Biomarkers distinguishing mammary fibroepithelial neoplasms: a tissue microarray study.

OBJECTIVE

To determine whether any markers in biopsy specimens were useful for distinguishing (1) fibroadenoma (FA) from benign phyllodes tumors (PTs); and (2) from benign borderline PTs of the breast.

METHODS

Specimens of 80 breast tumors (20 FA, 38 benign, 12 borderline, and 10 malignant PTs) diagnosed at Tri-Service General Hospital from 1986 to 2006 and 10 normal breast tissue were investigated immunohistochemically for the expression of 11 biomarkers including p53, Ki-67, topoisomerase IIα, p16, retinoblastoma protein (pRb), fascin-1, estrogen receptor-β, CD117, osteopontin, hypoxia-inducible factor-1α, and cyclooxygenase-2. The binary logistic regression method was used to generate functions that discriminate between benign and borderline PT and also between FA and benign PT.

RESULTS

On the basis of the most active area of stained stromal cells, the staining intensity, and the immunoscore, the expression of Ki-67, topoisomerase IIα, p16, and pRb was significantly higher in borderline or malignant PTs than in benign PTs. Ki-67 could discriminate benign from borderline PTs singly with a high sensitivity (91.7%), specificity (100%), and accuracy (98%). In addition, expression of Ki-67, p16, and pRb was significantly higher in benign PT than in FA. Binary logistic regression identified p16 and pRb as the only marker combination capable of distinguishing FA from benign PTs with sensitivity (94.7%), specificity (75%), and accuracy (87.9%).

CONCLUSIONS

Ki-67 may be a useful marker for discriminating benign from borderline PTs, and p16 and pRb may be a useful combination of markers for distinguishing FA from benign PTs in core biopsy specimen.