Technical accuracy and clinical efficacy of thoracic computed tomography

The role of CT and MRI in staging of the mediastinum

Lung Cancer Study Group (LCSG) Protocol 883, the comparative study of the results of magnetic resonance imaging (MRI) and computerized tomography (CT) for staging of tumor, nodal, and selected metastatic sights in patients with surgically staged lung cancer was activated in August 1988 but was not completed because of termination of LCSG funding. A literature review was therefore undertaken to determine the results of other studies that were performed to evaluate the relative efficacy of MRI and CT in the staging of patients with lung cancer. These studies determined that CT and MRI are approximately equal in the staging of N2 disease with a sensitivity of 70 to 90%, a specificity of 60 to 90%, and an accuracy of 66 to 90% depending on the criteria used for determining positive nodes and the compulsiveness of surgical staging. Magnetic resonance imaging is probably better in the assessment of superior sulcus tumors, tumors involving the aorta-pulmonary window, hilar nodes, in assessing chest wall or diaphragmatic invasion, and in evaluating patients whose CT findings are equivocal. Computed tomography and MRI reveal adrenal abnormalities in 10 to 20% of patients but only one third of these have metastases. Mediastinoscopy has a sensitivity of 85 to 90%, a specificity of 100%, and an accuracy of about 95% and is therefore the gold standard for N2 staging. If the CT examination reveals no N2 disease, one can proceed directly to thoracotomy with approximately a 15% chance of finding N2 disease. It was concluded that because CT is much cheaper, it should therefore be used for the noninvasive staging of patients with lung cancer unless the above-noted special circumstances are present that have been shown to favor MRI. Because of the limited accuracy of CT and MRI, however, positive findings must be confirmed by biopsy specimens and pathologic study.

Can computed tomography of the chest stage lung cancer? Yes and no

To determine the accuracy of computed tomography (CT) of the chest in the staging of lung cancer, we studied 418 patients with primary pulmonary carcinoma between 1979 and 1986. Each had a preoperative scan performed before detailed operative staging. Each CT scan was analyzed for components of the current TNM staging system. Computed tomography sensitivity and specificity for mediastinal lymph node metastasis were 84.4% and 84.1%, with corresponding positive and negative predictive accuracies of 68.7% and 92.9%, respectively. When TNM stages were derived from CT scans, only 190 of 418 (45.4%) completely agreed with operative staging. An additional 53 of 418 (12.7%) predicted the correct stage, although components of the TNM system were incorrect. In 94 of 418 scans (22.5%) CT overestimated the stage, whereas in 81 (19.4%) CT downgraded the stage. Computed tomography suggested metastatic lesions in liver, lung, adrenal gland, bone, or abdominal lymph nodes in 40 of 373 scans (10.7%); only five of 40 (12.5%) had documented metastasis. In summary, CT of the chest cannot accurately stage primary lung carcinoma according to the TNM classification. Because the negative predictive accuracy for mediastinal lymph node metastasis remains high (92.9%), invasive staging can be deferred for definitive thoracotomy when no lymphadenopathy is evident on CT. The high negative predictive accuracy for scans of the chest and upper abdomen makes CT a useful tool for exclusion of metastatic disease.

Intrathoracic lymph node evaluation by CT and MRI with histopathologic correlation in non-small cell bronchogenic carcinoma

Twenty patients with non-small cell bronchogenic carcinoma were prospectively studied for intrathoracic lymphadenopathy using computed tomography (CT) and magnetic resonance imaging (MRI). The CT and MRI results were correlated with the surgical histopathology results of 103 harvested lymph nodes. Based on node dimension (utilizing a 1-cm cutoff value), neither imaging modality was accurate in predicting the histopathology of the nodes. CT had a sensitivity of 21%, specificity of 95%, and an overall accuracy of 85%; MRI had a sensitivity of 14%, specificity of 97%, and an overall accuracy of 85%. Moreover, CT and MRI detected preoperatively only 17% and 14%, respectively, of the harvested nodes. In summary, lymph node sizes measured by CT and MRI are inaccurate in predicting the benignity or malignancy of lymph nodes. Lymphadenopathy can be present even with a "negative" CT or MRI scan of the thorax.