Three-dimensional ultrasound evaluation of the effects of pomegranate therapy on carotid plaque texture using locality preserving projection

https://doi.org/10.1016/j.cmpb.2019.105276Get rights and content

Highlights

  • A texture-based biomarker was developed that detected changes in atherosclerotic plaque due to dietary intervention with high sensitivity.

  • Established the effect of pomegranate juice and tablets in a statistically significant manner for the first time.

  • The proposed biomarker requires 20 times fewer subjects than total plaque volume (TPV) to detect the effect of pomegranate.

Abstract

Background and objective: Dietary supplements are expected to confer a smaller beneficial effect than medical treatments. Therefore, there is a need to develop cost-effective biomarkers that can demonstrate the efficacy of such supplements for carotid atherosclerosis. The aim of this study is to develop such a biomarker based on the changes of 376 plaque textural features measured from 3D ultrasound images.

Methods: Since the number of features (376) was greater than the number of subjects (171) in this study, principal component analysis (PCA) was applied to reduce the dimensionality of feature vectors. To generate a scalar biomarker for each subject, elements in the reduced feature vectors produced by PCA were weighted using locality preserving projections (LPP) to capture essential patterns exhibited locally in the feature space. 96 subjects treated by pomegranate juice and tablets, and 75 subjects receiving placebo-matching juice and tablets were evaluated in this study. The discriminative power of the proposed biomarker was evaluated and compared with existing biomarkers using t-tests. As the cost of a clinical trial is directly related to the number of subjects enrolled, the cost-effectiveness of the proposed biomarker was evaluated by sample size estimation.

Results: The proposed biomarker was more able to discriminate plaque changes exhibited by the pomegranate and placebo groups than total plaque volume (TPV) according to the result of t-tests (TPV: p=0.34, Proposed biomarker: p=1.5×105). The sample size required by the new biomarker to detect a significant effect was 20 times smaller than that required by TPV.

Conclusion: With the increase in cost-effectiveness afforded by the proposed biomarker, more proof-of-principle studies for novel treatment options could be performed.

Introduction

Stroke is the second leading cause of death worldwide, with 5.5 million stroke-related deaths in 2016 [1], [2]. China, accounting for one-fifth of the population, has a disproportionate share of stroke mortality, which stood at 1.8 million in 2016 and accounted for one-third of the total number of deaths from stroke worldwide [1], [2]. Carotid atherosclerosis is a major source of atherosclerotic emboli (platelet aggregates and plaque debris) that would block cerebral arteries, leading to ischemic strokes. Fortunately, for patients with high stroke risk, 75–80% of stroke can be prevented by lifestyle/dietary changes and medical therapies [3]. With recent advances in the pathogenesis of atherosclerosis, the numbers of new interventions are expected to rise rapidly. Effects of these treatment or management strategies are required to be thoroughly validated in clinical trials. Therefore, in parallel to the development of novel therapies, there is a critical requirement of measurement tools for cost-effective serial monitoring of carotid atherosclerosis.

Carotid intima-media thickness (IMT) measured from 2D B-mode ultrasound was an early imaging biomarker introduced to increase the sensitivity of carotid disease monitoring, but recent investigations suggest two major weaknesses of IMT. First, the annual change of IMT is small ( ~ 0.15 mm, below the spatial resolution of carotid ultrasound) and does not permit the measurement of change in individuals within a clinically affordable timeframe [4]. Second, IMT measures vascular wall thickening, which is not directly related to atherosclerosis [5], rendering it a weak predictor of cardiovascular risk [6], [7]. Although total plaque area (TPA) measurement has emerged as a more accurate metric for stroke risk prediction [8], both IMT and TPA are measured from 2D ultrasound images, acquisition of which requires an operator to choose an imaging plane to be scanned, making an image difficult to reproduce, even for the same operator. Therefore, 2D ultrasound measurements are sub-optimal in serial monitoring of plaque [9].

