- Academic Editor
Background: Clear cell renal cell carcinoma (ccRCC) is a common malignant tumor of the urinary system characterized by abundant immunocytes infiltration. The impact of guanosine triphosphatases (GTPases) of immunity-associated proteins (GIMAPs) on the tumor immune microenvironment (TIME) and prognosis of ccRCC is unclear. Methods: The expression of GIMAPs in ccRCC was determined through multiple datasets (ONCOMINE, TCGA and UALCAN). The relationship between GIMAP family members was analyzed through Spearman correlation analysis. The interaction among the GIMAPs protein was analyzed using STRING. Prognostic values of GIMAPs were evaluated by Survival analysis, Lasso and Cox regression analysis; Prognostic risk model and nomogram were constructed. The correlation between GIMAPs and TIME was explored using TIMER, Cibersort and Pearson correlation analysis. Gene set enrichment analysis (GSEA) was performed to discuss their function and mechanism in ccRCC. Results: GIMAPs were over-expressed in ccRCC and significantly related to overall survival (OS) of the patients. GIMAPs were positively correlated with each other, the risk model based on GIMAPs had good prognostic value in ccRCC. GIMAPs mainly expressed in TIME and were associated with abundant immunocytic infiltration in ccRCC, the risk model also had close correlation with TIME. Our results showed GIMAPs may affect the development of ccRCC by regulating the amount and antitumor activity of immunocytes in TIME. Conclusions: GIMAPs were over-expressed in ccRCC, and their expression levels were significantly related to the OS of patients and immunocytic infiltration in TIME. GIMAPs are potential therapeutic targets and prognostic biomarkers for ccRCC.
Renal cell carcinoma is an aggressive
urinary system malignancy, endangering human health. In 2020, over 4,000,000 new
cases and about 1,800,000 deaths had been reported globally [1]. Clear cell renal
cell carcinoma (ccRCC) is the most common histological subtype of renal cell
carcinoma, accounted for approximately eighty percent of all
cases [2]. Patients with early ccRCC can achieve favorable outcome by radical
surgery, but patients with advanced ccRCC and postoperative metastasis have poor
prognosis [3, 4]. CcRCC is insensitive to chemotherapeutic drugs and
radiotherapy. Interleukin-2 and
Human guanosine triphosphatases of immunity-associated proteins (GIMAP) family genes are located on chromosome 7, spanning about 500 KB, and include seven functional genes (GIMAP1, GIMAP2, GIMAP4, GIMAP5, GIMAP6, GIMAP7, GIMAP8) and a pseudogene [15]. The GIMAP proteins are similar in N-end sequence and contain guanine nucleotide binding domain called GTPase [15, 16]. Most of these proteins participate in the maintenance and development of lymphocytes. The deficiency of GIMAP5 leads the decrease of peripheral T, B cells and natural killer (NK) cells in mice [17, 18]. GIMAP1 is important for the maintenance of T cells’ proliferation and mature the function of B cells [19, 20]. GIMAP4 may promote T cell apoptosis [21]. Knocking out GIMAP6 makes Jurkat T cells more susceptible to apoptosis inducers [22]. Recent studies have found the dysregulation of GIMAPs in a number of tumor types, including hepatocellular cancer, endometrial cancer and non-small cell lung cancer, GIMAPs were significantly associated with the prognosis of patients with lung adenocarcinoma and endometrial cancer [23, 24, 25, 26]. CcRCC is characterized by the infiltration of abundant immunocytes, nonetheless, the influence of the GIMAP family on the immunological microenvironment of ccRCC and the prognosis of ccRCC patients is unclear.
In this study, we sought to investigate the impact of GIMAPs on the immunological microenvironment and prognosis of ccRCC. We identified GIMAPs’ expression in ccRCC from multiple databases. Then, we discussed the impact of GIMAPs on outcomes in ccRCC patients. We then elucidated the interrelationship between GIMAPs and the immunocytes in ccRCC and discussed the molecular mechanism of GIMAPs affecting the development of ccRCC. Our results showed that GIMAPs could regulate TIME, affect the prognosis of these patients and are potential therapeutic targets for ccRCC.
