Does arterial shear explain the magnitude of flow-mediated dilation?: a comparison between young and older humans

Abstract

Flow-mediated dilatation (FMD) has become a commonly applied approach for the assessment of vascular function and health in humans. Recent studies emphasize the importance of normalizing the magnitude of FMD to its apparent eliciting stimulus, the postdeflation arterial shear. However, the relationship between shear stress and the magnitude of FMD may differ between groups. The aim of this study was to examine the relationship between the brachial FMD and four different indexes of postdeflation shear rate (SR) in healthy children (n = 51, 10 ± 1 yr) and young (n = 57, 27 ± 6 yr) and older (n = 27, 58 ± 4 yr) adults. SR was calculated from deflation (time 0) until 9 s (peak), 30 s (0–30), 60 s (0–60), or until the time-to-peak diameter in each individual (0-ttp). Edge detection and wall tracking of high resolution B-mode arterial ultrasound images were used to calculate the conduit artery diameter. In young adults, the brachial artery FMD demonstrated a significant correlation with the area under the SR curve (SR_AUC) 0–30 s (r² = 0.12, P = 0.009), 0–60 s (r² = 0.14, P = 0.005), and 0-ttp (r² = 0.14, P = 0.005) but not for the peak SR_AUC 0–9 s (r² = 0.04, P = 0.12). In children and older adults, the magnitude of the brachial artery FMD did not correlate with any of the four SR_AUC stimuli. These findings suggest that in young subjects, postdeflation SR_AUC correlates moderately with the magnitude of the FMD response. However, the relationship between FMD and postdeflation shear appears to be age dependent, with less evidence for an association in younger and older subjects. Therefore, we support presenting SR_AUC stimuli but not normalizing FMD responses for the SR_AUC when using this technique.

flow-mediated dilatation (FMD) describes the vasodilatation of a conduit artery following an increase in shear stress, typically induced by 5 min ischemia provoked by a limb cuff placed distal to the recording probe (2). With the assumption that the appropriate methodology is used (12, 39), the FMD response is largely nitric oxide mediated (14, 20) and provides information about the endothelial function (40). Since impaired endothelial vasodilator function is considered an early and integral manifestation of vascular disease, which predicts cardiovascular events (11, 16, 18, 21, 26), the FMD approach has become a popular and frequently applied technique in clinical and physiological research.

In a series of experiments, Pyke et al. (28–30) explored the relationship between FMD and its apparent eliciting stimulus. They concluded that the area under the shear rate (SR) curve following cuff deflation (SR_AUC), rather than peak SR, is the critical determinant of the FMD response. As a result of recommendations arising from these studies, including the suggestion that the appropriate strategy for normalization of the FMD response is to relate FMD to the total SR_AUC from cuff deflation to the time of peak diameter measurement (29), recent studies have begun to correct the magnitude of the peak diameter for the eliciting shear (1, 5, 15, 17, 24, 42). However, the suggestion that SR_AUC be used to normalize FMD is based on repeated-measure studies of relatively small samples (n = 8–24) of healthy, young adults (28, 29, 33), and this relationship may conceivably differ with age and cardiovascular risk. Furthermore, confusion persists regarding the appropriate time frame across which SR_AUC should be calculated, with different studies using different predetermined time windows (3, 8, 15, 22, 25, 38) (e.g., 0–30 or 0–60 s) or the time-to-peak diameter (1). The aim of this study was to examine the relationship between FMD and four different SR_AUC stimuli that have been used in the literature in a cross section of children and young and older adults. Our aim was to identify the best method to normalize FMD for differences in its eliciting SR stimulus. We hypothesized that correcting the FMD for the SR_AUC from cuff deflation to the time of peak diameter, but not peak SR or a SR_AUC calculated within a predetermined time window, will abolish individual differences in the magnitude of the FMD response within and between the age groups.

METHODS

Subjects.

Healthy volunteers (n = 135) were recruited from the community and were stratified based on age into three different groups: children (n = 51, 9–10 yr), young adults (n = 57, 20–41 yr), and older adults (n = 27, 50–66 yr) (Table 1). All subjects were nonsmokers and normotensive (<140/90 mmHg), and none had any history of diabetes, insulin resistance, or cardiovascular disease. None of the subjects used any medications known to interfere with the cardiovascular system. The Ethics Committee of the Liverpool John Moores University approved the study protocols, which adhered to the Declaration of Helsinki. Informed consent from subjects and informed consent of parents/guardians of children were obtained before their participation in the study.

