Skip to main content

Prognosis prediction model based on competing endogenous RNAs for recurrence of colon adenocarcinoma

Abstract

Background

Colon adenocarcinoma (COAD) patients who develop recurrence have poor prognosis. Our study aimed to establish effective prognosis prediction model based on competing endogenous RNAs (ceRNAs) for recurrence of COAD.

Methods

COAD expression profilings downloaded from The Cancer Genome Atlas (TCGA) were used as training dataset, and expression profilings of GSE29623 retrieved from Gene Expression Omnibus (GEO) were set as validation dataset. Differentially expressed RNAs (DERs) between non-recurrent and recurrent specimens in training dataset were screened, and optimum prognostic signature DERs were revealed to establish prognostic score (PS) model. Kaplan-Meier survival analysis was conducted for PS model, and GEO dataset was used for validation. Prognosis prediction efficiencies were evaluated by area under curve (AUC) and C-index. Meanwhile, ceRNA regulatory network was constructed by using signature mRNAs, lncRNAs and miRNAs.

Results

We identified 562 DERs including 42 lncRNAs, 36 miRNAs, and 484 mRNAs. PS prediction model, consisting of 17 optimum prognostic signature DERs, showed that high risk group had significantly poorer prognosis (5-year AUC = 0.951, C-index = 0.788), which also validated in GSE29623. Prognosis prediction model incorporating multi-RNAs with pathologic distant metastasis (M) and pathologic primary tumor (T) (5-year AUC = 0.969, C-index = 0.812) had better efficiency than clinical prognosis prediction model (5-year AUC = 0.712, C-index = 0.680). In the constructed ceRNA regulatory network, lncRNA NCBP2-AS1 could interact with hsa-miR-34c and hsa-miR-363, and lncRNA LINC00115 could interact with hsa-miR-363 and hsa-miR-4709. SIX4, GRAP, NKAIN4, MMAA, and ERVMER34–1 are regulated by hsa-miR-4709.

Conclusion

Prognosis prediction model incorporating multi-RNAs with pathologic M and pathologic T may have great value in COAD prognosis prediction.

Peer Review reports

Background

Colon cancer is the first leading cancer type in digestive system, and approximately 101,420 new cases of colon cancer will be diagnosed in United States in 2019 with the estimated deaths of 51,020 [1]. Colon adenocarcinoma (COAD) counts for more than 80% of all colon cancers, and the remaining small proportion of colon cancers are related to sarcomas and squamous cell carcinomas [2, 3]. Retrospectively study conducted by Cass et al. has revealed that 37% patients developed local recurrence and distant metastases after complete primary resection, and local recurrence without clinical evidence of distant metastases was the most common cause of death within 5 years [4]. Even though survival of patients with colon cancer has been improved along with the developing of new chemotherapeutics and advanced techniques, the prognosis of patients developed recurrence still remains poor and early recurrence have the possibility to be cured by salvage surgery [5, 6]. Thus, it is urgently to find the prognostic markers and construct effective prognosis prediction model for the recurrence of COAD.

It is widely acknowledged that competing endogenous RNAs (ceRNAs) could interact with mRNAs by competing with miRNAs, and miRNA-mediated interactions between long non-coding RNAs (lncRNAs) and mRNAs exist in the progressing of various diseases [7,8,9]. The lncRNA H19 has been reported to function as a ceRNA for miR-138 and miR-200a to suppress the target gene expressions of Vimentin, ZEB1, and ZEB2, thereafter to promote epithelial to mesenchymal transition in colorectal cancer [10]. Moreover, lncRNA H19 could also up-regulate cancer-related mRNA expressions by functioning as a ceRNA to influence phosphatidylinositol-3-kinase (PI3K)/Akt signaling pathway, and lncRNA H19 is associated with poor prognosis in colorectal cancer [11]. The lncRNA ATB could inhibit the expression of E-cadherin, thus contributing to the colon cancer progression via epithelial to mesenchymal transition, and higher levels of lncRNA ATB are related to poor prognosis [12]. Recently, the ceRNA regulatory networks of lncRNAs, miRNAs and mRNAs have been also constructed in many studies to reveal the molecular mechanisms involved in initiation and progression of human colon adenocarcinoma [13,14,15]. However, there are few studies about ceRNAs based prediction models of COAD recurrence. In the study of Chen et al., they established ceRNA network for differentially expressed RNAs (DERs) between 15 non-recurrent and 98 recurrent COAD patients from The Cancer Genome Atlas (TCGA) database [16]. Meanwhile, Chen et al. identified the prognostic lncRNA markers by using multivariate regression and established a nomogram for recurrence prediction based on the identified lncRNA biomarkers and clinical covariates [16].

