Total brachial artery reactivity and first time incident coronary heart disease events in a longitudinal cohort study: The multi-ethnic study of atherosclerosis
Total brachial artery reactivity and first time incident coronary heart disease events in a longitudinal cohort study: The multi-ethnic study of atherosclerosis
Joseph F. PolakID 0 1
Pamela Ouyang 1
Dhananjay Vaidya 1
0 Ultrasound Reading Center, Department of Radiology, Tufts Medical Center , Boston, MA , United States of America, 2 Department of Medicine, Johns Hopkins University School of Medicine , Baltimore, MD , United States of America
1 Editor: Giacomo Pucci, University of Perugia , ITALY
Brachial artery reactivity (BAR) is usually determined as the maximum brachial artery diameter (BAD) following release of an occluding pressure cuff compared to a BAD before cuff inflation. BAD early after cuff deflation can also serve as baseline for estimating total brachial artery reactivity (TBAR). We investigate whether TBAR is associated with first time coronary heart disease events.
Data Availability Statement: The authors confirm
that the following Data Availability statement is
complete, correct, and acceptable to be published
alongside our manuscript. The original brachial
reactivity data has been integrated with the MESA
data already on BioLINCC (https://biolincc.nhlbi.
nih.gov/studies/mesa/), the NHLBI data repository.
Participants of the Multi-Ethnic Study of Atherosclerosis (n = 5499) consisting of whites,
African-Americans, Chinese and Hispanics were followed longitudinally for a mean of 12.5
years. Brachial artery ultrasound was performed following five minutes of cuff occlusion at
the forearm. TBAR was estimated from BAD following cuff release as the difference
between maximum and minimum brachial artery diameters divided by the minimum
diameter multiplied by 100%. TBAR was added to multivariable Cox proportional hazards models
with Framingham risk factors as predictors and time to first coronary heart disease event as
Average TBAR was 9.7% (9.7 SD). Mean age was 61.7 years, 50.9% women. Increased
TBAR was associated with lower risk of CHD events with a hazard rate of 0.78 per SD
increase (95% C.I. 0.67, 0.91; p = 0.001). A TBAR below the median of 7.87% (Inter Quartile
Range: 4.16%, 13.0%) was associated with a 31% lower risk of coronary heart disease
event (Hazard Ratio: 0.69; 95% C.I.: 0.55, 0.87).
/NHLBI (R01 HL069003 and R01 HL081352 to
JFP). The funders had no role in study design, data
collection and analysis, decision to publish, or
preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
TBAR is an independent predictor of first time coronary heart disease events and is
exclusively measured after release of a blood pressure occlusion cuff.
The brachial artery dilates in response to the endogenous release of nitric oxide (NO) that
occurs during reactive hyperemia [
]. Brachial artery flow-mediated dilation (FMD), also
called brachial artery reactivity (BAR), is typically seen on ultrasound imaging following the
release of an occlusive blood pressure cuff that has been kept inflated for five minutes in order
to induce forearm ischemia. An increase in brachial artery diameter is a marker of an ?healthy
endothelium? while lessened degrees of diameter increase are associated with increased risk of
cardiovascular outcomes [
]. The calculation of BAR typically relies on obtaining a baseline
diameter before cuff inflation [
] and is ideally done with the aid of a stereotactic device that
stabilizes the location of the ultrasound imaging probe over the brachial artery [
While most investigators use baseline brachial artery diameters before cuff inflation for
BAR calculations, an alternate approach is to use a baseline brachial artery diameter following
cuff deflation [
]. This approach can help compensate for the absence of a stereotactic device
and lessen the possibility of inadvertent probe displacement while the occlusion cuff is inflated
or when it is deflated.
The brachial artery responds to vasodilator stimuli in a fashion similar to the coronary
]. As such, it is considered a surrogate for the effects of atherosclerosis on the coronary
artery. This hypothesis is supported by outcomes studies linking traditional FMD
measurements with coronary heart disease events [
We propose to measure total brachial artery reactivity (TBAR) as the difference between
the minimum and maximum brachial artery diameters following release of the occlusion cuff
divided by the minimum diameter. No paper has investigated whether this response is
associated with incident coronary heart disease (CHD) events in individuals free of CHD.
We hypothesize that TBAR is independently associated with first time coronary heart
disease events in the Multi-Ethnic Study of Atherosclerosis (MESA), a longitudinal follow-up
study of a multi-ethnic cohort of individuals free of CHD at baseline.
Materials and methods
MESA is a multiethnic population of 6814 men and women aged 45?84 with no history of
clinical cardiovascular disease [
] recruited between July 2000 and August 2002. MESA is a
cohort that includes white, African-American, Hispanic-American, and Chinese participants
from six separate sites in the United States. Participants were excluded from enrollment if they
had a physician diagnosis of heart attack, stroke, transient ischemic attack, heart failure,
angina, atrial fibrillation or a history of any cardiovascular procedure, a weight above 300 lbs,
pregnancy, or any medical conditions that would prevent long-term participation. MESA
protocols and all studies described herein have been approved by the Institutional Review Boards
of all collaborating institutions in the United States: Columbia University, New York NY;
Johns Hopkins University, Baltimore MD; Northwestern University, Chicago IL; University of
California at Los Angeles CA; University on Minnesota, Twin Cities MN; and Wake Forest
2 / 20
Includes 765 Curves
with Response set to 0%
4734 Curves with
Maximum and Minimum
Without a Clear Cut Minimum
in the Early Part of the Post-Release
Fig 1. Study cohort composition. Various physiological and technical issues lead to the exclusion of 578 individuals as listed in the text. In addition, 144
participants had incomplete risk factor profiles, 114 failed to show up at the examination site. Digital copies of brachial artery diameter curves were not
available in 623 cases. As discussed later, 5499 processed curves were analyzed. Of these, 4734 had the expected time course of an increase in diameter at least
30 seconds following cuff release preceded by a minimum diameter. This pattern was not seen on the remaining 765 curves where a maximum diameter was
not seen starting 20 seconds after cuff release.
University, Winston-Salem NC. The participants gave informed consent and underwent an
evaluation of brachial artery endothelial function at the baseline visit.
