Endothelial dysfunction assessed by brachial artery
ultrasound in severe sepsis and septic shock
Leandro Becker MD, MSc, Karen Prado MD, MSc, Murilo Foppa MD, ScD,
Nidiane Martinelli BSc, Cynthia Aguiar MD, MSc, Thiago Furian MD, MSc,
Nadine Clausell MD, PhD, Luis Eduardo Rohde MD, ScD⁎
Post-Graduate Program in Health Sciences, Cardiology and Cardiovascular Sciences, Medical School from Rio Grande do
Sul Federal University, Non-Invasive Cardiac Laboratory from the Cardiology Division and Intensive Medicine Division,
Hospital de Clínicas de Porto Alegre
Purpose: Noninvasive evaluation of endothelial function may be accomplished by ultrasound
assessment of flow-mediated vasodilation (FMD) of the brachial artery. This study aims to investigate
the role of FMD analysis on intrahospital prognosis of patients with sepsis.
Methods: Adult patients admitted to the intensive care unit with severe sepsis or septic shock were
consecutively included. Brachial artery FMD was measured upon admission, after 24 and 72 hours. A
group of apparently healthy subjects paired for sex and age was used as controls. Patients were followed
up to discharge or death.
Results: We studied 42 patients (mean age, 51 ± 19 years) with sepsis predominantly of abdominal or
respiratory etiology (75%). Acute Physiology And Chronic Health Evaluation II risk score was 23 ± 7,
and intrahospital mortality rate was 33%. Flow-mediated vasodilation in septic patients was
significantly lower than in healthy controls (1.5 ± 7% vs 6 ± 4%, P b .001). Most of the nonsurvivors
(86%) showed a decline in sequential FMD analyses, whereas only 43% of survivors showed a
reduction of FMD (P = .01). In nonsurvivors, FMD was significantly lower 72 hours after sepsis onset
(−3.3% ± 10% vs 5.2% ± 4%; P b .05; time-group interaction P value = .03).
Conclusions: Brachial FMD is altered in septic patients with hemodynamic instability, and its
deterioration may be an early marker of unfavorable prognosis.
© 2012 Elsevier Inc.
Open access under the Elsevier OA license.
Sepsis is a clinical syndrome caused by severe infection
and is characterized by a systemic inflammatory reaction
with varying degrees of organ dysfunction . Organ failure
has a cumulative effect on sepsis-related mortality, which
may reach 70% in cases with severe hypotension . The
endothelium plays a central role in sepsis physiopathology,
⁎Corresponding author. Hospital de Clínicas de Porto Alegre, Serviço
de Cardiologia, Unidade de Métodos Não Invasivos, Rua Ramiro Barcelos
2350, Porto Alegre – RS – Brasil.
E-mail addresses: firstname.lastname@example.org (L. Becker),
email@example.com (K. Prado), firstname.lastname@example.org (M. Foppa),
email@example.com (N. Martinelli), firstname.lastname@example.org (C. Aguiar),
email@example.com (T. Furian), firstname.lastname@example.org (N. Clausell),
email@example.com (L.E. Rohde).
0883-9441 © 2012 Elsevier Inc.
Journal of Critical Care (2012) 27, 316.e9–316.e14
Open access under the Elsevier OA license.
producing several biologic mediators of vasomotor function
and balancing the release of nitric oxide and endothelin into
the bloodstream .
Endothelial dysfunction can be indirectly assessed by
detection of various inflammatory markers released by the
vascular endothelium, measured in plasma or serum .
Noninvasive assessment of endothelial vasomotor function
can also be achieved by ultrasound measurement of
peripheral artery diameters, exploring flow-mediated vaso-
dilation (FMD)—an indicator of endothelial nitric oxide
bioavailability . In humans, this technique has been
extensively validated in several cardiovascular scenarios,
particularly in patients with risk factors or proven athero-
sclerosis [6-14]. Application of ultrasound-based techniques
for endothelial evaluation in septic patients has not been fully
explored so far, possibly because this environment is marked
by intense vasomotor fluctuations and frequent use of
vasoconstrictors. Vaudo et al  have recently published
a pioneer study that described the presence of significant
brachial FMD alterations in early stages of sepsis. In septic
patients with endothelial dysfunction, a reduction in the
number of leukocytes and high sequential organ failure
scores were also observed.