In order to further reveal the recurrence prediction markers, the expression profilings of 367 COAD samples (279 non-recurrent and 88 recurrent ones) downloaded from TCGA were used as training dataset, and expression profilings of 65 COAD samples from Gene Expression Omnibus (GEO) were considered as validation dataset. Optimum prognostic signature DERs were identified via the L1 penalized lasso estimation in addition to multivariate regression analysis. Prognosis prediction efficiency of the ceRNAs-based prognostic score prediction model was compared with that of clinical prognostic prediction model.

Materials and methods

Data source and preprocessing

The mRNA, lncRNA and miRNA sequencing data of COAD samples based on the platform of Illumina HiSeq 2000 RNA Sequencing were downloaded from TCGA (https://portal.gdc.cancer.gov/projects/TCGA-COAD). The mRNA along with lncRNA expression profilings (RNA-seq data) of 512 COAD samples and miRNA expression profilings (miRNA-seq data) of 461 COAD samples were downloaded from TCGA A total of 367 COAD samples that had both of mRNA expression profilings and miRNA expression profilings were obtained by mapping the clinical data for each sample. These expression profilings of the 367 COAD samples, including 279 non-recurrent and 88 recurrent specimens, were used as training dataset.

Meanwhile, RNA expression profilings of human colon adenocarcinoma were retrieved by searching the keywords of “colon adenocarcinoma” and “Homo sapiens” from National Center for Biotechnology Information (NCBI) GEO (https://www.ncbi.nlm.nih.gov/geo/) for validation. The datasets were retained when meet the following criteria: expression profilings of tumor tissue samples from COAD patients; containing lncRNA, mRNA and miRNA expression profilings; with corresponding available clinical information about recurrence and prognosis. Subsequently, GSE29623 [17] consisted of lncRNA, mRNA and miRNA expression profilings of 65 colon adenocarcinomas based on the platforms of GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array and GPL11162 NIH Taqman Human MicroRNA Array v.2 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29623) was used as validation dataset.

Identification of differentially expressed RNAs

The lncRNAs, mRNAs and miRNAs in training and validation datasets were annotated by using Human Genome Organisation (HUGO) Gene Nomenclature Committee (HGNC, http://www.genenames.org/) [18]. Afterwards, the differentially expressed RNAs between non-recurrent and recurrent specimens in training dataset were screened by Limma (Linear Models for Microarray Data) package (Version 3.34.7; https://bioconductor.org/packages/release/bioc/html/limma.html) [19] of R3.4.1. The cut-off criteria were set as false discovery rate (FDR) < 0.05 and |log2 fold change (FC) | > 0.5. Hierarchical clustering analysis based on centered Pearson correlation was performed for all the identified DERs via the pheatmap (Version 1.0.8, https://cran.r-project.org/web/packages/pheatmap/index.html) of R3.4.1 [20, 21].

Construction and validation of prognosis prediction model based on prognostic score

The DERs related to prognosis were firstly uncovered by using the univariate Cox regression analysis in Survival package (Version2.41–1, http://bioconductor.org/packages/survivalr/) [22]. Then, multivariate Cox regression analysis was conducted to screen the independent prognostic DERs with the threshold of log-rank p value < 0.05. Thereafter, the optimum prognostic signature DERs were identified via the L1 penalized lasso estimation based Cox-Proportional Hazards (Cox-PH) model in penalized package (Version0.9–50, http://bioconductor.org/packages/penalized/) of R3.4.1. When the maximal cross-validation likelihood run 1000 times, the optimal lambda was determined. The prognostic score was calculated based on the identified optimum prognostic signature DERs according to the following formula:

$$ \mathrm{Prognostic}\ \mathrm{score}\ \left(\mathrm{PS}\right)=\sum {\upbeta}_{\mathrm{DERs}}\times {\mathrm{Exp}}_{\mathrm{DERs}} $$

Where β DERs referred to the LASSO coefficients of signature DERs, and Exp DERs denoted the expression levels of signature DERs.

