Author + information
- Received July 17, 2006
- Revision received October 20, 2006
- Accepted October 27, 2006
- Published online March 6, 2007.
- Harry A. Silber, MD, PhD⁎,⁎ (, )
- Joao A.C. Lima, MD, FACC⁎,†,
- David A. Bluemke, MD, PhD†,
- Brad C. Astor, PhD‡,
- Sandeep N. Gupta, PhD§,1,
- Thomas K. Foo, PhD§,1 and
- Pamela Ouyang, MD, FACC⁎
- ↵⁎Reprint requests and correspondence:
Dr. Harry A. Silber, Cardiology A-1 East, Johns Hopkins Bayview Medical Center, 4940 Eastern Avenue, Baltimore, Maryland 21224.
Objectives Our goal was to investigate whether the association between established cardiovascular risk factors and arterial reactivity differs between the lower and upper extremities.
Background Resistance artery reactivity in the arm is associated with cardiovascular risk factors, coronary disease, and events. However, the relationship of lower versus upper extremity vasoreactivity to increasing cardiovascular risk factors has not been determined.
Methods We studied 82 subjects in 3 groups: 33 young healthy (YH) (21 to 41 years), 30 older healthy (OH) (>50 years), and 19 older type 2 diabetic subjects (OD). We directly measured systolic shear rate, flow, and radius in brachial and femoral arteries at rest and during post-occlusion hyperemia using magnetic resonance imaging.
Results Brachial and femoral systolic shear rate, flow, and radius were similar among the groups at rest. Brachial hyperemic shear rate and hyperemic flow normalized as a function of baseline radius were not statistically different when YH were compared with OH and OH with OD. In contrast, femoral hyperemic shear rate and hyperemic flow normalized to baseline radius were lower in OH than YH (680 ± 236 s−1 vs. 843 ± 157 s−1, p = 0.001, and 0.84 ± 0.25 mm1.27/s vs. 1.01 ± 0.16 mm1.27/s, p = 0.001) and lower in OD than OH (549 ± 183 s−1, p = 0.02, and 0.74 ± 0.19 mm1.27/s, p = 0.046).
Conclusions Persons with increasing cardiovascular risk factor burden had progressively reduced arterial reactivity in lower but not upper extremities. This may help to explain why atherosclerosis usually develops more severely in legs than in arms, and suggests that legs may be more sensitive than arms for assessing early global atherosclerotic risk.
Improved methods are needed for predicting atherosclerosis development and cardiac events. Peripheral arterial reactivity noninvasively tests for nitric oxide-dependent endothelial dysfunction, a key early event in atherosclerosis development (1). Although the upper extremity is conveniently available for testing arterial reactivity, there is evidence to suggest that the lower extremity may be a more sensitive site for assessing global cardiovascular risk. Atherosclerosis usually involves the lower extremities more severely than the upper extremities (2), and endothelial function is more impaired in the lower than upper extremity in persons with coronary artery disease (3). This suggests that endothelial function may be impaired earlier or more severely in lower extremities than upper extremities in persons with cardiovascular risk factors but without clinical coronary disease.
Certain tests of arterial reactivity are only partially dependent on nitric oxide, but this does not limit their ability to identify groups at risk. The stimulus for flow-mediated dilation (FMD) is post-occlusion hyperemic shear stress, which is hyperemic shear rate times viscosity (4–7). Hyperemic shear rate and flow are partially dependent on endothelial nitric oxide release from resistance arteries (8–10). However, the hyperemic response is also dependent on other molecular mechanisms, including prostaglandins and adenosine (11). Although hyperemic flow (HypQ) is only partially regulated by endothelially released nitric oxide, reduced hyperemia is associated with cardiac risk factors, coronary artery disease, and cardiac events (8,12–17). This may be partly because other properties that are associated with cardiovascular risk factors, including increased central arterial stiffness, increased local arterial stiffness, and increased sympathetic nervous system activity, are also associated with reduced hyperemic shear or flow (18–20). Indeed, a large study showed that the association between impaired FMD and cardiovascular risk factors was attributable to a reduced hyperemic shear stimulus for FMD (13).
