The ovary is a special organ that organizes menstruation, hormonal balance, bone metabolism and fertilization. Periodically, the adult ovarian stroma demonstrates active tissue remodeling during and after ovulation. In the women of reproductive ages and perimenopausal period, several pathologic events cause unfavorable conditions such as metrorrhagia, sterility and symptoms associated with climacteric disorders. Ectopic endometrial tissues in the ovary bleed periodically, causing longstanding inflammation, i.e., ovarian endometriosis. A series of investigations on ovarian neoplasms have improved our understanding of proinflammatory microenvironment including unfavorable cytokines, chemokines and imbalanced hormone production. Compared with the tumors of other organs, ovarian neoplasms are composed of more heterogeneous histologic types; not only epithelial tumors but also sex-cord stromal tumors and germ cell tumors frequently develop in this organ. Epithelial cells are lined only in monolayer on the surface, and they are exposed to mechanical stress such as periodical ovulation. The majority of ovarian parenchyma is composed of dense fibrotic stroma and follicles of various stages. Under physiologic hormonal milieu, follicular development, ovulation and luteinization are finely controlled. Therefore, stromal cells of adult ovary actively contribute to the synthesis and remodeling of extracellular matrix (ECM) and blood vessels.

Ovarian cancer represents one of the most common malignant conditions in adult women [1,2]. The macroscopic morphology of ovarian cancer is in most cases multi-cystic, regardless of histologic subtypes. On the other hand, the pathogenesis of ovarian cancers differs in histologic types, genetic backgrounds and pre-cancerous inflammatory conditions. Once the cysts rupture/leak and cancer cells are spread directly into the peritoneal cavity, disseminated cancer cells easily cause lethal peritonitis that involves the omentum and peritoneum, that is, peritonitis carcinomatosa. In this review, I introduce current understanding of ovarian cancers: (I) General view of adult human ovaries, (II) Pathologic backgrounds of ovarian cancers of four common histologic types, and in the latter sections, I highlight (III) Cancer stroma including ovarian parenchyma and peritoneal mesenchymal tissue. Cancer-related proinflammatory microenvironment is discussed from the points of cancer progression, angiogenesis and ascites production.

From prepubertal to postmenopausal periods, the ovary exhibits a wide spectrum of appearance and functional activities. The adult ovary is composed of three zones: outer cortex, inner medulla and the hilus (Figure 1A). Follicular structures such as follicle and corpus luteum are located in the cortex (Figure 1B). After ovulation, the corpus luteum is gradually converted to a scar, i.e., the corpus albicans. Aged corpus albicans is either replaced by stroma or embedded in dense collagen fibers in the medulla.

The surface of the ovary consists of a single layer of peritoneal cells (Figure 1C,D). These surface cells are fragile and often invisible in resected ovaries. These mesodermally derived lining cells are not pure “epithelial” cells. They are immunoreactive for epithelial markers such as cytokeratin, Ber-EP4 and Wilms tumor gene (WT1) [3]. In addition, they are positively stained for mesothelial markers such as calretinin and vimentin. Therefore, these cells may be called “modified mesothelium/peritoneal cells”. Epithelial inclusion cysts are often observed in normal ovaries, and they are embedded in the cortex (Figure 1C), and infrequently, in the medullary stroma. The lining cells of inclusion cysts (Figure 1E) sometimes connect with ovarian surface cells, and are also stained for WT1 and other epithelial and mesothelial markers, supporting the notion that they are originated from surface lining cells. It is accepted that ovarian surface cells can undergo epithelial-mesenchymal transition (EMT) both in vitro and in vivo, especially in the process of post-ovulatory remodeling [4]. Although most of inclusion cysts remain inactive, the notion is widely accepted that inclusion cyst is potential precursor lesion of ovarian cancer. Biological dynamics of the cyst is great research interest among gynecologic oncologists. Several studies demonstrate differential molecular expression patterns between inclusion cyst-lining cells and quiescent surface cells [5,6,7].

Dysregulation of some transcription factors are noted in the formation of inclusion cysts, such that paired box gene 8 (PAX8) and CCAT/enhancer binding protein-β are upregulated, whereas GATA4 is frequently lost in inclusion cyst-lining cells [5,6]. Loss of the stroma-specific gene Hoxa5 may also be involved. Hoxa5 gene knockout mice tend to develop inclusion cysts in which cyst lining cells are strongly stained for PAX8 [7]. The finding indicates that Hoxa5 in ovarian stroma may play an important role in the physiologic regulation of ovarian surface epithelial cells.

