CLINICAL RESEARCH: ASSESSING VASCULAR FUNCTION
Methodological Approaches to Optimize Reproducibility and Power in Clinical Studies of Flow-Mediated Dilation
Ann E. Donald, AVS*,*,
Julian P. Halcox, MD, MA, FRCP*,
Marietta Charakida, MD, PhD*,
Clare Storry, BSc, AVS*,
Sharon M.L. Wallace, BA, DipHE, RN*,
Tim J. Cole, ScD ,
Peter Friberg, MD, PhD* and
John E. Deanfield, BA, BChir, MB, FRCP*
* Vascular Physiology Unit, UCL Institute of Child Health, London, United Kingdom
Centre for Pediatric Epidemiology and Biostatistics, UCL Institute of Child Health, London, United Kingdom.
Manuscript received October 31, 2007;
revised manuscript received January 7, 2008,
accepted February 5, 2008.
* Reprint requests and correspondence: Ms. Ann E. Donald, Vascular Physiology Unit, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, United Kingdom. (Email: A.Donald{at}ich.ucl.ac.uk).
 |
Abstract
|
|---|
Objectives: Our aim was to determine reproducibility of the flow-mediated dilation (FMD) response profile, and discriminatory ability of the components.
Background: Brachial FMD is widely used to study conduit artery endothelial function. Automated B-mode image edge detection (B-ED) provides a full response profile. Reproducibility and biological relevance of these additional components have not been fully explored.
Methods: Forty-two healthy adults underwent FMD using B-ED repeated at fixed time intervals up to 3 months. The FMD profile was assessed for diameter changes, area under the curve, and time course. Measures were compared in 25 adults with hypercholesterolemia, 25 subjects with diabetes, and 50 matched control subjects.
Results: The maximum change in FMD was the most reproducible (coefficient of variation = 9.8%, 10.6%, 6.6%, and 9.2% at 4 to 6 h, 1 week, 1 month, and 3 months, respectively). Most of the variability occurred between subjects rather than within. All FMD measures except time course were significantly reduced in hypercholesterolemia and diabetes. Power curves were generated to indicate the appropriate number of subjects for parallel and crossover study designs.
Conclusions: Maximum FMD percentage change from baseline is the most reproducible of the response curve measures and best identifies those with risk factors. Flow-mediated dilation measured by B-ED is robust and practical to assess the effect of interventions on endothelial function in clinical trials.
|
Abbreviations and Acronyms
| | A-WT = A-mode wall tracking | | B-ED = B-mode edge detection | | CV = coefficient of variation | | FMD = flow-mediated dilation | | FMDmax = maximum flow-mediated dilation percentage change from baseline | | NO = nitric oxide | | T2DM = type 2 diabetes mellitus |
|
Atherosclerosis begins in early life. Strategies to study and manage ‘lifetime’ risk from atherosclerosis require robust intermediate phenotypes, on the causal pathway for disease. Endothelial function is linked to cardiovascular risk factors, provides prognostic information, and can be studied noninvasively by measurement of flow-mediated dilation (FMD) (1). This reflects local nitric oxide (NO) bioavailability and enables both examination of mechanisms involved in the initiation and progression of pre-clinical vascular disease and the impact of acute and long-term interventions (2).
While the principles of FMD have not changed, operator technique and analysis have been refined (3). We now report the reproducibility of components of the FMD response over time periods up to 3 months, and their ability to discriminate between healthy subjects and those with diabetes or hypercholesterolemia. Power curves have been generated to assist design of both crossover and parallel trials to evaluate the impact of interventions on endothelial function. We have also compared B-mode edge detection (B-ED) software (Medical Imaging Applications, Coralville, Iowa) to A-mode wall tracking (A-WT) (Vadirec, Medical Systems Arnhem, Oosterbeek, the Netherlands) an early semiautomatic method relying on M-mode vascular images (4,5).
|
Methods
|
|---|
Study protocols.
Study 1: Reproducibility of FMD Components by B-ED
We studied 42 healthy adults (19 men, mean age 43 years [range 20 to 75 years]) free from cardiovascular risk factors at 4- to 6-h, 1-week, 1-month, and 3-month intervals.
Study 2: FMD Profile in Subjects With Type 2 Diabetes Mellitus (T2DM) or Hypercholesterolemia
Twenty-five patients (16 men, age 53 years [range 39 to 67 years]) with T2DM  2 years duration with fasting glucose 6.9 mmol/l (124 mg/dl) and/or on hypoglycemic treatment (16 sulphonylureas, 9 metformin, 7 combination, and 7 diet) were matched with 25 healthy control subjects recruited contemporaneously.
Twenty-five patients (18 men, age 50 years [range 19 to 67 years]) with total fasting cholesterol >6.0 mmol/l (232 mg/dl) (7 on statins) were compared with 25 healthy control subjects matched for age, gender, and vessel diameter.