Relatively greater reductions in lower compared with upper extremity HypQ have been demonstrated in subjects with hypercholesterolemia and in subjects with peripheral arterial disease (21,22). However, regional differences in hyperemic shear rate have not been studied in persons with risk factors.
We previously reported a method of directly measuring hyperemic systolic shear rate and hyperemic systolic flow in the brachial and femoral arteries from velocity-encoded images obtained using magnetic resonance imaging (MRI) (23). The purpose of this study was to determine if there are differences in lower versus upper extremity hyperemic shear rate or flow responses with increasing cardiovascular risk factors.
Eighty-two subjects from 3 groups were studied: 1) 33 healthy young adults, ages 21 to 41 years (16 men, 17 women), with no cardiovascular risk factors of hypertension, diabetes, hyperlipidemia, smoking, obesity, or cardiac disease in a first-degree relative; 2) 30 older healthy subjects, ages 50 to 74 years (14 men, 16 women), with no other cardiovascular risk factors; and 3) 19 older subjects with type 2 diabetes, ages 51 to 69 years (14 men, 5 women). Subjects had no acute illness or documented symptoms of intermittent claudication, and were not on nitrate medicine. The study protocol was approved by the Institutional Review Board at the Johns Hopkins School of Medicine. All subjects gave written informed consent.
Subjects abstained from eating or drinking except water for at least 6 h before the study. All scans were performed in the morning. Baseline blood pressure was recorded in the right arm. Phase-contrast MRI was performed as described previously (23), using a 1.5-T scanner (CV/i, General Electric Healthcare Technologies, Milwaukee, Wisconsin) equipped with cardiac gradient coils (40 mT/m, 120 T/m/s). Electrocardiographic leads were placed on the thorax. To image the brachial artery, a 3-inch receiver coil was placed medial to the upper arm. An inflatable cuff was placed on the forearm, extending to just above the elbow. To image the femoral artery, a 4-element phased array receiver coil was placed anterior and posterior to the upper thigh. An inflatable cuff was placed on the lower thigh. The cuff was inflated at least 30 mm Hg above the subject’s measured systolic blood pressure for 5 min, then released. For each artery, phase-contrast images using the same fixed cross-sectional axial prescription were obtained before and immediately after cuff release. Serum values of glucose, hematocrit, and fasting lipid panel were obtained in all groups, and hemoglobin A1C in groups 2 and 3, after the scanning portion of the study.
Coronal and axial scout images were obtained to locate the brachial artery, to locate the superficial femoral artery at 3 to 5 cm distal to the bifurcation of the common femoral artery, and to verify that the arteries were parallel to the magnet bore. Phase-contrast scans were gated to the electrocardiogram signal. A single imaging plane perpendicular to the artery of interest was prescribed. The imaging parameters were: matrix size 256 × 128, field-of-view 8 × 8 cm for the brachial artery and 10 × 10 cm for the femoral artery, slice thickness 3 mm, flip angle 25°, bandwidth 31.2 kHz, repetition time 11.43 ms, echo time 5.25 ms, 8 views per segment, first-order flow compensation, no phase-wrap, and no magnitude weighting. Settings of 16 signal averages (NEX) were used at baseline, and 2 NEX after cuff release. During peak hyperemia, 10 views per segment were used if necessary to obtain a scan time of 35 s or less. The velocity encoding gradient was 50 to 70 cm/s during baseline and 120 to 150 cm/s during peak hyperemia. Resulting temporal resolution was about 90 to 180 ms. Each scan provided 5 to 9 magnitude (anatomic) and phase (velocity) images of the arterial cross section during the cardiac cycle.
