Skip to main content
. Author manuscript; available in PMC: 2008 Jun 4.
Published in final edited form as: Clin Exp Metastasis. 2008 Feb 27;25(4):357–369. doi: 10.1007/s10585-008-9146-6

Fig. 4. ECM eQTL gene expression signatures predict survival in human breast cancer.

Fig. 4

Diasporin Pathway candidate genes accurately predict overall survival in the Dutch Rosetta dataset. (a) The cumulative survival for the NDN signature was estimated to be 70% vs. 57% for the good and poor prognosis NDN signatures, respectively (NDN signature hazard ratio = 1.94, 95% confidence interval [CI] = 1.16–3.26). (b) The cumulative survival for the LUC7L signature was estimated to be 67% vs. 53% for the good and poor prognosis LUC7L signatures, respectively (LUC7L signature hazard ratio = 1.55, 95% CI = 1.32–2.89). (c) The cumulative survival for the PI16 signature was estimated to be 72% vs. 53% for the good and poor prognosis PI16 signatures, respectively (PI16 signature hazard ratio = 1.57, 95% CI = 1.04–2.36). (d) The cumulative survival for the CENTD3 signature was estimated to be 74% vs. 50% for the good and poor prognosis CENTD3 signatures, respectively (CENTD3 signature hazard ratio = 2.77, 95% CI = 1.80–4.25). (e) The cumulative survival for the CSF1R signature was estimated to be 69% vs. 51% for the good and poor prognosis CSF1R signatures, respectively (CSF1R signature hazard ratio = 2.38, 95% CI = 1.58–3.60). (f) Indeed, it appears that these candidate gene signatures have a similar ability to predict survival in this dataset than the 70-gene signature described by van’t Veer et al [4]. Specifically, the survival for the good and poor prognosis 70-gene signatures was estimated to be 73% vs. 47%, respectively (70 gene signature hazard ratio = 4.49, 95% CI = 2.65–7.61).