Predicting prognosis of breast cancer with gene signatures: Are we lost in a sea of data?

Takayuki Iwamoto, Lajos Pusztai

Research output: Contribution to journalComment/debatepeer-review

39 Citations (Scopus)


A large number of prognostic and predictive signatures have been proposed for breast cancer and a few ofthese are now available in the clinic as new molecular diagnostic tests. However, several other signatures have not fared well in validation studies. Some investigators continue to be puzzled by the diversity of signatures that are being developed for the same purpose but that share few or no common genes. The history of empirical development of prognostic gene signatures and the unique association between molecular subsets and clinical phenotypes of breast cancer explain many of these apparent contradictions in the literature. Three features of breast cancer gene expression contribute to this: the large number of individually prognostic genes (differentially expressed between good and bad prognosis cases); the unstable rankings of differentially expressed genes between datasets; and the highly correlated expression of informative genes.

Original languageEnglish
Article number81
JournalGenome Medicine
Issue number11
Publication statusPublished - Nov 12 2010
Externally publishedYes

ASJC Scopus subject areas

  • Molecular Medicine
  • Molecular Biology
  • Genetics
  • Genetics(clinical)


Dive into the research topics of 'Predicting prognosis of breast cancer with gene signatures: Are we lost in a sea of data?'. Together they form a unique fingerprint.

Cite this