DNA Methylation-Based Panel Predicts Survival of Patients With Clear Cell Renal Cell Carcinoma and Its Correlations With Genomic Metrics and Tumor Immune Cell Infiltration

Abstract

DNA methylation based prognostic factor for patients with clear cell renal cell carcinoma (ccRCC) remains unclear. In the present study, we identified survival-related DNA methylation sites based on the differentially methylated DNA CpG sites between normal renal tissue and ccRCC. Then, these survival-related DNA methylation sites were included into an elastic net regularized Cox proportional hazards regression (CoxPH) model to build a DNA methylation-based panel, which could stratify patients into different survival groups with excellent accuracies in the training set and test set. External validation suggested that the DNA methylation-based panel could effectively distinguish normal controls from tumor samples and classify patients into metastasis group and non-metastasis group. The nomogram containing DNA methylation-based panel was reliable in clinical settings. Higher total mutation number, SCNA level, and MATH score were associated with higher methylation risk. The innate immune, ratio between CD8⁺T cell versus Treg cell as well as Th17 cell versus Th2 cell were significantly decreased in high methylation risk group. In inclusion, we developed a DNA methylation-based panel which might be independent prognostic factor in ccRCC. Patients with higher methylation risk were associated genomic alteration and poor immune microenvironment.

Keywords: clear cell renal cell carcinoma; elastic net; methylation CpG island; prognostication panel; survival analysis.

FIGURE 1

Differential methylation analysis. (A) Volcano plot for differentially methylated CpG sites between normal renal tissue and renal cell carcinoma. (B) The methylation levels of the 11 CpG site in the high risk group, low risk group, and normal renal tissue group.

FIGURE 2

The prognosis relevance of the DNA methylation panel. (A) Time dependent Receiver operating characteristic (ROC) analysis of the prediction ability at the DNA methylation panel at different time points in the training set. (B) The overall survival (OS) of patients in the methylation low risk group and the high risk group in the training set. (C) ROC analysis of the prediction ability at the DNA methylation panel at different time points in the test set. (D) The OS of patients in the methylation low risk group and the high risk group in the test set.

FIGURE 3

External validation of the prediction ability of the DNA methylation-based panel. (A) The methylation risk of patients in normal renal tissue group and ccRCC group and its diagnostic accuracy (B); (C) the methylation risk of patients who develop metastasis and not develop metastasis and its diagnostic accuracy (D).

FIGURE 4

The clinical application of the DNA methylation panel containing nomogram. (A) Nomogram incorporating the DNA methylation-based panel, age, hemoglobin level, serum calcium level, tumor stage, and gender. (B) Calibration analysis of the prediction ability of the nomogram at 3 years. (C) Calibration analysis of the prediction ability of the nomogram at 5 years. (D) Decision curve analysis on the nomogram.

FIGURE 5

Comparison of prognostication performance between the DNA methylation panel and other existing prognostic markers.

FIGURE 6

The associations between the DNA methylation risk and total mutation number (A), mutant-allele tumor heterogeneity (B), and somatic copy number alterations (C).

FIGURE 7

The associations between the DNA methylation risk and the tumor microenvironment (TME) of patients, including T cell infiltration score (A), NK cells (B), mast cells (C), the ratio between CD8⁺ T cells versus Treg cells (D), the ratio between Th17 cells versus Th2 cells (E), the expression of PDCD1 (F), the expression of PDCD1LG2 (G), the expression CD274 (H), and the expression of CTLA4 (I).