Table 2 Multivariable DFS analyses in each cohort.

From: High-throughput proteomics profiling-derived signature associated with chemotherapy response and survival for stage II/III colorectal cancer

Variables

HR

95% CI

P-value

Training cohort (n = 203)

Age (≥ 65 vs. < 65)

2.02

1.21−3.63

0.007

Tumor location (rectum vs. colon)

1.96

1.19−3.22

0.008

pN stage (N1-2 vs. N0)

1.97

1.15−3.37

0.014

PS (high vs. low)

2.62

1.38−4.96

0.003

Internal validation cohort (n = 204)

pT stage (T4 vs. T1-3)

2.75

1.46−5.17

0.002

pN stage (N1-2 vs. N0)

2.01

1.15−3.52

0.014

PS (high vs. low)

2.81

1.33−5.96

0.007

External validation cohort (n = 333)

pT stage (T4 vs. T1-3)

2.29

1.40−3.75

< 0.001

N stage (N2 vs. N1)

2.95

1.90−4.56

< 0.001

CEA ( ≥ 5 vs. <5 ng/L)

2.28

1.51−3.43

< 0.001

PS (high vs. low)

2.84

1.61−5.02

< 0.001

  1. We calculated hazard ratios and p values using an adjusted multivariate Cox proportional hazards regression model, including proteomic signature (high risk vs. low risk), sex (male vs. female), age (≥ 60 years vs. 60 years), pN stage (N1-2 vs. N0), pT stage (T4 vs. T1-3), MMR, (dMMR vs. pMMR) histology (low vs. high), location (rectum vs. colon), lymph nodes examined (≥ 12 vs. < 12). and CEA ( ≥ 5 vs. < 5 ng/L). DFS Disease-free survival, PS Proteomic signature, CEA Carcinoembryonic antigen, HR Hazard ratio, CI Confidence interval. We selected variables with the backward stepwise approach, the p value threshold was 0·05 (p > 0·05) for removing insignificant variables from the model. Only variables that were significantly associated with survival are presented.
Back to article page