Identification of a 6-gene signature predicting prognosis for colorectal cancer

Abstract

BackgroundAn accurate and robust gene signature is of the utmost importance in assisting oncologists to make a more accurate evaluation in clinical practice. In our study, we extracted key mRNAs significantly related to colorectal cancer (CRC) prognosis and we constructed an expression-based gene signature to predict CRC patients’ survival.MethodsmRNA expression profiles and clinicopathological data of colon adenocarcinoma (COAD) cases and rectum adenocarcinoma (READ) were collected from The Cancer Genome Atlas database to investigate gene expression alteration associated to the prognosis of CRC. Differentially expressed mRNAs (DEMs) were detected between COAD/READ and normal tissue samples. Relying on a univariate and multivariate Cox regression analyses, a mRNA panel signature was established and used for predicting the overall survival (OS) in CRC patients. Receiver operating characteristic curve was used to evaluate the prognosis performance of our model through calculating the AUC values corresponding to the 3-year and 5-year survival. To assess the performance of gene signature in the given cancer subgroups (CRC entire cohort, COAD cohort, and READ cohort), a stratified analysis was carried out according to clinical factors.ResultsA total of 5341 and 5594 DEMs were collected from COAD vs. normal tissue samples, and READ vs. normal samples respectively. A univariate regression analysis for the common DEMs between COAD and READ cohorts resulted in 14 common mRNAs related to OS. The multivariate Cox regression analysis revealed that 6 of these mRNAs (EPHA6, TIMP1, IRX6, ART5, HIST3H2BB, and FOXD1) had significant prognostic value allowing the discrimination between high- and low-risk patients, implying poor and good outcomes, respectively. The stratified analysis identified 6-gene signature as an independent prognostic signature in predicting CRC patients’ survival.ConclusionsThe 6-gene signature could act as an independent biomarker for survival prediction of CRC patients.Electronic supplementary materialThe online version of this article (10.1186/s12935-018-0724-7) contains supplementary material, which is available to authorized users.

Network Properties of Cancer Prognostic Gene Signatures in the Human Protein Interactome

“…Although the number of genes in prognostic signatures had a decreasing trend as a whole [50,51], most of the previous prognostic signatures were often in the form of a union set of dozens or hundreds of genes, or even a network module [6,[52][53][54]. This would considerably weaken the importance of individual genes, which may be one of the reasons why the four network centralities of the single PG were not high.…”

OncoNetExplainer: Explainable Predictions of Cancer Types Based on Gene Expression Data

“…MAI; they are shown in table III.To further validate our findings, the saturation analyses of cancer genes across 33 tumor types (except for COAD) are obtained from the TumorPortal [26]. Validation for the COAD cancer follows a signature-based approach [3], which was used for predicting the progression of colorectal cancer. However, our approach makes some false identifications, as 21 out of 25 genes are validated to be correct, making only 4 false identifications.…”

“…Since intensities did not follow any regular pattern, we chose top 660 genes across 33 tumor types (top-20 genes per class) as more significant based on the measure of MAI. Since we have more than 20K protein-coding genes, our choice of 660 is still a reasonable choice, since the number of important biomarkers should be small whose GE changes are sensitive to cancer growth [3]. All genes in the top-20 list can thus be viewed as tumor-specific biomarkers, which contribute most toward making the predictions.…”

Identification and verification of 3 key genes associated with survival and prognosis of patients with colon adenocarcinoma via integrated bioinformatics analysis

“…Moreover, knockdown of TIMP1 leads to decreased proliferation of HT29 cells [39]. Huang [40], Zeng [41], and Zuo [42] et al demonstrated that the expression levels of TIMP1 may be negatively correlated with the prognosis of COAD patients. Song et al [43] found that TIMP1 knockdown could result in decreased proliferation, migration, and invasion as well as increased apoptosis in vitro, whereas there is weakened tumorigenesis and metastasis in vivo via the FAK-PI3K/Akt and MAPK pathways.…”