The population studied was further restricted to individuals with available brachial artery
diameter tracings, complete Framingham risk factor evaluations (n = 5499; Fig 1). The 578
cases with specific exclusions included 140 individuals with systolic blood pressures of 180
mmHg or above, a blood pressure difference between both arms > 15 mmHg (n = 151), 97
3 / 20
with prior mastectomy, various medical issues with the right hand or arm (n = 31), Raynaud?s
phenomenon in 54 individuals, patient related factors (n = 64) and 41 instances of technical
difficulties. The identification of minimum and maximum diameters within defined time
windows following the release of an occlusion cuff was possible in 4734 individuals (Fig 1).
Risk factors and anthropomorphic variables
The risk factors used in this paper are derived from the original CHD Framingham Risk Score
]: age, smoking and diabetes status, systolic blood pressure, LDL and HDL cholesterol with
sex and race/ethnicity added.
Age, sex, race/ethnicity, and medical history were self-reported. Current smoking was
defined as self-report of a cigarette in the last 30 days. Resting blood pressure (BP) was
measured in the seated position using a Dinamap model Pro 100 automated oscillometric
sphygmomanometer (Critikon, Tampa, Florida); pressures were the average of the last two of three
performed measurements. Lipid levels were measured after a twelve-hour fast. Total cholesterol
was measured using a cholesterol oxidase method (Roche Diagnostics), as was HDL cholesterol
following precipitation of non HDL-cholesterol with magnesium/dextran. Triglycerides were
measured with Triglyceride GB reagent (Roche Diagnostics) and LDL cholesterol estimated
]. The presence of diabetes mellitus was based on self-reported physician diagnosis, use of
insulin and/or oral hypoglycemic agent, or a fasting glucose value 126 mg/dL [
Total brachial artery reactivity measurements
Trained technicians at each of the six field centers acquired B-mode ultrasound images with a
Logiq-700 ultrasound device (GE Healthcare, Waukesha, WI) and an ultrasound transducer
(M12L) set at 9 MHz. The sonographers performed the study by holding the transducer and
did not have access to a stereotactic holder. The ultrasound probe was placed on the medial
aspect of the right arm a few centimeters above the elbow with a slight angulation in order to
best visualize the brachial artery. An occlusion cuff was placed on the upper right forearm. A
blood pressure cuff was inflated to a pressure 50 mm Hg above maximal systolic pressure. The
cuff was kept inflated for 5 minutes with the ultrasound probe held centered over the same
brachial artery segment. Images were videotaped starting 15 seconds before cuff deflation and
continuing for 90 seconds after cuff release. Videotape recordings were made using super VHS
The acquired images were sent to Tufts Medical Center Ultrasound Reading Center for
blinded processing. Digital streams of the brachial artery ultrasound images were acquired
from the videotapes at a frame-rate of 30 frames-per-second as MJPEG compressed images
(compression ratio six to one) using a Pinnacle DC-30 Video board (Corel Inc., Mountain
View CA) and a Compaq AP-200 workstation (Compaq Computer Corporation, Houston,
TX) equipped with a Pentium III processor (Intel Corporation, Santa Clara, CA). A reader
reviewed the images and identified the point at which the blood pressure cuff had been
released. The reader then identified an appropriate brachial artery segment and placed a
rectangular region-of-interest on a selected image frame. Customized software was used to
calculate the location of the near and far wall media-adventitia interfaces in this region-of-interest
and to generate brachial artery diameter versus time curves without manual editing. These
were transferred to Access (Microsoft, Redmond WA) databases for archiving. The archived
brachial artery diameter curves were subsequently retrieved and processed using a MATLAB
(The MathWorks Inc., Natick MA) program that smoothed the diameter versus time curves
using a finite impulse response digital filter and processed the resultant curves to identify the
maximum diameter and time to maximum diameter starting 20 seconds following release of
4 / 20
the blood pressure cuff. The algorithm then searched for a minimum diameter going
backwards until 10 seconds after cuff release (Fig 2A?2C) based on previous observations [
Responses were reported as (maximum diameter?minimum diameter) / (minimum diameter)
multiplied by 100% or total brachial artery reactivity (TBAR). Although we processed 5499
brachial diameter curves, there were 765 instances where the curve analysis algorithm detected
a maximum diameter in a time window 20 to 30 seconds after cuff release and failed to find a
smaller diameter in the preceding time interval (Fig 2D). These cases likely represented low
amplitude responses. We therefore assigned them a zero TBAR value and included them in the
Reproducibility was assessed by blinded review of replicate studies performed on the same
participant and acquired on the same day (n = 88). Because of the blinded design, the same
reader performed the diameter extractions on participants with ?dummy? identification
numbers. The correlation coefficient for determining the maximum brachial artery diameter was
0.90 (95% CI: 0.85, 0.94). Replicate measurements of TBAR had a moderate correlation
coefficient of 0.50 (95% CI: 0.33, 0.64). The mean TBAR at the first acquisition was 11.2% and 10.9%
for the second acquisition for a non-significant difference of 0.2% (95% CI: -1.2%, 0.7%).
Brachial artery baseline diameters and then peak-diameters following release of the occlusion
cuff were measured using three approaches: 1) the current software, named Funky Work
Station (FWS), was used to extract the diameter curves that were subsequently processed to
generate TBAR, 2) Brachial Analyzer, a commercial widely distributed software tool (Vascular
Research Tools 5; Medical Imaging Application, LLC, Coralville, IA, USA) and, 3) the software
used to measure FMD in the Multi-Ethnic Study of Atherosclerosis (MESA) and the
Cardiovascular Health Study (CHS)[
]. Baseline (BASE) BAD measurements and peak-response
(PR) BAD were used to calculate FMD according to the equation: FMD = ((PR BAD?BASE
BAD)/ BASE BAD)) 100.
Studies performed on 90 participants were used. Mean age was 61.8 years (10.8 SD).
Fiftysix percent were women and the race/ethnicity breakdown was 40% non-Hispanic whites,
11.1% Chinese-American, 30% African-American and 18.9% Hispanic American. The same
operator used the FWS software and Brachial Analyzer, while one different reader used the
MESA FMD software. In all three instances, the videotaped image sequences were digitized
and then processed by the respective software tools.
To further test the predictive value of TBAR against the more traditional method of
estimating FMD, we obtained the original MESA FMD dataset, hereon referred to as ?classic? FMD.
