Abstract B35: Defining multiethnic genome-wide transcriptional signatures in normal and cancerous breast tissue using paraffin-embedded samples

Background and Rational: Ethnic-specific disparities in breast cancer (BC) stage of presentation and survival rates are well documented. To further investigate possible ethnic-specific genetic contributions to these disparities, we are completing gene expression profiling studies in a South Florida multiethnic cohort consisting of thirty “Triple Negative” BC patients [10 each African-American (AA), His (His) and non-Hispanic white (Cauc) women] matched for age of diagnosis and hormone receptor status, as well as in a cohort of “Triple Negative” patients from West Africa. For comparison purposes, we are conducting similar analyses in bona fide normal breast tissue surgical specimens from Cauc and AA women. The overall study aim is an increased understanding of the biological basis of ethnic-specific BC disparities, leading ultimately to individualized, ethnic-specific diagnostic and therapeutic approaches. Two immediate study goals are to demonstrate the utility of FFPE samples in obtaining consistent, reproducible data from gene expression arrays, and secondly, to identify differentially expressed genes between tumor and normal breast tissue that are common or unique among the three ethnic groups.

Methods: Pathology specimens were freshly cut from FFPE blocks and marked by a pathologist as to normal vs. tumor tissue. RNA isolation, labeled cDNA preparation, and hybridization of tumor and normal cDNAs to a breast focused gene expression microarray (Breast Cancer DSA Research Tool) was performed by Almac Diagnostics. Each South Florida patient had self-matched (tumor vs. normal) tissues for gene expression studies.

Results: Using 36 matched tumor and normal FFPE samples from 18 patients, approximately 17516 transcripts were detected on the Breast Cancer DSA with intensity significantly greater than background. For normal and tumor tissue samples, 9399 and 10,296 transcripts respectively, were detected in all three ethnic groups. Importantly, a subset of transcripts (hundreds to one thousand) was detected in only one or two ethnic groups. Using two-way ANOVA (disease state and ethnicity), a subset of 6479 transcripts was identified with p-value less than 0.01 in the statistical test and was selected and further used in data quality control. Data QC indicated that patient samples clustered well with respect to both ethnicity and normal versus tumor tissue. Additional analytical methods included K-means 2-Dimensional clustering and Principal Component Analysis. From these analyses, we identified ethnic-specific expression patterns in the matched normal and tumor tissue samples, which are being validated by qPCR, as well as through comparisons to data derived from the ethnically-matched normal tissues. We are refining these current studies by focusing further data analyses on lymph node negative triple negative samples, and through comparisons with similar study data from the native African samples.

Summary: These analyses indicate that consistent, high quality gene expression data can be generated from FFPE samples, and that even with a small sample size, ethnic specific gene expression differences can be detected in tumor and matched normal breast tissue samples across ethnic groups. Once validated, these results have vast future implications for addressing breast cancer health disparities.