Skip to main content
. Author manuscript; available in PMC: 2017 Sep 15.
Published in final edited form as: Clin Cancer Res. 2016 Apr 8;22(18):4664–4675. doi: 10.1158/1078-0432.CCR-16-0316

Table 3.

Stepwise forward selection of early detection markers in combined pre-diagnostic models for variable lag-times since blood donationa

Model 1: CA125 Model 2: CA125 + HE4 Model 3: CA125 + HE4 + CA72.4 Model 4: CA125 + HE4 + CA72.4 + CA15.3
# sets C-Statistic (95%-CI) pb C-Statistic (95%-CI) pc C-Statistic (95%-CI) pd C-Statistic (95%-CI) pe
All women 197 0.70 (0.66-0.74) <0.001 0.71 (0.67-0.75) <0.001 0.71 (0.67-0.75) 0.002 0.71 (0.67-0.75) 0.05
        corrected C f - 0.71 0.71 0.71
≤ 6 months 27 0.92 (0.86-0.98) <0.001 0.89 (0.83-0.96) 0.58 0.90 (0.83-0.96) 0.36 0.92 (0.86-0.97) 0.06
        corrected C f - 0.89 0.90 0.92
≤ 1 year 62 0.82 (0.76-0.88) <0.001 0.82 (0.77-0.88) 0.005 0.84 (0.78-0.89) 0.002 0.84 (0.78-0.89) 0.55
        corrected C f - 0.81 0.83 0.83
1 to 2 years 76 0.72 (0.65-0.78) <0.001 0.73 (0.66-0.79) 0.004 0.73 (0.67-0.80) 0.03 0.73 (0.67-0.80) 0.08
        corrected C f - 0.72 0.72 0.72
2 to 3 years 59 0.56 (0.48-0.64) 0.42 0.58 (0.50-0.67) 0.18 0.58 (0.50-0.66) 0.57 0.58 (0.50-0.66) 0.35
        corrected C f - 0.53 0.53 0.53
a

Conditional logistic regression model with log2-transformed markers as continuous variables

b

Likelihood ratio test, comparing Model 1 with empty model

c

Likelihood ratio test, comparing Model 2 with Model 1

d

Likelihood ratio test, comparing Model 3 with Model 2

e

Likelihood ratio test, comparing Model 4 with Model 3

f

C-Statistic in multivariate models corrected for over optimism obtained by bootstrap sampling