Image data was imported via Scion Image (Scion Corporation, Frederick, Maryland) into a spreadsheet-based (Excel, Microsoft Corporation, Mountain View, California) program created in our laboratory. The program employs a user-independent algorithm to measure arterial radius, blood flow, and shear rate from the phase images. The algorithm located the precise center of the arterial cross section as where the velocity datapoints in a radial plot were optimally correlated. An approach simplified from Oyre et al. (24) was used to calculate shear rate and radius. The cardiac phase containing peak flow was used. A 1-mm wide segment of velocity datapoints in the radial plot near the lumen wall was fit by least-squares method to a parabola, with the assumption that blood flow velocity at the lumen wall is zero. The outer edge of the segment was defined where a smoothed average of the velocity profile decreased to 20% of the peak velocity in the cross section. Systolic shear rate was calculated as the slope of the best-fit parabola where the parabola equaled zero. The distance from this point to the center was defined as the radius of the artery. Systolic flow was calculated by summing the velocity pixels within the radius of the artery. This approach provided sub-pixel precision in determining shear rate and radius.
We previously showed in young healthy subjects that HypQ is strongly related to baseline radius squared when the brachial and femoral arteries are considered together (23). To compare HypQ between arm and leg, we normalized HypQ by dividing it by a function of baseline radius derived for the brachial and femoral arteries separately. The function was derived in young healthy subjects using a log-linear regression. For log (HypQ) = (a) log (baseline radius) + b, this means that HypQ = (b)(baseline radius)a. By design, HypQ normalized to baseline arterial radius should be close to 1 for both brachial and femoral arteries in the young healthy group. Hyperemic flow being normalized to baseline radius can be seen as a surrogate for normalizing HypQ to limb volume, since arterial diameters grow in proportion to volume of the downstream tissue (25).
Results are expressed as mean value ± SD. Student t tests were used for pairwise comparisons between groups. Paired t tests were used to compare measured parameters between baseline and hyperemia, and to compare normalized HypQ between femoral and brachial arteries. A p value <0.05 was considered significant.
Subject characteristics are summarized in Table 1. Compared with the younger healthy subjects, the older healthy subjects had higher body mass index, systolic blood pressure, fasting glucose (although still within the normal range), total cholesterol, and low-density lipoprotein, but they had greater high-density lipoprotein (HDL). Compared with the older healthy subjects, the older diabetic subjects had greater body mass index, systolic blood pressure, fasting glucose, triglycerides, hemoglobin A1C, and lower HDL. Only the diabetic subjects were on vasoactive medications. Of the 19 older diabetic subjects studied, 9 (47%) were on angiotensin-converting enzyme inhibitors, 2 (11%) were on angiotensin receptor antagonists, 2 (11%) were on calcium-channel blockers, 2 (11%) were on beta-blockers, and 12 (63%) were on statins. One subject was on a nitrate, which was held for 24 h before the scan.
MRI data analysis
Figure 1 shows the phase image obtained of typical brachial and femoral and arterial cross sections during systole at baseline. Beside each image is a radial plot of the datapoints. Each plot shows the best-fit parabola segment and measured shear rate. Figure 2 illustrates radial plots of velocity data during systole in the brachial and femoral arteries of a typical young control subject at baseline and during peak hyperemia. Systolic shear rate increased substantially during peak hyperemia.
The relationship between hyperemic systolic flow (HypQ) and baseline radius (R) was: Hyperemic systolic flow in brachial arteries was normalized for baseline radius: Hyperemic systolic flow in femoral arteries was normalized for baseline radius: Magnetic resonance imaging-derived measurements are summarized in Table 2. The groups were similar at baseline with regard to brachial arterial radius, brachial systolic flow, brachial systolic shear rate, femoral arterial radius, femoral systolic flow, and femoral systolic shear rate.