We contrasted the expression of GIMAPs mRNA in normal renal tissues and ccRCC using ONCOMINE (https://www.oncomine.org, accessed on 14 October 2021), which can provide powerful and reliable function to analyze multiple expression of gene characteristics of tumors in the Gene Expression Omnibus, TCGA and published literature [27]. Then, we downloaded RNA-seq and clinical data of patients with ccRCC from TCGA (https://portal.gdc.cancer.gov/, accessed on 16 December 2021) and UCSC Xexa (https://xenabrowser.net/datapages/, accessed on 31 December 2021). We were able to analyze clinical data for 529 patients and RNA sequencing data for 530 tumor samples and 72 normal samples, the expression of genes are normalized as log2 (TPM + 1). The expression of GIMAPs mRNA between them were compared to verify the results in ONCOMINE. GIMAPs’ effect was evaluated by receiver operating characteristic (ROC) curves and multivariate logistic regression analysis for the diagnosis of ccRCC. The protein expression of GIMAPs in normal renal tissues and ccRCC were compared by UALCAN (http://ualcan.path.uab.edu, accessed on 21 October 2021). UALCAN is an online tool that can analyze tumor patients’ transcriptome data and clinical parameter in TCGA, also the expression of proteins of tumors in the Clinical Proteomic Tumor Analysis Consortium (CPTAC) [28, 29].
STRING (version: 11.5, https://cn.string-db.org/, accessed on 3 November 2021) was utilized to evaluate the interreaction among the GIMAPs protein and construct the network diagram. STRING is a database containing many types of protein-protein interactions [30].
Patients’ inclusion criteria: the patients with RNA-seq data and complete survival data (OS and survival status). 527 patients were included. On the basis of the median expression of GIMAPs, the patients were divided into Low and High groups. The effect of GIMAPs on OS of patients with ccRCC in TCGA was assessed using the Kaplan Meier (KM) curve. To further analyze the effect of GIMAPs and clinical parameters (tumor stage, tumor grade, tumor longest dimension, gender and age) on ccRCC patients’ prognosis, patients were randomly split into two sets (a training set: 322 patients (61%) and a test set: 205 patients (39%)). In the training set, Lasso regression analysis was used to select the genes in GIMAPs to construct the prognostic risk model, then risk scores of patients both in the training set and test set were calculated. Cox regression analysis was utilized to determine the independent prognostic factors in risk scores and clinical parameters, then the independent prognostic factors were used to construct a nomogram. The performance of the risk model and nomogram were comprehensively evaluated using the KM curve, calibration curve, concordance index, and ROC curve.
The interrelationship among GIMAPs family members and between GIMAPs expression and immunocytes in ccRCC was evaluated using TIMER (https://cistrome.shinyapps.io/timer/, accessed on 21 October 2021). The amount of infiltrating immunocytes were estimated by the TIMER algorithm. The correlation was identified by Spearman correlation analysis. TIMER is online tool that can analyze immunocytes infiltration of different cancers from TCGA [31]. TME scores of ccRCC in TCGA were downloaded from ESTIMATE (https://bioinformatics.mdanderson.org/estimate/, accessed on 18 October 2022), immunocytes infiltration scores of ccRCC were calculated using CIBERSPRT algorithms [32].
The co-expression analysis was carried out using LinkedOmics (http://linkedomics.org, accessed on 2 November 2021) to obtain the genes co-expressed with GIMAPs in ccRCC. The ccRCC patients’ data were from TCGA, and the statistical method was Pearson correlation analysis. LinkedOmics is a web service that can analyze various tumor patients’ clinical parameters and gene expressions in CPTAC and TCGA. Webgestalt (http://www.webgestalt.org, accessed on 2 November 2021) was used for gene set enrichment analysis (GSEA) of the genes co-expressed with GIMAPs. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were chosen for GSEA, Rank Criteria: Pearson correlation coefficient (PCC); Minimum Number of Genes: 3; Simulations: 1000. WebGestalt is an online tool that can perform functional enrichment analysis of genes of multiple species [33].
