Abstract
Background: Experimental evidence has suggested that transient receptor potential (TRP) channels play a crucial role in tumor biology. However, clinical relevance and significance of TRP channels in cancer remain largely unknown. Materials and Methods: We applied a data-driven approach to dissect the expression landscape of 27 TRP channel genes in 14 types of human cancer using International Cancer Genome Consortium data. Results: TRPM2 was found overexpressed in most tumors, whereas TRPM3 was broadly down-regulated. TRPV4 and TRPA1 were found up- and down-regulated respectively in a cancer type-specific manner. TRPC4 was found to be closely associated with incidence of head and neck cancer and poor survival of patients with kidney cancer. TRPM8 was identified as a new molecular marker for lung cancer diagnosis and TRPP1 for kidney cancer prognosis. Conclusion: Our data-driven approach demonstrates that the variation in the expression of TRP channel genes is manifested across various human cancer types and genes, for certain TRP channels have strong predictive diagnostic and prognostic potential.
Transient receptor potential (TRP) channels generate electrochemical signals in terms of membrane potential or intracellular Ca2+ in response to various internal and external stimuli (1, 2). In human, the TRP channel superfamily consists of 27 isotypes that are classified into six subfamilies (3): canonical (TRPC), vanilloid (TRPV), melastatin (TRPM), polycystin (TRPP), mucolipin (TRPML), and ankyrin (TRPA). Emerging evidence has shown that the aberrant functions of TRP channels are closely associated with cancer hallmarks, such as sustaining proliferative signaling, evading growth suppressors, resisting cell death, and activating invasion and metastasis (4, 5). In addition, the TRP channel network suggests that TRP channels are involved in tumor biology by interacting with oncogenes or tumor suppressors (6-8). However, the clinical relevance and significance of TRP channels in patients with cancer has not been investigated.
Recently, an alliance of computational biology with high-throughput technologies has provided useful frameworks for linking biological information to clinical significance. In particular, integration and analysis of a large volume of heterogeneous biological and clinical datasets in silico has expanded our epistemic scope of biomedical knowledge (9-11). With advances in genomic technologies, such as next-generation sequencing and bioinformatics, data-driven approaches have been reforming the way in which we understand tumor biology, discover tumor-associated genes, and develop anticancer therapeutic strategies (12). Consequently, data-driven cancer research can deliver the promise of early diagnosis and medical treatments of patients with cancer (12). Therefore, data-driven approaches may be useful to ascertain clinical relevance and significance of TRP channel in cancer.
In the present study, we investigated the clinical significance of 27 TRP channels in 14 human cancer types using the International Cancer Genome Consortium (ICGC) dataset. Our study provides a novel conceptual framework for translating biological knowledge on TRP channels into clinical practice.
Materials and Methods
Data selection. The normalized gene-expression data of all cancer types were downloaded in data repository (ftp site) from the ICGC data portal (https://dcc.icgc.org). The downloaded ICGC data includes gene-expression data from 42 projects (Data release 15.1, February 11th, 2014). Of 42 projects, 28 projects were filtered out: 16 projects did not have gene-expression data and 12 projects did not include normal samples. Finally, 14 projects containing matched tumor and normal samples (552 pairs) were chosen to analyze gene expression data (Table I). Table II shows the clinical information of these 552 patients. Because the gene-expression data from each project use different normalization methods, the expression levels of TRP channels were determined by the ratio of the normalized gene-expression levels between normal and tumor samples.
Statistical analysis. All statistical analyses were performed using program R 3.1.2 (https://www.r-project.org/). To calculate odds ratios (ORs), the best threshold values were chosen by calculating F1 score based on receiver operating characteristic (ROC) analysis for all combinations of cancer types and TRP channel genes. ORs and their 95% confidence interval (CI) were estimated using logistic regression. Using those threshold values, the expression values of each TRP channel gene were classified into high and low expression groups. Univariate analysis was then applied to calculate p-values, ORs, and CI between high- and low-expression groups. Finally, multivariate logistic regression was applied for significant TRP channel genes whose p-values of univariate analysis were less than 0.01 and the area under the ROC curve (AUC) values of univariate analysis were greater than 0.8. The criteria of p-value and AUC were empirically chosen.
Survival analysis to evaluate the discriminatory power and the predictive accuracy of TRP channel gene expression were applied to only one project, such as kidney cancer [clear cell carcinoma (CCC)] among 14 projects because the CCC kidney cancer type included cancer survival data (13). The non-parametric Kaplan–Meier method was used to determine survival curves and the log-rank test was used to determine overall survival rates. Using median gene expression values as bifurcating point, the samples were divided into high- and low-expression groups and the survival rates of groups were compared. Cox proportional hazards model was applied to estimate hazard ratios (HRs) and 95% CIs. Harrell's concordance index (c-index), widely used as a surrogate for the ROC analysis (14), was calculated on the basis of HR and 95% CI.
