Molecular characterization of clinical grade in breast cancer (BC) challenges the existence of “grade 2” tumors

Abstract

506 Background: Histological grade (HG) in BC provides important prognostic information. Grade 2 tumors are observed to have an intermediate clinical outcome if compared to grade 1 (good prognosis) and grade 3 (poor prognosis) tumors. A major problem is the lack of inter-observer consistency, particularly for grade 2 tumors. The aim of this study was to correlate gene expression profiles (GEP) with HG and clinical outcome. Methods: GEP from 137 early BC patients receiving only locoregional therapy at 2 institutions was performed using Affymetrix U133 A chips. Median follow-up was 6.8 years. Differentially expressed genes were identified by contrasting grade 1 and grade 3 determined by the local pathologist using a stratified t-test. Central pathology review is ongoing. False discovery proportion was done using 10,000 permutations. A multiple independent validation was performed using three publicly available datasets from Van’t Veer et al. 2002, Sorlie et al. 2001 and Sotiriou et al. 2003 (total=281 tumor samples). Results: Grade 1 and 3 tumors were associated with distinct GEP. A “genetic grade index” (GGI) based on the 80 most significant genes was computed. Interestingly, grade 2 tumors were found to have GGI similar to either grade 1 or 3 tumors. At 50% cutoff, GGI segregated grade 2 tumors into two distinct subgroups namely grade 1- and grade 3-like with statistically different relapse free survival (RFS; HR: 3.89, p=0.013) and distant metastasis free survival (DMFS; HR: 4.67, p=0.049), similar to those observed in HG 1 and 3 tumors respectively (RFS; HR: 2.61, p=0.019 and DMFS; HR: 3.59, p=0.026). We then applied the 80-gene list to 3 independent publicly available datasets. Similarly in all three, grade 2 tumors were divided into grade 1- and 3-like subgroups with statistically distinct clinical outcome. Conclusions: HG 1 and 3 in BC is associated with distinctive GEP. GEP reveals grade 2 tumors to consist of either grade 1 or 3 molecular signatures, associated with different prognoses. This has been independently confirmed in multiple validation sets across different microarray platforms and could have important implications for optimal patient management. No significant financial relationships to disclose.