Purpose: We hypothesized that comprehensive breast cancer methylation profiling might provide biomarkers for diagnostic assessment of suspicious breast lesions using fine needle aspiration biopsy (FNA).
Experimental design: Twenty-three gene promoters were surveyed by quantitative methylation-specific PCR in bisulfite-modified DNA from 66 breast carcinomas (BCa), 31 fibroadenomas (FB) and 12 normal breast (NT) samples to define a set of genes differentially methylated in malignant and non-malignant tissues. This set was tested in 78 FNA washings obtained pre-operatively (66 malignant, 12 benign), with histopathological diagnosis. Receiver operator characteristic (ROC) curve analysis identified a gene panel which might distinguish cancer from non-cancerous lesions. Finally, this panel was validated in an independent series of FNA washings (45 cases) in which cytomorphology did not reach definitive diagnosis.
Results: In tissue samples, 14-3-3-sigma, DAPK, CCND2, RASSF1A, CALCA, APC, HIN1, RARbeta2, TIG1, and GSTP1 methylation levels differed significantly among BCa, FB, and NT. ROC curve analysis identified a panel of four gene loci (CCND2, RASSF1A, APC, and HIN1) that discriminated BCa from benign lesions in a set of 78 FNA washings from histologically characterized breast lesions. When this panel was tested in the validation dataset of 45 FNA washings, breast cancer was identified with perfect specificity (100%) when 3 of 4 gene loci tested positive, providing estimated added information of 91% over cytomorphologic evaluation alone.
Conclusions: Our data provide evidence that multigene methylation analysis augments diagnostic accuracy of cytological assessment of suspicious breast lesions, and might be a valuable ancillary tool for breast cancer diagnosis.