Development of 3D ultrasound has allowed for more sensitive and reproducible quantification of carotid atherosclerosis. Total plaque volume (TPV) measured from 3D ultrasound has been shown to be reproducible [10], able to predict vascular events [11] and sensitive to disease progression and regression in a clinical trial involving high-dose atorvastatin [12]. The risk of cardiovascular events is related to the presence of vulnerable plaques and the vulnerability of a plaque is determined by its composition. Changes in plaque textural features have served as a surrogate for plaque composition assessment. Textural analysis was shown to be more accurate than TPV in showing plaque regression with high-dose atorvastatin versus placebo subjects [13]. Van Engelen et al. [14] showed that textural change was more accurate than TPV change in predicting cardiovascular events. Zhou et al. [15] showed that fractal dimension texture analysis was more able than TPV to discriminate subjects receiving atorvastatin and placebo. A number of investigations performed in 2D ultrasound have also shown that textural features were useful in identifying symptomatic patients [16], [17]. Roy-Cardinal et al. [18] showed that quantitative tissue information extracted from ultrasound imaging was useful in classifying carotid artery plaque components, including lipid, calcification and ruptured fibrous cap. These results suggest that a biomarker based on plaque textural features would be more able than TPV to highlight treatment effects. The increased discriminative power conferred by such a texture-based biomarker could substantially reduce the sample size, duration and therefore cost required to establish the efficacy of novel dietary/lifestyle interventions.

Many textural features have been extracted from carotid ultrasound images in investigations involving the stratification of stroke risk [16], [19], [20]. Christodoulou et al. [16] extracted 61 texture features and shape parameters from 2D longitudinal carotid ultrasound images to classify symptomatic statuses using a combination of classifiers. Kyriacou et al. [19] extracted 43 texture features from 2D longitudinal ultrasound images in a stroke risk assessment study. The probabilistic neural network (PNN) and support vector machine (SVM) were used to classify the symptomatic statuses of patients. Araki et al. [20] extracted 16 textural features from 2D longitudinal images and classified the stroke risk into high and low using SVM. The ground truth classification was determined by lumen diameter. Huang et al. [21] extracted 300 texture features in classifying plaques into 3 categories (i.e., hyperechoic, intermediate and anechoic) using the K-nearest neighbour classifier. The ground truth labels were manually assigned by expert observers. These techniques were evaluated in cross-sectional analyses based on 2D carotid ultrasound images, but not in longitudinal studies of plaque change. The increased reproducibility afforded by 3D carotid ultrasound imaging techniques has allowed for 3D textural analysis of plaque change. Awad et al. [13] extracted 270 textural features to detect statin-related changes in carotid atherosclerosis and showed that textural features were more accurate than TPV in classifying patients who received high-dose atorvastatin and placebo. Van Engelen et al. [14] extracted 376 features to predict cardiovascular events in high-risk patients and showed that textural change was more accurate than TPV change in predicting cardiovascular events. The 376 features extracted in this study were generated by 9 textural extraction techniques and encompass most features used in the 2D and 3D studies discussed above.

In the current study, we aim to develop a cost-effective texture-based biomarker capable of quantifying the degree of plaque change. The 2D and 3D textural analyses discussed above primarily involved feature-based classification. The classifiers generated discrete class labels but did not allow the quantification of how much change has occurred in a plaque. To address this issue, we developed a scalar, easy-to-interpret and discriminative biomarker integrating the 376 plaque features extracted in Ref. [14], which encompass textural features used in previous studies. In addition, a physical understanding of the biomarker should be established for it to be accepted by clinicians. One way of characterizing the biomarker is to identify important textural features that contribute to the biomarker’s ability for discriminating treatment and placebo subjects. A metric was proposed in this study to determine the weight of each of the 376 features in the proposed biomarker (Eq. (7)).

The proposed biomarker was validated in a clinical trial aiming to evaluate the effect of pomegranate juice and tablets. Lipid oxidation in arterial macrophages and lipoproteins is a major contributor to carotid atherosclerosis. As a rich source of potent anti-oxidants [22], pomegranate juice has been shown to inhibit LDL oxidation and attenuate atherosclerosis development in animal studies [23], [24]. However, a recent 18-month randomized placebo-controlled trial aiming at assessing the effect of pomegranate juice in 289 subjects at risk of coronary disease and stroke showed that no difference in carotid IMT change was found between the pomegranate and the control group (p = 0.65) [25], thereby casting a doubt on whether pomegranate juice is effective in slowing the progression of atherosclerosis.