The PS value of each sample in training dataset was calculated, and samples were divided into high risk group (PS > median value) and low risk group (PS < median value) by using median value as the cut-off point. The Kaplan-Meier curves were plotted by survival package (Version2.41–1) of R3.4.1 [22]. At the same time, the PS value of each sample in validation dataset was also calculated using the expression levels, and samples were divided into high risk group and low risk group with median value as the cut-off point. The Kaplan-Meier curves for samples in validation dataset were plotted by survival package (Version2.41–1) of R3.4.1 [22].

Screening independent prognostic clinical factors and prognosis prediction efficiencies

The independent prognostic clinical factors for training dataset were screened by using the univariate and multivariate Cox regression analyses in Survival package (Version2.41–1, http://bioconductor.org/packages/survivalr/) [22] with the threshold of log-rank p value < 0.05. Afterwards, the clinical prognostic prediction model was constructed based on the screened independent prognostic clinical factors.

The prognosis prediction efficiencies of clinical prognostic prediction model and the PS prediction model were evaluated by pROC package (Version 1.14.0, https://cran.r-project.org/web/packages/pROC/index.html) of R3.4.1 [23] to calculate the area under ROC (AUC; AUC = 0.5: no discriminatory power, AUC = 1: perfect discriminatory power) and by survcomp (Version 1.34.0, http://www.bioconductor.org/packages/release/bioc/html/survcomp.html) of R3.4.1 [24, 25] to calculate the C-index (C-index > 0.7: acceptable discriminatory power; C-index > 0.8: excellent discriminatory power; C-index > 0.9: outstanding discriminatory power).

ceRNA regulatory network

The regulatory interactions between signature lncRNAs and signature miRNAs identified above were retrieved from the database of DIANA-LncBase V2 (http://carolina.imis.athena-innCOADation.gr/diana_tools/web/index.php?r=lncbasev 2%2Findex-experimental). Cytoscape (Version 3.6.1, https://cytoscape.org/) [26] was used to visualize the lncRNA-miRNA regulatory network.

The target genes of differentially expressed miRNAs were predicted based on the starBase (Version 2.0, http://starbase.sysu.edu.cn/) [27], which contains prediction data from five databases including targetScan, picTar, RNA22, PITA and miRanda. The regulatory interactions predicted in at least one of the five databases were remained. Signature mRNAs identified above were mapped into the predicted target genes to construct the signature miRNA-mRNA regulatory network, which was also visualized by Cytoscape (Version 3.6.1, https://cytoscape.org/) [26]. Subsequently, lncRNA-miRNA regulatory network was integrated with signature miRNA-mRNA regulatory network to construct the ceRNA regulatory network by Cytoscape (Version 3.6.1, https://cytoscape.org/) [26].

Results

Screened DERs between non-recurrent and recurrent COAD samples

After annotation, 13,834 mRNAs, 539 lncRNAs and 540 miRNAs were obtained in training and validation datasets. In total, 562 DERs were identified by Limma between 279 non-recurrent and 88 recurrent COAD samples in the TCGA training dataset, including 42 lncRNAs (20 up-regulated and 22 down-regulated ones), 484 mRNAs (393 up-regulated and 131 down-regulated ones), and 36 miRNAs (30 up-regulated and 6 down-regulated ones) (Fig. 1a, b). Meanwhile, the hierarchical clustering analysis results showed the expression levels of differentially expressed lncRNAs, miRNAs or mRNAs could well distinguish the non-recurrent COAD samples from recurrent COAD samples (Fig. 1b).

Fig. 1
figure1

Identification and hierarchical clustering analysis of differentially expressed RNAs (DERs) between 279 non-recurrent and 88 recurrent colon adenocarcinoma (COAD) samples in training dataset. A, the identified DERs with the thresholds of false discovery rate (FDR) < 0.05 and |log2 fold change (FC) | > 0.5. Blue dots represent DERs. B, the ratios of up-regulated and down-regulated lncRNAs, miRNAs and mRNAs. Pink column and blue column represent up-regulated and down-regulated DERs, respectively. C, the hierarchical clustering analysis of differentially expressed lncRNAs, miRNAs or mRNAs. Obvious color contrast was observed suggesting DERs could well distinguish the non-recurrent COAD samples from recurrent COAD samples

