Table 4

Stratified Cox-regression model of seven-gene Heterogeneous Mixture Discriminant Analysis (MDAhet) classifier

Training set

Combined test set

Annotated

183

286

MDAhet

0.29 (0.16–0.56) p = 0.0002

0.15 (0.07–0.36) p < 0.000001

LN

1.31 (0.73–2.33) p = 0.36

3.25 (1.61–6.58) p = 0.001

CT

NA

0.68 (0.34–1.39) p = 0.29

LN+MDAhet

MDAhet

0.29 (0.15–0.55) p = 0.0002

0.06 (0.01–0.27) p = 0.0002

LN

1.59 (0.81–3.11) p = 0.18

3.68 (1.32–10.13) p = 0.012

CT+MDAhet

MDAhet

NA

0.27 (0.15–0.48) p = 0.00001

CT

NA

0.76 (0.27–2.13) p = 0.6

Stratified Cox-proportional hazards regression performance of the seven-gene MDAhet classifier, lymph node status (LN) and chemotherapy (CT) across oestrogen receptor-negative training and test sets, with strata defined by cohorts. For the univariate analysis, Hazard ratio (HR), 95% confidence intervals (CI) and LR-test p-value are given. In the multivariate models, p-values quoted are from the corresponding Wald test.

Teschendorff and Caldas Breast Cancer Research 2008 10:R73   doi:10.1186/bcr2138