**3. Mechanism of peritoneal metastasis**

**2. Overview of ovarian cancer metastasis**

46 Tumor Metastasis

vessels during hematogenous metastasis [20, 21].

The lack of an anatomic barrier allows the ovarian cancer cells to very conveniently spread into the peritoneal cavity. The cancer cells on the surface of the primary tumors start loosing cell–cell contact and become loosely attached to each other. As a result of this, they become prone to exfoliation into the peritoneal cavity (**Figure 1**). Exfoliation is promoted by the mechanical forces like rubbing of neighboring peritoneal organs during respiratory move‐ ments and flow of the peritoneal fluids. The cancer cells may come off as single cells or as clumps. This is a passive mode of dissemination unlike the typical invasion followed by intravasation observed in tumors undergoing hematogenous metastasis [13, 17, 18]. The peritoneal fluid naturally flows within the peritoneal cavity upward, toward the head, and then back downward, toward the feet, as a result of the diaphragm movement during respiration and gravitational pull, respectively [19]. The exfoliated ovarian cancer cells from the primary tumor are disseminated throughout the peritoneal cavity by this natural flow of the peritoneal fluid (**Figure 1**). Since normally there is only a small volume of the peritoneal fluid present, dissemination is predominantly limited to the organs in the vicinity of the primary tumor [17]. As the disease progresses, more and more ascites is produced and this enables the spread of the cancer cells to more distant sites in the abdomen. One of the pre‐ dominant sites of ovarian cancer metastasis is the omentum which is a fatty double fold of the peritoneal membrane, about 8 by 8 inches in size, covering the bowels [13]. It is important to note that this mode of spreading typical of ovarian cancer is very different in terms of the hydrodynamic forces experienced by the cancer cells when they are carried rapidly in the blood

Epithelial cells tend to undergo anoikis in the absence of attachment to a substratum. Therefore, the main challenge faced by the cancer cells floating in the peritoneal fluid is overcoming anoikis and surviving floatation. In addition, they have to avoid immune surveillance. The cancer cells either form aggregates or spheroids or exist as single cells (**Figure 1**) [22]. The spheroids may also contain embedded cancer‐associated fibroblasts as well as activated mesothelial cells, which contribute to the development of the ascetic microenvironment [22]. The subsequent challenge for these floating cancer cells is to successfully attach to the surface of the organs in the peritoneal cavity (**Figure 1**). Debulking surgery often reveals such spheroids loosely attached to the peritoneum. The mesothelial cells covering the peritoneum and the bowels secrete mucus like substances, which help in reducing friction between surfaces as they brush against each other during the course of the organs' natural movements. The same also helps in preventing attachment of the cancer cells to some extent. However, the integrins expressed by the metastasizing cells help them to attach to the extra cellular matrix proteins (ECMs) secreted by the mesothelial cells. Thereafter, the cancer cells are able to push apart the

mesothelial cells forming the protective barrier and invade into the organ [23, 24].

Having invaded through the mesothelium of the site of metastasis, the cancer cells have to now revert back to their normal self of growing attached to a substratum. However, since they are now encountering a new microenvironment with a potentially different ECM and secreted factors, they have to now adapt to these new conditions. The adaptive process involves The first step in the peritoneal metastasis is exfoliation of the ovarian cancer cells from the primary tumor into the peritoneal cavity. The prerequisite for this step is the loss of cell–cell contact between the cancer cells. As mentioned earlier, ovarian cancer can potentially arise from the fallopian tube epithelial cells or ovarian surface epithelium. Both express the classic epithelial marker epithelial cadherin (E‐cadherin) [29]. E‐cadherin plays a key role in epithelial cell behavior, tumor suppression, and tissue architecture through its function as a cell–cell adhesion molecule [30]. It is associated with the actin cytoskeleton through α, β, and γ catenins. While E‐cadherin is directly involved in the formation of adherens junctions between adjacent epithelial cells, it can also regulate the formation of tight junctions and desmosomes [30, 31].

