Table 3.

Cox proportional hazard models exploring Ang1 and Tie2 as a joint biomarker in the training dataset (N = 91)

CovariatesHR (95% CI)PBootstrap frequency (%)
FIGO stage
 FIGO stage I1.000 (referent)
 FIGO stage II0.978 (0.203–4.694)0.978
 FIGO stage III2.827 (0.838–9.535)0.094
 FIGO stage IV14.775 (3.478–62.760)< 0.001
Size of residual disease
 ≤1-cm residual disease1.000 (referent)
 >1-cm residual disease1.363 (0.774–2.402)0.283
Treatment
 Standard arm1.000 (referent)
 Bevacizumab arm0.626 (0.228–1.722)0.4779.9
Individual biomarkers
 Ang1
  <median1.000 (referent)
  ≥median0.413 (0.146–1.168)0.09661.9
 Tie2
  ≥median1.000 (referent)
  <median0.463 (0.138–1.546)0.21023.3
Interaction termsa
 Ang1 × Treatment5.677 (1.387–23.241)0.01651.7
 Tie2 × Treatment2.369 (0.539–10.422)0.25412.5
 Ang1 × Tie28.132 (1.684–39.255)0.00965.7
 Ang1 × Tie2 × Treatment0.038 (0.004–0.139)0.00384.3

NOTE: There were no missing data.

  • aIn all models, standard arm, 0; bevacizumab arm, 1; ang1 < median, 0; ang2 ≥ median, 1; tie2 < median, 1; tie2 ≥ median, 0. All interaction terms are multiplicative.