Cancer Therapy Vol 3, 299-320, 2005

1Hospital das Clinicas da Faculdade de Medicina da Universidade de Sao Paulo, Sao Paulo (Brasil)

__________________________________________________________________________________

*Correspondence: Paul H. Sugarbaker, M.D., Washington Cancer Institute, 106 Irving Street, NW, Suite 3900, Washington, DC 20010, USA; Phone: (202) 877-3908; Fax: (202) 877-8602; e-mail: Paul.Sugarbaker@medstar.net

Key words: Intraperitoneal chemotherapy, cytoreductive surgery, peritonectomy, appendiceal cancer, colon cancer, gastric cancer, peritoneal surface malignancy

Abbreviations: Completeness of cytoreduction score, (CCS); cytoreduction score, (CC); diffuse peritoneal adenomucinosis, (DPAM); early postoperative intraperitoneal chemotherapy, (EPIC); hyperthermia, (HIIC); long-term intraperitoneal chemotherapy, (LTIC); peritoneal cancer index, (PCI); peritoneal mucinous adenocarcinomas, (PMCA); Prior surgical score, (PSS)

Peritoneal carcinomatosis represents a significant adverse step in the natural history of any gastrointestinal cancer using conventional management. It eliminates any hope of cure and with disease progression leads to poor quality of life. Patients who relapse through peritoneal seeding will experience a succession of complications secondary to the progressive carcinomatosis. Those patients who present with peritoneal implants with the primary cancer in place may suffer not only from progressive carcinomatosis but also from the complications of the primary tumor, most commonly obstruction, bleeding and perforation.

Despite advances for early diagnosis of colorectal cancer, peritoneal carcinomatosis persists as a major problem. Peritoneal implants are present in 10% of patients with colorectal cancer at the time of diagnosis and are the second cause of death after liver metastasis (Sugarbaker, 1990). In contrast to the other two main sites of colorectal cancer metastasis, liver and lymph nodes, peritoneal seeding is considered a condition uniformly lethal with no perspective of cure. From a database of 3019 colorectal cancer patients, Jayne and colleagues identified 349 patients with peritoneal carcinomatosis (Jayne et al, 2002). The median survival of this group was 7 months. Unfortunately this recent data showed no improvement in the survival of these patients if compared with the first study of the natural history of peritoneal carcinomatosis published 13 years before (Chu et al, 1989). Also, a European multicenter trial (EVOCAPE 1) evaluated prospectively 118 patients with peritoneal carcinomatosis arising from colorectal cancer. The mean survival of those patients was 6.9 months (Sadeghi et al, 2000).

In the United States 20 to 30% of patients with gastric cancer being explored for potentially curative resection will be found to have peritoneal seeding at the time of surgical exploration (Sugarbaker, 2003 or 2004). Current standard treatment is systemic chemotherapy which may delay onset of symptoms but is not curative. The median survival of these patients is 5 months with virtually no long-term survivors (Sadeghi et al, 2000). Yoo and colleagues reviewed 2328 patients with gastric cancer who underwent curative resection with at least 5-years follow-up (Yoo et al, 2000). Documented evidence of relapse of the disease was found in 508 patients. Isolated peritoneal recurrence was noted in 34% of patients who relapsed. Hematogenous recurrence occurred in 26% and local-regional persistence of the tumor was seen in 19%. Two or more sites of recurrence were documented in the remaining patients. Serosal invasion and lymph node metastasis were risk factors of relapse in all patterns of recurrence.

This high incidence of peritoneal carcinomatosis following curative resections is shared by others, with an average incidence between 20% and 50% (Gunderson and Sosin, 1982; Koga et al, 1984; Wisbeck et al, 1986; Landry et al, 1990). These data show that in an impressive number of patients the recurrence is isolated within the peritoneal cavity. It also suggests that if an effective treatment could be targeted toward peritoneal dissemination, at least a third of the patients with advanced gastric cancer could experience a better outcome.

Systemic chemotherapy for gastric patients presenting with peritoneal seeding at the time of abdominal exploration or as a manifestation of disease recurrence after a curative surgery is uniformly disappointing. Preusser and colleagues published a response rate for advanced gastric cancer of 50%; nevertheless patients with peritoneal dissemination obtained the worst response (Preusser et al, 1989). Ajani and colleagues, treated patients prior to gastrectomy (Ajani et al, 1991). At exploration, peritoneal carcinomatosis was the most common cause of failure of intensive neoadjuvant chemotherapeutic treatment. Also, radiation showed limited results in this situation and is expected to cause significant morbidity when applied to such a large field.

However, during the last two decades new concepts have arisen with the intent to develop treatment options for carcinomatosis patients. Recognition of the peritoneal layer as the first defense barrier to tumor progression allows one to assume that cure might be possible if tumor nodules were eradicated from all peritoneal surfaces (Sugarbaker, 1990). Also in 1984, Flessner and colleagues showed in pharmacokinetics experiments that the clearance of a drug from the peritoneal cavity is inversely proportional to its molecular weight (Flessner et al, 1984). This new pharmacologic concept, known as Peritoneal-Plasma Barrier, represents the basis of intraperitoneal use of chemotherapeutic agents. In 1995, Sugarbaker published new surgical strategies collectively referred to as peritonectomy procedures (Sugarbaker, 1995). These surgical techniques allow the resection of all parietal peritoneum involved by peritoneal seeding. Visceral peritoneum invaded by tumor may require organ resection.