Ovarian stroma is composed of spindle-shaped cells arranged in whorls or storiform pattern (Figure 1C). Other cells than these spindle cells include luteinized stromal cells. These luteinized cells may exist in the periphery of follicles and lutein cysts. They contain lipid, and look polygonal-shaped. These active stromal cells are generally indistinguishable from non-reactive stromal cells in shape, but they may produce steroid converting enzymes. Postmenopausal ovaries exhibit varying degrees of stromal cells, and it is often the case that hormonally inactive ovaries show relative stromal hyperplasia composed of fibroma-like spindle cells. Both the ovarian artery and the branch of the uterine artery penetrate ovarian hilus and supply the ovary with blood. Thus the ovarian hilus contains numerous blood vessels (Figure 1F). Apart from common epithelial tumors of the ovary, sex cord-stromal tumors originate from granulosa cells, theca cells, Sertoli cells, Leydig cells and fibroblasts of stromal origin. Hormonal disturbance is often observed in the patients with sex cord-stromal tumors because normal counterparts of these cell types produce estrogen, androgen and related steroids, which will be discussed later.

Ovarian cancers belong to surface epithelial-stromal tumors that are the most common types among ovarian neoplasms. The etiology of sporadic ovarian cancers is not fully understood, but a series of investigations suggest that certain conditions such as genetic predisposition and benign inflammatory diseases are involved in the molecular mechanism of carcinogenesis. The pathologic studies have revealed differential background of ovarian cancers depending on histologic types [8,9]. There are four representative histologic types; serous, mucinous, endometrioid and clear cell types (Figure 2A–D). The other types such as transitional cell and squamous cell cancers are reported, if any, at low frequency. The ovarian surface is exposed to the peritoneal cavity. In most cases, cancer cells proliferate inward, and when thin cyst walls rupture, cancer cells spread into peritoneal cavity. Cancer cells may grow outward in papillary form, then directly expose themselves to the peritoneal cavity. Therefore, ovarian cancers preferentially metastasize to multiple surface sites of intraperitoneal organs including the omentum (Figure 2E). Hematogenous and lymphatic metastases are also associated with distant metastasis of this malignancy.

Serous cancer is divided into low- and high-grade subtypes [10,11]. Currently, it is widely accepted that low-grade and high-grade serous cancers develop through different carcinogenic pathways, respectively [9]. The former is characterized by mutations of KRAS, BRAF and ERBB2 [11,12]. The latter is characterized by TP53 mutations and not accompanied by mutations of KRAS, BRAF and ERBB2 [11,12,13]. Intensive studies on the pathogenesis of serous cancers have proposed that a majority of high-grade subtype may arise from the fimbriated end of the fallopian tube [14,15,16,17]. Ovarian surface is in close apposition with fimbria. Therefore, it may not be surprising that tubal intraepithelial cancer cells contact with ovarian surface and then subsequently destruct and replace the ovarian parenchyma, leading to “primary ovarian cancer” as a cystic mass. With regard to familial ovarian cancers, the histology of the patients with germline mutations in either BRCA1 or BRCA2 is in most cases serous type [18,19]. This may be consistent with the findings that serous type is the highest in frequency of tubal cancers, and that tubal cancers are found in a high number of patients with BRCA mutations who undergo prophylactic salpingo-oophorectomy [20].

Mucinous cancer is another frequent type of surface epithelial-stromal tumors. It requires careful clinicopathologic examination whether the ovary is either primary or metastatic. Advanced mucinous tumors spread and produce mucin lakes in the peritoneal cavity, i.e., pseudomyxoma peritonei. It is now widely accepted that the most cases of pseudomyxoma peritonei result from ruptured mucinous appendiceal adenomas/adenocarcinomas, and that the ovary is involved as the secondary lesion [8]. Metastatic ovarian cancers from the gastrointestinal tract frequently show mucinous architecture [21], thus it is critically important to define correctly the origin of the tumor by investigating the immunostaining patterns of cytokeratin 7, cytokeratin 20, CDX2 and some other diagnostic antibodies [22,23,24,25]. With regard to primary ovarian mucinous cancers, mutations of KRAS are observed at higher rates (46–75%) compared with the other types [26,27,28].

Endometrioid cancer may also show genetic changes in PTEN (20–43%) [29,30]. In addition, mutations of β-catenin are frequently observed in this type of tumor (32–64%) [29,31,32]. On the other hand, it is less well investigated the responsible genes contributing to the carcinogenesis of clear cell cancer. A recent study has revealed frequent mutations of chromatin remodeling gene ARID1A (BAF250a) in clear cell (57%) and endometrioid (30%) types of ovarian cancers [33,34]. Endometriosis, a chronic inflammatory disease, may exist in the vicinity of tumor nests of endometrioid and clear cell ovarian cancers in histology. Although the frequency of endometriosis lesion in these types of cancers varies among the studies, it is widely accepted that some endometrioid and clear cell ovarian cancers consecutively develop from the epithelial cells of endometriotic cysts (Figure 2F), which will be discussed later.

Apart from these ovarian cancers belonging to surface epithelial-stromal tumors, many other types of neoplasms develop in the ovary. Teratoma, dysgerminoma and yolk sac tumor are classified into germ cell tumors, and mature cystic teratoma (dermoid cyst) is the most common type of germ cell neoplasm of the ovary [35]. Stromal tumors, if not as common as surface epithelial-stromal tumors, account for 8% of ovarian neoplasms in the U.S. [36], and they exhibit characteristic features. The tumor cells of this group such as granulosa cells and theca cells may produce estrogen, androgen and related hormones [35]. Non-neoplastic counterparts of these cells are sometimes activated in the vicinity of cancer/adenoma glands, and they are called “ovarian tumor with functioning stroma” (Figure 3A), which will be discussed later.

The etiology of sex cord-stromal tumors and germ cell tumors is poorly understood. The mutation in FOXL2 gene, which encodes a forkhead transcription factor, is reported to be present in granulosa cell tumors and less frequently in theca cell tumors with lutenization [37,38]. Germ cell tumors frequently occur to children and adolescent females, therefore, primitive germ cells that colonize the ovary during embryonic stage may aberrantly develop in the early decades of life.

As mentioned above, endometrioid and clear cell ovarian cancers occasionally co-exist with endometriosis (Figure 2F) [39]. Endometriosis is a major gynecologic disease that causes longstanding inflammation, pain and infertility. This disorder is pathologically defined as endometrial glands and stroma outside the uterus. Endometriosis primarily involves pelvic organs such as the peritoneum and ovaries, and the ovarian endometriosis may grow as the cysts containing menstrual effluent (chocolate cyst). The etiology of endometriosis remains to be the question under debate. The implantation theory (Sampson’s theory), that is, endometriosis is formed by implantation via retrograde menstruation, is widely accepted among physicians [40]. However, endometriosis of the other lesions than the urogenital system such as endometriosis in the lung and diaphragm cannot be explained solely by this mechanism. The metaplasia theory that proposes coelomic tissue transformation into endometrium [41], and the embryonic rest theory that explains Mullerian remnant differentiation into endometriotic cells, are also supported by some, if not all, gynecologic and pathologic research groups [42]. Although the pathogenesis may be different among the cases, clinicopathologic features are same; various types of immune cells including hemosiderin-laden macrophages infiltrate in endometriosis lesions where bleeding and tissue remodeling are repeated (Figure 3A,B). Inflammatory cytokines/chemokines are important in the progression of this disease. Ectopic endometrium demonstrates increased expression of various soluble factors, including heat shock protein 27, interleukin (IL)-1β, IL-6, IL-11, IL-12 and vascular endothelial growth factor (VEGF) [43,44,45]. Dysregulated immune responses are also noted. Natural killer (NK) cells show impaired function in the women with endometriosis [46,47,48]. T cells of both inflammatory and immunosuppressive subtypes are increased in the lesions and peritoneal fluid [45,49]. Although the profiles and functions of immune cells are not fully understood, a series of studies elucidate impaired immune responses in the local lesions of endometriosis.

The pathogenesis of endometriosis-associated ovarian cancers (EAOC) is, therefore, likely to be supported by endometriosis-based dysregulated inflammation. Relative risks for ovarian cancers among endometriosis patients are varied among reports [50]. It has been noted in several clinical studies that endometrioid and clear cell cancers are the most frequent histologic types preceded by endometriosis [50,51]. Incidence of EAOC is not fully understood precisely. It is because in part that the cancer tissue destructs and remodels its surrounding stroma in which previous endometriosis lesion is localized. A cohort study reported that cancer risk is elevated significantly among Japanese patients who have ovarian endometriosis, with standardized incidence ratios (SIR) of 8.95 (95% confidence intervals = 4.12–15.3) [52]. Swedish women with early diagnosed and long-standing endometriosis have a higher risk of ovarian cancer with SIR of 2.01 and 2.23, respectively [53]. Not only clinical data but also histopathologic findings support the notion that endometriosis stands as a precursor lesion of ovarian cancer [54,55]. Atypical endometriosis, namely that atypical cells lining in endometriotic cyst and atypical endometrial hyperplasia arising in ovarian endometriosis, is occasionally observed in the vicinity of EAOC (Figure 3E,F). With regard to epithelial abnormalities, molecular studies have revealed that ovarian cancer cells and coexisting endometriosis epithelium share common methylation pattern of the androgen receptor gene and loss of heterozygosity (LOH) at tumor suppressor gene loci [56,57,58]. Since the lesions are exposed to special hormonal and inflammatory factors associated with endometriosis, the role of stroma in the carcinogenesis of EAOC of endometrioid and clear cell types should be distinguished from that in advanced serous cancers with BRCA1 and/or TP53 mutations.