Prognostic score prediction model

There were 227 DERs (204 mRNAs, 11 lncRNAs and 12 miRNAs) related to prognosis according to the univariate Cox regression analysis in Survival package with the log-rank p value < 0.05. Furthermore, 48 independent prognostic DERs (4 lncRNAs, 6 miRNAs and 38 mRNAs) were identified via the multivariate Cox regression analysis in Survival package. The L1 penalized estimation based Cox-PH model in penalized package revealed 17 optimum prognostic signature DERs (6 miRNAs, 2 lncRNAs and 9 mRNAs), including 3 protective RNAs (hsa-mir-105-1, hsa-mir-3609, and MMAA) and 14 risky RNAs (LINC00115, NCBP2-AS1, hsa-mir-34c, hsa-mir-363, hsa-mir-4709, hsa-mir-891a, ALOX15, ERVMER34–1, GRAP, KCP, NKAIN4, PADI1, SIX4, and SLC16A8) based on the hazard ratio (HR) by multivariate Cox regression analysis (Table 1).

Table 1 The optimum prognostic signature DERs identified by L1 penalized estimation based Cox-Proportional Hazards (Cox-PH) model

The PS value of each sample in training dataset was calculated based on the LASSO coefficients (Fig. 2a) and expression levels of the 17 signature DERs. Samples were divided into high risk group and low risk group with the median value of 1.218 as the cut-off point (Fig. 2b). The survival time and recurrent status of patients in high risk group and low risk group showed that patients in low risk group had obviously longer survival time and lower proportion of recurrence (Fig. 2c). Meanwhile, expression levels of signature DERs in each sample revealed that the 3 protective RNAs were up-regulated in low risk group, while the other 14 risky RNAs were up-regulated in high risk group (Fig. 2d).

Fig. 2
figure2

The 17 optimum prognostic signature DERs. A, the lasso coefficient curves of the 17 signature DERs. B, calculated prognostic score (PS) of each sample in training dataset. C, prognosis of patients in training dataset between high risk group and low risk group. D, expression levels of 3 protective RNAs and 14 risky RNAs of samples in high risk group and low risk group

The Kaplan-Meier curves of training dataset showed high risk group and low risk group classified by PS prediction model had significantly different prognosis (p = 1.062e-10) (Fig. 3a). The AUC of 1-year, 3-year, and 5-year ROC of training dataset was 0.971, 0.949, and 0.951, respectively. With regarding to the validation dataset, high risk group also showed significantly lower survival ration in comparison with the low risk group (p = 3.491e-02) (Fig. 3b). Meanwhile, the AUC of 1-year, 3-year, and 5-year ROC of validation dataset was 0.841, 0.837, and 0.909, respectively.

Fig. 3
figure3

The Kaplan-Meier curves and receiver operating characteristic curves for training and validation datasets. A, the Kaplan-Meier curves of high risk group and low risk group and receiver operating characteristic (ROC) curves for training dataset. B, the Kaplan-Meier curves of high risk group and low risk group and receiver operating characteristic curves for validation dataset

Independent prognostic clinical factors and prognosis prediction efficiencies of different models

Three independent prognostic clinical factors, i.e. pathologic distant metastasis (M) (p = 2.65E-02), pathologic primary tumor (p = 2.23E-02), and PS model status (p = 1.29E-05), were found according to univariate and multivariate Cox regression analyses (Table 2). Moreover, the Kaplan-Meier curves of pathologic M and pathologic T showed low pathologic M or pathologic T stage had significantly higher survival ratio (Fig. 4a, b). The prognosis prediction model constructed by both of pathologic M and pathologic T had better prognosis prediction efficiency (5-year AUC = 0.712, C-index = 0.680) than prediction model of pathologic M or pathologic T alone. The prognosis prediction model constructed by mRNAs (C-index = 0.746), or multi-RNAs (C-index = 0.788) had a C-index value larger than 0.7, suggesting acceptable discriminatory power. Moreover, prognosis prediction model constructed by multi-RNAs combined pathologic M and pathologic T had a 5-year AUC of 0.969 and a C-index of 0.812, suggesting excellent discriminatory power (Fig. 4c).

