Comparative Diagnostic Accuracy of Contrast-Enhanced Ultrasound and Shear Wave Elastography in Differentiating Benign and Malignant Lesions: A Network Meta-Analysis

Background: We performed a network meta-analysis to compare the diagnostic accuracy of contrast-enhanced ultrasound (CEUS) and shear wave elastography (SWE) in differentiating benign and malignant lesions in different body sites. Methods: A computerized literature search of Medline, Embase, SCOPUS, and Web of Science was performed using relevant keywords. Following data extraction, we calculated sensitivity, specificity, positive likelihood ratio (LR), negative LR, and diagnostic odds ratio (DOR) for CEUS, and SWE compared to histopathology as a reference standard. Statistical analyses were conducted by MetaDiSc (version 1.4) and R software (version 3.4.3). Results: One hundred and fourteen studies (15,926 patients) were pooled in the final analyses. Network meta-analysis showed that CEUS had significantly higher DOR than SWE (DOR = 27.14, 95%CI [2.30, 51.97]) in breast cancer detection. However, there were no significant differences between CEUS and SWE in hepatic (DOR = −6.67, 95%CI [−15.08, 1.74]) and thyroid cancer detection (DOR = 3.79, 95%CI [−3.10, 10.68]). Interestingly, ranking analysis showed that CEUS achieved higher DOR in detecting breast and thyroid cancer, while SWE achieved higher DOR in detecting hepatic cancer. The overall DOR for CEUS in detecting renal cancer was 53.44, 95%CI [29.89, 95.56] with an AUROC of 0.95, while the overall DOR for SWE in detecting prostate cancer was 25.35, 95%CI [7.15, 89.89] with an AUROC of 0.89. Conclusion: Both diagnostic tests showed relatively high sensitivity and specificity in detecting malignant tumors in different organs. Network meta-analysis showed that CEUS had higher diagnostic accuracy than SWE in detecting breast and thyroid cancer, while SWE had higher accuracy in detecting hepatic cancer. However, the results were not statistically significant in hepatic and thyroid malignancies. Further head-to-head comparisons are needed to confirm the optimal imaging technique to differentiate each cancer type.


INTRODUCTION
Ultrasound (US) has been used for decades in differentiating benign and malignant lesions because of its low cost, ease of access, and non-invasiveness. For example, it belongs to the triad (physical examination, mammography and US), commonly used to assess the risk of breast cancer (1). Moreover, it can detect thyroid nodules as small as 2 mm in size and predicts malignancy based on features like irregular border, hypo-echogenicity, and calcification (2,3). However, none of these features can individually predict malignancy and conventional US alone has shown moderate accuracy in detecting malignant lesions (4). Therefore, improvements to US technique have been sought.
The introduction of contrast agents (contrast-enhanced US/CEUS) allows for visibility of blood flow within the lesion, which improves its characterization (5). The current in-use contrast media are second-generation agents as SonoVue. These agents remain within the intravascular space, which increases their safety and allows for continuous imaging over the enhancement period (6). Several studies have reported high sensitivity and specificity for CEUS in differentiating malignant lesions with the breast, thyroid, liver and kidneys (5,(7)(8)(9). A recent meta-analysis showed no significant difference between CEUS and contrast-enhanced computed tomography (CECT) and magnetic resonance imaging (CEMRI) in terms of the diagnostic accuracy in characterizing focal liver lesions (FLLs) (8).
Shear wave elastography (SWE) relies on the degree of lesion stiffness when subjected to external pressure. Malignant nodules have harder consistency (less elasticity) than benign ones due to the uncontrolled proliferation of cancer cells (10). Therefore, SWE has been investigated for differentiating benign and malignant nodules. Compared to conventional US, SWE is more quantitative and is less operator-dependent, allowing more effective detection of malignant tumors (11). Recent diagnostic test accuracy (DTA) studies and metaanalyses showed high sensitivity and specificity for SWE in detecting malignant lesions within the breast and hepatic tissues (11)(12)(13).
According to our knowledge, data are lacking on the direct comparison between CEUS and SWE; therefore, we performed a meta-analysis to evaluate the diagnostic accuracy of CEUS and SWE in differentiating malignant tumors in the breast, liver, thyroid, kidneys, and prostate tissues in comparison to histopathology as a reference test. Moreover, we used network meta-analysis (NMA) to compare the diagnostic accuracy of both tests in malignant tumor differentiation.

MATERIALS AND METHODS
This meta-analysis has been conducted and reported in accordance with the Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies (The PRISMA-DTA Statement) (14); Supplementary File I.

Literature Search
We searched Medline (via PubMed), Embase, SCOPUS and Web of Science for diagnostic accuracy studies that evaluated the use of CEUS and SWE in the differentiation of malignant tumors in different body organs. The following search terms were used with different combinations in different databases: Contrast-enhanced Ultrasound OR CEUS OR Ultrasound OR SonoVue OR Shear Wave Elastography OR SWE OR Sonoelastography OR Elastosonography AND Malignant OR Cancer OR Tumor OR Benign OR Adenoma OR Adenocarcinoma OR Carcinoma OR Nodule. No search filters of any sort were used during the search. All retrieved search results from database search (including bibliographic data and abstracts) were imported into EndNote (X7) for duplicate removal and then were transferred to a Microsoft Excel Sheet for screening.

