Autophagy-Related Long Non-coding RNA Is a Prognostic Indicator for Bladder Cancer

Abstract

Bladder cancer (BC) is one of the most common malignant urinary system tumors, and its prognosis is poor. In recent years, autophagy has been closely linked to the development of BC. Therefore, we investigated the potential prognostic role of autophagy-related long non-coding RNA (lncRNA) in patients with BC. We obtained the lncRNA information and autophagy genes, respectively, from The Cancer Genome Atlas (TCGA) data set and the human autophagy database (HADb) and performed a co-expression analysis to identify autophagy gene-associated lncRNAs. Then, we divided the data into training group and testing group. In the training group, 15 autophagy-related lncRNAs were found to have a prognostic value (AC026369.3, USP30-as1, AC007991.2, AC104785.1, AC010503.4, AC037198.1, AC010331.1, AF131215.6, AC084357.2, THUMPD3-AS1, U62317.4, MAN1B1-DTt, AC024060.1, AL662844.4, and AC005229.4). The patients were divided into low-risk group and high-risk group based on the prognostic lncRNAs. The overall survival (OS) time for the high-risk group was shorter than that for the low-risk group [risk ratio (hazard ratio, HR) = 1.08, 95% CI: 1.06-1.10; p < 0.0001]. Using our model, the defined risk value can predict the prognosis of a patient. Next, the model was assessed in the TCGA testing group to further validate these results. A total of 203 patients with BC were recruited to verify the lncRNA characteristics. We divided these patients into high-risk group and low-risk group. The results of testing data set show that the survival time of high-risk patients is shorter than that of low-risk patients. In the training group, the area under the curve (AUC) was more than 0.7, indicating a high level of accuracy. The AUC for a risk model was greater than that for each clinical feature alone, indicating that the risk value of a model was the best indicator for predicting the prognosis. Further training data analysis showed that the gene set was significantly enriched in cancer-related pathways, including actin cytoskeleton regulation and gap junctions. In conclusion, our 15 autophagy-related lncRNAs have a prognostic potential for BC, and may play key roles in the biology of BC.

Keywords: TCGA; autophagy; bladder cancer; lncRNA; prognostic indicator.

Figure 1

Prognostic network of long non-coding RNAs (lncRNAs) and co-expressed autophagy genes in bladder cancer (BC). In the center, the red node represents lncRNA, and the blue represents autophagy genes. The co-expression network was constructed by using CYTOSCAPE 3.7.2 software.

Figure 2

Analysis of autophagy-related lncRNA risk scores of patients with BC in The Cancer Genome Atlas (TCGA) training group. (A) Patients with BC were divided into low-risk group (n = 102) and high-risk group (n = 101) based on the median-risk score. (B) Survival status and duration of the survival in patients with BC. (C) Heat map of the expression of 15 key lncRNAs in BC. The color from green to red shows the expression trend from low level to high level.

Figure 3

Kaplan–Meier (KM) survival curve of the autophagy-related lncRNA BC risk score in TCGA training group. In the TCGA data, the 5-year survival rate of high-risk patients is lower than that of low-risk patients.

Figure 4

Multi-factor receiver operating characteristic (ROC) curve. The area under the curve (AUC) value of the model established in the training group is significantly more than 0.7, and is greater than the predicted value of clinical data.

Figure 5

Visualization of the clinical correlation between autophagy-related lncRNA model and patients with BC. The risk score is closely related to grade, union for international cancer control (UICC) stage, and pathological T stage (*p < 0.05).

Figure 6

Analysis of autophagy-related lncRNA risk scores of patients with BC in TCGA testing group. (A) Autophagy-related lncRNA low-risk group (n = 102) and high-risk group (n = 101) in patients with BC. (B) Survival status and duration of BC cases. (C) Heat map of the expression of 15 key lncRNAs in BC. The color from green to red shows the expression trend from low level to high level.

Figure 7

KM survival curve of the autophagy-related lncRNA risk score for BC in TCGA testing group. TCGA data shows that the 5-year survival rate of high-risk patients is lower than that of low-risk patients (log rank p < 0.01).

Figure 8

Multi-factor receiver operating curve details. The AUC value of the testing group is significantly more than 0.7, which is greater than the predicted value of clinical data.

Figure 9

Gene set enrichment analysis (GSEA) based on TCGA data set shows that cancer-related pathways are significantly enriched in highly expressed populations. (A) extracellular matrix (ECM) receptor interaction, (B) focal adhesion, (C) gap junction, (D) regulation of actin cytoskeleton, (E) wnt signaling pathway, (F) small cell lung cancer, (G) glioma. The highly expressed population is significantly enriched in ECM receptor interaction, focal adhesion, gap junction, and the regulation of the actin cytoskeleton pathways.