BluePrint™: Development and Validation of the Breast Cancer Subtyping Profile

Classification of breast cancer into molecular subtypes may be important for accurate selection of therapy for patients as tumors with a seemingly similar biology but different molecular characteristics can have different outcomes. BluePrint is a multi-gene profile which has been developed for the classification of breast cancer into molecular subtypes. The profile separates tumors into Basal-type, Luminal-type and ERBB2-type subgroups. The addition of this molecular subtyping profile will further complement MammaPrint^® by providing further improvement of outcome prediction.

Development

The BluePrint signature determines the RNA levels of 80 genes that best discriminate among these three distinctive subtypes. Tumors from an initial cohort of 295 patients^(1,2) were used for the development of gene expression profiles specific for the Basal-type, Luminal-type and ERBB2-type breast cancers.

Using state-of-the-art bioinformatics tools, Agendia identified genes whose expression ratios best discriminate between the three subgroups. Subtype specific gene expression profiles were identified in a 3-fold cross-validation procedure. Optimal classification of the training samples in the corresponding Basal-type, Luminal-type and ERBB2-type subgroups was reached with a set of 80 genes. Next, a nearest-centroid classification procedure utilizing the 80-gene profile was developed that most accurately classified the breast cancer molecular subtypes on all samples.⁽³⁾

Validation

The BluePrint molecular subtyping profile was subsequently validated on 374 independent samples hybridized to 44k Agilent arrays resulting in 90% concordance between the molecular subtyping profile and the classification based on ER, PR and HER2/neu status.  Additionally, the profile was further validated using publicly available datasets (n=251 ⁽⁴⁾ and n=159 ⁽⁵⁾) generated using Affymetrix arrays with concordance of 79% between the BluePrint profile and classification using ER, PR and HER2/neu receptor status.⁽³⁾ 

When combined, BluePrint characterized a total of 13% (116) of the breast tumors as Basal-type, 66% (603) as Luminal-type and 22% (198) as ERBB2-type. Compared to single-marker readout for the presence of ER, PR and HER2, 13% of the samples that were scored positive for the presence of ER/PR did not express a luminal-type gene profile.^(3-6)

Figure 1. On the left scatterplot of samples from the first validation samples (n=374) and on the right of samples from the second vali­dation set (n=410) with the classifier indices for the Basal-type shown on the x-axis, Luminal-type on the z-axis and ERBB2-type on the y-axis.(6)

Outcome

MammaPrint risk-classification, followed by the 80-gene, BluePrint molecular subtyping profile, can help identify specific groups of breast cancer patients who are at increased risk of distant recurrence. Moreover, groups with low risk of recurrence are recognized, thereby, aiding in treatment planning. Implementation of this knowledge can improve the clinical management of breast cancer patients.