Sage Journals: Discover world-class research

Abstract

BACKGROUND:

There has been variability between laboratories in the identification of cancer stem cells (CSCs) markers for epithelial ovarian cancer (EOC). We have evaluated three new surface markers for EOC to identify CSCs precisely.

METHODS:

Three new putative CSCs specific surface markers CD9, CD24 and EPHA1 identified by a bioinformatics approach were evaluated in normal ovary, fallopian tube and ovarian tumours.

RESULTS:

The expression of CD9 alone was observed in normal ovarian surface epithelium and fallopian tube whereas CD24 and EPHA1 were not expressed ( $n=$ 5). CD24 was expressed in all tumours ( $N=$ 101) while CD9 and EPHA1 were expressed in 89 and 71 tumours, respectively. The statistical analysis showed significant correlation of the stage of the disease ( $p<$ 0.0001), type of surgery ( $p<$ 0.0001) and residual disease ( $p<$ 0.0001) with overall survival. Although expression of CD9, CD24 and EPHA1 was observed in the majority of tumours there was no significant correlation with outcome. In patients who underwent primary surgery, increased expression of CD24 significantly correlated with poor survival. The expression of CD24 was significantly reduced ( $p<$ 0.002) upon analysis of paired sections from patients prior to surgery and at interval debulking surgery ( $n=$ 16).

CONCLUSION:

These findings suggest that overexpression of these new markers may be useful in identifying and targeting ovarian CSCs and CD24 may be a putative CSCs marker in ovarian cancer.

Keywords

Ovarian cancer stem cells serous ovarian cancer CD24 CD9 EPHA1 primary surgery interval debulking surgery clinical outcome

1. Introduction

Tumors are a heterogeneous mass of phenotypically different cells consisting of blood vessels, infiltrating macrophages, connective tissue and terminally differentiated malignant cells. A very small proportion of cells in this tumour are clonogenic, both in vitro and in vivo [1]. The development of a tumour according to the stochastic model suggests that every malignant cell in a heterogeneous tumour is capable of forming tumours. While the hierarchical model suggests that cancer stem cells (CSCs) alone are responsible for initiation and maintenance. CSCs shows similarity to normal stem cells in properties like self-renewal and differentiation [2].

Ovarian cancer is one of the leading causes of death and has the highest mortality rate among gynecologic cancers. The five-year overall survival is nearly 80% in the early stages, which declines to 30% with advanced disease. The survival rate for epithelial ovarian cancer (EOC) has remained low despite improved debulking surgery and platinum-based chemotherapy [3, 4]. Survival is still poor as majority of patients have recurrence of the disease. One of the reasons for the recurrence is drug resistance due to persistence of CSCs. These cells are usually not sensitive to conventional chemotherapy comprising of taxol and carboplatin. Although CSCs have been identified in many tumours, there has been variability between laboratories in the identity of cell surface protein that marks these cells in EOC [5, 6]. It has become increasingly apparent that a composite approach is required to identify CSCs with certainty [7].

Using publicly available microarray databases, we have identified CD9, CD24 and EPHA1 cell surface proteins as putative CSCs markers in high grade serous adenocarcinoma (Manuscript submitted). This study reports the expression of CD9, CD24 and EPHA1 in a larger cohort of ovarian tumours and their clinical significance with respect to overall survival (OS) and event free survival (EFS). To our knowledge, there are no immunohistochemical studies in EOC correlating the expression of CD9, CD24 and EPHA1 with clinical outcome.

2. Materials and methods

2.1 Patients

Patients diagnosed with EOC between January 2005 and December 2007 ( $N=$ 293) and treated at Cancer Institute (WIA), Chennai who had adequate archival paraffin material were selected for this study ( $N=$ 101). Clinical data of these patients with follow up till December 2016 was recorded. Staging of patients was performed according to the International Federation of Gynecology and Obstetrics Classification (FIGO). Primary surgery was performed in 45% of patients while the rest received neoadjuvant chemotherapy followed by IDS (Interval debulking surgery). A proportion of patients had primary surgery performed in other hospitals. The chemotherapy comprised of carboplatin and paclitaxel in all patients. The median duration of follow up of all patients was 4.07 years.

The study was approved by the Institutional Ethics Committee. Selection of patients for immunohistochemical study was performed depending upon the availability of the formalin fixed paraffin embedded (FFPE) tumours. We have also examined the FFPE sections from consecutive patients ( $N=$ 16) diagnosed with high grade serous ovarian cancer (HGSOC) at Cancer Institute (WIA) between 2014 and 2016 on the basis of availability of biopsy and IDS.

The diagnosis of EOC was established using a combination of ascitic fluid cytology, CA 125 estimation, histology and imaging (Ultrasonography or Computed axial Tomography-CT scan).

2.2 Bioinformatic analysis

A bioinformatic approach was used to identify new putative ovarian CSCs specific cell surface proteins. The analysis was based on the hypothesis that genes expressed at the embryonic stage may be expressed in cancer. Public microarray databases like Oncomine [8], TCGA [9] and Amazonia [10] were analyzed for genes commonly expressed in embryonic stem cells and ovarian cancer. Further, Human protein atlas was used to analyze the expression of these genes in normal ovary and ovarian cancer at the protein level [11]. The analysis identified 13 differentially overexpressed genes in ovarian cancer. CD24, CD9 and EPHA1 were selected on the basis of their expression and number of databases in which they were differentially overexpressed (Table S1).