Study 3: Reproducibility of FMD by A-WT
We studied 36 healthy adults free of cardiovascular risks (14 men, 33 years [range 23 to 61 years]) on 2 occasions 1 month apart.
Assessment of endothelial function.
All subjects gave informed written consent. Studies were approved by the local ethics committee and were carried out under standardized conditions by trained operators:
- • Rest 10 min in a warm, quiet room (22°C to 26°C) after 4 h abstinence from food and caffeinated drinks.
- • Right arm extended laterally and brachial artery imaged longitudinally 5 to 10 cm above antecubital fossa (image stability achieved by anterior plane, scanning through the biceps muscle with the probe in a stereotactic holder with micrometer movement).
- • Pneumatic cuff (7 x 30 cm) (Scanmed, Moreton-in-Marsh, Gloucestershire, United Kingdom) placed immediately below the medial epicondyle. Inflation time 5 min at 300 mm Hg. The arm, hand, and head position noted, along with the distance between the pneumatic cuff and the probe. A thermal print of the arterial image was taken for matching at subsequent visits.
- • B-mode ultrasound images were updated on the R-wave of the electrocardiogram, with Doppler flow and electrocardiogram displayed continuously.
B-ED methodology.
A standard protocol was used as previously described (3). The lumen diameter was measured from within the same region-of-interest box for each image (Fig. 1A). Diameters were plotted for FMD and area under the curve measurement (Fig. 1B). Additionally, the incremental area under the diameter versus time curve was calculated, accounting for baseline diameter over intervals post-cuff release of 0 to 1 min, 0 to 3 min, and 0 to 5 min (0 = cuff deflation point). The times to reach maximum dilation and to recover to half the absolute change in diameter after cuff release were also assessed to examine the response time course after reactive hyperemia. All analyses were conducted by an experienced investigator (A.E.D.) blinded to subject identity and visit order.
View larger version (43K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1 Image Acquisition and Readout for FMD Using B-ED and A-WT
(A) B-mode ultrasound image of the brachial artery, with region-of-interest box for edge detection diameter measurement. (B) Flow-mediated dilation (FMD) using B-mode edge detection (B-ED). (C) B-mode image with perpendicular M-line for A-mode wall tracking (A-WT). (D) A-mode wall tracking output. (Upper trace) A-mode radio-frequency signal. (Middle trace) Anterior and posterior wall motion. (Lower trace) Distension waveform generating end-diastolic diameter. AUC = area under the curve.
|
|
A-WT methodology.
A standard protocol was used as previously described (6). The cuff was placed 4 to 6 cm below the elbow crease to reduce image movement on cuff inflation/deflation. A single M-line was placed perpendicular to the vessel walls on the image to record the systolic to diastolic excursion of the brachial artery (Fig. 1C) providing the distension waveform and end-diastolic diameter (Fig. 1D).
Statistical analysis.
All measures are expressed as mean ± standard deviation unless otherwise stated.
Reproducibility for each time interval between pairs of visits was expressed as the coefficient of variation (CV) of a single measurement, defined as the standard deviation of the difference between paired values divided by the mean and divided by  2, as previously described (3). Paired t tests were used to compare patients and control subjects. To assess the number of subjects required for crossover and parallel studies, for 80% statistical power and 5% significance, and to assess the effect of varying the number of readings pre- and post-treatment, power curves were constructed from the between- and within-subject variances as previously reported (7). Reproducibility of B-ED and A-WT over 1 month was compared using Bland-Altman plots. SPSS (version 12, SPSS Inc., Chicago, Illinois) was used for analyses.
|
Results
|
|---|
Study 1.
Six of 144 scans were excluded (1 noncompliance with fasting protocol, 2 poor image quality, 2 subject movement, 1 equipment failure). Baseline vessel diameter, baseline blood flow, reactive hyperemia, blood pressure, and heart rate were similar for each interval. The maximum change in diameter expressed as a percentage and as an absolute value, maximum flow-mediated dilation percentage change from baseline (FMDmax), and absolute FMD change from baseline, respectively, were the most reproducible components of the FMD profile at each interval (Table 1). Maximum FMD percentage change from baseline correlated inversely with age (p < 0.01), female subjects had significantly smaller vessels than male subjects (p < 0.01), and vessel size correlated with age (p < 0.01). Between- and within-subject variance was similar at all visit intervals (13.3, 11.9, 14.3, and 11.1 and 1.3, 1.3, 0.6, and 1.1 for studies done 4 to 6 h, 1 week, 1 month, and 3 months apart, respectively). We constructed power curves from the data recorded 4 to 6 h and 3 months apart for both crossover and parallel studies for protocols using 1, 2, or 4 FMD measurements before and after an intervention (Fig. 2). Increasing the number of assessments in the protocol had minimal impact in a parallel study design. By contrast, in a crossover design, increasing the number of FMD scans reduced the number of subjects slightly, but at the cost of a more complex protocol.