As the cancer progresses from a benign to a malignant form, the cells undergo an epithelial to mesenchymal transition (EMT). This involves molecular and morphological changes wherein they loose their epithelial characteristics and gain mesenchymal traits. This includes a loss of the compact cell‐to‐cell attachment, polarity, and cuboidal shape. The cells become more spindle shaped and motile. EMT also involves a change in the expression of epithelial and mesenchymal markers [32]. A very important aspect of this transition is the loss of expression of E‐cadherin and a concomitant increase in the expression of neural cadherin (N‐cadherin). This results in a reduction in the cell–cell interaction between cancer cells through their adherens junctions and an increase in the ability of the cancer cells to interact with the normal stromal cells present in the microenvironment. In ovarian cancer, E‐cadherin expression can be regulated transcriptionally and post‐transcriptionally [33]. ZEB‐1, ZEB‐2, Snail, and Slug are known to repress E‐cadherin and can be regulated by several external cues. The signaling pathways that regulate EMT and E‐cadherin expression include transforming growth factor β (TGF‐β), epidermal growth factor (EGF), hepatocyte growth factor (HGF), endothelin‐1 (ET‐ 1), and bone morphogenetic protein 4 (BMP‐4) [32]. Moreover, the miR‐200 family of micro‐ RNAs can also indirectly regulate EMT by targeting ZEB‐1 and ZEB‐2, which results in the derepression of E‐cadherin [34]. Decreased expression of miR‐200 family resulted in an increase in the expression of ZEB‐1 and ZEB‐2, which repressed E‐cadherin transcription and induced EMT in ovarian cancer.

The loss of E‐cadherin expression and the resulting decrease in the cell–cell attachment promotes the dissemination of the cells into the peritoneal cavity. Interestingly, the loss in E‐ cadherin expression was found to lead to an induction of expression of α5‐integrin [35]. α5‐ integrin forms a heterodimer with β1‐integrin that binds to fibronectin and hence is called the fibronectin receptor. The induction of α5‐integrin was not through the canonical β‐catenin pathway. Instead, it was through the epithelial growth factor receptor (EGFR)/focal adhesion kinase (FAK)/mitogen‐activated protein kinase (MAPK) pathway. The increase in fibronectin receptor expression was found to help the disseminated ovarian cancer cells attach to the fibronectin secreted by the mesothelial cells lining the omentum and peritoneum [35]. This is an evidence of how the loss of E‐cadherin—which facilitates shedding—is coupled to prepar‐ ing the cells to reattach at the distant metastatic site.

Once the cancer cells have been shed into the peritoneal fluid, it significantly affects the prognosis of the patient as evidenced by the 29% relapse rate of stage 1A ovarian cancer compared to 59% relapse rate of stage 1C [36]. However, once detached from the tumor mass, the cancer cells face several challenges in surviving in the peritoneal fluid. The peritoneal fluid is a result of continuous secretion of fluids by the peritoneal capillaries. This helps in lubricating the adjacent organs in the peritoneal cavity and allows uptake of soluble factors through the peritoneum. A majority of the peritoneal fluid is returned to the circulation through lymphatic drainage. However, in ovarian cancer patients, the increased leakiness of vasculature induced by high vascular endothelial growth factor (VEGF) levels accompanied by blocking of the lymphatic vessels by cancer cells results in ascites formation [22]. This ascites is called malignant ascites because of the presence of floating cancer cells. The malignant ascites facilitates the spread of the cancer cells throughout the peritoneal cavity.

The disseminated ovarian cancer cells floating in the ascites either as spheroids or as single cells develop resistance to anoikis and acquire cancer stem cell‐like properties [37, 38]. Interestingly, the single‐cell population was found to have a greater percentage of cancer stem cells [39]. The cancer stem cells enriched from ascites have highly elevated ability to form mouse xenograft tumors [40]. Just like cancer stem cells, the floating spheroids and single cells are resistant to chemotherapy. The compact nature of the spheroids serves as an additional physical barrier for the chemotherapeutic agents, preventing the inner cells from exposure to the drug [22]. Taken together, this indicates that the cancer cells floating in the ascites are stem‐ like and chemoresistant and have the potential to seed new metastatic tumors within the peritoneal cavity.