Of the 3501 entries in this dataset, there were 3162 individuals also having TBAR
Coronary heart disease events
Events were identified during follow-up examinations and by telephone interview conducted
every 9 to 12 months to inquire about all interim hospital admissions, cardiovascular
outpatient diagnoses, and deaths. Copies were obtained of all death certificates and of all medical
records for hospitalizations and outpatient cardiovascular diagnoses. Two physicians from the
MESA study events committee independently reviewed all medical records for end-point
classification and assignment of incidence dates. Coronary heart disease (CHD) events included
myocardial infarction, resuscitated cardiac arrest, and death secondary to coronary heart
disease. A total of 16 suspected CHD events occurred in close proximity to the participant
enrollment date. These participants were excluded from the analyses.
5 / 20
Fig 2. Total brachial artery reactivity (TBAR) after the release of an occlusion cuff inflated for 5 minutes. The rapidly varying diameter values are raw
diameter measurements (green) while the solid smooth lines (red) are the results of curve smoothing. Maximum and minimal brachial artery diameters were
made from the smoothed curves. TBAR is the difference between maximum and minimum diameters divided by the minimum diameter and then multiplied by
100%. Visual review of some of the studies showed subjective differences in the responses: (A) a strong decrease in brachial artery diameter followed by a strong
maximum; (B) a small decrease in diameter followed by a strong increase; and (C) a marked decrease in diameter followed by a small increase. Equivocal results
(D) in 765 instances (765/5499: 13.9%) were due to a failure to detect a minimum diameter in the 20 seconds following cuff release (dotted line) since at least one
diameter value within this time window (arrowhead) was greater than subsequent diameters.
6 / 20
Variables are presented as means and standard deviation (SD) values if continuous and as
percentages if categorical. Median values are shown with inter-quartile ranges (IQR).
The baseline multivariable Cox proportional hazards regression model used robust error
handling to take account of outliers. The model was created with the components of the
traditional Framingham risk score for coronary artery disease: age, systolic blood pressure, diabetes,
HDL-cholesterol, LDL-cholesterol and smoking history. We added sex and race/ethnicity.
TBAR measurements were added to the baseline model and the hazards ratio for TBAR
obtained. We reported the results for continuous variables by their respective standard
deviation values. Kaplan-Meier survival rates were plotted by quartiles of TBAR values.
We assigned a TBAR value of 0.0% to the participants with low amplitude responses (Fig
2D) for a full analytical sample of 5499 participants and then excluded 16 cases with outcomes
near enrolment. We also performed analyses in the subset of 4734 participants (excluding 15
participants with early outcomes for a total of 4719) whose curves showed a clear minimum
and maximum diameter response.
We looked at possible outliers by setting two plausible boundary values i.e., at 0% if below
zero and 40.93% if above the 99th percentile. We repeated the analysis after excluding the
We separately compared the measurements made with FWS, the current software, with
Brachial Tools and the MESA FMD software. We also compared the results of Brachial
Analyzer with those of the MESA FMD software. In all three instances, the intra-class correlation
coefficient (ICC) was calculated and the 95% confidence intervals generated using a mixed
model, i.e., the software tool was considered a ?fixed? effect. Since the purpose was to compare
to overall reproducibility of obtaining brachial artery diameters, we reported the ICC for the
mean response. We further set to zero all negative responses seen with FWS and Brachial
Analyzer since this was the convention used for MESA FMD.
Finally, we compared ?traditional? FMD to TBAR with two Cox proportional hazards
models. In the first we used TBAR (per 1 SD value of 10.5%) as predictor and time to CHD event as
outcome. In the second model we used ?classic? FMD (per 1 standard deviation value of 8.7%)
as the predictor. In both cases we report the hazard ratios, calculated the C-statistics and
compared them. We repeated these analyses in minimally adjusted models: age, sex, and
Analyses were performed using Stata version 11.2 (Stata Corporation, College Station, TX).
Level of significance was two-sided at p < 0.05.
Total brachial artery reactivity
Average participant age was 62.0 years (10.2 years SD) with 50.9% women (Table 1). The mean
cohort follow-up was 12.5 years. The ethnic composition of the cohort was white (36.4%),
Chinese, (13.4%) African-American (27.8%) and Hispanic (22.4%). The distribution of
Framingham risk factors is shown in Table 1. Mean TBAR was 9.7% (? 9.7% SD). The maximal
brachial artery diameter was 4.80 mm. Because of missing minimal diameter values, the other
brachial artery reactivity parameters will be shown in Table 2. A total of 328 first time coronary
artery disease events occurred during follow-up.
The distribution of risk factors for the 4719 participants who had clearly identified maximal
and minimal diameters following release of the occlusion cuff is shown in Table 2. Overall
patient demographics were similar to those of the full cohort (Table 1).
7 / 20
Average maximum diameter at peak-artery dilation was 4.90 mm (0.87 mm SD) and the
early minimum diameter was 4.33 mm (0.86 mm SD). Average time taken to attain maximum
artery diameter was 55.2 (?14.6 SD) seconds and 19.4 sec ? 8.9 SD) seconds for minimum
diameter. Mean TBAR was 11.3% (? 9.6% SD). The number of events decreased to 275 for a
similar follow-up interval.
The results of the multivariable model are shown in Table 3. All traditional Framingham
risk factors were significantly associated with incident coronary heart disease with the
exception of LDL cholesterol. TBAR was a significant independent predictor of lower risk of CHD
when added to the Framingham risk factors with a hazard ratio of 0.78 for each 9.6% increase
(p = 0.003). This corresponds to a slightly larger than 2% decrease in risk for each percent
increase in TBAR.
Similar results were obtained when we repeated the analyses using the median TBAR as a
marker (Table 4). The risk of CHD events in the full cohort decreased 31% when participants
with a TBAR greater than 7.87% were compared to those with values below the median.
We repeated the analysis restricting the study population to the participants with clearly
defined minima and maxima on their brachial artery diameter curves (Table 5). Again, all
traditional Framingham risk factors were significantly associated with incident coronary heart
disease with the exception of LDL cholesterol. TBAR remained a significant independent
predictor of lower risk of CHD when added to the Framingham risk factors with a hazard ratio of
0.73 for each 9.6% increase (p = 0.003). This corresponds approximately to a 3% decrease in
risk for each percent increase in TBAR.