In the present prospective study, we evaluated if
ultrasound-based FMD is feasible in severe sepsis and
septic shock and might offer prognostic information in
2.1. Patients and study design
The study included 42 consecutive adult patients admitted
to the intensive care unit at Hospital de Clínicas de Porto
Alegre within 24 hours of diagnosis of severe sepsis or septic
shock according to internationally accepted consensus
definitions . Sepsis was defined based on clinical evidence
of infection and 2 or more of the following: (1) fever axial
temperature greater than 38°C or hypothermia (axial
temperature b36°C), (2) tachycardia (heart rate N90 beats
per minute), (3) tachypnea (N20 breaths per minute) or need
for mechanical ventilation, (4) leukocytosis (N12 000
cells/mm³) or leukopenia (b4000 cells/mm³), or a ratio of
greater than 10% band cells to polymorphonuclear cells.
Severe sepsis was defined as clinical signs of sepsis
associated with organ dysfunction, alterations in perfusion,
or hypotension. Septic shock was defined as sepsis with
hypotension even after initial volume expansion. Exclusion
criteria were as follows: (1) age older than 80 years, (2) heart
failure, (3) liver failure (Child-Pugh class C), (4) bone
marrow failure (leukocytes b500/μL) or (5) immunosup-
pression (acquired human immunodeficiency; use of immu-
nosuppressants, including corticosteroids [prednisoneN5
mg/kg/day]; or cancer), and (6) infective endocarditis.
Assessment of endothelial-dependent FMD of the bra-
chial artery was carried out in a convenience sample of
apparently healthy control subjects (n = 38), matched by age
and sex. Most control subjects were health care pro-
fessionals, with no history of cardiovascular disease and no
risk factors for atherosclerotic disease.
2.3. Study protocol
Clinical features and laboratory data of patients were
collected at admission, including Acute Physiology And
Chronic Health Evaluation II score , hemoglobin and
leukocytes, C-reactive protein, and serum lactate levels.
Brachial artery ultrasound to determine FMD was also
carried out at admission. These parameters were reassessed
in survivors after 24 hours and 72 hours. Informed consent
was signed by all patients or guardians before inclusion in
the study. The study was approved by the institution's
research ethics committee.
2.4. Flow-mediated vasodilation evaluation
Flow-mediated vasodilation was assessed in the brachial
artery. Measurements were obtained with the patient in the
supine position, in an arm without venous or arterial lines.
Brachial artery images were obtained using a high-frequency
transducer (7.5-10 MHz) and a commercially available
ultrasound system (Philips EnVisor; Philips, Andover,
Mass). Images were obtained simultaneously with electrocar-
diographic tracing and digitally recorded. To minimize
operational errors, both the transducer and the arm were
maintained in the same position during the entire procedure.
Baseline images were recorded, and brachial artery poster-
oanterior diameter was measured in diastole, in 3 adjacent
segments, at the best angle of interrogation to determine the
intima-media thickness. This procedure was repeated for 3
consecutive beats. A pressure cuff was then placed on the
forearm and inflated to 230 to 250 mm Hg for 5 minutes.
Brachial artery diameter was measured again 45 to 60 seconds
after sudden cuff deflation, following the study protocol .
The mean of 9 measurements of baseline and posthyperemia
diameters was used for statistical analysis. Flow-mediated
vasodilation was expressed as the relative change in brachial
artery diameter during hyperemia and defined as 100 ×
([posthyperemia diameter − baseline diameter]/baseline
diameter). Therefore, positive percentage values indicate
vasodilation, whereas negative percentage values indicate
constriction. Brachial artery diameter measurements were
performed off-line, but immediately after the protocols were
finished, on a day-by-day basis. As such, the operator was
always blinded to the final clinical outcome. Intraobserver
variability was measured by the same investigator (LB) in 9
subjects, with excellent reproducibility between examinations
316.e10L. Becker et al.
(mean variation of baseline diameter of 0.3 mm and mean
FMD variation of 0.63%). These values are substantially
below the differences observed between survivors and
nonsurvivors and are in accordance with the guidelines from
the International Brachial Artery Reactivity Task Force .
2.5. Blood-derived markers
Quantitative determination of human endothelin 1 (ET-1)
in extracted plasma samples from septic patients was
performed using commercially available enzyme-linked
immunosorbent assay (R&D Systems, Minneapolis, Minn,),
according to the manufacturer's recommendations. The limit
plasma samples underwent solvent extraction (acetone:1 N
HCl: water [40:1:5]), were lyophilized in a centrifugal
evaporator, and reconstituted into 0.25 mL of sample diluent.
Interleukin 6 (IL-6) and soluble vascular cell adhesion
molecule 1 (sVCAM-1) concentrations were also quantified
in plasma samples using commercially available enzyme-
linked immunosorbent assay (R&D Systems), according to
the manufacturer's recommendations. Interleukin 6 detect-
able levels ranged from 3 to 300 pg/mL and VCAM-1
detectable levels ranged from 6.25 to 200 pg/mL.
2.6. Statistical analysis
Quantitative variables are presented as mean ± standard
deviation or median and interquartile range; categoric
variables are expressed as absolute numbers and percent-
ages. Variables without normal distribution underwent
logarithmic transformation. Student t test, χ2, or Fisher
exact test was used for comparisons between the groups, as
appropriate. To study the behavior of brachial artery FMD
over time, 2-way analysis of variance for repeated measures
were used using general linear models (SAS Software 8.0;
SAS, Cary, NC). Two-tailed P b .05 was considered as
3.1. Patients' clinical characteristics
We enrolled 42 septic patients (62% women; mean
age, 51 ± 19 years; 75% of abdominal and respiratory
etiology). Mean length of stay in the intensive care unit
was 8 days (median, 6 days; 3-13 days). In 79% of
cases, vasoactive drug therapy was required. During
hospitalization, 14 (33%) patients died because of sepsis.
The baseline characteristics of survivors were similar to
septic patients who died during follow-up, except for the
use of noradrenalin and for baseline lactate levels, which
were higher in nonsurvivors (Table 1).
In the control group, we included 38 apparently healthy
individuals without risk factors for cardiovascular disease
(57% women). Mean age of controls was 47 ± 14 years (P =
.18 for comparison with septic patients).
3.2. FMD response
Endothelium-dependent FMD response was 4 times lower
in septic patients compared with apparently healthy controls
(P b .001; Fig. 1). In addition, we did not observed a
Clinical characteristics of studied patients
(n = 42)
(n = 28)
(n = 14)
Source of sepsis
Time in shock (h)
Use of noradrenalin
Days in ICU
51 ± 19
48 ± 20
57 ± 15
16 ± 6
23 ± 7
8 ± 7
16089 ± 12089
0.9 ± 0.8
217 ± 129
2.0 ± 0.5
250 ± 109
2243 ± 1603
17 ± 6
22 ± 6
8 ± 7
16694 ± 12267
0.7 ± 0.7
198 ± 117
2.1 ± 0.6
234 ± 111
1917 ± 1197
16 ± 5
25 ± 8
9 ± 7
14880 ± 12087
1.3 ± 1.0
262 ± 149
1.96 ± 0.5
281 ± 101
2872 ± 2097
Data are expressed as mean ± SD, median (25-75 interquartile range), or absolute number (%).
APACHE indicates Acute Physiology And Chronic Health Evaluation; ICU, intensive care unit; Log, natural logarithm; CRP, C-reactive protein.
316.e11Endothelial dysfunction in severe sepsis and septic shock
statistically significant difference in FMD in patients with or
without intravenous vasopressors at baseline (P = .56).
Table 2 shows the prognostic value of FMD for
intrahospital mortality in patients with severe sepsis or
septic shock. Although we observed a trend toward
increased mortality rates in patients with baseline FMD
less than 7% (internationally accepted cutoff point) and in
patients with baseline FMD less than 5.7% (study
median), statistical significance was not achieved for the
comparison between survivors and nonsurvivors. None-
theless, 86% of nonsurvivors showed a decrease in
endothelial function in FMD sequential analyses (com-
parison between baseline measurement and measurements
after 24 or 72 hours), whereas only 43% of survivors had
a reduction in FMD (P = .01; Table 2).
