Accuracy and consistency in application of a probabilistic approach to reporting breast fine needle aspiration

To determine the application of a probabilistic/categorical approach for reporting breast fine needle aspiration (FNA) and its dependence on the cytopathologist's level of experience.

All breast surgical specimens that had preoperative breast FNA at our institution during a 3-year period were identified. The cytologic results were reported as 1 of 6 categories: positive, suspicious, atypical, epithelial proliferative, unremarkable and nondiagnostic, according to well-defined criteria. Five cytopathologists were responsible for all cytology sing-out during the study period. The histologic and cytologic diagnoses were correlated.

A total of 297 cases were identified. Overall, there were no false positive cases (positive predictive value [PPV] = 100%). Two false negative cases (negative predictive value [NPV] = 96%) were due to sampling error. This indicates that the PPV and NPV for each of the 5 pathologists were also all 100% except for the 1 pathologist who had two false negative cases due to sampling errors. The probability of finding carcinoma on histology for suspicious and atypical cytologic categories ranged from 67% to 100% and 8% to 31%, respectively, for the individual pathologists. Fifteen cases were signed out by > or = 2 pathologists. The involvement of consultants was significantly associated with diagnosis (P = .02). Ten of the 15 cases were in the suspicious (5) or atypical (5) category.

The probabilistic approach with defined diagnostic criteria is an accurate method and can be consistently applied in reporting breast FNA. Although use of the indeterminate (suspicious and atypical) categories is variable, a definite and considerable difference in the probability of carcinoma between these 2 categories was observed for all pathologists. The involvement of consultants did not move the cases out of these indeterminate categories.