The spheroids have elevated levels of E‐cadherin and EpCAM and concomitant diminished expression of vimentin, matrix metalloproteinases (MMPs), and CD44 [37, 39]. Therefore, the metastasizing cancer cells demonstrate plasticity in terms of their ability to switch back and forth from epithelial and mesenchymal phenotypes as per the demands of the different steps of metastasis. In addition to the cancer cells, the ascites has several normal cell types that together form the malignant ascites microenvironment and supports the floating cancer cells. The main non‐cancer‐cell types include cancer‐associated fibroblasts (CAFs), mesothelial cells, immune cells, mesenchymal stem cells, and platelets [22]. These cells can be associated with the cancer spheroids or the single cells. They can also exist by themselves, floating in the peritoneal fluid. These supporting cells produce a milieu of factors that assist the cancer cell survival and subsequent colonization of the metastatic site. Cells like platelets also offer protection from immune surveillance by coating the cancer cells.

be regulated transcriptionally and post‐transcriptionally [33]. ZEB‐1, ZEB‐2, Snail, and Slug are known to repress E‐cadherin and can be regulated by several external cues. The signaling pathways that regulate EMT and E‐cadherin expression include transforming growth factor β (TGF‐β), epidermal growth factor (EGF), hepatocyte growth factor (HGF), endothelin‐1 (ET‐ 1), and bone morphogenetic protein 4 (BMP‐4) [32]. Moreover, the miR‐200 family of micro‐ RNAs can also indirectly regulate EMT by targeting ZEB‐1 and ZEB‐2, which results in the derepression of E‐cadherin [34]. Decreased expression of miR‐200 family resulted in an increase in the expression of ZEB‐1 and ZEB‐2, which repressed E‐cadherin transcription and

The loss of E‐cadherin expression and the resulting decrease in the cell–cell attachment promotes the dissemination of the cells into the peritoneal cavity. Interestingly, the loss in E‐ cadherin expression was found to lead to an induction of expression of α5‐integrin [35]. α5‐ integrin forms a heterodimer with β1‐integrin that binds to fibronectin and hence is called the fibronectin receptor. The induction of α5‐integrin was not through the canonical β‐catenin pathway. Instead, it was through the epithelial growth factor receptor (EGFR)/focal adhesion kinase (FAK)/mitogen‐activated protein kinase (MAPK) pathway. The increase in fibronectin receptor expression was found to help the disseminated ovarian cancer cells attach to the fibronectin secreted by the mesothelial cells lining the omentum and peritoneum [35]. This is an evidence of how the loss of E‐cadherin—which facilitates shedding—is coupled to prepar‐

Once the cancer cells have been shed into the peritoneal fluid, it significantly affects the prognosis of the patient as evidenced by the 29% relapse rate of stage 1A ovarian cancer compared to 59% relapse rate of stage 1C [36]. However, once detached from the tumor mass, the cancer cells face several challenges in surviving in the peritoneal fluid. The peritoneal fluid is a result of continuous secretion of fluids by the peritoneal capillaries. This helps in lubricating the adjacent organs in the peritoneal cavity and allows uptake of soluble factors through the peritoneum. A majority of the peritoneal fluid is returned to the circulation through lymphatic drainage. However, in ovarian cancer patients, the increased leakiness of vasculature induced by high vascular endothelial growth factor (VEGF) levels accompanied by blocking of the lymphatic vessels by cancer cells results in ascites formation [22]. This ascites is called malignant ascites because of the presence of floating cancer cells. The malignant ascites

The disseminated ovarian cancer cells floating in the ascites either as spheroids or as single cells develop resistance to anoikis and acquire cancer stem cell‐like properties [37, 38]. Interestingly, the single‐cell population was found to have a greater percentage of cancer stem cells [39]. The cancer stem cells enriched from ascites have highly elevated ability to form mouse xenograft tumors [40]. Just like cancer stem cells, the floating spheroids and single cells are resistant to chemotherapy. The compact nature of the spheroids serves as an additional physical barrier for the chemotherapeutic agents, preventing the inner cells from exposure to the drug [22]. Taken together, this indicates that the cancer cells floating in the ascites are stem‐ like and chemoresistant and have the potential to seed new metastatic tumors within the

facilitates the spread of the cancer cells throughout the peritoneal cavity.

induced EMT in ovarian cancer.

48 Tumor Metastasis

peritoneal cavity.

ing the cells to reattach at the distant metastatic site.

Having successfully survived flotation in the peritoneal fluids, the next goal of the metasta‐ sizing ovarian cancer cells is to attach to the various organs present in the peritoneal cavity. Electron micrographs of sections of normal peritoneum and omentum have revealed the architecture of the mesothelium covering them. The mesothelium consists of a monolayer of mesothelial cells that are very tightly joined end to end to form a protective barrier [24]. These mesothelial cells serve to provide a slippery surface—through the secretion of glycosamino‐ glycans and lubricants—facilitating normal coelomic movement as well as preventing infection and attachment of cancer cells [41]. The mesothelial cells can perform diverse functions such as secretion of ECMs, growth factors, and inflammatory cytokines for tissue repair and regeneration, proteases for fibrinolysis, and prevention of adhesions [41]. They are also actively involved in the movement of fluids and solutes across serosal cavities [42].

Early *in vitro* experiments revealed that the ovarian cancer cells force the retraction of the mesothelial cells upon attachment to the mesothelium [43]. More recently, Iwanicki et al. have demonstrated the role of the fibronectin receptor (α5β1‐integrin) expressed on the surface of the ovarian cancer cells help them attach to the fibronectin secreted on the surface of the mesothelial cells and promote the displacement of mesothelial cells through myosin‐mediated traction forces [24]. Subsequent studies revealed the ovarian cancer cells with a mesenchymal phenotype had a greater propensity for mesothelial clearance [44]. The fibronectin secretion by the mesothelial cells was found to be induced by their interaction with the metastasizing ovarian cancer cells. The TGF‐β secreted by the cancer cells activated a RAC1/SMAD‐mediated signaling pathway in the mesothelial cells, which resulted in the transcriptional upregulation of the fibronectin gene and also induced an EMT‐like phenotype in the mesothelial cells [45]. This would probably help in subsequent mesothelial clearance and also may potentially serve as a source of cancer‐associated fibroblasts in the microenvironment of the metastatic tumor [46].

The increased expression of the fibronectin receptor in the ovarian cancer cells is also beneficial in coupling attachment to growth factor signaling to promote metastasis. Inhibition of the interaction of α5β1‐integrin on the cancer cells with the fibronectin on the surface of the omentum and peritoneum in mouse xenograft models of ovarian cancer metastasis resulted in a decreased metastatic burden in both prevention and intervention settings [28]. Since inhibition of α5β1‐integrin can also inhibit angiogenesis, it was further investigated whether the effects on metastasis were actually due to disruption of the human cancer cell α5β1‐integrin interaction with fibronectin or that of the mouse endothelial cell α5β1‐integrin. The effect of an anti‐murine α5β1‐integrin‐blocking antibody was compared to that of the anti‐human α5β1‐ integrin‐blocking antibody. Interestingly, the murine‐blocking antibody did not show any significant effect and, therefore, confirmed the key role of the interactions of the cancer cell α5β1‐integrin with the fibronectin of the microenvironment in promoting ovarian cancer metastasis [28]. Further investigation revealed that the activation of α5β1‐integrin resulted in the activation and phosphorylation of the receptor tyrosine kinase c‐Met independent of its ligand—hepatocyte growth factor (HGF) [28]. This attachment induced activation of the growth factor receptor lead to increased invasiveness and growth through the subsequent activation of the FAK/Src signaling pathways in the cancer cells. The expression of a consti‐ tutively active FAK could abrogate the inhibitory effects of the α5β1‐integrin‐blocking antibody on the ovarian cancer cells [28].

