Main

Epithelial ovarian cancer accounts for about 3% of total cancer cases in women. However, ovarian cancer is one of the most lethal gynecological malignancies. An estimated 21, 550 new cases were expected in the United States in 2008. An estimated 15, 520 deaths were expected in 2008.1 Ovarian cancer is a group of heterogeneous diseases and consists of different histological types, which can be readily differentiated by histological evaluation. Genome-wide global gene analysis further defines distinct expression profiles of different types of ovarian cancer.2 Different histological types of ovarian cancer seem to be regulated by different pathogenic pathways.3

HMGA2, a high-mobility-group AT-hook (HMGA) protein, is a nonhistone DNA-binding factor and binds to AT-rich sequences in the minor groove of the DNA helix. HMGA2 is expressed in embryonic tissue, but not in most adult tissues,4, 5 and is an important regulator for cell growth, differentiation, apoptosis and malignant transformation.6

HMGA2 is overexpressed in many malignant neoplasms,6 including ovarian cancer.7, 8, 9, 10 In animal models, overexpression of HMGA2 is found to be an early event in ovarian tumorigenesis.7, 9 Studies have shown that silencing of HMGA2 gene expression prevents RAS-induced transformation of thyroid cells,11 inhibits ovarian cancer growth in nude mice7 and results in growth inhibition and increased apoptosis of liposarcoma cells.12 HMGA2 may be an important tumorigenic factor associated with cancer development. The expression of HMGA2 is regulated by microRNA. Studies show that let-7s can specifically repress HMGA2 expression both in vivo and in vitro, revealing the regulatory role of let-7 in HMGA2 expression.9, 13, 14, 15, 16, 17 Downregulation of let-7s is common in ovarian cancer.16, 18

To understand the function of HMGA2 in human ovarian cancer, it is important to characterize which type of ovarian cancer is associated with HMGA2 overexpression. In this study, we examined HMGA2 expression along with HMGA1, p53 and Ki-67 in various types of ovarian cancer. Our results show that HMGA2 overexpression was significantly higher in high-grade papillary serous carcinoma than in other types of ovarian cancer. In addition, we found that let-7b, the negative regulator of HMGA2, was significantly downregulated in high-grade papillary serous carcinoma. We suggest that HMGA2 can be one of the important molecules in the tumorigenesis of high-grade papillary serous carcinoma.

Materials and methods

Case Selection

The cases were collected after surgery at Northwestern Memorial Hospital and New York University from 1996 to 2009. Approval of the institutional research board from both institutions was obtained. A total of 115 cases of ovarian cancer were collected for the study. This included histological diagnosis of high-grade papillary serous carcinoma (N=30), serous borderline tumor (N=10), malignant mixed Mullerian tumor (carcinosarcoma, N=15), endometrioid ovarian carcinoma (N=30), mucinous ovarian carcinoma (N=15) and clear cell ovarian carcinoma (N=15). Patients’ ages at surgery, tumor sizes, FIGO grades and stages and lymph node metastasis are summarized in Table 1.

Table 1 Clinical summary of the different types of ovarian carcinoma in this study
Full size table

Tissue Preparation, Antibodies and Immunohistochemistry

All cases were reviewed by two pathologists. Tissue cores were collected from tumor and control sections of each case (normal fallopian tubes for serous carcinoma, endometrium for endometrioid carcinoma) for tissue microarray. Antibodies used for this study included HMGA1 (kindly provided by Dr E Hernando), HMGA2 (BioCheck Inc., Foster City, CA, USA), Ki-67 (cell proliferation marker; Neomarkers, Fremont, CA, USA) and p53 (Dako, Carpinteria, CA, USA). Tissue microarrays were sectioned 4 μm in thickness. After deparaffinization and antigen retrieval, all immunohistochemical staining was performed on a Ventana Nexus automated system (Tucson, AZ, USA). In brief, endogenous peroxidase activity was blocked with 3% hydrogen peroxide. Primary antibodies were detected using standard biotinylated anti-mouse or anti-rabbit secondary antibodies.

let-7b and let-7c microRNA In Situ Hybridization

The hybridization system and probes of miRCURY LNA, including let-7b, let-7c and U6, were purchased from Exiqon (Vedbaek, Denmark). The detailed procedure for in situ hybridization was followed as per manufacturer's protocol.19 In brief, 4 μm tissue microarray slides were prepared. After deparaffinization and deproteinization, the slides were prehybridized with 1 × hybridization buffer without probe. The hybridization was carried out overnight in a 1 × hybridization buffer (30–70 μl) with predenatured miRCURY LNA, let-7b, let-7c or U6 probes. After washing, the slides were blocked and incubated with AP-conjugated anti-DIG Fab fragments (1:1500; Roche, Indianapolis, IN, USA) and visualized for color detection.

