1. MKI67 expression in human cancer types
We identified the MKI67 mRNA expression characteristics between pan-cancer and adjacent normal tissues using the TIMER2 database and TCGA data. As shown in Fig. 2A, MKI67 was upregulated across diverse cancer types, including BLCA, BRCA, CESC, CHOL, COAD, ESCA, GBM, HNSC, KICH, KIRC KIRP. LIHC, LUAD, LUSC, PCPG, PRAD, READ, SKCM, STAD, THCA and UCEC. Figure 2B exhibited that the MKI67 expression was upregulated among diverse cancer types consistent with the results of the TIMER2 dataset. We further evaluated MKI67 protein levels between pan-cancer and adjacent normal tissues using the CPTAC dataset in transcriptional levels. Figure 2C indicated MKI67 protein expression was significantly higher among COAD, BRCA, LIHC, LUAD, PAAD, and OV consistent with TIMER2 database and TCGA data. Overall, MKI67 expression was significantly upregulated across varied cancer types.
2. Pan-cancer analysis of the correlation between MKI67 expression and clinicopathology
After identifying the characteristics of MKI67 expression at the mRNA and protein levels, we explored the association between MKI67 expression and clinicopathological features and clinical parameters across different cancer types using the TCGA database and the10028 cancer patients we collected, respectively. Figure 3 revealed that there were significantly different MKI67 expressions of stage I, II, III, and IV among LUADLUSC, LIHC, BRCA, and THCA based on the TCGA database. Additionally, there were not statistical different MKI67 expressions of stage I, II, III, and IV among COADREAD, CESC, STAD, PAAD, and ESCA in the TCGA database. We then probed the correlation of MKI67 expression with clinical parameters in10028 cancer patients collected including LUADLUSC, COADREAD, THCA, LIHC, STAD, BRCA, ESCA, CESC, PRAD and PAAD. Table S1 shown poor differentiation, TNM classification, classification, and staging in BRCA patients were correlated with high expression of MKI67. Table S2 indicated the high expression of Ki67 was closely related to TNM classification, clinical stage and pathological type in LUADLUSC patients. Table S3 indicated the BCLA stage, Edmondson grade, tumor size and tumor nodule increase in LIHC patients were closely related to the high expression of Ki67. Table S4 displayed the high expression of Ki67 was associated with N classification, M classification, pathological type, and differentiation degree in COADREAD patients. Whereas, the clinical information of CESC, STAD, and ESCA was too little to be analyzed (in Annex 1), and the amount of data of THCA, PRAD and PAAD were too small to be validated. Above all, the results of TCGA database and our validation present that the high MKI67 expression was associated with certain clinicopathological features and clinical parameters, for instance, poor differentiation, TNM classification, and clinical stage, among some cancer types. However, Table S5-7 showed that the expression of MKI67 was unrelated to the clinicopathology of STAD, ESCA, and CESC.
3. Prognostic value of MKI67 expression in our validation cohort and TCGA database by pan-cancer analysis
To investigate the prognosis value of MKI67 expression among various cancer types, we then explored the correlation between MKI67 expression and prognosis of patients with different cancer types based on the TCGA database. We used several survival metrics including OS, DSS, DFS, and PFS to evaluate the prognosis value of MKI67 expression by Cox regression analysis. As shown in Fig. 4A, MKI67 expression was significantly associated with OS in 10 cancer types, including ACC, KIRP, LGG, LIHC, LUAD, MESO, PAAD, SARC, SKCM, and THYM. MKI67 was a risk factor in these cancer types except THYM. Figure 4B displayed MKI67 expression was remarkably related to DSS of 9 cancer types, including ACC, KIRC, KIRP, LGG, LIHC, LUAD, MESO, PAAD, and SKCM. Figure 4C indicated that increased MKI67 expression was significantly associated with KIRP, LIHC, PAAD, STAD, and THCA of DFS. Notably, MKI67 was a risk factor for death in patients with KIRP, LIHC, PAAD and THCA but a protective factor for STAD. Figure 4D demonstrated MKI67 expression appreciably affected PFS in patients with KIRP, LIHC, PAAD, STAD, and THCA. Similarly, STAD was a protective factor for STAD but the others were a risk factor.
