This study aimed to evaluate the performance characteristics of endoscopic ultrasound (EUS) elastography combined with contrast-enhanced EUS (CE-EUS) for the diagnosis of pancreatic malignancy among solid pancreatic lesions (SPLs).
A comprehensive literature search using MEDLINE, Embase, and the Cochrane Library was performed from January 1991 through December 2020. The pooled performance characteristics, including sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR-), and diagnostic odds ratio (OR), were calculated using Meta-DiSc 1.4.
In total, 430 patients with 282 pancreatic malignancies were included in the metaanalysis. The pooled sensitivity, specificity, LR+, LR-, and diagnostic OR were 84% (95% confidence interval [CI], 80% to 88%), 85% (95% CI, 78% to 90%), 5.31 (95% CI, 2.57 to 10.97), 0.15 (95% CI, 0.07 to 0.34), and 67.72 (95% CI, 12.84 to 357.26) for the combination of CE-EUS and EUS elastography, respectively. In these same studies, the corresponding performance characteristics for EUS elastography were 87% (95% CI, 82% to 90%), 56% (95% CI, 48% to 64%), 2.28 (95% CI, 1.43 to 3.63), 0.12 (95% CI, 0.03 to 0.42), and 22.60 (95% CI, 5.81 to 87.92), respectively. In these same studies, the respective performance characteristics for CE-EUS were 84% (95% CI, 80% to 88%), 78% (95% CI, 70% to 84%), 3.80 (95% CI, 1.92 to 7.53), 0.13 (95% CI, 0.04 to 0.41), and 31.29 (95% CI, 6.12 to 159.87).
CE-EUS and EUS elastography are reliable supplemental techniques for the characterization of SPLs and diagnosis of pancreatic malignancies. However, more studies assessing the combined utilization of both procedures are needed.
Previous studies have studied the performance of endoscopic ultrasound (EUS) elastography combined with contrast-enhanced EUS. Currently, the efficacy of the combined modality is unclear. This meta-analysis revealed no statistically significant improvement for the combined method versus each modality used separately. More studies and greater experiences are needed before the role of these modalities in the diagnostic approach to solid pancreatic lesions can be delineated.
Endoscopic ultrasonography (EUS) is a widely used diagnostic modality to evaluate pancreaticobiliary diseases, especially solid pancreatic lesions (SPLs), and it has a good ability to detect pancreatic masses. However, EUS has a limited ability to distinguish the diagnoses of lesions, and thus, EUS with fine needle biopsy (EUS-FNB) is the current gold standard for diagnosing pancreatic tumors (largely replacing fine needle aspiration [FNA]), with sensitivity, specificity and accuracy of 54%-96%, 96%-98%, and 83%-95%, respectively [
CE-EUS is a newer imaging modality utilizing an ultrasonographic contrast agent to visualize blood flow in fine vessels and to evaluate SPLs [
The combination of EUS elastography and CE-EUS has not been extensively studied in the evaluation of SPLs. Because of the continued evolution of endosonographic techniques, the aim of the present study was to assess the diagnostic effectiveness of EUS elastography combined with CE-EUS for the diagnosis of SPLs by calculating the pooled sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR-), and diagnostic odds ratio (OR) from studies that compared the two modalities.
The systematic review and meta-analysis was prepared and revised according to the Preferred Reporting Items for Systematic Review and Meta-Analyses 2020 (PRISMA 2020) extension for diagnostic test accuracy.
A comprehensive literature search using the MEDLINE, Embase, and Cochrane Library databases was performed from January 1991 through December 2020. The search formula for all three databases was: (eus OR 'endoscopic ultrasound' OR ultrasonography OR endosonography) AND ('contrast-enhanced' OR 'contrast enhancement' OR 'contrast imaging' OR 'ce eus' OR 'contrast enhanced endosonography' OR 'contrast enhanced endoscopic ultrasound') AND (elastography OR endoelastography OR elasticity) AND (pancreas OR 'pancreatic lesion' OR 'solitary pancreatic lesion' OR 'pancreatic mass' OR 'pancreatic cancer'). There were no language restrictions. This study was limited to articles published after 1991, when elastography was invented.
The articles eligible for inclusion were limited to retrospective and prospective manuscripts or abstracts studying the diagnostic accuracy of CE-EUS, EUS elastography, and their combined use for distinguishing pancreatic adenocarcinoma among SPLs. In addition, the studies were required to use histology of EUS-FNB samples and/or surgical specimens to confirm the histopathologic diagnosis and to have a minimum follow-up of 6 months for patients without surgical or EUS-FNB samples to classify as chronic pancreatitis. Furthermore, the studies were required to have used chronic pseudotumoral pancreatitis as a control group and provide contingency tables or adequate data to extrapolate contingency tables based on sensitivity, specificity, and/or accuracy.
The exclusion criteria were the inclusion of cystic pancreatic lesions and/or malignancies other than pancreatic adenocarcinoma with a lack of a subgroup analysis to extrapolate contingency tables for the targeted research aim, overlap within selected studies such as same study group, research institution, and period of inclusion, and lack of confirmatory diagnostic testing with either EUS-FNB, surgical histology, or appropriate follow-up/surveillance intervals. All reviews, editorials, and letters that did not report their own data were excluded from the study, as were case reports or case series. Technical studies that focused on the technology of CE-EUS and EUS elastography and studies including non-pancreatic cancer or either EUS elastography or CE-EUS were excluded.
A meta-analysis for CE-EUS, EUS elastography, and their combined use was performed by calculating pooled estimates of diagnostic accuracy indices, including the sensitivity, specificity, LR+, LR-, and diagnostic OR based on provided or extrapolated contingency tables. Summary receiver operating characteristics (SROC) curves were generated, and the area under the curve (AUC) and the Q* index (the point on the SROC curve where sensitivity and specificity are equal) were then calculated. Statistical analysis was conducted using the Meta-DiSc 1.4 software program (Ramóny Cajal Hospital, Madrid, Spain). Pooled results with 95% confidence intervals (CIs) were derived using the DerSimonian-Laird method (a random-effects model) [
The selection process is illustrated in
The characteristics of the studies included are summarized in
The demographics of patients and characteristics of cancer from each study included in the meta-analysis are described in
A true-positive was defined as increased stiffness or a hard lesion by elastography and hypovascular or hypodense lesion on CE-EUS. A true negative was defined as decreased stiffness or soft lesion by elastography and a hypervascular or hyperdense lesion on CE-EUS. When comparing EUS elastography results from all six studies, high heterogeneity was observed for sensitivity (I2=91.6%, P<0.001), specificity (I2=84.7%, P<0.001), LR+ (I2=82.6%, P<0.001), LR-(I2=82.0%, P<0.001), and the diagnostic OR (I2=67.1%, P=0.010). The pooled analysis of diagnostic tests of EUS elastography with regard to the diagnosis of SPLs for pancreatic adenocarcinoma showed a sensitivity of 87% (95% CI, 82% to 90%), specificity of 56% (95% CI, 48% to 64%), LR+ of 2.28 (95% CI, 1.43 to 3.63), LR- of 0.12 (95% CI, 0.03 to 0.42), and diagnostic OR of 22.60 (95% CI, 5.81 to 87.92), as shown in
When comparing the CE-EUS results from all six studies, high heterogeneity was observed for sensitivity (I2=87.3%, P<0.001), specificity (I2=66.7%, P=0.010), LR+ (I2=80.0%, P<0.001), LR- (I2=84.7%, P<0.001) and the diagnostic OR (I2=83.8%, P<0.001). The pooled analysis of diagnostic tests of CE-EUS with regard to diagnosis of SPLs for pancreatic adenocarcinoma showed a sensitivity of 84% (95% CI, 80% to 88%), specificity of 78% (95% CI, 70% to 84%), LR+ of 3.80 (95% CI, 1.92 to 7.53), LR- of 0.13 (95% CI, 0.04 to 0.41), and diagnostic OR of 31.29 (95% CI, 6.12 to 159.87), as shown in
When comparing the results of CE-EUS combined with EUS elastography from all six studies, high heterogeneity was observed for sensitivity (I2=84.5%, P<0.001), specificity (I2=77.7%, P<0.001), LR+ (I2=62.2%, P=0.021), LR- (I2=80.3%, P<0.001), and the diagnostic OR (I2=73.5%, P=0.002). The pooled analysis of diagnostic tests of CE-EUS combined with EUS elastography with regard to diagnosis of SPLs for pancreatic adenocarcinoma showed a sensitivity of 84% (95% CI, 80% to 88%), specificity of 85% (95% CI, 78% to 90%), LR+ of 5.31 (95% CI, 2.57 to 10.97), LR- of 0.15 (95% CI, 0.07 to 0.34), and diagnostic OR of 67.72 (95% CI, 12.84 to 357.26), as shown in
The QUADAS-2 questionnaire for assessment of the quality of the selected studies demonstrated high quality of the studies (