2.3 Immunohistochemistry

We have evaluated the expression of CD24, CD9 and EPHA1 in sections from normal surface epithelium (OSE), fallopian tube (FT) and normal ovary (each $N=$ 5) along with tumour sections obtained from patients with EOC ( $N=$ 101) [12]. These paraffin sections were de-paraffinized by series of treatment with xylene (3 times $\times$ 5 min) followed by absolute ethanol, 90% ethanol and 70% ethanol (each 3 times $\times$ 5 min). These sections were washed in deionized water to remove residual solvent. Antigen retrieval for CD9 was performed in Citric acid (pH 6) buffer while Tris- EDTA (pH 9) buffer was used for CD24 and EPHA1 using microwave. After PBS washes (5 $\times$ 5 minutes), sections were blocked with 5% skimmed milk. Paraffin sections were incubated with respective primary antibody CD9 (Epitomics, 1:50 dilution), CD24 (Thermo Fischer, 1:25 dilution) and EPHA1 (Epitomics, 1:50 dilution) at 4 ${}^{\circ}$ C overnight followed by PBS wash. These sections were labeled with poly-HRP conjugated secondary antibodies for 1 hour at 4 ${}^{\circ}$ C. Further, sections were treated with chromogen DAB (Sigma, 1.2 mg/mL in PBS) followed by haematoxylin counterstain.

2.4 Evaluation of staining

The expression of CD9, CD24 and EPHA1 was evaluated in each section on the basis of percentage of tumour cells stained (coded as 0: no tumour cells stained, 1: 1–40% staining, 2: 40–80% staining and 3: $\geqslant$ 80%) and intensity of staining (1: 1 $+$ , 2: 2 $+$ and 3: 3 $+$ ). The score for staining was the product of multiplication of percentage of tumour cells stained and intensity of staining. The final score was coded as Negative (0), Low (1&2), Moderate (3&4) and High (6&9). The stained slides were evaluated by the pathologist who was blinded to clinical data.

2.5 Statistical analysis

Overall survival (OS) and event-free survival (EFS) were evaluated as the measures of outcome for the selected patients ( $N=$ 101). An event was defined as recurrence (local or distant), doubling of CA125 values on minimum two occasions separated by atleast one week duration. OS is defined as a period calculated from the date of diagnosis until either last date of follow up or death. EFS is defined as a period between dates of diagnosis and occurrence of any one of the events mentioned. Patient characteristics were analyzed using Cox’s regression analysis. The Kaplan-Meier method was used to analyze patient’s survival. The Cox proportional hazards model was used to evaluate the prognostic factors and to assess any impact on EFS and OS. All statistical analyses were performed using SPSS version 20. All statistical values were considered significant at $p<$ 0.05. For data without normal distribution Wilcoxon signed rank sum test was employed.

Table 1
Patients characteristics

Sr. no.	Characteristics	Number of patients ( $N=$ 101)
1. Age
	$\leqslant$ 40	73
	$\geqslant$ 40	28
2. Stage
	Stage I	17
	Stage II	08
	Stage III	63
	Stage IV	11
	Not available	02
3. Treatment
	Chemotherapy	53
	Surgery	47
	Not available	01
4. Surgery
	Primary	46
	IDS	41
	Biopsy	14
5. Residual bulk
	Optimal	66
	Sub-optimal	17
	Inoperable (Includes biopsy)	18
6. Histology
	Serous	77
	Non-serous	24
7. Grade
	I	04
	II	28
	III	68
	Not available	01
8. Recurrence
	Yes	42
	No	59
9. Survival status
	Alive	41
	Dead	60

Figure 1.

Expression of CD24, CD9 and EPHA1 in normal tissues. An approximate expression of CD24, CD9 and EPHA1 in different types of normal tissues at RNA (GTEX) and protein level derived from online database (Human Protein Atlas).

Figure 2.

Expression of CSC markers in normal ovarian surface epithelium (OSE), Fallopian tube (FT) and ovary. The immunohistochemistry of normal OSE (Top panel), fallopian tube (Middle panel) and ovary (Lower panel) sections with antibodies against CD9, CD24 and EPHA1 with their respective H and E sections. (Representative image, 400X).

