Transcriptomic profiling reveals an enhancer RNA signature for recurrence prediction in colorectal cancer

doi:10.21203/rs.3.rs-1735174/v1

Download PDF

Article

Transcriptomic profiling reveals an enhancer RNA signature for recurrence prediction in colorectal cancer

https://doi.org/10.21203/rs.3.rs-1735174/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Colorectal cancer (CRC) is one of the most fatal malignancies worldwide, and this is in part due to high rates of tumor recurrence in these patients. Currently, TNM staging remains the gold standard for predicting prognosis and recurrence in CRC patients; however, this approach is inadequate for the identification of high-risk patients who have the highest likelihood of disease recurrence. Recent evidence has revealed that enhancer RNAs (eRNAs) represent a higher level of cellular regulation and their expression is frequently dysregulated in several cancers including CRC. However, the clinical significance of eRNAs as recurrence predictor biomarkers in CRC remains unexplored, which was the primary aim of this study.

Results

We performed a systematic analysis of eRNA expression profiles in colon cancer (CC) and rectal cancer (RC) patients, from the TCGA dataset. By using rigorous biomarker discovery approaches, by splitting the entire dataset into a training and the testing cohort, we identified a 22-eRNA panel in CC, and a 19-eRNA panel in RC for predicting tumor recurrence. The Kaplan-Meier analysis showed that biomarker panels robustly stratified low and high-risk CC (P = 7.29e-05) and RC (P = 6.81e-03) patients with recurrence. Multivariate and LASSO Cox regression models indicated that both biomarker panels were independent predictors of recurrence, and significantly superior to TNM staging in CC (HR = 11.89, P = 9.54e-04) and RC (HR = 3.91, P = 3.52e-02). Importantly, the ROC curves demonstrated that both panels exhibited excellent recurrence prediction accuracy in CC (AUC = 0.833; 95% CI: 0.74–0.93) and RC (AUC = 0.834; 95% CI: 0.72–0.92) patients. Subsequently, a combination signature which included the eRNA panels along with TNM staging achieved an even greater predictive accuracy in patients with CC (AUC = 0.85).

Conclusion

Herein, we report a novel eRNA signature for predicting recurrence in patients with CRC. Further experimental validation in independent clinical cohorts, these biomarkers have the potential to improve current risk stratification approaches for guiding precision oncology treatments in patients suffering from this lethal malignancy.

colorectal cancer

enhancer RNA

recurrence prediction biomarker

transcriptome expression profiling

LASSO Cox regression

Despite advances in diagnosis and treatment of colorectal cancer (CRC), the disease still accounts for approximately 10% of the global cancer cases [1]. According to the American Cancer Society, an estimated of 52,580 deaths are expected from CRC in the United States in 2022 [2]. CRC is a highly heterogeneous disease [3]. Depending on the anatomical location where the disease initiates, it can be classified into colon cancer (CC) or rectal cancer (RC); however, both disease types are often grouped together because they share similar molecular characteristics [4]. The five-year survival rate for CRC patients is approximately 65% for all stages [2]. However, only an estimated 38% of patients are diagnosed with localized disease for which the five-year survival rate can be more than 90% [2]. But when looked anatomically, the five-year survival rate for all stages in CC is 63% and for RC is 67% [2]. In the current clinical practice, clinicopathological features such as patient age, gender, primary site location, tumor grade and tumor-node-metastasis (TNM) staging are used for predicting prognosis and recurrence in CRC patients [5]. However, these conventional clinical practices alone are not precise, which results in inaccurate patient stratification [6, 7]. Several protein-coding genes have been developed as molecular markers for CRC prognosis [8–11] yet the survival rate is unsatisfactory [12]. Therefore, there is an imperative need to discover novel recurrence-associated biomarkers that may contribute to improved risk stratification and identify high-risk CRC patients for appropriate clinical decision-making and treatments.

