Histopathological Validation of Grayscale Carotid Plaque
Characteristics Related to Plaque Vulnerability
Carol C. Mitchell, PhD
a
, James H. Stein, MD, FAHA
a
, Thomas D. Cook, PhD
b
, Shahriar
Salamat, MD, PhD
c
, Xiao Wang, PhD
d
, Tomy Varghese, PhD
d
, Daren C. Jackson, PhD
e,*
,
Carolina Sandoval Garcia, MD
f
, Stephanie M. Wilbrand, PhD
f
, and Robert J. Dempsey, MD
f
a
Department of Medicine, Cardiovascular Medicine Division, University of Wisconsin School of
Medicine and Public Health, USA, Madison WI, 53792
b
Department of Biostatistics and Medical Informatics, University of Wisconsin School of Medicine
and Public Health, USA, Madison WI, 53792
c
Department of Pathology and Laboratory Medicine, University of Wisconsin School of Medicine
and Public Health, USA, Madison WI, 53792
d
Department of Medical Physics, University of Wisconsin School of Medicine and Public Health,
USA, Madison, WI 53705
e
Department of Neurology, University of Wisconsin School of Medicine and Public Health, USA,
Madison, WI 53705
f
Department of Neurological Surgery, University of Wisconsin School of Medicine and Public
Health, USA, Madison, WI 53792
Abstract
Inflammation and angiogenesis play major roles in carotid plaque vulnerability. The purpose of
this study was to determine if grayscale features of carotid plaques are associated with histological
markers for inflammation. Thirty-eight individuals completed a dedicated research carotid
ultrasound exam prior to carotid endarterectomy. Grayscale analysis was performed on plaque
Corresponding Author: Carol Mitchell, University of Wisconsin, Department of Medicine/Division of Cardiovascular Medicine,
University of Wisconsin School of Medicine and Public Health, K6/322 Clinical Science Center, 600 Highland Ave MC 3248,
Madison, Wisconsin 53792, Phone: 608-262-0680, Fax: 608-263-0405, ccm@medicine.wisc.edu.
*
now at Wisconsin Surgical Outcomes Research Program, Department of Surgery, University of Wisconsin School of Medicine and
Public Health, USA, Masison, WI 53792
Clinical Trial Registration:
Unique Identifier: NCT02476396
Conflicts of Interest
C.C. Mitchell: Other; Davies Publishing Inc., authorship for two echocardiography textbooks, currently under review, may have future
royalties. Elsevier, Wolters Kluwer, author textbook chapters, may have future royalties.
J.H. Stein: Other; Wisconsin Alumni Research Foundation-patent related to carotid wall thickness and vascular age.
T.D. Cook: Consultant/Advisory Board; GlaxoSmithKline, Bristol-Myers Squibb, Merck, Mast Therapeutics.
S. Salamat: None.
X. Wang: None.
T. Varghese: Other; Siemens Ultrasound, Research Agreement for use of Ultrasound Research Interface. No financial benefit.
D.C. Jackson: None.
C. Sandoval Garcia: None.
S.M. Wilbrand: None.
R.J. Dempsey: None.
HHS Public Access
Author manuscript
Ultrasound Med Biol
. Author manuscript; available in PMC 2018 January 01.
Published in final edited form as:
Ultrasound Med Biol
. 2017 January ; 43(1): 129–137. doi:10.1016/j.ultrasmedbio.2016.08.011.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
images to measure plaque echogenicity (grayscale median [GSM] pixel brightness), plaque area,
presence of discrete white areas [DWAs], and the percent of black area near the lumen on any one
component of the plaque. Plaques with higher ultrasound GSM had greater percent calcification
(p=0.013) on histopathology. Presence of an ultrasound DWA was associated with more plaque
hemosiderin (p=0.0005) and inflammation (p=0.019) on histopathology examination. The percent
of plaque black area in any one component was associated with a higher score for macroscopic
ulceration (p=0.028). Ultrasound plaque characteristics (GSM, DWAs and Black areas) represent
histopathological markers associated with plaque vulnerability.