8 / 20
Repeating the analysis with median TBAR as a cut-point showed that total brachial artery
reactivity remained significantly associated with events, hazard ratio of 0.66 (95% CI: 0.51,
0.85; p = 0.001), when participants with a TBAR above the median were compared to those
below (Table 6). This corresponds to a decreased risk of 34% for individuals with a TBAR
above the median of 9.04%.
The Kaplan-Meier curves for TBAR quartiles are shown in Fig 3 for all 5483 participants.
There is decreased risk of a CHD event with time in individuals with the highest two quartiles
while the two lowest quartiles show poor discrimination. The plotted quartiles were -5.18 to
4.16%, 4.16 to 7.78%, 7.87 to 13.02%, and 13.02 to 190.22%
A sensitivity analysis was performed accounting for plausible outliers. The one negative TBAR
value and the 59 values above the 99th percentile (40.93%) were excluded. The Cox
proportional hazards model was then applied to the data (Table 7). The mean TBAR was now 9.46%
(8.0% SD). The TBAR hazard ratio increased to 0.81, i.e. a decrease risk of 19% for a TBAR but
remained statistically significant at p = 0.002.
The Kaplan-Meier curves for TBAR quartiles are shown in Fig 4 for the 5423 participants
after excluding outliers. The hazard ratio was 0.77 (95% CI: 0.66, 0.90) for TBAR after
9 / 20
Population size is 5483 (5499 discounting 16 events close to enrollment).
? Reported for a change of one standard deviation.
765 participants had low amplitude TBAR responses. These TBAR values were assigned a value of 0.0%. The statistical model is also adjusted for race/ethnicity.
excluding all 60 outliers and remained significant at p = 0.0001 for a mean TBAR of 9.15%
(7.38 SD). As before, there is decreased risk of a CHD event with time in individuals with the
highest two quartiles while the two lowest quartiles show poor discrimination. The TBAR
quartiles were 0.0 to 4.13%, 4.13 to 7.79%, 7.79 to 12.81%, and 12.81 to 40.11%.
The Kaplan-Meier curves for TBAR quartiles are shown in Fig 5 for the 4659 participants
after excluding outliers, instances of low amplitude responses and events noted at enrollment.
As before, there is decreased risk of a CHD event with time in individuals with the highest two
quartiles while the two lowest quartiles now show better scaling for events than on Fig 4. The
TBAR quartiles were 0.0 to 5.75%, 5.76 to 8.95%, 8.95 to 13.9%, and 13.9 to 40.11%.
The base, post-release brachial artery diameters as measured by the three different software
tools, FWS, Brachial Analyzer, and MESA FMD and the calculated FMD values are shown in
Table 8. The brachial artery diameters are larger and the calculated FMD is smaller for FWS
when compared to the two other tools.
Population size is 5483 (5499 discounting 16 events close to enrollment).
? Reported for a change of one standard deviation.
765 participants had low amplitude TBAR responses possibly due to noise or low amplitude responses; the TBAR values were assigned a value of 0.0%.
Statistical model is also adjusted for race/ethnicity.
The ICCs between FWS, Brachial Analyzer and MESA FMD are shown in Table 9. In
summary, the ICCs for diameters were all greater than 0.9 for BASE BAD and PR BAD. However,
while the estimated FMD between FWS FMD and MESA FMD was moderate at 0.76 (95%
confidence intervals: 0.64, 0.84), the ICCs between these two tools and Brachial Analyzer were weaker.
Comparing TBAR to ?classic? FMD in an un-adjusted model, TBAR had a hazard ratio of
0.73 (95% CI: 0.60, 0.90) for a p = 0.002, similar to that of FMD with a hazard ratio of 0.78
(95% CI: 0.87, 0.90), p = 0.001. The respective C-statistics were 0.566 (95% CI: 0.53, 0.602),
p < 0.001 and 0.566 (95% CI: 0.529, 0.603). The differences between both were not significant
(p = 0.99). In the minimally adjusted models, TBAR had a lower hazard ratio (HR 0.83; 95%
CI: 0.69, 0.99) than FMD (HR 0.94; 95% CI: 0.81, 1.10). However TBAR was a borderline
significant predictor (p = 0.049) of events whereas FMD was not (p = 0.44). The C-statistics for
both models were similar and not statistically different (p = 0.35) with the TBAR model having
0.716 (95% CI: 0.683, 0.749) and the FMD model 0.713 (95% CI: 0.680, 0.746).
We have measured total brachial artery reactivity (TBAR) by relying on the brachial artery
diameters obtained following the release of a blood pressure occlusion cuff. We have shown
Population size is 4719. Statistical model is adjusted for race/ethnicity.
? Reported for a change of one standard deviation.
The median value of Total Brachial Artery Reactivity is 9.04% (IQR: 5.8, 14.1).
Number at risk
Quartile 1 1370
Quartile 2 1372
Quartile 3 1373
Quartile 4 1368
Fig 3. Kaplan-Meier curves showing the likelihood of remaining event free with time. The curves for the event free
survival of all individuals are plotted by quartiles, quartile 1 (Q1) being the lowest and Q4 the largest. The plotted
quartiles were -5.18 to 4.16%, 4.16 to 7.78%, 7.87 to 13.02%, and 13.02 to 190.22%.
that this measurement is an independent predictor of future coronary heart disease events in a
population free of cardiovascular disease at baseline after accounting for the Framingham risk
Our measurement process is different from the method typically used in the assessment of
flow mediated brachial artery dilation / brachial artery reactivity (FMD / BAR) since we
obtained the baseline diameters used to calculate TBAR after cuff release and not before cuff
]. TBAR as measured by us includes a previously described early decrease in
artery diameter following release of the occlusion cuff [
]. This diameter decrease is believed
Kaplan?Meier Survival Estimates
Population size is 5423. Statistical model is also adjusted for race/ethnicity.
? Reported for a change of one standard deviation.