Table 2 The independent prognostic clinical factors according to univariate and multivariate Cox regression analyses
Fig. 4
figure4

The Kaplan-Meier curves of pathologic distant metastasis (M) and pathologic primary tumor (T) in training dataset, and prediction efficiencies of different prognosis prediction models. A, Kaplan-Meier curves of pathologic M0 and M1. B, Kaplan-Meier curves of pathologic T1, T2, T3, and T4. C, ROC curves of different prognosis prediction models

ceRNA regulatory network

Based on the DIANA-LncBasev2 database, the regulatory interactions between signature lncRNAs and miRNAs were obtained. The lncRNA NCBP2-AS1 could interact with hsa-miR-34c and hsa-miR-363, and lncRNA LINC00115 could interact with hsa-miR-363 and hsa-miR-4709. The target genes of signature miRNAs were predicted by starBase and regulatory interactions involving signature mRNAs were remained. SIX4, GRAP, NKAIN4, MMAA, and ERVMER34–1 are regulated by hsa-miR-4709. All these regulatory interactions were used for ceRNA regulatory network construction. Three miRNAs (hsa-miR-34c, hsa-miR-363 and hsa-miR-4709), two lncRNAs (NCBP2-AS1 and LINC00115), and five mRNAs (SIX4, GRAP, NKAIN4, MMAA, and ERVMER34–1) were included in the ceRNA regulatory network (Fig. 5).

Fig. 5
figure5

The constructed competing endogenous RNA (ceRNA) regulatory network. Square, triangle and circle nodes represent lncRNAs, miRNAs, and mRNAs, respectively

Discussion

Colon adenocarcinoma is one of the most common causes of death in adults, causing a major public health problem over the world [1]. The current strategies for treating colon cancer are mainly radical surgery and chemotherapy, while the prognosis of colon cancer patients who develop distant metastasis and local recurrence after treatments would become worse [28, 29]. Therefore, effective novel biomarkers and reliable prognosis prediction model for COAD recurrence are of great help in selecting treatment strategies for patients according to the recurrence risk. In our study, 562 DERs (42 lncRNAs, 36 miRNAs, and 484 mRNAs) were identified between 279 non-recurrent and 88 recurrent COAD sample profilings downloaded from the TCGA. According to the univariate Cox regression analysis, multivariate Cox regression analysis, and L1 penalized estimation based Cox-PH model, 17 optimum prognostic signature DERs were revealed, including 3 protective RNAs (hsa-mir-105-1, hsa-mir-3609, and MMAA) and 14 risky RNAs (LINC00115, NCBP2-AS1, hsa-mir-34c, hsa-mir-363, hsa-mir-4709, hsa-mir-891a, ALOX15, ERVMER34–1, GRAP, KCP, NKAIN4, PADI1, SIX4, and SLC16A8), which were used to establish the PS prediction model.

Kaplan-Meier curves showed that high risk group and low risk group in TCGA dataset divided by the PS prediction model had significantly different prognosis with a 5-year AUC of 0.951, which was also validated in GEO validation dataset with the 5-year AUC of 0.909. The prognosis prediction model constructed by independent prognostic clinical factors, including pathologic M and pathologic T, had a prognosis prediction efficiency with 5-year AUC of 0.712 and C-index of 0.680. The PS prediction model constructed by 17 optimum prognostic signature DERs had a better prognosis prediction efficiency with a 5-year AUC of 0.951 and C-index of 0.788. Huang et al. have developed a radiomics nomogram for preoperative prediction of lymph node metastasis with a C-index of 0.736, and validated in another cohort with a C-index of 0.778, suggesting this radiomics nomogram may be helpful for preoperative prediction of lymph node metastasis in colorectal cancer [30]. Another prognostic model composed of six significant prognostic factors (age, first-degree relative cancer history, differentiation grade, vessels/nerves invasion, TNM stage and HALP) has a 5-year AUC of 0.73 for patients with locally advanced colorectal cancer [31]. In our study, prognosis prediction model constructed by incorporating multi-RNAs with pathologic M and pathologic T has a 5-year AUC of 0.969 and a C-index of 0.812, suggesting this prognosis prediction model may have great value in COAD.