Keywords
carotid plaque; vulnerable plaque; ultrasound grayscale imaging
Introduction
Treatment options for patients with carotid artery atherosclerosis are largely influenced by
the presence of clinical symptoms and medical imaging estimates of the degree of stenosis
and plaque surface characteristics (Liapis et al. 2009; Dempsey et al. 2010; Brott et al. 2011;
Ricotta et al. 2011; Salem et al. 2014). However, these criteria do not account for cumulative
arterial damage that can lead to clinically unrecognized (“silent”) strokes due to the presence
of vulnerable plaques and microemboli (Dempsey et al. 2010). It is estimated that for every
clinical stroke diagnosis there are approximately five silent strokes which go unnoticed and
are associated with cognitive decline, especially in executive function skills (Seshadri 2006;
Dempsey et al. 2010; Rocque et al. 2012; Wang et al. 2013). Therefore, better ways to
evaluate plaque vulnerability are essential to optimize patient treatment and management not
only for major stroke prevention but also to delay cognitive impairment due to arterial injury.
Vulnerable plaques are associated with thin fibrous caps, large lipid cores, intraplaque
hemorrhage, inflammation (Stary et al. 1995; Fleiner et al. 2004; Salem et al. 2013; Marnane
et al. 2014; Salem et al. 2014), and in some reports calcification (Shaalan et al. 2004).
Ultrasound methods used to assess carotid plaque for features of vulnerability include;
integrated backscatter (IBC) (Bridal et al. 2000; Kawasaki et al. 2001; Nagano et al. 2008),
midband, slope and intercept values of straight-line fit (MBF) to the apparent backscatter
transfer function, (Waters et al. 2003) carotid strain imaging, (Maurice et al. 2005; Shi et al.
2008; Shi et al. 2009; Wang et al. 2013; Wang et al. 2016a; Wang et al. 2016b), acoustic
radiation impulse force imaging (ARFI) based approaches (Czernuszewicz et al. 2015),
shearwave elastography (Garrard et al. 2015) and grayscale analyses of plaque features (El-
Barghouty et al. 1996; Tegos et al. 2000; Grogan et al. 2005; Salem et al. 2014). IBC, MBF,
zero frequency intercept, and carotid strain imaging utilize the raw radiofrequency echo
signal to determine associations between scattering properties of the plaque tissues and
plaque composition. In grayscale analyses, B-mode ultrasound images are digitized and
processed with specialized software to calculate a grayscale median (GSM) value and to
demonstrate pixel brightness distribution based on the grayscale value of groups of pixels
(Lal et al. 2002; Nicolaides et al. 2010). Images are normalized with blood as the reference
for black, and the vessel wall (adventitia) as the reference point for white. Comparing pixel
Mitchell et al.
Page 2
Ultrasound Med Biol
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
brightness values to histology specimens allows for determination of what type of tissue is
most likely composing a plaque. Lal et al. (2002) imaged volunteers and patients undergoing
carotid endarterectomy (CEA). Volunteers were scanned to determine pixel brightness
values of bone (utilizing the tibial head and cranium), fat (utilizing the sub-cutaneous fat of
the abdomen), muscle (biceps muscle), and fibrous tissue (iliotibial tract). A blood vessel
was imaged in the same frame as all of these different tissue types so that images could be
normalized with blood representing a grayscale value of 0 and adventitia representing a
grayscale value of 190. Patients undergoing CEA had their plaques analyzed and the pixel
brightness distribution was compared to values attained in normal tissue of volunteers and
histopathology specimens. Findings demonstrated good agreement of ultrasound grayscale
values predicting the type of tissue present in plaques (i.e. calcium, blood, lipid, fibrous
tissue), with a significant higher amount of blood and lipid associated with symptomatic
plaques compared to asymptomatic plaques (p=0.0048 for blood, and p=0.026 for lipid) (Lal
et al. 2002). Evaluation of plaque surface characteristics have also been utilized to
characterize features of plaques associated with vulnerability (Kanber et al. 2013a). Kanber
et al 2013a utilized a plaque surface irregularity index to objectively quantify plaque surface
irregularities. The authors found that plaques with a higher surface irregularity index were
associated with ipsilateral cerebrovascular symptoms (Kanber et al. 2013a).
