Improving clinical efficacy of computed tomographic scan in the preoperative assessment of patients with non-small cell lung cancer.

The criterion of choice for computed tomographic scan identification of metastatic mediastinal nodes is not clearly fixed. This prospective study was designed to define the most suitable computed tomographic criterion for detection of nodal metastasis, enabling improvement of the test's clinical efficacy. One hundred twenty-three patients with potentially operable non-small cell lung cancer underwent mediastinal evaluation by computed tomographic scan and cervical mediastinoscopy followed by thoracotomy with mediastinal node dissection. There were 116 men and seven women; the mean age was 59.3 +/- 9.1 years. Forty-six tumors were classified after operation as stage I, 20 as stage II, 27 as stage IIIa, and 30 as stage IIIb. Mediastinal nodes were classified as metastatic according to the following computed tomographic scan criteria: (1) shorter axis 1 cm or larger; (2) shorter axis 1.5 cm or larger (nodes less than 1 cm were classified as negative and those 1 to 1.5 cm as indeterminate); and (3a) shorter axis 1 cm or larger, plus evidence of central necrosis or discontinued capsule, or (3b) shorter axis 2 cm or more, regardless of the nodal morphologic condition. The highest sensitivity rate was achieved by using criterion 1 (90%) and the poorest by criterion 3 (75%). The greatest specificity rate was obtained by applying criterion 3 (90%) and the lowest by criterion 1 (54%). The prediction by using computed tomographic criterion 3 correlated better with pathologic findings than that derived by adopting the criterion 1 or 2. When mediastinal nodes were identified as negative according to criterion 1, 2, or 3, the complete resection rate was 92%, 92%, or 95%, respectively, rendering cervical mediastinoscopy unnecessary. When mediastinal nodes were classified as positive, the resectability rate was 55%, 27%, or 13%, respectively. In these instances cervical mediastinoscopy allowed identification of different degrees of mediastinal involvement; it proved to be the most useful procedure for preoperative selection of those patients with N2 tumors who are amenable to a complete resection. In conclusion, the use of computed tomographic criterion 3 does improve the clinical efficacy of the test, by sparing a large number of unnecessary mediastinal explorations, without increasing the rate of useless thoracotomies.