Distinction of phyllodes tumor from fibroadenoma: a reappraisal of an old problem.

BACKGROUND: Using fine-needle aspiration (FNA) smears, it is difficult to distinguish low grade phyllodes tumor (PT) from fibroadenoma (FA) due to overlapping cytologic features between the two lesions. The authors retrospectively studied 45 histologically proven fibroepithelial breast tumors of which 33 were FA and 12 were PT (1 malignant, 8 borderline, and 3 benign) to define cytologic features that can help in the accurate categorization of these lesions by using FNA samples.
METHODS: The cytologic features analyzed included: 1) epithelial component for number (<5 or >5), architecture, apocrine metaplasia, squamous metaplasia, nuclear pleomorphism, and mitosis; 2) stromal fragments for number (<5 or >5), cellularity (on a scale of 1+ to 3+), borders, cell characteristics, nuclear pleomorphism, and mitosis; 3) individual dispersed stromal cells in the background for cellularity (on a scale of 1+ to 3+), and cellular shape (short/round/oval or long spindle) based on whether they were smaller or larger than 2 times the size of a small round lymphocyte.
RESULTS: The mean age of patients with FA was 34 years and of those with PT 44 years. The average size of FA was 2.0 cm, and the average size of PT was 4.0 cm. The characteristics of the epithelial fragments of PT and FA were not significantly different. Stromal fragments were noted in 60% of FA and 83% of PT samples examined. Fifty-six percent of PT and 30% of FA exhibited hypercellular stromal fragments (3+ cellularity), and the difference was not statistically significant. Large club-shaped hypercellular stromal fragments were present only in FA (in 21% of the samples). There was no difference in the overall cellularity of the background stromal nuclei in the two types of lesions. Long spindle nuclei averaging greater than 30% of the dispersed stromal cell population in the background were found only in cases of PT (in 57% of the samples; P < 0.001). Short/round/oval nuclei characterized most FAs. Long spindle nuclei constituting 10-30% of the dispersed stromal cells, however, occurred in both PT and FA to the extent of 43% and 21%, respectively.
CONCLUSIONS: Hypercellular stromal fragments occur not only in PT, but also in FA, and hence they cannot be used as the sole criterion for making a diagnosis of PT on FNA. The proportion of individual long spindle nuclei (>30%) amid the dispersed stromal cells in the background is the most reliable discriminator between the two lesions. Lesions in which long spindle nuclei constitute between 10% and 30% may represent either PT or FA, and therefore such lesions should be categorized as indeterminate on FNA.