In the constructed ceRNA regulatory network, lncRNA NCBP2-AS1 could interact with hsa-miR-34c and hsa-miR-363, and lncRNA LINC00115 could interact with hsa-miR-363 and hsa-miR-4709. SIX4, GRAP, NKAIN4, MMAA, and ERVMER34–1 were regulated by hsa-miR-4709. It has been reported that lncRNA LINC00115 might be a prognostic lncRNA in lung adenocarcinoma, and functions as a ceRNA by interacting with hsa-miR-7 to regulate FGF2 [32]. Zhang et al. have identified the differentially expressed lncRNAs related to cancer recurrence, and revealed that lncRNA LINC00115 is one of the markedly up-regulated lncRNAs in cancer associated with disease-free survival [33]. In the study of Li et al., hsa-miR-4709 was identified to be significantly up-regulated in colon cancer and high levels of hsa-miR-4709 was associated with poor prognosis [34]. As one member of the sine oculis homeobox (SIX) homolog family, e SIX4 have been found up-regulated in colorectal patients, which is significantly related to lymph node metastasis, advanced Tumor Node Metastasis (TNM) stages, and unfavorable prognosis [35]. Moreover, knocking down of SIX4 could suppress colorectal cell metastasis via in-activation of the PI3K/Akt signaling pathway. Additionally, SAM68 (Src-associated in mitosis, 68 kDa) has been reported overexpressed in colorectal cancer and higher expressions were related with poorer prognosis in colorectal cancer [36]. SAM68 could interact with GRAP to active oncogenic pathways, such as epidermal growth factor and PI3K/Akt signaling pathways, thereby contributing to cancer progressing [37, 38]. Thus, lncRNA LINC00115 identified as an optimum prognostic signature DERs in our study, may highly correlated with COAD recurrence via functioning as a ceRNA by interacting with hsa-miR-4709 to regulate expressions of SIX4, GRAP, NKAIN4, MMAA, and ERVMER34–1. Moreover, in order to be further used in clinical, the prediction efficacy of the constructed prognosis prediction model, which incorporating multi-RNAs with pathologic M and pathologic T, should be considered to be validated by using a larger number of clinical samples.

Conclusion

In conclusion, 562 DERs (42 lncRNAs, 36 miRNAs, and 484 mRNAs) were identified between 279 non-recurrent and 88 recurrent COAD samples downloaded from the TCGA. PS prediction model based on 17 optimum prognostic signature DERs was established, and validated in 65 COAD samples downloaded from GEO. Prognosis prediction model incorporating multi-RNAs with pathologic M and pathologic T has a 5-year AUC of 0.969 and a C-index of 0.812. The lncRNA LINC00115 may highly correlated with COAD recurrence via functioning as a ceRNA by interacting with hsa-miR-4709 to regulate expressions of SIX4, GRAP, NKAIN4, MMAA, and ERVMER34-1.

Availability of data and materials

The raw data were collected and analyzed by the Authors, and are not ready to share their data because the data have not been published.

References

  1. 1.

    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019;69(1):7–34.

    Article  Google Scholar 

  2. 2.

    Lavrsen K, Dabelsteen S, Vakhrushev SY, Levann AM, Haue AD, Dylander A, et al. De novo expression of human polypeptide N-acetylgalactosaminyltransferase 6 (GalNAc-T6) in colon adenocarcinoma inhibits the differentiation of colonic epithelium. J Biol Chem. 2018;293(4):1298–314.

    CAS  Article  Google Scholar 

  3. 3.

    Kriegsmann M, Longuespée R, Wandernoth P, Mohanu C, Lisenko K, Weichert W, et al. Typing of colon and lung adenocarcinoma by high throughput imaging mass spectrometry. Biochim Biophys Acta (BBA)-Proteins Proteomic. 2017;1865(7):858–64.

    CAS  Article  Google Scholar 

  4. 4.

    Cass AW, Million RR, Pfaff WW. Patterns of recurrence following surgery alone for adenocarcinoma of the colon and rectum. Cancer. 1976;37(6):2861–5.

    CAS  Article  Google Scholar 

  5. 5.

    Sargent D, Sobrero A, Grothey A, O'Connell MJ, Buyse M, Andre T, et al. Evidence for cure by adjuvant therapy in colon cancer: observations based on individual patient data from 20,898 patients on 18 randomized trials. J Clin Oncol. 2009;27(6):872.

    Article  Google Scholar 

  6. 6.

    Group QC. Adjuvant chemotherapy versus observation in patients with colorectal cancer: a randomised study. Lancet. 2007;370(9604):2020–9.

    Article  Google Scholar 

  7. 7.

    Salmena L, Poliseno L, Tay Y, Kats L, Pandolfi PP. A ceRNA hypothesis: the Rosetta stone of a hidden RNA language? Cell. 2011;146(3):353–8.

    CAS  Article  Google Scholar 

  8. 8.

