Skip to main content

Table 3 Comparison of prognostic indexes with logistic regression models

From: Integrated analysis of independent gene expression microarray datasets improves the predictability of breast cancer outcome

 

Rosetta

Stanford

 

-2LL

L. R.

p value

-2LL

L. R.

p value

Intercept

108.7

  

79.9

  

SEP

93.1

15.6

0.0001

65.9

14.0

0.0002

ER status

98.0

10.8

0.001

74.3

5.5

0.02

PR status

103.5

5.3

0.02

NA

NA

NA

Tumor size

101.2

7.6

0.006

76.6

3.3

0.35

Lymph node status

NA

NA

NA

78.2

1.7

0.44

Histological grade

92.9

15.9

0.0004

70.4

9.5

0.009

Age at diagnosis

98.9

9.8

0.002

78.6

1.24

0.27

Angioinvasion

104.4

4.4

0.04

NA

NA

NA

Multivariate model (include SEP)

67.7

41.0

0.000002

54.8

25.1

0.005

Multivariate model (not include SEP)

75.2

33.6

0.00002

56.9

23.0

0.006

  1. Model fitness was represented by -2LL (-2 log likelihood), L.R. (likelihood ratio), and p value. Multivariate models were built with all presented indexes with or without SEP. NA: value of index was uniform or not given. Two Stanford samples, svcc76 and svn007, were not included because of incomplete clinical information.