Immunoscores, MicroRNA Scores and Statistical Analysis

One-score system for immunointensity and microRNA intensity was used for the markers HMGA1, HMGA2 and p53, and microRNA let-7b and let-7c. The semiquantitation for intensity was scored on a scale of 0, negative; 1, weak; 2, moderate and 3, strong. Another score system for percentage of immunopositivity was used for Ki-67 that was immunoreactive for only a portion of the tumor cells. The mean values and standard errors were calculated. Paired t-test was used. The P-values <0.05 were considered as statistically significant.

Results

Expression of HMGA2 in Different Types of Ovarian Cancer

In this study, we collected a total of 115 ovarian carcinomas from 6 different histological types, including high-grade papillary serous carcinoma, carcinosarcoma, serous borderline tumor, endometrioid ovarian carcinoma, mucinous ovarian carcinoma and clear cell ovarian carcinoma. In addition, sections from normal fallopian tubes and normal endometrium were collected as normal control. The detail biomedical factors of these tumor types are summarized in Table 1.

HMGA2 was weakly and occasionally moderately expressed in ciliated cells of fallopian tube, but completely negative in secretory cells and endometrial stromal and epithelial cells (data not shown).20 In tumor sections, we defined HMGA2-positive tumors as those tumors with at least moderate immunoreactivity for HMGA2 (score ≥2).

In high-grade papillary serous carcinomas, we found 64% of tumors (18 out of 28) were moderately to strongly immunoreactive for HMGA2 (Table 2). The immunopositive cells for HMGA2 ranged from 40 to 100%. In contrast, only 10% serous borderline tumors (1 out of 10) were immunoreactive for HMGA2 (Table 2 and Figure 1).

Table 2 Differential expression of the selected markers in ovarian cancer
Full size table
Figure 1
figure 1

Photomicrographs illustrate examples of immunoreactivity for HMGA1, HMGA2, p53 and Ki-67 in six different histological types of ovarian cancer. Hematoxylin and eosin (H/E) and immunohistochemical stains for the selected markers were performed in serial sections of tissue core with low magnification ( × 4) and inset with high magnification ( × 40). PSC, high-grade serous papillary carcinoma; MMMT, carcinosarcoma; EOC, endometrioid ovarian carcinoma; SBT, serous borderline tumor; MOC, mucinous ovarian carcinoma; CCOC, clear cell ovarian carcinoma.

Full size image

In carcinosarcoma, the rate of immunoreactivity for HMGA2 was 60% of cases (9 out of 15) in carcinoma component and 47% (7 out of 15) in sarcoma (Table 2 and Figure 2). Most cases of carcinosarcoma contained carcinoma of serous differentiation. The similar rate of immunoreactivity for HMGA2 in papillary serous carcinomas and carcinoma component of carcinosarcoma suggested a similar function of HMGA2 in these two types of ovarian malignancies. Among 30 cases of endometrioid ovarian carcinomas, only 2 cases (7%) were immunoreactive for HMGA2. One of endometrioid ovarian carcinoma was strongly immunoreactive for both HMGA2 and p53 (Figure 1). Our findings indicated that endometrioid carcinoma of ovaries was mostly negative for HMGA2. About 23% of clear cell ovarian carcinomas (3 out of 13) were immunoreactive for HMGA2. Further study revealed that all three HMGA2-positive clear cell ovarian carcinomas had strong immunoreactivity for p53 (Figures 1 and 3). Of 15 mucinous ovarian carcinomas, 1 (6%) was immunoreactive for HMGA2, which was the lowest rate of HMGA2 overexpression in all ovarian carcinomas (Table 2 and Figure 3).

Figure 2
figure 2

Photomicrographs illustrate immunoreactivity for HMGA2 (a2) and p53 (a3) in carcinosarcoma. Examples showed immunoreactivity for HMGA2 and p53 were present in both epithelial and stromal components. (a1) H/E stain.

Full size image
Figure 3
figure 3

Immunoprofile analyses of HMGA2 and p53 in six different histological types of ovarian cancer. (a) Histogram (each gray bar representing one case) shows the relative levels (y axis, immunointensity) and rates (above histobars) of HMGA2 (lower) and p53 (upper) expression in six different types of ovarian cancer. The relative expression was sorted based on the HMGA2 expression from negative to the highest levels. Arrow bars separate each tumor type (indicated at bottom). HGPSC, high-grade serous papillary carcinoma; MMMT, carcinosarcoma; EOC, endometrioid carcinoma; SBT, serous borderline tumor; MOC, mucinous ovarian carcinoma; CCOC, clear cell ovarian carcinoma. (b) Correlation analysis of HMGA2 (red dots) with p53 (blue dots) in 115 cases of ovarian cancer.

Full size image

Statistical analysis indicated that the rate and levels of immunoreactivity for HMGA2 were significantly higher in high-grade papillary serous carcinomas and carcinosarcomas than in other types of ovarian cancers (P<0.05) (Table 2).

HMGA1 and p53 Expression in Different Types of Ovarian Cancer

High-mobility-group A1 gene (HMGA1) is one of the gene family members of HMGA2. HMGA1 expression in ovarian cancer is largely unknown. We examined HMGA1 expression in the serial sections of the same population of ovarian cancer. As in Table 2 and Figure 1, HMGA1 is weakly and moderately immunoreactive for all types of ovarian cancer. The percentage of HMGA1-positive cases ranged from 82% for endometrioid carcinoma to 100% for clear cell carcinoma. The average levels of HMGA1 ranged from 0.91 to 1.43 (Table 2). The levels and rates of immunoreactivity for HMGA1 were statistically insignificant among different types of ovarian cancer (P>0.05). We further examined the expression of p53 in the serial sections of all selected cases. We defined p53-positive cases as at least moderate and diffuse immunoreactivity for p53. We found that the rate of immunoreactivity for p53 varied widely among different types of ovarian cancers. High-grade papillary serous carcinomas and carcinosarcomas had the highest rate of p53 immunopositivity, 64% and 67%, respectively. All other types of ovarian cancers showed low rates of p53 immunoreactivity ranging from 13 to 28% of cases (Table 2 and Figure 3). Statistical analysis indicated that p53 was significantly higher in high-grade papillary serous carcinomas and carcinosarcomas than in other types of ovarian cancers (Table 2). In serial sections, we found that expression of p53 was moderately correlated with HMGA2 expression (r=0.51) as shown in Figure 3b.

The proliferation index (immunopositivity for Ki-67) in six different types of ovarian cancers varied widely. Overall, carcinosarcoma had the highest Ki-67 index (50% positive cells), followed by high-grade papillary serous carcinoma (39%), endometrioid ovarian carcinoma (31%) and clear cell ovarian carcinoma (27%). Mucinous ovarian carcinoma and serous borderline tumor had the lowest index at 18 and 16%, respectively. By correlation analysis, there were weak correlations between HMGA2 and p53, and HMGA2 and Ki-67 (r=0.35 and 0.21, respectively). The weak correlation of HMGA2 with Ki-67 was negatively contributed by endometrioid ovarian carcinoma, in which, only lower rate of HMGA2 and higher rate of Ki-67 were observed (Table 2).

let-7 Expression in High-Grade Papillary Serous Carcinomas and Endometrioid Ovarian Carcinomas

let-7s were found to be downregulated in ovarian cancer. To characterize which types of ovarian cancer had significant downregulation of let-7s, we selected two let-7 family members, let-7b and let-7c for the study. In contrast to HMGA2 that is only expressed in tumor cells, but not in normal counterpart, let-7s are expressed in both normal and tumor cells. Therefore, to determine the differential examination of microRNAs let-7s, matched normal controls have to be used. As shown in Materials and methods section, in ovarian cancer of high-grade papillary serous carcinoma and endometrioid ovarian carcinoma, the case-matched normal tissue controls from fallopian tube and endometrium were collected. We acknowledged that normal endometrium and fallopian tube might not be the best controls for ovarian high-grade papillary serous carcinomas and endometrioid ovarian carcinomas, but they were the closest one we could get. We examined let-7b and let-7c expression by microRNA in situ hybridization, along with U6 as internal small RNA control.

We found that there was significant downregulation of let-7b in high-grade papillary serous carcinomas (1.04±0.14) in comparison to matched fallopian tube epithelium (2.04±0.10) (P<0.05) (Figure 4). Downregulation of let-7c had a similar pattern but less significant as let-7c had lower expression level in both normal fallopian tube epithelia (1.31±0.11) and tumor cells (0.69±0.13). No significant differences of let-7 expression were noted between endometrioid ovarian carcinomas (let-7b, 1.35±0.11; let-7c, 0.77±0.16) and endometrium (let-7b, 1.45±0.11; let-7c, 0.86±0.12) (Figure 4).