Table 1.

Subject characteristics of the participants divided into a group of children and young and older adults

	Children	Young Adults	Older Adults	ANOVA
n	51	57	27
Age, yr	10±1*	27±6*	58±4*	<0.001
Sex, male/female	20/31	48/9	15/12
Systolic blood pressure, mmHg	106±8*	116±10	121±16	<0.001
Diastolic blood pressure, mmHg	63±4*	67±7*	71±10*	<0.001
Mean blood pressure, mmHg	78±5*	81±11*	87±11*	<0.001
Height, cm	142±7*	177±8*	168±9*	<0.001
Weight, kg	39.7±9.1*	76.5±11.3	75.1±17.6	<0.001
BMI, kg/m2	19.5±3.2*	24.3±2.8*	26.2±4.6*	<0.001

Values are means ± SD; n, number of participants.

^*P < 0.05, post hoc significant from all other groups.

Experimental design.

After the participants reported to the laboratory, the brachial artery FMD response was measured after a resting period of at least 20 min. Testing was performed between 9:00 am and 4:00 pm. Previous studies found, despite an attenuated FMD immediately after waking, that the FMD does not appear to be influenced across the time frame adopted in our study, when testing is performed under well-controlled conditions (23, 34). Accordingly, all measures were performed under standardized conditions in a quiet, temperature-controlled room and after at least 6 h of fast and at least 8 h of abstinence from caffeine or alcohol. No subject performed strenuous physical activity for at least 24 h before testing, since this can result in a transient change in FMD (4).

Brachial artery FMD.

Patients rested supine with the right arm extended and immobilized with foam, supported at an angle of ∼80° from the torso. Heart rate and mean arterial pressure were determined from an automated sphygmomanometer (GE Pro 300V2, Dinamap, Tampa, FL) on the contralateral arm. For the assessment of the FMD response, a rapid inflation/deflation pneumatic cuff was positioned on the imaged arm distal to the olecranon process to provide a stimulus to forearm ischemia. A 7.5- (Aspen, Acuson; Mountain View, CA) or 10-MHz (T3000, Terason, Aloka, UK) multifrequency linear array probe attached to a high-resolution ultrasound machine was used to image the brachial artery in the distal third of the upper arm. Ultrasound parameters were set to optimize longitudinal B-mode images of the lumen/arterial wall interface. A continuous Doppler velocity assessment was obtained simultaneously, and data were collected using the lowest possible insonation angle (always <60°), which did not vary during each study (27). After a resting period of at least 20 min, 1 min of baseline recording of the brachial artery diameter and velocity was performed. Subsequently, the occlusion cuff was inflated to >200 mmHg for 5 min. The brachial artery diameter and velocity recordings were restarted at least 30 s before cuff deflation and continued for at least 3 min after deflation. Peak artery diameter and flow, and the time to reach this peak after cuff deflation, were recorded.

Brachial artery diameter and blood flow.

Posttest analysis of the brachial artery diameter was performed using custom-designed edge-detection and wall-tracking software, which is independent of investigator bias (41). Briefly, the echo-Doppler signal was real-time encoded and stored as a digital file when using the Terason ultrasound machine. When using the Aspen machine, the video signal was taken directly from the ultrasound machine and, using an IMAQ-PCI-1407 card, was encoded and stored as a digital DICOM file on the PC. In both situations, subsequent software analysis of the data was performed at 30 Hz using an icon-based graphical programming language and toolkit (LabView 6.02, National Instruments, Austin, TX).

