[Transcription markers of survival in breast neoplasms measuring the signals from the proliferation].

Abstract

We show that proliferation-related signals are omnipresent in the breast cancer transcriptome. As a result, many transcriptional signatures generated at random are valuable for the prognosis of disease-free survival: despite their biological rationale, 30-60% of published prognostic signatures are not significantly better. We propose a mathematical transformation, the super PCNA decovolution, which removes proliferation-related signals from tumours transcriptional profiles. Both random and published signatures loose nearly all their prognostic value after removal of these signals.