With completion of the Human Genome Project in 2003, it is evident that protein-coding genes account for merely 1–2% of the human genome [13], while an astounding proportion of the genome serves as noncoding component, which was previously characterized as the dark matter of the genome. Noncoding RNAs are rich assortment of RNA molecules spanning from short RNAs (including miRNAs, piRNAs, snoRNAs) to long RNAs (including lncRNAs and circular RNAs) [14]. Accumulating evidence suggests that these noncoding RNAs play a regulatory role in the biological processes and are involved in cancer development and progression [14, 15]. Recently, enhancer RNAs (eRNAs) that are between 50-2000 nucleotide long, have been discovered as one class of the noncoding RNAs that are bidirectionally transcribed from the enhancer regions [16]. Several studies have reported dysregulation of eRNAs in cancers, which leads to the altered expression of oncogenes and tumor suppressor genes [17, 18]. Likewise, eRNAs regulate active enhancers, promote gene expression and affect cell-specific transcriptional regulation [16, 19]. Expression level of eRNAs within the enhancer regions positively correlates with expression level of nearby protein-coding genes [16], suggesting that eRNAs can be considered as genome wide markers for active transcriptional regulation. In view of the emerging interest for eRNAs as a promising category of biomarkers, herein we developed an unbiased, systematic bioinformatics approach to identify a novel and robust eRNA panel that can predict tumor recurrence in patients with stage II and III colorectal cancer.

Building survival model to identify clinically relevant eRNAs

We developed an eRNA based survival model to identify clinically relevant eRNA for CC and RC through the following steps (Figure 1).

We first randomly selected 70% of the entire cohort from the TCGA as the training set and reserved the remaining 30% of the cohort as testing set. To avoid deviation affecting stability of the model we maintained the distribution of disease-free and relapsed patients from the entire cohort in both training and testing sets.
Multivariate Cox regression analysis was carried out in the training set on 477 eRNAs for CC and on 460 eRNAs for RC, along with controlling the effects from other clinical risk factors including patient age, gender and TNM stage.
eRNAs significantly associated (P < 0.05 & z-score > 1.96) for predicting patient DFS were retained and termed as prognostic eRNAs and were further selected by LASSO Cox regularization with 10-folds cross validation.
Following LASSO Cox regularization and eRNA selection, a risk score formula was established. The risk score for each patient was calculated by a linear combination of expression and multivariate Cox coefficient of eRNAs.
Patients were classified into low-risk or high-risk groups using the median risk score as the cut-off threshold from the training set. The coefficient for each eRNA and the cut-off value of the risk score from the training set was used to calculate the risk score and to stratify patients into individual risk groups in the testing set.
Survival difference between the two groups were estimated using Kaplan-Meier curve and compared using log-rank test.

The above-mentioned steps 1-6 were repeated for 100 times to obtain 100 different eRNA sets. Only those sets that significantly stratified patients into two groups (high-risk groups exhibited poor survival than low-risk groups) in the testing were termed as successful survival model, else failed model. The results obtained from one such trial is shown in Additional file 1: Figure S1. For CC, we applied the model to the entire cohort (n = 379), however, the model was generalized, and we found very few prognostic eRNAs signature which successfully stratified patients into two different risk groups within the testing cohort. Due to CC heterogeneity, we performed the identification of prognostic eRNAs signature within CC subtypes based on Consensus Molecular Subtypes (CMS) classification. The entire CC cohort (n = 379) was classified into 5 subtypes with CMS1 comprising of 13%, CMS2 23%, CMS3 14%, CMS4 29%, and mixed subtype 21% (Figure 2A). The proportion of patients classified into CMS subtypes were approximately similar as described in Guinney et al [20] study. We further checked the five-year DFS probability of each CMS subtypes with CMS1: 89%; CMS2: 64%; CMS3: 81%; CMS4: 40%; and mixed subtype: 60% as shown in Figure 2B. As CMS4 displayed the worst survival compared to other CMS subtypes, we focused our study to identify a clinically relevant eRNA signature for this subtype. We found 22 prognostic eRNAs for CC-CMS4 subtype and 19 prognostic eRNAs for RC that significantly distinguished patients into two risk groups in testing set (Figure 2, C and D).

