Expression Profiling, Mammaprint Assay
TERMINOLOGY
Abbreviations
MammaPrint breast cancer assay (MPA)
Definitions
Gene expression profiling (GEP) studies have added greatly to our understanding of biologic diversity of breast cancer
Expression array studies allow simultaneous examination of global changes in gene expression from clinical breast cancer samples
Target Gene or Antigen
GEP studies allow simultaneous examination of thousands of genes without need to define function or relationship among genes
Application to breast cancer has potential to provide important insights into tumor biology and clinically meaningful tumor subtypes
Large data sets generated by these studies require computer and statistical analysis to look for biologically meaningful patterns
Statistical and computer analysis need to identify groups of genes (gene signatures) that correlate with clinical outcome
Types of Tests
GEP of clinical breast cancer samples using cDNA microarray
Requires fresh or snap frozen tissue samples from a cohort of breast cancer patients with outcome data
RNA isolated from tumor tissue for GEP analysis
Supervised Analysis of GEP
Goal of supervised classification of breast cancer using GEP data
Detect gene expression patterns that are predictive of outcome in clinically well-defined patient cohorts
Initial step: Separate breast cancer patients in the large patient cohort into clinically defined subsets or groups
Different patient groups usually defined by clinical outcome
Disease recurrence = poor outcome
No disease recurrence = good outcome
Mathematical models and statistics used to identify gene sets or “genomic classifiers”
Should predict which outcome group each patient belongs to (“class prediction”)
Validation of gene sets for classification of outcome must be done in additional patient cohorts to help establish clinical utility
Validation of assay results in additional patients important to help exclude “false discovery”
GEP studies are associated with high “false discovery” due to larger number of genes evaluated in limited number of patients
Numbers of genes tested is quite large compared with number of clinical samples used to create these models
Potential for high “false discovery” rate for genes that appear to be correlated with outcome
MammaPrint (70 Gene Prognostic Panel)
MPA developed by investigators from the Netherlands Cancer Institute
MPA: Test development
1st cohort used frozen tumor tissue from 98 patients with node-negative invasive breast cancer
RNA from individual cases was hybridized to 25,000 gene microarray
Preliminary unsupervised statistical clustering analysis
5,000 genes showed significant variation between different breast cancer cases
Tumor samples could be separated into 2 general classes of tumor
1 class correlated with mostly ER(+) tumors
1 class correlated with mostly ER(-) tumors
Statistical methods that incorporated clinical outcome data were then applied
Patients separated into 2 groups based on outcome
Relapse within 5 years (34 patients) and disease free at 5 years or more (44 patients)
Identified 231 genes whose expression was correlated with outcome
Further statistical methods identified 70 genes as optimal number to use to develop GEP prognostic classifier
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