Table 2

Expression of CD24, CD9 and EPHA1 in ovarian tumours

Characteristics ( $N=$ 101)	Low (89)	Moderate (12)	Negative (12)	Low (29)	Moderate (32)	High (27)	Negative (30)	Low (40)	Moderate (17)	High (14)
	CD24 ${}^{*}$		CD9				EPHA1
Age
$\leqslant$ 40 ( $n=$ 28)	25	03	01	09	08	10	08	11	04	05
$\geqslant$ 40 ( $n=$ 73)	64	09	11	20	25	17	22	29	13	09
Histology
Serous ( $n=$ 77)	12	65	10	23	22	22	25	30	12	10
Non-serous ( $n=$ 24)	24	00	02	06	11	05	04	08	03	04
Surgery
Primary ( $n=$ 46)	40	06	04	13	16	13	11	21	07	07
IDS ( $n=$ 41)	36	05	06	12	11	11	15	12	08	06
Biopsy ( $n=$ 14)	13	01	02	04	05	03	03	07	02	01
Residual bulk
Optimal ( $n=$ 67)	56	10	07	19	20	21	21	23	12	11
Sub-optimal ( $n=$ 17)	16	01	02	06	06	03	04	08	03	02
Inoperable ( $n=$ 17)	17	01	03	04	07	03	05	08	02	01
Grade
I ( $n=$ 04)	04	00	00	01	03	00	01	01	02	00
II ( $n=$ 29)	27	01	00	09	15	05	02	16	04	07
III ( $n=$ 68)	57	11	12	19	15	22	27	23	11	07
Stage
I ( $n=$ 17)	17	00	02	06	03	05	04	07	04	02
II ( $n=$ 08)	04	04	00	01	05	02	02	05	00	01
III ( $n=$ 65)	56	07	08	18	23	16	18	25	12	10
IV ( $n=$ 11)	10	01	02	04	02	03	06	03	01	01
Recurrence
Yes ( $n=$ 42)	37	05	06	11	12	13	13	13	10	06
No ( $n=$ 59)	52	07	06	18	21	14	17	27	07	08
Survival status
Alive ( $n=$ 41)	38	03	02	15	13	11	10	20	09	05
Dead ( $n=$ 60)	51	09	10	14	20	16	20	26	06	05

${}^{*}$ CD24 high and negative are absent.

3. Results

3.1 Patient characteristics

Consecutive patients who presented to the Cancer Institute (WIA) between 2005 and 2007 were selected ( $N=$ 293). Archival paraffin blocks of good quality were available from 101 patients which forms the basis of analysis. Patient’s characteristics are summarized (Table 1). The majority of patients were HGSOC (76.23%) followed by non-serous (24.76%). The non-serous tumours comprised of endometrioid, poorly differentiated, mucinous and clear cell carcinoma. These were classified as malignant in histology. The majority of patients presented with stage III/IV disease (73.26%) and with grade III tumour (67.32%) at histology. Sixty patients have died (59.40%) due to cancer while 41 patients were alive (40.59%) with median follow-up 4.07 years. All patients who were operable at presentation underwent primary surgery that included total abdominal hysterectomy, salpingo-oopherectomy, omentectomy and removal of all macroscopic disease. This was followed by chemotherapy for 6 cycles with taxol and carboplatin. In those who were inoperable, neoadjuvant chemotherapy was administered and surgery performed after 3–4 cycles.

3.2 Expression of CD9, CD24 and EPHA1

We have used normal OSE, FT and ovary (each $N=$ 5) to evaluate the expression of putative stem cell markers CD9, CD24 and EPHA1. The combined data for expression of CD9, CD24 and EPHA1 in different normal tissues at RNA (GTEX database) [13] and protein level (Human protein atlas) [11] is represented in Fig. 1.

3.2.1 Normal ovary

CD9 was expressed in both the surface epithelium of the ovary and fallopian tube. CD24 and EPHA1 were not expressed in both OSE and FT. None of the three markers were expressed in the stroma and follicles of the normal ovary (Fig. 2). For comparison, we have also evaluated previously reported ovarian CSCs markers such as CD133, CD117 and CD44. CD117 alone was expressed in FT while CD44 and CD133 were absent in all tissue sections (Data not shown).

3.2.2 Ovarian tumours

The expression of these cell surface markers was evaluated in primary ovarian tumours ( $N=$ 101). All the tumour sections showed expression of CD24. The expression of CD24 was moderate in 11.88% ( $N=$ 12) patients and low in 88.12% ( $N=$ 89) tumours. The CD9 expression was absent in 12 (11.88%), low in 29 (28.71%), moderate in 32 (31.68%) and high in 27 tumours (26.73%). EPHA1 expression was absent in 30 (29.70%), low in 40 (39.6%), moderate in 17 (16.83%) and high in 14 tumours (13.86%) (Table 2). The representative images are shown in Fig. 3.

Figure 3.

Expression of CD9, CD24 and EPHA1 in ovarian tumours. Immunohistochemistry of representative ovarian tumour sections ( $N=$ 101) with antibodies against CD9, CD24 and EPHA1 and their respective H and E sections. Arrows indicate the expression of an individual marker. (Representative image, 400X).

3.3 Simultaneous analysis of CD24, CD9 and EPHA1 expression

The co-expression of putative CSCs specific surface markers (CD24, CD9 and EPHA1) markers was evaluated in individual FFPE sections of same tumours from the patients with EOC ( $N=$ 101). In combined analysis, 89 tumours (88.11%) had expression of CD24 and CD9 (CD24 $+$ CD9), 71 (70.29%) tumours expressed CD24 and EPHA1 (CD24 $+$ EPHA1) while CD9 and EPHA1 (CD9 $+$ EPHA1) were expressed in 68 (67.32%) tumours. The comparison of sections identified set of 68 tumours which expressed all the three markers (CD24 $+$ CD9 $+$ EPHA1) when stained individually (Table 3). The combination did not show any significant correlation with OS ( $p=$ 0.303) and EFS ( $p=$ 0.649).