Kaplan?Meier Survival Estimates
Fig 4. Kaplan-Meier curves showing the likelihood of remaining event free with time for TBAR values after
excluding 60 outliers. The curves for the event free survival of 5423 individuals after excluding outliers are plotted by
quartiles, quartile 1 (Q1) being the lowest and Q4 the largest. The TBAR quartiles were 0.0 to 4.13%, 4.13 to 7.79%, 7.79
to 12.81%, and 12.81 to 40.11%.
to represent a form of brachial artery vasoconstriction [
]. The early decrease in brachial
artery diameter appears associated with the pressure drop (20 to 24 mmHg) that accompanies
maximal brachial artery blood flow [
]. The mechanisms responsible for this early
vasoconstriction have not yet been fully explained but its existence was noted by Dobrosielski et al. [
and recently confirmed to have a prevalence of above 60% in children and young adults [
]. Because of our experimental design and the lack of a stereotactic stabilizer, we are unable
to identify individuals with a true vasoconstrictor response since we lack an appropriate
We let our algorithm find the point of maximum diameter and the time to maximal
diameter rather than perform a measurement at 60 seconds [
] based on observations [
1, 4, 20
the peak response does not necessarily occur at 60 seconds and found, on average, a
peakresponse at 55.2 seconds following cuff release. Variations in the time-to-peak dilation have
been speculated to be secondary to factors linked to aging and level of physical fitness [
Our algorithm did not find a minimum diameter before a maximum in 765 members of our
cohort (13.5% men and 14.3% women) (Fig 2D). We believe that there are two possibilities for
this finding: (1) un-interpretable brachial artery diameter curves due to physiological
vasoconstriction of the brachial artery or, (2) diameter mis-registration due to artifacts and noise.
The first possibility was brought up by a group of investigators who described a negative flow
mediated responses or brachial artery constriction (BAC) [
]. Sedlak et al found that 11% of
their study population exclusively composed of women had such a response [
]. The second
possibility is also likely given the known technical difficulties in obtaining precise brachial
artery diameter measurements. We took into consideration both possibilities by assigning a
TBAR value of 0.0% to these cases with the following logic: (1) if the response represented
13 / 20
Fig 5. Kaplan-Meier curves showing the likelihood of remaining event free with time for TBAR values after
excluding 60 outliers, instances of poorly defined maxima and minima, and early events. The curves for the event
free survival of 4659 individuals after excluding outliers are plotted by quartiles, quartile 1 (Q1) being the lowest and
Q4 the largest. The TBAR quartiles were 0.0 to 5.75%, 5.76 to 8.95%, 8.95 to 13.9%, and 13.9 to 40.11%.
vasoconstriction, we conservatively biased the measurement to the null and, (2) if the error
due to noise overwhelmed the TBAR, then the value of TBAR was likely low and close to zero.
Flow mediated dilation measured with respect to a diameter obtained before cuff inflation
has been shown to be associated with CHD events [
2, 17, 23, 24
]. We add to these observations
by reporting associations between CHD events and total brachial artery reactivity, a
measurement made solely following the time when the occlusion cuff is deflated. Brachial artery
reactivity calculated using a baseline diameter measured after cuff deflation might therefore be
equivalent to the response seen with a baseline diameter measurement made before cuff
inflation . It has the advantage of limiting the errors associated with ultrasound probe
displacement when a stereotactic device is not available.
Our study is limited in the following ways. It has technical limitations, may possess a
participant selection bias, and is a multi-center study. We believe that the major limitations of our
study are technical, being related to the acquisition protocol and the equipment used at the
time of data analysis. All acquisitions were performed without the help of a stereotactic holder
5.6 (? 5.9)
3.6 (? 3.3)
14 / 20
to help stabilize the position of the ultrasound transducer during an acquisition lasting up to 7
]. This can cause sonographer fatigue and may therefore be a source of variability.
The automated edge detection process used to determine brachial artery diameters took more
than 10 minutes to generate the brachial diameter curves given the limitations of the computer
hardware available in early 2000 (Pentium III processor; Intel Corporation, Santa Clara, CA).
The length of the measurement process limited the reader?s ability to make adjustments to the
region-of-interest position once processing was started. The process used to generate brachial
artery diameter curves was therefore an automatic one with minimal operator involvement:
one region-of-interest was drawn and the edge detector parameters were adjusted once.
At the time this study was undertaken there was no validated software that would have
permitted the tracking of carotid and brachial artery diameters. The diameter tracking software
was designed in house. Performance of this software was since validated in prior studies
linking carotid diameters to left ventricular mass [
] and incident stroke [
]. Carotid artery
distensibility measurements made with this software have also shown associations with left
ventricular dynamics [
] and aortic wall calcification [
]. However, we also compared the
ability of our software to perform traditional FMD measurements by comparing the results of
measurements made with it to those obtained with Brachial Analyzer and the MESA software
in a subset of 90 participants. Similar to the results of Faita et al., we found strong correlations
between diameter measurements [
]. We found much weaker associations between
techniques for FMD estimates. The strongest correlation was between our software and that used
in MESA (Table 9).
Our TBAR reproducibility studies were performed in a completely blinded fashion. Due to
time constraints and participant burden, the Coordinating Center at the University of
Washington (Seattle, WA, USA) generated a list of participants that would have replicate studies.
This was mostly done near the end of the study. The Ultrasound Reading Center received
these studies with the subject identification numbers having been scrambled. Results of the
analyses were matched after the analyses had been completed. Because of the design, almost all
studies were performed by the same sonographer and read by the same reader. The timing of
the replicate study is not known nor is any interval ingestion of food. As such, part of the
variability might be secondary to these factors that are well recognized as modifying flow mediated
]. This might have contributed to the low correlation between replicate
The lack of cardiac gating might also have blunted associations between TBAR and events.
However, diameter measurements made on images gated to diastole, to systole, or averaged
throughout the cardiac cycle seem to give similar estimates of flow mediated dilation [
We address possible selection bias by listing the source of exclusions in Fig 1. This is either
unexplained since the participant was not seen at the testing station, caused by a specific
medical condition that precludes performance of the test, linked to the lack of availability of a digital
record of the study at the time of processing, and an incomplete set of risk factors.
