Abstract
Context.-Pulmonary bronchial brushing specimens can be processed by liquid-based or conventional methods. The ability to accurately diagnose a pulmonary malignancy with a liquid-based preparation (LBP) versus a conventional preparation may differ.
Objective.-To compare the performance of LBPs of malignant pulmonary bronchial brushing specimens with the performance of conventional preparations.
Design.-Participant responses from 553 malignant pulmonary bronchial brushing samples were evaluated for concordance with the general diagnosis. The performance of LBPs was compared with that of classic preparations. A nonlinear mixed model was used to analyze the performance by reference diagnosis, preparation type, program years, participant type, and the interaction terms between these 4 factors.
Results.-Concordance with the general category of malignant disease was observed in 95.2% of conventional Papanicolaou-stained, 90.9% of modified Giemsa-stained, and 96.9% of LBP (P<.001) samples. The results were significantly different between individual reference diagnoses (P<.001). The performance of LBPs was consistently higher for most diagnoses and was significantly better for squamous cell carcinoma (P=.005), small cell carcinoma (P<.001), and metastatic carcinoma not otherwise specified (P<.001). All participant types performed significantly better with LBPs of small cell carcinoma. Pathologists and cytotechnologists performed significantly better with LBPs of squamous cell carcinoma. A significantly higher concordance was observed between the general diagnosis and program years 2007-2011 versus 2001-2006 (P=.006).
Conclusions.-Liquid-based preparations performed better than conventional methods, with significantly higher performance in squamous cell, small cell, and metastatic carcinomas. Improved performance over time may reflect more frequent use of LBP methods and increased familiarity with interpreting the morphologic findings.