Table 3
Co-expression of CD24, CD9 and EPHA1 in ovarian tumours

Patient’s characteristics	CD24 $+$ CD9		CD24 $+$ EPHA1		CD9 $+$ EPHA1		CD24 $+$ CD9 $+$ EPHA1
( $N=$ 101)	Negative	Positive	Negative	Positive	Negative	Positive
	00	89	00	71	09	68	68 (67.3)
Age
$\leqslant$ 40 ( $n=$ 28)	00	27	00	20	01	20	20 (71.4)
$\geqslant$ 40 ( $n=$ 73)	00	62	00	51	08	48	48 (65.7)
Histology
Serous ( $n=$ 77)	00	67	00	52	07	49	49 (63.6)
Non-serous ( $n=$ 24)	00	22	00	19	02	19	19 (79.2)
Surgery
Primary ( $n=$ 46)	00	41	00	35	04	35	35 (76.1)
IDS ( $n=$ 41)	00	35	00	26	03	23	23 (56.1)
Biopsy ( $n=$ 14)	00	12	00	10	02	10	10 (71.4)
Residual bulk
Optimal ( $n=$ 67)	00	60	00	46	04	43	43 (64.2)
Sub-optimal ( $n=$ 17)	00	15	00	13	02	13	13 (76.5)
Inoperable ( $n=$ 17)	00	14	00	12	03	12	12 (70.6)
Grade
I ( $n=$ 04)	00	04	00	03	00	03	03 (75.0)
II ( $n=$ 29)	00	29	00	27	00	27	27 (93.1)
III ( $n=$ 68)	00	56	00	41	09	38	38 (56.0)
Stage
I ( $n=$ 17)	00	15	00	13	01	12	12 (70.6)
II ( $n=$ 08)	00	08	00	06	00	06	06 (75.0)
III ( $n=$ 65)	00	57	00	47	06	45	45 (69.2)
IV ( $n=$ 11)	00	09	00	05	02	05	05 (45.4)
Recurrence
Yes ( $n=$ 42)	00	36	00	29	05	28	28 (66.7)
No ( $n=$ 59)	00	53	00	42	04	40	40 (67.8)
Survival status
Alive ( $n=$ 41)	00	39	00	31	01	30	30 (73.2)
Dead ( $n=$ 60)	00	50	00	40	08	38	38 (63.3)

3.4 Prognostic factor analysis

The prognostic factors were evaluated in all the 101 patients with EOC by univariate and multivariate analysis. In this cohort of patients, OS and EFS did not significantly correlate with the expression of CD9 ( $p=$ 0.955 and $p=$ 0.868), CD24 ( $p=$ 0.466 and $p=$ 0.820) and EPHA1 ( $p=$ 0.114 and 0.613) (Fig. 4A). The clinical correlation with survival of these markers was also evaluated using RNA sequencing based database, GEPIA derived from TCGA analysis [14]. In silico analysis showed significant correlation between CD9 expression and OS ( $p=$ 0.026) (Fig. S1) while CD24 and EPHA1 did show any significant correlation ( $p=$ 0.77 and 0.08 respectively) (data not shown). These markers did not show any significant correlation with any histotype.

Figure 4.

Correlation of expression of surface markers with outcome. Clinical correlation for expression of CD9, CD24 and EPHA1 with OS and EFS. A. in patients with EOC ( $N=$ 101) and B. with primary surgery ( $N=$ 46) as an initial treatment using Kaplan Meier analysis. The solid line indicates patients with low or negative expression while dotted line indicates moderate to high expression of markers.

The univariate analysis in patients ( $N=$ 101) showed significant correlation between characteristics like stage ( $p\leqslant$ 0.0001 and $p=$ 0.003), histology ( $p=$ 0.012 and 0.025), type of surgery ( $p\leqslant$ 0.0001 and 0.001) and residual bulk ( $p\leqslant$ 0.0001 and 0.027) with OS and EFS respectively. Other clinical factors age ( $p=$ 0.08 and 0.073) and grade ( $p=$ 0.16 and 0.91) did not show correlation with OS and EFS, respectively (Table 4). In multivariate analysis, the stage of the disease ( $p\leqslant$ 0.0001 and $p=$ 0.003), type of surgery ( $p\leqslant$ 0.0001 and $p=$ 0.005) and residual disease ( $p\leqslant$ 0.0001 and $p=$ 0.023) were significantly correlated with OS and EFS, respectively (Table 5).

Table 4

Univariate analysis

Variable	Number	Overall survival					Event free survival
		HR	95% CI		$p$ value		HR	95% CI		$p$ value
Age
$\leqslant$ 48	51	Baseline					Baseline
$>$ 48	50	1.565	0.93	–2.6	0.	08	1.748	0.94	–3.24	0.	073
Stage
I	19	Baseline					Baseline
II	08	2.8	0.7	–11.2	0.	145	6.76	1.2	–36.9	0.	027
III	63	3.9	1.4	–10.9	0.	009 ${}^{*}$	7.04	1.6	–29.5	0.	008 ${}^{*}$
IV	11	11.5	3.5	–37.09	$<$ 0.	0001 ${}^{*}$	12.7	2.4	–66.4	0.	003 ${}^{*}$
Histology
Serous	78	Baseline					Baseline
Non-serous	23	0.38	0.18	–0.81	0.	012 ${}^{*}$	0.343	0.13	–0.87	0.	025 ${}^{*}$
Grade
I	05	Baseline					Baseline
II	28	1.4	0.18	–11.2	0.	72	1.07	0.1	–4.07	0.	92
III	68	4.03	0.55	–29.2	0.	16	3.9	0	–1.4	0.	91
Surgery
Primary	46	Baseline					Baseline
IDS	41	1.49	0.8	–2.7	0.	18	0.20	0.1		$<$ 0.	0001 ${}^{*}$
Biopsy	14	5.6	2.7	–11.5	$<$ 0.	0001 ${}^{*}$	0.33	0.41		0.	001 ${}^{*}$
Residual disease
$<$ 1 cm	66	Baseline					Baseline
$>$ 1 cm	17	2.7	1.4	–5.29	0.	002 ${}^{*}$	3.1	1.5	–6.5	0.	002 ${}^{*}$
Inoperable	18	3.6	1.9	–6.7	$<$ 0.	0001 ${}^{*}$	2.5	1.1	–5.6	0.	027 ${}^{*}$
CD24
Low	89	Baseline					Baseline
High	12	1.4	0.69	-2.86	0.	349	1.07	0.42	-2.7	0.	87
CD9
Low	41	Baseline					Baseline
High	60	0.943	0.56	–1.58	0.	825	0.945	0.5	–1.75	0.	857
EPHA1
Low	70	Baseline					Baseline
High	31	0.6	0.33	–1.09	0.	094	1.18	0.63	–2.2	0.	59

