CorePAM: a 24-gene PAM50-derived expression score with cross-platform external validation for breast cancer prognosis

Area of Science:

• Oncology
• Genomics
• Biostatistics

Background:

• The 50-gene PAM50 classifier is crucial for breast cancer prognosis but requires extensive resources.
• A data-driven approach was used to identify a smaller gene subset that retains prognostic accuracy.

Purpose of the Study:

• To derive and validate a minimal gene set (CorePAM) from the PAM50 classifier that maintains non-inferior prognostic performance.
• To assess the incremental prognostic value of CorePAM over clinical factors and its association with pathologic complete response (pCR).

Main Methods:

• Cox elastic-net regression with cross-validation was applied to the SCAN-B cohort (N=3069) for gene selection.
• External validation was performed in four independent cohorts (TCGA-BRCA, METABRIC, GSE20685, GSE1456) using RNA-seq and microarray data.
• Incremental value was assessed using bootstrap C-index, and pCR prediction was evaluated in neoadjuvant cohorts.

Main Results:

• The 24-gene CorePAM classifier achieved an out-of-fold C-index of 0.670, with a minimal gap (0.009) compared to the 50-gene model.
• CorePAM demonstrated independent association with survival across all validation cohorts (pooled HR=1.37) and provided incremental value over a clinical model (CORE-A).
• CorePAM was significantly associated with pCR in neoadjuvant analyses (pooled OR=1.69).

Conclusions:

• The 24-gene CorePAM classifier meets non-inferiority criteria for breast cancer prognosis compared to the 50-gene PAM50.
• CorePAM simplifies assay development and shows potential for clinical deployment, pending platform-specific validation.
• The classifier's prognostic ability is retained even after adjustment for anatomical staging and it predicts pCR.