Grayscale findings that have been associated with plaque vulnerability are 1) low grayscale
median (GSM) values (Nicolaides et al. 2010; Ruiz-Ares et al. 2011; Ibrahimi et al. 2014;
Ruiz-Ares et al. 2014), 2) juxtaluminal black area (JBA) size (Griffin et al. 2010; Kakkos et
al. 2013), 3) discrete white area (DWA) presence (Nicolaides et al. 2010), and 4) large
plaque area (Nicolaides et al. 2010; Salem et al. 2014). Studies using these parameters have
demonstrated correlations between low echogenicity and worrisome histopathological
plaque features, few have described how grayscale features (Salem et al. 2014) are
associated with inflammation (Griffin et al. 2010; Nicolaides et al. 2010; Kakkos et al. 2013;
Salem et al. 2014; Kanber et al. 2015), except when ultrasound contrast agents are used
(Shah et al. 2007; Hjelmgren et al. 2014). The purpose of this study was to determine if
grayscale features of carotid plaques are associated with histological markers for
inflammation.
Methods
Participants
Patients were recruited to subjects in the “Structural Stability of Carotid Plaque and
Symptomatology” (NIH funded study: R01 NS064034) study from 2010–2015. This study
was approved by the University of Wisconsin Health Sciences Institutional Review Boards
and all subjects provided informed consent. All subjects met criteria for surgical carotid
endarterectomy (>60% carotid arterial stenosis [percent stenosis determined by computed
tomography angiography, magnetic resonance angiography, or ultrasound imaging];
symptomatic patients had symptoms of motor and/or language deficits on examination and
asymptomatic patients had no deficits but may have had silent strokes seen on imaging)
( North American Symptomatic Carotid Endarterectomy Trial (NASCET) Steering
Committee; North American Symptomatic Carotid Endarterectomy Trial Collaborators
Mitchell et al.
Page 3
Ultrasound Med Biol
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
(NASCET), Walker et al. 1995). Subjects were recruited from the neurosurgery clinics and
inpatient units. Potential participants were excluded if they had prior carotid artery surgery,
prior carotid artery endovascular procedures, prior cervical radiation, or were otherwise
deemed unsuitable for carotid endarterectomy. Thirty-eight patients completed the
ultrasound research clinical imaging session and had pathology results available for
comparison at the time of this analysis.
Ultrasound Imaging
Ultrasound images were acquired on average (standard deviation) 10.6 (13.8) days prior to
carotid endarterectomy. The clinical imaging portion of the protocol was performed with an
Acuson S2000 ultrasound system (Siemens Medical Solutions USA, Inc., Malvern, PA) and
9L4 transducer. The ultrasound clinical imaging protocol consisted of transverse and
longitudinal imaging of the common carotid, carotid bulb, internal carotid and external
carotid arteries. Pulsed wave Doppler velocities were recorded in the common carotid artery
(proximal, mid and distal segments), carotid bulb, internal carotid artery (proximal, mid and
distal segments) and external carotid artery. Color Doppler images were acquired in the
transverse and longitudinal planes in the common carotid, carotid bulb, internal and external
carotid arteries. All images were saved digitally in DICOM format and analyzed with
Access Point Software (Freeland Systems, LLC, Alpharetta, GA) to select the frames most
representative of the plaque, which were then saved as a bitmap for plaque grayscale
analysis. All scans were performed by two certified ultrasonographers.
