An Aging-Related Gene Signature-Based Model for Risk Stratification and Prognosis Prediction in Breast Cancer

Figures & data

Table 1 Clinical Co-Variates of the Training and Validation Cohorts

Characteristics	Training Cohort TCGA (N=1000)	Validation Cohort GSE20685 (N=327)
Age (years)
≤45	176 (17.6)	151 (46.2)
>45	824 (82.4)	176 (53.8)
Mean	57.93	47.89
Range	26–89	24–84
T classifications^a
1	265 (26.5)	101 (30.9)
2	582 (58.2)	188 (57.5)
3	123 (12.3)	26 (8.0)
4	30 (3.0)	12 (3.7)
N classifications^a
0	479 (47.9)	137 (41.9)
1	345 (34.5)	87 (26.6)
2	108 (10.8)	63 (19.3)
3	68 (6.8)	40 (12.2)
M classifications^a
0	984 (98.4)	319 (97.6)
1	16 (1.6)	8 (2.4)
Risk score^b
≤0.26	589 (58.9)	268 (82.0)
>0.26	411 (41.1)	59 (18.0)

Notes: ^aDiagnosis based on the AJCC 2010, seventh edition. ^bCut-off values were determined by the maximally selected Log rank statistics.

Figure 1 Flow chart of data collection and analysis.

Figure 2 Identification of a prognosis-related ARG-based signature in the TCGA training cohort. (A) Selection of the optimal candidate genes in the LASSO model. (B) LASSO coefficients of prognosis-associated ARGs, each curve represents a gene. (C) Forest plots showing results of univariate Cox regression analysis between the candidate ARGs expression and overall survival.

Figure 3 Assessment of prognostic value of the ARG signature model in the TCGA training cohort. (A) Determination of cut-off value of ARGs risk scores by the maximally selected Log rank statistics. (B) The distribution of risk scores in the TCGA. (C) Patient distribution in the high- and low-risk group according to overall survival status. (D) The heat map showing expression profiles of the six ARGs. (E) Kaplan–Meier curves for the overall survival of patients in the high- and low-risk group. (F) Multivariate Cox regression analysis of ARGs signature and other clinicopathological factors.

Figure 4 Assessment of prognostic value of the ARG signature model in the GSE20685 validation cohort. (A) The distribution of risk scores in the GSE20685. (B) Patient distribution in the high- and low-risk group according to overall survival status. (C) The heat map showing expression profiles of the six ARGs. (D) Kaplan–Meier curves for the overall survival of patients in the high- and low-risk group. (E) Multivariate Cox regression analysis of ARGs signature and other clinicopathological factors.

Figure 5 The landscape of immune cell infiltration between the high- and low-risk group in the TCGA training cohort. (A) Barplot of different immune cell infiltrations. (B) Heat map of the tumor-infiltrating cell proportions. (C) Correlation matrix of the association between the expression level of the six ARGs and tumor-infiltrating immune cell infiltrations. (D) Violin plot showing differences of infiltrating immune cell types between the low- and the high-risk group.

Figure 6 Development of a nomogram based on ARGs signature for predicting overall survival of patients with BC. (A) The nomogram plot integrating ARG risk score, age, N- and M-classification in the TCGA training cohort. (B) The calibration plot for the probability of 1-, 3-, and 5-year OS in the TCGA training cohort. (C) The calibration plot for the probability of 1-, 3-, and 5-year OS in the GSE20685 validation cohort.