Pathophysiology of colorectal peritoneal carcinomatosis: Role of the peritoneum.

Colorectal cancer (CRC) is the third most common cancer and the fourth most common cause of cancer-related death worldwide. Besides the lymphatic and haematogenous routes of dissemination, CRC frequently gives rise to transcoelomic spread of tumor cells in the peritoneal cavity, which ultimately leads to peritoneal carcinomatosis (PC). PC is associated with a poor prognosis and bad quality of life for these patients in their terminal stages of disease. A loco-regional treatment modality for PC combining cytoreductive surgery and hyperthermic intraperitoneal peroperative chemotherapy has resulted in promising clinical results. However, this novel approach is associated with significant morbidity and mortality. A comprehensive understanding of the molecular events involved in peritoneal disease spread is paramount in avoiding unnecessary toxicity. The emergence of PC is the result of a molecular crosstalk between cancer cells and host elements, involving several well-defined steps, together known as the peritoneal metastatic cascade. Individual or clumps of tumor cells detach from the primary tumor, gain access to the peritoneal cavity and become susceptible to the regular peritoneal transport. They attach to the distant peritoneum, subsequently invade the subperitoneal space, where angiogenesis sustains proliferation and enables further metastatic growth. These molecular events are not isolated events but rather a continuous and interdependent process. In this manuscript, we review current data regarding the molecular mechanisms underlying the development of colorectal PC, with a special focus on the peritoneum and the role of the surgeon in peritoneal disease spread.

Core tip: Colorectal peritoneal carcinomatosis (PC) is associated with poor prognosis and bad quality of life for patients in their terminal stages of disease. A loco-regional treatment, combining cytoreductive surgery and hyperthermic intraperitoneal peroperative chemotherapy, is associated with significant morbidity and mortality. Therefore, a comprehensive understanding of the molecular events involved in peritoneal disease spread and subsequent careful patient selection is paramount to avoid unnecessary toxicity. This manuscript highlights current data regarding the molecular mechanisms underlying the development of colorectal PC with a special focus on the peritoneum and the role of the surgeon in peritoneal disease spread.

INTRODUCTION

Colorectal cancer (CRC) is the third most common cancer and the fourth most common cause of cancer-related death worldwide[1,2]. Besides the lymphatic and haematogenous routes of dissemination, CRC frequently gives rise to the transcoelomic spread of tumor cells in the peritoneal cavity, which ultimately leads to peritoneal carcinomatosis (PC)[3]. Colorectal PC is associated with a poor prognosis and bad quality of life for these patients in their terminal stages of disease[4-6]. Recent genomic profiling studies have demonstrated distinct gene expression patterns, determining CRC spreading to either the liver, the peritoneum or both[7,8].

The precise incidence of PC is unknown due to the low sensitivity of preoperative imaging techniques (CT, MRI, PET, PET/CT) and heterogeneity of published methods and findings[9,10]. Using a database of 3019 CRC patients, Jayne et al[11] reported that 8% of these patients presented with synchronous PC and 5% presented with metachronous disease. In a recent population-based cohort study from Stockholm County in Sweden, 4.3% of 11124 CRC patients were diagnosed with synchronous PC and 4% with metachronous PC[12]. In the synchronous group, more than 50% of patients presented with peritoneal metastases as the only site of tumor dissemination.

In the past, oncologists and surgeons assumed PC was identical to distant metastases and as such, regarded it as an incurable component of intra-abdominal malignancy only open to palliative treatment options. Systemic chemotherapy in patients with PC resulted in no long-term survival with a median overall survival of 15.2 mo and poor quality of life of these patients in their terminal stages of disease[13,14]. A new loco-regional treatment modality combining cytoreductive surgery (CRS) and hyperthermic intraperitoneal peroperative chemotherapy (HIPEC) has demonstrated promising clinical results in patients with colorectal PC. In this setting, CRS aims to remove all macroscopic tumor through standardized peritonectomy procedures and multivisceral resections, whereas the subsequent intraoperative chemotherapy seeks to eliminate all residual microscopic tumor[15]. This novel approach has demonstrated encouraging clinical results in several phase II and III trials[16-18]. However, CRS and HIPEC are associated with a significant morbidity (grade III-IV complications) of approximately 34% and a 30-d mortality of 4%[19,20]. Therefore, careful patient selection is paramount to avoid unnecessary toxicity[21-23]. The aim of this manuscript is to review current data regarding the molecular mechanisms underlying the development of colorectal PC, with a special focus on the peritoneum.

THE PERITONEUM AS THE FIRST LINE OF DEFENCE IN PC

The peritoneum is the largest and most complex serous membrane of the human body[24]. The visceral peritoneum, covering the intra-abdominal organs and mesenteries, forms a continuous layer with the parietal peritoneum, which lines the abdominal wall and pelvic cavities[24]. It is composed of a monolayer of mesothelial cells supported by a basement membrane that rests on a layer of connective tissue, also referred to as the submesothelium (Figure 1)[25].

Figure 1

Structure of the peritoneum. The peritoneum is composed of a mesothelium supported by a basement membrane that rests on a layer of submesothelium. The mesothelium consists of a monolayer of either flattened, stretch, squamous-like or cuboidal mesothelial cells. The luminal surface of mesothelial cells has numerous microvilli varying in shape, size and density. Cilia have also been identified on the surface of resting mesothelial cells. The basement membrane consists of a thin laminar network containing type I and IV collagen, proteoglycans and glycoproteins. The submesothelium consists of a complex network of extracellular matrix made up of different types of collagen, glycoproteins, glycosaminoglycans and proteoglycans. Blood vessels, lymphatics, and various cells types (fibroblasts, resident tissue macrophages, and mast cells) are also found is this layer. ECM: Extracellular matrix.

The mesothelium consists of a monolayer of either flattened, stretched, squamous-like or cuboidal mesothelial cells. The latter can be found in various areas including the liver, the spleen, the “milky spots” of the omentum and the peritoneal side of the diaphragm overlying the lymphatic lacunae (cfr. Attachment to distant peritoneum)[26-28]. Cuboidal mesothelial cells are also observed within an injured mesothelium. Both squamous and cuboidal mesothelial cells can be distinguished by their ultrastructural differences. Squamous-like mesothelial cells contain few mitochondria, a poorly developed Golgi apparatus and little rough endoplasmatic reticulum (RER), which are located centrally near the round or oval nucleus[29]. Cuboidal mesothelial cells contain a central prominent nucleolus, abundant mitochondria and RER, a well developed Golgi apparatus, microtubules and microfilaments[30]. The luminal surface of mesothelial cells has numerous microvilli varying in shape, size and density; increasing the functional mesothelial surface area[31]. Cilia have also been identified on the surface of resting mesothelial cells[32]. The mesothelium functions as a dynamic layer that contributes substantially to the structural, functional, and homeostatic properties of the peritoneum[29]. The underlying basement membrane, a thin laminar network containing type I and IV collagen, proteoglycans and glycoproteins, acts as a selective barrier to macromolecules entering the submesothelial layer[29]. The submesothelium consists of a complex network of extracellular matrix (ECM) made up of different types of collagen, glycoproteins, glycosaminoglycans and proteoglycans. Blood vessels, lymphatics, and various cell types (fibroblasts, resident tissue macrophages, and mast cells) are also found in this layer[24,29].

The first major function of the peritoneum is facilitating transport of fluid and cells across the serosal cavities[33]. The microvilli on the luminal surface of the mesothelial cells play an important role in this process as they increase the surface area and bind fluids in their glycosaminoglycan-rich glycocalyx thereby aiding absorption[34]. Secondly, the peritoneum provides a slippery and non-adhesive surface to allow intracoelomic movement[35]. This slippery and non-adhesive surface is established by the secretion of a small amount of sterile fluid containing phosphatidylcholine produced by each mesothelial cell. Thirdly, the peritoneum acts as a first line of defence in host resistance[36,37]. The fourth function of the peritoneum involves tissue repair by releasing growth factors[38,39]. In conclusion, the peritoneum should be considered an organ with a structural and protective function for the contents of the abdominal cavity[24,25,40].

PATHOPHYSIOLOGY OF COLORECTAL PC

The emergence of PC is the result of a molecular crosstalk between tumor cells and host elements, comprising several well-defined steps. First, individual or clumps of tumor cells detach from the primary tumor and gain access to the peritoneal cavity. In the second step, these free tumor cells become susceptible to the regular peritoneal transport along predictable routes. The third step involves the attachment to the distant peritoneum where the tumor cells, during the fourth step, invade the subperitoneal space. The underlying connective tissue provides the necessary scaffold for tumor proliferation. The final step involves angiogenesis, which sustains tumor proliferation and enables further metastatic growth[41]. It is important to realise that these steps, known as the “peritoneal metastatic cascade”, do not necessarily occur in isolation, but rather describe a continuous and interdependent process (Figure 2, Table 1)[41].

Figure 2

Pathophysiology of colorectal peritoneal carcinomatosis: the peritoneal metastatic cascade. The emergence of PC is the result of a molecular crosstalk between tumor cells and host elements, comprising several well-defined steps. A: Individual or clumps of tumor cells detach from the primary tumor and gain access to the peritoneal cavity. Spontaneous exfoliation of tumor cells from the primary tumor can be promoted by the down-regulation of E-cadherin, increased interstitial fluid pressure, and iatrogenically during surgery; B: The free tumor cells become susceptible to the regular peritoneal transport. Peritoneal transport is due to changes in the intra-abdominal pressure during respiration, gravity and peristalsis of the bowel; which results in a clockwise flow from the pelvis, along the right paracolic gutter and to the subdiaphragmatic space and finally towards the pelvis again; C: Attachment of tumor cells to distant peritoneum occurs via two processes, denominated transmesothelial and translymphatic metastasis. During transmesothelial metastasis, loose tumor cells directly adhere to distant mesothelium through adhesion molecules. During translymphatic metastasis, free tumor cells gain access to the submesothelial lymphatics through lymphatic stomata. Preferential tumor growth in the milky spots of the greater omentum has been observed; D: Tumor cells invade the submesothelium. In areas of absent or rounded (cuboidal) mesothelial cells, tumor cells interact with the laminar network of the basement membrane through integrin-mediated adhesion. Subsequent invasion of the submesothelial tissue occurs via degradation by proteases (MMPs); E: Systemic metastasis. PC: Peritoneal carcinomatosis.

