CRD summary

This review concluded that endoscopic ultrasound was highly accurate for the staging of superficial oesophageal cancers, and the location and type of lesion, ultrasound method, and operator's experience could affect accuracy. The lack of assessment of study quality and limitations in the conduct and interpretation of the meta-analyses mean that these conclusions may not be reliable.

Authors' objectives

To assess the diagnostic accuracy of endoscopic ultrasound in differentiating mucosal versus submucosal invasion of superficial oesophageal cancers.

Searching

MEDLINE, Scopus, Cochrane Database of Systematic Reviews, and CINAHL Plus were searched and the Internet was searched using Google Scholar, for articles from 1980 to June 2010. The search terms were reported. The bibliographies of included studies were screened for additional articles.

Study selection

Prospective or retrospective studies of 15 or more patients with suspected oesophageal cancer or oesophageal cancer confirmed by endoscopic biopsy and imaging were eligible for inclusion. Studies were required to compare endoscopic ultrasound with a reference standard, which included histopathologic staging of an endoscopic mucosal resection or surgically removed sample, and to report sufficient data to construct 2x2 contingency tables (numbers of true-positive, false-positive, true-negative, and false-negative results). Studies where staging could not be separated into mucosal or submucosal, and those of superficial oesophageal cancers in which lymph nodes that could be metastatic (over 1cm in size) were detected during endoscopic ultrasound, were excluded.

Most of the included studies were conducted in Japan, and the others were conducted in the USA or Europe. Most studies assessed the performance of endoscopic ultrasound using a radial scanning echoendoscope. The frequencies of endoscopic ultrasound ranged from 7.5MHz to 30MHz.

Two reviewers independently assessed studies for inclusion and any disagreements were resolved by discussion with two other reviewers.

Assessment of study quality

The authors did not state that they assessed the methodological quality of the included studies.

Data extraction

Data were extracted to populate 2x2 contingency tables for mucosal (T1a) and submucosal (T1b) invasion. These data were used to calculate the sensitivity, specificity, and positive and negative likelihood ratios, with 95% confidence intervals. Where any cell was zero for a study, a correction of 0.5 was added to all cells for that study to allow the calculation of sensitivity and specificity.

Two reviewers independently extracted the data.

Methods of synthesis

Pooled estimates of sensitivity, specificity, positive and negative likelihood ratios, and the diagnostic odds ratio, with 95% confidence intervals, were calculated using a DerSimonian and Laird random-effects model. Summary receiver operating characteristic curves were constructed using the Moses-Littenberg model.

Between-study heterogeneity was assessed using Χ², Cochran's Q and Ι². Subgroup analyses and meta-regression analysis were used to investigate potential sources of between-study heterogeneity. Subgroups were studies from Japan or outside Japan, studies focusing on squamous cell carcinoma or adenocarcinoma, studies using only radial scanning endoscopic ultrasound or a high-frequency probe (15MHz or higher), and studies from Japan using a high-frequency mini probe) The Fisher exact test was used to test for significant differences in the under-diagnosis and over-diagnosis rates with endoscopic ultrasound for squamous cell carcinoma and adenocarcinoma.

Publication bias was assessed using the Egger and fail-safe N tests, and the trim-and-fill method. Funnel plots were constructed using the standard error and diagnostic odds ratio.

Results of the review

Nineteen studies, with a total of 1,019 participants (range 17 to 102), were included in the review.

For T1a staging, the pooled sensitivity of endoscopic ultrasound was 85% (95% CI 82 to 88) and the specificity was 87% (95% CI 84 to 90). The pooled positive likelihood ratio was 6.62 (95% CI 3.61 to 12.12) and the negative ratio was 0.20 (95% CI 0.14 to 0.30).

For T1b staging, the pooled sensitivity of endoscopic ultrasound was 86% (95% CI 82 to 89) and the specificity was 86% (95% CI 83 to 89). The pooled positive likelihood ratio was 5.13 (95% CI 3.36 to 7.82) and the negative likelihood ratio was 0.17 (95% CI 0.09 to 0.30).

Between-study heterogeneity, as assessed by Ι², was high for all analyses. Subgroup and regression analyses indicated that studies conducted in Japan had higher estimates of diagnostic accuracy (diagnostic odds ratio) than those conducted outside Japan. No other variables were significant. The Fisher exact test indicated significantly more under-staging of squamous cell carcinomas than of adenocarcinomas. There was some evidence of publication bias in the Egger test.

Authors' conclusions

Endoscopic ultrasound had high accuracy for the staging of superficial oesophageal cancers. Variation between studies suggested that factors, such as the location and type of lesion, method and frequency of ultrasound probe, and the operator's experience, could affect diagnostic accuracy.

CRD commentary

This review assessed the accuracy of endoscopic ultrasound for the staging of superficial oesophageal cancers. The inclusion criteria were fully defined and a range of sources were searched for relevant studies. It was not clear whether any language restrictions were applied, but studies published in Japanese were translated. Measures to minimise error and bias were applied throughout the review process, but the methodological quality of included studies was not assessed and it is therefore not possible to judge the impact of any weaknesses in the included studies upon the findings of the review.

The models used to pool the estimates of sensitivity and specificity were not those currently recommended for test accuracy studies. The authors' conclusions that the location and type of lesion, method and frequency of ultrasound probe, and the operator's experience could affect the diagnostic accuracy of endoscopic ultrasound were not supported by the data presented.

The lack of an assessment of the methodological quality of the included studies and limitations in the conduct and interpretation of the meta-analyses mean that the conclusions may not be reliable.

Implications of the review for practice and research

Practice: For suspected early stage oesophageal cancer, the authors recommended examination with endoscopic ultrasound first to rule out infiltration of the muscularis propria (T2) and regional lymph node metastasis (stage II and beyond), followed by endoscopic mucosal resection with histological examination of the removed specimen.

Research: The authors stated that further prospective studies were needed to explore factors affecting the sensitivity and specificity of endoscopic ultrasound in superficial oesophageal cancer staging and to standardise the endoscopic ultrasound process to improve its accuracy.

Funding

Not stated.

Bibliographic details

Thosani N, Singh H, Kapadia A, Ochi N, Lee JH, Ajani J, Swisher SG, Hofstetter WL, Guha S, Bhutani MS. Diagnostic accuracy of EUS in differentiating mucosal versus submucosal invasion of superficial esophageal cancers: a systematic review and meta-analysis. Gastrointestinal Endoscopy 2012; 75(2): 242-253. [PubMed: 22115605]

Original Paper URL

http://www.giejournal.org/article/S0016-5107(11)02204-8/abstract

Indexing Status

Subject indexing assigned by NLM

MeSH

Adenocarcinoma /pathology /ultrasonography; Endosonography; Esophageal Neoplasms /pathology /ultrasonography; Humans; Neoplasm Invasiveness; Sensitivity and Specificity

AccessionNumber

12012004881

Database entry date

29/10/2012

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.