When the concept of peritonectomy is combined with new pharmacokinetics studies, strategies for the effective management of peritoneal surface malignancies can be formulated for a significant proportion of patients. Accumulated experience shows that success against carcinomatosis from gastrointestinal tumors depends on a coordinated and dose intensive effort. That means maximal surgical cytoreduction plus maximal chemotherapy delivered directly into the abdominal and pelvic cavity during the perioperative period. Unless both strategies are used as a planned part of the surgical intervention little success can be expected.

The most direct test of this new concept of the peritoneum as the first line of defense against peritoneal dissemination is pseudomyxoma peritonei arising from an appendiceal adenoma. It represents a paradigm of this new approach to peritoneal surface malignancies with approximately 85% 5-year survival (Sugarbaker et al, 1999). No other treatments for this group of patients have been shown to be curative. From the remarkable results achieved in appendiceal mucinous tumors using this new combined treatment a rationale for the treatment of peritoneal carcinomatosis from colorectal cancer and gastric cancer emerges. The treatment modalities are further described, selection criteria discussed, results of clinical studies presented and morbidity, mortality and quality of life among treated patients evaluated.

Until recently any patient with peritoneal seeding from a gastrointestinal primary had no surgical option with intent to cure. This was based on the assumption that even if the peritoneal nodules were the only anatomical sites of advanced disease, they were untreatable by conventional surgical techniques. However, the study of the pathobiology of the peritoneal surface component of cancer and also of the patterns of this dissemination allowed the evolution of six surgical procedures for the treatment of macroscopic disease spread on peritoneal surfaces (Sugarbaker, 1995).

An important concept regarding tumor behavior derives from the studies of Weiss. He showed that even though a large number of malignant cells reached the liver through the blood stream only a few adhere to the endothelium and develop into true metastasis. This phenomenon known as "metastatic inefficiency" characterizes hematogenous dissemination of disease (Weiss, 1986). In contrast, free cancer cells in the peritoneal cavity implant and grow with great efficiency. Schott and colleagues evaluated the prognostic significance of isolated tumor cells detected in the bone marrow and in the peritoneal lavage of 84 patients with gastric cancer and in 109 patients with colorectal cancer (Schott et al, 1998). Although cancer cells identified in the bone marrow showed little prognostic significance, free cancer cells in the peritoneal cavity were highly correlated with limited long-term survival.

From observations collected from reoperative surgical procedures, Sugarbaker, (1996) described four patterns of intracoelomic cancer dissemination. In the absence of peritoneal fluid aggressive tumors tend to spread randomly to the peritoneum surrounding the primary tumor. However, tumors that are mucus producing or those that lead to ascitic fluid accumulation, present a characteristic redistributed pattern. The tumor cells tend to adhere at the peritoneal resorption sites in the greater omentum and beneath the diaphragm. Also, gravity is a determinant in tumor distribution since the peritoneal fluid accumulates in the pelvis creating a fluid rich in tumor contamination. Thus, these malignant cells have the opportunity to implant and progress on pelvic peritoneal surfaces.

The third pattern of intracoelomic cancer dissemination was described by Carmignani and colleagues (2003). They documented the influence that peritoneal motion has on tumor distribution. Many structures in the abdomen are stationary, whereas others such as the jejunum and ileum are in continuous peristalsis. When cancer cells are present within the peritoneal space, especially when ascitic or mucoid fluid is present, the malignant cells tend to deposit within non-mobile anatomic sites such as the rectosigmoid junction, around the cecum and the appendix, over the liver and in the subpyloric space. In contrast, the surfaces of the small bowel and its mesentery may remain almost free of disease. This is one of the observations that led to a rationale for peritonectomy procedures and a curative approach to peritoneal surface malignancies.

A fourth mechanism influencing tumor distribution into the abdominopelvic cavity occurs as a result of surgical manipulation and is called "tumor cell entrapment." The malignant cells present in the peritoneal cavity at the time of the first surgery have a tendency to implant on wounded surfaces such as the laparotomy scar, along the bed of the resected primary tumor, or in suture lines. These cells become entrapped by fibrin and are stimulated by inflammatory growth factors released during the healing process. The concept that tumor cell entrapment and enhancement can result from surgical procedures should profoundly affect the practice of cancer surgery (Sugarbaker, 1990).