Carotid Plaque Grayscale Analysis
Ultrasound images were analyzed using the “Plaque Texture Analysis Software” package
(LifeQ Medical, Nicosia, Cyprus). Images were normalized with the blackest area of the
blood representing a grayscale value of 0 and the brightest white area of the adventitia
representing a grayscale value of 190 (Nicolaides et al. 2010). After normalization, images
were standardized to a pixel density of 20 pixels per mm (Nicolaides et al. 2010). The
following grayscale features were analyzed for each plaque image: GSM, plaque area,
presence of a discrete white area (DWA), presence of a black area adjacent to the lumen
either with or without an echogenic border, and the percent black area of any one component
of the plaque. The GSM value was calculated using the median grayscale value of the traced
plaque. The presence of a DWA was noted by the identification of a white area in the plaque
without acoustic shadowing and a grayscale value greater than 126 (Nicolaides et al. 2010)
(color coded red, figure 1). Juxtaluminal black areas (JBAs) without an echogenic border
and black areas with a border (BABs) were identified as areas with a grayscale value of less
than 25 near the lumen and color coded as black with the plaque analysis software. In
subjects with more than one black area, the largest black area was measured. The black area
was traced to obtain an area measure (figure 2) and from this measurement the percent black
area of any one component was calculated. The entire plaque was manually traced and the
plaque area calculated (reported in mm
2
). In images with acoustic shadowing, the plaque
was traced to include the echogenic portion of the plaque and exclude the portion containing
acoustic shadowing. All image selection and analyses were performed by a single, trained
investigator.
Mitchell et al.
Page 4
Ultrasound Med Biol
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Histopathology
Paraffin-embedded plaque sections were stained using standard hematoxylin and eosin
methods. Sections were assessed by a single pathologist blinded to subject characteristics for
the percent calcification, cholesterol content, hemorrhage, hemosiderin content, and
inflammatory cell infiltration. Histologic classification of plaques was made using the
updated classification of atherosclerotic plaques recommended by the American Heart
Association (Tureyen et al. 2006). Plaque specimens also were scored on an ordinal scale
(0–3) for the presence of hemosiderin and inflammation (figure 3). A score of 0 represented
no hemosiderin or inflammation, 1 was minimal, 2 was moderate and 3 represented
extensive presence of hemosiderin and/or inflammation.
Surgical Ulceration Scoring
All carotid endarterectomies and ulceration scores were provided by a board-certified
physician. At the time of carotid endarterectomy, the surgeon scored the plaque on a
subjective, ordinal scale of 1 to 4. A score of 1 represented minimal ulceration and a score of
4 represented extensive ulceration.
Statistical Analysis
Carotid plaque characteristics (GSM, presence of DWA, presence and size of JBA or BAB,
and plaque area) were analyzed and compared to histopathology. Statistical calculations
were performed using R Core Team (R Development Core Team 2015). Kendall’s Rank
Correlation Tau was utilized to assess the percent calcification reported in the plaque
histopathology and its relationship to the GSM. The Wilcoxon Rank Sum Test was utilized
to assess the relationships between diffuse histopathological inflammation and hemosiderin
and the presence of DWAs, the surgeon’s ulceration score, and the percent black areas of a
component of a plaque.
Results
Subjects
There were 38 subjects that completed the ultrasound research protocol that had
histopathology results available for analysis. Subjects were mean (standard deviation) 69
(9.5) years old (range 43–87 years); 16 (42%) were female; 24 (63%) were symptomatic.
Characteristics for all patients are presented in Table 1. For each subject a single plaque
image, that was the most representative of the grayscale features, was selected for analysis,
except for one patient in which two images were selected. Most representative was defined
as the longitudinal image plane in which the plaque appeared the largest, based on maximum
lumen encroachment, maximum length extended and best border definition. Color Doppler
loops were also used to ensure that the image selected also defined the plaque borders the
best. In one patient two images were required to demonstrate the entire length of the plaque.
One image demonstrated plaque extending from the distal CCA to the bulb and the other
image demonstrated the plaque extending into the ICA. In this patient, the plaque extending
into the ICA demonstrated the DWAs. Grayscale features of GSM, presence of DWA and/or
black area and plaque area were selected based on the correlates of these features with
Mitchell et al.
Page 5
Ultrasound Med Biol
. Author manuscript; available in PMC 2018 January 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript