Differential expression of CDKN2A in breast cancer patients based on TGCA data
First, the CDKN2A expression data of 122 normal tissues and 1,066 tumor tissues from breast cancer patients were downloaded from TCGA in March 2019. We detected that CDKN2A levels were markedly higher in tumor tissues (P= 0.000) (Figure 1A). Besides, we compared CDKN2A mRNA levels in normal tissues and matched tumor tissues in breast cancer patients. Similarly, CDKN2A levels were higher in the tumor tissues (P= 0.000) (Figure 1B), which implies that CDKN2A may play a relevant role in breast cancer.
Clinicopathological parameters
As displayed in Table 1, we analysed Clinicopathological parameters of 832 female breast cancer patients according to clinical data downloaded from TCGA after excluding incomplete clinical prognostic information. The median age of confirmed diagnosis was 58 years old. Most of pathologic types (73.9%, n= 615) were of infiltrating ductal carcinoma, 19.6% (n= 163) were infiltrating lobular carcinoma, least were mixed histology (n= 25) and another specific carcinoma (n= 29). Moreover, clinical stage I of breast cancer was found in 143 patients (17.19%), stage II was found in 482 patients (57.93%), stage III was found in 193 patients (23.20%), and stage IV was found in 14 patients (1.68%). Besides, there were four different subtypes of breast cancer, including Luminal A (73.6%, n= 613), Luminal B (7.7%, n= 64), HER2-enriched (17.1%, n= 142) and Triple negative (1.6%, n= 13). The mean and median follow-up times after surgery were 3.2 and 2.3 years respectively. During the 23 years of follow-up in this survey, 109 (13.1%) patients died.
Table 1. Clinicopathological parameters of the breast cancer patients.
Clinicopathological parameters
|
N
|
Age
|
|
≤58
|
416
|
>58
|
416
|
Histological type
|
N
|
Infiltrating ductal carcinoma
|
615
|
Infiltrating lobular carcinoma
|
163
|
Mixed histology
|
25
|
Another specific carcinoma
|
29
|
Tumor classification
|
|
T1
|
220
|
T2
|
486
|
T3
|
104
|
T4
|
22
|
Lymphatic classification
|
|
N0
|
396
|
N1
|
281
|
N2
|
105
|
N3
|
50
|
Distant classification
|
|
M0
|
818
|
M1
|
14
|
Clinical stage
|
|
I
|
143
|
II
|
482
|
III
|
193
|
IV
|
14
|
Estrogen receptor
|
|
Positive
|
666
|
Negative
|
166
|
Progesterone receptor
|
|
Positive
|
583
|
Negative
|
249
|
HER2
|
|
Positive
|
77
|
Negative
|
755
|
Molecular subtypes
|
|
Luminal A
|
613
|
Luminal B
|
64
|
Triple negative
|
142
|
HER2-enriched
|
13
|
Survival status
|
|
Death
|
109
|
Survival
|
723
|
CDKN2A expression in breast cancer tissues according to clinical characteristics
We further investigated CDKN2A expression between tumour tissues and normal tissues in breast cancer patients based on clinical characteristics (Figure 2). Our results revealed that there were differences in CDKN2A expression among four molecular subtypes (P= 0.000) (Figure 2I). In addition, we found that as the clinical stage increased, CDKN2A expression decreased inversely (P= 0.491). However, no differences were observed according to TNM stage, distant metastasis and HER2 status (P= 0.917, P= 0.458, P= 0.5 and P= 0.579 respectively).
CDKN2A expression is associated with clinicopathological features
To evaluate the clinical significance of CDKN2A, we analysed the associations between CDKN2A expression and the clinicopathological characteristics by chi-square test. CDKN2A expression was correlated with age (P= 0.018), histological types (P= 0.028), ER status (P= 0.000), PR status (P= 0.000) and molecular subtypes (P= 0.000), respectively. However, no significant correlation was detected in TNM stage and HER2 status (Table 2).
Table 2. Relationships of CDKN2A expression with clinical features based on TGCA data.
Variable
|
CDKN2A (n)
|
low
|
high
|
P1
|
n=416
|
n=416
|
|
Age(years)
|
≤58
|
191
|
225
|
0.018
|
>58
|
225
|
191
|
Histological types
|
Infiltrating ductal carcinoma
|
315
|
300
|
0.028
|
Infiltrating lobular carcinoma
|
69
|
94
|
Mixed histology
|
18
|
7
|
Other specific carcinoma
|
14
|
15
|
T classification
|
T1
|
113
|
107
|
0.924
|
T2
|
239
|
247
|
T3
|
52
|
52
|
T4
|
12
|
10
|
Lymphatic metastasis
|
N0
|
195
|
201
|
0.051
|
N1
|
135
|
146
|
N2
|
65
|
40
|
N3
|
21
|
29
|
Distant metastasis
|
M0
|
409
|
409
|
1.000
|
M1
|
7
|
7
|
Clinical stage
|
I
|
69
|
74
|
0.831
|
II
|
238
|
244
|
III
|
102
|
91
|
Ⅳ
|
7
|
7
|
ER
|
Positive
|
364
|
302
|
0.000
|
Negative
|
52
|
114
|
HR
|
Positive
|
316
|
267
|
0.000
|
Negative
|
100
|
149
|
HER2
|
Positive
|
41
|
36
|
0.550
|
Negative
|
375
|
380
|
Molecular subtypes
|
Luminal A
|
334
|
279
|
0.000
|
Luminal B
|
35
|
29
|
Triple negative
|
41
|
101
|
HER2-enriched
|
6
|
7
|
The bold number is on behalf of the P-values with significant differences. 1P Value was calculated by χ2 test. ER: estrogen receptor; PR: progesterone receptor; HER2: human epidermal growth factor receptor 2.
High CDKN2A expression is an independent protective factor for OS
Survival analysis showed that high CDKN2A expression was positively associated with overall survival (OS) (P=0.000, Figure 3A). We also found that increased CDKN2A expression was related to superior overall survival, in spite of age, lymphatic metastasis and HER2 status. Beyond that, high CDKN2A expression was correlated with superior OS, especially in breast cancer patients with T2-T3 classification, clinical stage II-III and luminal-like subtype but without distant metastasis (Figure 3B-P).
Univariate Cox regression analysis showed that CDKN2A expression, age, TNM stage and HER2 status were significantly associated with OS (Table 3). Moreover, multivariate analysis confirmed that CDKN2A expression and age were independent predictors of OS in breast cancer patients (Table 3).
Table 3. Univariate and multivariate analyses in breast cancer patients
Variable
|
Univariate analysis
|
Multivariate analysis
|
|
HR
|
95% CI
|
P value
|
HR
|
95% CI
|
P value
|
CDKN2A
|
0.701
|
0.584-0.842
|
0.000
|
0.976
|
0.953-0.998
|
0.037
|
Age
|
1.033
|
1.018-1.049
|
0.000
|
1.030
|
1.014-1.046
|
0.000
|
T
|
1.624
|
1.280-2.059
|
0.000
|
1.260
|
0.964-1.645
|
0.091
|
N
|
1.624
|
1.334-1.978
|
0.000
|
1.267
|
0.999-1.607
|
0.051
|
M
|
3.251
|
1.961-5.391
|
0.000
|
1.885
|
0.925-3.842
|
0.081
|
STAGE
|
0.571
|
0.419-1.049
|
0.000
|
0.704
|
0.482-1.028
|
0.069
|
ER
|
1.067
|
0.697-1.632
|
0.765
|
|
|
|
PR
|
1.225
|
0.831-1.806
|
0.304
|
|
|
|
HER-2
|
0.501
|
0.312-0.811
|
0.004
|
0.743
|
0.441-1.253
|
0.265
|
Molecular types
|
1.143
|
0.939-1.391
|
0.182
|
|
|
|
HR: Hazard ratio; CI: Confidence interval. The bold number is on behalf of P-values with significant differences.
CDKN2A-related signaling pathway
We implemented GSEA to recognize the signalling pathways activated in breast cancer by comparing the low and high CDKN2A expression datasets. GSEA showed significant differences (FDR<0.25, NOM P-value <0.05) in the enrichment of the MSigDB Collection pathways. As shown in Table 4, we chose the signalling pathways that were most significantly enriched in breast cancer patients with high CDKN2A expression, including pathways related to bladder cancer, CAMs, the cell cycle, cytokine-receptor interactions, cytosolic DNA sensing, DNA replication, killer cell-mediated cytotoxicity and P53 (Figure 4).
Table 4. Gene sets enriched in the high CDKN2A expression.
Gene set name
|
NES
|
NOM p-val
|
FDR q-val
|
BLADDER_CANCER
|
1.857
|
0.006
|
0.108
|
DNA_REPLICATION
|
1.853
|
0.013
|
0.094
|
CELL_ADHESION_MOLECULES_CAMS
|
1.763
|
0.032
|
0.062
|
CELL_CYCLE
|
1.780
|
0.040
|
0.058
|
CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION
|
1.656
|
0.042
|
0.086
|
NATURAL_KILLER_CELL_MEDIATED_CYTOTOXICITY
|
1.634
|
0.043
|
0.085
|
P53_SIGNALING_PATHWAY
|
1.602
|
0.047
|
0.093
|
CYTOSOLIC_DNA_SENSING_PATHWAY
|
1.556
|
0.048
|
0.114
|
NES: normalized enrichment score; NOM: nominal; FDR: false discovery rate. Gene sets with NOM p-val<0.05 and FDR q-val<0.25 are regarded as significant.