Computerised tomography does not predict N2 disease in patients with lung cancer.

Background: Mediastinal node involvement in primary lung cancer determines the staging and prognosis of the patient, and as these nodes can be seen on the computerised tomography (CT) scan of the chest it is a temptation to diagnose malignant involvement if the nodes appear enlarged. However, initial experience with mediastinal node mapping at lung resection demonstrated this extrapolation to be unreliable and misinterpretation of enlarged nodes on CT may lead to misdiagnosis and prejudice the patient's management.

Aim: To demonstrate that the sensitivity, specificity, and accuracy of the CT to detect malignant mediastinal nodes is too low to use size of node on CT as representative of malignant involvement.

Methods: One hundred and fifty-three sequential patients with resectable lung cancer were studied with preoperative CT. Two radiologists determined the preoperative T and N status from these studies with nodes of 1.5 cm or larger diagnosed abnormal. These results were compared to the results of subsequent node mapping performed after lung resection.

Results: Sensitivity was found to be 26%, specificity to be 81% and overall accuracy 69%--too low to justify the diagnosis of N2 disease on size of 1.5 cm or larger. CT is not a valid means of diagnosing malignant involvement of mediastinal nodes.