Expression of CDK7 correlates with molecular subtypes and predicts clinical outcomes in breast cancer.

BACKGROUND: Cyclin-dependent kinase 7 (CDK7) belongs to CDK family, which is involved in transcriptional activity and cell cycle progression. Recent studies have suggested that CDK7 inhibitor is effective in the treatment of breast cancer, whereas, the efficacy of antitumor response varies significantly in different subtypes of breast cancer. However, the underlying connection between CDK7 and molecular subtypes of breast cancer is currently unknown.
METHODS: To further elucidate the role of CDK7 in breast cancer, we investigated the association of CDK7 expression with different molecular subtypes and clinical outcomes in breast cancer using the Oncomine, GENT2, UALCAN and PrognoScan database. The correlation between CDK7 mRNA expression and promoter methylation was established using MEXPRESS datasets. And the CDK7 expression with different stages and subtypes was also examined by immunohistochemistry (IHC) assay in 140 breast cancer patients.
RESULTS: We found that the expression of CDK7 was increased in breast cancer. Besides, our data demonstrated that both CDK7 hypomethylation status and copy number variations (CNVs) were highly correlated with mRNA overexpression. Moreover, based on the tissue array of 140 patients and more than 4,000 samples from bc-GenExMiner database, CDK7 protein levels were significantly higher in luminal and human epidermal growth factor receptor 2 positive (HER2+) breast cancer in comparison with triple-negative breast cancer (TNBC) subtype. The expression of CDK7 was prominently correlated with molecular markers of different subtypes of breast cancer. Furthermore, the results of the univariate analysis indicated that high CDK7 expression was dramatically associated with poor overall survival (OS) (P=0.0323, HR =2.08).
CONCLUSIONS: Our results highlighted that elevated CDK7 expression was correlated with molecular subtypes and acted as a candidate biomarker of poor prognosis in breast cancer. 2021 Translational Cancer Research. All rights reserved.

Introduction

Breast cancer is one of the most prominent cancer types that lead to female mortality around the world (1). Great improvement has been achieved in the development of diagnosis and treatment for breast cancer patients. Breast cancer is a genetic and phenotypic heterogeneity disease with multiple distinct subtypes (2). Therefore, more reliable markers need to be pinpointed for early diagnosis and therapeutic strategies for individual patients.

Preclinical research indicated various potential interventional targets related to tumorigenesis including cell cycle process, which was regulated by cyclin-dependent kinases (CDKs) and other regulatory molecules. For example, CDK7 binds to MAT1 and cyclin H to constitute the CDK activating kinase (CAK), which executes its function by phosphorylating other CDKs involved in cell cycle control (3,4). Previous studies reported that CDK7 inhibitors affected transcriptional activity and cell cycle progression in various cancers (5-9).

Based on estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) protein expression status, breast cancers could be classified into luminal A carcinomas (ER+ and/or PR+, HER2–), luminal B carcinomas (ER+ and/or PR+, HER2+), HER2 carcinomas (ER–, PR–, HER2+) and triple-negative carcinomas (ER–, PR–, HER2–) (10). In contrast to ER+ breast cancer, triple-negative breast cancer (TNBC) was reported to be uniquely dependent on CDK7 (7). Another study has shown that HER2+ breast cancers may be more sensitive to CDK7 inhibition THZ1 than ER+ breast cancers (11). To date, recent studies have shown that CDK7 inhibition is effective in breast cancer with significant subtype selectivity. However, little information has been published on the underlying connection between CDK7 and molecular subtypes of breast cancer patients.

Thus, we attempted to elucidate the function of CDK7 in breast cancer using public bioinformatics databases and tissue microarrays process to establish the correlation of CDK7 expression with promoter methylation status, clinical outcomes and makers of different subtype of breast cancer. Meantime, we further identified that CDK7-related carcinogenic pathways changed at the molecular level in breast cancer. Collectively, our results might further support the application of CDK7 selective inhibitors in the treatment of certain subtypes of breast cancer. We present the following article in accordance with the REMARK reporting checklist (available at http://dx.doi.org/10.21037/tcr-20-2911).

Methods

Public database and bioinformatics analysis

CDK7 mRNA expression in various types of cancer and normal tissues was ascertained by the Oncomine database (https://www.oncomine.org/resource/main.html) (12) and GENT2 database (http://gent2.appex.kr/gent2/) (13). CDK7 levels and clinical parameters in TCGA (https://www.cancer.gov/tcga) were extracted and examined in the UALCAN platform (http://ualcan.path.uab.edu) (14). The PrognoScan database (http://dna00.bio.kyutech.ac.jp/PrognoScan/index.html) (15) and the Human Protein Atlas database (https://www.proteinatlas.org) (16) were used to investigate the prognostic role of CDK7 mRNA in breast cancer. The bc-GenExMiner database (http://bcgenex.centregauducheau.fr/BC-GEM) (17) was used to analyze CDK7 gene on ER, PR, and HER2. The correlation of CDK7 with molecular subtype markers from breast cancers was performed on GEPIA database (http://gepia.cancer-pku.cn/index.html) (18). The cBioPortal for Cancer Genomics (http://www.cbioportal.org/) (19) was a platform which was applied to visualize and analyze genetic alterations of CDK7. We evaluated the relationship between CDK7 expression and its methylation and copy number variations (CNVs) using MEXPRESS (https://mexpress.be) (20). The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethical Committee of the First Affiliated Hospital of Nanjing Medical University (2019-SRFA-197) and individual consent for this retrospective analysis was waived.

Tissue microarray, patients, and follow-up

CDK7 was detected by immunohistochemistry (IHC) according to standard methods. The breast cancer tissue microarray (HBre-Duc140Sur-01) was obtained from Shanghai Outdo Biotech (Shanghai, China), and included tissue samples collected from January 2001 to August 2004. The tissue microarray had been previously described (21). The follow-up time was up to July 2013.

Analysis and quantification of staining

Anti-CDK7 antibody (Cell Signaling Technology, USA, #2916, 1:500) was used to perform IHC staining. CDK7 expression intensity was divided into: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The staining percentage was as follows: 0 (<10% positive cells), 1 (10–25% positive cells), 2 (25–49% positive cells), 3 (50–74% positive cells) and 4 (≥75% positive cells). We combined two parameters to calculate the total score. Immunohistochemical staining was performed blindly by two pathologists. While, if the two pathologists disagree, we also consulted a third party for judgment. IHC score ≥6 indicated positive staining (22).

Statistical analysis

The Chi-squared and Fisher’s exact test were executed to analyze the relationship between CDK7 expression and clinical pathological characteristics via SPSS 26. We set the significance level for this study to be P values <0.05.

Results

The mRNA levels of CDK7 in breast cancer

By analyzing the GENT databases and Oncomine databases, we investigated CDK7 expression in cancer and normal samples. In GENT2 database, CDK7 levels were significantly upregulated in breast carcinoma (). Oncomine database was used to detect the expression of CDK7 mRNA in tumors and normal clinical samples, and the mRNA levels of CDK7 in breast carcinoma were displayed in . Besides, analysis of the Oncomine database revealed that CDK7 mRNA levels were notably higher in lobular breast carcinoma (LBC), ductal breast carcinoma (DBC) and invasive ductal breast carcinoma (IDBC) () (10,23).

Figure 1

Transcription levels of CDK7 in breast cancer. (A) CDK7 mRNA levels in normal and different types of cancer was searched in the GENT2 database. (B) The plot showed the number of datasets with highly significant mRNA down-expression (blue) or over-expression (red) of CDK7 (cancer vs. normal) from the Oncomine database. Threshold parameters: fold change of 1.5, and P value of 0.001. (C,D,E,F) Box plots compared CDK7 levels in normal and breast carcinoma using Oncomine database. The analysis was shown in (C,D) LBC, (E) DBC, (F) IDBC relative to normal breast tissue. CDK7, cyclin-dependent kinase 7; LBC, lobular breast carcinoma; DBC, ductal breast carcinoma; IDBC, invasive ductal breast carcinoma.

Analysis of GENT2 and the Oncomine databases indicated that CDK7 levels were significantly increased in breast carcinoma. Therefore, we next investigated the correlation between CDK7 levels and clinical parameters of breast carcinoma samples in UALCAN database. According to the subgroup analysis based on the patient’s race, age, disease stage, lymph node metastasis status and TP53 mutation status, CDK7 expression in breast carcinoma patients was significantly higher than the normal control group. Thus, expression of CDK7 could be served as a potential diagnostic indicator in breast cancer ().

Figure 2

Relationship between CDK7 expression and clinical parameters in breast cancer. The UALCAN database was used to assess the relationship between CDK7 expression and clinical parameters such as: (A) sample types, (B) race, (C) age, (D) tumor stages, (E) nodal metastasis status, (F) TP53 mutation status. ***, P<0.001 versus normal. CDK7, cyclin-dependent kinase 7.

Integrated analysis of CDK7 DNA methylation and CNVs in breast cancer

DNA methylation was a major epigenetic mechanism that modulated gene expression, which was a vital biological mechanism for tumor occurrence and development (24). Here, we investigated whether CDK7 mRNA expression was correlated with its methylation status using MEXPRESS. As shown in , the plot indicated that there was a negative correlation between CDK7 expression and methylation around the CpG island and promoter region. Seven abnormal methylation sites (cg26166102, r=–0.073; cg11859953, r=–0.151; cg18275539, r=–0.118; cg01558324, r=–0.160; cg07850415, r=–0.131; cg0227054, r=–0.076; cg16181978, r=–0.071) were observed and the Pearson correlation coefficients were indicated on the right-hand side. Here, DNA methylation of the CDK7 promoter may further support CDK7 overexpression in breast carcinoma. In addition to epigenetic mechanism, we speculate that CDK7 gene mutation may also affect its expression.

Figure 3

Relationship between CDK7 levels and its methylation and CNVs in breast cancer. (A) This view showed the relationship between CDK7 expression and CNVs, as well as methylation around CpG island and promoter region. (B) Genetic CDK7 mutation in TCGA breast cancer cohort. (C) CDK7 mutation type in TCGA breast cancer cohort. CDK7, cyclin-dependent kinase 7; CNV, copy number variation.

Our finding indicated that CDK7 expression was positively associated with its CNVs changes using MEXPRESS (r=0.542, P<0.0001). Our results subsequently indicated that CNVs of CDK7 frequently occurred in breast cancer samples. Meanwhile, the cBioCancer Genomics Portal was accustomed to investigating the frequency of CDK7 mutations in breast cancer. CDK7 contained multiple mutations in breast cancer such as amplification, fusion, deep deletion and missense (), among which, missense was the major forms of mutation ().

CDK7 levels predict prognosis in breast cancer

We evaluated whether CDK7 levels were associated with clinical outcomes among breast cancer. Notably, CDK7 levels significantly impacted prognosis in breast cancer. The correlation between CDK7 levels and prognosis of breast cancer patients was displayed in . Four cohorts (GSE9893, GSE2990, GSE3494-GPL96, GSE4922-GPL96) including 155 cases, 125 cases, 236 cases, and 249 cases at various stages of breast carcinoma showed that high CDK7 expression was correlated with poor prognosis (). To further unveil the prognostic value of CDK7 in breast carcinoma, we used Human Protein Atlas database (P=0.0087) (). Therefore, our results suggested that high CDK7 expression could be a risk factor for poor prognosis in breast cancer.

Table 1

Relationship between CDK7 levels and survival trends in breast cancer

Dataset	Endpoint	Number	Probe ID	In (HRhigh/HRlow)	COX P value	In (HR)	HR (95% CI)
GSE9893	OS	155	3581	0.91	0.008742	0.2	1.22 (1.05–1.41)
GSE2990	DMFS	125	211297_s_at	1.25	0.000905	1.19	3.28 (1.63–6.63)
GSE2990	RFS	125	211297_s_at	0.67	0.044463	0.58	1.78 (1.01–3.12)
GSE3494-GPL96	DSS	236	211297_s_at	1.19	0.020482	0.90	2.46 (1.15–5.26)
GSE4922-GPL96	DFS	249	211297_s_at	0.96	0.046541	0.62	1.86 (1.01–3.44)

CDK7, cyclin-dependent kinase 7; OS, overall survival; DMFS, distant metastasis free survival; RFS, relapse free survival; DSS, disease specific survival; DFS, disease free survival.

Figure 4

Survival assay of CDK7 in breast cancer patients. Survive curves evaluating the prognostic value of CDK7 mRNA expression in the PrognoScan, including: (A) OS (GSE9893 n=155), (B) DMFS (GSE2990 n=125), (C) RFS (GSE2990 n=125), (D) DSS (GSE3494-GPL96 n=236), (E) DFS (GSE4922-GPL96 n=249). (F) Impact of CDK7 level on OS in TCGA breast cancer cohort (n=1,040). CDK7, cyclin-dependent kinase 7; OS, overall survival; DMFS, distant metastasis free survival; RFS, relapse free survival; DSS, disease specific survival; DFS, disease free survival.

CDK7 expression is related to molecular subtype and poor prognosis of breast cancer

We evaluated the clinicopathological features correlated with CDK7 levels in 140 breast cancer samples by immunohistochemical analysis. CDK7 protein was mainly existed in the nucleus. Representative photographs of immunohistochemical staining for breast cancer tissues were presented in . As shown in , patients with higher CDK7 levels appeared to be associated with advanced stages. What’s more, the expression of CDK7 was positively correlated with ER status (P=0.02) and PR status (P=0.011), which suggested that luminal breast cancer might positively express CDK7.

