Value of Contrast-Enhanced Ultrasonography in the Differential Diagnosis of Enlarged Lymph Nodes : a Meta-Analysis of Diagnostic Accuracy Studies

Lymph node status is one of the major predictors of prognosis in patients with cancer (Innace et al., 2010;Deng et al., 2014; Fayaz et al., 2014; Gasparri et al., 2014; Kawada et al., 2014). Furthermore, correctly diagnosing the enlarged lymph nodes in patients with or without primary tumors is essential to allow selection of an appropriate treatment strategy (Esen., 2010). A large number of modalities may be used to characterize lymph nodes, such as computed tomography, magnetic resonance imaging and gray scale ultrasound; these depend mainly on morphological characteristics for the identification of enlarged lymph nodes (Schröder et al., 2002; Riegger et al., 2012). Gray scale ultrasound combined with color Doppler ultrasonography can be applied to estimate the shape (L/T ratio), margins, internal structure and vascularization of lymph nodes (Ahuja et al., 2002; Stramare et al., 2004). Alternative diagnostic modalities include ultrasonography-guided fine needle aspiration


Introduction
Lymph node status is one of the major predictors of prognosis in patients with cancer (Innace et al., 2010;Deng et al., 2014;Fayaz et al., 2014;Gasparri et al., 2014;Kawada et al., 2014).Furthermore, correctly diagnosing the enlarged lymph nodes in patients with or without primary tumors is essential to allow selection of an appropriate treatment strategy (Esen., 2010).A large number of modalities may be used to characterize lymph nodes, such as computed tomography, magnetic resonance imaging and gray scale ultrasound; these depend mainly on morphological characteristics for the identification of enlarged lymph nodes (Schröder et al., 2002;Riegger et al., 2012).Gray scale ultrasound combined with color Doppler ultrasonography can be applied to estimate the shape (L/T ratio), margins, internal structure and vascularization of lymph nodes (Ahuja et al., 2002;Stramare et al., 2004).Alternative diagnostic modalities include ultrasonography-guided fine needle aspiration

Value of Contrast-Enhanced Ultrasonography in the Differential Diagnosis of Enlarged Lymph Nodes: a Meta-Analysis of Diagnostic Accuracy Studies
Ya Jin 1 , Yu-Shuang He 1 , Ming-Ming Zhang 2 , Shyam Sundar Parajuly 3 , Shuang Chen 1 , Hai-Na Zhao 1 , Yu-Lan Peng 1 * biopsy (US-FNAB).However, each of these various methods has its own limitations in the clinical diagnosis of lymph nodes, meriting the development of improved techniques.How to improve the diagnostic accuracy of enlarged lymph nodes and reduce unnecessary puncture of benign lymph nodes remains a challenge.
Recent advances in ultrasound technology, including commercially available ultrasonographic contrast agents (Levovist and SonoVue), contrast-specific ultrasonographic modes, and quantitative software (Qontraxt (Rubaltelli et al., 2007) and TIC analysis (Steppan et al., 2010)), have improved the accuracy of ultrasonography in the diagnosis of lymphadenopathy.This has particularly been the case for contrast-enhanced ultrasonography-guided fine needle aspiration cytology, which can improve the puncture success rate (Sun et al., 2012;Karina et al., 2013).
Nowadays, more and more studies are focusing on the use of contrast-enhanced ultrasonography (CEUS) in the differential diagnosis of benign and malignant lymph nodes (including cervical, axillary, inguinal, mediastinal and abdominal lymph nodes) and our meta-analysis provides summaries of the results of relevant studies, estimates of the average diagnostic accuracy of CEUS, the uncertainty of this average, and the variability of the study findings around the estimates.

Materials and Methods
The methodology used in this meta-analysis based on the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy, version 1.0 (Deeks et al., 2010).

Data sources and search strategies
We carried out a systematic literature search of Pubmed, Embase, Cochrane Library Central and SCI databases, with the last search undertaken on September 1, 2014.The search terms used were: "Contrast media OR Contrast-enhanced OR Microbubble", "Ultrasonography OR Ultrasound" and "Lymph nodes".These keywords were identified in the medical subject heading, title or abstract.To identify additional relevant studies, the literature search was also performed manually.

Selection of studies
All titles and abstracts of the retrieved studies were screened independently by two reviewers.Duplicates, reviews, letters, comments, case reports, and articles reporting other diseases, other diagnostic techniques or other types of result were excluded.The corresponding author was contacted by email with a request for the full text when this could not be obtained online; if the full text or original data was not provided, the study was excluded from our analysis.The remaining studies were considered potentially eligible for inclusion, and their full text was retrieved.

Inclusion and exclusion criteria
All potentially eligible articles were assessed independently by two reviewers, using predefined inclusion and exclusion criteria.Discrepancies were resolved by consensus.If no consensus could be reached, a third reviewer was consulted.
The inclusion criteria were as follows: (a) original full paper publication; (b) human study; (c) evaluation of CEUS for the differentiation of benign and malignant lymph nodes; (d) inclusion of at least 20 lesions; (e) published in English; (f) included an accepted reference method, using specimens obtained from surgery or lymph node biopsy; (g) reported data that allowed construction of 2×2 contingency tables and calculation of the true-positive (TP), false-positive (FP), true-negative (TN) and falsenegative (FN) rates for the use of CEUS for the diagnosis of benign and malignant lymph nodes.
The exclusion criteria were as follows: (a) no evaluation of the value of CEUS for the differential diagnosis of benign and malignant lymph nodes; (b) no relevant data on the sensitivity and specificity, or the number of TNs, TPs, FNs and FPs; (c)

Data extraction
For each study, the following information was extracted: (a) author, publication year, the type and dose of contrast agent, the type of scanners, imaging modality; (b) participant characteristics (age, sex, the number of patients and lesions and the number of lesions histologically proven to be malignant); (c) statistics for the meta-analysis: TP, FP, TN, FN, sensitivity and specificity.

Quality assessment
Methodological quality was assessed according to the revised tool for Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) (Smidt et al., 2008;Deeks et al., 2010;Whiting et al., 2011).The full QUADAS-2 tool consists of four domains: patient selection, index test, reference standard, and flow and timing.Each domain was assessed in terms of the risk of bias according to the signaling questions, and the first three domains were judged in terms of concerns regarding applicability.Each question was scored "yes" if reported, "no" if not reported, or "unclear" if there was inadequate information in the article to make a judgment.

Data synthesis and statistical analysis
To assess data inhomogeneity, a random-effects model was applied to control for differences in the reported data (e.g., patient characteristics and methods used).This represents a classic, non-iterative method to account for inter-study heterogeneity.χ 2 and I 2 statistics were computed.I 2 values were interpreted according to the proposal of Higgins and Thompson (Higgins et al., 2002), with heterogeneity determined as either low (I 2 ≤25%), medium (25% <I 2 ≤50%) or high (50% <I 2 ≤75%).Factors influencing diagnostic accuracy were assessed by means of formal meta-regression analysis (Lijmer et al., 2002).The parameters listed in the Data Extraction and Quality Assessment sections (see above) were used as covariates.p<0.05was considered to indicate a significant difference.
Summary sensitivities, specificities and the summary receiver operating characteristic curve with corresponding 95% confidence intervals (CIs) were used to examine the accuracy of CEUS in the differential diagnosis of enlarged lymph nodes.Data synthesis was performed within the bivariate mixed-effects binary regression modeling framework (Reitsma et al., 2005).It is possible that the accuracy of the following clinical subgroups could differ, and therefore act as potential source of heterogeneity: (1) with a primary tumor versus without; (2) imaging modality: low MI with SonoVue.We evaluated subgroups according to quality assessment and data extraction.
In addition, a bivariate box plot was used to assess the distributional properties of sensitivity versus specificity and for identifying possible outliers.After omitting these outliers and according to the results of the subgroups analysis, sensitivity analysis was performed and the change in heterogeneity was observed.The Deeks' funnel plot asymmetry test was also used to investigate whether all the studies were from a single population, and to search for publication bias (Deeks et al., 2005).
The QUADAS figure was drawn using Revman 5.0 (Cochrane Collaboration).All statistical analyses were performed using the MIDAS and METANDI modules in Stata 12.0 (Stata Corp, Texas, USA).

Assessment of study quality
According to the QUADAS scale, the methodological quality was rated as "not good" for all the included studies.Only one study fulfilled over 10 items (Xue et al., 2011); three studies met seven items (Rubaltelli et al., 2007;Wang et al., 2009;Poanta et al., 2014), five met eight (Moritz et al., 2000;Schmid-Wendtner et al., 2002;Kanamori et al., 2006;De Giorgi et al., 2010;Podkrajsek et al., 2011), and the remaining studies met nine.Our assessment of methodological quality is summarized in Figure 2.
The pooled estimates for sensitivity, specificity, PLR and NLR in the subgroups were presented in Table 2.

Sensitivity analysis of the factors influencing the diagnostic performance of CEUS
According to the subgroups analysis, we excluded the studies of power and color Doppler modes that were used with Levovist and evaluated the diagnostic accuracy of CEUS using low MI and SonoVue (Figure 5).A bivariate boxplot (Figure 6) showed that three studies (Moritz et al., 2000;Hocke et al., 2008;Slaisova et al., 2013) were heterogeneous with respect to the other studies.After omitting the three studies, sensitivity analysis was performed: the heterogeneity (I 2 ) of sensitivity decreased from 85.68% (95%CI, 79.68-91.68) to 60.92% (95%CI, 37.28-84.59)and from 91.61% (95%CI, 88.6-94.63) to 85.21% (95%CI, 78.24-92.19) of specificity.The results of the summary estimates were presented in Table 3.

Assessment of publication bias
To address publication bias, a funnel plot was constructed of log DOR against the standard error of the estimate of log DOR.According to Deeks' funnel plot asymmetry test, there was no publication bias among the included studies (p=0.15; Figure 7).

Discussion
The prevalence of enlarged lymph nodes is quite high.In past years, B-mode and Color-Doppler sonography were usually used as a first-line procedure to differentiate benign and malignant lymph nodes.There is no single criterion or even a combination of criteria that is sensitive or specific enough to diagnose malignancy; therefore, fine-needle aspiration is currently the primary diagnostic procedure.However, it has been shown to be a costeffective method, and is limited by sampling difficulties and reliable fine-needle aspiration is dependent on the sampler's experience and the cytologist's expertise (Fatima et al., 2011).
Our meta-analysis has investigated the diagnostic performance of CEUS in differentiating between benign and malignant lymph nodes.Analysis of 1563 lesions demonstrated consistently high pooled sensitivity, specificity, positive LR and DOR, but it also showed a lower negative LR.According to these results, CEUS can be used in clinical practice as an excellent diagnostic tool for diagnosis of malignant lymph nodes.This finding is of great importance because of the limitations of current approaches to identification of lymph nodes.
Subgroup analysis in this research showed that the sensitivity of CEUS seemed to be lower in patients without a primary tumor, indicating that the capability of CEUS for recognizing malignant lymph nodes in patients with a primary tumor is higher than in patients without.Furthermore, subgroup analysis revealed that the I 2 value of the subgroup with a primary tumor was much lower than that of the subgroup without a primary tumor, implying that the homogeneity of the studies included in the subgroup with a primary tumor is better.Overall, our analysis suggests that CEUS shows promise as a screening method in clinical practice for use in patients with enlarged lymph nodes, especially those with a primary tumor.
In the subgroup analysis, the sensitivity, specificity, positive likelihood ratio of the Doppler mode subgroup were higher than the subgroup of low MI.This result could be elaborated by the fewer number of studies of Doppler mode and because the methodology of CUES has changed dramatically since its introduction in the late 90s, the earlier high MI and Doppler mode used with Levovist are only of historical interest.We excluded the studies of power and color Doppler modes that were used with Levovist and evaluated the diagnostic accuracy of CEUS using low MI and SonoVue, the sensitivity and specificity are 90% and 81%, respectively and this result may reveal the real clinical value.
The major limitation of this meta-analysis was the extent of the observed heterogeneity.We used a random effects approach to analyze the heterogeneous data, and a bivariate box plot was used to identify possible outliers.In the evaluation of the accuracy of CEUS, excluding three outliers reduced the heterogeneity without substantially changing the summary estimates.Possible reasons underlying the differences between the three outliers and the other studies include higher malignant lymph nodes rates, different doses of contrast agent, and the lymph node site.
In addition, it is plausible that part of the heterogeneity was caused by a large variation in the qualitative and quantitative criteria used for determining the stage of lymph nodes.Among the included studies, fifteen used qualitative criteria, and one used both qualitative and quantitative criteria, as follows: (1) intense homogeneous enhancement with no perfusion defects for benign lymph nodes, and intense inhomogeneous enhancement in the arterial phase with perfusion defects or hypo-enhancement for malignant lymph nodes;(2) in the color code image, vessels have a regular appearance with both venous and arterial vessels visible for benign lymph nodes, and an irregular appearance with only arterial vessels visible for malignant lymph nodes ;(3) predominantly hilar vessels for benign lymph nodes, and predominantly peripheral vessels or an absence of vessels for malignant lymph nodes;(4) differentiation between benign and malignant lymph nodes made on the basis of the rapidity of enhancement, intensity of enhancement (5) quantitative criteria: analysis of the time-intensity curve.Since more than three studies are needed for a meta-analysis using Stata, subgroup analysis could not be performed to evaluate the accuracy of quantitative CEUS and CEUS used with the Doppler mode and the contrast agent of Levovist.This may be another reason for the significant heterogeneity in our analysis.Additional sources of heterogeneity may include the sex proportion, the mean lymph node size, the primary tumor of the included patients, the choice of imaging modality, the design of the procedure, and QUADAS score.Nonetheless, the heterogeneity in this type of diagnostic study remains a concern, and to some extent it influences the certainty of the conclusions.
In conclusion, our meta-analysis showed that CEUS has the potential to be used as a valuable examination, with high sensitivity and specificity, to help characterize lymph nodes.However, the clinical value of CEUS needs further examination.Future large-scale studies, the development of the analytical software (like Stata, Metadisc and Revman), particularly with regard to the accuracy of quantitative CEUS for evaluating lymph nodes, are required to evaluate the screening improvement that has been hypothesized.