The initial phase of image analysis involved the identification of regions of interest (ROIs) on the first frame of every individual study. These ROIs allowed for an automated calibration of diameters on the B-mode image and velocities on the Doppler strip. A ROI was drawn around the optimal area of the B-mode image, and within this ROI, a pixel-density algorithm automatically identified the angle-corrected near and far-wall e-lines for every pixel column within the ROI. The algorithm begins by dividing the ROI into an upper half, containing the near-wall lumen-intima interface, and a lower half containing the far-wall interfaces. The near-wall intimal edge is identified by a Rake routine that scans from the bottom to the top of the upper half of the ROI. The position of the edge is established by determining the point where the pixel intensity changes most rapidly. Typical B-mode ROIs, therefore, contained ∼200–300 diameter measures per frame, the average of which was calculated and stored. This process occurred at 30 frames/s. A final ROI was drawn around the Doppler waveform and automatically detected the peak of the envelope of this waveform 30 times/s for off-line analysis. The mean diameter measure derived from within the B-mode ROI (above) was synchronized with the velocity measure derived from the Doppler ROI at 30 Hz. Ultimately, from this synchronized diameter and velocity data, blood flow (the product of cross-sectional area and Doppler velocity) and SR (4 times velocity divided by diameter) were calculated at 30 Hz. All data were written to the file and retrieved for analysis in a custom-designed analysis package. We have shown that the reproducibility of diameter measurements using this semiautomated software is significantly better than manual methods, reduces observer error significantly, and possesses an intraobserver coefficient of variation of 6.7% (41).

Data analysis.

Baseline diameter and blood flow were determined during the 1 min before cuff inflation. Peak diameter following cuff deflation was automatically detected according to an algorithm that identified the maximum bracket of data subsequent to the performance of a moving window smoothing function. This smoothing routine calculates the median value from 100 consecutive frames (∼3 s) before the window shifts to the next bracket of data that shares 20% overlap with the preceding bracket (see Ref. 1 for more details). The maximum value of all the calculated median values is then automatically detected and chosen to represent the peak of the diameter curve. FMD was calculated as the percent rise of this peak diameter from the preceding baseline diameter.

From the simultaneously acquired blood flow velocity estimates and diameter measures, SR was calculated at 30 Hz. The postdeflation SR data were exported to a spreadsheet, and the SR_AUC was calculated based on the Riemann sum technique. To examine the relationship between diameter dilation (FMD%) and the eliciting SR, we then calculated the most frequently applied methods for SR correction used in the literature: peak SR (AUC from 0–9 s) (3, 29) and the SR_AUC from cuff deflation up to the arbitrary time point of 30 (15) and 60 s (8) and from deflation to the time of peak diameter for each individual (1).

Statistics.

Statistical analyses were performed using SPSS 14.0 (SPSS, Chicago, Illinois) software. All data are reported as means (SD), and the statistical significance was assumed at P ≤ 0.05. Pearson's correlation coefficient was used to examine the correlation between the different correction methods for the eliciting shear stimulus and the FMD for the whole group, but also for the different age groups. These correlations were calculated for the FMD presented as a relative (FMD%) as well as the absolute change in millimeter from baseline (FMD_mm) (35). A linear regression model was used to examine the impact of various factors [SR_AUC stimuli, diameter, blood pressure, and body mass index (BMI)] upon the magnitude of the FMD%. In addition, an ANOVA procedure was used to examine differences in baseline characteristics between the three groups. Post hoc analysis was used to identify the groups that differed significantly.

RESULTS

Subject characteristics.

Age, height, BMI, and diastolic blood pressure differed significantly between all three groups (Table 1). Systolic arterial blood pressure and weight were significantly lower in children, compared with young and older adults, whereas older men had a higher systolic blood pressure than children and young adults (Table 1).

Brachial artery FMD% versus SR stimuli.

Baseline and peak brachial artery diameter was significantly lower in children compared with young and older adults (Table 2). The FMD, when expressed as the relative change from baseline, was significantly higher in children than in young and older adults. The time-to-peak diameter did not significantly differ between groups (Table 2).

Table 2.