Prognostic association of eRNA signature risk score with disease-free survival in patients

A risk score was constructed using the median Cox-coefficients from the successful survival models. For the CC-CMS4 subtype, compared with patients in the low-risk score group, those with high-risk scores exhibited a poor DFS (Figure 3A). Only 17% of CC patients in the high-risk group were disease-free at 5 years, compared with 93% of the patients in the low-risk score group. We further randomly divided the entire cohort into 50% and using the same risk scoring formula and cut-off thresholds from the entire cohort, wherein the 22 eRNA signature still significantly stratified patients into two risk groups for DFS (Figure 3B). For RC, 54% of the patients in the high-risk score group were disease-free at 5 years, compared with 84% of the patients in the low- risk groups (Figure 3C). Similarly, upon random splitting the RC cohort into 50% and using the same cut-off threshold, the 19 eRNA signature significantly stratified the patients into two risk groups (Figure 3D)

Recurrence prediction by eRNA signature is independent of clinical risk factors

To evaluate the prognostic independence of eRNA signature against the known clinical risk factors, multivariate Cox regression analysis revealed that the 22 eRNA signature was an independent predictor of DFS in CC patients (HR= 11.89, P = 9.54e-04; Figure 4A). Additionally, we found that patients with relapsed tumors manifested with significantly higher risk scores than disease free patients (P= 5.72e-07; Figure 4B). Similarly, the 19 eRNA signature was independently associated with DFS in RC (HR=3.91, P = 3.52e-02; Figure 4C), and their expression was significantly higher in relapsed patient (P = 2.61e-06; Figure 4D).

The eRNA signature is a better predictor of recurrence with high sensitivity and specificity

To confirm the prediction accuracy for DFS, we performed ROC curve analysis for the eRNA signature. In both CC and RC, we observed eRNA signature exhibited excellent prediction power for relapse. For CC, the AUC and 95% confidence interval (CI) for the 22-eRNA signature was 0.83 and 0.739-0.928 (Figure 5A), and for RC the 19-eRNA signature achieved a remarkable prediction accuracy (AUC=0.834, 95% CI= 0.737-0.931; Figure 5B). Subsequently, when eRNA signature was combined with the TNM stage, it achieved an even higher prediction accuracy in CC (AUC=0.847) suggesting its potential clinical utility in this malignancy. However, no increment in prediction accuracy was observed in RC when combined with TNM stage.

Higher expression of eRNAs associated with tumor recurrence compared to its target genes

We performed the annotation of each prognostic eRNA (as suggested by Zhang et al [21]) and examined the expression profiles of eRNAs and their neighboring target genes, which were highly positively correlated. For example, AADAC-associated eRNA (AADACe, ENSR00000160285) was highly correlated with the AADAC gene in CC (SCC = 0.43, P= 1.03e-05) as shown in Figure 6A. Similarly, TRMT6-associated eRNA (TRMT6e, ENSR00000106418) was highly correlated with the expression of the TRMT6 gene in RC (SCC = 0.38, P= 1.53e-05) as shown in Figure 6B. Several studies have reported that eRNA expression has a better survival prediction potential compared to its target genes. In support of the previous studies, we also observed that high levels of AADACe was associated with worse survival (log-rank test P = 0.02058, Figure 6C). Interestingly, AADAC gene itself was not associated with CC patient survival, suggesting AADACe clinical utility regardless of AADAC expression in CC patients. Similar results were observed for TRMT6-associated eRNA (TRMT6e) in RC as shown in Figure 6D.