Table 1

General overview molecules/molecular pathways involved in the peritoneal metastatic cascade

Steps peritoneal metastatic cascade	Molecules/molecular pathways
Detachment from the primary tumor	Spontaneous tumor cell shedding:
	E-cadherin ↓
	N-cadherin ↑
	EMT
	PC1 and PC2 ↑
	Interstitial fluid pressure ↑
	Peroperative seeding tumor cells during surgery
Peritoneal transport	Muscinous ascites
	Actin microfilament system
	Lammelipodia, filipodia
Attachment to distant peritoneum	Transmesothelial dissemination:
	ICAM-1 ↑, PECAM-1, VCAM-1 ↑
	TNF-α, IL-1β, IL-6, IFN-γ
	β1 integrin subunit
	CD43, CD44
	Hyaluronan
	Translymphatic dissemination:
	Lymphatic stomata
	Milky spots
Invasion into the subperitoneal space	Rounding of mesothelial cells:
	HGF/SF ↑
	c-Met ↑
	Destruction of the mesothelial monolayer:
	Tumor-induced apoptosis
	Fas ligand/Fas
	Adherence to the basement membrane:
	Integrines
	Invasion of the peritoneal-blood barrier:
	MMP-1, MMP-2, MMP-7, MMP-9, MMP-13, MMP-14 ↑
	TIMP-1, TIMP-2, TIMP-3, TIMP-4
	uPA/uPAR
	plasminogen activator inhibitor -1 and -2
Proliferation and angiogenesis	Proliferation:
	EGFR, EGF, TGFα
	IGF-1, IGF-Binding Protein-3
	Angiogenesis:
	HIF-1α, HIF-1β
	VEGF/VEGFR

E-cadherin: Epithelial-cadherin; N-cadherin: Neural-cadherin; EMT: Epithelial to mesenchyme transition; PC: Polycystin; ICAM: Intercellular adhesion molecule-1; PECAM: Platelet-endothelial cell adhesion molecule-1; VCAM-1: Vascular adhesion molecule-1; TNF-α: Tumor necrosis factor-α; IL-1β: Interleukin-1β; IL-6: Interleukin-6; IFN-γ: Interferon-γ; CD43: Sialophorin; HGF: Hepatocyte growth factor; SF: Scatter factor; MMP: Matrix metalloproteinases; TIMP: Tissue inhibitor metalloproteinases; uPA: Urokinase plasminogen activator; uPAR: Urokinase plasminogen activator receptor; EGFR: Epidermal growth factor receptor; EGF: Epidermal growth factor; TGFα: Tumor growth factor α; IGF-1: Insulin like growth factor-1; HIF: Hypoxia inducible factor; VEGF: Vascular endothelial growth factor; VEGFR: Vascular endothelial growth factor receptor.

Detachment of tumor cells from the primary tumor

The metastatic pathway begins with the detachment of individual, or clumps of tumor cells from the primary tumor. This can be the result of spontaneous exfoliation of tumor cells from cancers that have invaded though the full thickness of the bowel wall and its investing serosa[41,42]. Serosal involvement of colon adenocarcinoma (pT4 stage) is an unfavourable independent prognostic marker for the development of PC[43-45]. Spontaneous exfoliation of malignant cells can be promoted by the down-regulation of intracellular adhesion molecules on the tumor cell surfaces, more specifically E (epithelial)-cadherin[46]. E-cadherin belongs to the type I subfamily of cadherins[47]. The general structure comprises an extracellular part, a membrane-spanning domain and a cytoplasmic tail. E-cadherin binds homotypically to E-cadherin on neighbouring cells through its Ca2+-dependent extracellular domains. The cytoplasmic tail associates with p120, α-, β-, and γ-catenin, which is responsible for the connection with the actin cytoskeleton and allows in- and outward signal transduction[46-49]. It has been confirmed that the down-regulation of E-cadherin expression levels is associated with dedifferentiation, progression, and metastasis of CRC[50,51]. This down-regulated tumor suppressor or metastasis suppressor function has also been reported in gastric[52,53] and ovarian cancer[54,55] with PC. Furthermore, reduction of cell-cell adherence, by the loss of E-cadherin, and the upregulation of mesenchymal N (neural)-cadherin are established hallmarks of the epithelial to mesenchyme transition (EMT)[56]. This reversible reprogramming process allows cells to separate, lose their apico-basal polarity typical of epithelial cells, demonstrate heightened resistance to apoptosis, and revert to a more motile mesenchymal phenotype[56]. This is believed to play a major role in the invasion and metastasis of tumor cells[49,57]. Gargalionis et al[58] report that overexpression of the epithelial polycystins PC1 and PC2 in a human colon carcinoma cell line, HCT116, is able to induce EMT-related alteration in E-cadherin, N-cadherin, Snail and Twist (EMT trigger) mRNA expression. PC2 exogenous expression was also found to increase cell migration[58]. PC1 and PC2 are both membrane-spanning proteins, which were first identified as mechanosensors involved in the pathophysiology of polycystic kidney disease[59]. PC1 is a mechanosensor with G-protein coupled receptor properties that perceives extracellular mechanical signals and translates them into biochemical responses[60-63]. PC2 constitutes a mechanosensitive Ca2+ channel[64-67]. Both receptors physically interact through their C-terminal, forming heterodimeric complexes at the cell membrane and at primary cilia[68,69].

Secondly, spontaneous tumor cell shedding from the primary tumor can also occur due to increased interstitial fluid pressure, a well-known phenomenon in solid tumors such as in CRC. Hayashi et al[70] reported that the pressure in the tumor is important, not only for the number of tumor cells shed but, also for the size of emboli shedding into lymphatics around the primary tumor. Interstitial hypertension can be the result of high osmotic pressure, increased vessel permeability and hyperperfusion, rapid cell proliferation, lack of effective lymphatic drainage, hyperplasia around blood vessels and increased production of ECM components[71].

