Research-based PAM50 signature and long-term breast cancer survival

Minya Pu,John P Pierce,Shirley W Flatt,Loki Natarajan,Emily Pittman,Tammi L Vickery,Barbara A Parker,Catherine R Marinac,Karen Messer,Sandahl H Nelson,Elaine R Mardis,Sherri R Davies,Matthew J Ellis

doi:10.1007/s10549-019-05446-y

Abstract

PurposeMulti-gene signatures provide biological insight and risk stratification in breast cancer. Intrinsic molecular subtypes defined by mRNA expression of 50 genes (PAM50) are prognostic in hormone-receptor positive postmenopausal breast cancer. Yet, for 25–40% in the PAM50 intermediate risk group, long-term risk remains uncertain. Our study aimed to (i) test the long-term prognostic value of the PAM50 signature in pre- and post-menopausal breast cancer; (ii) investigate if the PAM50 model could be improved by addition of other mRNAs implicated in oncogenesis.MethodsWe used archived FFPE samples from 1723 breast cancer survivors; high quality reads were obtained on 1253 samples. Transcript expression was quantified using a custom codeset with probes for > 100 targets. Cox models assessed gene signatures for breast cancer relapse and survival.ResultsOver 15 + years of follow-up, PAM50 subtypes were (P < 0.01) associated with breast cancer outcomes after accounting for tumor stage, grade and age at diagnosis. Results did not differ by menopausal status at diagnosis. Women with Luminal B (versus Luminal A) subtype had a > 60% higher hazard. Addition of a 13-gene hypoxia signature improved prognostication with > 40% higher hazard in the highest vs lowest hypoxia tertiles.ConclusionsPAM50 intrinsic subtypes were independently prognostic for long-term breast cancer survival, irrespective of menopausal status. Addition of hypoxia signatures improved risk prediction. If replicated, incorporating the 13-gene hypoxia signature into the existing PAM50 risk assessment tool, may refine risk stratification and further clarify treatment for breast cancer.

Highlights

Breast cancer is a heterogeneous disease with large variations in relapse rates even among patients with similar clinical profiles
The focus in the current work was to examine mRNA expression and breast cancer outcomes, and as a first step we investigated the original research-use PAM50 signature which classified tumors into five distinct subtypes: Luminal A, Luminal B, Basal, Her2-enriched, and normal-like [10]
Formalin-fixed paraffin-embedded (FFPE) tissue samples from the primary tumor were available for 60% (n = 1723) of the Women’s Healthy Eating and Living (WHEL) cohort

Summary

Introduction

Breast cancer is a heterogeneous disease with large variations in relapse rates even among patients with similar clinical profiles. More information is needed regarding how these tests perform in well-characterized cohorts of breast cancer patients with long-term follow-up [8]. The breast cancer intrinsic molecular subtypes, defined by mRNA expression of 50 genes (PAM50), have been shown to improve prognostication significantly compared to standard tumor characteristics and other genomic signatures [3, 9,10,11,12,13,14,15]. The related proprietary Prosigna gene signature, FDA-approved in 2013, was validated to estimate relapse risk in postmenopausal women with early stage, hormonereceptor positive breast tumors, but has not been validated in pre-menopausal women [16]. For the researchbased PAM50 signature and the Prosigna test [16, 17], an estimated 25–40% of patients are characterized as having “intermediate risk,” and for these patients, the long-term risk of relapse remains uncertain

Methods

Results

Conclusion