View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2 Number of Subjects Required in Crossover and Parallel Designs for Given FMD Effect
Relation between effect on maximum percent change in flow-mediated dilation (FMD) (%) and number of subjects required in crossover and parallel study designs at 80% power and 5% significance, 4 to 6 h and 3 months apart with 3 monitoring strategies: 1, 2, or 4 measures pre- and post-treatment.
|
|
Study 2.
Disease groups and control groups had similar diameters and flows at baseline. All components of the FMD response except peak time and recovery half time were lower in subjects with T2DM or hypercholesterolemia, but did not further differentiate patients from control subjects compared with FMDmax. Absolute FMD change from baseline and FMDmax showed the greatest differences between patients and control subjects for both disease groups (Table 2).
Study 3.
Comparing A-WT with B-ED over a 1-month interval, reproducibility of baseline diameter was comparable (CV 2.6% vs. 3.5%). However, FMD at 60 s after cuff deflation using A-WT was lower (4.3 ± 3.6% vs. 7.5 ± 3.8% at Visit 1, and 4.6 ± 3.1% vs. 7.2 ± 4.0% at Visit 2, both p < 0.01) and more variable (CV 17.8% vs. 8.6%) than B-ED (Fig. 3).
View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3 Bland Altman Plots for FMD60s in Study 3 Over 1 Month Showing B-ED and A-WT
The dotted lines represent the 95% confidence intervals. The x-axis shows the average flow-mediated dilation (FMD) and the y-axis the difference between pairs. A-WT = A-mode wall tracking; B-ED = B-mode edge detection; FMD60s = change in flow-mediated dilation at 60 s after cuff deflation.
|
|
|
Discussion
|
|---|
We report the opportunity, afforded by current ultrasound methodology, to evaluate endothelial function in a reproducible manner and the potential application to detect abnormality in disease states. Power curves have been constructed demonstrating that FMD measurement is practical in both crossover and parallel studies.
In 1992, we described the noninvasive method for FMD, which has been shown to be NO dependent and also to reflect coronary endothelial function (1,8,9). The technique has been widely adopted in clinical research, but, manual measurements are time consuming and subjective. Newer semiautomatic measurement software is faster, less subjective, and potentially more reproducible. This, combined with greater image stability afforded by fixing the ultrasound probe in an adjustable stereotactic clamp and adherence to a rigorous protocol, improves the accuracy and reproducibility. In contrast to early analyses, B-ED measures the whole dilation profile. We show that, of all the potential measures, FMDmax is both the most reproducible and the best discriminator of endothelial function abnormalities between health and disease. The ‘area under the response curve’ adds little value compared with the simpler FMDmax. The time to FMDmax and the recovery time were similar in health and disease suggesting that the magnitude rather than time course of vasodilation best reflects differences in endothelial function.
A-mode wall tracking is more variable than B-ED, despite excellent spatial resolution. This may reflect difficulty in maintaining positional stability of a single M-line without 2-dimensional image guidance. Measurements must be performed at set time points (60 s), limiting reproducibility and also producing lower FMD, missing the maximal response in most cases. Maximal dilation occurred between 24 s and 117 s (mean 59 s) after cuff release so that readings taken at 60 s are maximal for only 30% of subjects (10).
The low CV with B-ED was accomplished by meticulous standardization of the technique and operator training. Physiological variability can be reduced by controlling factors known to influence endothelial function, such as temperature, recent food and drink intake, exercise, and infection (2,11). Previous studies have reported variability during the menstrual cycle (12). Although our study was not designed to test this issue, reproducibility was similar in women and men.
We investigated subjects age 20 to 71 years, with a wide range of vessel diameters and FMD responses, using a protocol that reflects predominantly NO bioavailability (9). Dilation induced by stimuli such as hand warming, prolonged ischemia, or cuff inflation above the site of measurement are less NO dependent, less extensively studied, and require further evaluation of their biological significance (9,13).
A-mode wall tracking provided consistently lower FMD than B-ED, likely due to a more distal cuff position used to minimize movement artifact. Standardization of cuff position is crucial in multicenter and interventional studies and may explain, in part, previously reported variation in FMD (14).