The Student’s t-test in SPSS (25.0.0.0, IBM Corp., Chicago, IL, USA)
was used to compare the GIMAPs mRNA expression in the ccRCC tumor and normal
renal tissues. In evaluating the value of GIMAPs for ccRCC’s diagnosis,
“plotROC” package was used to plot ROC curves and calculated Area Under Curve
(AUC), and the “glm” function was used for the multivariate logistic regression
analysis. In evaluating the impact of GIMAPs on the prognosis of ccRCC patients,
“glmnet” package was used to perform Lasso regression analysis, and
“survival”, “survminer” and “survivalROC” packages were used to perform Cox
regression analysis, plot Kaplan Meier survival curves, nomogram, the ROC curves
and calibration curves, and calculate concordance index. R (v.4.2.1) was used to
perform the statistical analysis, p
In ONCOMINE, Yusenko et al. [34] and Lenburg et al. [35] Renal dataset both showed higher mRNA
expression of GIMAP2, GIMAP4, GIMAP5, GIMAP6 and GIMAP7 in ccRCC tumor tissues
than normal renal tissues (p
GIMAPs | ccRCC tumor cases | Normal renal cases | Fold Change | p-value | t-score | Dataset |
GIMAP1 | 26 | 5 | 2.622 | 4.29 × 10 |
4.554 | Yusenko Renal |
GIMAP2 | 9 | 9 | 2.536 | 9.24 × 10 |
4.326 | Lenburg Renal |
GIMAP2 | 26 | 5 | 4.974 | 3.98 × 10 |
12.909 | Yusenko Renal |
GIMAP4 | 10 | 10 | 3.300 | 8.75 × 10 |
5.008 | Gumz Renal |
GIMAP4 | 23 | 23 | 2.317 | 1.93 × 10 |
10.214 | Jones Renal |
GIMAP4 | 9 | 9 | 2.112 | 2.32 × 10 |
6.106 | Lenburg Renal |
GIMAP4 | 26 | 5 | 3.300 | 1.40 × 10 |
9.996 | Yusenko Renal |
GIMAP5 | 10 | 10 | 2.119 | 6.05 × 10 |
5.432 | Gumz Renal |
GIMAP5 | 23 | 23 | 1.440 | 5.38 × 10 |
6.170 | Jones Renal |
GIMAP5 | 9 | 9 | 1.927 | 6.12 × 10 |
5.496 | Lenburg Renal |
GIMAP5 | 26 | 5 | 2.599 | 0.001* | 4.924 | Yusenko Renal |
GIMAP6 | 10 | 10 | 4.199 | 5.73 × 10 |
7.705 | Gumz Renal |
GIMAP6 | 23 | 23 | 2.325 | 6.32 × 10 |
6.307 | Jones Renal |
GIMAP6 | 9 | 9 | 2.374 | 0.002* | 3.667 | Lenburg Renal |
GIMAP6 | 26 | 5 | 3.321 | 3.07 × 10 |
8.988 | Yusenko Renal |
GIMAP7 | 9 | 9 | 1.814 | 0.019* | 2.314 | Lenburg Renal |
GIMAP7 | 26 | 5 | 3.364 | 1.63 × 10 |
5.671 | Yusenko Renal |
GIMAP8 | 9 | 9 | –1.037 | 0.758 | –0.719 | Lenburg Renal |
GIMAP8 | 26 | 5 | 1.033 | 0.469 | 0.081 | Yusenko Renal |
* p
The GIMAPs mRNA expression pattern was verified in TCGA. CcRCC tumor tissues
showed higher mRNA expression of GIMAP1 compared with normal renal tissues
(p
The expression of GIMAPs mRNA in ccRCC tumor and normal renal
tissues. (A–G) Compared to normal renal tissues, the ccRCC tumor had higher
mRNA expression of GIMAPs. (H) ROC curves showed high sensitivity and specificity
of these GIMAPs in distinction between normal renal tissues and ccRCC tumors. **
p
Gene symbol | Logistic regression analysis | ||
OR | 95% CI | p‑value | |
GIMAP1 | 0.0269 | 0.0036–0.1673 | |
GIMAP2 | 1.1541 | 0.4149–3.3329 | |
GIMAP4 | 4113.901 | 414.5919–70,187.91 | 0.787 |
GIMAP5 | 827.8691 | 80.2729–15,301.57 | |
GIMAP6 | 0.0024 | 0.0003–0.0136 | |
GIMAP7 | 4.1545 | 1.0478–19.7302 | 0.054 |
GIMAP8 | 0.0675 | 0.0158–0.2911 |
* p
We then studied the GIMAPs protein expression of ccRCC in CPTAC. There was no
total protein but only phosphoprotein data about GIMAP5. The expression of GIMAP5
phosphoprotein in normal renal tissues and ccRCC was not significantly different
(Fig. 2D), but GIMAP1 (Fig. 2A), GIMAP2 (Fig. 2B), GIMAP4 (Fig. 2C), GIMAP6 (Fig. 2E), GIMAP7 (Fig. 2F), and GIMAP8 (Fig. 2G) total proteins were expressed at
lower levels in normal renal tissues than ccRCC (p
GIMAPs protein expression in normal renal tissues and ccRCC
tumors. (A–C,E–G) Most GIMAPs (GIMAP1, GIMAP2, GIMAP4, GIMAP6, GIMAP7, and
GIMAP8) had higher total protein expression in ccRCC tumors than in normal renal
tissues. (D) It had no significant difference between GIMAP5 phosphoprotein
expression of normal renal tissues and ccRCC tumors. *** p
First, we used STRING to detect the interaction between GIMAPs protein, and
found that all combined scores except with GIMAP2 were more than 0.6 (Fig. 3A,
Supplementary Table 2). Then, the relationship among GIMAP family
members in ccRCC was analyzed. All Spearman
correlation coefficients (SCCs) between GIMAP
members were more than 0.5 (p
Relationship between GIMAP family members. (A) Protein-protein interaction network diagram of GIMAP family members. (B) Heatmap of spearman correlation coefficients between GIMAP family members in ccRCC.
