Objective: The aim of our study was to compare the assessment of peripancreatic lymph nodes using CT with the gold standard of detailed histopathologic assessment of resected specimens in patients with pancreatic ductal adenocarcinoma.
Subjects and methods: Sixty-two patients with presumed pancreatic carcinoma were prospectively studied with dual-phase contrast-enhanced helical CT, and images were interpreted in consensus by three radiologists. Complete surgical resection was performed in 28 patients. A detailed nodal classification system was used for radiologic, surgical, and pathologic staging in the nine patients whose final diagnosis at histology was pancreatic ductal adenocarcinoma.
Results: Forty lymph nodes were prospectively identified on CT in these nine patients. Two of 23 nodes (9%) measuring less than 5 mm in the short-axis diameter were malignant, four of 11 nodes (36%) measuring 5-10 mm were malignant, and one of six nodes (17%) larger than 10 mm was malignant. Using a short-axis diameter of greater than 10 mm as the criterion for nodal involvement, we found a sensitivity of 14% (1/7) and a specificity of 85% (28/33), with a positive predictive value of 17% (1/6), a negative predictive value of 82% (28/34), and an overall accuracy of 73% (29/40). Ovoid nodal shape, clustering of nodes, and the absence of a fatty hilum were not useful predictors of malignancy on CT.
Conclusion: In resectable pancreatic ductal adenocarcinoma, CT is not accurate overall for the prediction of nodal involvement. In a patient with presumed pancreatic carcinoma that is considered to be resectable, the depiction on CT of peripancreatic nodes should not prevent attempted curative resection.