A prospective analysis of touch preparation cytology for intraoperative detection of mediastinal lymph node metastases

Abstract

Patients diagnosed with lung cancer may require immediate evaluation of mediastinal lymph nodes to determine treatment plan. Typically, frozen section (FS) analysis has been used, but this analysis can be time-consuming and uses more tissue than touch preparation (TP) cytologic analysis. TP accuracy has been studied in other organs, but no prospective studies comparing TP to FS have been performed on mediastinal lymph nodes in lung cancer. Our goal was to compare the accuracy of TP to FS in these cases. After obtaining institutional review board approval, all patients undergoing mediastinal lymph node evaluation for a diagnosis of lung cancer were asked to participate. If consent was given, TP and FS analyses were performed on all mediastinal lymph node stations in all patients and compared to permanent hematoxylin and eosin analysis. Data were collected prospectively. Twenty patients were enrolled. Mean age was 67.7 years. Fifty-five percent (11 of 20) of patients were men. The mean number of lymph node stations sampled in each patient was 3.4. In predicting the stage of the patient, TP had a sensitivity and specificity of 95% and 100%, respectively. FS had a lower sensitivity, 85%, and a specificity of 100%. On permanent analysis, metastatic foci ranged in size from 0.15 mm to 1.5 mm. TP was more sensitive than FS in detecting mediastinal lymph node metastases. The technical difficulty of obtaining full-thickness sections without creating significant artifact may contribute to the lower sensitivity of FS in detecting micrometastases.