Brachial artery characteristics at baseline and during the FMD response of participants divided into a group of children and young and older adults

	Children	Young Adults	Older Adults	ANOVA
n	51	57	27
Baseline diameter, mm	2.6±0.3*	4.1±0.6	4.1±0.9	<0.001
Peak diameter, mm	2.9±0.4*	4.4±0.7	4.4±1.0	<0.001
Change from baseline, mm	0.3±0.1	0.3±0.1	0.2±0.1†	0.047
Change from baseline, FMD%	10.7±4.9*	7.5±3.1*	6.0±2.9*	<0.001
Time-to-peak dilation, s	71±32	71±37	86±32	0.14
Peak SRAUC, s−1, ×103	5.0±2.1*	4.3±2.2	3.2±2.2	<0.001
SRAUC 0–30 s, s−1, ×103	18.6±5.4*	11.4±5.9	11.9±5.9	<0.001
SRAUC 0–60 s, s−1, ×103	31.3±8.8*	17.0±8.8	17.5±7.9	<0.001
SRAUC 0-ttp, s−1, ×103	35.6±16.7*	17.5±8.8	20.4±9.5	<0.001
FMD/SRAUC 0-ttp, s−1, ×10−4	4.3±2.4	5.3±3.1	3.4±1.9†	0.045

Values are means ± SD; n, number of participants. Four different methods for postdeflation shear rate area-under-the-curve (SR_AUC) calculation were presented: peak shear rate and shear rate 0–30 s, 0–60 s, and 0 up to time-to-peak (0-ttp) diameter. FMD%, flow-mediated dilatation, presented as %change from baseline diameter.

^*P < 0.05, post hoc significant from all other groups.

^†P < 0.05, post hoc significant between young and older adults.

The SR_AUC differed significantly between the four methods (peak, 0–30 s, 0–60 s, and 0-ttp) for children and young and older adults (ANOVA, all P < 0.001), with the largest SR_AUC stimulus when calculating SR until the time-to-peak diameter (Table 2).

Correlations across and within different age groups.

The correlation between the magnitude of the FMD% response and the four different SR_AUC stimuli for the pooled data set (n = 135) revealed no significant correlations for the peak SR_AUC (Fig. 1). In contrast, the FMD% correlated significantly with the SR_AUC stimulus when calculated to 30 s, 60 s, and until the individual time of peak diameter (Fig. 1). The magnitude of the FMD_mm response and the four different SR_AUC stimuli for the pooled data set revealed no correlations.

Fig. 1.

Correlations between brachial artery flow-mediated dilatation, presented as %change from baseline diameter (FMD%) and 4 different shear rate stimuli. Peak shear rate (A), 0–30-s shear rate (B), 0–60-s shear rate (C), and 0 time-to-peak diameter (0-ttp) shear rate (D) are shown in the pooled data set of children (n = 51, ▴), young adults (n = 57, ), and older adults (n = 27, ○).

In parallel with the pooled data, we found no correlation between the brachial artery FMD% and the peak SR_AUC stimulus in the individual groups of children and young and older adults (Figs. 2, 3, and 4). Younger adults showed a significant correlation between the FMD% and the SR_AUC stimulus until 30 s, 60 s, and the time of peak diameter (Fig. 3). In marked contrast, children and older adults reported no significant correlation between the FMD% and the SR_AUC stimuli (Figs. 2 and 4, respectively). The FMD_mm response and the four different SR_AUC stimuli did not correlate in children or young or older adults.

Fig. 2.

Correlations between brachial artery FMD% and 4 different shear rate stimuli. Peak shear rate (A), 0–30-s shear rate (B), 0–60-s shear rate (C), and 0-ttp shear rate (D) are shown in children (n = 51, ▴).

Fig. 3.

Correlations between brachial artery FMD% and 4 different shear rate stimuli. Peak shear rate (A), 0–30-s shear rate (B), 0–60-s shear rate (C), and 0-ttp shear rate (D) are shown in young adults (n = 57, ).

Fig. 4.

Correlations between brachial artery FMD% and 4 different shear rate stimuli. Peak shear rate (A), 0–30-s shear rate (B), 0–60-s shear rate (C), and 0-ttp shear rate (D) are shown in older adults (n = 27, ○).

Although FMD% correlated significantly with age (r = −0.39, r² = 0.15, P = 0.004), BMI (r = −0.36, r² = 0.13, P < 0.001), mean arterial blood pressure (r = −0.17, r² = 0.03, P = 0.03) and SR_AUC stimuli (Fig. 1), a stepwise linear regression model performed to examine the contribution of these various parameters to the FMD% in the pooled data set (n = 135) identified baseline diameter as the strongest predictor for the magnitude of the FMD% (r = −0.53, r² = 0.28, P < 0.001). The other variables listed above did not demonstrate a significant β-coefficient and could not be added to the regression model.

