MicroRNA Signature for HER2-positive Breast and Gastric Cancer

Materials and Methods

Discovery sets. Discovery set samples were collected at the time of surgery from patients with cancer at the National Cancer Center, Keimyung University Dongsan Hospital, and Asan Medical Center in Korea from 2001 to 2012. Specimens were collected with Institutional Review Board approval (NCCNCS12581) and all patients gave informed consent, signing Institutional Review Board approved forms. A 10 μm-thick top slide was stained with hematoxylin and eosin. Guided by this top slide, remaining tissue was macrodissected to trim non-tumorous stromal components. Macrodissected, frozen tissue sample was mechanically crushed in liquid nitrogen and subjected to total RNA and genomic DNA isolation using mirVana™ Kit (Ambion, Austin, TX, USA) and DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA, USA), respectively. DNAse I-treated RNA (500 ng) was poly-A tailed, and FlashTag Biotin HSR Labeling Kit (Genisphere LLC, Hatfield, PA, USA) was used to join a biotin-labeled dendrimer molecule to the 3’ end using DNA ligase. Labeled samples were hybridized to GeneChip miRNA 2.0 microarrays (Affymetrix, Santa Clara, CA, USA) based on miRbase version 15. All cell files were robust multiarray average-normalized. After filtering out star-form microRNAs, we subjected 913 human microRNAs to further analyses for this study. BRB-Arraytools software (version 4.3; National Cancer Institute, Bethesda, MD, USA) was used to carry out unsupervised and supervised microRNA analyses. Principal component analyses were performed using 1–correlation as a distance metric. Copy number states of gastric carcinomas were determined using SNP6 arrays and Genotyping Console 4.1 (Affymetrix). In breast carcinomas, HER2 immunohistochemistry results were scored as 0-3+ according to the method recommended for the HercepTest (Dako, Glostrup, Denmark) (5). Breast carcinomas with scores of 3+ or ERRBB2 gene amplification by fluorescence in situ hybridization were identified as being HER2-positive.

Validation sets. To validate these candidate HER2-related microRNAs in breast carcinomas, we downloaded microRNA datasets for HER2-positive breast cancer from the gene expression omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/; GSE40525 and GSE39543). ArrayTools was used to normalize the data to the median array and to evaluate the differential expression of the HER2 microRNA signature in each GEO dataset. Since there were no public microRNA datasets for HER2-positive gastric cancer, we used our own microRNA data from endoscopic biopsy of 14 patients with metastatic gastric cancer (6). These microRNA microarray data were generated using custom-synthesized 8×15k microRNA microarrays containing 4,361 microRNAs (Sanger miR 9.0 database) (Agilent Technologies, San Jose, CA, USA). A mixture of total RNA isolated from three gastric cancer cell lines (SNU-601, SNU-638, and AGS; Korean Cell Line Bank, Seoul, Korea) was used as the reference RNA for competitive hybridization. Copy number status of the 14 gastric cancer samples was previously assessed using Agilent 4×44k HD-CGH microarray (6). Based on the ADM-2 algorithm of Agilent's CGH Analytics software, the aberration filter was set at a minimum of five probes in the region, and a log ratio that was equal to the DLRSpread (the spread of the ratio differences between consecutive probes) of each sample (7).