Skip to main content

Table 5 Multivariable Cox regression of disease-free survival

From: Development and validation of a new tumor-based gene signature predicting prognosis of HBV/HCV-included resected hepatocellular carcinoma patients

Parameter

HR (95% CI)

P value

ci training (95% CI)

ci internal validation (95% CI)

ci in silico validation (95% CI)

9-gene signature

4.44 (2.36, 8.33)

< 0.0001

0.70 (0.58, 0.82)

0.74 (0.53, 0.95)

0.65 (0.55, 0.83)

Clinical parameters

Tumor diameter

1.13 (1.06, 1.20)

0.0001

   

Tumor differentiation

1.58 (0.89, 2.80)

0.115

0.64 (0.50, 0.78)

0.67 (0.24, 0.99)

0.57 (0.41, 0.73)

9-gene signature and clinical parameters

9-gene signature

3.95 (0.68, 7.45)

< 0.0001

   

Tumor diameter

1.08 (1.01, 1.12)

0.010

   

Tumor differentiation

1.29 (0.68, 2.50)

0.422

0.79 (0.55, 1.03)

0.83 (0.57, 1.36)

0.70 (0.58, 0.92)

Improvement of combined model compared to

  

dLL

Degrees of freedom

P value

9-Gene signature only

  

10.53

2

< 0.001

Clinical parameters only

  

83.21

3

0.01

  1. Three multivariable Cox regression models were built using the training cohort: a model consisting of only the 9-gene signature (top), a model consisting only of the clinical tumor diameter and tumor differentiation, and a model combining both the 9-gene signature and clinical parameters (bottom). HRs are given with their 95% CIs and the corresponding P values. For each model, the concordance index (ci) is given for the training and internal validation cohort as well as for the patients of the or in silico validation cohort. Its 95% CI is determined from 1000 bootstrap samples of the respective cohort. The improvement of the combined model, including the 9-gene signature and the clinical parameters, compared with the 9-gene signature and clinical parameters alone is shown (bottom) based on the difference in log-likelihood (dLL)