Fig. 2 illustrates the temporal evolution of stratified
endothelial function in survivors and nonsurvivors. We
observed that patients with a favorable outcome depicted
gradual FMD improvement, reaching an absolute differ-
ence in FMD greater than 8% 72 hours after the onset of
sepsis (P for time-group interaction = .03; comparison at
72 hours: P b .05).
No significant association was observed between FMD
and the number of leukocytes (P = .89). However, a
comprehensive analysis of all assessments revealed a weak
but statistically significant negative correlation between
FMD and plasma lactate levels (r = −0.26, P = .007; n =
104). Flow-mediated vasodilation was not significantly
correlated to ET-1 levels at any time point. We observed
significant but modest negative correlations between FMD
and IL-6 (r = −0.32, P = .05) and vascular cell adhesion
molecule 1 (r = −0.43, P = .007). Finally, aggregated
analysis demonstrated that patients that did not deteriorate
endothelial function and had low levels of sVCAM-1 or
lactate (below median levels) had the best prognosis; patients
that deteriorate endothelial function had an intermediate
prognosis; and patients that had deterioration of endothelial
function and increased levels of sVCAM-1 or lactate had the
worse prognosis (Fig. 3A [sVCAM-1] and Fig. 3B [lactate];
P for trend ≤ .01).
In this prospective study, we have shown that septic
patients depicted remarkable changes in vasomotor endo-
thelial function evaluated by brachial artery ultrasound in the
early phases of severe sepsis or septic shock. We also
observed that nonsurvivors had a progressive decline in
FMD, a finding that was statistically significant 72 hours
after inclusion in the protocol.
Over the past decade, there has been great interest in the
assessment of endothelial function during sepsis. Most
published studies, however, have used serum markers
directly or indirectly associated to the infectious process
and with endothelial cells' activation. Reinhart et al 
have shown that IL-6 plasma levels greater than 1000 pg/mL
were highly predictive of increased risk of death by sepsis.
Interleukin 6, however, is a nonspecific inflammatory
marker, and increased levels of this cytokine may not
directly reflect changes on endothelial function. Endothelin
1, a powerful vasoconstrictor produced by endothelial cells
and released in response to physical and chemical stimuli,
has also been studied in the context of sepsis. In the early
stages of infection, ET-1 plasma levels raise considerably, an
event that has a potential beneficial effect on arterial pressure
and organ perfusion. However, long-lasting increments of
ET-1 concentration in the blood trigger a profound
vasoconstriction inducing tissue hypoperfusion, alterations
that are notably harmful to tissue and hemodynamic
homeostasis . We have shown previously that increased
levels of ET-1 might be predictive of increased risk of sepsis-
related mortality if measured very early (approximately 6
hours) after clinical deterioration . In the present
analysis, ET-1 measured during the first 24 hours of severe
sepsis was not predictive of increased risk of death, in
accordance with our previous findings .
Noninvasive point-of-care techniques might allow safe,
reproducible, and accurate evaluation of endothelial function
at bedside. A noninvasive technique for viewing the
microcirculation has been recently developed (video micro-
scope), thus enabling in-depth observation of small blood
vessels [19,20]. De Backer et al  have used this technique
to assess microvascular blood flow in the sublingual region
of patients with severe sepsis and healthy controls. Vessel
Different cutoff points of FMD and intrahospital
(n = 42)
(n = 28)
(n = 14)
Decline in FMDc
aFisher exact test.
bCutoff point = sample median.
cRefers to patients who presented FMD reduction 24 or 72 hours
after initial assessment.
Brachial Artery FMD (%)
N = 42 N = 37
1.5 ± 7
6 ± 4
P < .001
Comparison of FMD in healthy controls and patients with
316.e12L. Becker et al.
density was found to be significantly lower in patients with
severe sepsis (4.5 vs 5.4 mm, P b .01). Endothelium-
dependent microvascular reactivity has also been evaluated
by peripheral arterial tonometry in septic patients. Bedside
microvascular reactivity was impaired in proportion to sepsis
severity  and is inversely correlated to plasma levels of
angiopoietin 2 , an angiogenic peptide released by
endothelial cells that increases endothelial activation and
permeability. As such, this novel and exciting technique may
have a role as a user-independent method of monitoring
endothelial dysfunction in sepsis, if validated in large-scale
multicentric prospective studies.