Figure 5

CDK7 levels correlated with molecular subtypes. (A) CDK7 staining was shown as representative photographs. The staining results were negative (0), weak (1+), moderate (2+), and strong (3+). (B) IHC scores indicated that CDK7 levels were higher in luminal and HER2+ compared with TNBC subtype. (C) IHC scores indicated that CDK7 levels were higher in luminal compared with not luminal subtype. (D) IHC scores indicated that CDK7 levels were lower in TNBC compared with not TNBC subtype. (E) IHC scores indicated that CDK7 levels were not significantly associated with HER2 status at whole level. (F) CDK7 levels were higher in ER+ than ER– breast cancer at the absence of HER2 enriched cases. (G) Kaplan Meier survival curves compared the high and low expression of CDK7 in breast cancer patients. *, P<0.05; **, P<0.005. IHC, immunohistochemistry; TNBC, triple-negative breast cancer.

Table 2

Relationship between CDK7 expression and clinicopathologic characteristics of breast cancer samples

Variables	CDK7a, n (%)		Total, n (%)	P value
	Low	High
Age				0.369
<50 years	32 (47.8)	26 (40.0)	58 (43.9)
≥50 years	35 (52.2)	39 (60.0)	74 (56.1)
Pathology gradeb				0.145
Grade 1	8 (12.3)	3 (4.7)	11 (8.5)
Grade 2	54 (83.1)	54 (84.4)	108 (83.7)
Grade 3	3 (4.6)	7 (10.9)	10 (7.8)
T stage				0.001
1	22 (32.8)	5 (7.7)	27 (20.5)
2	37 (55.2)	54 (83.1)	91 (68.9)
3	8 (11.9)	6 (9.2)	14 (10.6)
Nodal statusc				0.747
0	23 (35.4)	25 (38.5)	48 (36.9)
1	20 (30.8)	19 (29.2)	39 (30.0)
2	16 (24.6)	18 (27.7)	34 (26.2)
3	6 (9.2)	3 (4.6)	9 (6.9)
ER statusd				0.020
Positive	35 (55.6)	46 (75.4)	81 (65.3)
Negative	28 (44.4)	15 (24.6)	43 (34.7)
PR statuse				0.011
Positive	27 (42.9)	40 (65.6)	67 (54.1)
Negative	36 (57.1)	21 (34.4)	57 (45.9)
HER2 statusf				0.754
Positive	18 (28.6)	19 (31.1)	37 (29.8)
Negative	45 (71.4)	42 (68.9)	87 (70.2)
Subtypes				0.047
Luminal	35 (55.6)	46 (75.4)	81 (65.3)
HER2-positive	11 (17.5)	8 (13.1)	19 (15.3)
TNBC	17 (26.9)	7 (11.5)	24 (19.4)

a, eight samples staining missed; b, two and one samples of pathology grade status of CDK7 low and high expression group were missed, respectively; c, two samples of nodal status of CDK7 low expression group were missed; d, four and four samples of ER status of CDK7 low and high expression group were missed, respectively; e, four and four samples of PR status of CDK7 low and high expression group were missed, respectively; f, four and four samples of HER2 status of CDK7 low and high expression group were missed, respectively. CDK7, cyclin-dependent kinase 7; ER, estrogen receptor; PR, progesterone receptor; TNBC, triple-negative breast cancer.

Next, we assessed the association between CDK7 protein levels and molecular subtypes. IHC scores showed that CDK7 protein was significantly higher in luminal (P=0.0018) and HER2+ (P=0.0194), compared with CDK7 protein levels in TNBC (). Also, CDK7 levels were significantly different between luminal vs. not luminal (P=0.0437), and TNBC vs. not TNBC (P=0.0039) (). However, CDK7 levels were not associated with HER2 status (P=0.0523) (). Samples of HER2 positive status were excluded in the next analysis because CDK7 levels were significantly higher in HER2 positive samples, which could improve CDK7 expression in ER– groups. CDK7 protein levels were significantly positively correlated with ER status after excluding HER2-positive cases (P=0.0179) ().

More importantly, we investigated the relationship between CDK7 levels and clinical outcome by extracting survival data. The results of the univariate analysis indicated that CDK7 overexpression was remarkably correlated with shorter OS in breast cancer patients (P=0.0323, HR =2.08) ().

In bc-GenExMiner database, CDK7 expression was enriched in luminal and HER2+ compared with basal-like subtype (). ER and PR status were both positively associated with CDK7 expression (). In addition, there were no obvious difference of CDK7 levels between HER2+ and HER2– group (). Moreover, compared with TNBC and basal-like breast cancer, CDK7 levels were apparently upregulated in breast cancer subtypes other than TNBC and basal-like breast cancer ().

