EUS versus magnetic resonance imaging in staging rectal adenocarcinoma: a diagnostic test accuracy meta-analysis
Additional Document Info
BACKGROUND AND AIMS: EUS and magnetic resonance imaging (MRI) are both used for locoregional staging of rectal cancer, which determines treatment options. There is a lack of consensus on the best modality for locoregional staging, with studies supporting both EUS and MRI. In this study, we performed the first diagnostic test accuracy meta-analysis to compare the diagnostic accuracy, sensitivity, and specificity of EUS and MRI in the staging of rectal cancer. METHODS: A comprehensive electronic literature search up to June 2018 was performed to identify prospective cohort studies directly comparing the accuracy of EUS with MRI in staging nonmetastatic rectal cancer with surgical pathology as the reference standard. Quality of the included studies was measured by using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool. A bivariate hierarchical model was used to perform the meta-analysis of diagnostic test accuracy according to the Cochrane approved methodology. Summary receiver operating characteristics were developed, and the area under the curve was calculated for overall and individual T and N staging, for EUS, MRI, and head-to-head comparison. RESULTS: Six of 2475 studies including 234 patients were eligible. Pooled sensitivity and specificity in T staging were .79 (95% confidence interval [CI], .72-.85) and .89 (95% CI, .84-.93) for EUS and .79 (95% CI, .72-.85) and .85 (95% CI, .79-.90) for MRI, respectively. Pooled sensitivity and specificity in N staging were .81 (95% CI, .71-.89) and .88 (95% CI, .80-.94) for EUS and .83 (95% CI, .73-.90), and .90 (95% CI, .82-.95) for MRI, respectively. In area under the curve head-to-head analysis, EUS was superior to MRI in overall T staging (P < .05). EUS outperformed MRI in overall T, overall N, T1, and T3 staging (P < .01), after excluding studies using an endorectal coil for MRI. MRI was superior to EUS in T2 staging (P = .01) in both analyses. CONCLUSIONS: EUS and MRI both provide reasonable diagnostic accuracy in the staging of nonmetastatic rectal cancer. EUS was superior to MRI in overall T staging and overall T and N staging after adjusting for MRI technology. Practitioners should be aware of advantages and disadvantages of both modalities and choose appropriate methods while considering diagnostic accuracy of each test and institutional practices and limitations.