Hyperemic shear rate
Brachial hyperemic systolic shear rate tended to be lower in older healthy subjects than young healthy subjects (996 ± 305 s−1 vs. 1,041 ± 280 s−1, p = 0.28), and tended to be lower in older diabetic than older healthy subjects (909 ± 246 s−1 vs. 996 ± 305 s−1, p = 0.26); however, the groups were not statistically significantly different. In contrast, femoral hyperemic systolic shear rate was significantly lower in older healthy than young healthy subjects (680 ± 236 s−1 vs. 843 ± 157 s−1, p = 0.001), and was significantly lower in older diabetic than older healthy subjects (549 ± 183 s−1 vs. 680 ± 236 s−1, p = 0.02).
Change in shear rate from baseline to hyperemia
Change in brachial systolic shear rate from baseline to hyperemia tended to be lower in older healthy subjects than young healthy subjects (572 ± 196 s−1 vs. 654 ± 225 s−1, p = 0.06), and tended to be lower in older diabetic than older healthy subjects (530 ± 176 s−1 vs. 572 ± 196 s−1, p = 0.40) (Fig. 3A), but the groups were not statistically significantly different. In contrast, change in femoral systolic shear rate from baseline to hyperemia was significantly lower in older healthy subjects than young healthy subjects (251 ± 150 s−1 vs. 450 ± 120 s−1, p < 0.001), and was significantly lower in older diabetic than older healthy subjects (150 ± 142 s−1 vs. 251 ± 150 s−1, p = 0.01) (Fig. 3B).
HypQ normalized as a function of baseline arterial radius
Hyperemic brachial flow normalized as a function of brachial arterial radius was similar between older healthy and young healthy subjects (1.05 ± 0.28 mm1.30/s vs. 1.02 ± 0.19 mm1.30/s, p = 0.29), and was similar between older diabetic and older healthy subjects (1.05 ± 0.22 mm1.30/s vs. 1.05 ± 0.28 mm1.30/s, p = 0.48). In contrast, hyperemic femoral flow normalized as a function of femoral arterial radius was significantly lower in older healthy than young healthy subjects (0.84 ± 0.25 mm1.27/s vs. 1.01 ± 0.16 mm1.27/s, p = 0.001), and was lower in older diabetic than older healthy subjects (0.74 ± 0.19 mm1.27/s vs. 0.84 ± 0.25 mm1.27/s, p = 0.046). Femoral normalized HypQ was similar to brachial normalized flow in the young healthy group (1.01 ± 0.16 mm1.27/s vs. 1.02 ± 0.19 mm1.30/s, p = 0.45); however, femoral normalized HypQ was less than brachial normalized HypQ in the older healthy group (0.84 ± 0.25 mm1.27/s vs. 1.05 ± 0.28 mm1.30/s, p = 0.0001) and in the older diabetic group (0.74 ± 0.19 mm1.27/s vs. 1.05 ± 0.22 mm1.30/s, p < 0.0001).
The major finding of this study was that resistance artery reactivity was progressively reduced in the lower extremity but not the upper extremity as cardiovascular risk factors increased. This suggests that measuring resistance artery reactivity in the lower extremity may be more sensitive than in the upper extremity for assessing global risk of atherosclerotic cardiovascular disease. Also, to our knowledge, this is the first study to measure hyperemic shear rate directly in peripheral arteries of subjects with cardiovascular risk factors and to compare them with healthy subjects.
