Predicting relapse in estrogen receptor (ER) positive breast cancer (luminal) subgroups treated with adjuvant tamoxifen.

Loi S, Desmedt C, Haibe-Kains B, Lallemand F, Gillett C, Tutt A, Ryder K, Ellis P, Harris A, Smeds J, Bergh J, Cardoso F, Piccart MJ, Sotiriou C

Background: ER-positive breast cancers (BC) exhibit considerable clinical heterogeneity. Microarray profiling has lead to the notion of distinct molecular subgroups within ER-positive BC; these are often described as luminal subtypes (Sorlie, Sotiriou et al). The luminal subtypes seem to be directly associated with proliferative genes, with the luminal A group correlated with low grade, and the B and C groups with high grade. We sought to further define these subtypes and identify genes associated with BC recurrence within the subgroups of ER-positive BC. 

Material and Methods: We determined gene expression profiles from 168 tamoxifen-only treated ER positive early stage BC samples (training set) using Affymetrix U133 2.0 Plus Genechips. Median follow-up was 7.3 yrs. We divided these tumors into 2 groups according to their degree of proliferation. These groups, which had distinct clinical outcomes (HR:4.1, 95%CI:2.0-8.5, p<0.001), were named low and high grade ER-positive subtypes and are molecularly consistent with the previously described luminal subtypes of Sorlie and Sotiriou et al. Genes associated with distant relapse within the subtypes were identified using Cox proportional hazards model. The independent validation set consisted of 86 ER-positive tamoxifen-only treated BC from a different institution.

Results: For the low grade subgroup we identified 330 genes that were significantly associated with distant relapse. Here, many genes were related to immune function. In the high grade subgroup 195 genes were identified, including many that were HER-2 and ER-related. The probability of selecting these genes by chance was estimated to be p=0.01 and 0.04 after 1000 random permutations respectively. There were only 7 genes in common. In the independent validation set, these genes could identify distinct prognostic groups within the low (n=44) and high grade subtypes (n=42): low grade: HR: 3.9 (95%CI: 0.9-19.2), p=0.06; high grade HR: 2.7 (95%CI:1.1-7.1), p= 0.04. We are currently validating this concept in a further 100 tumor samples.

Discussion: ER-positive BC can be divided into low grade or high grade subtypes. Furthermore, different groups of genes can identify women at high risk of relapse on adjuvant tamoxifen within these ER-positive subgroups for whom other therapies could be targeted. Further investigation into the biology of these diverse gene sets could substantially aid treatment tailoring for BC patients.