We further focused on the relationship between MKI67 expression and OS using Kaplan-Meier analysis based on the 10028 cancer patients collected and TCGA database. We used ROC curve to determine the cut-off value of several cancer types based on the results of IHC, to divide patients into low and high Ki67 expression groups. According to the maximized Youden Index indexes, we identified the cut-off value among different cancer types, including LUADLUSC, LIHC, BRCA, COADREAD, STAD, ESCA, and CESC, with an optimal cut-off value of 21.5%, 31.5%, 36.5%, 66.5%, 69%, 67.5%, and 74%, sequentially. Figure 5 displayed typical IHC landscapes of low and high Ki67 expression groups in several cancer types. Based on the top 10 cancers in China, i.e. LUADLUSC, COADREAD, THCA, LIHC, STAD, BRCA, ESCA, CESC, PRAD and PAAD, we probed the association between MKI67 expression and OS by the 10028 cancer patients collected. However, the analysis of THCA, PRAD and PAAD, were not available due to limited data. Fig. 5. Protein expression level of MKI67 in human multiple cancer tissues of COADREAD (A), BRCA (B), STAD (C), CESC (D), LIHC (E), LUADLUSC (F) and ESCA (G). Representative images of MKI67expression in pan-cancer tissues are shown. Original magnification, ×200 and ×400.
Specially, we added the subtype of COADREAD, i.e. COAD and READ to further explore the prognosis of MKI67 expression. Figure 6(A) exhibited increased MKI67 expression was significantly associated with poor prognosis in LUADLUSC, LIHC, and BRCA patients, but good prognosis in COADREAD and READ patients. Meanwhile, we verified the association between MKI67 expression and OS using TCGA database. As shown in Fig. 6(B), the results indicated increased expression of MKI67 was notably associated with poor prognosis in patients with LUADLUSC, LIHC, and PAAD. To better compared the results of our validation and TCGA database, we used Table 1 to clearly demonstrate the similarities and differences. Table 1 indicated the results of our validation were generally consistent with TCGA database, except for BRCA and COADREAD. Table 1 suggested MKI67 expression had no effect on the prognosis of BRCA patients and COADREAD patients in TCGA database, while the increased MKI67 expression was significantly associated with poor prognosis of BRCA patients, but with good prognosis of COADREAD patients in our verification results. This may be due to inter-tumor/intra-tumor heterogeneity or the difference in patient population, such as racial or regional disparities. Above all, the prognosis value of MKI67 expression in several cancer types was precisely identified in our validation cohort and TCGA database, of which the results were generally consistent and reliable.
Table 1
The comparisons of results between our validation and TCGA database.
Clinical parameters | Verify OS | OS in TCGA database |
---|
n | Hazard ratio(95%CI) | p-Value | n | Hazard ratio(95%CI) | p-Value |
---|
LIHC | 1454 | 2.039(1.735–2.397) | <0.001 | 424 | 1.639(1.160–2.315) | 0.005 |
LUADLUSC | 3140 | 1.682(1.531–1.848) | <0.001 | 1149 | 1.226(1.008–1.492) | 0.042 |
LUAD | 2521 | 1.725(1.554–1.915) | <0.001 | 598 | 1.397(1.049–1.860) | 0.022 |
LUSC | 615 | 1.212(0.851–1.728) | 0.3165 | 551 | 0.881(0.673–1.153) | 0.353 |
BRCA | 2065 | 2.752(2.105–3.597) | 0.0001 | 1226 | 1.122 (0.816–1.542) | 0.479 |
COADREAD | 1261 | 0.726(0.561–0.940) | 0.0136 | 643 | 0.815(0.577–1.152) | 0.243 |
STAD | 904 | 1.127(0.954–1.331) | 0.149 | 409 | 0.842(0.607–1.168) | 0.299 |
CESC | 962 | 0.991(0.763–1.287) | 0.945 | 309 | 0.843(0.530–1.340) | 0.467 |
ESCA | 138 | 1.172 (0.610–2.252) | 0.614 | 174 | 1.046(0.646–1.695) | 0.853 |
4. Correlation analysis between MKI67 expression and immune cell infiltration
