Abstract
BACKGROUND: To investigate and compare the diagnostic performance of (18)F-Fluorodeoxyglucose ((18)F-FDG) PET/MR and MR alone in whole-body primary staging and restaging of patients with rectal cancer. METHODS: A retrospective analysis was performed to evaluate diagnostic accuracies of combined reading of PET/MR and MR alone in T, N and M staging against the reference standard. Inter-observer agreement regarding TNM staging was calculated separately for PET/MR and MR alone. RESULTS: A total of 39 studies of 34 patients could be evaluated. Diagnostic accuracies of PET/MR and MR alone were the same in locoregional T staging. For predicting N+ stage, the specificity of combined reading of PET and MR (0.917 and 0.833 for reader 1 and 2, respectively) was slightly higher than MR alone (0.833 and 0.75) without significantly increasing the overall accuracy (0.783 vs. 0.783 and 0.783 vs. 0.739). For detecting distant metastasis, the sensitivities of PET/MR and MR alone were shown equal (1.0 vs. 1.0 and 0.938 vs. 0.938), while the specificity of PET/MR was marginally lower (0.87 vs. 0.913 and 0.826 vs. 0.87). The inter-observer agreements were good to excellent in M (κ = 0.64 and 0.637 for PET/MR and MR alone, p < 0.001) and N staging (0.819 and 0.738, p < 0.001). CONCLUSION: PET did not yield a significant improvement in diagnostic accuracy of PET/MR in TNM staging of rectal cancer, since MR alone facilitated accurate classification of disease stage with good to excellent inter-observer agreement.