Another effect of adhesion of the ovarian cancer cells to the surface of the omentum is the increased secretion of the extracellular protease MMP‐2. It cleaves fibronectin and vitronectin present on the surface of the mesothelium into smaller fragments, which enhances binding of the cancer cells to these ECMs through their specific integrin receptors α5β1‐integrin and αvβ3‐ integrin, respectively [47]. Inhibition of MMP‐2 in the ovarian cancer cells as a prevention measure inhibited their adhesion to the omentum in nude mice. However, the host MMP‐2 did not play a role in this process as evidenced in MMP‐2 knockout mouse xenograft experi‐ ments [47].

Once the cancer cells attach to the mesothelial cells on the surface of the omentum, they embark on a process of adapting to the new microenvironment of the site of metastasis. As evidenced by the ECM-cancer cell interactions and their consequences above, more productive reciprocal interactions between the cancer cells and their new microenvironment are essential for successful establishment of the metastatic tumors. The cancer cells have to revert from surviving anoikis while floating in the peritoneal fluid to an attached growth in the presence of new ECMs and growth factors available in the microenvironment of the omentum and peritoneum. This involves significant changes in the gene expression profiles of the colonizing cancer cells and, therefore, would involve the activation/repression of transcriptional/ translational regulators dependent on microenvironmental cues. One such important micro‐ environment regulated translational regulator was reported to be the micro‐RNA miR‐193b [6]. miR‐193b is a tumor suppressor micro‐RNA that was found to be downregulated in the metastasizing ovarian cancer upon their interaction with the mesothelial cells covering the surface of the omentum. This downregulation promoted growth and invasiveness of the cancer cells *in vitro,* colonization of human omentum *ex vivo,* and decreased metastasis in mouse xenografts [6]. Interestingly, the miR‐193b downregulation was induced by the hypermethy‐ lation of its promoter as a result of the cross talk between the cancer cells and the mesothelial cells. The promoter hypermethylation was catalyzed by the increased expression of DNMT1 in the cancer cells stimulated by their interaction with the mesothelium [6]. miR‐193b was found to directly target urokinase. A decrease in miR‐193b expression resulted in increased expression of urokinase, which mediated the functional effects of miR‐193b in driving metastatic colonization of the omentum [6].

inhibition of α5β1‐integrin can also inhibit angiogenesis, it was further investigated whether the effects on metastasis were actually due to disruption of the human cancer cell α5β1‐integrin interaction with fibronectin or that of the mouse endothelial cell α5β1‐integrin. The effect of an anti‐murine α5β1‐integrin‐blocking antibody was compared to that of the anti‐human α5β1‐ integrin‐blocking antibody. Interestingly, the murine‐blocking antibody did not show any significant effect and, therefore, confirmed the key role of the interactions of the cancer cell α5β1‐integrin with the fibronectin of the microenvironment in promoting ovarian cancer metastasis [28]. Further investigation revealed that the activation of α5β1‐integrin resulted in the activation and phosphorylation of the receptor tyrosine kinase c‐Met independent of its ligand—hepatocyte growth factor (HGF) [28]. This attachment induced activation of the growth factor receptor lead to increased invasiveness and growth through the subsequent activation of the FAK/Src signaling pathways in the cancer cells. The expression of a consti‐ tutively active FAK could abrogate the inhibitory effects of the α5β1‐integrin‐blocking antibody

Another effect of adhesion of the ovarian cancer cells to the surface of the omentum is the increased secretion of the extracellular protease MMP‐2. It cleaves fibronectin and vitronectin present on the surface of the mesothelium into smaller fragments, which enhances binding of the cancer cells to these ECMs through their specific integrin receptors α5β1‐integrin and αvβ3‐ integrin, respectively [47]. Inhibition of MMP‐2 in the ovarian cancer cells as a prevention measure inhibited their adhesion to the omentum in nude mice. However, the host MMP‐2 did not play a role in this process as evidenced in MMP‐2 knockout mouse xenograft experi‐

