In a prospective pilot study, we performed breast fine needle aspirations (FNAs) on 213 high-risk and 30 low-risk women and analyzed these aspirates for cytologic changes and biomarker abnormalities of aneuploidy and overexpressed estrogen receptor (ER), epidermal growth factor receptor (EGFR), p53 and HER-2/neu. High-risk women were those with a first degree relative with breast cancer (73%), prior biopsy indicating premalignant breast disease (26%), a history of breast cancer (13%), or some multiple of these risk factors (11%). Median ages of the high-risk and low-risk groups were 44 and 42, respectively. Sixty-three percent of the high-risk and 73% of the low-risk group were premenopausal. Sixty-eight percent of the high-risk and 17% of low-risk women had cytologic evidence of hyperplasia with or without atypia (P<.0001). Aneuploidy and overexpression of EGFR and p53 occurred in 25%, 36%, and 28% of high-risk subjects but in less than 4% of low-risk subjects (P<.0002). Overexpression of ER and HER-2/neu occurred in 8% and 19%, respectively of high-risk women; no low-risk women had these abnormalities. Sixty-eight percent of high-risk women and 7% of low-risk women had abnormalities of one or more of these biomarkers exclusive of cytology. Thirty-one percent of high-risk women, but no low-risk women had abnormalities of two or more biomarkers (P=.0004). Biomarker abnormalities were more frequent with increasing cytologic abnormality. Eighteen percent of women with normal cytology, 29% of women with epithelial hyperplasia and 60% of women with hyperplasia with atypia had abnormalities of two or more biomarkers (P=.048 and <.0001, respectively). Restricting the analysis to those three biomarkers most frequently overexpressed in the high-risk group (ploidy, EGFR, p53), 13% of high-risk women with normal cytology, 20% of high-risk women with epithelial hyperplasia and 51% of high-risk women with atypical hyperplasia had abnormalities of 2 or more of these 3 biomarkers. At a median follow up of two years, 8 of 213 women have been diagnosed with in situ (n=5) or invasive (n=3) cancer. Later detection of neoplasia was associated with prior FNA evidence of atypical hyperplasia (P <.0001) and multiple biomarker abnormalities in the 5 test battery (P=.006) by univariate analysis. By multivariate analysis, development and/or detection of cancer was primarily predicted by atypical hyperplasia (P=.0047) and secondarily by multiple biomarker abnormalities (P=0.021). Atypical hyperplasia, ECFR, and p53 in breast FNAs have promise as risk markers and as surrogate endpoint biomarkers for breast cancer chemoprevention trials. (C) 1997 Wiley-Liss, Inc.