The differential expression of NMB between CRC tumor tissues and normal tissues
Scatter difference plot(p<0.001, Figure.2A) and paired difference plot(p<0.001, Figure.2B) showed that the expression level of NMB in CRC tumor tissues was significantly higher than that in normal tissues. The same results were confirmed in GEPIA(p<0.001, Figure.2C).
High expression of NMB in CRC tumor tissue suggests poor overall survival
We evaluated the association between NMB expression and prognosis in CRC patients using Kaplan-Meier risk estimates. Compared with low NMB expression, high NMB expression was associated with significantly poorer overall survival(OS)(p<0.001, Figure.2D).The same results were obtained in GEPIA(p=0.012, Figure.2E). We used expression data from 44 normal samples and 568 tumor samples to draw ROC curves(Figure.2F) to evaluate the diagnostic value of NMB in CRC.The area under the ROC curve(AUC) AUC at 1 years: 0.622;AUC at 3 years: 0.576;AUC at 5 years: 0.669.Therefore, NMB has considerable diagnostic value for CRC.
We performed survival analysis on 8 datasets in the GEO database using Kaplan-Meier risk estimates. (Figure.3A-H).Then,Meta-analysis was performed on the above analysis results( HR =1.0459,95%-CI:1.0067-1.0865,z=2.31 ,p-value=0.0211).The results of the meta-analysis were visualized using the forest map(Figure.3I).Because I2<50% and P>0.05(I2=1%,P=0.43),we chose the fixed effect model.The results of this meta-analysis suggest that NMB is indeed a high-risk gene in colorectal cancer.(HR=1.05,95%-CI:1.01-1.09)
Relationship between expression of NMB and clinicopathological features.
Clinical data from 711 patients with CRC from TCGA were analyzed and unknown and incomplete clinical information was deleted.The expression of NMB was only correlated with age(p=0.01,Figure.4A),but not with gender(p=0.293,Figure.4B), T classification(p=0.209,Figure.4C) and N classification(p=0.088,Figure.4D), distant metastasis(M)(p=0.838,Figure.4E) and clinical stage(p=0.919,Figure.4F).Similarly, Logistic regression analysis showed that the expression of NMB in colorectal cancer was only correlated with age(OR = 1.504 for >=65 versus <65, P = 0.019), and there was no significant correlation with gender(OR = 1.178 for Male versus female, P = 0.343), T classification(OR = 0.327 for T2 versus T1, P = 0.071;OR = 0.368 for T3 versus T1, P = 0.092;OR = 0.310 for T4 versus T1, P = 0.066),N classification(OR = 0.807 for N1 versus N0, P = 0.306;OR = 1.475 for N2 versus N0, P = 0.104),distant metastasis(M)(OR = 1.139 for M1 versus M0, P = 0.603), and stage (OR = 0.969 for stage II versus stage I, P = 0.901; OR = 0.927 for stage III versus stage I, P = 0.0.775;OR = 1.059 for stage IV versus stage I, P = 0.855)(Table 1).
NMB Is an Independent Predictor of Poor Survival in CRC
The univariate and multivariate Cox proportional hazard regression analyses were used to evaluate whether the expression of NMB could be an independent predictor of poor survival in CRC patients.After the deletion of unknown or incomplete clinical information.We performed univariate and multivariate Cox proportional hazard regression analyses in 472 patients with CRC.The univariate analysis showed that age (HR, 1.039; 95% CI, 1.017-1.062; p <0.001), stage (HR, 2.293, 95% CI, 1.794-2.932, p < 0.001), T classification (HR,2.890; 95% CI, 1.887-4.427; p <0.001),lymph node (HR, 2.073; 95% CI,1.606-2.675; p <0.001), distant metastasis (HR, 4.512; 95%CI, 2.888-7.049; p < 0.001), and high-NMB expression (HR,1.046; 95% CI, 1.005-1.088; p =0.027) were important predictors of survival(Figure 5A,Table 2).The multivariate analysis showed that age (HR, 1.47; 95% CI, 1.025-1.069; p <0.001) ,T classification (HR,1.733; 95% CI, 1.068-2.813; p =0.026)and high-NMB expression (HR,1.054; 95% CI, 1.008-1.102; p =0.021)were the important independent predictors of poor overall survival of CRC(Figure 5B and Table 2).
GSEA identifies the NMB-related signaling pathway.
