An immunohistochemical panel of antibodies that predicts recurrence in estrogen receptor positive breast cancer patients correlates with the luminal B gene expression subtype

Robert S. Seitz,Brian Ring,Rodney A. Beck,Douglas Ross,Hege G. Russnes,Therese Sørlie,Elin Borgen,Rolf Kåresen,Anne Lise Børresen-Dale,Bjørn Naume

IF: 11.2

2007-01-01

Cancer Research

Abstract:5079 Introduction: Gene expression profiling of breast tumor specimens using hierarchical cluster analysis has consistently identified a subtype of estrogen receptor positive (ER+) disease associated with poor prognosis (Sorlie et al, PNAS 2003). This subtype is generally referred to as luminal B (LumB) and classifies patients in ways similar to other prognostic gene expression profiles trained to identify poor prognosis ER+ patients (Fan et al, NEJM, 2006). We have previously reported the translation of gene expression signatures into an immunohistochemistry (IHC) based test that uses five antibodies to identify a group of ER+ patients at higher risk of recurrence (Ring et al, JCO, 2006, Ross et al, ASCO, SABCS 2006). Here we measure gene expression patterns and tumor IHC on 123 patient samples to determine the association between classification as high risk for tumor progression by the five antibody predictor and classification as LumB by gene expression profiling. Methods: RNA was prepared from fresh tumor material from surgical specimens from 123 breast cancer patients at primary diagnosis. Gene expression profiles were measured using whole-genome cDNA microarrays and tumor class assigned by correlation to classifier centroids using the previously defined intrinsic gene list (Sorlie, 2003). Tissue Microarrays were constructed from Fresh Frozen Paraffin Embedded (FFPE) tissue from the same patients and IHC stains performed using an anti-ER antibody as well as the five monoclonal antibody prognostic panel (SLC7A5, HTF9C, p53, NDRG1, CEACAM5). A previously established algorithm for combining staining results into a risk index was applied to the ER+ patients and patients were classified as low, moderate, or high risk for tumor recurrence. Of the 123 patients, 47 were identified as ER+ and had sufficient gene expression data and FFPE material to compare IHC and gene expression-based classification. Of the eight patients classified as LumB, one patient was classified as low risk (1 of 16 total), one as moderate risk (1 of 13 total) and six as high risk (6 of 18 total) (p=0.041 Fisher’s exact test; high risk group compared to moderate and low risk groups combined). Each antibody had a positive association with LumB status by logistic regression, with SLC7A5 and HTF9C being significant (p=0.015 and 0.043 respectively). Conclusion: In this small sample set, 88% of lumB subtype ER+ breast cancer patients were classified as high risk for tumor recurrence by a five antibody prognostic IHC test. This demonstrates that this test is recognizing the aggressive ER+ tumor phenotype identified by gene expression profiling. Further exploration of the clinical associations of this IHC-based tumor classification in archived paraffin blocks will help in defining the clinical relevance of the emerging taxonomy-based classification of breast cancer.