Background: To study different imaging criteria for prediction of lymph node metastases (Stage III disease) in colon cancer using CT. Methods: In a retrospective setting, 483 consecutive patients with histology proven colon cancer underwent elective primary resection during 2008-2011, a cohort of 119 patients were included. Contrast enhanced CT examinations, in portal-venous phase, were reviewed with assessment of the number of lymph nodes, their anatomical distribution, size, size ratio, internal heterogeneity, presence of irregular outer border and attenuation values. Sensitivity, specificity, PPV and NPV for each studied criteria for prediction of stage III disease was calculated. Results: According to histopathology 80 patients were stage I-II and 39 were stage III. Of the studied CT-criteria for lymph node metastases per patient, internal heterogeneity in at least one lymph node resulted in the best performance with sensitivity, specificity, PPV and NPV of 79, 84, 70 and 89%, Odds ratio (OR) 20. Presence of irregular outer border resulted in a sensitivity, specificity, PPV and NPV of 59, 81, 61 and 82%, OR 6.2. If both internal heterogeneity and/or irregular outer border was used as a criterion this resulted in a sensitivity, specificity, PPV and NPV of 85, 75, 62 and 91%, OR 16.5. None of the size criteria used were predictive for stage III disease. Conclusions: When performing preoperative CT in patients with colon cancer, the imaging criteria that allow best prediction of stage III disease on CT are either presence of at least one lymph node with internal heterogeneity or internal heterogeneity and/or irregular outer border. These criteria have to be validated in a prospective study.
CITATION STYLE
Rollvén, E., Abraham-Nordling, M., Holm, T., & Blomqvist, L. (2017). Assessment and diagnostic accuracy of lymph node status to predict stage III colon cancer using computed tomography. Cancer Imaging, 17(1), 1–8. https://doi.org/10.1186/s40644-016-0104-2
Mendeley helps you to discover research relevant for your work.