To identify differentially activated signaling pathways in CRC, we performed the Gene Set Enrichment Analysis (GSEA) between low NMB and high NMB expression datasets. Various cancer-related KEGG pathways are enriched in high NMB phenotypes(Figure 6), for example, Cell cycle, DNA replication, P53 Signaling pathway, and VEGF signaling pathway. While KEGG pathways enriched in low NMB phenotypes were Colorectal cancer, ERBB Signaling pathway, JAK-STAT Signaling pathway, MAPK Signaling pathway, MTOR Signaling pathway, NOTCH Signaling pathway, pancreatic cancer,pathways in cancer, TGF-BETA Signaling pathway, and WNT Signaling pathway.
Enrichment analysis of NMB co-expressed genes
The co-expressed genes have similar functions and mechanisms. To further investigate the underlying mechanisms of NMB regulation, we used the Linkedomics Platform to identify 6512 significant genes. The volcano map (Figure 7A) shows that there is a correlation between global genes and NMB by Pearson test. Heat maps (Figure 7B) show the top 50 genes in CRC that are negatively and positively correlated with NMB.ORA showed that co-expressed genes were involved in Hepatocellular carcinoma cell cycle, RNA transport, RNA processing, Metabolism of RNA, negative regulation of gene expression, chromosome, EGF/EGFR Signaling Pathway, VEGFA-VEGFR2 Signaling Pathway, TGF-beta Signaling Pathway, Negative regulation of NOTCH4 signaling, Gene expression (Transcription), and Metabolism of proteins(Figure 8). Then, GSEA was performed to investigate the potential functions and pathways of NMB induction. We explored three main types of GO enrichment: biological process (BP), cellular component (CC), and molecular function (MF). In the BP category, we explored ribonucleoprotein complex biogenesis, tRNA metabolic process, generation of precursor metabolites and energy, establishment of protein localization to membrane, protein-containing complex disassembly, regulation of GTPase activity, regulation of vasculature development, cell-cell adhesion via plasma-membrane adhesion molecules, peptidyl-serine modification, and Ras protein signal transduction(Figure 9A).In the CC category, we explored mitochondrial inner membrane, mitochondrial matrix, cytosolic part, condensed chromosome, cell-cell junction, early endosome, cell leading edge, and apical part of cell(Figure 9B).In the MF category, we explored structural constituent of ribosome, catalytic activity, acting on RNA, lyase activity, guanyl-nucleotide exchange factor activity, cofactor binding, catalytic activity, acting on DNA, modification-dependent protein binding, protein serine/threonine kinase activity, nucleoside-triphosphatase regulator activity, and phospholipid binding(Figure 9C).For KEGG pathway analysis, we explored Ribosome, Proteasome, Spliceosome, Metabolic pathways, RNA transport, Parathyroid hormone synthesis, secretion and action,Proteoglycans in cancer, ECM-receptor interaction, Ras signaling pathway, and JAK-STAT signaling pathway. (Figure 9D)
Relationship Between NMB Expression and Immune Infiltration in CRC
Tumor infiltrating lymphocytes is an independent prognostic factor for survival in cancer patients.Like breast cancer[10], ovarian cancer, colorectal cancer, and gastric cancer. Therefore, we investigated the relationship between NMB expression and Immune Infiltration in colorectal cancer. The histogram shows the number of immune cells in each sample(Figure 10A). The difference test between the expression of NMB and immune cells showed that T cells CD8(p<0.001), T cells CD4 memory resting(p=0.023), NK cells activated(p=0.029), and Macrophages M0 (p=0.016)were different in the high expression group and the low expression group(Figure 10B).Correlation test showed that the expression of NMB was correlated with T cells CD8(R=0.24,p<0.001),T cells CD4 naive(R=-0.14,p=0.014),T cells CD4 memory resting(R=-0.14,p=0.011),T cells CD4 memory activated(R=0.16,p=0.0029),T cells follicular helper(R=0.19,p<0.001),Macrophages M0(R=-0.13,p=0.017) and Macrophages M1(R=0.12,p=0.026).After the intersection of the two test results, it can be concluded that the expression level of NMB is significantly correlated with T cells CD8, T cells CD4 memory resting, Macrophages M0(Figure 10C).Besides,different mutational forms of NMB in CRC were associated with immune infiltration of 6 leukocytes(B cell,CD4+ T cells,CD8+ T cells,macrophage,neutrophil,dendritic).So it follows that NMB plays an important role in immune infiltration in CRC.