There were 322 cases in the training set and 205 cases in the test set (Table 3). Patients with higher GIMAP1 (Fig. 4A), GIMAP2 (Fig. 4B), GIMAP4 (Fig. 4C),
GIMAP5 (Fig. 4D), GIMAP6 (Fig. 4E), GIMAP7 (Fig. 4F), and GIMAP8 (Fig. 4G)
expression levels had longer OS than those with lower expression levels,
according to survival analysis (p
Clinical factor | Training set (n = 322) | Test set (n = 205) | Overall (n = 527) | |
Survival time (day) | 1354.97 |
1313.23 |
1338.74 | |
Status | ||||
Live | 219 | 136 | 355 | |
Dead | 103 | 69 | 172 | |
Gender | ||||
Female | 114 | 69 | 183 | |
Male | 208 | 136 | 344 | |
Age (year) | ||||
65 | 42 | 107 | ||
50–59 | 83 | 55 | 138 | |
60–69 | 96 | 54 | 150 | |
78 | 54 | 132 | ||
Tumor longest dimension (cm) | ||||
0–0.9 | 17 | 12 | 29 | |
1–1.9 | 167 | 106 | 273 | |
2–2.9 | 49 | 35 | 84 | |
21 | 15 | 36 | ||
Missing | 69 | 36 | 105 | |
Tumor grade | ||||
1 | 9 | 4 | 13 | |
2 | 133 | 93 | 226 | |
3 | 128 | 77 | 205 | |
4 | 48 | 27 | 75 | |
Missing | 4 | 4 | 8 | |
Tumor Stage | ||||
I | 162 | 101 | 263 | |
II | 35 | 22 | 57 | |
III | 74 | 48 | 122 | |
IV | 50 | 32 | 82 | |
Missing | 1 | 2 | 3 |
Kaplan Meier curves showed the impact of GIMAPs on OS of ccRCC patients. (A–G) Patients with higher expression levels of all GIMAPs had longer OS than those with lower expression levels. OS, overall survival.
Because of the close correlation between GIMAP family members, a risk model base
on GIMAPs was constructed by Lasso regression to comprehensively evaluate the
impact of GIMAPs on prognosis of patients with ccRCC. In the training set, when
the model lambda (
GIMAPs-based risk model for ccRCC patients’ prognosis. (A) The
coefficients of GIMAPs when the model lambda (
High risk score, old age, high tumor grade and stage were found to be
unfavorable for OS of ccRCC patients according to the Cox regression analysis
based on risk score and clinical parameters (p
Univariate analysis | Multivariate analysis | ||||||
Covariates | HR | 95% CI | p-value | HR | 95% CI | p-value | |
Gender | |||||||
(male vs. female) | 0.946 | 0.693–1.291 | 0.724 | ||||
Age (year) | |||||||
(51–59 vs. |
1.569 | 0.915–2.691 | 0.101 | 1.296 | 0.751–2.237 | 0.351 | |
(60–69 vs. |
1.863 | 1.107–3.135 | 0.019* | 1.282 | 0.752–2.184 | 0.361 | |
( |
3.076 | 1.862–5.083 | 2.569 | 1.543–4.277 | |||
Stage | |||||||
(II vs. I) | 1.221 | 0.657–2.267 | 0.528 | 1.192 | 0.638–2.227 | 0.583 | |
(III vs. I) | 2.611 | 1.734–3.931 | 1.877 | 1.220–2.887 | 0.004* | ||
(IV vs. I) | 6.467 | 4.412–9.478 | 4.471 | 2.856–6.999 | |||
Tumor longest dimension | 1.242 | 0.996–1.548 | 0.054 | ||||
Tumor grade | 2.317 | 1.890– 2.842 | 1.395 | 1.106–1.760 | 0.005* | ||
Risk score | 3.199 | 2.350–4.355 | 1.914 | 1.363–2.686 |
* p
The impact of risk scores on OS of ccRCC patients in TCGA. (A–G) In all clinical subgroups and total patients, the patients in Low risk groups had longer OS than patients in High risk groups. (H) Predictive nomogram base on risk score and clinical parameters for OS in ccRCC patients. (I) ROC curve respecting nomogram’s prognostic prediction’s specificity and sensitivity. (J) Calibration curve respecting the accuracy of nomogram for predicting overall survival (OS) at 1 year, 3 years and 5 years.
