Clinical relevance of morphologic MRI criteria for the assessment of lymph nodes in patients with rectal cancer.

Abstract

The aim of this study is the evaluation of lymph node staging by magnetic resonance imaging (MRI) within clinical routine in patients with rectal cancer. Routine MRI reports (3 T) of 65 consecutive patients with rectal cancer were retrospectively categorized in lymph node tumor positive or negative (mriN+; mriN0) and compared to the final histopathological results (pN+; pN0). Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and accuracy were calculated. The original MRI readings were then reanalyzed in order to identify the longest short-axis lymph node diameter for each patient. A receiver operating characteristic (ROC) curve was used to calculate a possible cutoff value for the short-axis lymph node diameter. Overall sensitivity was 94 %, specificity 13 %, NPV 86 %, PPV 28 %, and accuracy 34 %. The best accuracy could be calculated for a short-diameter cutoff of ≤5 mm (83 %); pN+ and pN0 groups were then significantly different (p < 0.0001). In clinical routine, lymph node assessment in patients with rectal cancer through MRI tends to overstage malignant lymphadenopathy. A ≤5-mm cutoff value for the short-axis lymph node diameter of benign nodes is able to improve the accuracy and has potential to lower the risk of overstaging.