15 / 20
Given the range of responses reaching up to 190%, we performed one analysis where we
excluded plausible outliers from our analyses (Fig 4). The results remained similar to those of
the original analyses. We also looked at the scenario where we excluded cases with poorly
defined maxima and minima (Fig 5). Comparing the Kaplan-Meier curves shown in Fig 4 to
those on Fig 5, an incremental increase in the likelihood of events between quartiles 1 and 2
was subjectively more apparent. This might indicate that some of the responses that were set at
0.0% represented cases where the technical quality of the brachial artery diameter acquisitions
One of the major limitations of our paper remains the lack of information on the
magnitude of the nitric oxide (NO) release. While most studies in the literature do not capture this
information, concurrent recording of the blood flow velocities (shear rate and shear stress)
] as a surrogate of NO release might improve overall reliability of brachial artery flow
mediated dilation and plausibly that of TBAR. However, even NO release only partly accounts for
the flow mediated response of the brachial artery following release of an occlusion cuff [
Local brachial artery distensibility might also be modulating the brachial artery responses
during both the early decrease in diameter and the flow mediated increase in diameter. Witte
et al. showed that decreased distensibility of the brachial artery is associated with decreased
FMD responses [
]. It is not clear that brachial artery reactivity is a completely different
phenotype than distensibility or is partly linked to it. This has implications for possible linkages
between aging and heritability as modulators of local brachial artery compliance, large artery
stiffening, and flow mediated dilation [
]. For example, if brachial artery distensibility is
linked to flow mediated dilation and distensibility is in part heritable, then flow mediated
dilation may also have a genetic component. However, our experimental design does not permit
us to address these issues directly. In a review of the factors associated with vascular aging,
Paneni et al. point out a very likely linkage between endothelial dysfunction due to age related
decreased nitric oxide production and increased breakdown . A plausible effect of
decreased NO activity on arterial smooth muscle tone would be to decrease distensibility. It is
likely that this would also apply to FMD. Paneni et al. also suggest that certain negative traits
may be passed on through epigenetic mechanisms and contribute to a milieu favoring vascular
]. This might apply to the distensibility of muscular arteries and to flow mediated
Our comparisons between TBAR and FMD should be viewed from a qualitative
perspective. The results, although showing statistical equivalence between both variables, should be
viewed very cautiously. The MESA ?classic? FMD measurements were made using a
case-control design and, as such, a weighing scheme is needed to properly interpret the results. Using
the available data to compare TBAR to FMD without a new weighing strategy limits the
interpretation of the results.
The multi-center nature of our study implicitly introduced variability into the measurement
process linked to the clinic site, number of sonographers, as well as limiting quality assurance
processes due to off-site supervision by a core laboratory. A major limitation of our study
includes imaging at six separate centers and by 20 different sonographers albeit 15 performed
more than 20 studies each. The number of sonographers likely increased variability and
attenuated the precision of our measurements. Such limitations can be overcome in specialized
laboratories that study the brachial artery responses to reactive hyperemia in a systematic fashion
1, 25, 41, 42
]. It is unclear whether this level of expertise can be consistently promulgated to
the clinic. For example, some of the six clinic sites in our study hired trained and certified
vascular sonographers while others relied on individuals with less formal training. However, one
advantage of using total brachial artery reactivity as a measurement is limiting the length of
16 / 20
data acquisition to a short time period following release of the occlusion cuff. This likely
reduced measurement variability. While we recognize that our measurement process has
greater variability than those generated in a specialized laboratory, our measurements of total
brachial artery reactivity made solely in the time period when the occlusion cuff is released
were significantly associated with coronary heart disease events.
Total brachial artery reactivity is a significant independent predictor of first time coronary
heart disease events. It is exclusively measured during the post-release phase of brachial
reactivity studies. Further studies are needed to confirm this finding.
The authors would like to thank the investigators, the staff, and the participants of the MESA
study for their valuable contributions.
A full list of participating MESA investigators and institutions can be found at http://www.
Data curation: Joseph F. Polak.
Formal analysis: Joseph F. Polak.
Funding acquisition: Joseph F. Polak.
Conceptualization: Joseph F. Polak, Pamela Ouyang, Dhananjay Vaidya.
Investigation: Joseph F. Polak, Pamela Ouyang, Dhananjay Vaidya.
Methodology: Joseph F. Polak, Pamela Ouyang, Dhananjay Vaidya.
Project administration: Joseph F. Polak.
Supervision: Joseph F. Polak.
Validation: Joseph F. Polak.
Writing ? original draft: Joseph F. Polak.
Writing ? review & editing: Pamela Ouyang, Dhananjay Vaidya.
17 / 20
18 / 20
19 / 20
1. Thijssen DH , Black MA , Pyke KE , Padilla J , Atkinson G , Harris RA , et al. Assessment of flow-mediated dilation in humans: a methodological and physiological guideline . [Review]. Am J Physiol Heart Circ Physiol . 2011 ; 300 ( 1 ): H2 - 12 . https://doi.org/10.1152/ajpheart.00471. 2010 PMID: 20952670
2. Green DJ , Jones H , Thijssen D , Cable NT , Atkinson G . Flow-mediated dilation and cardiovascular event prediction: does nitric oxide matter? . [Review]. Hypertension . 2011 ; 57 ( 3 ): 363 - 9 . https://doi.org/ 10.1161/HYPERTENSIONAHA.110.167015 PMID: 21263128
3. Corretti MC , Anderson TJ , Benjamin EJ , Celermajer D , Charbonneau F , Creager MA , et al. Guidelines for the ultrasound assessment of endothelial-dependent flow-mediated vasodilation of the brachial artery: a report of the International Brachial Artery Reactivity Task Force . J Am Coll Cardiol . 2002 ; 39 ( 2 ): 257 - 65 . PMID: 11788217
4. Donald AE , Halcox JP , Charakida M , Storry C , Wallace SML , Cole TJ , et al. Methodological approaches to optimize reproducibility and power in clinical studies of flow-mediated dilation . J Am Coll Cardiol . 2008 ; 51 ( 20 ): 1959 - 64 . https://doi.org/10.1016/j.jacc. 2008 . 02 .044 PMID: 18482664
5. Dobrosielski DA , Arce AA , Allen JA , Wood RH , Welsch MA , and the Louisiana Healthy Aging Study . Biphasic responses of the brachial artery diameter following forearm occlusion: a blunted response in the elderly . Dynamic Medicine . 2006 ; 5:4 . https://doi.org/10.1186/ 1476 -5918-5-4 PMID: 16597328
6. Jiang B , Seddon M , Fok H , Donald A , Chowienczyk P . Flow-mediated dilation of the radial artery is offset by flow-induced reduction in transmural pressure . Hypertension . 2011 ; 57 ( 6 ): 1145 - 50 . https://doi. org/10.1161/HYPERTENSIONAHA.110.163113 PMID: 21502570
7. Green DJ , Thijssen DH. De Motu Arteriarum: hemodynamics and arterial function in humans . Hypertension . 2011 ; 57 ( 6 ): 1049 - 50 . https://doi.org/10.1161/HYPERTENSIONAHA.110.168591 PMID: 21502565
8. Ostrem JD , Evanoff NG , Ryder JR , Steinberger J , Sinaiko AR , Bisch KL , et al. High-flow-mediated constriction in adults is not influenced by biomarkers of cardiovascular and metabolic risk . J Clin Ultrasound . 2017 : 45 ( 1 ); 35 - 42 . https://doi.org/10.1002/jcu.22387 PMID: 27492803
9. Anderson TJ , Uehata A , Gerhard MD , Meredith IT , Knab S , Delagrange D , et al. Close relation of endothelial function in the human coronary and peripheral circulations . J Am Coll Cardiol . 1995 ; 26 ( 5 ): 1235 - 41 . PMID: 7594037
10. Matsuzawa Y , Kwon T-G , Lennon RJ , Lerman LO , Lerman A . Prognostic Value of Flow-Mediated Vasodilation in Brachial Artery and Fingertip Artery for Cardiovascular Events: A Systematic Review and Meta-Analysis . Journal of the American Heart Association . 2015 ; 4 ( 11 ): 13 .
