Prognostic and predictive values of Nrf2, Keap1, p16 and E-cadherin expression in ovarian epithelial carcinoma

Abstract

Objectives: Despite considerable interest in the Nuclear factor-erythroid 2-related factor 2 (Nrf2)/Kelch-like ECH-associated protein-1 (Keap1), p16 and epithelial cadherin (E-cadherin) activation in carcinoma progression, contradictory results regarding association of Nrf2/Keap1/E-cadherin and p16 expression with clinico-pathological features and prognosis have been reported. The predictive value of these markers in ovarian carcinoma is unknown. Methods/Materials: In this retrospective study, 108 cases were evaluated immunohistochemically with antibodies to Nrf2, Keap1, estrogen receptor (ER), p16 and E-cadherin. The results were compared with histological and clinical data, disease-free survival (DFS) and overall survival (OS). Results: A cohort of 108 ovarian carcinomas (47 serous, 23 mucinous, 13 endometrioid and 25 clear cell), including 68 FIGO stage I-II cases and 40 FIGO stage III-IV cases was studied. The age of patients (P=0.005), FIGO stage (P<0.001), immunohistochemical expression of Keap1 (P<0.000), E-cadherin (P=0.045), p53 (P=0.003), p16 (P<0.001) and ER (P=0.004) were significant factors between different histological subtypes. Patients with serous carcinoma were older in age, presented with more advanced stage disease, worst prognosis, highest Keap1 expression and least percentage of E-cadherin immunoreactivity. In univariate analysis, FIGO staging (P=0.000 for DFS; P=0.000 for OS), Nrf2 (P=0.010 for DFS; P=0.001 for OS), and p16 (P=0.004 for DFS; P=0.019 for OS) were associated with worse prognosis. After multivariate analysis, FIGO staging and Nrf2 remained significance prognostic factors. Conclusions: There were differences in the expression of Nrf2, Keap1, p16 and E-cadherin between different ovarian carcinoma subtypes. In multivariate analysis, FIGO stage and Nrf2 expression were associated with poorer DFS and OS.

Introduction

Ovarian cancer is the main cause of death from gynecological malignancies in the developed and developing countries [1]. The prognosis is poor and over 70% of the patients diagnosed with ovarian cancer have advanced stage of disease at the time of diagnosis. Stage of disease, age at diagnosis, histological subtype and tumor grade, and effect of chemotherapy after primary surgery are important prognostic factors.

Nrf2 (nuclear factor erythroid-derived 2-like 2) is a transcription factor that control the genes expression of antioxidant enzymes, metal-binding proteins, and drug-metabolizing enzymes [2]. Nrf2 is thought to be a double-edged sword in cancer biology which has a dual function [3]. Dysfunction of Nrf2 function may be involved in the cancer development. In cases of already established cancer, Nrf2 may protect tumor cells from the oxidative damage induced by chemotherapeutic drugs and radiation, which may account for therapy resistance in cancer. Under normal physiological levels, Keap1 (Kelch-like ECH-associated protein 1) binds in a complex with Nrf2 and orchestrates Nrf2 ubiquitination [4]. When oxidative stress condition occurs, Nrf2 is released from a complex formed with Keap1 and moves from cytoplasm to nucleus of the cell [5-7]. This process up regulates expression of antioxidant genes by binding to the antioxidant response element (ARE).

E-cadherin is a transmembrane protein that involves in the establishment of cell-cell adhesion and motility of epithelial cells [8-10]. Oxidative stress disrupts the cell-cell adhesion complex, which might also affect Nrf2 activity. Thus, loss of E-cadherin expression might be associated with the resistance of cancer cells to chemotherapy, tumor invasion and distant metastasis [11,12]. In addition, Keap1 has the ability to bind actin filaments as a scaffold protein in the cytoplasm under normal physiology conditions. Since E-cadherin regulates actin dynamic, it is proposed that the activity of Nrf2/Keap1 might be affected by E-cadherin/actin binding ability [13]. However, the relationship between E-cadherin and Nrf2/Keap1 signaling pathway in ovarian carcinomas has not been examined.

