PCA-PAM50

PCA-PAM50 refines PAM50 intrinsic breast cancer subtyping using Principal Component Analysis (PCA) to improve concordance with clinical subtypes derived from immunohistochemistry (IHC) assays of 3-4 biomarkers.

Key Features:

• Iterative subtyping approach: Employs an iterative process integrating Principal Component Analysis (PCA) to refine PAM50 intrinsic subtyping in cohorts with unbalanced estrogen receptor (ER) status.
• Gene expression-based ER status: Derives a gene expression-based estimation of ER status via PCA to enable selection of an ER-balanced subset for accurate gene centering.
• Improved concordance across cohorts: Application across three distinct breast cancer study cohorts demonstrated enhanced consistency between intrinsic and clinical subtyping by 6-9.3%.
• Reclassification of aggressive luminal A tumors: Identifies a more aggressive subset within luminal A (LA) characterized by higher MKI67 gene expression and poorer patient survival, reclassifying these tumors as luminal B (LB) and increasing concordance with IHC by 25-49%.

Scientific Applications:

• Alignment of intrinsic and clinical subtypes: Bridges discrepancies between PAM50-derived intrinsic subtypes and IHC-derived clinical subtypes to increase subtype concordance.
• Prognostic refinement within luminal subtypes: Enables identification and reclassification of aggressive LA tumors (high MKI67, poorer survival) as LB for improved prognostic resolution.
• Subtype-based patient stratification: Supports research and clinical efforts to refine patient stratification and subtype-specific decision making based on adjusted gene expression subtyping.

Methodology:

PCA-PAM50 uses an iterative workflow in which Principal Component Analysis (PCA) is applied to derive gene expression-based ER status, select an ER-balanced subset for gene centering, and recalibrate PAM50 intrinsic subtyping in ER-unbalanced cohorts.