A Comparison of Withdrawn and Nonwithdrawn Brushes in the Diagnosis of Lung Cancer

Abstract

Brush specimens are one of several methods for establishing a diagnosis of lung cancer at the time of flexible bronchoscopy. Two acceptable methods of retrieving the brush cytology specimen are to withdraw the brush through the working channel of the bronchoscope and to withdraw the brush and bronchoscope as a unit, with the brush protruding through the distal tip of the bronchoscope. In this prospective study 53 patients with proven primary bronchogenic carcinoma underwent bronchoscopy by one senior staff pulmonary physician; there were 32 patients with peripheral tumors and 21 patients with central tumors. The sequence of sampling (withdrawn or nonwithdrawn) was assigned randomly. The specimens were interpreted by one cytopathologist in a blinded fashion and scored for both cellularity (1–4 according to the density of cells and number of quadrants involved) and presence or absence of recognizable malignant cells. Cellular yield was significantly greater for nonwithdrawn brushes [difference = 1.43 ± 1.21 for central tumors (p < 0.001), and difference = 0.65 ± 1.26 for peripheral tumors (p < 0.01)]. Malignant cells were identified in 66.7% (14/21) of withdrawn versus 71.4% (15/21) of nonwithdrawn brushes (p > 0.25; McNemar's test for matched pairs) for the central tumors. Malignant cells were identified in 53.1% (17/32) of withdrawn brushes versus 59.4% (19/32) or nonwithdrawn brushes (p > 0.25) for the peripheral tumors. This lack of a statistically significant difference could be the result of a type 2 error (inadequate sample size to detect a true difference). There were qualitative differences in the cellular preservation and nuclear details in favor of the nonwithdrawn brush specimens. We conclude, however, that withdrawing the brush through the bronchoscope decreases the cellularity of the specimen, but it does not affect the diagnostic accuracy for malignancy in either central or peripheral tumors.