, with all data expressed as means ± SD. The trial was designed (based on a power analysis using previous data and conservative estimates of the variances for FMAD) with 10 subjects per group, to have at least 80% power to detect a 1-mm2 absolute (3% difference) change in area between groups.
CMR examinations and exercise tests were well tolerated in all individuals. Image data from one NO (due to an image artifact) and two individuals with HFREF (due to evidence of a luminal narrowing >30%) were excluded from further analysis. The remaining 27 subjects formed the study population, and their detailed demographic data are displayed in Table 1. Two subjects were African American, and the remainders were Caucasian. Demographics and the incidence of hypertension, smoking, diabetes, and medication use were similar to those reported previously in population-based studies of elderly subjects with and without heart failure (8, 21, 37). Subjects with heart failure displayed a higher incidence of hypertension and use of angiotensin-converting enzyme inhibitors compared with their NO counterparts.
As shown in Table 2, patients with HFNEF and HFREF had severe reductions in V̇o2 peak and exercise time compared with their NO controls. Compared with NO controls, the absolute changes in FMAD compared with baseline were substantially reduced in patients with HFREF but not in patients with HFNEF (Fig. 2 and Table 3). FMAD was substantially reduced in HFREF compared with HFNEF patients by 2.24 mm2 (P < 0.001). These intergroup differences in FMAD persisted and were strengthened in analyses performed using the percent change in vessel area (Table 3), which, relative to absolute area, is independent of differences in gender and body size. Importantly, after adjustment for age and gender differences in our groups, FMAD was associated with V̇o2 peak in NO (r = 0.71, P = 0.05) and HFREF (r = 0.89; P = 0.02) subjects but not those with HFNEF (r = 0.26, P = 0.58). Absolute change in FMAD was not associated with participants' systolic (P = 0.56) or diastolic (P = 0.73) blood pressure. LV wall thickness was 9.5 ± 0.7, 11.3 ± 3.5, and 8.6 ± 0.5 mm in NO, HFNEF, and HFREF, respectively. The correlation between percent FMAD and wall thickness (r = −0.06) was not significant (P = 0.83).
To determine whether the differences in FMAD between patient groups were independent of intergroup differences in patient characteristics previously associated with FMAD, group differences adjusted for these variables (except for LVEF, which was part of the group definition) were derived (Table 4). All the factors were nonsignificant predictors of FMAD, with all P values >0.25. As shown in Table 4, after multivariate analysis with all of these characteristics entered in the model, the difference in FMAD of 2.91 mm2 between HFNEF and HFREF remained significant (P < 0.05), and the absolute change in area between NO and HFNEF (3.57 and 3.59 mm2, respectively, for NO and HFNEF) became more similar.
To assess for the potential influence of intergroup differences in circulating neurohormones and markers of inflammation, we performed additional analyses after adjusting FMAD for these serum measures listed in Table 1. Although renin levels were elevated in participants with HFNEF compared with other participants, renin, aldosterone, norepinephrine, and endothelin were not predictors of FMAD (P values ranging from 0.63 to 0.92). Angiotensin II (P < 0.08) and TNF-α (P < 0.14) trended toward predicting FMAD, and, after adjustment for both of them, FMAD was 12.3, 12.5, and 4.7% in our NO, HFNEF, and HFREF groups, respectively. The difference in FMAD between HFNEF and HFREF groups remained significant (P < 0.009).
An example of resting and post-cuff release laminar shear stress from a patient in each group is shown in Fig. 3. As shown in Table 3, after cuff release, laminar shear stress tended to amplify to a greater degree in NO and HFNEF patients than it did in patients with HFREF. In addition, the resting and post-cuff release velocity measures paralleled the changes seen in laminar sheer stress and FMAD 60–90 s after cuff release.
Arterial blood flow to the exercising leg is an important determinant of exercise capacity, and studies performed using invasive techniques indicate that reduced leg arterial blood flow during exercise contributes to the severe exercise intolerance found in patients with HFREF (1, 39). FMAD correlates reasonably well with exercise-induced medium-sized arterial blood flow (13), and, to date, several studies have suggested that there is a relationship between the abnormal brachial artery FMAD found in HFREF patients and their reduced exercise capacity (7, 12, 22).
