BACKGROUND: Computed tomography (CT) has been widely used for preoperative mediastinal lymph node evaluation in lung cancer. But its accuracy has remained controversial. We studied the predictability of N-staging by CT scan. METHODS: From 1981 to 1996, 546 patients had preoperative CT scan and underwent a surgical resection with mediastinal lymph node dissection for primary pulmonary adenocarcinoma and squamous cell carcinoma. Nodes larger than 1.0 cm at short axis were considered abnormal. The discrepancy between clinical and pathologic findings in N factor was analyzed. RESULTS: Sensitivity (ST), specificity (SP) and accuracy (AC) were 33.0%, 90.4% and 78.9%, respectively. No statistically significant difference in the results is detected for individual years or types of scanning device. There were statistically significant differences as follows: ST and SP by histologic type, SP by gender, SP and AC by tumor size, SP by Brinkman index, ST by tumor location, and AC by serum CEA value. CONCLUSIONS: We should pay attention to false positive nodes in heavy smokers (or males), and positive nodes in adenocarcinoma, tumor larger than 3 cm or rising of serum CEA value, regardless of negative lymph node on CT scan.
BACKGROUND: Computed tomography (CT) has been widely used for preoperative mediastinal lymph node evaluation in lung cancer. But its accuracy has remained controversial. We studied the predictability of N-staging by CT scan. METHODS: From 1981 to 1996, 546 patients had preoperative CT scan and underwent a surgical resection with mediastinal lymph node dissection for primary pulmonary adenocarcinoma and squamous cell carcinoma. Nodes larger than 1.0 cm at short axis were considered abnormal. The discrepancy between clinical and pathologic findings in N factor was analyzed. RESULTS: Sensitivity (ST), specificity (SP) and accuracy (AC) were 33.0%, 90.4% and 78.9%, respectively. No statistically significant difference in the results is detected for individual years or types of scanning device. There were statistically significant differences as follows: ST and SP by histologic type, SP by gender, SP and AC by tumor size, SP by Brinkman index, ST by tumor location, and AC by serum CEA value. CONCLUSIONS: We should pay attention to false positive nodes in heavy smokers (or males), and positive nodes in adenocarcinoma, tumor larger than 3 cm or rising of serum CEA value, regardless of negative lymph node on CT scan.