Predicting prognosis, immunotherapy and distinguishing cold and hot tumors in clear cell renal cell carcinoma based on anoikis-related lncRNAs

Abstract

Background: Clear cell renal cell carcinoma (ccRCC) is the most frequently occurring malignant tumor within the kidney cancer subtype. It has low sensitivity to traditional radiotherapy and chemotherapy, the optimal treatment for localized ccRCC has been surgical resection, but even with complete resection the tumor will be eventually developed into metastatic disease in up to 40% of localized ccRCC. For this reason, it is crucial to find early diagnostic and treatment markers for ccRCC.

Methods: We obtained anoikis-related genes (ANRGs) integrated from Genecards and Harmonizome dataset. The anoikis-related risk model was constructed based on 12 anoikis-related lncRNAs (ARlncRNAs) and verified by principal component analysis (PCA), Receiver operating characteristic (ROC) curves, and T-distributed stochastic neighbor embedding (t-SNE), and the role of the risk score in ccRCC immune cell infiltration, immune checkpoint expression levels, and drug sensitivity was evaluated by various algorithms. Additionally, we divided patients based on ARlncRNAs into cold and hot tumor clusters using the ConsensusClusterPlus (CC) package.

Results: The AUC of risk score was the highest among various factors, including age, gender, and stage, indicating that the model we built to predict survival was more accurate than the other clinical features. There was greater sensitivity to targeted drugs like Axitinib, Pazopanib, and Sunitinib in the high-risk group, as well as immunotherapy drugs. This shows that the risk-scoring model can accurately identify candidates for ccRCC immunotherapy and targeted therapy. Furthermore, our results suggest that cluster 1 is equivalent to hot tumors with enhanced sensitivity to immunotherapy drugs.

Conclusion: Collectively, we developed a risk score model based on 12 prognostic lncRNAs, expected to become a new tool for evaluating the prognosis of patients with ccRCC, providing different immunotherapy strategies by screening for hot and cold tumors.

Keywords: anoikis; clear cell renal cell carcinoma; hot and cold tumors; immunotherapy; lncRNA; prognostic biomarkers.

Figure 1

Flow chart of the whole design.

Figure 2

Anoikis-related lncRNAs. (A) The co-expression relationship network diagram of anoikis-related genes and lncRNAs. (B, C) The heat map and volcano map of differentially expressed lncRNAs. (D, E) The LASSO regression was performed, using the minimum criterion.

Figure 3

Construction of the ARlncRNA signature model in the train, test, and overall test. Overall survival analysis, risk score distribution, individual survival status, and heat map of 12 ARlncRNAs expression in high and low-risk groups for (A-D) the overall set, (E-H) the train set, and (I-L) the test set.

Figure 4

The validation of 12 lncRNAs prognostic model. (A, B) Single-factor and multi-factor Cox regression analysis of the risk score combined with clinical features. (C) Comparison of ROC curves for risk score and other clinical factors. (D) The ROC curves for 1-, 3-, and 5-year overall survival predictions by the risk score model in the training set. (E, F), PCA, and t-SNE analyses of the train set.

Figure 5

Risk score difference analysis of different clinical features. (A-D) Box line diagram and (E-L) Kaplan-Meier survival curves in groups stratified by gender, age, stage, and M status.

Figure 6

Construction and Verification of Predictive Nomogram. (A) The nomogram for predicting 1-year, 3-year, and 5-year survival rates (‘***’ p< 0.001). (B) The calibration curves of the nomogram. (C) ROC curve of risk score and clinical features. (D, E) The univariable and multivariable Cox analysis of nomogram. **p < 0.01; ***p < 0.001.

Figure 7

Risk score and Immune infiltration landscape evaluation. (A) Evaluation of immune cell infiltration in two groups. The correlation between the risk score and infiltration of CD8+T cells (B), CD4+T cells (C), neutrophils (D), M1 macrophages (E), M0 macrophages (F), naïve B cells (G), M1 macrophages (H), and activated mast cells (I) was examined.

Figure 8

Prediction of immunotherapy. (A-C) Differences in TME scores between high- and low-risk groups. (D, E) Analysis of immune cells and related functions in two groups. (F) Differential expression of immune checkpoint in two groups. **p < 0.01; ***p < 0.001.

Figure 9

The differences in drug sensitivity of patients. (A) Axitinib, (B) Pazopanib, and (C) Sunitinib.

Figure 10

Using consensus clustering analysis, patients can be grouped into two clusters. (A) Consensus Cluster Plus divides patients into two categories. (B), Kaplan-Meier survival analysis. (C) Correspondence between typing and high and low risk groups. (D, E) PCA and t-SNE analyses were performed for the two clusters.

Figure 11

Infiltration of different immune cells in two clusters. (A-C) The degree of the immune score, stromal score, and estimated score in distinct clusters. (D) A Heat map of immune cells in two clusters based on different platform algorithms. (E) The immune-related functions of two clusters were analyzed by ssGSEA analysis. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 12

IC50 of Dasatinib (A), Bosutinib (B), Nilotinib (C), and Lapatinib (D) targeted drugs in distinct clusters.