Correlation Between Expression Level of NMB and Immune Marker Sets
To further investigate the relationship between NMB expression and immunoinfiltrating cells in colorectal cancer,we used TIMER database to detect the immune markers of T cells,CD8 + T cells, B cells,monocytes,neutrophils, NK cells, TAMS,M1 and M2 macrophages, and dendritic cells.Then,we also analyzed T cells with different functions, such as Th1 cells, Th2 cells, Tregs, Tfh cells, Th17 cells, and depleted T cells.Our results show that the expression level of NMB in CRC is closely related to immune marker sets in most immune cells(Table 3).After adjustment for correlation of tumor purity,It turns out that CD3E,HLA-DPB1,CD3D,CD79A,TGFB1,CD2,HLA-DPA1,ITGAX,HLA-DRA,CD86,CCR7,CD19,NRP1,HAVCR2,CTLA4,TBX21 showed a significant correlation with NMB expression in colon cancer(P < 0.001; Cor value ≥ 0.40).And in rectal cancer,CD3E,HAVCR2,IL10,CD1C,CEACAM8,CCR7,STAT3,CTLA4,ITGAM,HLA-DPB1,FOXP3,CCL2,KIR3DL1,STAT1 showed a significant correlation with NMB expression(P < 0.001; Cor value ≥ 0.40).
Construction of the Prognostic Signature for CRC Patients
Using the survival information of CRC patients to perform univariate Cox regression analysis on the 162 DEGs, it was found that there are 6 DEGs with significant prognostic differences. Perform multivariate Cox analysis on 6 genes with prognostic significance, and construct a prognostic signature composed of 4 genes, including NDUFB10, SERF2, DPP7, and NAPRT.Based on the prognosis signature, the risk score calculation formula was obtained: risk score = (0.547443833 ×DPP7 expression) + (-1.222239025 × NDUFB10 expression) + (0.400602782 × NAPRT expression) + (0.792523684 × SERF2 expression).The risk score was calculated for each CRC patient, and patients were divided into high a risk-group (n = 270) and a low-risk group (n = 270) based on the median.We constructed a heat map to show the expression of 4 genes in the high-risk group and the low-risk group, and the expression of 4 genes in the high-risk patients was higher than that in the low-risk patients(Figure.11A).Figure.11B shows the distribution of risk scores for CRC patients. Patients are divided into two groups, with risk scores increasing from left to right.Figure.11C shows the distribution of risk scores for CRC patients. Patients are divided into two groups, with risk scores increasing from left to right. K-M curve was used to compare the difference in OS time between the high-risk group and the low-risk group (Figure.11D). Results showed that CRC patients with a high-risk score had significantly lower OS than GC patients with a low-risk score(P<0.001). We plotted a time-dependent ROC curve to predict survival in CRC patients, showing that the risk score had high sensitivity and specificity.AUC of risk score(AUC=0.711) was higher than that of age(AUC=0.646),gender(AUC=0.433),stage(AUC=0.709),T stage(AUC=0.673),N stage(AUC=0.656) and M stage(AUC=0.647) (Figure.11E).Figure.11F reflects the univariate Cox analysis of the relationship between the clinical features, risk score, and OS of CRC patients.Age(P <0.001),stage(P < 0.001),T(P <0.001),N(P <0.001),M(P <0.001) and risk score(P < 0.001) significantly affect the prognosis of GC patients.Figure.11G reflects a multivariate Cox analyzed the relationship between the clinical features, risk score, and OS of GC patients. Age (P < 0.001) ,T(P=0.019) and risk score (P < 0.001) are independent prognostic risk factors for CRC.
Construction and Validation of the Nomogram
We used factors such as age, stage, T, M, N, and risk score to construct a nomogram to predict the survival rate of CRC patients more conveniently(Figure.11H). According to the nomogram, the scores of CRC patients are calculated and then added to obtain the total score, thereby predicting the survival probability of 1 year and 3 years, which is beneficial to guide clinical decision-making. Because the closer the calibration curve is to the diagonal, the more accurate the prediction result will be. The calibration curves of the nomogram show that the nomogram has good accuracy in predicting survival rates at 1 and 3 years(Figure.11I, J). The 1-year(AUC = 0.711) and 3-year(AUC = 0.712) ROC curves also show that the forecasting ability of the nomogram is very accurate(Figure.11K).
Analysis of drug sensitivity
We selected 162 NMB-related genes through the co-expression network(fdrFilter=0.00000001,logFCfilter=10), and performed Spearman correlation analysis with small molecule/drug sensitivity (IC50) to explore the correlation between DEGs and drug sensitivity. The results showed that NMB was significantly related to the drug resistance of many chemotherapeutic drugs and tumor-targeted drugs, including 5-Fluorouracil, Methotrexate, Belinostat, CUDC-101, vorinostat, and so on(Figure.12).