Once the cancer cells attach to the mesothelial cells on the surface of the omentum, they embark on a process of adapting to the new microenvironment of the site of metastasis. As evidenced by the ECM-cancer cell interactions and their consequences above, more productive reciprocal interactions between the cancer cells and their new microenvironment are essential for successful establishment of the metastatic tumors. The cancer cells have to revert from surviving anoikis while floating in the peritoneal fluid to an attached growth in the presence of new ECMs and growth factors available in the microenvironment of the omentum and peritoneum. This involves significant changes in the gene expression profiles of the colonizing cancer cells and, therefore, would involve the activation/repression of transcriptional/ translational regulators dependent on microenvironmental cues. One such important micro‐ environment regulated translational regulator was reported to be the micro‐RNA miR‐193b [6]. miR‐193b is a tumor suppressor micro‐RNA that was found to be downregulated in the metastasizing ovarian cancer upon their interaction with the mesothelial cells covering the surface of the omentum. This downregulation promoted growth and invasiveness of the cancer cells *in vitro,* colonization of human omentum *ex vivo,* and decreased metastasis in mouse xenografts [6]. Interestingly, the miR‐193b downregulation was induced by the hypermethy‐ lation of its promoter as a result of the cross talk between the cancer cells and the mesothelial cells. The promoter hypermethylation was catalyzed by the increased expression of DNMT1 in the cancer cells stimulated by their interaction with the mesothelium [6]. miR‐193b was found to directly target urokinase. A decrease in miR‐193b expression resulted in increased

on the ovarian cancer cells [28].

ments [47].

50 Tumor Metastasis

As the cancer cells adapt to the new microenvironment of the metastatic site and start to proliferate, they also recruit resident and non‐resident normal cells and convert them into the tumor‐associated stroma or 'activated stroma' [48]. It is well known that the tumors consist of 10–50% of non‐cancer cells or the tumor stroma [49]. The key components of this tumor stroma are the cancer‐associated fibroblasts (CAFs), tumor‐associated macrophages (TAMs), and other immune cells, endothelial cells, pericytes, adipocytes, extracellular matrix proteins, etc. [50]. All these stromal components are essential for successful growth and progression of the tumors as they are a critical source of growth and tropic factors, help in evasion of immune surveillance, angiogenesis, ECM remodeling, invasiveness, etc. Therefore, the eventual success of the cancer cells in colonizing the omentum will depend upon their ability to develop an active tumor stroma.

The metastasizing ovarian cancer cells were found to recruit the resident normal fibroblasts in the basement membrane of the omentum and reprogram them into CAFs. This reprogramming was driven by the decreased expression of miR‐214, miR‐31, and an increase in the expression of miR‐155 in the normal fibroblasts induced by the cancer cells [49]. The resulting CAFs promoted ovarian cancer cell migration, invasion, and colony formation *in vitro* and tumor growth and metastasis *in vivo*. Interestingly, CAFs could be converted back into normal fibroblasts by the combined overexpression of miR‐214 and miR‐31 and inhibition of miR‐155. The micro‐RNAs mediated their effects through an array of targets, most of which were identified as chemokines and cytokines. The main mediator was found to be CCL5, which was a direct target of miR‐214. Inhibition of CCL5 in nude mice injected with a mixture of ovarian cancer cells and CAFs significantly decreased the ability of the CAFs to promote tumor growth and metastasis [49].

An important and abundant cellular component of the omentum is adipocytes. Until recently, not much was known about the direct role of the omental adipocytes in promoting ovarian cancer metastasis to the omentum even though it is well established that omentum is one of the main sites of ovarian cancer metastasis and that it is a predominantly fatty tissue. Dr. Lengyel's group went on to demonstrate that the omental adipocytes secrete adipokines that promote the homing of the metastasizing ovarian cancer cells to the omentum [51]. The cancer cells, thereafter, could induce metabolic reprogramming of the adipocytes and induce lipolysis in them. The adipocytes in turn induced the expression of FABP4—a fatty acid transporter in the cancer cells. As a result of this, the cancer cells efficiently take up the free fatty acids released by the adipocytes and utilize them as a source of energy and building blocks to drive tumor growth [51]. This explains why the omentum tumor is usually the largest one in the peritoneal cavity with sometimes the whole omentum getting converted into a solid, hard omental cake. By that time, all the adipocytes have been depleted and used for the growth of the metastatic tumor.