DISCUSSION

In recent years, several studies have provided support for the concept that the FMD response should be normalized for the magnitude of its eliciting postdeflation shear stress stimulus to make valid comparisons between subjects (17, 29, 32). While this principle is based on studies adopting a “within” subjects approach in small samples of young subjects (28, 29, 33), we examined the relationship between artery dilation and four commonly adopted methods to correct for the eliciting shear stimulus between subjects and between groups (children and young and older adults). We found a significant correlation between the FMD% response and the SR_AUC using a predetermined window (0–30 and 0–60 s) or between deflation and the individual time-to-peak diameter, but not for the peak SR. This finding confirms the above studies, which were performed in similarly young and healthy subjects, indicating that the SR_AUC, rather than the peak SR, determines the magnitude of the FMD (29). However, relatively weak correlations were found in this group of 57 subjects, with the SR_AUC (0–60 s) explaining only ∼14% of the FMD% response. More importantly, the correlation between FMD and the eliciting SR_AUC stimulus appears to be group specific. Since young adults showed modest significant correlations between FMD and SR_AUC to 30 s, 60 s, or the time-to-peak diameter, no correlations were reported between any of these measures of SR and FMD in either children or older subjects. The correlation between the brachial artery FMD and the SR stimulus, therefore, appears to be age dependent. These observations indicate that the SR_AUC normalization approaches explain only a relatively small portion of the magnitude of the FMD response and that any relationships present are dependent on the groups studied. This finding has important implications for future studies using the FMD, where a consideration is given to normalization procedures for comparisons between individuals or groups.

In an attempt to improve the validity of FMD testing and methodology, Pyke and Tschakovsky performed an elegant series of experiments to gain better insight into the nature of the eliciting shear stress stimulus for the FMD response. They found that the SR_AUC stimulus, rather than the peak shear, is the critical determinant of the FMD response (29). Our data in a large group of young, healthy adults reinforce these findings since we demonstrated a significant correlation between FMD% and the SR_AUC using a predetermined time window (0–30 or 0–60 s) or up to the peak diameter, but not using the peak SR_AUC. The correlations found between FMD% and the three distinct methods to calculate the SR_AUC stimulus did not importantly differ. This indicates that extending the time window for the calculation of the SR_AUC after 30 s postdeflation does not significantly improve the correlation between the magnitude of the FMD and the eliciting SR stimulus in our dataset.

A recent influential study suggested, based on of the finding that the SR_AUC stimulus explained 56% of the variation in the FMD response, that the FMD should be corrected for this SR_AUC stimulus (29). In our study, however, the SR_AUC stimulus predicts only 9% of the magnitude of the FMD%, and other methods were even more modestly predictive (e.g., other SR_AUC stimuli and FMD_mm). A possible explanation for these contrasting findings is that we examined a heterogeneous age group (9–66 yr), whereas most previous studies examined young subjects. A subgroup analysis on our data demonstrated enhanced correlations between the magnitude of the FMD response and the SR_AUC stimuli in healthy, young subjects (Fig. 2). However, our correlations are still markedly lower than reported previously (r = 0.74; see Ref. 29). A crucial difference between both studies is that we used a between-subject design in 57 subjects, whereas Pyke and Tschakovsky performed a repeated-measures design of six repeated measurements within 10 subjects. The latter approach is likely to overestimate the correlations typical of between-subject comparisons. We suggest that, if the FMD response is primarily, or even largely, dependent on the postdeflation eliciting shear evident between subjects, then it might be expected that the variance in FMD responses between subjects would be more than ∼15% dependent on measured indexes of shear.

To explore which variables, other than SR_AUC stimuli, might contribute to the magnitude of the FMD response, we performed a stepwise linear regression analysis. Interestingly, this analysis revealed that the baseline diameter provided the most powerful prediction of the FMD% response (r = 0.53, r² = 0.28, P < 0.001). This finding reinforces previous studies that identified artery diameter as a strong predictor for the FMD (13, 17, 28, 29, 33). Some studies have attributed this relationship to the impact of diameter on postdeflation shear. On the basis of a modest relationship between SR_AUC and FMD, the impact of baseline diameter is unlikely to be fully explained through its impact on the SR_AUC stimulus alone (35). Possibly, inherent differences in the wall architecture of different sized arteries may partly explain the relationship between size and functional responsiveness, as originally proposed by Folkow et al. (9, 10). Indeed, in a recent study, we observed similar relationships between endothelium-independent vasodilation and baseline artery size as those present between artery size and FMD, suggesting an important role for the structural aspects of the vessel wall determining the magnitude of FMD (35).

