Skip to main content
. 2015 Oct 30;7(12):13372–13387. doi: 10.18632/oncotarget.5618

Table 3. Univariate and multivariate survival analyses in CRCs (validation cohort; n = 725)a .

Variables n Univariate analysis P-value Multivariate analysisb P-value
H.R. (95% C.I.) H.R. (95% C.I.)
EPCAM expression status
 EPCAM-intact 675 Reference Reference
 EPCAM-PL 50 2.8 (1.95–4.01) <0.001 1.57 (1.04–2.39) 0.03
AJCC/UICC cancer stage
 Stage I/II 346 Reference Reference
 Stage III/IV 379 4.54 (3.38–6.09) <0.001 3.18 (2.3–4.39) <0.001
Tumor differentiation
 WD/MD 697 Reference Reference
 PD 28 3.4 (2.15–5.37) <0.001 1.62 (0.95–2.76) 0.08
Lymphovascular invasion
 Absent 404 Reference Reference
 Present 321 2.62 (2.04–3.36) <0.001 1.27 (0.96–1.68) 0.09
Perineural invasion
 Absent 552 Reference Reference
 Present 173 3.19 (2.49–4.08) <0.001 1.9 (1.45–2.48) <0.001
Tumor budding
 Absent 221 Reference Reference
 Present 504 2.13 (1.56–2.89) <0.001 1.29 (0.93–1.78) 0.13
CIMP status
 CIMP-low/negative 679 Reference Reference
 CIMP-high 46 1.81 (1.19–2.76) 0.006 1.08 (0.68–1.72) 0.73

H.R., Cox hazard ratio; 95% C.I., 95% confidence interval of H.R.; EPCAM-intact, intact EPCAM expression; EPCAM-PL, partial loss of EPCAM expression; AJCC/UICC, American Joint Committee on Cancer/International Union against Cancer; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated; CIMP, CpG island methylator phenotype

a

EPCAM-CL (n = 1) was excluded from these Cox proportional hazards regression model-based survival analyses due to the extremely small sample size for this variable.

b

Variables statistically significant in the univariate analysis were entered into the multivariate analysis.