Table 5

Multivariate analysis

Variable		Number	Overall survival					Event free survival
			HR	95%CI		$P$ value		HR	95%CI		$P$ value
Model I
Stage	I	19	Baseline					Baseline
	II	08	1.96	0.45	–8.5	0.	36	8.06	1.3	–48.8	0.	02
	III	63	3.96	1.4	–11.17	0.	009 ${}^{*}$	7.15	1.7	–30.1	0.	007 ${}^{*}$
	IV	11	12.1	3.7	–40.02	$<$ 0.	0001 ${}^{*}$	12.7	2.4	–67.5	0.	003 ${}^{*}$
CD24	Low	89	Baseline					Baseline
	High	12	1.33	0.6	–2.9	0.	47	0.87	0.3	–2.4	0.	8
CD9	Low	41	Baseline					Baseline
	High	60	1.28	0.74	–2.2	0.	37	0.93	0.4	–1.8	0.	84
EPHA1	Low	70	Baseline					Baseline
	High	31	0.576	0.3	–1.08	0.	085	1.2	0.6	–2.5	0.	503
Model II
Surgery	Primary	46	Baseline					Baseline
	IDS	41	0.18	0.34	–1.50	$<$ 0.	0001 ${}^{*}$	0.22	0.11	–0.45	$<$ 0.	0001 ${}^{*}$
	Biopsy	14	0.28	0.57	–1.60	$<$ 0.	0001 ${}^{*}$	0.38	0.19	–0.75	0.	005 ${}^{*}$
CD24	Low	89	Baseline					Baseline
	High	12	0.72	0.34	–1.5	0.	397	0.78	0.37	–1.65	0.	531
CD9	Low	41	Baseline					Baseline
	High	60	0.96	0.57	–1.6	0.	896	1.00	0.59	–1.6	0.	991
EPHA1	Low	70	Baseline					Baseline
	High	31	1.71	0.92	–3.1	0.	088	0.648	0.34	–1.2	0.	171
Model III
Residual disease	$<$ 1 cm	66	Baseline					Baseline
	$>$ 1 cm	17	2.9	1.5	–5.6	0.	002 ${}^{*}$	3.2	1.5	–6.7	0.	002 ${}^{*}$
	Inoperable	18	4.1	2.1	–7.8	$<$ 0.	0001 ${}^{*}$	2.5	1.14	–5.9	0.	023 ${}^{*}$
CD24	Low	89	Baseline					Baseline
	High	12	1.5	0.74	–3.3	0.	23	1.4	0.5	–3.8	0.	48
CD9	Low	41	Baseline					Baseline
	High	60	1.07	0.63	–1.8	0.	78	1.013	0.5	–1.8	0.	96
EPHA1	Low	70	Baseline					Baseline
	High	31	0.59	0.32	–1.1	0.	09	1.302	0.67	–2.5	0.	43

3.5 Expression of CD9, CD24 and EPHA1 in sections from primary surgery

The expression of CD9, CD24 and EPHA1 was further evaluated in tissue sections from patients who underwent primary surgery ( $N=$ 46). The clinical correlation for expression of these surface markers with OS and EFS was evaluated using Kaplan-Meier analysis. Increased expression of CD24 adversely correlated ( $p=$ 0.012) only with OS in patients who had undergone primary surgery. The other CSC markers did not correlate significantly with either EFS or OS (Fig. 4B).

3.6 Comparative expression of CD24 in paired sections from biopsy and IDS

Among the three new markers, only CD24 showed significant correlation with OS in patients who underwent primary surgery. Therefore, expression of CD24 was further evaluated in paired tumour sections (a diagnostic biopsy and IDS; $N=$ 16). The second sample was obtained during IDS after treatment of patients with neoadjuvant chemotherapy for 3–4 cycles. The biopsies showed significantly higher expression of CD24 ( $p=$ 0.0025) with respect to IDS (Fig. 5A and B). The difference in expression of CD24 between presentation biopsy and following IDS was compared as follows. All these patients ( $n=$ 16) were on follow up following completion of treatment. There was recurrence of disease in 60% patients while 37.5% continued to be in remission. The probability of outcome correlated with the difference in expression of CD24 in the paired samples. The expression of CD24 was observed in IDS sections from patients with recurrence as well as complete remission. The magnitude of difference for expression of CD24 was higher in patients with recurrence on follow-up. Due to smaller size of the sample the magnitude of difference in expression was not significant ( $p=$ 0.55) between those who recurred later versus those in remission (Fig. 5C).