Thirdly, peroperative seeding of viable tumor cells can also be induced iatrogenically during surgery, when the tumor is inadvertently ruptured, opened or cut into. Alternatively, the presence of tumor cells in the peritoneal cavity can be the result of transected lymphatics and blood vessels during the course of surgical resection[72].

Using peritoneal lavage cytology, several studies have aimed to assess the presence of free peritoneal tumor cells in gastric and CRC patients. They relate the presence of these cells to various clinical and pathological parameters, including locoregional recurrence rate and survival[73-77]. A wide range of 2.1% to 52% positive peritoneal cytology is reported across several studies, which reflects the heterogeneity of the techniques and the timing to detect malignant cells in peritoneal washes (Table 2)[45,76,78]. Lloyd et al[75] and Altomare et al[79] performed both pre and post resection peritoneal lavage cytology analysis using polymerase chain reaction. They both report a similar pre-resection rate of positive cytology, 12% to 14%, but differ in the post resection detection rate, 3%[79] and 20%[75]. Recently, Bae et al[76] performed peritoneal lavage cytology in patients with CRC without distant metastasis who underwent curative resection, immediately after making a midline abdominal incision and just before manipulation of the tumor. They reported a rate of positive cytology of 4.1%.

Table 2

Intra-operative peritoneal lavage: detection method, timing and outcome data

Ref.	Patients, n	Method of detection	Peritoneal lavage fluid	Timing of sampling	Intraperitoneal free cancer cells
Kirstensen et al[73]	237	PCR	200-600 mL 0.9% NaCl	After	8.01%
Nishikawa et al[74]	410	Cytology	200 mL 0.9% NaCl	Before	7.60%
Lloyd et al[75]	125	Immunobead RT-PCR	100 mL 0.9% NaCl	Before	12.80%
				After	29.60%
Bae et al[76]	145	Cytology	100 mL 0.9% NaCl	Before	4.10%
Noura et al[77]	697	Cytology	100 mL 0.9% NaCl	Before	2.20%
Altomare et al[79]	29	Thin-Prep	150 mL 0.9% NaCl	Before	13.80%
				After	2.60%
		RT-PCR	150 mL 0.9% NaCl	Before	37.90%
				After	41.40%
Rossi Del Monte et al[78]	48	Cytology	250 mL 0.9% NaCl	Before	0.00%
		Immunofluorescence	250 mL 0.9% NaCl	Before	17.00%
		qRT-PCR	250 mL 0.9% NaCl	Before	42.00%

PCR: Polymerase chain reaction; RT-PCR: Real-time polymerase chain reaction; qRT-PCR: Quantitative real-time polymerase chain reaction.

Peritoneal transport of tumor cells

Once the tumor cells have been detached from the primary tumor and seeded in the peritoneal cavity, they become susceptible to the regular peritoneal transport. Previously, it was assumed that intraperitoneal cancer dissemination was a random process independent of the physical and biological properties of the tumor and the host[80]. However, several studies report that the direction taken by the loose tumor cells and their ultimate destination are dependent on the anatomic site of the primary tumor and the continued cephalic circulation responsible for the clearance of fluid from the peritoneal cavity[81-83]. The latter is due to changes in intra-abdominal pressure during respiration, gravity and peristalsis of the bowel; which results in a clockwise flow from the pelvis, along the right paracolic gutter and to the subdiaphragmatic space and finally towards the pelvis again[40,81]. As a result, certain areas of the peritoneal cavity; the subphrenic region, lesser sac, mesentery, diaphragm and the paracolic gutters, will have an increased risk of occurrence of metastases[80]. Carmignani et al[81] and Hugen et al[82] report the presence of mucinous ascites to be a prominent facilitator of widespread intraperitoneal cancer distribution with colonic mucinous adenocarcinoma and mucinous colonic adenocarcinoma having different peritoneal surface distribution patterns. The presence of peritoneal adhesions and fibrin entrapment resulting from a surgical trauma are also influencing factors in peritoneal transport. Moreover, during the EMT, malignant tumor cells gain migratory and invasive properties that involve a dramatic reorganisation and activity of the actin microfilament system resulting in the formation of actin-rich membrane protrusions: lammelipodia and filipodia. This process is stimulated by pathological expression of growth factors, their receptors and signalling intermediates, which are the products of proto-oncogenes[56,84].

Attachment to the distant peritoneum

The ultimate destination of intraperitoneal dissemination depends not only on the physical and biological properties of the free tumor cells but also on the tissue that will harbour the mestastatic implantation. The attachment of free CRC cells to the distant peritoneum can occur via two processes, denominated transmesothelial and translymphatic mestastasis.

During transmesothelial dissemination, loose tumor cells directly adhere to the distant mesothelium, the innermost layer of the peritoneum. The possible role of adhesion molecules in tumor-mesothelial interactions has been investigated, based on parallels drawn between the mesothelial cell and the endothelial cell[47]. The mesothelium expresses a distinct pattern of adhesion molecules, which are known to play an important role in leukocyte traffic during peritoneal inflammation and are believed to be exploited by invading tumor cells during the peritoneal metastatic cascade[29].

