Transcriptional information underlying the generation of CSCs and the construction of a nine-mRNA signature to improve prognosis prediction in colorectal cancer

Figures & data

Figure 1. (a) Unsupervised hierarchical clustering analysis of the differentially expressed mRNAs between CD133- and CD133+ Caco-2 cells. (b) The differentially expressed genes between CRC and normal tissues, in which red indicates high expression, and green indicates low expression.

Table 1. Summary of colorectal cancer patient clinical characteristics based on the inclusion criteria.

Characteristic	Patients (n = 476)
	n	%
Age category
>65 y	269	56.513
≤65 y	207	43.487
Gender
Male	256	53.782
Female	220	46.218
Race
White	219	46.008
Asian	9	1.891
Black or African American	52	10.924
Unknown	196	41.177
Pathological Stage
Stage I	85	17.857
Stage II	180	37.815
Stage III	126	26.471
Stage IV	70	14.706
Unknown	15	3.151
Eventual prognosis
Alive	394	82.773
Dead	82	17.227

Table 2. Enrichment analysis of GO terms for the 155 differentially expressed genes.

ID	Description	P value	Adj. p value	Count
GO:0048018	Receptor ligand activity	7.85e-08	3.12e-05	18
GO:0005126	Cytokine receptor binding	5.81e-07	0.0001154	13
GO:0008083	Growth factor activity	9.00e-06	0.0011907	9
GO:0005125	Cytokine activity	1.57e-05	0.0015566	10
GO:0045236	CXCR chemokine receptor binding	0.0003522	0.0236768	3
GO:0008201	Heparin binding	0.0003578	0.0236768	7
GO:0001664	G protein-coupled receptor binding	0.0004484	0.0254319	9
GO:0031418	L-ascorbic acid binding	0.0005798	0.0287703	3
GO:0005160	Transforming growth factor beta receptor binding	0.0007285	0.0321348	4
GO:0043621	Protein self-association	0.001053	0.0391582	4
GO:0019842	Vitamin binding	0.001085	0.0391582	6

Figure 2. Acquisition of 155 differentially expressed genes and GO and KEGG functional enrichment analyses. (a) There were 393 differentially expressed genes between CD133- and CD133+ Caco-2 cells, 11,832 genes differentially expressed between colorectal tumor and normal tissues, and 155 genes coexisted among them. (b) GO functional enrichment and (c) KEGG pathway analyses of these 155 genes.

Figure 3. (a) Protein association networks of 155 genes. (b) The top 30 genes with the highest degree of correlation.

Figure 4. Genes associated with patient’s survival outcomes by applying the K-M method in “survival” R package.

Table 3. Overall information of 9 prognostic mRNAs associated with OS in CRC patient.

Ensembl ID	Gene symbol	Relative coefficient	Z value	P value
ENSG00000002726	AOC1	−0.268184925	−3.279204172	0.001041003
ENSG00000163794	UCN	0.145612613	1.455843705	0.145435831
ENSG00000129422	MTUS1	−0.361412144	−2.749916882	0.005961038
ENSG00000117399	CDC20	−0.46844737	−3.342579065	0.000830037
ENSG00000074317	SNCB	0.202331807	1.924286475	0.054318685
ENSG00000151224	MAT1A	0.176725272	3.168886703	0.00153024
ENSG00000137285	TUBB2B	0.115172814	1.81508945	0.069510164
ENSG00000109158	GABRA4	−0.086924131	−1.656027435	0.097716263
ENSG00000163283	ALPP	0.125962491	2.278719861	0.022683723

Figure 5. (a) Distribution of mRNA-related RSs in patients. (b) The expression heatmap of nine prognostic mRNAs. (c) The mortality of patients with an increase in the RS. (d) K-M survival curve analysis for the OS of CRC patients using the nine-mRNA signature. (e) ROC curve analysis of the nine-mRNA signature.

Table 4. Multivariate Cox regression analysis of overall survival.

Variable	HR	SE	Z value	P value	95% CI of HR
Age	1.04852	0.01141	4.154	3.27e-05	1.0253–1.072
Gender	0.91535	0.2382	−0.371	0.71	0.5739–1.46
Race	0.92524	0.16807	−0.462	0.644	0.6656–1.286
Stage	2.39071	0.13337	6.535	6.37e-11	1.8408–3.105
Nine-mRNA RS (high vs. low)	3.8438	0.2488	5.411	6.25e-08	2.36–6.26

Figure 6. (A ~ F) Patients divided according to the pathological stages of the tumor or age and analyzed for differences in the corresponding survival based on the signature.

Figure 7. The survival difference of stage IIA and stage IVA between high and low risk groups of CRC patients.

Figure 8. Expression levels of the nine mRNAs across four colorectal cancer subtypes.

Availability of data and materials

The data that support the findings of this study are openly available in the TCGA Research Network: http://cancergenome.nih.gov/ and GEO database: https://www.ncbi.nlm.nih.gov/geo/. Also can be availabled from the corresponding author upon reasonable request.