Figure 5.

Comparative expression of CD24 in paired sections from biopsy and IDS. A. Representative images of CD24 expression in paired sections. H and E staining (left panel), Immunohistochemical staining for CD24 expression in paired biopsy (middle panel) and interval debulking surgery (right panel). B. Expression of CD24 in paired biopsy and IDS sections from all patients ( $N=$ 16). The graph represents the mean values of CD24 expression in paired biopsy and IDS sections. C. CD24 expression in patients who developed recurrence on follow-up ( $N=$ 10) and those in complete remission ( $N=$ 6). The graph represent mean values of the difference calculated between biopsy and IDS sections obtained from patients who developed either recurrence or in complete remission on follow up. The Wilcoxon signed rank test was applied for calculation of significance.

4. Discussion

A small population of malignant cells termed as cancer stem cells (CSCs), are thought to be mainly responsible for initiation and maintenance of tumour [2]. These CSCs have been identified consistently using surface markers in different solid tumours like breast, brain, colon and melanoma. Unlike other solid tumours, there has been inconsistency in identifying specific markers for ovarian cancer. Due to this variability it has proved to be challenging to isolate CSCs in ovarian cancer [7]. We have identified new surface markers CD9, CD24 and EPHA1 using publicly available microarray databases. These were the most common genes differentially overexpressed in ovarian cancer when compared with normal ovary and embryonic stem cells [15].

Ovarian cancer is considered to originate from either the ovarian surface epithelium (OSE) or fallopian tube (FT) [16]. It is therefore possible that potential CSC markers are expressed in these locations. To address this question, we evaluated the expression of these new surface markers in normal OSE, FT and ovary sections along with primary tumors from patient specimens.

Microarray data from various laboratories has identified several genes in ovarian cancer with amplification or deletion relative to normal ovarian surface epithelial cells. Among them, CD24 has been found to be highly expressed in the epithelial ovarian cancer [17]. The expression of CD24 was observed in serous as well as endometrioid ovarian cancer subtypes [18]. It is now apparent that CD24 is expressed at high levels in many cancers [19]. CD24 appears to play an important role in metastasis via JAk-2 pathway in ovarian cancer [20]. Further, CD24 has been also shown to promote self-renewal through Nanog mediated by STAT3 phosphorylation that leads to metastasis [21]. Interestingly CD24 is a transcriptional target of hypoxia inhibitory factor and its expression increases in the presence of hypoxia [22].

CD24 is an important molecule for B-cell function and interestingly, it is one of the 30 genes that are expressed in the embryonic stem cells [23]. Ovarian cancer cells identified using CD24 as a marker, have shown stem cell properties like self-renewal, quiescence and resistance to chemotherapy. In assays CD24 $+$ cells showed higher tumorigenic capacity when compared with CD24 $-$ fraction isolated from ovarian tumours. The expression of stem cell specific transcription factors (Nestin, Oct3/4 and Notch 1/2) in CD24 $+$ cells suggests that CD24 that it is an important marker in identifying ovarian cancer stem cells [24, 25, 26, 27].

On careful scrutiny of the literature, none of the reports have examined the expression of CD24 either in primary tumours or paired samples and correlated with outcome in the context of cancer stem cells. When evaluated in selected number of patients who had undergone surgery as primary treatment ( $N=$ 46), the expression of CD24 alone correlated significantly with overall survival ( $p=$ 0.012) (Fig. 4). This observation needs to be confirmed in tumours from a larger sample of patients who undergo primary surgery.

The expression of CD24 was further evaluated in paired tumour sections (biopsy and IDS). CD24 expression was significantly higher in biopsies as compared to IDS (Fig. 5A and B). This was further reflected in a higher magnitude of difference in patients who have had a recurrence as compared to those in complete remission on follow up. The the magnitude of difference did not correlate significantly with the probability of recurrence (Fig. 5C). This is related to the quantum of CSCs at presentation. This is variable between patients. The level of expression of CSC markers in the initial biopsy or surgical specimen has been the factor that correlated with outcome in many cancers. When paired samples from primary ovarian cancer and recurrent tumours have been examined there has been enrichment for a CSC marker [28, 29].

In this study the expression of CD9 observed in high grade serous ovarian adenocarcinoma (HGSOC) was similar to that obtained from Human Protein Atlas at the RNA level. The expression of CD9 is an adverse prognostic factor in different tumours. However, in gastro-intestinal stromal tumour, over expression was correlated with better survival [30]. In an effort to identify new surface markers in acute lymphoblastic leukemia other than CD34, cells expressing CD9 showed stem cell properties both in vitro and in vivo serial transplantation assays [31]. This is the first report on CD9 as a putative ovarian cancer stem cell marker. CD9 alone was expressed in OSE while FT showed the expression of both CD9 and CD117. We have shown that these cell surface markers were differentially overexpressed in primary tumours when evaluated in a larger cohort of patient specimens (Figs 2 and 3). In contrast to normal ovary, there was higher expression of CD9 (59.40 %) in tumour cells (Table 3) but there was no significant correlation observed with overall survival ( $p=$ 0.33). However in TCGA database, at RNA level higher expression of CD9 was an adverse prognostic factor for survival (HGSOC, $N=$ 426) (Fig. S1). The difference between our result and that of the TCGA data is most likely due to sample size.