All GIMAPs’ expression had negative correlation with the purity of tumor
(p
Correlation between GIMAPs and immunocytes infiltration in
ccRCC. All GIMAPs mRNA expression were negative to the purity of tumor,
indicated GIMAPs were mainly expressed in tumor immune microenvironment. The
degree of immunocyte infiltration (B cell, CD4
The risk model based on GIMAPs was closely correlated with the TIME of ccRCC,
the risk score was highly associated with immune score (SSC = 0.23, p
Relationship between the immune microenvironment of ccRCC and
the risk score. (A–C) Risk scores’ relationship with stromal score, immune
score and ESTIMATE score. (D) Heatmap showed immunocytes infiltration in ccRCC of
Low and High risk groups. (E) The comparison of immunocytes infiltration in the
Low and High risk groups. (F–H) The comparison between the immune checkpoints
(CTLA4, PD1 and PDL1) expression in High and Low risk groups. * p
We used co-expression analysis to identify genes that correlated with GIMAPs in
ccRCC, and then performed GSEA of genes co-expressed with each GIMAP. The results
of co-expression analysis are shown in Supplementary Table 3. There were
total 93 KEGG pathways (p
Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways
correlated with GIMAPs in ccRCC. (A) Heatmap of pathways correlated with GIMAP
family members in ccRCC. (B) The common pathways positively correlated with GIMAP
family members. NES, normalized enrichment score; FDR, false discovery rate. ***
p
Recent research showed that TIME was crucial to the occurrence and development of multiple tumors [38]. CcRCC is considered as an immunogenic tumor characterized of abundant immunocytes infiltration, in particular T cell infiltration [39]. Different types and functional states of immunocytes have different effects on ccRCC [40]. Therefore, identification of key factors affecting TIME is beneficial to predict the prognosis and explore new treatments for ccRCC patients. In this study, we found that GIMAPs were over-expressed in TIME of ccRCC, and they had close correlation with clinical outcomes and infiltration of immunocytes in this tumor.
The GIMAP family proteins are associated with immunity, which all have binding domains for GDP/GTP. GIMAPs can regulate biological functions and the states of a variety of immunocytes [17, 18, 19, 20, 21, 22]. GIMAPs are closely related to the autoimmune regulation of diabetes, allergy and asthma [41]. The dysregulation of GIMAPs is not only related to immune related diseases, but also various tumors. The blood and tumor tissues of hepatocellular carcinoma patients showed down-regulated levels of GIMAP6 and GIMAP5 expression, suggesting GIMAP6 and GIMAP5 possibly participated in the pathogenic mechanism of hepatocellular carcinoma [24]. It has been identified that the majority of GIMAPs (GIMAP1, GIMAP4, GIAMP6, GIMAP7 and GIMPA8) were down-regulated in endometrial cancer, and that low GIMAPs expression were associated with a poor prognosis and closely linked to immunocytes infiltration [25]. Studies indicated that GIMAPs showed low expression in lung adenocarcinoma tissues, and low GIMAPs expression were closely related to poor clinical outcomes [23, 26, 42]. Similarly, our study showed that the prognosis for patients with low GIMAP expression was poor, and in different clinical subgroups, the patients in the Low risk groups had longer OS than those in High risk groups, suggesting that the GIMAPs-based risk model has a good capacity for predicting the prognosis of patients with ccRCC. GIMAP family members were negatively correlated with tumor purity of lung adenocarcinoma [26]. In our research, GIMAP family members had the similar correlation with tumor purity of ccRCC, indicating that GIMAPs were mainly located in TIME, and that the risk score had a positive relationship with the immune score and a negative relationship with the stromal score. We found that GIMAPs’ high expression reflected the characteristics of high immunocytes infiltration in ccRCC. However, different from other tumors, GIMAPs’ expression was notably up-regulated in ccRCC. High immunocytes infiltration in ccRCC dose not always means favorable prognosis for patients. The types and states of infiltrating immunocytes are key factors for prognosis of ccRCC patients [43]. For this reason, we analyzed the biologic function of GIMAPs in TIME.
