Diagnostic accuracy of imaging modalities in detection of histopathological extranodal extension: A systematic review and meta-analysis

To present an up to date systematic review and meta-analysis evaluating the diagnostic accuracy of the most used imaging modalities in detection of histopathological extra nodal extension (ENE) in head and neck squamous cell carcinoma. Medline, Embase, and Cochrane databases were systematically searched on March 27th 2020. Screening, inclusion, quality assessment, and data extraction were done by two reviewers. Meta-analysis was conducted using the bivariate model approach after pooling the studies according to imaging modality. Heterogeneity was explored by meta-regression. Comparison was done by meta-regression and sub-group analyses. Out of 476 initial hits, 25 studies were included for analysis. Of these, 14 dealt with CT, nine with PET/CT, four with MRI, two with ultrasound, and none with PET/MRI. Meta-analysis based on a total sample size of 3391 showed that CT had a sensitivity of 76% [67–82%] and specificity of 77% [69–83%], MRI a sensitivity of 72% [64–79%] and specificity of 78% [57–90%], and PET/CT a sensitivity of 80% [76–84%] and specificity of 83% [74–90%] in the ability to predict ENE. No meta-analysis could be done on ultrasound. There were no significant differences between modalities in overall accuracy; however, PET/CT had significantly higher sensitivity than CT and MRI. There was no significant difference in the ability of CT, MRI, and PET/CT to diagnose histopathological ENE, except that PET/CT had a significantly higher sensitivity than CT and MRI.