Putative biological functions of eRNA signature in colorectal cancer

The eRNAs have no protein-coding capacity; thus, we applied guilt-by-association strategy to explore the putative potential biological functions of the eRNA signature in CRC as shown in Figure 7A. We found biological functions related to metabolic process, DNA methylation, transcription initiation, cell cycle, myeloid leukocyte mediated immunity for CC. Similarly, biological functions related to regulation of DNA replication, ncRNA metabolic process, mRNA processing, oxidative phosphorylation were enriched for RC. The highest enriched biological functions for TATDN2-associated eRNA (TATDN2e, ENSR00000148476) in CC was histone H3K4 methylation and CDH17-associated eRNA (CDH17e, ENSR00000227152) in RC was regulation of mRNA splicing by spliceosomes (Figure 7, B and C)

The human eRNAs are a specific class of regulatory noncoding RNAs transcribed from the enhancer genomic regions and act in cis-direction to influence the transcription of the neighboring protein-coding genes. To the best of our knowledge, our study is the first to report the analysis and discovery of an RNA-seq based prognostic eRNA signature in CRC. Cox proportional hazard regression is widely used to decipher the association of molecular marker expression profiles with disease outcomes. Using the eRNA expression profile for CC and RC cohort from the TCGA database, we identified a 22 and 19 eRNA panel for predicting prognosis in CC and RC, respectively. Using a machine learning approach for the random selection of eRNAs from the training set and further validation of the survival model in the testing set with 100 iterations increased the robustness of the identified prognostic eRNAs. Risk scoring formula was applied to integrate prognostic eRNA expression into a signature. Patients were stratified into low-risk score group and high-risk score group based on the median risk score of the entire cohort. Kaplan-Meier survival analysis with Mantel log-rank test evaluated the association of eRNA signature with patient DFS. We validated the prognostic association of the signature by randomly selecting 50% of the cohort and using the median risk score of the entire cohort to stratify the patient into two risk score groups. Multivariable Cox analysis was used to estimate the prognostic independence of the eRNA signature against the established clinical risk factors in CRC.

The human CC is a highly heterogeneous disease, therefore we identified consensus molecular subtypes using gene sets for consensus molecular subtype classification. As, CMS4 subtype was associated with worst prognosis we chose to develop prognostic signature for this subtype. In this study, we also incorporated stage 1 patients, as we hypothesized that these patients in CMS4 subtypes might have higher risk of relapse than in other subtypes. We found that among 477 eRNAs from CC and 460 eRNAs from RC only 412 eRNAs (78%) were in common suggesting similarity in the eRNA landscape in the two-cancer type, however we found only 1 prognostic eRNA ENSR00000021608 similar between the two cancer types (Additional file 2: Figure S2) suggesting prognostic specificity of eRNA in their respective cohort.

We would like to acknowledge potential limitations of our study. First, the CC and RC cohorts included in our study were only from the TCGA. Therefore, this study used random selection of prognostic eRNA signature from the training set and further validated successful stratification of the cohort into two risk score groups in the testing set. Second, mutational status of KRAS and BRAF genes, CIMP and MSI status were not included as covariate in the multivariate Cox analysis as information for these clinical risk factors were not available for these patient cohorts. Third, in our study we could not significantly stratify the patients into two risk score groups for overall survival because only few deaths were observed (death vs censored in CC; 14/82 and in RC; 5/116). Despite these drawbacks, significant stratification of patient into two risk score groups in the testing sets in various iterations and further validation in the entire cohort with sophisticated statistical analysis offers a high level of confidence in the overall analyses.

To conclude, we developed a signature consisting of 22 eRNAs in CC and 19 eRNAs in RC, that can predict tumor relapse in these two malignancies. In addition, the expression level of eRNA signature is sensitive and profiled using RNA-seq platform. The expression of eRNA signature possesses a superior prediction power for recurrence compared to other clinical risk factors. The eRNA signature was independent in predicting patient tumor relapse. Thus, our study provides novel evidence for recurrence-associated eRNAs for experimental validation in clinical cohorts and deeper investigations to better understand the CRC etiology and progression, while providing clinical application for precise risk stratification to identify those group of patients who are in higher risk for CRC recurrence.