    Huang M, Zhong Z, Lv M, Shu J, Tian Q, Chen J. Comprehensive analysis of differentially expressed profiles of lncRNAs and circRNAs with associated co-expression and ceRNA networks in bladder carcinoma. Oncotarget. 2016;7(30):47186.

    Article  Google Scholar 

  9. 9.

    Zhou M, Wang X, Shi H, Cheng L, Wang Z, Zhao H, et al. Characterization of long non-coding RNA-associated ceRNA network to reveal potential prognostic lncRNA biomarkers in human ovarian cancer. Oncotarget. 2016;7(11):12598.

    Article  Google Scholar 

  10. 10.

    Liang W-C, Fu W-M, Wong C-W, Wang Y, Wang W-M, Hu G-X, et al. The lncRNA H19 promotes epithelial to mesenchymal transition by functioning as miRNA sponges in colorectal cancer. Oncotarget. 2015;6(26):22513–25. https://doi.org/10.18632/oncotarget.4154.

    Article  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Zhong M-E, Chen Y, Zhang G, Xu L, Ge W, Wu B. LncRNA H19 regulates PI3K–Akt signal pathway by functioning as a ceRNA and predicts poor prognosis in colorectal cancer: integrative analysis of dysregulated ncRNA-associated ceRNA network. Cancer Cell Int. 2019;19(1):148. https://doi.org/10.1186/s12935-019-0866-2.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Yue B, Qiu S, Zhao S, Liu C, Zhang D, Yu F, et al. LncRNA-ATB mediated E-cadherin repression promotes the progression of colon cancer and predicts poor prognosis. J Gastroenterol Hepatol. 2016;31(3):595–603.

    CAS  Article  Google Scholar 

  13. 13.

    Zhang Z, Qian W, Wang S, Ji D, Wang Q, Li J, et al. Analysis of lncRNA-associated ceRNA network reveals potential lncRNA biomarkers in human colon adenocarcinoma. Cell Physiol Biochem. 2018;49(5):1778–91.

    CAS  Article  Google Scholar 

  14. 14.

    Qian W, Feng Y, Li J, Peng W, Gu Q, Zhang Z, et al. Construction of ceRNA networks reveals differences between distal and proximal colon cancers. Oncol Rep. 2019;41(5):3027–40.

    CAS  PubMed  Google Scholar 

  15. 15.

    Gao Z, Fu P, Yu Z, Zhen F, Gu Y. Comprehensive analysis of lncRNA–miRNA–mRNA network ascertains prognostic factors in patients with colon cancer. Technol Cancer Res Treat. 2019;18:1533033819853237.

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16.

    Chen F, Li Z, Deng C, Yan H. Integration analysis for novel lncRNA markers predicting tumor recurrence in human colon adenocarcinoma. J Transl Med. 2019;17(1):299. https://doi.org/10.1186/s12967-019-2049-2.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Chen DT, Hernandez JM, Shibata D, McCarthy SM, Humphries LA, Clark W, et al. Complementary strand microRNAs mediate acquisition of metastatic potential in colonic adenocarcinoma. J Gastrointest Surg. 2012;16(5):905–12.

    Article  Google Scholar 

  18. 18.

    Povey S, Lovering R, Bruford E, Wright M, Lush M, Wain H. The HUGO gene nomenclature committee (HGNC). Hum Genet. 2001;109(6):678–80.

    CAS  Article  Google Scholar 

  19. 19.

    Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.

    Article  Google Scholar 

  20. 20.

    Wang L, Cao C, Ma Q, Zeng Q, Wang H, Cheng Z, et al. RNA-seq analyses of multiple meristems of soybean: novel and alternative transcripts, evolutionary and functional implications. BMC Plant Biol. 2014;14(1):169.

    Article  Google Scholar 

  21. 21.

    Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci. 1998;95(25):14863–8.

    CAS  Article  Google Scholar 

  22. 22.

    Wang P, Wang Y, Hang B, Zou X, Mao J-H. A novel gene expression-based prognostic scoring system to predict survival in gastric cancer. Oncotarget. 2016;7(34):55343.

    Article  Google Scholar 

  23. 23.

    Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinform. 2011;12(1):77.

    Article  Google Scholar 

  24. 24.

    Mayr A, Schmid M. Boosting the concordance index for survival data–a unified framework to derive and evaluate biomarker combinations. PLoS One. 2014;9(1):e84483.

    Article  Google Scholar 

  25. 25.