Several meta-analyses evaluating the diagnostic performance of EUS elastography for diagnosing malignant pancreatic tumors have shown a diagnostic sensitivity of 92%-98% [
Studies that investigated the efficacy of CE-EUS only or elastography were excluded from this meta-analysis. Including data from a single-method study was deemed to be not applicable to the goal of the study, which was to compare a combined method to a single method, and may introduce confounding and bias. Most of the studies included in this meta-analysis used hue histograms. Elasticity can be measured both qualitatively and quantitatively [
Studies have also demonstrated good performance outcomes of CE-EUS, with diagnostic sensitivity of 89%-96% and specificity of 64%-100% across several studies [
Significant heterogeneity was observed among the studies, likely owing to the high number of false-positives in the studies of Hocke et al. [
Another factor contributing to significant heterogeneity was that the studies by Hocke et al. [
A method to potentially standardize heterogeneity with regard to histogram and strain ratio cutoffs is to generate artificial intelligence–based systems with images and videos generated from the adjunctive EUS techniques of CE-EUS and EUS elastography and the corresponding pathology results. Aside from this potentially exciting application, CE-EUS and EUS elastography do have the potential to serve as clinically useful adjuncts to conventional EUS in the diagnosis of SPLs, but more studies–particularly in the United States–and greater experiences are needed before their role in the diagnostic approach to SPLs can be delineated.
B-mode is the conventional first method used to target a lesion. B-mode makes it possible to find the lesion, and then EUS elastography and CE-EUS are applied. B-mode only enables a characterization of a lesion as hypoechoic, hyperechoic, or heterogeneous, but not whether it is malignant or not. After a lesion is identified using B-mode, FNB of the identified lesion is performed. CE-EUS and EUS elastography for the purposes of this study are used to help further stratify the potential for malignancy. When comparing B-mode EUS to CE-EUS, CE-EUS remains superior for detecting pancreatic cancer [
EUS-guided FNB is regarded as the technique of choice for diagnosis of focal pancreatic masses. EUS elastography and CE-EUS have emerged as newer techniques to characterize SPLs, with both techniques having excellent accuracy and high sensitivity. Neither technique, however, is currently approved by the Food and Drug Administration in the United States, and while the techniques have been increasingly adopted as adjunctive, supplemental tools for diagnosis of SPLs and pancreatic malignancy, they are still not widely available on the global scale. With regard to future applications of these two modalities, these techniques may bring significant improvement and guidance during clinical procedures by isolating high-yield targets within SPLs that may increase the diagnosis of pancreatic malignancies in questionable lesions, which, in turn, could limit unnecessary surgical resections and reduce postoperative morbidity and mortality, while also reducing the number of needle passes required to obtain good samples [
This meta-analysis demonstrated that when EUS elastography was combined with CE-EUS, there was a pooled sensitivity of 84% and specificity of 85%. There was no statistically significant difference in measures of diagnostic accuracy compared to the use of each modality alone. The analysis was limited by the low sample sizes of eligible studies and the paucity of true-positives from both diagnostic tests in the same patient. However, EUS elastography combined with CE-EUS had the highest specificity and diagnostic OR among the three comparisons. Most prospective studies assessing EUS elastography and CE-EUS involved investigating each modality without incorporation of the other, which significantly limited the comparison of assessment of the efficacy of the combined techniques. Furthermore, a "true-positive" of the combination was defined as both techniques being separately positive for malignancy instead of having a combined criterion for positivity. This could lead to potential false-negatives, thus decreasing the sensitivity. However, this would not explain the poor specificity, as there was a relatively high number of false-positive results with both techniques combined. Insufficient data were provided to calculate the performance characteristics of the combination of both modalities if at least one of the methods was positive. Presumably, the diagnostic performance would be increased, and this would be of interest for future investigations.