Mesothelial cells express adhesion molecules belonging to the Immunoglobulin Superfamily: intercellular adhesion molecule-1 (ICAM-1), platelet-endothelial cell adhesion molecule-1 (PECAM-1) and vascular adhesion molecule-1 (VCAM-1)[85-87]. Several pro-inflammatory cytokines released following surgery or secreted by circulating tumor cells [tumor necrosis factor-α, interleukin (IL)-1β, IL-6 and interferon-γ] are known to cause a beneficial environment for the tumor-mesothelial interactions[88-93]. These cytokines enhance the expression of the adhesion molecules, ICAM-1 and VCAM-1, on mesothelial cells and induce the contraction of mesothelial cells, thereby exposing the basement membrane[94]. In these areas of absent or rounded mesothelial cells, the interaction between the tumor cells and the laminar network of the basement membrane is mediated through the β1 integrin subunit[95-97]. In an in vitro study, Ziprin et al[98] demonstrated tumor-mesothelial adhesion by an interaction between mesothelial ICAM-1 and tumor expressed CD43 (sialophorin).

The glycosaminoglycan, hyaluronan, is secreted by the mesothelial cells and subsequently assembled into a pericellular coat[37]. First, the hyaluronan coat protects the mesothelium from vital infections and the cytotoxic effect of lymphocytes. Secondly, hyaluronan is involved in tumor-mesothelial adhesion through the interaction with tumor expressed CD44[29,99]. This interaction is an important step in the peritoneal metastatic cascade of ovarian, colon and CRC[100,101]. CD44 is a cell surface glycoprotein, which is widely expressed in several non-neoplastic cells as well as neoplastic cells and is involved in migration of cells, homotypic and heterotypic cell-cell adhesion. The cd44 gene is composed of 20 exons, 10 of which are variably expressed[102,103]. The smallest and most abundant isoform is the standard form, CD44s. Alternative splicing of 10 variant exons, which account for sequences located in the extracellular part of the CD44, results in the expression of CD44v1 up to CD44v10[104]. The variant isoforms CD44v3 and CD44v6 are believed to play a role in the metastatic cascade of CRC. Expression of CD44v6 is largely restricted to the advanced stages (T3/T4) of CRC and was higher in metastatic cancer than in nonmetastatic cancer[102,105]. Saito et al[106] investigated the clinical importance of CD44s and CD44v6 and their relevance to EMT in 113 patients with stage II/III CRC treated with curative surgery. They report that high expression of CD44v6 is an independent poor prognostic factor for disease-free survival and overall survival.

During translymphatic dissemination, loose tumor cells gain access to the submesothelial lymphatics through openings at the junction of two or more mesothelial cells, the lymphatic stomata. Lymphatic stomata are small openings of lymphatic capillaries, which are involved in immunoregulation but most importantly serve as drainage channels for active absorption of fluids and cells from the serous cavities. They can be found in the greater omentum, appendices epiploicae of the colon, the peritoneal side of the diaphragm, falciform ligament, Douglas pouch and the small bowel mesentery[107]. Specialized structures, called “milky spots”, are also found in these anatomical regions, distributed around the lymphatic stomata. Milky spots are immunocompetent cell aggregates, which resorb peritoneal fluid through their lymphatic stomata and mainly serve as gateways for and providers of macrophages for the abdominal cavity[108-110]. They play a role in the formation of PC as they provide a highly vascular microenvironment, which permits early survival of circulating tumor cells. The production of VEGF by the mesothelium in the milky spots also promotes angiogenesis, contributing to preferential tumor growth in the milky spots[111]. However, the precise mechanisms are not well understood. Lopes Cardozo et al[112] investigated the spread of the syngeneic CC531 colon cancer cells in Wag/Rij rats, after inoculation in the abdominal cavity. They observed tumor cells in the milky spots of the greater omentum within 4 h after intraperitoneal inoculation, demonstrating preferential tumor growth in these immune aggregates.

Invasion into the subperitoneal space

For adhered tumor cells to invade the subperitoneal space, they must first penetrate the mesothelial monolayer. This can occur either at areas of peritoneal discontinuity, by invading the intercellular spaces between adjacent rounded mesothelial cells or by destroying the monolayer.

Rounding of mesothelial cells occurs in response to several pro-inflammatory cytokines, thereby exposing the basement membrane[97]. Hepatocyte growth factor/scatter factor (HGF/SF) produced by mesothelial cells induces detachment, motility and proliferation of these cells in the process of mesothelial wound repair[39]. Binding of HGF to its tyrosine kinase receptor, encoded by the c-MET proto-oncogene, initiates an invasive growth program[113]. This program is required during embryonic development for tissue and organ morphogenesis, but is exploited by tumor cells to promote invasive and metastatic ability[114,115]. Sawada et al[116] already reported that overexpression of c-Met is a prognostic factor in ovarian cancer and targeting this receptor in cultured ovarian carcinoma cells inhibited peritoneal dissemination through an α5β1 integrin-dependent mechanism. In CRC, Osada et al[117] investigated the effect of HGF on progression of liver metastasis. They reported that malignancy with high c-Met expression and under high level of HGF, led to unfavourable patient prognosis and poor survival. The presence of malignant ascites was also reported to contain factors, which induce changes in the morphology of the mesothelial cells, resulting in the separation of cell-cell contacts and the subsequent establishment of the characteristic round morphology[94,118].

Destruction of the mesothelial monolayer can occur through tumor-induced apoptosis. Using a three-dimensional in vitro model of the human peritoneum, Jayne et al[94] demonstrated that CRC cell lines rapidly adhered to the outer mesothelial monolayer. The majority of the adhered tumor cells displayed proliferative growth on the mesothelial surface without invasion. A proportion of the tumor cells invaded the mesothelium, which was characterized by apoptosis of the mesothelial cells involving membrane blebbing, cell shrinkage and nuclear fragmentation. Heath et al[119] explored the role of the death ligand/receptor system, Fas Ligand/Fas, in the process of apoptosis using human mesothelial cells co-cultured in vitro with the SW480 colonic cancer cell line. They demonstrated that invasion of the peritoneal mesothelium occurs via tumor-induced mesothelial apoptosis, at least in part mediated by a Fas-dependent mechanism.

After penetrating the mesothelium, the tumor cells adhere to the basement membrane through integrin-mediated adhesion. Integrins are calcium/magnesium-dependent heterodimer molecules, consisting of an α and a β subunit, located on the cell membrane. They are involved in both homotypic cell-cell and heterotypic cell-ECM adhesion and mediate in- and outward signal transduction to the actin cytoskeleton via cytoplasmic proteins[47].

Subsequent invasion of the peritoneal-blood barrier, the submesothelial tissue between the peritoneal mesothelium and the submesothelial arterial blood capillaries, occurs via degradation by proteases[120]. Tumor cells, mesothelial cells, surrounding fibroblasts, inflammatory cells and macrophages secrete matrix metalloproteinases (MMPs), which are responsible for the degradation of several ECM components[121]. Destruction of the peritoneal-blood barrier by these enzymes results from a disturbed equilibrium between the activation of pro- MMPs and their inhibition by tissue inhibitor metalloproteinases (TIMPs)[47]. In CRC, increased levels of MMP-1, MMP-2, MMP-7, MMP-9, MMP-13 and MMP-14 have been reported to play a role in the formation of PC. The MMPs are a family of zinc- and calcium dependent multifunctional enzymes currently comprising 23 members in humans, either membrane-anchored or secreted[122,123]. Many MMPs have overlapping substrate specificity and are involved in a network of mutual activation by MMPs and plasmin activation (Table 3)[124]. Four types of TIMPs (TIMP-1 - TIMP-4) have been reported, which control the activity of the MMPs[125-129].

Table 3

Overview matrix metalloproteinases[124]

MMP	Name	Producing Cells	Substrates
MMP-1	Collagenase	Fibroblasts, synovial cells, chondrocyte	Collagen I, II, III, X
MMP-2	Gelatinase	Fibroblasts, chrondrocyte, mesangium	Gelatin, collagen IV, V, VII, XI
		endothelial cells, cancer cells	Laminin, fibronectin, elastin
MMP-3	Stromelysin-1	Synovial cells, chondrocyte, fibroblasts	Proteoglycan, collagen III, IV, VII, IX, elastin
MMP-7	Matrilysin	Cancer cells, macrophage	Proteoglycan, gelatin, fibronectin
			elastin, collagen IV, laminin
MMP-8	Neutrophil, Collagenase	Neutrophil	Collagen I, II, III
MMP-9	Gelatinase B	Neutrophil, macrophage, thromboblast	Gelatin, collagen III, IV, V
		osteoclast, cancer cells, T-lymphocyte	α2 chain, Elastin
MMP-10	Stromelysin-2	Cancer cells, T-lymphocyte	Collagen III, IV, V, fibronectin, gelatin
MMP-11	Stromelysin-3	Cancer cells, macrophage, mesangium	Fibronectin, laminin, proteoglycan, gelatin
MMP-12	Metalloestelase	Macrophage	Elastin
MMP-13	Collagenase-3	Chondrocyte, cancer cells	Collagen I, II, III
MMP-14	MT1-MMP	Cancer cells, fibroblasts	Collagen I, II, III, gelatin, Laminin
			Fibronectin, vitronectin, proteoglycan
MMP-15	MT2-MMP	Cancer cells, fibroblasts	Fibronectin, aggrecan, tenascin
MMP-16	MT3-MMP	Neuronal cell	Collagen III, gelatin, fibronectin
MMP-17	MT4-MMP	Unknown	Unknown
MMP-20	Enamelysin	Odontoblast	Amelogenin, gelatin
MMP-24	MT5-MMP	Unknown	Unknown
MMP-25	MT6-MMP	Unknown	Unknown
TIMP-1		Tissue, extracellular fluid	Complex formation with pro-MMP-9 and MMPs
TIMP-2		Tissue, extracellular fluid	Complex formation with pro-MMP-9, inhibition of MMP-2 degradation

MMP: Matrix metalloproteinases; TIMP: Tissue inhibitor metalloproteinases.

Elevated expression of MMP1 has been reported in several studies to be correlated with metastasis, reduced overall and/or disease-free survival[129-132]. However, some controversy exists regarding the role of MMP-1. Hettiaratchi et al[133] published a study, including 503 CRC patients and 471 healthy individuals, demonstrating that a single nucleotide polymorphism in the mmp-1 gene promoter resulted in a significantly improved 5-year survival rate.