CRC Patients Datasets

The expression profiles of eRNAs for CC and RC were downloaded from the eRNA data portal in cancer (eRic) database [21]. Here, RNA-seq expression profiles from The Cancer Genome Atlas (TCGA) database were mapped to eRNA regions, followed by calculating their expression levels as reads per million (RPM) for each patient. For CC and RC, we retrieved 477 and 460 eRNA transcripts. The datasets were log2 (RPM +1) normalized. The clinical information for CRC patients were downloaded from the cBioPortal database [22, 23]. Only CRC patients with TNM stage 1, 2 and 3 were included. This study included 96 CC patients with CMS4 subtype and 121 RC patients for which both RNA-seq and survival information were available.

Consensus Molecular Subtypes (CMS) classification

The CMS classification for each patient with CC was performed using gene sets with Synapse ID syn4983432 (5973 genes available) downloaded from the Sage-Bionetworks (https://sagebionetworks.org/research-projects/colorectal-cancer-subtyping-consortium-crcsc/). The R package CMS classifier was used for CMS subtype calling using random forest classifier function with default posterior probability of 0.5 [20]. In this method, based on similarity of gene expression profiles, all CC patients were classified into four subtypes CMS1 – CMS4. All CC patients with mixed features or intra-tumoral heterogeneity were classified as mixed subtypes.

Identification of prognostic eRNAs associated with tumor relapse

We randomized 70% of the entire cohort as training set and the rest 30% as testing set. The distribution of disease-free and relapse-free patient population in both training and testing sets were comparable. Since, clinical information including patient age, gender and tumor stage can affect patient survival, we accordingly examined the prognostic potential of each eRNA using multivariate Cox regression analysis in the training set. We used zscore >1.96 (which corresponds to P < 0.05) as a cut-off threshold to select prognostic eRNAs associated with disease-free survival (DFS). We used least absolute shrinkage and selection operator (LASSO) Cox regression with 10-fold cross validation to identify subset of prognostic eRNAs and established an eRNA signature. The prognostic potential of eRNA signature was thereafter evaluated in the testing set. We repeated the above split cohort approach into training set and a testing set for 100 times, which generated 100 different sets of eRNA signatures.

Risk Score Calculation

During each iteration, risk score for each patient was calculated using a linear combination of expression and coefficients from multivariate Cox regression analysis as follows:

where e_j is the multivariate coefficient of eRNA j, Exp_ij is the expression value of eRNA j in patient i, and n is the number of prognostic eRNAs (selected after lasso Cox analysis). To calculate risk score of the prognostic eRNAs selected from all the successful survival models (which significantly stratify the patients into two groups in testing set), we used median Cox coefficients obtained from lasso Cox analysis.

Statistical Analysis

Kaplan-Meier survival analysis (low-risk score groups vs high-risk score groups) with Mantel log-rank test was performed to access differences between the two survival probabilities. Multivariable Cox proportional hazard regression was performed to determine the prognostic independence of risk scores and clinical risk factors. Receiver operating characteristic (ROC) curve and area under the curve (AUC) analyses were performed to evaluate the sensitivity and specificity of eRNA risk scores for predicting DFS. Survival analysis was performed using the survival R package [24]. LASSO cox proportional hazard regression analysis was performed using glmnet R package [25, 26]. ROC curve and AUC analysis were performed using pROC R package [27] and plots were generated using ggplot2 R package [28].