EPHA1 receptor was first isolated from an Erythropoietin producing hepatoma cell line [32]. The increased expression of EPHA1 was observed in advanced ovarian tumours with respect to normal controls [33]. The Eph family plays an important role in signaling of cancer cells and co-expression of ephrin ligands and Eph receptors may be responsible for progression [34]. The altered expression of these receptors contributes to migration and invasion of cancer cells in vitro [35]. High expression of EPHA1 correlated with aggressive tumour type and could be useful for identifying low grade tumours from high grade [36]. We observed low expression of EPHA1 in majority of stage III/IV (68.91 %), high grade (73.52%) and serous sub-type (71.42%) of epithelial ovarian cancer while high EPHA1 expression was observed in only 31 (30.69%) tumours (Table 3). Similarly, reduced expression of EPHA1 correlated with poor clinical outcome in colorectal cancer [37]. Our study is the first to report EPHA1 as a putative cancer stem cell marker in ovarian cancer based on expression of this receptor.

In general, the identification of cell surface markers that target CSCs has to be evaluated in primary tumours. Although, there are well established approaches to prove that a marker targets CSCs in vitro and in vivo, it is more challenging to demonstrate its significance in primary tumours [7]. First, the evaluation of the expression of such a marker in primary tumours by immunohistochemistry is important. However, as this is only semi-quantitative it is difficult to be sure of the proportion of malignant cells that express a particular marker. The evaluation of any CSC marker in primary tumours for expression and correlation with outcome does not definitively prove the clinical relevance. Perhaps the best way is to examine tissue sections from diagnostic biopsies and surgery at recurrence for level of expression of these markers. If indeed, persistence of CSCs leads to recurrence after initial treatment, then the second biopsy should be enriched for the marker [29]. This approach can be evaluated thoroughly for all the tumour types. Ultimately, only by developing treatments that target CSCs and improve survival can the hypothesis be proven.

In conclusion, this report has identified and evaluated 3 new potential CSC markers in ovarian cancer. This data sets the tone for a more detailed evaluation of these markers to delineate their role in the pathogenesis of ovarian cancer.

Footnotes

Acknowledgments

The work is supported by Indian Council for Medical Research (ICMR), Department of Biotechnology (DBT) and University Grant Commission (UGC), Government of India. We are also thankful to Departments of Pathology, Surgical Oncology and Epidemiology, Biostatistics and Cancer Registry.

Conflict of interest

Authors declare no conflict of interest.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-201463.

References

Reya

Morrison

S.J.

Clarke

M.F.

et al., Stem cells, cancer, and cancer stem cells, Nature 414 (2001), 105–111.

Dick

J.E.

, Looking ahead in cancer stem cell research, Nature Biotechnology 27 (2009), 44–46.

Aletti

G.D.

and Hartmann

L.C.

, Ovarian cancer care: It’s time for “personalized” approaches, Oncology 24 (2010), 728–736.

Shan

and Liu

, Epithelial ovarian cancer: focus on genetics and animal models, Cell Cycle 8 (2009), 731–735.

Zhang

Balch

Chan

M.W.

et al., Identification and characterization of ovarian cancer-initiating cells from primary human tumors, Cancer Research 68 (2008), 4311–4320.

Zhang

Cui

Lai

et al., Ovarian cancer stem cells express ROR1, which can be targeted for anti-cancer-stem-cell therapy, Proceedings of the National Academy of Sciences 111 (2014), 17266–17271.

Nagare

R.P.

Sneha

Priya

S.K.

et al., Cancer stem cells – are surface markers alone sufficient, Current Stem Cell Research & Therapy 12 (2017), 37–44.

Rhodes

D.R.

Shanker

et al., ONCOMINE: A cancer microarray database and integrated data-mining platform, Neoplasia 6 (2004), 1–6.

Weinstein

J.N.

Collisson

E.A.

Mills

G.B.

et al., The cancer genome atlas pan-cancer analysis project, Nature Genetics 45 (2013), 1113–1120.

10.

Carrour

T.L.

Lhermitte

Lamb

Reme

Pantesco

Hamamah

Klein

and De Vos

, Amazonia!: An online resource to google and visualize public human whole genome expression data, The Open Bioinformatics Journal (2010), 5–10.

11.

Uhlen

Bjorling

Agaton

et al., A human protein atlas for normal and cancer tissues based on antibody proteomics, Molecular & Cellular Proteomics 4 (2005), 1920–1932.

12.

Priya

S.K.

Kumar

Hiran

K.R.

et al., Expression of a novel endothelial marker, C-type lectin 14A, in epithelial ovarian cancer and its prognostic significance, International Journal of Clinical Oncology 22 (2017), 107–117.

13.

Lonsdale

Thomas

Salvatore

et al., The genotype-tissue expression (GTEx) project, Nature Genetics 45 (2013), 580–585.

14.

Tang

Kang

et al., GEPIA: A web server for cancer and normal gene expression profiling and interactive analyses, Nucleic Acids Research 45 (2017), W98–W102.

15.

Nagare

R.P.

Sneha

Krishnapriya

et al., Abstract 1392: ALDH1A1 is the best marker by a composite approach for cancer stem cells in ovarian cancer whose integrity is controlled by the hedgehog pathway, Cancer Research 75 (2015), 1392–1392.

16.

Dubeau

, The cell of origin of ovarian epithelial tumours, The Lancet Oncology 9 (2008), 1191–1197.