Moreover, previous studies have documented that patients with HFNEF exhibit a reduction in V̇o2 that is similar in magnitude to patients with HFREF (21). Several factors may be responsible for this reduction of V̇o2 in patients with HFNEF, including heightened proximal aortic stiffness, diminished lower extremity skeletal muscle mass (14, 17), or an elevation of circulating neurohormones (19). Because prior studies have shown an association between reduced FMAD and exercise capacity in patients with HFREF, we sought to assess whether FMAD in medium-sized arteries was abnormal in patients with HFNEF. We felt this question important to address, given the preponderance of elderly patients with HFNEF and the desire to identify abnormalities suitable for therapeutic intervention.
Our results demonstrate that, in contrast to the severe reduction of FMAD seen in HFREF patients and contrary to our original hypotheses, leg FMAD in HFNEF patients was relatively preserved and similar to that observed in elderly healthy control participants. This occurred even though both heart failure groups had similarly severe reductions in exercise V̇o2. Perhaps other variables, including proximal aortic stiffness and muscle tone, may be more responsible for the decrease in V̇o2 associated with this disorder. Ours is a clinically relevant finding because exercise intolerance is the primary symptom of patients with HFNEF. Importantly, exercise capacity and the demographic characteristics of the elderly subjects in the present study (Tables 1 and 2) were similar to those reported previously (8, 21, 37), and patients with HFREF had significantly reduced leg FMAD compared with healthy age-matched controls (Table 3), congruent with observations reported by others (7, 12, 13, 22).
To understand why FMAD was preserved in HFNEF relative to HFREF participants, we performed analysis of co-variance, adjusting for a wide range of variables known to influence peripheral arterial endothelial function, including patient demographics and co-morbidities (such as tobacco use, diabetes, or hypercholesterolemia) (15, 34), medication use (25), circulating neurohormones and cytokines (11, 23), and change in longitudinal shear stress (16, 30), one of the most potent stimulators of endothelial cell nitric oxide release, which in turn relaxes vascular smooth muscle. After adjustment for demographics and co-morbidities, the difference in FMAD between HFNEF and HFREF groups widened and remained significant (P < 0.05). In addition, FMAD in HFNEF subjects became even more similar to that in NO control subjects (Table 4). Although angiotensin-converting enzyme inhibitors (25) were by definition used more frequently in both types of heart failure subjects compared with NO control subjects, there was no difference in use of these vasoactive medications between both groups of heart failure participants. Additionally, after adjustment for chronic medication use in our sample, the difference in FMAD between our heart failure groups remained significant (P < 0.01). Thus it seems unlikely that the use of vasoactive medications is responsible for the relatively preserved FMAD seen in HFNEF patients.
Likewise, the intergroup differences in FMAD persisted, and were strengthened, after assessment for the potential effect of the circulating neurohormones and inflammatory modulators. While circulating endothelin and TNF-α levels were not significantly different among our patient groups, renin levels were higher in HFREF vs. HFNEF patients. Modulation of FMAD via the renin-angiotensin-aldosterone axis may be achieved via direct effects of angiotensin II through type 1 angiotensin receptor (AT-1)-mediated generation of superoxide anions from smooth muscle and endothelial cell membrane-bound reduced nicotinamide adenine dinucleotide-dependent oxidase (9, 29). Because circulating levels of angiotensin II were not significantly different in our patient groups, the relevance of isolated elevation of renin is uncertain. None of our participants used the aldosterone-blocking agents spironolactone and eplerenone. It is possible that the “aldosterone escape” may have contributed to the severely reduced FMAD seen in HFREF via the adverse effects of increased circulating aldosterone on arterial function (2).
As shown in Table 3, we were able to utilize the velocity information derived from the phase contrast component of the study to examine the relationship among velocity, longitudinal sheer stress, and FMAD. Resting longitudinal sheer stress and blood flow velocity were similar in the patient groups, but after cuff release, there was a significant increase in blood flow velocity in NO and HFNEF subjects vs. those with HFREF (Fig. 3). It is plausible that our observed changes in FMAD in NO and HFNEF subjects were associated with augmentation of blood flow velocity and heightening of the longitudinal sheer stimulus for nitric oxide release after cuff release. Failure to augment blood flow velocity and longitudinal sheer stress in HFREF subjects could be related to their reduced LVEF (mean of 26% vs. >50% in HFREF and HFNEF subjects, respectively), to endothelium-dependent or -independent impairment of the microcirculatory and arteriolar reservoir distal to the medium-sized arteries, or to other mechanisms. Consistent with previous reports (20, 35), resting cardiac output (Table 1) was not reduced in either heart failure groups compared with normal control subjects, implying that forward flow was similar among the patient groups. Further resolution of this mechanistic question requires additional study.