After identifying the prognosis value of MKI67 expression, we investigated the potential relationship between MKI67 expression and tumor-infiltrating immune cells, an essential component of the Tumor Microenvironment (TME), across various cancer types by xCell algorithm. As demonstrated in Fig. 7A, it is revealed that MKI67 expressed in 38 immune cell subtypes were generally significantly contributed to the level of tumor-infiltrating immune cells in several cancer types. In particular, MKI67 expression were most positively correlated with Th1 and Th2 CD4+ T cells across various cancer types, while MKI67 expression were largely negatively correlated with Macrophage M2 cells across various cancer types. Notably, MKI67 expression have different effect on diverse cancer types, such as positively related to Macrophage M1 cells in BLCA, BRCA, KIRC, LUAD, and THCA; negatively related to Macrophage M1 cells in CESC, GBM, LUSC, READ, TGCT, and THYM. We then verified the relationship between MKI67 expression and tumor-infiltrating immune cells via the TIMER algorithm. Figure 7B indicated that the expression of MKI67 was significantly correlated with the tumor purity of 14 cancers and the degree of B-cell invasion of 23 cancers. MKI67 was also associated with CD4+ T cell invasion in 22 cancers, CD8+ T cell invasion in 19 cancers, DC invasion in 25 cancers, neutrophil invasion in 24 cancers, and macrophage invasion in 17 cancers. Above all, the results revealed that MKI67 expression plays diverse functions in different cancer types, which may partially explain MKI67 performed an opposing impact against the prognosis of various cancer types.
5. Correlation analysis of MKI67 expression with TMB and MSI
We further explored the relationship between MKI67 expression and dynamic immune-related features, including TMB and MSI. TMB and MSI are two emerging biomarkers related to immunotherapy response. The results showed that MKI67 expression was significantly positively correlated with TMB in various cancer types, including ACC, KICH, STAD, PAAD, BRCA, LUAD, CHOL, and UCS, while it was negatively correlated with TMB in THYM (Fig.S1A). As shown in Fig.S1B, MKI67 expression was positively correlated with MSI in LUSC, STAD, ACC, UCEC, UVM, UCS, and MESO, but negatively correlated with SKCM, PCPG, and DLBC. Moreover, we compared the potential association of MKI67 expression with eight immune checkpoint pathway genes in pan-cancer. The results showed that MKI67 was significantly positively correlated with several of the pan-cancer immune checkpoint genes in some cancer types, such as THCA, STAD, LIHC, and BRCA, but also negatively correlated with the immune checkpoint genes in some cancer types, including THYM and GBM(Fig.S1C). These results revealed that MKI67 was capable of playing a vital role in immunotherapy response.
6. Enrichment analysis of MKI67 related genes in pan-cancer
Next, we analyzed the MKI67 protein-protein interaction using the String database to further explore the probable molecular mechanisms of it in tumor prognosis(Fig.S2). Then, to investigate the functional impact of the MKI67 gene, we used the GEPIA2 database to extract the top 100 genes with similar expression patterns to MKI67 in all tumor types. Among them, the first 10 genes were positively correlated with MKI67, i.e., IKZF1, DOCK2, NCKAP1L, ARHGAP30, DOCK8, FLI1, VAV1, AKNA, ARHGAP9, and PTPRC, orderly (Fig.S3). We then performed GO and KEGG analyses for these 100 genes. The GO analysis is divided into three parts: GO_MF, GO_BP, and GO_CC (Fig.S4A). Under the GO_BP component, MKI67 is associated with cell proliferation and division in several tumors, such as organelle fission, nuclear Division, and chromosome segregation. For GO_CC, MKI67-related gene products are located in the spindle, chromosomal region, chromosome, and centromeric region and simultaneously perform functions. In the GO_MF analysis, the results indicated that MKI67 related genes mostly had tubulin binding, microtubule binding, and ATP hydrolysis activity. As demonstrated in Fig.S4B, KEGG pathway analysis showed that the 100 genes were mainly related to Cell cycle, Oocyte meiosis, Progesterone mediated oocyte maturation and Human T-cell leukemia virus 1 infection ", "Cellular senescence", and "p53 signaling pathway".