Background
Intraplaque neovascularization is considered an important indicator of plaque vulnerability. Contrast-enhanced ultrasound (CEUS) of carotid arteries improves imaging of carotid intima-media thickness and permits real-time visualization of neovascularization of the atherosclerotic plaque. The authors conducted a systematic review and meta-analysis to evaluate the accuracy of CEUS-detected carotid atherosclerotic plaque.
Methods
A systematic search was performed to identify studies published in the MEDLINE, Embase, Scopus, and Web of Science databases from 2004 to June 2015. Studies evaluating the accuracy of quantitative analysis and qualitative analysis (visual interpretation) for the diagnosis of intraplaque neovascularization compared with histologic specimens and/or clinical diagnosis of symptomatic plaque were included. Parameters evaluated were plaque quantitative CEUS intensity and the visual grading of plaque CEUS. A random-effects meta-analysis was used to pool the likelihood ratios (LRs), diagnostic odds ratios, and summary receiver operating characteristic curves. Corresponding areas under the curves were calculated.
Results
The literature search identified 203 studies, 20 of which were selected for systematic review; the final meta-analysis included seven studies. For qualitative CEUS, pooled sensitivity was 0.80 (95% CI, 0.72–0.87), pooled specificity was 0.83 (95% CI, 0.76–0.89), the pooled positive LR was 3.22 (95% CI, 1.67–6.18), the pooled negative LR was 0.24 (95% CI, 0.09–0.64), the pooled diagnostic odds ratio was 15.57 (95% CI, 4.94–49.03), and area under the curve was 0.894. For quantitative CEUS, pooled sensitivity was 0.77 (95% CI, 0.71–0.83), pooled specificity was 0.68 (95% CI, 0.62–0.73), the pooled positive LR was 2.34 (95% CI, 1.69–3.23), the pooled negative LR was 0.34 (95% CI, 0.25–0.47), the pooled diagnostic odds ratio was 7.06 (95% CI, 3.6–13.82), and area under the curve was 0.888.
Conclusions
CEUS is a promising noninvasive diagnostic modality for detecting intraplaque neovascularization. Standardization of quantitative analysis and visual grading classification is needed to increase reliability and reduce technical heterogeneity.
A number of molecular changes, including inflammation, angiogenesis, lipid accumulation, and thrombosis, may occur at sites of atherosclerotic plaque development, resulting in further physiologic progression and, ultimately, rupture. Although the precise mechanism of remodeling of an asymptomatic plaque into a vulnerable plaque remains unknown, neovascularization within the atherosclerotic plaque is considered an important factor in plaque development and vulnerability to rupture. These changes in the artery wall are not specific to any particular arterial bed; rather, they are seen in systemic peripheral vascular disorders. The carotid artery is an accessible window for assessing characteristics of atherosclerotic plaque. B-mode and Doppler ultrasound are most commonly used in daily clinical practice to find carotid atherosclerotic plaques and to evaluate carotid stenoses resulting from plaque. Feinstein first reported the feasibility of using ultrasound contrast agents for identifying carotid intraplaque neovascularization (IPN). Recently, multiple studies involving carotid contrast-enhanced ultrasound (CEUS) have shown that echocardiographic contrast agents composed of microbubbles can be used to detect IPN by evaluating the increase of microbubble signal intensity in the plaque, which can be visually scored. We pooled available evidence on the diagnostic methodology relevant to CEUS and its prognostic value in a systematic review and performed a meta-analysis to assess the diagnostic accuracy of CEUS.
Methods
Our systematic review and meta-analysis adhered to standards for meta-analysis of observational studies in epidemiology, and diagnostic accuracy initiatives guidelines for conducting systematic reviews of diagnostic studies were followed.
Eligibility Criteria
We included peer-reviewed studies that assessed the diagnostic accuracy of quantitative and qualitative CEUS in the evaluation of IPN, with CEUS serving as the index test. We selected human studies that included plaque histology or clinically diagnosed symptomatic plaque (patients had symptoms of stroke, transient ischemic attack, or amaurosis fugax clinically related to the index plaque neurovascular territory <12 months before entry into the study), or both, as the reference test. No language restriction was applied.
