Research Paper Volume 13, Issue 4 pp 5718—5747

Comprehensive analysis of macrophage-related multigene signature in the tumor microenvironment of head and neck squamous cancer

Construction and analysis of risk score through random forest model. (A) According to the optimal cutoff value of 73.341, patients were divided into high-risk and low-risk groups. (B) Most of the patients with long survival time are concentrated in the low-risk group (blue dots), while the high-risk group patients have significantly reduced survival time (yellow dots), indicating that this model can predict the prognosis well. (C) Cluster analysis of the five genes. Highly expressed UTP3 was significantly associated with high risk, while up-regulated of DNASE1L3, AMAMDEC1, FANCE, and DNAJC13 were associated with low-risk patients. (D) Receiver-operating characteristic (ROC) curve showed predictive ability of five-gene model, and the area under the curve (AUC) of the 1-, 3- and 5-year OS were 0.959, 0.983 and 0.963, respectively.

Figure 5. Construction and analysis of risk score through random forest model. (A) According to the optimal cutoff value of 73.341, patients were divided into high-risk and low-risk groups. (B) Most of the patients with long survival time are concentrated in the low-risk group (blue dots), while the high-risk group patients have significantly reduced survival time (yellow dots), indicating that this model can predict the prognosis well. (C) Cluster analysis of the five genes. Highly expressed UTP3 was significantly associated with high risk, while up-regulated of DNASE1L3, AMAMDEC1, FANCE, and DNAJC13 were associated with low-risk patients. (D) Receiver-operating characteristic (ROC) curve showed predictive ability of five-gene model, and the area under the curve (AUC) of the 1-, 3- and 5-year OS were 0.959, 0.983 and 0.963, respectively.