ABSTRACT: The aim of this study was to compare the gene expression profile changes breast tumors after the treatment with Anthracyclines and Taxanes. To this end, an oligonucleotide microarray was performed (Affymetrix’s HG-U133 Plus 2.0 array). This gene expression study was carried out on the biopsied tumor samples previous being treated with chemotherapy, and subsequently compared with themselves once treatment schedule ended. The post-chemotherapy biopsy was obtained from the surgical piece. The goal of this study was the finding of several genes related to apoptosis, proliferation, differentiation, survival and transformation-related genes and correlating their differences in expression with the degree of response to chemotherapy, determined by the Miller and Payne histological grading system. After informed consent, patients with a histologically confirmed diagnosis of breast cancer and scheduled chemotherapy treatment based on Anthracyclines and Taxanes (Treatment A: Epirubicin 90 mg/m2-Cyclophosphamide 600 mg/m2, 3 cycles bi-weekly and Taxol 150 mg/m2-Gemcitabine 2500 mg/m2, 6 cycles bi-weekly ± weekly Herceptin 4 mg/Kg during the first week, 2 mg/Kg for the remaining 11 cycles; Treatment B: Doxorubicin 60 mg/m2-Pemetrexed 500 mg/m2, 4 cycles tri-weekly and Taxotere 100 mg/m2, 4 cycles tri-weekly; Treatment C: Doxorubicin 60 mg/m2-Cyclophosphamide 600 mg/m2, 4 cycles tri-weekly and Taxotere 100 mg/m2, 4 cycles tri-weekly) were recruited for this study. Pre-chemotherapy and post-chemotherapy biopsies were examined by a pathologist who determined the Miller & Payne grade for each patient. Matching pairs of pre-chemotherapy and post-chemotherapy samples were divided into 3 groups according to Miller & Payne grade: group of bad response (Miller & Payne grades 1 and 2), group of mid response (Miller & Payne grade 3) and group of good response (Miller & Payne grades 4 and 5). Gene expression analysis was performed in paired samples as follows: bad response group post-chemotherapy biopsy vs pre-chemotherapy biopsy (Bad Final vs Initial). For this assay were necessary 13 samples being chosen according to histopathologic criteria (Miller & Payne grades 4 and 5). Of them, 10 samples were paired and 3 pre-chemotherapy samples from patients that experienced a good response to chemotherapy were also included in this experimental series. Other comparisons in which this group of samples was involved include: Initial Bad vs Good, Initial Good vs Mid, Final Mid vs Good and Final Bad vs Good. This gene expression profiling was carried out making use of Affymetrix’s GeneChip technology, with the Affymetrix’s HG-U133 Plus 2.0 array from this provider. All the protocols and apparatus were recommended by Affymetrix. Total RNA from frozen mammary tumors was directly extracted by a RNeasy Mini kit and homogenized by QIAshredder columns under manufacturer’s instructions. The quality and quantity of the obtained RNA, was checked out through agarose electrophoresis and later spectrophotometry at 260/280 nm. Biotinylated cRNA was synthesized following the IVT labeling kit from Affymetrix and purified by the GeneChip Sample Cleanup Module from Affymetrix. Once again, the quality and quantity of the obtained cRNA, was checked out through agarose electrophoresis and posterior spectrophotometry at 260/280 nm. After hybridization, slides were washed and scanned following the manufacturer’s standard protocol. Intensity values were normalized by Robust Multichip Average method and subsequently these were filtered for remove the control sequences and those with a hybridization signal near to background. The spike controls were: BioB, BioC, BioD and Cre; because BioB was the less presented in the samples, it was used to estimate the sensitivity of the experiment. The housekeeping control was GAPDH. After non-supervised PCA analysis and clustering, gene expression statistical significances were identified by two regression models taking into account the pathologic response to chemotherapy and if the sample was obtained before or after chemotherapy treatment. Supervised PCA analysis and clustering were performed with processed data. Partek Genomics Suite v7.3.1 (Partek) software was employed for the statistic analysis and clustering.