Network visualization. The open-source program, Gephi 0.8.2-beta (http://gephi.github.io/) was used to visualize the relation between genes and cancer types.
Results
Variation in the expression of TRP channel genes in human cancer. To gain deeper insight into the roles of TRP channels in tumor biology, we dissected the expression landscape of 27 TRP channel genes in 14 human cancer types (Table I). Ubiquity and specificity of the altered expression of TRP channels was found throughout cancer types (Figure 1 and Table III). TRPM2, TRPM3, and TRPM6 are a typical example of the ubiquity of altered expression: TRPM2 was found up-regulated in most cancer types (by 1.47- to 7.56-fold), whereas TRPM3 and TRPM6 were broadly down-regulated (by 0.13- to 0.56-fold), suggesting the isotype-specific functions of TRP channels in cancer biology.
The International Cancer Genome Consortium gene expression data used in this study. Among 4,854 tumor samples, 552 matched normal samples were used in this study.
The altered expression of certain TRP channels was specific for cancer types (Figure 1 and Table III). Interestingly, in some cases, an opposing expression pattern was observed according to cancer type: TRPV4 was found to be overexpressed in cervical cancer (18.65-fold), whereas its expression was reduced in liver cancer (0.21-fold); TRPA1 was found to be up-regulated in kidney cancer (by 11.94- to 28.74-fold), whereas its expression was reduced in prostate cancer (0.15-fold). These results suggest that TRP channels have opposing roles depending on the cancer type. However, we found TRPV1 not to be significantly changed in different cancer types.
The association between TRP channel expression and cancer incidence or clinical outcome. We then questioned the clinical relevance and significance of TRP channels in human cancer. To identify whether the altered expression of TRP channels are associated with cancer incidence, we performed univariate and multivariate logistic regression analysis. Our results are summarized in Table IV. TRP channels significantly affect the risk of cancer incidence. We found higher expression of TRPM2 to be closely associated with an increased risk for four cancer types, namely bladder, head and neck, liver, and lung cancer (adenocarcinoma) (OR=14.260-389.563). In contrast, the higher expression of TRPM3 was found to be associated with a decreased risk for bladder, breast, and thyroid cancer (OR=0.062-0.102). Interestingly, higher expression of TRPC6 was associated with reduced risk for breast, colon and prostate cancer (OR=0.572, 0.012 and 0.153, respectively) but an elevated risk for head and neck cancer (OR=1.922).
Altered expression of transient receptor potential (TRP) channels in human cancer. A: Heat map representing the median values of expression of TRP channel genes (ratio of cancer to normal). N: Not available. B: Network visualizing the association between TRP channel isotypes (gray nodes) and cancer types (pale red nodes). Line colors represent up (red)- or down (blue)-regulation of TRP channel genes and line width indicates their expression levels in cancer. CCC, Clear cell carcinoma; PCC, papillary cell carcinoma; AC, adenocarcinoma; SCC, squamous cell carcinoma; TRPC, transient receptor potential channels canonical; TRPV, transient receptor potential channels vanilloid; TRPM, transient receptor potential channels melastatin; TRPML, transient receptor potential channels mucolipin; TRPP, transient receptor potential channels polycystin; TRPA, transient receptor potential channels ankyrin.
Clinical information of 552 tumor-normal matched samples in the International Cancer Genome Consortium data.
Alterations in the gene expression of transient receptor potential (TRP) channels in cancer. The data are expressed as median values (cancer-to-normal ratios).
To assess the effect of the altered expression of TRP channels on clinical outcomes, we performed univariate and multivariate survival analyses for clear cell kidney cancer (survival data are available only for this cancer type) (Figure 2). We divided the patients based on the expression levels of each TRP channel gene (i.e. high- and low-expression groups). Kaplan–Meier analysis indicated that the patients with CCC kidney cancer with low expression of TRPC4, TRPM3, TRPP1, and TRPA1 had significantly worse overall survival and higher risk of death than those with high expression (HR=3.754, 3.000, 3.355, and 2.649, respectively; log-rank test p=0.0068, 0.0229, 0.0147, and 0.0437, respectively).
Feasibility of TRP channels as diagnostic and prognostic markers. We also performed ROC analysis to assess the feasibility of TRP channels as diagnostic markers. Our results are summarized in Table V. TRP channels have a strong diagnostic potential for various cancer types, particularly in head and neck, kidney, and lung cancer, in which clinically useful diagnostic markers are not available: overexpression of TRPC4, TRPM2, and TRPM8 might be used as diagnostic markers, in terms of sensitivity and specificity, for cancer of the head and neck cancer, kidney (clear cell carcinoma and papillary cell carcinoma), and lung (adenocarcinoma and squamous cell carcinoma), respectively.