Groblewska et al[134] suggested that MMP-2 and TIMP-2 play a role in the process of CRC invasion and metastasis. They conducted a study with 72 CRC patients and 68 healthy individuals to assess the serum levels and tissue expression of MMP-2 and TIMP2. MMP-9 has been implicated in the progression, invasion and metastasis of CRC[135,136]. In a study conducted by Alkhamesi et al[137], bidirectional signalling was reported between mesothelial cells and tumor cells in generating cancer invasion. They demonstrated that the interaction of ICAM with its ligand, CD43, plays a vital role in both peritoneal adhesion of tumor cells and the preparation of the right environment for subsequent invasion by increasing the production of MMPs (MMP-2 and MMP-9).

MMP-7 is the smallest member of the MMP family and has been proposed to fulfil a dual role in the progression of peritoneal metastases. On the one hand, MMP-7 can have a potential role in tumor invasion and metastasis by degrading basement membrane and submesothelial components. On the other hand, MMP-7 can promote the development and progression of tumor cells by inhibiting tumor cell apoptosis, decreasing cell adhesion and inducing angiogenesis[138].

Yamada et al[139] reported MMP-13 as a useful predictor of liver metastasis in patients diagnosed with CRC.

Another mediator in the degradation of peritoneal-blood barrier is the urokinase plasminogen activating system, consisting of the urokinase plasminogen activator receptor (uPAR) and the urokinase plasminogen activator (uPA). uPA is a serine protease, which upon activation of the pro-enzyme (pro-uPA) catalyses the reaction in which plasminogen is converted to plasmin. Plasmin is in turn responsible for the degradation of several ECM components and the activation of pro-MMPs[140,141]. The catalytic activity of uPA is controlled by its inhibitors, plasminogen activator inhibitor-1 and plasminogen activator inhibitor-2, through the formation of an enzymatically inactive, trimeric receptor-protease-inhibitor complex[142]. Seetoo et al[143] investigated the expression levels of PAS in a series of human CRC tissues and correlated these results with patient outcome. They report uPA and uPAR to be possible independent predictors of liver metastasis, patient overall survival and cancer-specific survival after resection of colorectal tumors.

Proliferation and angiogenesis

A known hallmark of malignant tumor cells is their ability to trigger proliferation. Sustained proliferation is achieved through the production of growth factors and their receptors by tumor cells and their associated stromal cells, inducing autocrine and paracrine loops[144]. Both the epidermal growth factor receptor (EGFR) and the insulin like growth factor-1 (IGF-1) have been reported to be involved in this process[7,145].

EGFR belongs to the ErbB cell surface receptor family and can be activated by several ligands including EGF and TGFα[146]. Binding of its ligand results in homo- or heterodimerization of various ErbB family members, followed by internalisation of the EFGR receptor complex. Upon autophosphorylation of the EGRF tyrosine kinase domains in the cytoplasmic tails, a transduction signalling cascade is initiated, which in turn regulates tumor cell proliferation, differentiation and survival[147,148]. Yonemura et al[149] reported that gastric tumors with synchronous expression of EGF and EGFR had the highest malignant potential, causing autocrine secretions for self-replication. Ziober et al[150] conducted an in vitro assay with CRC cell lines and demonstrated that TGFα activated autocrine circuits with its receptor, EGFR, and determined growth factor independence of CRC cells[150,151]. Tampellini et al[152] reported that co-expression of immunoreactive EGFR and TGFα was significantly higher in CRC with distant metastases at diagnosis than in CRC presenting at a lower tumor stage.

IGF-1 and its transmembrane receptor are part of a family of cellular modulators that are important in the regulation of growth and development[153]. Varghese et al[7] performed microarray analyses on tumors from patients with CRC metastasized to either the liver or the peritoneum. IGF-1 was exclusively upregulated in tumor samples of patients with peritoneal metastases. Further evidence of the involvement of IGF-1 was provided by Fuchs et al[154]. They reported that increased plasma levels of IGF-Binding Protein-3, an endogenous antagonist of IGF-1, were associated with improved treatment response to first-line chemotherapy and a prolonged time to cancer progression.

Angiogenesis, the growth of new blood vessels from pre-existing vessels, is paramount for tumor growth and the formation of metastases. For their survival, tumor cells rely on the delivery of oxygen from pre-existing blood vessels and nutrients by the recruitment of stromal cells. However, when these tumor cells are located more than 150 μm from the submesothelial capillaries, oxygen and nutrients will not be able to pass the peritoneal-blood barrier resulting in hypoxia-induced apoptosis[155]. Therefore, angiogenesis is induced through the production of angiogenic factors by tumor cells[156,157]. Key players in this process are hypoxia inducible factor-1 (HIF-1) and VEGF.

HIF-1 is a heterodimeric protein composed of HIF-1α and HIF-1β, which activates the transcription of genes involved in the induction of angiogenesis, including VEGF[158,159]. HIF-1β is constitutively expressed and does therefore not depend on the hypoxic status of the cells. The expression of HIF-1α increases exponentially as oxygen levels decline in the cell[160]. In the same study discussed above, Varghese et al[7] reported that HIF-1α was upregulated only in tumor samples from peritoneal metastases. Greijer et al[161] investigated the expression of HIF-1 in normal colorectal mucosa, adenomas and carcinomas. They observed upregulation of HIF-1α in colorectal adenomas and carcinomas. Using immunohistochemistry, Wu et al[162] investigated the clinicopathologic significance of HIF-1 expression in primary human colon cancer. High expression levels correlated positively with an advanced TNM stage and were associated with increased metastatic potential[162].