It is well established that p16 contributes to the regulation of cell cycle progression by inhibiting the S phase. In malignant tumors, p16 overexpression appears to be a mechanism to arrest the uncontrolled proliferation caused by failure of the Rb pathway (secondary to viral infection, mutational silencing of the Rb gene, or other mechanisms) [14]. Roles of p16 in senescence, as a tumor suppressor, as a prognostic marker and overexpression in ovarian carcinoma remain poorly investigated.

In this study, we evaluated the immunohistochemical expression and correlation between Nrf2, Keap1, p16 and E-cadherin in histologically different ovarian carcinomas by tissue microarray (TMA). In addition, we aimed to identify prognostic marker of ovarian carcinomas associated with clinical outcome.

Materials and methods

Case selection

From 1998 to 2011, the tumors studied were retrieved from the files of Departments of Pathology (Wan Fang Hospital and Taipei Medical University Hospital). Tumor tissues were histologically diagnosed and classified using the guidelines of the World Health Organization classification system [1]. The grading and staging of tumors were assigned according to the system of the International Federation of Gynecology and Obstetrics (FIGO). Review of diagnostic material by at least 2 gynecologic pathologists to confirm diagnosis. Further immunohistochemical stains (such as p16, estrogen receptor and p53) are needed if controversial cases present. Tissue microarrays (TMA), where applicable, will be created from archival paraffin embedded tissue. One to four tumor regions were chosen from each sample, which were incorporated into TMA blocks. This study was approved by the institutional review board and informed consent has been obtained.

Immunohistochemistry

The primary antibodies used were Nrf2 (courtesy of Professor Huang of Academia Sinica; 1: 100), Keap1 (clone: 1G2; Origene; 1: 100), E-cadherin (clone: NCH-38; DAKO Carpinteria CA; 1: 100), p53 (clone: DO-7; Ventena; prediluted), p16 (clone: E6H4; Ventena; prediluted) and estrogen receptor (clone: 6F11; Novocastra; prediluted). Immunohistochemical staining was performed using an automated stainer (Ventana, BenchMark XT) with 4 μm-thick sections from formalin-fixed and paraffin-embedded tissues, Tissue sections were deparaffinised and rehydrated in graded alcohols and xylene. After retrieval by autoclaved retrieval technique (10 mM citric acid buffer; 10-20 minutes) and inhibited by endogenous peroxidase activity (0.3% H₂O₂; 5 minutes), sections were incubated with primary antibodies. Negative controls were performed by omitting primary antibodies. After washing with Tris buffer, sections were incubated for 30 min with secondary antibody (dilution rate 1:100), re-incubated with 100 μg/mL peroxidase-conjugated streptavidin, and colonized with 0.02% 3,3’-diaminobenzidine tetrahydrochloride (DAB) (0.05 M tris-HCL buffer with 0.03% H₂O₂). The sections were counterstained with hematoxylin, dehydrated, and mounted.

Immunohistochemical expression was quantified by two pathologists (P-L Liew and C-L Chen). Nuclear Nrf2 and cytoplasmic Keap1 staining patterns were quantitatively scored as percentage. Negative Nrf2 and Keap1 immunoexpression were defined as less than 50% positive tumor cells, and positive Nrf2 and Keap1 immunoexpression were defined 50% or more than 50% positive tumor cells. With respect to E-cadherin, cell membrane throughout the epithelium immunoreactivity was also scored as percentage. The cut-off level for E-cadherin positivity was cell membrane staining in more than 10% of tumor cells. ER status determination followed current College of American Pathologists (CAP) guideline. The cut-off level for ER positivity was nuclear staining in more than 1% of tumor cells. The staining results of immunohistochemistry p53, p16 and ER were scored semi quantitatively according to the percentages of positive cells: score 0 (0-1%), score 1 (2-24%), score 2 (25-75%), and score 3 (more than 75%). Score 0 and score 3 of p53 were considered mutant type; score 1 and score 2 of p53 were regarded as no significant.

Statistical analysis

The statistical analyses were performed with SPSS for Windows software (SPSS, Chicago, IL). Data were expressed as median (interquartile range), mean ± SD, and percentages. Statistical analyses were performed using the Wilcoxon rank sum Test, Chi-square test, Mann-Whitney U tests, and Student’s t-test. Multivariate logistic regression analysis was done on factors that were significant in the univariate analysis. Additionally, Spearman rank correlation and multivariate linear regressions with stepwise variable selection were used to assess the significance of associations between ordinal or continuous predictor variables. Survival curves were estimated using the Kaplan-Meier method and Cox regression. A P-value of less than 0.05 was considered statistically significance.

