Construction of a New Tumor Immunity-Related Signature to Assess and Classify the Prognostic Risk of Colorectal Cancer

Figures & data

Table 1 Clinical Information of Samples

Feature	GEO	TCGA-COAD
Relapse
NO	914	320
YES	216	117
T Stage
T1	12	11
T2	47	75
T3	349	300
T4	105	50
TX	617	1
N Stage
N0	293	256
N1	124	103
N2	87	78
N3	5
NX	621
M Stage
M0	485	323
M1	30	60
MX	615	54
Stage
I	124	73
II	548	167
III	410	126
IV	44	60
X	4	11
Gender
Female	481	203
Male	557	234
Unknown	92
Grade
G1	16
G2	134
G3	21
GX	959
Lymphatic_invasion
NO		242
YES		152
Unknown		43
Age
>65	552	255
≤65	396	182
Unknown	182

Figure 1 Work flow chart.

Figure 2 Immunoinfiltration analysis. (A) PCA analysis of five datasets prior to elimination of batch effect. (B) PCA analysis of five datasets after elimination of batch effect. (C) KM curves of StromalScore^high group and StromalScore^low groups in the merged GEO dataset. (D) KM curves of ESTIMATEScore^high group and ESTIMATEScore^low groups in the merged GEO dataset. (E) KM curves of ImmuneScore^high group and ImmuneScore^low groups in the merged GEO dataset. (F) Correlation analysis of immune scores.

Figure 3 WGCNA. (A) Cluster analysis of merged GEO dataset samples. (B) Analysis of network topology for various soft-thresholding powers. (C) Gene dendrogram and module color. (D) Correlations between the 18 modules and immune scores.

Figure 4 Functional enrichment analysis. (A) Gene Ontology (GO) functional annotation of pink module genes. (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation of pink module genes. (C) GO functional annotation of yellow module genes. (D) KEGG pathway annotation of yellow module genes. (E) GO functional annotation of the midnight blue module genes. (F) KEGG pathway annotation of the midnight blue module gene.

Table 2 Multivariate Cox Analysis of Risk Models

Gene	Coef	HR	HR (Lower,0.95)	HR (Upper,0.95)	P
APOL6	−0.401	0.670	0.475	0.944	0.022
ENPP2	0.363	1.438	1.175	1.761	4.4E-04
ETS1	0.214	1.239	0.941	1.631	0.127
GZMA	−0.242	0.785	0.632	0.974	0.028
MX2	0.256	1.291	0.991	1.683	0.058
NLRP3	0.366	1.443	0.914	2.276	0.115
ROCK2	−0.337	0.714	0.506	1.007	0.055
SDS	0.662	1.939	1.177	3.197	0.009
SIRPG	−0.966	0.381	0.205	0.707	0.002
STAT1	−0.271	0.763	0.545	1.068	0.115
ZBED2	0.272	1.313	1.108	1.557	0.002

Figure 5 (A and B) The distribution of RiskScore and corresponding survival states in the Gene Expression Omnibus (GEO) training dataset. (C) Heatmap of gene expression in the RiskScore model. (D and E) Kaplan-Meier curve and receiver operating characteristic curve of RiskScore in the GEO training dataset.

Figure 6 (A) The distribution of RiskScore and corresponding survival states in the Gene Expression Omnibus (GEO) validation dataset, and heatmap of gene expression in RiskScore model. (B and C) Kaplan-Meier (KM) curve and receiver operating characteristic (ROC) curve of RiskScore in GEO validation dataset. (D) The distribution of RiskScore and corresponding survival states in the entire merged GEO dataset, and heatmap of gene expression in RiskScore model. (E and F) KM curve and ROC curve of RiskScore in the entire merged GEO dataset.

Figure 7 (A) Comparison of RiskScore among T1-4 Stage groups in the merged Gene Expression Omnibus (GEO) dataset. (B) Comparison of RiskScore among N1-4 stage groups in the merged GEO dataset. (C) Comparison of RiskScore between M0 and M1 stage groups in the merged GEO dataset. (D) Comparison of RiskScore among stage I-IV groups in the merged GEO dataset. (E) Comparison of RiskScore among grade 1–3 groups in the merged GEO dataset.

Figure 8 (A) T stage distribution between high- and low-risk groups in the merged Gene Expression Omnibus (GEO) dataset. (B) N stage distribution between high- and low-risk groups in the merged GEO dataset. (C) M stage distribution between high- and low-risk groups in the merged GEO dataset. (D) Stage distribution between high- and low-risk groups in the merged GEO dataset. (E) Relapse status distribution between high- and low-risk groups in the merged GEO dataset. (F) Grade stage distribution between high- and low-risk groups in the merged GEO dataset. (G) Age distribution between high- and low-risk groups in the merged GEO dataset. (H) Gender distribution between high- and low-risk groups in the merged GEO dataset.

Figure 9 (A) Comparison of differences in ESTIMATE immunization scores in high- and low-risk groups in the merged Gene Expression Omnibus (GEO) dataset. (B) Differential expression of immunotherapeutic genes in high- and low-risk subgroups in the merged GEO dataset. (C) Analysis of the correlation between immune scores and immunotherapeutic genes in the merged GEO dataset. (D) Comparison of differences in ESTIMATE immunization scores in high- and low-risk groups in The Cancer Genome Atlas (TCGA) dataset. (E) Differential expression of immunotherapeutic genes in high- and low-risk subgroups in the TCGA dataset. (F) Analysis of the correlation between immune scores and immunotherapeutic genes in the TCGA dataset. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, ns, not significant.

Table 3 Univariate and Multivariate Survival Cox Analysis of RiskScore in Merged GEO Datasets

Feature	Univariable Analysis				Multivariable Analysis
	HR	95% CI of HR		P	HR	95% CI of HR		P
		Lower	Upper			Lower	Upper
Age	0.939	0.65	1.359	0.74	1.351	0.867	2.105	0.184
Gender	1.247	0.863	1.803	0.24	0.964	0.636	1.462	0.864
T.Stage	4.037	1.88	8.672	1.00E-05	1.829	0.772	4.336	0.17
N.Stage	2.723	1.876	3.951	1.00E-05	0.454	0.161	1.283	0.136
M.Stage	6.009	3.988	9.055	1.00E-05	3.299	1.908	5.703	1.00E-05
Stage	3.21	2.176	4.736	1.00E-05	2.951	0.94	9.26	0.064
Lymphatic invasion	2.378	1.612	3.509	1.00E-05	1.426	0.889	2.287	0.141
RiskType	1.888	1.296	2.75	0.001	1.229	1.044	1.446	0.013

Figure 10 (A) Nomogram of RiskScore and M stage. (B) Calibration diagram of the nomogram.