Mechanisms and risk associated with hyperemic shear rate and flow
Post-occlusion hyperemic shear rate and flow are partially dependent on endothelial nitric oxide release from resistance arteries (8–10). However, the hyperemic response is also dependent on other molecular mechanisms, including prostaglandins and adenosine (11). Nevertheless, reduced hyperemia is associated with cardiac risk factors, coronary artery disease, and cardiac events (8,12–17). The reasons why hyperemic response identifies groups at risk despite being only partially dependent on nitric oxide are not completely understood. Possible explanations are that: 1) partial dependence on nitric oxide is sufficient to detect abnormalities that precede overt atherosclerosis; or 2) additional mechanisms that are also related to cardiovascular risk underlie the hyperemic response. We believe that the latter is most likely because several other mechanisms that are associated with risk factors are also associated with reduced hyperemic shear or flow. These include increased central arterial stiffness, increased lower extremity arterial stiffness, and increased sympathetic nervous system activity (20,26). These processes promote or are markers for increased risk of global atherosclerosis or cardiac events (27–29), though they are not independent of endothelial dysfunction (29,30). Taken together with our findings, this suggests that reduced hyperemic shear or flow in the lower extremity may be a sensitive integrated marker of cardiovascular risk.
Comparison with other studies
We did not find differences in resting lower extremity shear rate, flow, or normalized flow between groups, in contrast with one study (31), but in agreement with others (21,22,26,32–34). Other studies found no differences in forearm hyperemia with risk factors (21,22,35), and also found that leg hyperemia may be reduced even when arm hyperemia is not (21,22). Also in agreement with our study, those studies showed that lower extremity HypQ measures compared with resting flow measures more clearly distinguished groups with different levels of risk (21,22). The insensitivity of brachial hyperemia to risk factors in prior studies may relate to their measurement of mean flow instead of systolic or diastolic hyperemic shear, as suggested by Mitchell’s study of over 2,000 subjects (13). Our results do not necessarily conflict with that study, but rather suggest that a much smaller sample size may be sufficient to demonstrate reduced lower extremity resistance artery reactivity in the presence of risk factors.
Angiotensin-converting enzyme inhibitors, angiotensin receptor antagonists, and statins improve endothelial function in the upper extremities of type 2 diabetic patients (36,37). Thus, hyperemic shear rate was lowest in older diabetic patients despite their use of medicines that improve endothelial function. Future studies should examine whether certain cardiovascular medicines affect vasoreactivity of lower extremity arteries differently than upper extremities.
Studies of reproducibility
Reproducibility of post-ischemic hyperemic shear rate and flow measurements are very good, whether measured by ultrasound, venous occlusion strain-gauge plethysmography, or MRI (23,38,39). Indeed, an ultrasound-based study found that reproducibility of brachial HypQ was better than that of FMD measurements (38). This enhances the potential utility of measuring resistance artery reactivity for assessing cardiovascular risk.
Lower versus upper extremity arteries
Recent studies show that arterial reactivity may be more impaired in legs than in arms in individuals with coronary artery disease or peripheral arterial disease (3,21). Coronary artery disease and peripheral arterial disease have common risk factors. The results of this study suggest that arterial dysfunction occurs earlier or more severely in lower than upper extremities of persons with risk factors but no overt atherosclerotic disease. Those studies, taken together with the results of our study, may help to explain why leg arteries usually develop more severe atherosclerosis than brachial arteries (2). Regional differences in arterial function may underlie regional differences in atherosclerotic development.
Resistance and conduit arteries
This study examined vasodilator function of resistance arteries—the microvasculature—by examining flow and shear rate in upstream conduit arteries, the brachial and femoral arteries. Future studies should investigate whether there are differences between legs and arms in macrovascular function—FMD of conduit arteries—in subjects with risk factors.
Magnetic resonance imaging has unique properties that compliment those of other methods in helping to understand certain aspects of vascular reactivity. Magnetic resonance imaging acquires the entire flow velocity profile within a cross section and is thus advantageous for directly measuring shear rate. For imaging the femoral artery, clothing and the depth of the artery are not barriers to MRI. Limitations of MRI compared with ultrasound include temporal and spatial resolution, expense, and lack of portability.