As pomegranate is a dietary supplement expected to confer smaller beneficial effect than high-dose atorvastatin, the requirement on the discriminative power for the proposed biomarker is higher than that required for detecting the effect of atorvastatin in Ref. [13]. We hypothesized that the proposed biomarker has enough discriminative power in detecting the effects of pomegranate juice/tablets in a placebo-controlled study. This hypothesis was extensively evaluated in the current study.

Section snippets

Study subjects and ultrasound image acquisition and preprocessing

Subjects were recruited at the Stroke Prevention & Atherosclerosis Research Centre at Robarts Research Institute (London, Ontario, Canada) for a registered clinical trial (ISRCTN30768139). The 171 subjects involved in this study were randomized into two groups. 96 subjects received pomegranate extract in a tablet and juice form once daily, and 75 subjects received a placebo-matching tablet and juice lacking the active ingredients once daily. A total of 629 carotid plaques were identified from

Results

Table 2 shows the background medications taken by pomegranate and placebo subjects. No statistically significant difference was observed between the background medications taken by the two groups. Plaque textural and volume measurements were obtained at baseline and 376 ± 23 days (range: 283 - 579 days) later. Since the time interval between baseline and follow-up varies with subjects, TPV and textural change per year were calculated to allow comparisons among subjects. Nx for PCA and K for

Discussion

We developed a cost-effective and computationally efficient texture-based biomarker that detected changes in atherosclerotic plaques due to dietary supplements such as pomegranate. A previous study of the effect of pomegranate juice on carotid IMT [25] showed no significant effect, but the finding may be attributed to the insensitivity of IMT in detecting treatment effect. TPV, as a three-dimensional measurement, has shown to be more able to detect change [51], [52] and more correlated to

Declaration of Competing Interest

The patient data were acquired in a clinical study sponsored by POM Wonderful, LLC (USA). Professor Spence is the Scientific Officer of Vascularis Inc.

Acknowledgments

Dr. Chiu is grateful for funding support from the Research Grant Council of the HKSAR, China (Project nos. CityU 11205917, CityU 11203218) and the City University of Hong Kong Strategic Research Grants (nos. 7004617, 7005226).

References (79)

  • H. Sillesen et al.

    Carotid plaque burden as a measure of subclinical atherosclerosis: comparison with other tests for subclinical arterial disease in the high risk plaque bioimage study

    JACC

    (2012)
  • F. Rakebrandt et al.

    Relationship between ultrasound texture classification images and histology of atherosclerotic plaque

    Ultrasound Med. Biol.

    (2000)
  • A. Leahy et al.

    Duplex ultrasonography and selection of patients for carotid endarterectomy: plaque morphology or luminal narrowing?

    J. Vasc. Surg.

    (1988)
  • M. Langsfeld et al.

    The role of plaque morphology and diameter reduction in the development of new symptoms in asymptomatic carotid arteries

    J. Vasc. Surg.

    (1989)
  • M. Giannoni et al.

    Minor asymptomatic carotid stenosis contralateral to carotid endarterectomy (CEA): our experience

    Eur. J. Vasc. Surg.

    (1991)
  • R. Holdsworth et al.

    Symptoms, stenosis and carotid plaque morphology is plaque morphology relevant?

    Eur. J. Vasc. Endovasc. Surg.

    (1995)
  • L.W. O’Holleran et al.

    Natural history of asymptomatic carotid plaque: five year follow-up study

    Am. J. Surg.

    (1987)
  • E. Cave et al.

    Carotid artery duplex scanning: does plaque echogenicity correlate with patient symptoms?

    Eur. J. Vasc. Endovasc. Surg.

    (1995)
  • L.M. Reilly et al.

    Carotid plaque histology using real-time ultrasonography: clinical and therapeutic implications

    Am. J. Surg.

    (1983)
  • M.-L.M. Grønholdt et al.

    Lipid-rich carotid artery plaques appear echolucent on ultrasound b-mode images and may be associated with intraplaque haemorrhage

    Eur. J. Vasc. Endovasc. Surg.