Figure 4
figure 4

let-7b expression in high-grade papillary serous carcinomas (HG-PSC) and endometrioid ovarian carcinomas (EOC). (a) Relative expression of let-7b (y axis) in high-grade papillary serous carcinomas (red dot line) and matched fallopian tube epithelia (FT, blue solid line) in a total 30 cases. (b) Relative expression of let-7b (y axis) in endometrioid ovarian carcinomas (red dot line) and matched endometrium (EM, blue solid line) in a total 20 cases. (c) Photomicrographs illustrate some examples of intensity of let-7b expression in high-grade papillary serous carcinomas and FT. U6 is RNA quality control. The intensity of blue color indicates the levels of let-7b expression.

Full size image

Discussion

HMGA2 overexpression in ovarian cancer has been reported. Malek et al7 found that over 65% of ovarian serous carcinoma were positive for HMGA2, whereas all normal ovarian epithelial lined tissues were negative for HMGA2.7 Shell et al16 examined HMGA2 expression by immunohistochemistry in 100 primary ovarian cancer tissues (77% were serous type) and they found that high expression of HMGA2 significantly correlated with an adverse prognosis.

Ovarian cancer is a group of heterogeneous diseases, involving different molecular pathways. To study the function of HMGA2 in ovarian cancer, it is important to define which types of ovarian cancer are associated with HMGA2 overexpression. Kurman and Shih21 proposed a model for ovarian tumorigenesis. Ovarian tumors are divided into two groups designated as type I and type II. Type I tumors are slow growing, generally confined to the ovary at diagnosis and develop from well-established precursor lesions. Type I tumors include low-grade micropapillary serous, mucinous, endometrioid and clear cell carcinomas. They are genetically stable and are characterized by mutations in a number of different genes, including KRAS, BRAF, PTEN and β-catenin.3, 22, 23

Type II tumors are rapidly growing highly aggressive neoplasms for which precursor lesions arising from Fimbria are currently under intensive studies.24 Type II tumors include high-grade serous carcinoma, carcinosarcoma and undifferentiated carcinoma. This group of tumors has a high level of genetic instability and is characterized by mutations of p53.3 In this study, we found HMGA2 overexpression was commonly seen in type II tumors, including high-grade papillary serous carcinoma and carcinosarcoma. In addition, we found that HMGA2 overexpression is moderately correlated with p53 expression (Figure 3). We propose that HMGA2 overexpression is an important biomarker, in conjugated with p53, for the diagnosis and possibly associated with tumorigenesis of type II ovarian carcinoma.

In a recent study, we found that serous intraepithelial carcinoma in fallopian tube had high rate and level of HMGA2 overexpression along with p53-dominant mutations.20 The findings indicate that HMGA2 overexpression is an early event in the tumorigenesis of high-grade papillary serous carcinoma. Further study by characterization of functional relationship between p53 mutations and HMGA2 overexpression will be valuable for understanding the functional task of HMGA2 in the tumorigenesis of high-grade papillary serous carcinoma.

HMGA2 overexpression is present in many different epithelial and mesenchymal neoplasms. HMGA2 in regulation of the molecular pathways responsible for tumorigenesis and tumor progression were only partially understood. In the study of 60 cancer cell lines, Peter's research group found that HMGA2 is one of a few gene markers that can distinguish most type I (express mesenchymal gene signature) from type II (express epithelial gene signature) cancer cell lines.9, 16 HMGA2 participates in epithelial–mesenchymal transition,25, 26 a key step in tumor plasticity by converting adherent epithelial cells to motile mesenchymal cells27 through the functional loss of E-cadherin, which is required to maintain epithelial cell–cell adhesion.27 HMGA2 overexpression is associated with tumor growth,28 differentiation,9, 16, 28 metastasis,29 unfavorable outcome and resistance to treatment.25, 30 The function of HMGA2 in behaviors of aggressive tumor growth in ovarian high-grade papillary serous carcinomas deserves further investigation.

Members of the let-7 microRNA family function as tumor suppressors through specific repression of its target gene, particularly of HMGA2 expression in some tumor cells both in vivo and in vitro.17 The biological importance of molecular pairing of let-7::HMGA2 has been shown by repression of HMGA2 by let-7s which impairs tumor cell proliferation in many different tumor types.14, 15, 17 To study whether overexpression of HMGA2 is associated with loss of let-7 expression, in this study, we examined two family members of let-7. We found that let-7b was significantly downregulated in high-grade papillary serous carcinoma against fallopian tube epithelia (Figure 4). We proposed that loss of let-7s can be part of pathogenesis in high-grade papillary serous carcinoma.

In conclusion, our study suggests that HMGA2 overexpression is an important molecular change specifically related to high-grade papillary serous carcinomas. HMGA2 overexpression is associated with p53-dominant mutations and let-7 downregulation. In addition, HMGA2 overexpression can be a potential biomarker in conjugation with p53 for the detection and differentiation of ovarian high-grade papillary serous carcinomas.