Direct quantifications of endothelium-dependent (FMAD) and -independent (nitroglycerin infusion) measures of medium-sized vessel function have been accomplished with high-resolution CMR (38). When endothelium-dependent mechanisms are impaired, other investigators have assessed endothelium-independent mechanisms to determine whether failure to vasodilate is due to an inability to respond (endothelial-independent mechanism) or an inability to produce nitric oxide (endothelium-dependent mechanism) (6, 18). While both endothelium-independent and -dependent mechanisms are responsible for increased medium-sized arterial blood flow after stress, our research shows that FMAD (an endothelium-dependent assessment) is not impaired in patients with HFNEF. In this study, we did not assess endothelium-independent mechanisms of arterial dilation. For those elderly subjects with HFREF and abnormal FMAD, further studies are required to assess the relative importance of endothelium-independent mechanisms on arterial dilation (38).
Consistent with population-based reports, nearly all of our HFNEF participants had a history of systemic hypertension or were treated with anti-hypertensive medications (8, 21, 37). The reported effect of hypertension on FMAD has been variable. While some studies of peripheral arterial endothelial function in hypertensive individuals have identified abnormalities of endothelial function in medium-sized arteries (24, 28), other studies in patients with isolated hypertension have not (5). Often it is difficult to exclude co-morbidities that could impair FMAD in participants with hypertension. Importantly, our results indicate that superficial femoral arterial endothelial function in HFNEF patients with a preponderance of hypertension but without overt arteriosclerosis is far less reduced than in elderly subjects with HFREF. Also, FMAD in our HFNEF participants is nearly the same as that observed in healthy aged individuals.
We capitalized on the heightened accuracy and low variability of established validated CMR techniques for the assessment of FMAD (1, 33, 34). This allowed us to appreciate changes in our patient population using relatively small sample sizes. Our study with 27 subjects exhibited estimates of variances that were smaller than those assumed in the study design. On the basis of the results of this study, a future study of similar size would have at least 80% power to detect the above-stated differences. Also, we studied FMAD in the leg because leg arterial function is known to be a major contributor to exercise performance in healthy young subjects (35, 39). Interestingly, the changes in vessel areas after cuff release that we found in the superficial femoral artery of our NO subjects correspond to the 5–6% changes in brachial artery diameters identified previously in older healthy individuals (4).
We recognize the following limitations to our study. First, several factors may have contributed to exercise intolerance in our HFREF patients, and we cannot determine the exact cause of reduced FMAD in our HFREF participants. For example, with the variability found in our measures of TNF-α, 103 patients would be needed in each group to detect a 25% difference with 80% power between HFNEF and HFREF groups. Importantly, while our FMAD results in HFREF patients are similar to those reported previously (7, 12, 13, 22), it was not a goal of the current study to define the etiology of exercise intolerance or reduced FMAD in these subjects. Second, our HFNEF participants exhibited few conditions (i.e., advanced arteriosclerosis or peripheral vascular disease) that would influence FMAD (3). Our HFNEF participants were highly representative of those found in the community (8, 21, 37). Peripheral arterial endothelial function may be reduced in HFNEF patients in proportion to the preponderance of other clinical conditions known to influence FMAD. Third, the majority of our participants were Caucasian, and therefore we are uncertain how our results apply to patients of other ethnicities. Finally, the 95% confidence intervals of the significant difference of 2.24 mm2 (P = 0.001) in FMAD between our heart failure groups ranged from 1.31 to 3.17 mm2. Additional post hoc power analyses indicated that, if we performed a second study with four times the number of subjects, this confidence interval would be reduced by approximately one-half, but it would not materially affect the overall findings of the study.
In conclusion, in subjects with heart failure and normal LVEF in the absence of advanced arteriosclerosis, diabetes, or hypercholesterolemia, superficial femoral artery flow-mediated dilation is preserved relative to elderly patients with heart failure and reduced LVEF. Future investigations of other components of the cardiovascular and musculoskeletal systems are warranted to determine the pathophysiological mechanisms of exercise intolerance in patients with heart failure and normal LVEF.
Research was supported in part by National Institutes of Health (NIH) General Clinical Research Center Grants, including the Clinical Associate Physician Award (MO1-RR-07122), North Carolina Baptist Hospital Technology Development Fund (B-03-97/98R), NIH Grant RO1-AG18915, and The Claude D. Pepper Older Americans Independence Center of Wake Forest University NIH Grant P60-AG-10484.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society