Cancer tissues preferentially generate special stroma for aberrant proliferation and invasion. The fibroblasts and some other cell types such as pre-existing vascular cells and mesenchymal stem cells potentially become cancer-associated fibroblasts (CAF) [59]. Distinctive genetic abnormalities were reported in CAF of some cancer types [60,61,62]. On the other hand, some other studies found no aneuploidy [63,64]. The genetic alteration in CAF of ovarian cancers was reported, such as allelic imbalance on chromosome 3p21 and 18q21 [65,66]. A later study, however, concluded that somatic genetic alterations are extremely rare in CAF of ovarian cancers [67]. Qiu et al., in the latter study, suggested that the quality of DNA accessible from paraffin-embedded tissues may not be reliable for the assessment of genetic alterations [67].

Several cytokines activate fibroblasts, such as hepatocyte growth factor (HGF), fibroblast growth factor-2 (FGF-2), transforming grow factor (TGF)-β, CXCL12 and platelet-derived growth factor (PDGF) [59]. TGF-β produced by ovarian cancer cell line SKOVA3 activates omental fibroblasts [68,69], and direct interaction between the cancer cells and fibroblasts promotes invasion through increased expression of HGF and matrix metalloproteinase (MMP)-2 [68]. The stroma in ovarian cancers should be considered in the light of progression stages. The ovarian cancer cells in advanced stages can be a double-edged sword; they do not only infiltrate for local invasion but also disengage stromal interaction for dissemination. In this review, some important stromal components are highlighted.

The inflammatory microenvironment of ovarian cancers varies among cases, depending on disease background, clinical stages, cancer immunogenicity, host immune response, and repertoire of cytokines/chemokines produced by both tumor and stromal sides. Once cancer cells spread into the peritoneal cavity, they potentially invade peritoneum and omentum, causing carcinomatous peritonitis. Such systemic inflammation does not necessary mean anti-cancer reaction because cancers may cause their favorite inflammation by taking advantage of immunosuppressive cells including regulatory T cells (Treg) and tumor associated macrophages (TAM). Some studies demonstrated that Treg are recruited to ovarian cancer lesions, leading to poor prognostic outcomes [100,101].

In advanced ovarian cancer cases, uncontrollable amounts of ascites are occasionally produced. Several angiogenic and pro-tumoral cytokines/chemokines in ascites have been investigated as prognosis-prediction markers [102,103,104,105]. Chemokines are a family of low molecular weight cytokines, and they exert the activities by binding to corresponding G protein-coupled receptors (GPCRs). There are CC, CXC, C and CX3C subfamilies in human based on the spacing or the presence of four N-terminal cysteine residues, and most members are classified into either CXC or CC subfamilies [106,107]. I highlight CXC and CC chemokines and a few other molecules.

In this review, the pathophysiology of ovarian cancers was discussed from the points of stroma and proinflammatory milieu. As most advanced ovarian cancers spread in the peritoneal cavity and grow in metastatic sites, “stroma of advanced ovarian cancer” should be considered in a wide range of view, including mesothelium, adipose tissue and cancer-producing fluid. Organ-specific stroma with the potency of hormonal production, proinflammatory molecules and peritoneal mesenchyma to which cancer cells migrate are all important factors for better understanding of the dynamics of ovarian cancers.

Apart from common epithelial-stromal ovarian cancers, peritoneal cancers so called “normal-sized ovary cancer syndrome” exhibit similar course to ovarian and tubal serous counterparts [139]. Cancer cells of this type show serous papillary morphology, and are positively immunostained for Ber-EP4 and WT1, and indistinguishable from serous ovarian cancers by cytology and immunohistochemistry [140,141]. Molecular analyses revealed that peritoneal cancers originate in a multifocal manner [142]. Germline mutations of BRCA1 are also associated with this type of cancer at a high rate [143]. Other malignant and pre-cancerous conditions with ovarian histologic subtypes also occur to the peritoneum such as endometriosis, EAOC and mucinous adenocarcinoma [144,145].

Series of investigations have improved our understanding of the nature of ovarian cancers. Many important findings have unveiled distinctive differences in the pathogenesis and microenvironment depending on histologic types and pre-cancerous conditions. Meanwhile, the complexity of proinflammatory milieu has also been elucidated. Further studies are necessary on ovarian cancer stroma including vasculature, inflammatory cells and mesenchymal components of the peritoneal cavity in different types and stages, which will improve the prognosis of the patients by exploring the frontier of more effective and safer therapeutic strategies.