Evaluating local lymph node metastasis with magnetic resonance imaging, endoluminal ultrasound and computed tomography in rectal cancer: a meta-analysis.

Abstract

Magnetic resonance imaging (MRI), endorectal ultrasound (EUS) and computed tomography (CT) are commonly used to evaluate lymph node (LN) metastasis for rectal cancer, but there is no agreement on which form of imaging is most accurate. The study aimed to review systematically the diagnostic performance of the three imaging modalities. The PubMed, Cochrane Library and EMBASE databases were systematically searched for English and Chinese language studies evaluating the diagnostic accuracy of MRI, EUS and/or CT for evaluating LN metastasis. Papers published before 31 December 2013 were included in the search. Subject-level data were included. Diagnostic odds ratios were calculated for each modality and summary receiver operating characteristic curves were constructed using hierarchical regression models. The performance of the three modalities was compared. The analysis included data from 123 studies evaluating LN metastasis. The sensitivity and specificity in patients having no chemoradiotherapy were 0.77 and 0.76 for MRI, 0.57 and 0.80 for EUS and 0.79 and 0.76 for CT. The three modalities showed similar accuracy (P=0.39). MRI showed higher accuracy than EUS for patients who received neoadjuvant therapy (P=0.04). MRI at a field strength >1.5T yielded better performance than EUS (P=0.03) and similar performance to CT (P=0.17). High resolution MRI showed similar diagnostic accuracy to EUS (P=0.18) and CT (P=0.16). MRI, EUS and CT show similar accuracy but none could provide reliable evaluation for LN metastasis. MRI rather than EUS is recommended for LN evaluation after neoadjuvant therapy. MRI at a field strength of 3.0T is the recommended method for MRI examination. But high resolution MRI does not improve the diagnostic performance in evaluating LN metastasis.