Study Screening
For a study to be eligible for inclusion, it must have matched all the following criteria: (1) Population: Patients, suspected or diagnosed with malignancy in any body organ, (2) Intervention: CEUS or SWE [no specifications by US system or probe type], (3) Comparator: Histopathology, (4) Outcomes: Sensitivity, specificity, positive predictive value [PPV], and negative predictive value [NPV], and (5) Study type: Diagnostic accuracy study. Two independent authors reviewed the title and abstract of retrieved records against our eligibility criteria and classified them into: eligible, non-eligible, or requires further screening (seems to fit the inclusion criteria, but further confirmation is required). The full-text articles of the latter type were retrieved and underwent a second wave of screening. Any discrepancy between the two reviewers' decisions was solved by a senior reviewer (with a 15-year experience in secondary analysis and evidence synthesis methods) after reviewing the debated studies in reference to the pre-specified PICO criteria.

Data Extraction and Quality Assessment
An extraction sheet (in Microsoft Excel) was formatted and pilot-tested before final extraction. The sheet was customized to extract the baseline data of the imaging device, enrolled patients, as well as the raw diagnostic data of each included study. For pilot testing, two reviewers extracted these data from 5 included studies and the datasets were matched and compared with the original studies by a third reviewer. Each set of data was extracted by two reviewers and discordant decisions were resolved by discussion. These discussions included reexamining the studies, inspecting their available additional data sources and re-evaluating the former decisions. When the discrepancies remained, a senior reviewer examined the studies and settled the differences. The extracted data included (I) baseline characteristics of enrolled participants, (II) study design, (III) diagnostic test parameters: Parameters, cutoff value and US system for SWE and contrast agent, US technique, probe and mechanical index for CEUS, and (IV) Outcome data: true positive (TP), true negative (TN), false positive (FP), and false negative (FN) values. When these values were not directly given, they were calculated from the processed data as sensitivity, specificity, PPV, and NPV, using the statistical calculator on RevMan software (Version 5.3 for Windows). We used the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) score to assess the quality of included studies. It consists of 14 (yes/no/unclear) questions to assess different forms of bias within DTA studies (15).

Data Analysis
Pairwise meta-analyses were done under the random-effects model when two or more studies investigated the same predefined research question with the same laboratory test. We extracted the sensitivity, specificity, positive likelihood ratio (LR), negative LR, and diagnostic odds ratio (DOR) values for CEUS and SWE compared to histopathology as a reference standard. The DOR is calculated as (TP X TN)/ (FP X FN) and defined as the odds of having a positive test result in a patient with disease compared with the odds of a positive test result in a patient without disease. Moreover, summary receiver operating characteristic (SROC) curves were constructed to examine diagnostic accuracy. All statistics were reported as absolute values with their 95% confidence interval (95% CI). A p-value <0.05 was considered statistically significant. The Chisquare and I-square statistics were calculated in order to assess heterogeneity. Significant heterogeneity was considered to be present if the chi-square p-value was<0.1 (as per the Cochrane Handbook for Systematic Reviews of Intervention). Data were presented into five subgroups according to cancer site: breast, liver, thyroid, kidneys, and prostate. Network meta-analyses were conducted to compare the diagnostic accuracy of CEUS vs. SWE in malignancy detection. Heterogeneity and inconsistency were checked by the I 2 and the corresponding p-value. All statistical analyses were conducted on MetaDiSc (version 1.4) and R software (version 3.4.3).

Literature Search and Study Characteristics
Database search retrieved 5896 unique citations. Following title and abstract screening, 422 full-text articles were retrieved for further scrutiny. Finally, 114 diagnostic accuracy studies (65 on SWE and 50 on CEUS; one study by 4 assessed both modalities), reporting data from 15926 patients (5680 for CEUS and 10392 for SWE) were included in our network meta-analysis (Figure 1, Bibliographic details in Supplementary File II). According to   Figure 2B. Significant heterogeneity was observed for sensitivity (p < 0.0001) and specificity (p < 0.0001).

Hepatic Cancer
The pooled sensitivity, specificity, positive LR, and negative LR for CEUS in differentiating malignant hepatic lesions were 0. 78 Figure 3A. The included studies were heterogeneous in the estimates of sensitivity (p < 0.0001) and specificity (p < 0.0001).

Outcomes of Network Meta-Analysis
Corresponding network plots and forest plots of network metaanalysis between CEUS and SWE are shown in Figure 6.

Ranking Diagnostic Tests
According to Glas et al. (116), the DOR is considered as an indicator of ranking of competing diagnostic tests. According to our results, CEUS achieved the highest DOR in detecting breast and thyroid malignant lesions, while SWE achieved the highest DOR in detecting hepatic malignant lesions.