17.

Bauerschlag

Brautigam

Moll

et al., Systematic analysis and validation of differential gene expression in ovarian serous adenocarcinomas and normal ovary, Journal of Cancer Research and Clinical Oncology 139 (2013), 347–355.

18.

Kristiansen

Denkert

Schluns

et al., CD24 is expressed in ovarian cancer and is a new independent prognostic marker of patient survival, Am J Pathol 161 (2002), 1215–1221.

19.

Emmanuel

Gava

Kennedy

et al., Comparison of expression profiles in ovarian epithelium in vivo and ovarian cancer identifies novel candidate genes involved in disease pathogenesis, PLoS One 6 (2011), e17617.

20.

Burgos-Ojeda

McLean

et al., CD24+ ovarian cancer cells are enriched for cancer-initiating cells and dependent on JAK2 signaling for growth and metastasis, Molecular Cancer Therapeutics 14 (2015), 1717–1727.

21.

Bretz

N.P.

Salnikov

A.V.

Perne

et al., CD24 controls Src/STAT3 activity in human tumors, Cellular and Molecular Life Sciences 69 (2012), 3863–3879.

22.

Thomas

Harding

M.A.

Smith

S.C.

et al., CD24 is an effector of HIF-1-driven primary tumor growth and metastasis, Cancer Research 72 (2012), 5600–5612.

23.

Assou

Le Carrour

Tondeur

et al., A meta-analysis of human embryonic stem cells transcriptome integrated into a web-based expression atlas, Stem Cells 25 (2007), 961–973.

24.

Davidson

, CD24 is highly useful in differentiating high-grade serous carcinoma from benign and malignant mesothelial cells, Human Pathology 58 (2016), 123–127.

25.

Zhu

Zhang

and Lu

, CD24, COX-2, and p53 in epithelial ovarian cancer and its clinical significance, Frontiers in Biosciences 4 (2012), 2745–2751.

26.

Kang

Choi

Gao

M.-Q.

et al., CD24 ovary cancer cells exhibit an invasive mesenchymal phenotype, Biochemical and Biophysical Research Communications 432 (2013), 333–338.

27.

Gao

M.Q.

Choi

Y.P.

Kang

et al., CD24+ cells from hierarchically organized ovarian cancer are enriched in cancer stem cells, Oncogene 29 (2010), 2672–2680.

28.

Steg

A.D.

Bevis

K.S.

Katre

A.A.

et al., Stem cell pathways contribute to clinical chemoresistance in ovarian cancer, Clinical Cancer Research 18 (2012), 869–881.

29.

Ayub

T.H.

Keyver-Paik

M.D.

Debald

et al., Accumulation of ALDH1-positive cells after neoadjuvant chemotherapy predicts treatment resistance and prognosticates poor outcome in ovarian cancer, Oncotarget 6 (2015), 16437–16448.

30.

Yang

Shen

Zhang

et al., Expression and clinicopathological significance of CD9 in gastrointestinal stromal tumor, Journal of Korean Medical Sciences 28 (2013), 1443–1448.

31.

Yamazaki

C.W.

Naito

et al., Regulation of cancer stem cell properties by CD9 in human B-acute lymphoblastic leukemia, Biochemical and Biophysical Research Communications 409 (2011), 14–21.

32.

Dodelet

V.C.

and Pasquale

E.B.

, Eph receptors and ephrin ligands: Embryogenesis to tumorigenesis, Oncogene 19 (2000), 5614–5619.

33.

Herath

N.I.

Spanevello

M.D.

Sabesan

et al., Over-expression of Eph and ephrin genes in advanced ovarian cancer: Ephrin gene expression correlates with shortened survival, BMC Cancer 6 (2006), 144.

34.

Pasquale

E.B.

, Eph receptors and ephrins in cancer: Bidirectional signalling and beyond, Nature Reviews Cancer 10 (2010), 165–180.

35.

Cui

B.O.

Flamini

et al., Knockdown of EPHA1 using CRISPR/CAS9 suppresses aggressive properties of ovarian cancer cells, Anticancer Res 37 (2017), 4415–4424.

36.

Jin

Zou

Wan

et al., Decreased Eph receptorA1 expression is related to grade in ovarian serous carcinoma, Molecular Medicine Reports 17 (2018), 5409–5415.

37.

Dong

Wang

Sheng

et al., Downregulation of EphA1 in colorectal carcinomas correlates with invasion and metastasis, Modern Pathology 22 (2009), 151–160.

Expression of cancer stem cell markers CD24,EPHA1 and CD9 and their correlation with clinical outcome in epithelial ovarian tumours

Abstract

BACKGROUND:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Materials and methods

2.1 Patients

2.2 Bioinformatic analysis

2.3 Immunohistochemistry

2.4 Evaluation of staining

2.5 Statistical analysis

Table 1 Patients characteristics

3.1 Patient characteristics

3.2 Expression of CD9, CD24 and EPHA1

3.2.1 Normal ovary

3.2.2 Ovarian tumours

Table 3 Co-expression of CD24, CD9 and EPHA1 in ovarian tumours

3.6 Comparative expression of CD24 in paired sections from biopsy and IDS

Footnotes

Acknowledgments

Conflict of interest

Supplementary data

References

Table 1
Patients characteristics

Table 3
Co-expression of CD24, CD9 and EPHA1 in ovarian tumours