11. Bild DE , Bluemke DA , Burke GL , Detrano R , Diez Roux AV , Folsom AR , et al. Multi-ethnic study of atherosclerosis: objectives and design . Am J Epidemiol . 2002 ; 156 ( 9 ): 871 - 81 . PMID: 12397006
12. Wilson PW , D'Agostino RB , Levy D , Belanger AM , Silbershatz H , Kannel WB . Prediction of coronary heart disease using risk factor categories.[see comment] . Circulation . 1998 ; 97 ( 18 ): 1837 - 47 . PMID: 9603539
13. Friedewald WT , Levy RI , Fredrickson DS . Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge . Clin Chem . 1972 ; 18 ( 6 ): 499 - 502 . PMID: 4337382
14. Genuth S , Alberti KGMM , Bennett P , Buse J , Defronzo R , Kahn R , et al. Follow-up report on the diagnosis of diabetes mellitus.[see comment] . Diabetes Care . 2003 ; 26 ( 11 ): 3160 - 7 . PMID: 14578255
15. Fan L , Santago P , Jiang H , Herrington DM . Ultrasound measurement of brachial flow-mediated vasodilator response . IEEE Transactions on Medical Imaging . 2000 ; 19 ( 6 ): 621 - 31 . https://doi.org/10.1109/ 42.870669 PMID: 11026465
16. Yeboah J , Reboussin DM , Waters D , Kowalchuk G , Herrington DM . Effects of estrogen replacement with and without medroxyprogesterone acetate on brachial flow-mediated vasodilator responses in postmenopausal women with coronary artery disease . American Heart Journal . 2007 ; 153 ( 3 ): 439 - 44 . https://doi.org/10.1016/j.ahj. 2006 . 11 .006 PMID: 17307425
17. Yeboah J , Folsom AR , Burke GL , Johnson C , Polak JF , Post W , et al. Predictive value of brachial flowmediated dilation for incident cardiovascular events in a population-based study: the multi-ethnic study of atherosclerosis . Circulation . 2009 ; 120 ( 6 ): 502 - 9 . https://doi.org/10.1161/CIRCULATIONAHA.109. 864801 PMID: 19635967
18. Wallace JM , Stead EA Jr. Fall in pressure in radial artery during reactive hyperemia . Circ Res . 1959 ; 7 : 876 - 9 . PMID: 13842594
19. Ostrem JD , Evanoff N , Kelly AS , Dengel DR . Presence of a high-flow-mediated constriction phenomenon prior to flow-mediated dilation in normal weight, overweight, and obese children and adolescents . J Clin Ultrasound . 2015 ; 43 ( 8 ): 495 - 501 . https://doi.org/10.1002/jcu.22267 PMID: 25801746
20. Liuni A , Luca MC , Lisi M , Dragoni S , di Stolfo G , Mariani JA , et al. Observations of time-based measures of flow-mediated dilation of forearm conduit arteries: implications for the accurate assessment of endothelial function . Am J Physiol Heart Circ Physiol . 2010 ; 299 ( 3 ): H939 - 45 . https://doi.org/10.1152/ ajpheart.00271. 2010 PMID: 20639219
21. Black MA , Cable NT , Thijssen DHJ , Green DJ . Importance of measuring the time course of flow-mediated dilatation in humans . Hypertension . 2008 ; 51 ( 2 ): 203 - 10 . https://doi.org/10.1161/ HYPERTENSIONAHA.107.101014 PMID: 18086954
22. Sedlak TL , Johnson BD , Pepine CJ , Reis SE , Bairey Merz CN . Brachial artery constriction during brachial artery reactivity testing predicts major adverse clinical outcomes in women with suspected myocardial ischemia: results from the NHLBI-sponsored women's ischemia Syndrome Evaluation (WISE) Study . PLoS ONE [Electronic Resource]. 2013 ; 8 ( 9 ): e74585 .