Regarding temporal resolution, the precise point of peak systolic flow within the cardiac cycle may have been missed slightly, thus the measured systolic shear rate and flow values may be slightly less than the true maximum systolic values. In addition, the magnetic resonance image acquisition begun immediately after cuff release was approximately 30 s in duration, although the primary data contributing to image formation was acquired at 15 s (middle of k-space) for the magnetic resonance pulse sequence used. A study by Leeson et al. (40) using ultrasound showed that brachial HypQ at 15 s after cuff release was not substantially decreased from 5 s after cuff release for a 4.5-min cuff occlusion duration. Nevertheless, systolic HypQ or shear rate could possibly be less sensitive to risk factors by magnetic resonance than by ultrasound. In any case, the increased sensitivity of femoral versus brachial measurements in distinguishing at-risk groups is unlikely to be an artifact of magnetic resonance properties. Our findings of increased sensitivity of the leg arteries to risk factors should be confirmed with ultrasound. A more accessible leg artery for ultrasound scanning, such as the popliteal artery, may provide similar sensitivity to the femoral artery.
The multiple statistical tests conducted, 28 in Table 2, increase the potential for significant findings by chance. Another limitation due to sample size is that the relative contribution of specific risk factors to decreased resistance artery reactivity could not be determined. However, future studies may be able to assess the relative importance of specific risk factors on leg artery reactivity with a smaller sample size than would be required if the arm were studied. Also, we did not investigate subjects with established atherosclerosis, which future studies should include.
Arterial reactivity in the lower but not upper extremity was progressively reduced with increasing cardiovascular risk factors. This implies early or more severe impairment of resistance artery vasodilator function in the lower extremities of persons with risk factors, and may help to explain why atherosclerosis usually develops more severely in the lower extremities than the upper extremities. Further, the findings suggest that resistance artery vasodilator activity may be a more sensitive indicator of global cardiovascular risk in the leg than the arm.
The authors gratefully acknowledge the assistance of Ann V. Munson, MS, RD, in coordinating subjects.
↵1 Drs. Gupta and Foo were employed by General Electric during this study
This study was supported by an NHLBI grant (K23 HL04477).
Michael O’Rourke, MD, DSc, FACC, served as guest editor for this article.
- Abbreviations and Acronyms
- flow-mediated dilation
- high-density lipoprotein
- hyperemic flow
- magnetic resonance imaging
- Received July 17, 2006.
- Revision received October 20, 2006.
- Accepted October 27, 2006.
- American College of Cardiology Foundation
- Lerman A.,
- Zeiher A.M.
- Silber H.A.,
- Bluemke D.A.,
- Ouyang P.,
- Du Y.P.,
- Post W.S.,
- Lima J.A.
- Koller A.,
- Kaley G.
- Nichols W.,
- O’Rourke M.F.
- Meredith I.T.,
- Currie K.E.,
- Anderson T.J.,
- Roddy M.A.,
- Ganz P.,
- Creager M.A.
- Klabunde R.E.
- Mitchell G.F.,
- Vita J.A.,
- Larson M.G.,
- et al.
- Cortella A.,
- Zambon S.,
- Sartore G.,
- et al.
- Silber H.A.,
- Ouyang P.,
- Bluemke D.A.,
- Gupta S.N.,
- Foo T.K.,
- Lima J.A.
- Radegran G.,
- Saltin B.
- Suzuki E.,
- Kashiwagi A.,
- Nishio Y.,
- et al.
- Herrington D.M.,
- Brown W.V.,
- Mosca L.,
- et al.
- Oliver J.J.,
- Webb D.J.
- Seals D.R.,
- Dinenno F.A.
- Wilkinson I.B.,
- Qasem A.,
- McEniery C.M.,
- Webb D.J.,
- Avolio A.P.,
- Cockcroft J.R.
- Dinenno F.A.,
- Jones P.P.,
- Seals D.R.,
- Tanaka H.
- Proctor D.N.,
- Le K.U.,
- Ridout S.J.
- Leeson P.,
- Thorne S.,
- Donald A.,
- Mullen M.,
- Clarkson P.,
- Deanfield J.