Breast Cancer Research and Treatment

, Volume 142, Issue 3, pp 505–514

A relative ordering-based predictor for tamoxifen-treated estrogen receptor-positive breast cancer patients: multi-laboratory cohort validation

  • Xianxiao Zhou
  • Bailiang Li
  • Yuannv Zhang
  • Yunyan Gu
  • Beibei Chen
  • Tongwei Shi
  • Lu Ao
  • Pengfei Li
  • Shan Li
  • Chunyang Liu
  • Zheng Guo
Preclinical study

DOI: 10.1007/s10549-013-2767-8

Cite this article as:
Zhou, X., Li, B., Zhang, Y. et al. Breast Cancer Res Treat (2013) 142: 505. doi:10.1007/s10549-013-2767-8

Abstract

Current predictors for estrogen receptor-positive (ER-positive) breast cancer patients receiving tamoxifen are often invalid in inter-laboratory validation. We aim to develop a robust predictor based on the relative ordering of expression measurement (ROE) in gene pairs. Using a large integrated dataset of 420 normal controls and 1,129 ER-positive breast tumor samples, we identified the gene pairs with stable ROEs in normal control and significantly reversed ROEs in ER-positive tumor. Using these gene pairs, we characterized each sample of a cohort of 292 ER-positive patients who received tamoxifen monotherapy for 5 years and then identified relapse risk-associated gene pairs. We extracted a gene pair subset that resulted in the largest positive and negative predictive values for predicting 10-year relapse-free survival (RFS) using a genetic algorithm. A predictor was developed based on the gene pair subset and was validated in 2 large multi-laboratory cohorts (N = 250 and 248, respectively) of ER-positive patients who received 5-year tamoxifen alone. In the first validation cohort, the patients predicted to be tamoxifen sensitive had a 10-year RFS of 91 % (95 % confidence interval [CI] 85–97 %) with an absolute risk reduction of 34 % (95 % CI 17–51 %). The patients predicted to be tamoxifen insensitive had a significantly higher relapse risk than the patients predicted to be tamoxifen sensitive (hazard ratio = 4.99, 95 % CI 2.45–10.17, P = 9.13 × 10−7). Similar performance was achieved for the second validation cohort. The predictor performed well in both node-negative and node-positive subsets and added significant predictive power to the clinical parameters. In contrast, 2 previously proposed predictors did not achieve significantly better performances than the baselines of the validation cohorts. In summary, the proposed predictor can accurately and robustly predict tamoxifen sensitivity of ER-positive breast cancer patients and identified patients with a high probability of 10-year RFS following tamoxifen monotherapy.

Keywords

Breast cancerEstrogen receptor-positiveTamoxifenPredictorRelative orderingGene expression

Supplementary material

10549_2013_2767_MOESM1_ESM.pdf (773 kb)
Supplementary material 1 (PDF 773 kb)

Copyright information

© Springer Science+Business Media New York 2013

Authors and Affiliations

  • Xianxiao Zhou
    • 1
  • Bailiang Li
    • 2
    • 3
  • Yuannv Zhang
    • 2
  • Yunyan Gu
    • 2
  • Beibei Chen
    • 2
  • Tongwei Shi
    • 1
  • Lu Ao
    • 2
    • 4
  • Pengfei Li
    • 2
  • Shan Li
    • 1
  • Chunyang Liu
    • 4
  • Zheng Guo
    • 1
    • 2
    • 4
  1. 1.Bioinformatics Centre, School of Life Science and TechnologyUniversity of Electronic Science and Technology of ChinaChengduChina
  2. 2.College of Bioinformatics Science and TechnologyHarbin Medical UniversityHarbinChina
  3. 3.Genomics Research CenterHarbin Medical UniversityHarbinChina
  4. 4.Department of Bioinformatics, School of Basic Medical SciencesFujian Medical UniversityFuzhouChina