Conceptualization: Villa E. Data acquisition: Shin CM. Data analysis or interpretation: Shin CM. Drafting of the manuscript: Shin CM, Villa E. Critical revision of the manuscript: Shin CM, Villa E. Approval of the final version of the manuscript: all authors.
No potential conflict of interest relevant to this article was reported.
The study flow diagram demonstrating the algorithm for identifying studies to be included in the meta-analysis. EUS, endoscopic ultrasonography; CE, contrast-enhanced.
Forest plots of the meta-analysis of sensitivity (A), specificity (B), LR+ (C), LR- (D), and diagnostic OR (E), as well as the SROC curve (F) for the differentiation of pancreatic adenocarcinomas with EUS elastography only [
Forest plots of the meta-analysis of sensitivity (A), specificity (B), LR+ (C), LR- (D), and diagnostic OR (E), as well as the SROC curve (F) for differentiation of pancreatic adenocarcinomas with CE-EUS only [
Forest plots of the meta-analysis of sensitivity (A), specificity (B), LR+ (C), LR- (D), and diagnostic OR (E), as well as the SROC curve (F) for differentiation of pancreatic adenocarcinomas with the combination of CE-EUS and EUS elastography [
Results from the Quality Assessment of Diagnostic Accuracy Studies questionnaire [
Characteristics of the studies, including country, contrast mode and agent, diagnostic standard, and final diagnosis method
Study | Country | Study type | Contrast mode | Contrast agent | Diagnostic standard | Final diagnosis |
---|---|---|---|---|---|---|
Saftoiu et al. (2010) [ |
Romania | Prospective | Power Doppler | Sonovue | Elastography: mean hue histogram >175 | Histology |
CE-EUS: hypovascularity | 6-Month follow-up | |||||
Hocke et al. (2012) [ |
Germany | Prospective | Color/power Doppler | Sonovue | Elastography: hue color map (blue coding) | Histology |
CE-EUS: hypoenhancement | 6-Month follow-up | |||||
Iordache et al. (2016) [ |
Romania | Retrospective | Power Doppler | Sonovue | Elastography: mean hue histogram >175 | Histology |
CE-EUS: hypoenhancement | 1-Year follow-up | |||||
Iglesias-Garcia et al. (2017) [ |
Spain | Prospective | Power Doppler | Sonovue | Elastography: mean hue histogram >50 and strain ratio >10 | Histology |
CE-EUS: hypovascularity | 2-Year follow-up | |||||
Chantarojanasiri et al. (2017) [ |
Japan | Retrospective | Color Doppler | Sonazoid | Elastography: hue color map (blue coding) | Histology |
CE-EUS: hypovascularity | 6-Month follow-up | |||||
Costache et al. (2020) [ |
Germany, Italy, France, Spain, China, and Romania | Prospective | Power Doppler | Sonovue | Elastography: mean strain histogram <80 | Histology |
CE-EUS: hypoenhancement | 6-Month follow-up |
CE-EUS, contrast-enhanced endoscopic ultrasonography.
The demographics of patients and characteristics of pancreatic adenocarcinoma from each study included in the meta-analysis
Study | No. of cancer patients (%) | Age (yr) | Sex of cancer patients, male (%) | Mean size of cancer (mm) | Mean hue histogram value for cancer | Location of cancer, head:body:tail (n:n:n) |
---|---|---|---|---|---|---|
Saftoiu et al. (2010) [ |
33 (61.1) | 61.6±1.36 | 22 (66.7) | 40.2±0.22 | 190.30±2.07 | 26:4:3 |
Hocke et al. (2012) [ |
19 (32.8) | n/a | n/a | n/a | n/a | n/a |
Iordache et al. (2016) [ |
19 (61.3) | 60.31±8.43 | 12 (63.2) | n/a | 187.35±13.18 | n/a |
Iglesias-Garcia et al. (2017) [ |
13 (26.5) | 69.1±13.5 | 33 (67.3) | n/a | n/a | 37:12:0 |
Chantarojanasiri et al. (2017) [ |
95 (62.5) | 65.1±10.2 | 57 (60.0) | 25 | 25 | 62:26:7 |
Costache et al. (2020) [ |
70 (72.2) | 68.6±10.0 | 48 (68.6) | 33.6±11.12 | Strain: 22.7±11.2 | 40:21:8 |
n/a, not available.