Literature Search Strategy
The literature search was performed by an expert librarian (A.M.F.) in MEDLINE, Embase, the Cochrane Database of Systematic Reviews, Web of Science, Scopus, DARE, and CENTRAL (from database inception to June 2015). The search terms used were the following: “carotid,” “contrast-enhanced carotid ultrasound,” and “intraplaque neovascularization.” All titles and abstracts were evaluated independently by two reviewers (R.H. and C.A.B.) for relevance and the inclusion criterion of having quantitative or visual grading CEUS for the detection of IPN. Animal studies were excluded. Discordant assessments between reviewers were resolved through direct discussion, and if uncertainty remained, oversight was provided by a third reviewer (S.S.A.), who reviewed and discussed the discordant titles and abstracts. After exclusion on the basis of titles and abstracts, full articles were evaluated. Articles meeting the inclusion criteria were identified. A manual search of the reference lists of the identified studies was reviewed for additional citations.
Data Collection
A standardized data extraction sheet was developed. The extracted data included the characteristics of the study population, the number of included patients, the mean age of patients, sex distribution, clinical cardiovascular risk factors, history of coronary artery disease (CAD), and CEUS methodologic details. We extracted parameters for identifying IPN by CEUS. There are two ways to assess IPN: (1) measurement of the change of signal intensity in the plaque before and after the administration of a contrast agent to determine the ratio of enhanced intensity in the plaque to that in the lumen of the carotid artery and (2) interpretation of enhancement in plaques visually with a three-grade scale, where grade 0 represents no enhancement within the plaque, grade 1 represents limited appearance of bubbles confined to the adventitial side of the plaque and/or the shoulder, and grade 2 represents extensive microbubble presence, seen moving into the plaque core. The absolute numbers of true positives, false positives, true negatives, and false negatives were extracted from the data. We calculated sensitivity as true positives/(true positives + false negatives) and specificity as true negatives/(true negatives + false positives). For each candidate study, two reviewers (R.H. and C.A.B.) independently assessed methodologic quality by using the Quality Assessment of Diagnostic Accuracy Studies tool. The tool includes 14 questions in three domains (generalizability, clarity, and validity). Disagreements were resolved by consensus.
Outcome of Interest
Our primary end point was the diagnostic odds ratio (DOR) for the diagnostic accuracy of CEUS in identifying IPN. Other end points included likelihood ratios (LRs) and summary receiver operating characteristic curves. The LR is defined as the probability of a given test result for patients with a disease divided by the probability of the same result for patients without the disease. The positive LR (LR+) equals sensitivity/(1 − specificity), and the negative LR (LR−) equals (1 − sensitivity)/specificity. The DOR describes the odds of a positive test result in a patient with a disease compared with the odds in a patient without the disease and equals LR+/LR−. For this analysis, the disease in question was IPN.
Statistical Analysis
For those articles reporting the sensitivity and specificity of CEUS, we performed a meta-analysis to evaluate the accuracy of CEUS as validated by histology, clinical diagnosis, or both. Those articles that did not report sensitivity and specificity were included only in the systematic review.
A priori, we decided that the included studies would be heterogeneous and chose the random-effects model for all pooled analysis. The inverse variance method was used to pool LRs and DORs. Moses linear models were used to draw summary receiver operating characteristic curves, with each study represented as a point on the graph, and to calculate the corresponding area under the curve. In these analyses, we inspected the data for inconsistency in results across studies and used the I 2 statistic to assess inconsistency. The I 2 statistic estimates the percentage of variability in results across studies due to true differences in patients, tests, outcomes, and design rather than due to chance, with values of 25%, 50%, and 75% indicating low, moderate, and high inconsistency, respectively. Analysis was done using Meta-DiSc version 1.4 (Ramony Cajal Hospital, Madrid, Spain). Subgroup analyses were planned a priori to examine the effect of methodologic diversity on the potential heterogeneity among studies, which included reference test used, patient selection standard, mechanical index, type of contrast agent used, and quantitative or qualitative analysis method.
Results
Study Selection and Characteristics
Two hundred three studies were identified in the systematic search, 20 of which met the inclusion criteria for the systematic review ( Figure 1 ). Agreement between the two reviewers in the final selection of studies was 97%, with a κ of 0.89 (standard error, 0.056; 95% CI, 0.776–0.996). The study population consisted of 1,960 patients (mean age, 67 years). Four hundred seventy-one patients (24%) had carotid stenosis≥50%. Patient demographic data are reported in Table 1 .
| Study | Mean age (y) | Population | Men | Patient type | DM | HTN | Smoking | Dyslipidemia | TIA/stroke | CAD |
|---|---|---|---|---|---|---|---|---|---|---|
| Shah et al . (2007) | NA | 17 | 10 (59%) | Carotid stenosis >50% | 7 (41) | 13 (76) | NA | NA | 5 (29) | NA |
| Huang et al . (2008) | 59 | 63 | 42 (67%) | Patients with plaques | NA | NA | NA | NA | NA | NA |
| Coli et al . (2008) | 69 | 32 | 27 (84%) | Carotid stenosis >30% | 11 (34) | 24 (75) | 9 (28) | 18 (56) | 4 (13) | 14 (44) |
| Xiong et al . (2009) | 63 | 104 | 83 (80%) | Symptoms of cerebrovascular disease, plaque thickness > 2 mm | 35 (34) | 77 (74) | 38 (37) | NA | 35 (34) | NA |
| Giannoni et al . (2009) | 67 | 77 | 51 (66%) | Carotid stenosis >50% | NA | NA | NA | NA | 9 (12) | NA |
| Owen et al . (2010) | 70 | 37 | 27 (73%) | Carotid stenosis >30% | 6 (16) | 29 (78) | 24 (65) | NA | 16 (43) | NA |
| Staub et al . (2010) | 64 | 147 | 89 (61%) | Patients with plaques | 45 (31) | 100 (68) | 68 (46) | NA | 17 (12) | 77 (52) |
| Huang et al . (2010) | 62 | 176 | 108 (61%) | Patient with ischemic stroke | 13 (7) | NA | 56 (32) | NA | 81 (46) | NA |
| Faggioli et al . (2010) | 72 | 22 | 15 (86%) | Carotid stenosis >70% | 8 (36) | 21 (95) | 5 (23) | 9 (41) | 7 (32) | 6 (27) |
| Hoogi et al . (2011) | 68 | 27 | 19 (70%) | Carotid stenosis >70% | 7 (26) | 18 (67) | NA | 17 (63) | 6 (22) | 14 (52) |
| Shalhoub et al . (2011) | 71 | 31 | 24 (77%) | Carotid stenosis >50% | NA | NA | NA | NA | 16 (52) | NA |
| Staub et al . (2011) | 67 | 175 | 113 (65%) | Patients with one plaque >0.5 mm | 48 (39) | 93 (75) | 64 (52) | NA | 20 (11) | 78 (63) |
| Varetto et al . (2012) | 73 | 51 | 38 (70%) | Carotid stenosis >70% | 15 (30) | 41 (80) | 26 (50) | 32 (62) | 12 (24) | NA |
| Müller et al . (2013) | NA | 33 | 24 (72%) | Carotid stenosis >50% | 16 (48) | 30 (91) | 13 (39) | 20 (61) | 17 (52) | NA |
| Deyama et al . (2013) | 69 | 304 | 245 (80%) | CAD patients with carotid plaque | 138 (45) | 153 (50) | 115 (38) | NA | 25 (8) | 101 (33) |
| Zhu et al . (2013) | 63 | 312 | 228 (73%) | Stable CAD and atherosclerotic CVD | 83 (27) | 230 (74) | 123 (39) | NA | NA | NA |
| Zhou et al . (2013) | 63 | 46 | 43 (73%) | Carotid stenosis >50% | 12 (26) | 22 (48) | 27 (59) | 27 (59) | 24 (52) | NA |
| Saito et al . (2014) | 72 | 50 | 49 (98%) | Carotid stenosis >50% | 14 (28) | 43 (86) | 34 (68) | 37 (74) | 31 (62) | NA |
| Iezzi et al . (2015) | 70 | 50 | 28(56%) | Carotid stenosis >70% | NA | NA | NA | NA | NA | NA |
| Nakamitsu et al . (2015) | 69 | 206 | 159(77%) | Patient with stable CAD with carotid stenosis <50% | 80 (39) | 148 (72) | 125 (61) | 152 (74) | 15 (7.3) | 206 |
| Summary | 67 | 1,960 | 1,422 (73%) | 538 (27) | 1,042 (53) | 727 (37) | 312 (16) | 340 (17) | 496 (25) |
Meta-Analysis
Among these 20 reports, sensitivity and specificity could be extracted from seven, all of which were included in the final meta-analysis. Four hundred sixty-one patients were included, 188 of whom were symptomatic. Reference tests were histology in three studies and clinical diagnosis of symptomatic plaque in four studies. IPN was identified by histologic staining for special vascular (CD31, CD34, von Willebrand factor, and hemosiderin), inflammation, and angiogenic (vascular endothelial growth factor, CD68, and mast cell tryptase) markers. In total, 140 patients had histologic results; of these, 32 were also symptomatic patients, of whom 28 (88%) had histology-proven IPN. A summary of methodologic differences among the studies included in the meta-analysis is presented in Table 2 .
| Study | Machine, probe | MI | Contrast agent | Contrast agent dose per bolus (mL) | Analysis software | Reference test | Analysis method | Cutoff value for diagnosis IPN |
|---|---|---|---|---|---|---|---|---|
| Shah et al . (2007) | GE Vivid 7, L7; ATL HDI 5000, L7-4 | 0.06–0.1 | Optison | 0.5–1.0 | NA | Histology | Quality | NA |
| Xiong et al . (2009) | GE Logiq 9, L9 | 0.13 | SonoVue | 1.5 | TIC | Symptomatic | Quantity | EI ∗ , ratio 10 dB, 0.46 |
| Giannoni et al . (2008) | Acuson Sequoia 512, L15-8 | 0.4–1.4 | SonoVue | 2.5 | NA | Histology | Quality | NA |
| Owen et al . (2009) | Philips iU22, L12-5 | 0.34 | SonoVue | 2 | TIC | Symptomatic | Quantity | LP † contrast enhancement > 0 |
| Huang et al . (2010) | Acuson Sequoia 512, L15-8 | 0.35 | SonoVue | 2.4 | TIC | Symptomatic | Quantity | EI, WT ‡ : 6.4 dB, 4.15 sec |
| Zhou et al . (2013) | Philips iU22, L9-3 | 0.07 | Definity | NA | NA | Symptomatic | Quality | NA |
| Iezzi et al . (2015) | Esaote MyLab ultrasound scanner, L8-14 | 0.13 | SonoVue | 2 or 4 | TIC | Histology | Quality | NA |
∗ Equal to peak intensity minus baseline intensity. Ratio is the ratio of peak signal intensity in the plaque to the lumen of the carotid artery.
For qualitative CEUS, pooled sensitivity was 0.80 (95% CI, 0.72–0.87), pooled specificity was 0.83 (95% CI, 0.76–0.89), the pooled LR+ was 3.22 (95% CI, 1.67–6.18), the pooled LR− was 0.24 (95% CI, 0.09–0.64), the pooled DOR was 15.57 (95% CI, 4.94–49.03), and the area under the curve was 0.894. For quantitative CEUS, pooled sensitivity was 0.77 (95% CI, 0.71–0.83), pooled specificity was 0.68 (95% CI, 0.62–0.73), the pooled LR+ was 2.34 (95% CI, 1.69–3.23), the pooled LR− was 0.34 (95% CI, 0.25–0.47), the pooled DOR was 7.06 (95% CI, 3.6–13.82), and area under the curve was 0.888 ( Figures 2–5 ).
Four studies reported quantitative evaluations of signal intensity after echocardiographic contrast agent infusion. The quantification methods of signal enhancement were different. Xiong et al . and Iezzi et al . used the ratio of enhanced videointensity in the plaque to that in the lumen of the carotid artery, Owen et al . analyzed late phase-contrast videointensity (6 min after contrast infusion) as a marker of damaged endothelium and plaque inflammation, and Huang et al . defined signal enhancement as peak videointensity minus baseline videointensity and also used a time parameter: the wash-in time, defined as time to peak minus arrival time.
Visual grading of CEUS was reported in three studies as follows: Iezzi et al . used a three-grade scale to describe contrast enhancement in the plaque: grade 1, no enhancement within the plaque; grade 2, enhancement confined to the adventitial side of the plaque and/or the shoulder; and grade 3, extensive flow of droplets of contrast media throughout the plaque core. Giannoni et al . categorized the pattern of contrast enhancement in the plaque as type I (high-intensity regions of microbubbles extending from the adventitial layers toward the vessel lumen) and type II (delayed appearance of contrast enhancement at base of the plaques). Zhou et al . used two grades to present contrast enhancement in the plaque: grade 1 represented no contrast enhancement in the plaque, while grade 2 represented contrast enhancement in the plaque.
All included studies were qualitatively evaluated using the Quality Assessment of Diagnostic Accuracy Studies tool; results are shown in Table 3 .
| Quality | QUADAS item | Study | ||||||
|---|---|---|---|---|---|---|---|---|
| Shah et al . | Xiong et al . | Giannoni et al . | Owen et al . | Huang et al . | Zhou et al . | Iezzi et al . | ||
| Generalizability | Q1 | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Clarity | Q2 | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Q8 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Q9 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Q13 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Q14 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Validity | Q3 | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Q4 | Unclear | Yes | Yes | Yes | Yes | No | Yes | |
| Q5 | Yes | Unclear | Yes | Yes | Yes | Yes | Yes | |
| Q6 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Q7 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
| Q10 | Yes | Yes | Unclear | No | No | Yes | Yes | |
| Q11 | Yes | Yes | Unclear | Yes | Yes | Yes | Yes | |
| Q12 | Yes | Yes | Yes | Yes | Yes | Yes | Yes | |
To explore other sources of heterogeneity in evaluating IPN, we conducted subgroup analyses for diagnostic accuracy end points ( Table 4 ). The heterogeneity analysis showed that lower heterogeneity was associated with carotid stenosis >50%, use of Optison or Definity as the contrast agents, use of a very low mechanical index (<0.2), and visual grading.
| Subgroup | Studies | DOR (95%CI) | Heterogeneity, I 2 ( P ) |
|---|---|---|---|
| Reference test | |||
| Histology | 3 | 20.11 (4.81–84.03) | 53.3% (.07) |
| Clinic diagnosis of symptomatic | 5 | 7.09 (4.00–12.56) | 41.4% (.15) |
| Patients | |||
| Carotid stenosis >50% | 2 | 12.30 (4.95–30.57) | 0% (.38) |
| Others | 5 | 8.53 (3.90–18.65) | 60.9 (.02) |
| Mechanical index | |||
| ≤0.2 | 4 | 7.37 (4.03–13.49) | 0% (.47) |
| >0.2 | 3 | 12.36 (3.71–41.18) | 77.5% (.004) |
| Contrast agent | |||
| Optison or Definity | 2 | 6.34 (1.87–21.47) | 0% (.76) |
| Others | 5 | 10.58 (4.99–22.46) | 65.6% (.0078) |
| Analysis | |||
| Visual grading | 4 | 15.57 (4.94–49.03) | 46.6% (.10) |
| Enhancement of intensity | 3 | 7.06 (3.60–13.82) | 56% (.08) |
Systematic Review
The 13 studies meeting the initial inclusion criteria were not included in the final meta-analysis, because we could not obtain sensitivity and specificity from these reports. These studies were thus included in a systematic review. Summary data for these studies are shown in Table 5 . Seven of these reports, including 181 patients, provided histologic validation of IPN identified on CEUS from carotid endarterectomy. Plaque with a higher amount of contrast enhancement had significantly increased density of small-diameter microvessels in the corresponding region on histology. Histologic staining for specific inflammatory cellular markers (CD31 and CD68) showed a correlation between intraplaque contrast enhancement and the amount of stain uptake. Echolucent plaque was likely to have a higher grade of contrast enhancement in the plaque, but both plaque echogenicity and the degree of plaque stenosis had a weak correlation with the grade of IPN ( r = −0.199 and r = 0.157, respectively). Zhou et al .’s research demonstrated that among patients with carotid artery stenosis >50%, carotid plaques with IPN were more likely to have ipsilateral transcranial color Doppler signals (thought to be secondary to microembolic phenomenon) than non-IPN plaques. Faggioli et al . also found that the presence of an ipsilateral embolic lesion on computed tomography was significantly correlated with increased neovascularization. Two retrospective studies investigated the relationship between IPN, the presence of cardiovascular disease (CVD), and the occurrence of cardiovascular events (CVEs). Staub et al . showed that presence of IPN was not associated with cardiovascular risk factors. The prevalence of CVD was similar in subjects with and without carotid IPN (65% vs 77%, P = .208), whereas subjects with IPN had a higher incidence of prior CVEs (38% vs 20%, P = .031). Multivariate logistic regression analysis showed that IPN was an independent risk factor associated with history of CVE (odds ratio, 4.0; 95% CI, 1.3–12.6; P < .05). Deyama et al . demonstrated that the grade of IPN was significant correlated with coronary extent score, as defined by the number of complex coronary lesions and the number of diseased coronary arteries. High-grade IPN was a risk factor for acute coronary syndrome, independent of traditional risk factors (odds ratio, 1.91; 95% CI, 1.04–3.53; P < .01). Zhu et al .’s prospective follow-up study showed that the presence of intraplaque microbubble contrast enhancement was more common in patients with acute coronary syndrome than those with stable CAD and is a significant independent predictor of coronary events in patients with stable CAD (odds ratio, 3.90; 95% CI, 1.60–9.46; P = .003). Nakamura et al . followed patients with stable CAD who had plaques for 3 years, in whom a multivariate Cox model showed that enhanced intensity of plaque was a significant predictor of cardiac events independent of traditional risk factors (hazard ratio, 1.13; 95% CI, 1.05–1.21; P < .001). Six studies evaluated inter- and intraobserver agreement for quantitative and visual grading methods, and both methods indicated good agreement.