23. Yeboah J , Crouse JR , Hsu F-C , Burke GL , Herrington DM . Brachial flow-mediated dilation predicts incident cardiovascular events in older adults: the Cardiovascular Health Study.[see comment] . Circulation . 2007 ; 115 ( 18 ): 2390 - 7 . https://doi.org/10.1161/CIRCULATIONAHA.106.678276 PMID: 17452608
24. Shimbo D , Grahame-Clarke C , Miyake Y , Rodriguez C , Sciacca R , Di Tullio M , et al. The association between endothelial dysfunction and cardiovascular outcomes in a population-based multi-ethnic cohort . Atherosclerosis . 2007 ; 192 ( 1 ): 197 - 203 . https://doi.org/10.1016/j.atherosclerosis. 2006 . 05 .005 PMID: 16762358
25. Ostrem JD , Dengel DR , Marlatt KL , Steinberger J . Comparison of baseline brachial artery measurements and effect on peak flow-mediated dilation . Clin Physiol Funct Imaging . 2015 ; 35 ( 1 ): 34 - 40 . https://doi.org/10.1111/cpf.12123 PMID: 24438447
26. Charakida M , Masi S , Luscher TF , Kastelein JJP , Deanfield JE . Assessment of atherosclerosis: the role of flow-mediated dilatation . Eur Heart J . 2010 ; 31 ( 23 ): 2854 - 61 . https://doi.org/10.1093/eurheartj/ ehq340 PMID: 20864485
27. Polak JF , Wong Q , Johnson WC , Bluemke DA , Harrington A , O'Leary DH , et al. Associations of cardiovascular risk factors, carotid intima-media thickness and left ventricular mass with inter-adventitial diameters of the common carotid artery: The Multi-Ethnic Study of Atherosclerosis (MESA) . Atherosclerosis . 2011 ; 218 ( 2 ): 344 - 9 . https://doi.org/10.1016/j.atherosclerosis. 2011 . 05 .033 PMID: 21726862
28. Polak JF , Sacco RL , Post WS , Vaidya D , Arnan MK , O'Leary DH . Incident stroke is associated with common carotid artery diameter and not common carotid artery intima-media thickness . Stroke . 2014 ; 45 ( 5 ): 1442 - 6 . https://doi.org/10.1161/STROKEAHA.114.004850 PMID: 24643408
29. Fernandes VRS , Polak JF , Cheng S , Rosen BD , Carvalho B , Nasir K , et al. Arterial stiffness is associated with regional ventricular systolic and diastolic dysfunction: the Multi-Ethnic Study of Atherosclerosis . Arterioscler Thromb Vasc Biol . 2008 ; 28 ( 1 ): 194 - 201 . https://doi.org/10.1161/ATVBAHA.107. 156950 PMID: 17962621
30. Blaha MJ , Budoff MJ , Rivera JJ , Katz R , O'Leary DH , Polak JF , et al. Relationship of carotid distensibility and thoracic aorta calcification: multi-ethnic study of atherosclerosis . Hypertension . 2009 ; 54 ( 6 ): 1408 - 15 . https://doi.org/10.1161/HYPERTENSIONAHA.109.138396 PMID: 19805639
31. Faita F , Masi S , Loukogeorgakis S , Gemignani V , Okorie M , Bianchini E , et al. Comparison of two automatic methods for the assessment of brachial artery flow-mediated dilation . J Hypertens . 2011 ; 29 ( 1 ): 85 - 90 . https://doi.org/10.1097/HJH.0b013e32833fc938 PMID: 20842049
32. Greyling A , van Mil ACCM , Zock PL , Green DJ , Ghiadoni L , Thijssen DH , et al. Adherence to guidelines strongly improves reproducibility of brachial artery flow-mediated dilation . Atherosclerosis . 2016 ; 248 : 196 - 202 . https://doi.org/10.1016/j.atherosclerosis. 2016 . 03 .011 PMID: 27023841
33. Gemignani V , Bianchini E , Faita F , Giannarelli C , Plantinga Y , Ghiadoni L , et al. Ultrasound measurement of the brachial artery flow-mediated dilation without ECG gating . Ultrasound Med Biol . 2008 ; 34 ( 3 ): 385 - 91 . https://doi.org/10.1016/j.ultrasmedbio. 2007 . 08 .006 PMID: 17964069
34. Tinken TM , Thijssen DH , Hopkins N , Black MA , Dawson EA , Minson CT , et al. Impact of shear rate modulation on vascular function in humans . Hypertension . 2009 ; 54 ( 2 ): 278 - 85 . https://doi.org/10.1161/ HYPERTENSIONAHA.109.134361 PMID: 19546374
35. Green DJ , Jones H , Thijssen D , Cable NT , Atkinson G . Flow-mediated dilation and cardiovascular event prediction: does nitric oxide matter? Hypertension . 2011 ; 57 ( 3 ): 363 - 9 . https://doi.org/10.1161/ HYPERTENSIONAHA.110.167015 PMID: 21263128
36. Green DJ , Dawson EA , Groenewoud HMM , Jones H , Thijssen DHJ . Is flow-mediated dilation nitric oxide mediated?: A meta-analysis . Hypertension . 2014 ; 63 ( 2 ): 376 - 82 . https://doi.org/10.1161/ HYPERTENSIONAHA.113. 02044 PMID: 24277765
37. Witte DR , van der Graaf Y , Grobbee DE , Bots ML , Group SS. Measurement of flow-mediated dilatation of the brachial artery is affected by local elastic vessel wall properties in high-risk patients . Atherosclerosis . 2005 ; 182 ( 2 ): 323 - 30 . https://doi.org/10.1016/j.atherosclerosis. 2005 . 02 .012 PMID: 16159605
38. Cecelja M , Jiang B , Bevan L , Frost ML , Spector TD , Chowienczyk PJ . Arterial stiffening relates to arterial calcification but not to noncalcified atheroma in women. A twin study . J Am Coll Cardiol . 2011 ; 57 ( 13 ): 1480 - 6 . https://doi.org/10.1016/j.jacc. 2010 . 09 .079 PMID: 21435518
39. Tarnoki AD , Tarnoki DL , Stazi MA , Medda E , Cotichini R , Nistico L , et al. Heritability of central blood pressure and arterial stiffness: a twin study . J Hypertens . 2012 ; 30 ( 8 ): 1564 - 71 . https://doi.org/10.1097/ HJH.0b013e32835527ae PMID: 22688268
40. Paneni F , Diaz Ca?estro C , Libby P , Lu?scher TF, Camici GG . The Aging Cardiovascular System: Understanding It at the Cellular and Clinical Levels . J Am Coll Cardiol . 2017 ; 69 ( 15 ): 1952 - 67 . https:// doi.org/10.1016/j.jacc. 2017 . 01 .064 PMID: 28408026
41. Thijssen DH , Bullens LM , van Bemmel MM , Dawson EA , Hopkins N , Tinken TM , et al. Does arterial shear explain the magnitude of flow-mediated dilation?: a comparison between young and older humans . Am J Physiol Heart Circ Physiol . 2009 ; 296 ( 1 ): H57 - 64 . https://doi.org/10.1152/ajpheart. 00980. 2008 PMID: 19028795
42. Thijssen DH , Dawson EA , Black MA , Hopman MT , Cable NT , Green DJ . Heterogeneity in conduit artery function in humans: impact of arterial size . Am J Physiol Heart Circ Physiol . 2008 ; 295 ( 5 ): H1927 - 34 . https://doi.org/10.1152/ajpheart.00405. 2008 PMID: 18775852