To our knowledge, this is the first study to examine the correlation between indexes of postdeflation shear (SR_AUC) and FMD in populations in whom, a priori, one might expect different levels of endothelial function. Indeed, we observed significant differences between FMD and corrected FMD between the groups we studied. However, we found no correlation between the FMD and any of the SR_AUC stimuli in either children or older subjects, which indicates that the relationship between FMD and the eliciting shear stimulus differs between groups of differing age. This raises the possibility that many typically compared groups, such as those with and without cardiovascular disease or risk factors, may also have a different relationship between the FMD response and the SR_AUC stimulus. More importantly, the different relationships between FMD and SR_AUC stimuli between groups indicate that FMD normalization for the SR_AUC stimulus could lead to a misleading reassurance that the normalization of artery function has been conferred for all subjects to a similar degree. Although postdeflation shear is clearly an important stimulus for artery dilation during the FMD (17, 28, 29, 32), other factors likely contribute to the magnitude of this response. We propose that FMD normalizing for SR, a procedure involving multiplication of error in the process of dividing of a ratio (i.e., FMD) by yet another measure (SR_AUC), may lead to less precision in vascular function studies than simply providing a comparison between subjects or groups of SR_AUC. We therefore suggest that future studies present SR_AUC stimulus data, but we question the validity of normalization of FMD measures for SR_AUC. Since the dependence of FMD on SR_AUC is modest and the relationship between FMD and SR as the eliciting dilator stimulus differs between groups, a normalization of FMD may even lead to misleading results when comparing groups.

We can only speculate as to the reasons for differences between age groups in the strength of the relationship between FMD% and SR_AUC stimuli. One possible reason relates to age-related differences in vascular wall characteristics, for example elasticity (19), wall thickness (6), and baseline diameter (7). These characteristics of the vasculature are suggested to influence the responsiveness of arteries to physiological stimuli (9, 10) and may, therefore, contribute to differences in the relationship between FMD and SR_AUC stimuli in different age groups. Another reason relates to age-dependent differences in the sensitivity of the vasodilator [e.g., nitric oxide (31)] and vasoconstrictor pathways [e.g., angiotensin-II and endothelin-1 (37)]. Age-dependent differences in response to vasoactive substances likely contribute to the differences in the relationship between FMD and SR_AUC stimuli. Future studies should further elucidate the potential mechanisms that explain the differences in the strength of the relationship between FMD% and SR_AUC stimuli in age groups but also focus on other groups that differ regarding this relationship (e.g., sex difference).

In conclusion, our data support previous studies that suggested that the SR_AUC stimulus, rather than the peak SR postdeflation, correlates with the magnitude of the FMD%, whereas we found no differences in the correlations between FMD and three typically used methods for SR_AUC stimulus calculation. However, in contrast to the current, widely held view, we found the eliciting SR_AUC stimulus explaining only 10–15% of the FMD response. This reinforces the suggestion that, besides individual differences in the SR_AUC stimulus, other shear-independent factors contribute to individual differences in the magnitude of FMD responses. Moreover, the correlation between the SR_AUC stimulus and the FMD response differs between age groups, with no association between SR_AUC and FMD evident in either children or older adults. While we strongly support measurement and presentation of the eliciting SR_AUC stimulus in future studies comparing FMD between groups, our results suggest that further studies will be required before the practice of normalizing FMD for differences in the SR_AUC is adopted. This procedure introduces error and may even lead to a misinterpretation of data, especially when analyzing FMD responses between groups that have a different relationship between the FMD and SR_AUC stimulus.

GRANTS

This study was supported by The Netherlands Organization for Scientific Research NWO-Grant 82507010 (to D. H. J. Thijssen) and British Heart Foundation Grant FS/05/117/19971 (to M. A. van Bemmel).