Gene Set Enrichment Analysis

Spearman correlation coefficient (SCC) was calculated between prognostic eRNAs and whole genome protein-coding genes (PCGs) in CC and RC patients, respectively. To assess the putative biological functions of each eRNA, gene set enrichment analysis (GSEA) was performed using MSigDB (C5.bp.v7.1.symbols.gmt gene set collection; 7530 gene set available), ranked list of eRNAs correlated genes and their SCC, maximum gene set size of 5000, minimum gene set size of 15, weighted enrichment statistics, and 1000 permutations. GSEA analysis was performed using GSEA pre-ranked tool [29]. Over-represented gene sets (FDR q-value 0.05 and overlap coefficient of 0.5) were filtered and visualized using Enrichment map-a Cytoscape plug-in [30].

CRC

Colorectal cancer

eRNA

Enhancer RNA

Colon cancer

Rectal cancer

LASSO

Least Absolute Shrinkage and Selection Operator

AUC

Area under curve

CMS

Consensus molecular subtype

DFS

Disease free survival

ROC

Receiver operating characteristics

CIMP

CpG island methylator phenotype

MSI

Microsatellite instability.

Ethics approval and consent to participate

Not applicable

Consent for publication

Not applicable

Availability of data and materials

The eRNA expression dataset used in this study can be found in the eric database https://hanlab.uth.edu/eRic/. The clinical information for CC and RC can be downloaded from the cBioPortal database https://www.cbioportal.org.

Competing interests

The authors declare no conflict of interest.

Funding: This work was supported by, CA184792, CA187956, CA227602, CA072851 and CA202797 grants from the National Cancer Institute, National Institutes of Health to AG.

Authors’ contributions

D.S. and A.G. conceived and designed the study. D.S. developed the computational methodology and carried out the analysis. D.S., CC.L. and A.G. helped in data analysis and interpreted the result. D.S. drafted the manuscript. D.S., CC.L. and A.G. revised the manuscript. All authors read and approved the final manuscript.

Acknowledgements

We thank Hsuan-Cheng Huang in the Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taipei, Taiwan for helpful discussions.

Author detail

¹Department of Molecular Diagnostics and Experimental Therapeutics, Beckman Research Institute of City of Hope Comprehensive Cancer Center, Duarte, CA, USA

²Institute of Biomedical Informatics, National Yang-Ming University, Taipei, Taiwan

