clusters necessary to ensure that adequate cellular material was present for accufornia, Los Angeles, California.
rate diagnosis. The series included 21 cases cytologically diagnosed as false-negative, 75 cases that had been correctly identified as benign, 47 cases cytologically designated as atypical, and 40 cases that on initial review had been correctly identified as malignant. In semiblind fashion, the smears from each case were assigned to low, medium, and high cellularity categories. Low cellularity was defined as 10 or fewer cell clusters, moderate cellularity was defined as 11-30 clusters, and high cellularity was defined as more than 30 clusters. A cell cluster was defined as five or more cells. Within the low cellularity group, exact numbers of cell clusters and the presence of individual cells were recorded. The presence of bipolar cells was used as an adjunct criterion for specimen adequacy, and the bipolar cells in each of 10 1 200 fields were counted. Cellularity was then correlated with diagnostic accuracy.
RESULTS.
Using a cutpoint of a cumulative score of 6 or more cell clusters or the prominence of bipolar cells (ยข10 in each of 10 medium-power, 1200 fields) for assessment of specimen adequacy, a false-negative rate of 1.5%, associated with an unsatisfactory rate of 20.2%, was obtained.
CONCLUSIONS.
Based on the data gathered in this study, the authors believe that the sampling false-negative and unsatisfactory rates can be minimized by selecting a cutpoint for satisfactory smears at a level of 6 or more cell clusters (cumulative total) or the presence ยข 10 intact bipolar cells per 10 medium-power fields (1200).
Use of these criteria will decrease the false-negative rate of sampling in epithelial lesions of the breast. A false-negative rate of approximately 1.5% was obtained in association with an unsatisfactory rate of 20.2%. Using a cutpoint of 1 or more cell clusters, a false-negative rate of 2.1%, associated with an unsatisfactory rate of 13.7%, was obtained. [See editorials on pages 1-2, 3-5, this issue.] Cancer (Cancer Cytopathol) 1997;81:16-21.