We calculated Harrell's concordance index (c-index) to evaluate the usefulness of TRP channels as prognostic markers. The c-index is defined as the proportion of all patient pairs in which the predictions and outcomes are concordant (15). TRPC4, TRPM3, TRPP1, and TRPA1 for kidney (CCC) cancer had c-indices of 0.636, 0.614, 0.643, and 0.598, respectively (Table VI). When four TRP channels were combined, the c-indices were elevated to 0.710. Therefore, genes for each of these TRP channels or their combination could be used as promising prognostic markers for patients with kidney cancer.
Discussion
Accumulating experimental evidence has suggested that TRP channels play crucial roles in tumor biology (16-20). However, the clinical relevance and significance of TRP channels in cancer is poorly understood. In the present study, we applied a data-driven approach to dissecting the expression landscape of 27 TRP channels in 14 human cancer types and to assessing clinical relevance and significance of TRP channels. We found distinct features of variation in the expression of genes for TRP channels according to cancer type. We also show that TRP channels are clinically valuable for cancer diagnosis and prognosis. Our study provides a novel conceptual framework for unraveling the role of TRP channels in cancer biology and clinical oncology.
Survival curve for patients with kidney cancer (clear cell carcinoma) based on the expression levels of transient receptor potential channels canonical 4 (TRPC4; A), TRPM3 (B), transient receptor potential channels polycystin 1 (TRPP1; C), and transient receptor potential channels ankyri 1 (TRPA1; D). HR, Hazard ratio (95% confidence interval).
Our study provides insight into understanding of the role of TRP channels in carcinogenesis. Normal cells evolve into cancer cells through many genetic and epigenetic changes (21, 22). During such somatic evolution processes, many cancerous cells are removed by various host mechanisms and microenvironmental selection. The expression patterns of TRP channel genes in cancer suggest that cancer type-specific TRP channel-mediated Ca2+ remodeling mechanisms may play a crucial role in tumor cell survival under the pressure of microenvironmental selection. Changes in TRP channel expression may confer selective growth and survival advantages over internal or external threats to cancerous cells. Our data-driven study will assist future investigations to enlight the molecular mechanisms of TRP channels in tumor evolutionary processes and to develop feasible tests for cancer diagnosis and prognosis. However, TRP signatures depend not only on the level of expression but also on the subcellular localization of the channels. Therefore, the location-specific expression of TRP channels needs to be investigated in future studies.
As far as we know, this study is the first data-driven approach in TRP channel research. We showed that our focused data-driven approach effectively links biological information to clinical and epidemiological knowledge. Our results demonstrate clinical relevance and significance of TRP channels in human cancer, supporting the previous experiment-driven findings that TRP channels play an important role in cancer development and progression (16, 17, 23, 24). These results imply that further accumulation of information-rich biological data will make substantial progress in answering biological and clinical questions on TRP channels. In addition, the data-driven approach will produce the integrated knowledge on TRP channels from biological and clinical data. Therefore, our efforts may facilitate a new way of future research on TRP channels for unraveling their roles in biology and disease.
Odd ratios (ORs), confidence interval (CI), and p-value between high- and low-expression groups for transient receptor potential (TRP) channel genes and cancer incidence. The table shows nine cancer types, each of which is associated with the overexpression or underexpression of at least one TRP channel at a p-value of multivariate logistic regression of less than 0.01.
Diagnostic accuracy of transient receptor potential (TRP) channels for detecting cancer. Threshold values were chosen by calculating F1 score based on receiver operating characteristic (ROC) analysis for all combinations of cancer types and TRP channel genes.
Prognostic value of transient receptor potential (TRP) channels in clear cell kidney cancer.
Acknowledgements
This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012R1A1A2002804 and 2014R1A2A1A11050616), and by a grant from the Seoul National University Hospital Research Fund (03-2013-0040). In addition, this work was supported by the Education and Research Encouragement Fund of Seoul National University Hospital.
Footnotes
↵* These Authors contributed equally to this study.
Conflicts of Interest
The Authors declare no conflicts of interest.
- Received September 21, 2015.
- Revision received October 22, 2015.
- Accepted October 23, 2015.
- Copyright © 2016 The Author(s). Published by the International Institute of Anticancer Research.
References
In this issue
Jump to section
Related Articles
Cited By...
- TRPM2 promotes pancreatic cancer by PKC/MAPK pathway
- Prognostic Significance of Transient Receptor Potential Vanilloid Type 1 (TRPV1) and Phosphatase and Tension Homolog (PTEN) in Epithelial Ovarian Cancer
- Characterization and Optimization of the Novel Transient Receptor Potential Melastatin 2 Antagonist tatM2NX
- New analysis framework incorporating mixed mutual information and scalable Bayesian networks for multimodal high dimensional genomic and epigenomic cancer data
- Depletion of the Human Ion Channel TRPM2 in Neuroblastoma Demonstrates Its Key Role in Cell Survival through Modulation of Mitochondrial Reactive Oxygen Species and Bioenergetics