CcRCC is infiltrated by abundant
immunocytes, mainly DCs, macrophages, NK cells, CD4
NK cells are another important antitumor immunocytes in TIME. Studies showed
that IL-2 inhibited development of ccRCC by increasing proliferation and
cytotoxicity of NK cells [60]. A study indicated that a high proportion of NK
cells in TIME had correlated with favorable prognosis of ccRCC patients [61]. As
important immunosuppressive cells, Tregs can suppress the cytotoxicity of NK
cells by releasing TGF-
Immunotherapy has become an important method for the treatment of ccRCC [62, 63], but only small proportion of patients benefit from it [64]. Studies have
shown that there are a large number of attenuated and functional defective
CD8
Metabolic reprogramming is an important signature of ccRCC, including increased
aerobic Glycolysis, decreased mitochondrial Oxidative phosphorylation, increased
Fatty acid metabolism and so on [68, 69, 70]. It can provide sufficient energy and
substances for tumor growth, meanwhile, it is beneficial for tumors to adapt to
hypoxic environments, resist oxidative stress, and evade host immune surveillance
[71]. Study showed that ccRCC patients with high levels of glycolytic enzymes had
lower progression free survival and cancer specific survival than patients with
low levels of glycolytic enzymes [72]. Inhibition of aerobic glycolysis by 2-DG
could reduce the proliferation and activity of low-grade ccRCC, and promotion of
fatty acid oxidation by Etomoxir could inhibit the proliferation and activity of
high-grade ccRCC [73]. Studying tumor metabolism reprogramming are important to
find new strategies for the diagnosis and treatment of ccRCC. Inhibition of
NDUFA4L2 could reduce the vitality of ccRCC cells, increase mitochondrial mass,
and induce ROS production during hypoxia [74]. Depletion of MUC1 could inhibit
the migration and proliferation of ccRCC cells [75]. A study about lung cancer
showed that serum metabolomic fingerprints could serve as a “collective”
biomarker for predicting immune checkpoint inhibitor responses, capable of
predicting individual treatment outcomes with an accuracy of
In this study, we comprehensively explored the role of GIMAPs in ccRCC. Previous studies and our own researches all showed that GIMAPs were important to regulate TIME. As an immunogenic tumor, ccRCC was significantly affected by GIMAPs, which could inhibit the development of tumor by increasing the amount and antitumor activity of infiltrating immunocytes.
There are some limitations in this research. All data is derived from databases, and the total samples sizes are limited. More samples are needed to verify our results, and further experiments are essential to reaffirm the role of GIAMPs in ccRCC.
Our research indicated that GIMAPs were over-expressed in ccRCC, and that the GIMAPs’ expression was closely related to the prognosis of ccRCC patients. In addition, GIMAPs were highly correlated with the immunocytes infiltration in TIME. We suggest that GIMAPs are potential prognostic biomarkers and therapeutic targets of ccRCC.
All data produced and detailed in this article is accessible from open databases. On reasonable request, the corresponding author will provide additional information.
XML and MZ conceived and designed the project. MJZ and XZ collected the data. JH and XML performed the interpretation of data. MZ and MJZ performed the statistical analysis. MJZ and MZ wrote the manuscript. XML, JH and MZ revised the article. All authors read and approved the final manuscript.
Not applicable.
We appreciate the generosity of the researchers from ONCOMINE, TCGA and CPTAC for sharing the huge amount of data.
This research received no external funding.
The authors declare no conflict of interest.
Publisher’s Note: IMR Press stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.