Figure 6

CDK7 levels in different groups were analyzed according to clinical parameters by the bc-GenExMiner database. The box plot showed as: (A) subtypes, (B) ER, (C) PR, (D) HER2, (E) triple-negative status, and (F) basal-like status. CDK7, cyclin-dependent kinase 7; ER, estrogen receptor; PR, progesterone receptor; IHC, immunohistochemistry.

Correlation between CDK7 and luminal, HER2+, and basal-like subtype markers

We evaluated whether there was any association between CDK7 levels and luminal, HER2+, and basal-like subtype markers from breast cancer using the GEPIA web tools. Correlation analysis showed that CDK7 levels were positively correlated with luminal subtype markers of estrogen receptor (ESR1) (R=0.37, P<0.001), progesterone receptor (PGR) (R=0.25, P<0.001), endothelial transcription factor 3 (GATA3) (R=0.42, P<0.001), and were negatively correlated with basal-like subtype marker epidermal growth factor receptor (EGFR) (R=–0.091,P<0.001), keratin 5 (KRT5) (R=–0.11, P<0.001), but were not correlated with HER2+ subtype markers of epidermal growth factor receptor-2 (ERBB2) (P=0.43) ().

Figure 7

Correlations between CDK7 levels and markers of different subtypes of breast cancer. Correlation analysis of TCGA exhibited that CDK7 mRNA was associated with (A) ESR1, (B) PGR, (C) GATA3, (D) ERBB2, (E) EGFR, and (F) KRT5. CDK7, cyclin-dependent kinase 7.

Discussion

Altogether, our analysis of the Oncomine database and UALCAN showed that CDK7 mRNA was significantly upregulated among breast cancer. Our study verified the correlation between CDK7 expression and methylation status or copy number amplification, which may lead to an increase in CDK7 expression levels. Consistently, CDK7 overexpression was significantly associated with disease stages, nodal metastasis status and TP53 mutation status, which indicated that high expression of CDK7 could be involved in breast carcinoma progression. Accordingly, high CDK7 levels were significantly associated with shorter OS.

Based on tissue microarray and database information, our analysis showed that CDK7 protein levels were significantly higher in luminal and HER2+ breast cancer compared with TNBC. We found the CDK7 levels were positively associated with luminal subtype makers. Although the potential mechanism of the relationship between CDK7 expression and ER remained unclear, previous studies had revealed that CDK7 was closely related to ER phosphorylation at Ser118, and in vitro studies had confirmed that CDK7 can phosphorylate this site (25). This may provide a potential explanation for the association between CDK7 and ER protein levels in breast carcinoma. The result supports the design of CDK7 inhibitor in treating ER+ breast cancer, either being administered alone or in coordination with hormone therapy, which was expected to be a promising strategy to overcome resistance to endocrine therapy. These results indicated that CDK7 was not only a biomarker with diagnostic and prognostic value, but also a potential therapeutic target.

For the moment, a great deal of selective small molecule CDK7 inhibitors have been developed. Preclinical studies have found that CDK7 inhibitors can be effective for a variety of solid tumors, including breast cancer (7,8,26-28). Based on these previous studies, currently four CDK7 inhibitors, CT7001, SY-1365, SY-5609 and LY3405105, have now progressed to phase I/II clinical trials to test monotherapy as well as combination treatments in breast cancer. Our research demonstrated that luminal subtype has the highest expression of CDK7. However, the sensitivity of CDK7 inhibitors to different subtypes of breast cancer did not depend on the expression levels. It had been reported that it was TNBC not ER+ breast cancer that was sensitive to CDK7 inhibitor THZ1 (7). Mechanistically, it may inhibit TNBC-related genes regulated by CDK7 mediated super-enhancers to increase the sensitivity of THZ1. Furthermore, recent results have also suggested that in HER2+ breast cancer CDK7 inhibitor may have a strong response to combined inhibition of CDK7 and HER2 (11). The result suggested that CDK7 inhibition can be an additional therapeutic option in refractory breast cancer. This may explain that although CDK7 inhibitor THZ1 was not selective for breast cancer subtypes, the sensitivity of different subtypes varies greatly. In other words, THZ1 sensitivity did not correlate with CDK7 expression levels.

Conclusions

In summary, the present study suggested that CDK7 was overexpressed in different subtypes of breast cancer and correlated with methylation of the CDK7 promoter. CDK7 could be used as a promising biomarker for predicting prognosis of breast carcinoma. However, further experiments and clinical trials are required to support the application of CDK7 selective inhibitors in different breast cancer subtypes.