Results

The clinical data, clinicopathological factors and results of immunohistochemical stains according to different histology subtypes of carcinoma are listed in Table 1. Our group of carcinomas comprised 47 serous, 23 mucinous, 13 endometrioid and 25 clear cell. Patients’ ages ranged from 25 to 93 years (mean, 54.3 years; median, 53 years). The carcinomas comprised 68 early stage cases (FIGO stage I-II) and 40 advanced stage cases (FIGO stage III-IV), with a median follow-up time of 41.0 months (mean, 49.6 months; range, 1 to 148 months). The standard surgical procedures included total hysterectomy, bilateral salpingo-oophorectomy, omentectomy and lymph node sampling. In total, 38 patients (35.2%) completed Cisplatin-based chemotherapy.

Table 1.

Baseline characteristics of 108 cases with ovarian carcinomas

	Serous carcinoma	Mucinous carcinoma	Endometrioid carcinoma	Clear cell carcinoma	P-value
Case (No. Patients)	47	23	13	25
Age (year)	58.1±12.8	45.8±16.8	53.0±11.6	55.8±10.1	0.005*
FIGO Staging (No. Patients)					<0.001*
I+II	16	20	10	22
III+IV	31	3	3	3
DFS (months)	37.5±33.1	52.2±50.1	52.8±40.5	52.5±39.5	0.289
OS (months)	40.5±32.6	53.2±49.6	57.2±39.6	59.5±41.8	0.205
Nrf2 (%)	73.3±37.4	65.9±34.9	84.5±27.9	67.6±39.6	0.459
Keap1 (%)	57.6±37.8	23.7±30.9	50.0±36.9	17.0±27.5	<0.001*
E-cadherin (%)	63.0±34.0	70.6±33.4	86.7±25.4	79.6±20.8	0.045*
p53 (score)	1(3)	1(1)	1(0)	1(0)	0.003*
p16 (score)	3(2)	0(0)	1(3)	1(3)	<0.001*
ER (score)	0(1)	0(0)	0(3)	0(0)	0.004*

NOTE: For all laboratory measures and continuous variables: mean±SD, P value Kruskal-Wallis Test. Proportions: percentage, P value Chi-square test analysis.

^*Significant difference.

The age of patients (P=0.005), FIGO stage (P<0.001), immunohistochemical expression of Keap1 (P<0.000), E-cadherin (P=0.045), p53 (P=0.003), p16 (P<0.001) and estrogen receptor (P=0.004) were significantly different between serous carcinoma, mucinous carcinoma, endometrioid carcinoma and clear cell carcinoma. Patients with serous carcinoma were older in age, presented with more advanced stage disease and worst overall and disease-free survival. Besides, increased immunoreactivity of Nrf2 (Figure 1A), highest percentage of immunoexpression of Keap1 (Figure 1B) and less percentage of E-cadherin staining (Figure 1C) were also noticed in this particular histological subtype.

Figure 1.

Ovarian serous carcinoma with corresponding (A), Nrf2 nuclear staining (original magnification × 100) (B), Keap1 cytoplasmic staining (original magnification × 100) and (C), E-cadherin membranous staining (original magnification × 200).

The prognostic effects of clinicopathological factors were summarized in Tables 2 and 3. FIGO staging was the most important prognostic predictor for ovarian carcinoma (P=0.000 for DFS and P=0.000 for OS). Immunohistochemical expression of Nrf2 (P=0.010 for DFS and P=0.001 for OS) p16 (P=0.004 for DFS and P=0.019 for OS) also had impact on DFS and OS. After multivariate analysis, FIGO staging and Nrf2 remained significance prognostic factors. Other factors including age, different histology subtypes, immunoexpression of Keap1, E-cadherin and ER were not significant prognostic factors.

Table 2.

Univariate analyses showing HRs for patient OS and DFS conferred age, FIGO stage, histological subtypes, p53, p16, ER, Nrf2, Keap1 and E-cadherin expression (n=108)

Variable	Disease-free survival			Overall survival
	HR	95% CI	P-value	HR	95% CI	P-value
Age			0.677			0.976
≤50 years	1.00#			1.00#
>50 years	1.090	0.705-1.640		1.006	0.670-1.512
FIGO staging			0.000*			0.000*
I	1.00#			1.00#
II	1.058	0.584-1.917		1.177	0.649-2.135
III	2.487	1.560-3.966		2.572	1.609-4.127
IV	7.692	2.266-26.118		10.292	2.970-35.665
Tumor subtype			0.137			0.092
Serous	1.00#			1.00#
Mucinous	0.554	0.321-0.957		0.560	0.324-0.967
Endometrioid	0.690	0.370-10287		0.680	0.365-1.266
Clear cell	0.668	0.405-1.099		0.592	0.358-0.979
Nrf2 (%)			0.010*			0.001*
Negative	1.00#			1.00#
Positive	1.972	1.178-3.300		2.386	1.405-4.054
Keap1 (%)			0.283			0.125
Negative	1.00#			1.00#
Positive	1.248	0.833-1.868		1.375	0.915-2.065
E-cadherin (%)			0.816			0.691
Negative	1.00#			1.00#
Positive	0.920	0.457-1.852		1.151	0.575-2.307
p53			0.148			0.245
Score 1 and 2	1.00#			1.00#
Score 0 and 3	1.339	0.901-1.989		1.265	0.851-1.880
p16			0.004*			0.019*
Negative	1.00#			1.00#
Positive	1.926	1.236-3.002		1.656	1.0895-2.527
ER			0.834			0.536
Negative	1.00#			1.00#
Positive	1.025	0.809-1.300		1.077	0.851-1.363

^#Reference category for HR (Hazard Ratio) calculation with variable;

CI: Confidence Interval.

^*Significant difference.

Table 3.

Multivariate analyses showing HRs for patient OS and DFS conferred FIGO stage, Nrf2 expression (n=108)

Variable	Disease-free survival			Overall survival
	HR	95% CI	P-value	HR	95% CI	P-value
FIGO staging
I	1.00#			1.00#
II	0.866	0.470-1.595	0.644	0.908	0.492-1.677	0.759
III	2.443	1.535-3.888	0.000*	2.493	1.559-3.985	0.000*
IV	6.487	1.906-22.070	0.003*	8.403	2.422-29.153	0.001*
Nrf2 (%)			0.006*			0.001*
Negative	1.00#			1.00#
Positive	2.084	1.229-3.536		2.487	1.443-4.286
p16			0.081			0.307
Negative	1.00#			1.00#
Positive	1.578	0.945-2.636		1.283	0.795-2.071

^#Reference category for HR (Hazard Ratio) calculation with variable;

CI: Confidence Interval.

^*Significant difference.

Discussion

The utility of new biomarkers could promote a more consistent method to identify the proliferation potential of a tumor and understanding the behavior of tumor biology. Previous studies have demonstrated the predictive values of Nrf2, Keap1 and E-cadherin in different neoplasms [15-20]. This study investigated the expression of Nrf2, Keap1, p16 and E-cadherin in a large set of ovarian carcinoma. Our aim was to investigate whether changes in expression of these markers would influence the behavior or clinical outcomes of ovarian carcinoma.

Nrf2 is important in maintaining cellular homeostasis (oxidative stress) and contributes to diverse cellular functions such as differentiation, proliferation and inflammation. Nrf2 expression was associated with poorer clinical outcome in non-small cell lung cancer (mainly squamous cell carcinoma) [17,19], gallbladder carcinomas [18] and squamous cell carcinoma from the esophagus, skin, pulmonary and head/neck [20]. Jiang T et al. [21] reported that Nrf2 was not detected in non-malignant endometrial lesions and it was marginally expressed in type I endometrial tumors; however, a large percentage of type II endometrial tumors had a high level of Nrf2. Therefore, Nrf2 may play a role in the aggressiveness, resistance, and poor prognosis, suggesting that combined therapy with an Nrf2 inhibitor may represent a new avenue in cancer treatment [22-26]. In our study, when comparing different histology subtypes of epithelial ovarian cancers, patients with serous carcinoma were older in age, had advanced FIGO stage, poorer overall and disease-free survivals. Serous carcinoma had a highest expression of cytoplasmic Keap1 positivity, increased percentage of nuclear Nrf2 immunoreactivity and lowest E-cadherin membrane positivity than other carcinomas. These results were consistent with a previous study suggested that nuclear expression of Nrf2 in malignant ovarian cancer cells may play a role in resistance to Cisplatin-based treatment in serous carcinoma [22]. However, when comparing different histology subtypes of ovarian carcinoma, our result was differed from previous study [27] that Nrf2 target genes were upregulated in tumors with nuclear immunopositivity for Nrf2 in the clear cell epithelial ovarian carcinoma subtype. More studies are needed to confirm the independent association of Nrf2 with Keap1 in ovarian epithelial carcinoma.