Flow-mediated dilation measured by B-ED is reproducible and practical for studying endothelial function in clinical research from a young age. As the major component of variability in FMD is between subjects, crossover studies can be performed with sufficient statistical power using substantially fewer subjects than for parallel trial designs. The additional complexity resulting from multiple assessments of FMD adds very little to the power of the study. This information on vascular disease biology complements studies of arterial structure and clinical outcome, which require larger and more expensive protocols.
|
Footnotes
|
|---|
Ms. Donald was supported by the Coronary Artery Disease Research Association; Drs. Halcox and Deanfield were supported by the British Heart Foundation; Dr. Charakida was supported by the Greek State Scholarship Foundation; and Dr. Friberg was supported by the Swedish Medical Research Council.
|
References
|
|---|
1. Celermajer DS, Sorensen KE, Gooch VM, et al. Non-invasive detection of endothelial dysfunction in children and adults at risk of atherosclerosis Lancet 1992;340:1111-1115.[CrossRef][Web of Science][Medline]2. Deanfield JE, Halcox JP, Rabelink TJ. Endothelial function and dysfunction: testing and clinical relevance Circulation 2007;115:1285-1295.[Free Full Text] 3. Donald AE, Charakida M, Cole TJ, et al. Non-invasive assessment of endothelial function: which technique? J Am Coll Cardiol 2006;48:1846-1850.[Abstract/Free Full Text] 4. Sonka M, Liang W, Lauer RM. Automated analysis of brachial ultrasound image sequences: early detection of cardiovascular disease via surrogates of endothelial function IEEE Trans Med Imaging 2002;21:1271-1279.[CrossRef][Web of Science][Medline] 5. Hoeks AP, Brands PJ, Smeets FA, Reneman RS. Assessment of the distensibility of superficial arteries Ultrasound Med Biol 1990;16:121-128.[CrossRef][Web of Science][Medline] 6. Leeson P, Thorne S, Donald A, Mullen M, Clarkson P, Deanfield J. Non-invasive measurement of endothelial function: effect on brachial artery dilatation of graded endothelial dependent and independent stimuli Heart 1997;78:22-27.[Abstract/Free Full Text] 7. Sorensen KE, Celermajer DS, Spiegelhalter DJ, et al. Non-invasive measurement of human endothelium dependent arterial responses: accuracy and reproducibility Br Heart J 1995;74:247-253.[Abstract/Free Full Text] 8. Anderson TJ, Uehata A, Gerhard, MD, et al. Close relation of endothelial function in the human coronary and peripheral circulations J Am Coll Cardiol 1995;26:1235-1241.[Abstract] 9. Mullen MJ, Kharbanda RK, Cross J, et al. Heterogenous nature of flow-mediated dilatation in human conduit arteries in vivo: relevance to endothelial dysfunction in hypercholesterolemia Circ Res 2001;88:145-151.[Abstract/Free Full Text] 10. Palinkas A, Toth E, Venneri L, Rigo F, Csanady M, Picano E. Temporal heterogeneity of endothelium-dependent and -independent dilatation of brachial artery in patients with coronary artery disease Int J Cardiovasc Imaging 2002;18:337-342.[CrossRef][Web of Science][Medline] 11. Charakida M, Donald AE, Terese M, et al. Endothelial dysfunction in childhood infection Circulation 2005;111:1660-1665.[Abstract/Free Full Text] 12. Hashimoto M, Akishita M, Eto M, et al. Modulation of endothelium-dependent flow-mediated dilatation of the brachial artery by sex and menstrual cycle Circulation 1995;92:3431-3435.[Abstract/Free Full Text] 13. Doshi SN, Naka KK, Payne N, et al. Flow-mediated dilatation following wrist and upper arm occlusion in humans: the contribution of nitric oxide Clin Sci (Lond) 2001;101:629-635.[Medline] 14. Bots ML, Westerink J, Rabelink TJ, de Koning EJ. Assessment of flow-mediated vasodilatation (FMD) of the brachial artery: effects of technical aspects of the FMD measurement on the FMD response Eur Heart J 2005;26:363-368.[Abstract/Free Full Text]
Related Articles
-
Listening to the Endothelium: A Story of Signal and Noise
- Robert Vogel
J. Am. Coll. Cardiol. 2008 51: 1965-1966.
[Full Text]
[PDF]
-
Inside This Issue of JACC
J. Am. Coll. Cardiol. 2008 51: A31-A32.
[Full Text]
[PDF]
This article has been cited by other articles:
|
|
|
|
 
A. J. Donato, L. B. Gano, I. Eskurza, A. E. Silver, P. E. Gates, K. Jablonski, and D. R. Seals
Vascular endothelial dysfunction with aging: endothelin-1 and endothelial nitric oxide synthase
Am J Physiol Heart Circ Physiol,
July 1, 2009;
297(1):
H425 - H432.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. Vogel
Listening to the Endothelium: A Story of Signal and Noise
J. Am. Coll. Cardiol.,
May 20, 2008;
51(20):
1965 - 1966.
[Full Text]
[PDF]
|
|
|
|
Top
Abstract
Methods