ABSTRACT
Background: Colorectal cancer (CRC) is a malignant tumor with high morbidity and mortality, but there is still no recognized prognostic prediction model to better predict and intervene its prognosis. Our aim is to establish a novel microRNA (miRNA) signature and identify hub target genes for simply and accurately predicting survival risk for CRC patients and to provide therapeutic targets. Methods: The miRNA expression profiles along with clinical data of 512 CRC patients were downloaded from the Cancer Genome Atlas (TCGA) database and randomly divided into training set and validation set. The signature was generated from the training set after a series of Cox regression analyses, including least absolute shrinkage and selectionator operator (LASSO)-Cox regression, and verified in the test set and the whole set. Furthermore, the signature was compared with clinical risk factors. Interaction network of target genes of the seven micoRNAs was established. Functional enrichment analysis was performed to reveal the biological processes and pathways. GEPIA2 was used for prognostic analysis. Results: A 7-micoRNA prognostic signature was generated from the training set with the areas under the receiver operating characteristic (ROC) curve (AUC) of 5-year survival rate was 0.889. Its performance was well verified both in the test set and the entire set by Kaplan-Meier analysis (P value <0.05). Further analysis demonstrated that the signature was an independent prognostic risk factor for CRC patients and its predictive ability was superior to age and tumor-node-metastasis (TNM) stage. Interaction network found two major gene modules, and they might be involved in the activation of PI3K-Akt-mTOR and p53 signaling pathways, which related to epidermal growth factor receptor (EGFR) resistance. The GEPIA2 revealed that CDKN1A, eIF4E and SNAI1 were associated with CRC prognosis. Conclusions: Our study demonstrated the potential of this novel 7-micoRNA signature to independently predict overall survival in patients with CRC and provided potential therapeutic targets.
ABSTRACT
Although prostate biopsy is the gold standard for the diagnosis of prostate cancer, it also leads to high incidence of negative biopsies and the diagnosis of clinically low-risk prostate cancer and the subsequent overtreatment. It remains an unmet need to discover new biomarkers in order to defer the unnecessary biopsies in clinical practice. In this study, we described a new method, LBXexo score, to measure the urine exosomal PCA3/PRAC expression from non-DRE urine as a noninvasive diagnosis to improve the detection rate in Chinese population with a low serum PSA level. First-voided urine samples were collected to isolate exosomes, and exosomal RNAs of PCA3 and PRAC were measured by quantitative reverse transcription PCR. A significant increase in exoPCA3/PRAC was observed in both any-grade and high-grade prostate cancer groups when compared with the biopsy-negative group. Receiver-operating characteristic curve analyses showed that the LBXexo score significantly improved diagnostic performance in predicting biopsy results, with AUCs of 0.723 (p=0.017) and 0.736 (p=0.038) for any-grade and high-grade (GS ≥ 7) prostate cancer, respectively. For high-grade cancer, LBXexo had the negative and positive predictive values of 100% and 27.59%, respectively, and could potentially avoid unnecessary biopsy. This is the first report in Chinese population that demonstrates the predictive value of the exosomal expression of PCA3 and PRAC derived from non-DRE urine in predicting prostate biopsy outcomes. It could be used in clinical practice to make a better informed biopsy decision and avoid unnecessary biopsies in Chinese population.
