Skip to main content

Table 2 Multivariable Cox proportional hazards models for each genomic prognostic signature adjusted for PREDICT in (a) ER-positive and (b) ER-negative patients

From: Evaluation and comparison of different breast cancer prognosis scores based on gene expression data

Model

Hazard ratio (95% CI)

Log-likelihood

Model fit p value

c-index (95%CI)

c-index optimism

(a)

PREDICT

2.72 (NA)

− 2860.6

0.687 (0.661–0.713)

Oncotype DX

1 (0.99–1)

− 2859.1

0.0835

0.693 (0.667–0.718)

− 4.17 × 10–4

EndoPredict

1.13 (1.02–1.24)

− 2858

0.024

0.692 (0.666–0.717)

0.003

MammaPrint

0.52 (0.17– 0.87)

− 2853.9

2.32 × 10–4

0.699 (0.674–0.724)

6.72 × 10–4

Prosigna

1.58 (1.11– 2.05)

− 2858.7

0.0546

0.694 (0.668–0.719)

− 8.68 × 10–4

(b)

PREDICT

2.72 (NA)

− 1072

0.667 (0.630–0.704)

Oncotype DX

0.998 (0.986– 1.01)

− 1072

0.77

0.667 (0.63–0.704)

0.011

EndoPredict

0.884 (0.657– 1.11)

− 1071.4

0.287

0.667 (0.630–0.705)

0.007

MammaPrint

0.854 (0.196– 1.51)

− 1071.9

0.638

0.669 (0.632–0.706)

0.005

Prosigna

0.252 (− 0.875–1.38)

− 1069.3

0.0166

0.661 (0.622–0.699)

7.84 × 10–4

  1. 95% CI, 95% confidence interval
  2. Hazard ratio has been constrained to this value