    (1997)
  • N. Liasis et al.

    The use of speckle reduction imaging (SRI) ultrasound in the characterization of carotid artery plaques

    Eur. J. Radiol.

    (2008)
  • C.O. Johnson et al.

    Global, regional, and national burden of stroke, 1990–2016: a systematic analysis for the global burden of disease study 2016

    Lancet Neurol.

    (2019)
  • J.D. Spence

    Intensive management of risk factors for accelerated atherosclerosis: the role of multiple interventions

    Curr. Neurol. Neurosci. Rep.

    (2007)
  • J.D. Spence

    Recent advances in pathogenesis, assessment, and treatment of atherosclerosis

    F1000Research

    (2016)
  • A.V. Finn et al.

    Correlation between carotid intimal/medial thickness and atherosclerosis: a point of view from pathology

    Arterioscler. Thrombosis Vasc. Biol.

    (2010)
  • H.M. Den Ruijter et al.

    Common carotid intima-media thickness measurements in cardiovascular risk prediction: a meta-analysis

    Jama

    (2012)
  • A. Zanchetti et al.

    Baseline values but not treatment-induced changes in carotid intima-media thickness predict incident cardiovascular events in treated hypertensive patients: findings in the European Lacidipine Study on Atherosclerosis (ELSA)

    Circulation

    (2009)
  • J.D. Spence

    Ultrasound measurement of atherosclerosis

    Stroke

    (2004)
  • B. Chiu et al.

    Development of 3D ultrasound techniques for carotid artery disease assessment and monitoring

    Int. J. Comput. Assist. Radiol. Surg.

    (2008)
  • T. Wannarong et al.

    Progression of carotid plaque volume predicts cardiovascular events

    Stroke

    (2013)
  • C.D. Ainsworth et al.

    3D ultrasound measurement of change in carotid plaque volume: a tool for rapid evaluation of new therapies

    Stroke

    (2005)
  • J. Awad et al.

    Texture analysis of carotid artery atherosclerosis from three-dimensional ultrasound images

    Med. Phys.

    (2010)
  • A. van Engelen et al.

    Three-dimensional carotid ultrasound plaque texture predicts vascular events

    Stroke

    (2014)
  • R. Zhou et al.

    Fractal dimension based carotid plaque characterization from three-dimensional ultrasound images

    Med. Biol. Eng. Comput.

    (2019)
  • C.I. Christodoulou et al.

    Texture-based classification of atherosclerotic carotid plaques

    IEEE Trans. Med. Imaging

    (2003)
  • U.R. Acharya et al.

    Atherosclerotic plaque tissue characterization in 2D ultrasound longitudinal carotid scans for automated classification: a paradigm for stroke risk assessment

    Med. Biol. Eng. Comput.

    (2013)
  • M.-H. Roy-Cardinal et al.

    Assessment of carotid artery plaque components with machine learning classification using homodyned-K parametric maps and elastograms

    IEEE Trans. Ultrason. Ferroelectr. Freq. Control

    (2018)
  • E.C. Kyriacou et al.

    An integrated system for assessing stroke risk

    IEEE Eng. Med. Biol. Mag.

    (2007)
  • X. Huang et al.

    Classification of carotid plaque echogenicity by combining texture features and morphologic characteristics

    J. Ultrasound Med.

    (2016)
  • Cited by (11)

    • Three-dimensional ultrasound assessment of effects of therapies on carotid atherosclerosis using vessel wall thickness maps

      2021, Ultrasound in Medicine and Biology
      Citation Excerpt :

      However, this may be due to the limited sensitivity of IMT to the effect of therapy. We have previously developed two plaque texture biomarkers (Chen et al., 2020; Lin et al., 2020) that established the effect of the pomegranate juice and extract in the same subject cohort investigated in the current paper. However, measurements of plaque texture require reproducible plaque segmentation.

    • Interpretable Deep Biomarker for Serial Monitoring of Carotid Atherosclerosis Based on Three-Dimensional Ultrasound Imaging

      2023, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    View all citing articles on Scopus
    View full text