Figure 1 .
Figure 1.The Flow Chart of Literature Screening

Figure
Figure 5. Forest Plots Showing the Sensitivity and Specificity of CEUS Used With Low Mi and Sonovue for the Differentiating Between Benign and Malignant Lymph Nodes DOI:http://dx.doi.org/10.7314/APJCP.2015.16.6.2361Contrast-Enhanced US in Differential Diagnosis of Enlarged Lymph Nodes: A Meta-Analysis of Accuracy

Table 1 . Main Characteristics of the Studies Included Evaluating the Performance of CEUS in the Differential Diagnosis of Benign and Malignant Lymph Nodes
.doi.org/10.7314/APJCP.2015.16.6.2361Contrast-Enhanced US in Differential Diagnosis of Enlarged Lymph Nodes: A Meta-Analysis of Accuracy *Sen: sensitivity; Spe: specificity; DCUS: double contrast-enhanced ultrasonography, an oral ultrasonic contrast agent combined with an intravenous contrast agent; CE-HUS: contrast-enhanced harmonic ultrasonography; CE-EUS: contrast-enhanced endoscopic ultrasonography DOI:http://dx

Table 2 . Summary Estimates for Each Subgroup
*Sono Vue: the type of contrast agent; MI: Mechanical Index

Table 3 . Sensitivity Analysis
*DOR, diagnostic odds ratio; AUROC, area under the summary receiver operating characteristics curves; NLR, negative likelihood ratio; PLR, positive likelihood ratio . Predicting lymph node status in patients early gastric carcinoma using double contrast-enhanced ultrasonography.Arch Med Sci, 7, 457-64.Yu M, Liu Q, Song HP, et al Clinical application of contrast-enhanced ultrasonography in diagnosis of superficial lymphadenopathy.J Ultrasound Med, 29, 735-40.Zheng Z, Yu Y, Lu M, et al (2011).Double contrast-enhanced ultrasonography for the preoperative evaluation of gastric cancer: a comparison to endoscopic ultrasonography with respect to histopathology.Am J Surg, 202, 605-11.