The use of brachial artery ultrasound as a marker of
endothelial dysfunction in sepsis was recently described in
an Italian observational study. Vaudo et al  have
analyzed brachial artery FMD in 45 patients with Gram-
negative sepsis. Clinical parameters and FMD were
measured at baseline and 3 days after admission. In this
study, one third of the included patients had ultrasound-
based endothelial dysfunction at admission (FMD b7.5%).
This subset of individuals showed a decline in FMD and
sequential organ failure score after 3 days. The authors
concluded that reduction in brachial artery FMD predicted
sepsis-associated organ failure. In our study, we also
observed reduced FMD in septic patients when compared
with healthy controls, but our mean values were significantly
lower than those measured by the Italian group (81% of our
sample had FMD b7% at baseline). This finding is probably
P = .01P < .01
EF improvement and low sVCAM-1
EF deteroriation and high sVACM-1
EF improvement and low lactate
EF deteroriation and high lactate
Aggregated analysis demonstrating the impact of FMD evaluation associated to sVCAM-1 or lactate levels on mortality.
NS NSP < .05
Basal 24 hours72 hours
Brachial Artery FMD (%)
5.2 ± 4
3.8 ± 4
2.9 ± 5
-1.4 ± 9
-3.3 ± 10
0.8 ± 6
Temporal variation of FMD in patients with septic shock stratified by intrahospital evolution.
316.e13Endothelial dysfunction in severe sepsis and septic shock
explained by differences in sepsis severity between studies. Download full-text
In the Italian report, all patients had sepsis without evidence
of organ failure at study entry, whereas severe sepsis and
septic shock were both inclusion criteria in our protocol.
Furthermore, intrahospital mortality was 33% in our sample,
as opposed to only 4% in their report.
Interestingly, ET-1, a renowned marker of endothelial
activation, was not correlated to brachial artery FMD in our
patients, demonstrating that both parameters might depict
different aspects of endothelial dysfunction. The potential
role of bedside endothelial function analysis as a prognostic
tool in sepsis was demonstrated by the significant differences
in FMD 72 hours after baseline evaluation (Fig. 2). In
addition, increased sVCAM-1 plasma levels appeared to add
prognostic information over FMD analysis.
Some aspects of our study design deserve consideration.
Sample size was calculated to detect differences in FMD
between septic patients and healthy controls. We acknowl-
edge that this sample size was relatively limited and some of
our findings should be interpreted as hypotheses generator.
In addition, endothelium-independent vasodilatation was not
evaluated by sublingual nitrate administration in the present
protocol because of the hemodynamic instability that
characterized our patients, thus limiting our conclusions to
the endothelial-dependent aspect of the vasomotor function.
The relatively low values of brachial FMD observed in the
control group might be explained, in part, by the relatively
high mean age of this subjects (47 ± 14 years old; range, 31-
85 years). In analysis of FMD restricted to the youngest
subgroup of controls (lowest quartile), mean FMD was in the
expected reference range (8 ± 5%) . Finally, it is difficult
to conclusively separate the effect of vasoactive drugs on
FMD evaluation. In our study, only 43% of patients were
receiving noradrenaline at 72 hours evaluation. However, we
have not observed a significant difference in the percentage
of patients on vasopressor at 72 hours between those that
improved or decreased FMD throughout the study (39% vs
46%; χ2P value = .65). We acknowledge that further studies
are needed to specifically address this issue.
In summary, we conclude that changes in FMD occur early
on in septic patients with hemodynamic instability, with
progressive FMD decline over the first 72 hours representing
poor prognosis. However, further prospective studies with
larger samples are required to validate the predictive value of
endothelial dysfunction assessed by brachial artery ultra-
sound for risk assessment in severe sepsis and septic shock.
 American College of Chest Physician Society of critical care medicine
Consensus conference. Definition for sepsis and organ failure and
guidelines for the use of innovative therapies in sepsis. Crit Care Med
 Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of
sepsis in the United States from 1979 through 2000. N Engl J Med
 Iskit AB, Guc O. Effects of endothelin and nitric oxide on organ injury,
mesenteric ischemia, and survival in experimental models of septic
shock. Acta Pharmacol Sin 2003;24:953-7.