Our study also observed highest Keap1 cytoplasmic immunoexpression in serous carcinoma compared to the other tumors. The activity of Nrf2 is primarily regulated via its interaction with Keap1 for proteasomal degradation. However, alternative Keap1-independent mechanisms of Nrf2 regulation including phosphorylation of Nrf2 by various protein kinases, interaction with other protein partners and epigenetic factors are also discovered. Keap1 genetic mutations was found in 29% of clear cell ovarian carcinomas and 8% of nonclear cell tumors, which suggested a previously unrecognized role for Keap1-Nrf2 pathway in mediating chemotherapeutic responses [23]. However, another study [28] in endometrial carcinoma failed to corroborate an association between clinical findings and mutations that disrupt Nrf2-Keap1 interaction. This finding was similar to a previous study in non-small cell lung cancers [19]. Therefore, the prognostic relevance of Keap1 or Nrf2 genetic mutations in ovarian epithelial carcinomas may differ from other cancers. Further investigation of other factors that may affect the transcription and translation of either Keap1 or Nrf2 are needed.

E-cadherin is important in the tumor cell growth and transdifferentiation. It is a useful prognostic marker for ovarian clear cell carcinoma patients, and downexpression of E-cadherin carried a worse prognosis [29]. E-cadherin may have an inhibitory effect on Nrf2 and form a quaternary complex with Nrf2, Keap1 and β-catenin that contribute to Keap1-dependent ubiquitylation of Nrf2 [13]. Our data might imply association of E-cadherin and Nrf2-Keap1 in promoting tumor proliferation and metastasis.

In our study, we also found increased immunoexpression of p16 related to serous carcinoma. Surprisingly, p16 immunoexpression had impact on DFS and OS in univariate analysis although its significance disappeared after multivariate analysis. Increased expression of p16 in high-grade ovarian serous carcinoma compared with low-grade ovarian serous carcinoma and serous borderline tumors was first reported in 2007 [30]. Striking p16 homogeneous expression pattern, frequently used as a marker of RB1 loss, seen in approximately 66% of high-grade serous carcinoma and associated with a decreased progression free survival and overall survival [31]. Epigenetic alterations of p16 and RAR-β also had an important role in ovarian carcinogenesis and p16 methylation may play a significant role in downregulation of RAR-β gene in ovarian cancer [32]. As a senescence-associated marker, p16 associated BRAF mutation (known as “oncogene-induced senescence” phenotype) may impede progression of ovarian serous borderline tumors to low-grade serous carcinoma in a recent study [33]. Many aspects of p16 function and regulation are still partially unresolved.

Briefly, FIGO staging and Nrf2 immunoexpression were the most important prognostic factors in our study. Our study also provided a potential target for platinum-based first-line chemotherapy resistant in ovarian epithelial cancers. However, our study has few limitations. We used tissue microarrays rather than whole tissue sections for immunohistochemistry in this study. Heterogeneity of nuclear Nrf2, cytoplasmic Keap1 and membranous E-cadherin expression in tumor tissues between whole histological sections and TMA may exist. Second, we did not evaluate Keap1 and Nrf2 genetic mutations and polymorphisms from our formalin-fixed cancer tissue specimens. More studies have to be conducted to clarify the molecular changes for this relationship for a better understanding of the disease pathogenesis, novel prognostic factors and therapeutic strategy.

In conclusion, our results revealed variable expression of Nrf2, Keap1, and E-cadherin in different histological ovarian carcinomas. The importance of Nrf2-Keap1 mechanism in ovarian carcinomas progression is reflected in the association of Nrf2 expression with poorer survival, which appeared as independent prognostic factor.

Disclosure of conflict of interest

None.