DISCUSSION
This meta-analysis of DTA studies provides a comprehensive assessment and comparison of the diagnostic accuracy of two US modalities in differentiating malignant tumors in different body organs. It showed relatively high sensitivity (between 78 and 89%) and specificity (between 84 and 89%) for CEUS in identifying malignant lesions in the breast, liver, thyroid and kidneys. Moreover, it demonstrated relatively high sensitivity (between 82 and 84%) and specificity (between 83 and 86%) for SWE in differentiating malignant tumors within the breast, liver and prostate. However, it had relatively lower sensitivity (67%) and specificity (77%) in identifying malignant nodules within the thyroid gland. Our results support some recent practice guidelines that endorse the use of CEUS and SWE in differentiating malignant lesions within the liver and the breast (117,118). Moreover, it provides new data on a comparison that can impact the clinical practice. Through NMA, we compared the diagnostic accuracy of CEUS and SWE in three organs (where data on both tests were available in the literature). Our network and ranking analysis showed that CEUS was more accurate than SWE in differentiating breast and thyroid lesions (although the difference was not significant in thyroid malignancy according to NMA). On the other hand, SWE ranked higher in terms of diagnostic accuracy in differentiating hepatic malignant lesions (although the difference was not significant according to NMA).
Our results are in agreement with a former meta-analysis by Sadigh et al. that showed high sensitivity and specificity for SWE in differentiating breast malignant lesions [88 and 83% in comparison to 84 and 86% in our analysis; (11)]. However, our sensitivity and specificity results are quite lower than those obtained by Liu et al. in a meta-analysis on SWE accuracy in differentiating thyroid malignancy [sensitivity 81% and specificity 84%; (12)]. Likewise, another meta-analysis reported high sensitivity and specificity (93 and 90%, respectively) for CEUS in identifying hepatic malignant lesions (119). The observed discrepancy between our findings and those of the aforementioned meta-analyses may be attributed to the different sample size (being larger in our analysis) or the lesional characteristics of enrolled patients (being easier to identify in the studies included in the other meta-analysis i.e., less depth and clear contrast from the surrounding tissue).
Interestingly, a meta-analysis by Guang et al. showed comparable diagnostic accuracy for SonoVue-enhanced US with contrast-enhanced computed tomography and magnetic resonance imaging (8). Moreover, CEUS has other advantages over these modalities as ease of access, lack of radiation exposure or nephrotoxic materials; limitations that affect the use of CT and MRI in several diagnostic applications (120,121). It is also fair to recognize that both tests have limitations as well. For example, SWE suffers from operator-dependency and manual compression, while the adverse effects of the contrast agent is a concern with CEUS use. Further technical improvements with both modalities would further enhance their clinical potential.

Strength Points
This NMA directly compares the diagnostic accuracies of CEUS and SWE in different cancer sites and using different analytic approaches as pairwise, network and ranking pooled analyses. Therefore, it provides a holistic evaluation of the comparison of both techniques in different body organs. We performed a thorough literature search and retrieved a large number of studies (relatively large sample size), which adds to the validity and generalizability of our findings. Unlike former reviews that retrieved a small number of studies and focused on one test in one organ, we aimed to provide a comprehensive assessment of both tests in different organs and a high quality comparison whenever suitable data were provided.

Limitations and Future Research Implications
Our meta-analysis has some limitations. First, the observed heterogeneity in the majority of our outcomes may be due to differences in study design and patient characteristics. Second, we could not examine the effects of lesion characteristics, such as size and depth on the diagnostic accuracy of both tests due to lack of data. Third, many of the included studies did not mention whether the results of CEUS or SWE were interpreted with blinding to the findings of histopathology or not. Future studies should report diagnostic accuracy data based on the size and depth of the lesions to allow more detailed analysis. Moreover, they should adhere to the Standards for Reporting of Diagnostic Accuracy "STRAD" checklist in reporting their methods and findings to allow a more thorough critical appraisal.

CONCLUSION
Both diagnostic tests (CEUS and SWE) showed relatively high sensitivity and specificity in detecting malignant tumors in different organs; CEUS had higher diagnostic accuracy than SWE in detecting breast and thyroid cancer, while SWE had higher accuracy in detecting hepatic cancer (the differences in the latter two cancer types were not statistically significant). These results endorse the use of both tests for malignancy detection and rank their accuracy in different organs. Future studies should provide more data to allow characterization of both tests in lesions of different size or depth.

AUTHOR CONTRIBUTIONS
YS developed the concept, designed the study, and prepared the manuscript. RH acquired the data, controlled quality of the work, analyzed the data, and prepared the manuscript. LJ acquired the data. YX analyzed the data. YG acquired the data. HR acquired the data and conducted the analysis. ZW analyzed the data and prepared the manuscript.

ACKNOWLEDGMENTS
We are extremely thankful to authors of all the included papers for proving suitable data for analysis.