The VEGF family constitutes five structurally related proteins, VEGF-A, VEGF-B, VEGF-C, VEGF-D and placental growth factor. VEGF-C and VEGF-D are important in the process of lymphangiogenesis, while VEGF-A, VEGF-B and placental growth factor are important in neovascularization[163-166]. The most potent pro-angiogenic growth factor, VEGF-A binds to its receptors VEGFR-1 and VEGFR-2 and thereby increases endothelial cell survival, proliferation, migration and differentiation[167,168]. VEGF-A displays sensitivity to hypoxia and its expression in growing tissue is regulated by HIF[169]. In a study conducted by Shaheen et al[170], nude mice were injected with KM12L4 human colon cancer cells to generate PC. The simultaneous blockage of the VEGF and the EGF receptors with anti-VEGFR (DC101) and anti-EGFR (C225) resulted in decreased tumor vascularity, growth, proliferation, formation of ascites and increased apoptosis of both tumor cells and endothelial cells. Using immunohistochemistry, Logan-Collins et al[171] investigated tumor samples from patients undergoing CRS and intraperitoneal hyperthermic perfusion for mucinous adenocarcinoma of appendiceal or colonic origin. They reported that overall survival was better in patients with low VEGF expression than in patients with high VEGF expression.

ROLE OF THE SURGEON IN PERITONEAL DISEASE SPREAD

The above-mentioned pathophysiology describes the formation of PC as a continuous and interdependent process, known as the peritoneal mestastatic cascade. The surgeon plays a dual role in this process, as promotor of PC by breaching the mesothelium during surgery but at the same time as opponent of PC by performing CRS and HIPEC.

The role of the surgeon as promotor of PC was first described by Sugarbaker[22] in what they called the “tumor cell entrapment” hypothesis. This hypothesis explains the rapid peritoneal disease progression in patients who have undergone surgery as sole treatment. In this setting, the surgeon is responsible for: (1) peroperative seeding of malignant tumor cells originating from transected lymphatics and blood vessels; and/or (2) the dissemination of malignant cells when the tumor is inadvertently ruptured, opened or cut into[25,72]; (3) The peritonectomized surfaces and areas where the peritoneal barrier is disrupted during the course of surgical dissection, provide a favourable niche for the re-implantation of these free tumor cells[96]. The tumor cells become entrapped in the local fibrin deposition present on the traumatised peritoneal surfaces, where they can progress in the presence of growth factors involved in wound healing[172,173]. The fibrin is infiltrated by platelets, neutrophils and monocytes as part of this wound healing process. Using a rat model, van den Tol et al[174] reported that growth of intraperitoneally administered rat colon carcinoma cells was enhanced when injected together with lavage fluid from intra-abdominally traumatized animals. Lee et al[175] investigated whether the wound healing response after an abdominal incision leads to locally increased MMP-9 activity, thereby contributing to peritoneal metastasis. Using a murine model, they concluded that wound-associated inflammation enhances pro-MMP-9 expression. This in turn plays a key role in the growth and progression of tumor cells associated with peritoneal metastases[176].

Port-site recurrences or recurrences situated at extraction site skin incisions have been established concerns since the introduction of diagnostic or cancer laparoscopy[177,178]. First, port-site recurrences are abdominal wall recurrences that occur in the subcutaneous tissue as result of slipping of trocars as well as poor extraction techniques[179,180]. Second, the increase in intraperitoneal pressure during laparoscopy promotes tumor invasiveness and tumor growth via a protease-determined pathway. Using an in vitro experiment, Paraskeva et al[181] reported that the exposure of a human colon carcinoma cell line to a laparoscopic environment enhances the production of MMP-2, MMP-9 and uPA.

The surgeon’s role as preventer of PC involves the administration of intraperitoneal chemotherapy following CRS during the peroperative period. Today’s treatment of colorectal PC involves the combined treatment modality of CRS and HIPEC. The rationale behind the use of HIPEC after CRS is to treat viable circulating tumor cells, residual microscopic lesions, and to eliminate viable platelets, leukocytes and monocytes from the peritoneal cavity. The latter reduces the promotion of tumor growth at traumatized peritoneal surfaces during the wound healing process. Combining CRS and HIPEC for the treatment of colorectal PC has demonstrated encouraging clinical results in several phase II en III trials[16-18]. In a multicentre French trial, Elias et al[182] report an overall 1-year, 3-year and 5-year survival rate of 81%, 41% and 27%.

CONCLUSION

PC is the result of a complex molecular crosstalk between cancer cells and host elements, comprising several well-defined steps, known as the peritoneal metastatic cascade. Individual or clumps of tumor cells detach from the primary tumor, gain access to the peritoneal cavity and become susceptible to the regular peritoneal transport. They attach to distant peritoneum, invade the subperitoneal space, where angiogenesis sustains proliferation and enables further metastatic growth. It is important to realise that these molecular events do not necessarily occur in isolation, but rather describe a continuous and interdependent process. Current treatment combines CRS and HIPEC. This approach is associated with significant morbidity and mortality. A comprehensive understanding of the molecular events involved in peritoneal disease spread, subsequent careful patient selection and knowledgeable management of peritoneal surface metastases is therefore paramount to avoid unnecessary toxicity.

ACKNOWLEDGMENTS

Professor Yutaka Yonemura for granting his permission to use his overview of matrix metalloproteinnases, their producing cells and substrates[124]. Professor Lars Grieten for the illustrations.