Project description:Breast cancer is a heterogeneous disease described in well-recognized biological subtypes. Particularly, gene expression profiling has revealed 5 intrinsic subtypes of breast cancer characterized by different biological and clinical features. The diversity of the intrinsic subtypes across human population has been limited described, and there is few information about the genomic architecture of breast tumors in Mexican or Hispanic populations. In this study, we performed PAM50 assay, based in Affymetrix microarray profiling of 128 fresh frozen tumors from Mexican Latino Hispanic population, to describe the overall distribution of subtypes, and characterize the relation to clinicopathologic characteristics. As well, we correlated the mRNA expression patterns with specific copy number alterations (CPA), in order to analyze their role in breast tumors. A total of 100 blood-tumor samples were assayed using Affymetrix 6.0 SNP arrays; segmentation analysis and GISTIC were performed to identify focal amplifications or deletions. The distribution of PAM50 intrinsic subtypes in our cohort was computed to be 44% luminal A, 20% luminal B, 12.0% HER2-enriched, 12% basal-like, and 12% normal-like. Study comparison with the literature mainly TCGA and METABRIC (most of the patients came from Caucasian population), as well as LACE (which describe a population study) show a similar distribution of the intrinsic subtypes within Hispanic and Caucasian populations. Interestingly, basal-like subtype is less represented than in African-American race. The sum of sensitivity and specificity between the clinicopathologic and intrinsic subtype categories across 4 groups (excluding normal-like) was 50% and 87.5%, respectively. Differentially expression profiles within the subtypes reveal a set of genes altered in each group with biological relevance to stablish the phenotypical characteristics of each subtype. Our analyses confirmed the already reported copy number data. Importantly, many of the copy number profiles correlated with mRNA subtype. With this analysis we can conclude that breast cancer intrinsic subtypes have been reproduced in Mexican population contributing to the description of the PAM50 subtypes among multiple ethnic groups based on a gene expression assay. Our observation based in the integrative genomic analysis of mRNA expression and CPA allowed us to define gene circuits and phenotypic characteristics that can explain the heterogeneity of breast cancer subtypes.

Project description:The experiment was designed to enable validation of pre-processing and test algorithms. The in-house produced cDNA-array contains 44 clones: 19 clones from mice, 4 clones from Pine (involved in photosynthesis) and 21 artificial clones from the Lucidea Universal ScoreCard (Amersham Bioscienses), where each clone was printed 480 times in 48 identically designed sub-grids. The experiment contains eight arrays with identical experimental design. Total RNA from murine cell line was divided in two samples; the reference sample was labeled with the fluorophore Cy5 and spiked with the Lucidea Universal ScoreCard reference reaction, the test sample was labeled with Cy3 and spiked with the Lucidea Universal ScoreCard test reaction. The concentrations of the Lucidea reactions are known and consequently so are the true ratios between the reference and test channels. Approximately 40% of the artificial genes were differentially expressed. The particular design of the experiment makes the data suitable for validating algorithms for pre-processing and tests for identifying differentially expressed genes. - The 80.I.1 normalization algorithm : all spots flagged as not found during image analysis considered missing values (complete filtration). Print-tip MA-loess dye normalization based on the calibration clones was applied to the raw data and the B-statistic was calculated. - The 80.II.1 normalization algorithm : all spots flagged as not found during image analysis considered missing values (complete filtration). Local background correction was applied and all spots with negative values were excluded. Print-tip MA-loess dye normalization based on the calibration clones was applied to the background corrected data and the B-statistic was calculated. - The 80.II.64 normalization algorithm : all spots flagged as not found during image analysis considered missing values (complete filtration). Local background correction was applied, negative values were excluded and all spots with a value below 64 were set to 64 (censored). Print-tip MA-loess dye normalization based on the calibration clones was applied to the background corrected data and the B-statistic was calculated. - The 80.V.1 normalization algorithm : The intensities of all spots flagged not found were treated as missing values during normalization, but prior to calculating test-statistics the spot's log-ratios were set to zero. In the special case when all arrays generated not-found spots, the gene was removed from the experiment (Partial filtration). Local background correction was applied and all negative values were considered missing values. Print-tip MA-loess dye normalization based on the calibration clones was applied to the background corrected data and the B-statistic was calculated.

Dataset Information

Prototypes for 4x44K PAM50 Subtyping

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets