PAM50 gene signatures and breast cancer prognosis with adjuvant anthracycline- and taxane-based chemotherapy: correlative analysis of C9741 (Alliance).

Minetta C Liu ,Larry Norton,Katherine Deschryver,Baljit Singh,Silvana Martino,Charles M Perou ,Jerry P Reed ,Paula N Friedman ,Tammi L Vickery ,Clifford A Hudis ,Lisa A Carey ,Jacqueline Snider ,Matthew J Ellis ,Eric P Winer ,Brandelyn N Pitcher ,Marc L Citron ,Edith A Perez ,William J Gradishar ,Torsten O Nielsen ,Philip S Bernard ,Sherri R Davies ,Elaine R Mardis ,William T Barry

doi:10.1038/npjbcancer.2015.23

Abstract

PAM50 intrinsic breast cancer subtypes are prognostic independent of standard clinicopathologic factors. CALGB 9741 demonstrated improved recurrence-free (RFS) and overall survival (OS) with 2-weekly dose-dense (DD) versus 3-weekly therapy. A significant interaction between intrinsic subtypes and DD-therapy benefit was hypothesized. Suitable tumor samples were available from 1,471 (73%) of 2,005 subjects. Multiplexed gene-expression profiling generated the PAM50 subtype call, proliferation score, and risk of recurrence score (ROR-PT) for the evaluable subset of 1,311 treated patients. The interaction between DD-therapy benefit and intrinsic subtype was tested in a Cox proportional hazards model using two-sided alpha=0.05. Additional multivariable Cox models evaluated the proliferation and ROR-PT scores as continuous measures with selected clinical covariates. Improved outcomes for DD therapy in the evaluable subset mirrored results from the complete data set (RFS; hazard ratio=1.20; 95% confidence interval=0.99–1.44) with 12.3-year median follow-up. Intrinsic subtypes were prognostic of RFS (P<0.0001) irrespective of treatment assignment. No subtype-specific treatment effect on RFS was identified (interaction P=0.44). Proliferation and ROR-PT scores were prognostic for RFS (both P<0.0001), but no association with treatment benefit was seen (P=0.14 and 0.59, respectively). Results were similar for OS. The prognostic value of PAM50 intrinsic subtype was greater than estrogen receptor/HER2 immunohistochemistry classification. PAM50 gene signatures were highly prognostic but did not predict for improved outcomes with DD anthracycline- and taxane-based therapy. Clinical validation studies will assess the ability of PAM50 and other gene signatures to stratify patients and individualize treatment based on expected risks of distant recurrence.

Highlights

The clinical and genomic heterogeneity of early-stage breast cancer is well-recognized
No differences in results from the complete data set (RFS) or overall survival (OS) were observed between sequential versus concurrent chemotherapy: hazard ratio (HR) = 1.06 (95% confidence interval (CI) = 0.91–1.24, P = 0.43) and HR = 1.04, respectively
Improved outcomes were seen in patients who received DD treatment: HR = 1.26 for RFS and HR = 1.21 for OS

Summary

Introduction

The clinical and genomic heterogeneity of early-stage breast cancer is well-recognized. The luminal A (LumA), luminal B (LumB), HER2-enriched (HER2E), basal-like, and normal-like breast cancer subtypes were initially defined through unsupervised clustering analysis of global gene expression from RNA extracted from frozen tissue.[1] A 50-gene qPCR assay (PAM50) was developed to identify the intrinsic biological subtypes using RNA isolated from more readily available formalinfixed, paraffin-embedded (FFPE) tissue. These subtypes can be assessed using a multiplexed gene-expression profiling technology (NanoString Technologies; Seattle, WA, USA). The PAM50 assay was used to develop a prognostic risk of relapse score based on the relative distance to the centroid of each subtype;[6] a proliferation score based on a subset of genes related to cell cycle progression;[7] and composite scores that include tumor size with molecular phenotypes.[6,7] each has prognostic capability, the utility of these scores to predict for specific treatment benefit and select therapy has not been studied

Methods

Results

Conclusion