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F: Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians 2021, 71(3):209–249.
American Cancer S: Cancer Facts & Figs. 2022. American Cancer Society 2022.
Blank A, Roberts DEI, Dawson H, Zlobec I, Lugli A: Tumor heterogeneity in primary colorectal cancer and corresponding metastases. Does the apple fall far from the tree? Frontiers in medicine 2018:234.
Hong TS, Clark JW, Haigis KM: Cancers of the colon and rectum: identical or fraternal twins? In., vol. 2: AACR; 2012: 117–121.
Marks K, West N, Morris E, Quirke P: Clinicopathological, genomic and immunological factors in colorectal cancer prognosis. Journal of British Surgery 2018, 105(2):e99-e109.
Dienstmann R, Mason MJ, Sinicrope FA, Phipps AI, Tejpar S, Nesbakken A, Danielsen SA, Sveen A, Buchanan DD, Clendenning M et al: Prediction of overall survival in stage II and III colon cancer beyond TNM system: a retrospective, pooled biomarker study. Ann Oncol 2017, 28(5):1023–1031.
Moccia F, Tolone S, Allaria A, Napolitano V, Ilaria F, Ottavia M, Cesaro E, Docimo L, Fei L: Lymph node ratio versus TNM system as prognostic factor in colorectal cancer staging. A single Center experience. Open Medicine 2019, 14(1):523–531.
Oh SC, Park Y-Y, Park ES, Lim JY, Kim SM, Kim S-B, Kim J, Kim SC, Chu I-S, Smith JJ: Prognostic gene expression signature associated with two molecularly distinct subtypes of colorectal cancer. Gut 2012, 61(9):1291–1298.
Kandimalla R, Ozawa T, Gao F, Wang X, Goel A, Nozawa H, Hata K, Nagata H, Okada S, Watanabe T: Gene expression signature in surgical tissues and endoscopic biopsies identifies high-risk T1 colorectal cancers. Gastroenterology 2019, 156(8):2338–2341. e2333.
Ye S-B, Cheng Y-K, Hu J-C, Gao F, Lan P: Development and validation of an individualized gene expression-based signature to predict overall survival in metastatic colorectal cancer. Annals of Translational Medicine 2020, 8(4).
Yuan Y, Chen J, Wang J, Xu M, Zhang Y, Sun P, Liang L: Development and clinical validation of a novel 4-gene prognostic signature predicting survival in colorectal cancer. Frontiers in oncology 2020, 10:595.
Brouwer NP, Bos AC, Lemmens VE, Tanis PJ, Hugen N, Nagtegaal ID, de Wilt JH, Verhoeven RH: An overview of 25 years of incidence, treatment and outcome of colorectal cancer patients. International journal of cancer 2018, 143(11):2758–2766.
International Human Genome Sequencing C: Finishing the euchromatic sequence of the human genome. Nature 2004, 431(7011):931–945.
Mattick JS, Makunin IV: Non-coding RNA. Human Molecular Genetics 2006, 15(suppl_1):R17-R29.
Esteller M: Non-coding RNAs in human disease. Nature reviews genetics 2011, 12(12):861–874.
Kim T-K, Hemberg M, Gray JM, Costa AM, Bear DM, Wu J, Harmin DA, Laptewicz M, Barbara-Haley K, Kuersten S: Widespread transcription at neuronal activity-regulated enhancers. Nature 2010, 465(7295):182–187.
Gu X, Wang L, Boldrup L, Coates PJ, Fahraeus R, Sgaramella N, Wilms T, Nylander K: AP001056. 1, a prognosis-related enhancer RNA in squamous cell carcinoma of the head and neck. Cancers 2019, 11(3):347.
Lee J-H, Xiong F, Li W: Enhancer RNAs in cancer: regulation, mechanisms and therapeutic potential. RNA biology 2020, 17(11):1550–1559.
Sartorelli V, Lauberth SM: Enhancer RNAs are an important regulatory layer of the epigenome. Nature structural & molecular biology 2020, 27(6):521–528.
Guinney J, Dienstmann R, Wang X, De Reynies A, Schlicker A, Soneson C, Marisa L, Roepman P, Nyamundanda G, Angelino P: The consensus molecular subtypes of colorectal cancer. Nature medicine 2015, 21(11):1350–1356.
Zhang Z, Lee J-H, Ruan H, Ye Y, Krakowiak J, Hu Q, Xiang Y, Gong J, Zhou B, Wang L: Transcriptional landscape and clinical utility of enhancer RNAs for eRNA-targeted therapy in cancer. Nature communications 2019, 10(1):1–12.
Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E: The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. In., vol. 2: AACR; 2012: 401–404.
Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E: Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Science signaling 2013, 6(269):pl1-pl1.
Therneau T: A package for survival analysis in S. R package version 2015, 2(7).
Friedman J, Hastie T, Tibshirani R: Regularization paths for generalized linear models via coordinate descent. Journal of statistical software 2010, 33(1):1.
Friedman J, Hastie T, Tibshirani R: Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw 2010, 33(1):1–22.
Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, Müller M: pROC: an open-source package for R and S + to analyze and compare ROC curves. BMC bioinformatics 2011, 12(1):1–8.
Hadley W: ggplot2: Elegant Graphics for Data Analysis. 2016.
Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES: Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences 2005, 102(43):15545–15550.
Merico D, Isserlin R, Stueker O, Emili A, Bader GD: Enrichment map: a network-based method for gene-set enrichment visualization and interpretation. PloS one 2010, 5(11):e13984.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Transcriptomic profiling reveals an enhancer RNA signature for recurrence prediction in colorectal cancer

Status:

Version 1

Abstract

Background

Results

Conclusion

Figures

Introduction

Results

Discussion

Conclusions

Methods

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1