 Reinhardt K, Bayer O, Brunkhorst F, Meisner M. Markers of
endothelial damage in organ dysfunction and sepsis. Crit Care Med
 Corretti MC, Anderson TJ, Benjamin EJ, et al, International Brachial
Artery Reactivity Task Force. Guidelines for the ultrasound assess-
ment of endothelial-dependent flow-mediated vasodilation of the
brachial artery. J Am Coll Cardiol 2002;39:257-65.
 Quyyumi AA. Prognostic value of endothelial function. Am J Cardiol
 Perrone-Filardi P, Cuocolo A, Brevetti G, et al. Relation of brachial artery
flow-mediated vasodilation to significant coronary artery disease in patients
with peripheral arterial disease. Am J Cardiol 2005;96:1337-41.
 Wu WC, Sharma SC, Choudhary G, et al. Flow-mediated vasodilation
predicts the presence and extent of coronary artery disease assessed by
stress thallium imaging. J Nucl Cardiol 2005;12:538-44.
 Meyer B, Mörtl D, Strecker K, et al. Flow-mediated vasodilation predicts
outcome in patients with chronic heart failure comparison with B-type
natriuretic peptide. J Am Coll Cardiol 2005;46:1011-8.
 Witte DR, Westerink J, de Koning EJ, et al. Is the association between
flow-mediated dilation and cardiovascular risk limited to low-risk
populations? J Am Coll Cardiol 2005;45:1987-93.
 Gokce N, Keaney Jr JF, Hunter LM, et al. Predictive value of
noninvasively determined endothelial dysfunction for long-term
cardiovascular events in patients with peripheral vascular disease.
J Am Coll Cardiol 2003;41:1769-75.
vascular surgery. Arterioscler Thromb Vasc Biol 2007;27:2113-9.
determined endothelial dysfunction for long-term cardiovascular events and
restenosis in patients undergoing coronary stent implantation: a prospective
study. Coron Artery Dis 2008;19:337-43.
 Corrado E, Rizzo M, Coppola G, et al. Endothelial dysfunction and
carotid lesions are strong predictors of clinical events in patients with
early stages of atherosclerosis: a 24-month follow-up study. Coron
Artery Dis 2008;19:139-44.
 Vaudo G, Marchesi S, Siepi D, et al. Human endothelial impairment in
sepsis. Atherosclerosis 2008;197:747-52.
 Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a
severity of disease classification system. Crit Care Med 1985;13(10):
 Reinhart K, Menges T, Gardlund B, et al. Randomized, placebo-
controlled trial of the anti-tumor necrosis factor antibody fragment
afelimomab in hyperinflammatory response during severe sepsis: the
RAMSES Study. Crit Care Med 2001;29:765-9.
 Brauner JS, Rohde LE, Clausell N. Circulating endothelin-1 and tumor
necrosis factor-alpha: early predictors of mortality in patients with
septic shock. Intensive Care Med 2000;26:305-13.
 Mathura KR, Vollebregt KC, Boer K, et al. Comparison of OPS
imaging to intravital capillaroscopy of nail fold microcirculation.
J Appl Physiol 2001;91:74-8.
 Lundy DJ, Trzeciak S. Microcirculatory dysfunction in sepsis. Crit
Care Clin 2009;25:721-31.
in patients with sepsis. Am J Respir Crit Care Med 2002;166:98-104.
 Davis JS, Yeo TW, Thomas JH, et al. Sepsis-associated microvascular
dysfunction measured by peripheral arterial tonometry: an observa-
tional study. Crit Care 2009;13(5):R155.
 Davis JS, Yeo TW, Piera KA, et al. Angiopoietin-2 is increased in
sepsis and inversely associated with nitric oxide-dependent microvas-
cular reactivity. Crit Care 2010;14(3):R89.
 Hamburg NM, Palmisano J, Larson MG, et al. Relation of brachial and
digital measures of vascular function in the community. The
Framingham Heart Study. Hypertension 2011;57:390-6.
316.e14 L. Becker et al.