    Shan S, Chen W, Jia J. Transcriptome analysis revealed a highly connected gene module associated with cirrhosis to hepatocellular carcinoma development. Front Genet. 2019;10:305.

    CAS  Article  Google Scholar 

  26. 26.

    Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504.

    CAS  Article  Google Scholar 

  27. 27.

    Jun-Hao L, Shun L, Hui Z, Liang-Hu Q, Jian-Hua Y. starBase v2.0: decoding miRNA-ceRNA, miRNA-ncRNA and protein-RNA interaction networks from large-scale CLIP-Seq data. Nucleic Acids Res. 2014;42(D1):D92.

    Article  Google Scholar 

  28. 28.

    Yifan P, Zhiwei Z, Zhongmin L, Lin W, Jin G. Role of blood tumor markers in predicting metastasis and local recurrence after curative resection of colon cancer. Int J Clin Exp Med. 2015;8(1):982–90.

    Google Scholar 

  29. 29.

    O’Connell MJ, Campbell ME, Goldberg RM, Grothey A, Seitz J-F, Benedetti JK, et al. Survival following recurrence in stage II and III colon cancer: findings from the ACCENT data set. J Clin Oncol. 2008;26(14):2336–41.

    Article  Google Scholar 

  30. 30.

    Y-q H, Liang C-h, He L, Tian J, Liang C-s, Chen X, et al. Development and validation of a radiomics nomogram for preoperative prediction of lymph node metastasis in colorectal cancer. J Clin Oncol. 2016;34(18):2157.

    Article  Google Scholar 

  31. 31.

    Jiang H, Li H, Li A, Tang E, Xu D, Chen Y, et al. Preoperative combined hemoglobin, albumin, lymphocyte and platelet levels predict survival in patients with locally advanced colorectal cancer. Oncotarget. 2016;7(44):72076–83.

    Article  Google Scholar 

  32. 32.

    Li DS, Ainiwaer JL, Sheyhiding I, Zhang Z, Zhang LW. Identification of key long non-coding RNAs as competing endogenous RNAs for miRNA-mRNA in lung adenocarcinoma. Eur Rev Med Pharmacol Sci. 2016;20(11):2285–95.

    PubMed  Google Scholar 

  33. 33.

    Zhang X, Ho TT. Computational analysis of lncRNA function in cancer; 2019.

    Google Scholar 

  34. 34.

    Li F, Li Q, Wu X. Construction and analysis for differentially expressed long non-coding RNAs and MicroRNAs mediated competing endogenous RNA network in colon cancer. PLoS One. 2018;13(2):e0192494.

    Article  Google Scholar 

  35. 35.

    Li G, Hu F, Luo X, Hu J, Feng Y. SIX4 promotes metastasis via activation of the PI3K-AKT pathway in colorectal cancer. Peerj. 2017;5(3):e3394.

    Article  Google Scholar 

  36. 36.

    Shi L, Zhao X-H, Chen X-T, Song L-B, Liao W-T. High expression level and nuclear localization of Sam68 are associated with progression and poor prognosis in colorectal cancer. BMC Gastroenterol. 2013;13(1):126.

    Article  Google Scholar 

  37. 37.

    Martin-Romero C, Sanchez-Margalet V. Human leptin activates PI3K and MAPK pathways in human peripheral blood mononuclear cells: possible role of Sam68. Cell Immunol. 2001;212(2):83–91.

    CAS  Article  Google Scholar 

  38. 38.

    Locatelli A, Lange CA. Met receptors induce Sam68-dependent cell migration by activation of alternate extracellular signal-regulated kinase family members. J Biol Chem. 2011;286(24):21062–72.

    CAS  Article  Google Scholar 

Download references

Acknowledgements

none.

Funding

none.

Author information

Affiliations

Authors

Contributions

LPJ and TL participated in the design of this study, and they both performed the statistical analysis. FQM carried out the study and collected important background information. JDT drafted the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Jian Dong Tai.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no conflict of interest.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Jin, L.P., Liu, T., Meng, F.Q. et al. Prognosis prediction model based on competing endogenous RNAs for recurrence of colon adenocarcinoma. BMC Cancer 20, 968 (2020). https://doi.org/10.1186/s12885-020-07163-y

Download citation

Keywords

  • Colon adenocarcinoma
  • Recurrence
  • Prognosis prediction model
  • ceRNAs