Background: A probabilistic approach to the classification of fine-needle aspirates (FNAs) of the breast recently was recommended and received endorsement from the National Cancer Institute (NCI). In this system, FNAs are classified as benign, indeterminate/atypical, suspicious/probably malignant, and malignant, but to the authors' knowledge the use of these diagnostic categories has not been evaluated on a large scale. Furthermore, this classification scheme has not been applied to FNAs of nonpalpable lesions of the breast obtained under imaging guidance. Thus, the current study focused on whether the diagnostic categories could be applied usefully to ultrasound-guided FNAs (US-FNAs) of nonpalpable breast lesions.
Methods: Between 1988-1996, 1885 US-FNAs were performed on 1639 patients. The original FNA diagnoses were reclassified into the NCI-supported recommendations for diagnostic categories of breast FNAs. The cytologic findings were correlated with the tissue specimens, which were available in 851 cases, or with clinical follow-up of a minimum of 2 years in 127 of the 274 patients with benign solid lesions.
Results: The 1885 cases were categorized as follows: 1057 (56.1%) as benign, 86 (4.6%) as atypical, 79 (4.2%) as probably malignant, 502 (26.6%) as malignant, and 161 (8.5%) as unsatisfactory (defined as < 6 epithelial cell groups on all slides). The benign US-FNAs included 480 (45.4%) cysts and 577 (54.6%) solid lesions. Combined clinical and surgical follow-up showed that the frequency of malignancy was 3.7% in US-FNAs classified as benign, 52.9% in those designated as atypical, 75.8% in those designated as suspicious, and 98.9% in those classified as malignant. Based on combined histologic and clinical follow-up, a sensitivity of 97.1% and specificity of 99.1% were found for US-FNAs when definitive benign and malignant diagnoses were considered. A false-negative rate of 3.7% was attributed to sampling error. A false-positive rate of 0.68% was secondary to interpretative error of proliferative lesions.
Conclusions: Application of the NCI-supported diagnostic categories to US-FNA of nonpalpable breast lesions is useful in stratifying aspirates based on the likelihood of underlying malignancy. The subcategories of US-FNAs diagnosed as atypical have similar probabilities of malignancy; this justifies their being grouped as a single category wherein tissue biopsy would be required to exclude carcinoma. Benign and inadequate FNA diagnoses must be correlated with the clinical and imaging findings and in noncorrelative cases the patient should undergo biopsy. US-FNA is a sensitive and specific means with which to diagnose nonpalpable breast lesions.