Development and internal validation of an immune-based prognostic modeling of early-onset colorectal cancer via machine learning

Table 2 Univariable Cox regression analysis for progression-free survival in 123 colorectal cancer patients

Variable	Contrast	Hazard ratio	95%CI	P value
Age	Per unit increase	1.00	2.59-2.86	0.984
CD16+CD56+NK	High vs low	1.54	2.11-23.73	0.243
CD19+B	High vs low	0.65	1.37-3.82	0.243
CD3+CD4+T	High vs low	0.90	1.55-6.35	0.776
CD3+CD8+T	High vs low	0.96	1.6-7.08	0.903
CD3+T	High vs low	0.97	1.61-7.36	0.943
CD4+T/CD8+T	High vs low	1.37	1.95-16.62	0.388
CEA	Per unit increase	1.01	2.73-2.75	< 0.001
CMOR	Yes vs no	4.24	5.01 to > 1000	0.003
Chemoradiotherapy	Yes vs no	7.29	5.68 to > 1000	0.007
Family	Yes vs no	0.92	1.52-7.55	0.836
M	1 vs 0	5.57	9.6 to > 1000	< 0.001
N	1 vs 0	4.51	7.09 to > 1000	< 0.001
N	2 vs 0	3.87	3.19 to > 1000	0.028
Nerve invasion	Yes vs no	2.30	3.03-119.07	0.025
Sex	Female vs male	1.14	1.75-10.34	0.718
Stage	Stage in III-IV	3.27	4.68 to > 1000	0.002
T	2 vs 1	8959622.29	1 to > 1000	0.997
T	3 vs 1	33760956.62	1 to > 1000	0.996
T	4 vs 1	63867920.40	1 to > 1000	0.996
Treg	High vs low	0.81	1.48-5.28	0.563
Tumor necrosis	Yes vs no	0.00	1 to > 1000	0.998
Vascular invasion	Yes vs no	2.12	2.74-86.15	0.048

NK: Natural killer; CEA: Carcinoembryonic antigen; CMOR: Combined multiple organ resection; M: Metastasis; N: Node; T: Tumor; Treg: Regulatory T cells; CI: Confidence interval.