Considering that a significant proportion of patients with carcinomatosis arising from gastrointestinal cancer have a preponderance of spread on parietal peritoneum and have limited disease over their small bowel surfaces, Sugarbaker described the peritonectomy procedures with the intent to reduce peritoneal surface dissemination to a microscopic level (Sugarbaker, 1995). These procedures are used for resection of peritoneum involved by cancer. On the small bowel surface tumor nodules are electroevaporated and normal peritoneum is spared in order to preserve gastrointestinal function. These techniques are summarized as follows:

• Epigastric peritonectomy: includes any prior midline scar in continuity with preperitoneal epigastric fat pad, the xiphoid process and the round and falciform ligaments of the liver.
• Anterolateral peritonectomy: removes the greater omentum with the anterior layer of the peritoneum from the transverse mesocolon. If the spleen is involved it is resected. The peritoneum of the right paracolic gutter along with the appendix, the spleen and in some patients, the peritoneum covering the left paracolic gutter must be stripped.
• Subphrenic peritonectomy: the peritoneum underneath the right and left portions of the diaphragm is removed along with the left triangular ligament of the liver; this includes the peritoneum lining the retrohepatic space. Glisson's capsule is removed in part or completely as required by cancer implants.
• Omental bursa peritonectomy and lesser omentectomy: begins with the cholecystectomy and is followed by removal of the peritoneum of the porta hepatis, anterior and posterior aspects of the hepatoduodenal ligament, hepatogastric ligament and the peritoneal floor of the omental bursa, including the peritoneum overlying the pancreas.
• Pelvic peritonectomy: removes the peritoneum from the rectovesical or rectouterine space (pouch of Douglas) and usually the rectosigmoid colon must be resected. In women, both ovaries and the uterus are also removed.
• Partial/total gastrectomy: if the subpyloric space is involved an antrectomy may be enough to eliminate the disease. However, especially in mucinous tumors, the entire lesser curvature of the stomach may be encased by tumor and only a total gastrectomy allows complete removal of disease.

Cytoreductive surgery alone can treat gross and macroscopic disease. However, it leaves behind microscopic disease which will progress not only on the peritoneal surfaces, but also within the scar tissues. Residual disease will be stimulated by growth factors released during the healing process (Sugarbaker, 1990). This iatrogenic mechanism for tumor dissemination (tumor cell entrapment and enhancement) is a major cause for both systemic and intraperitoneal chemotherapy failure.

Intraperitoneal chemotherapy was designed to treat the entire abdominal and pelvic cavity using high drug concentrations locally with reduced systemic toxicities. As would be expected, the results of intraperitoneal chemotherapy administered alone for the treatment of established peritoneal metastasis are disappointing. This failure occurs as a result of the very limited penetration of the drugs into tumor nodules. Experimental studies show that only the outer layers of the cancer implants will achieve cytotoxic concentrations of the chemotherapeutic agent (Ozols et al, 1979; Los and McVie JG, 1990). The most optimistic studies suggest a maximum of 3 mm depth of penetration by the cytotoxic drugs (van de Vaart, 1998). These data establish that intraperitoneal chemotherapy should be used only after complete cytoreduction.

Another important limitation for the use of intraperitoneal chemotherapy alone for the treatment of peritoneal surface malignancies is the non-uniform distribution of the drug inside the abdominal and pelvic cavity. The tumor itself may cause extensive adhesions; also, a significant number of these patients have had prior surgical interventions leading to scar formation. Thus, unless all adhesions are taken down one cannot expect a uniform distribution to all peritoneal surfaces (Elias et al, 2000).

Hyperthermia alone already has some cytotoxic properties that can be summarized as follows: inhibition of angiogenesis, induction of apoptosis, denaturation of essential proteins, impaired DNA repair and induction of heat-shock proteins which can be receptors for natural-killer cells (Christophi et al, 1999; Dahl et al, 1999).

Perhaps more important, chemotherapy has been shown to be potentiated by hyperthermia. The ideal temperature for specific chemotherapy agents have not been precisely determined, but it is between 41^oC and 44^oC. Above this temperature the heat will be cytotoxic to non-cancerous cells and cause destruction of the chemotherapy agents (Shimizu et al, 1991).

Hyperthermia decreases interstitial pressure of the tumor nodules to a level expected for normal tissues. It also increases cell membrane permeability allowing a higher drug concentration inside tumor nodules (Storm, 1989; Leunig et al, 1992; Jacquet et al, 1998). Although hyperthermic enhancement of penetration deep into the cancer nodule has only been demonstrated for cisplatin and doxorubicin, it should occur also with other cytotoxic drugs (van der Vaart et al, 1998).

Another benefit of higher temperatures is heat augmentation with selected drugs such as mitomycin C, cisplatin and doxorubicin (Storm, 1989). Some chemotherapy agents such as melphalan, cyclophosphamide and ifosfamide are synergized by heat to a great extent (Urano et al, 1999, 2002; Mohamed et al, 2003; Glehen et al, 2004).

The timing of drug delivery can be classified as preoperative intraperitoneal chemotherapy; intraoperative intraperitoneal chemotherapy, usually associated with hyperthermia (HIIC); early postoperative intraperitoneal chemotherapy (EPIC); and long-term intraperitoneal chemotherapy (LTIC).

Preoperative intraperitoneal chemotherapy has been used by Yonemura (Japan), for the treatment of gastric cancer presenting with established peritoneal metastasis. This group also utilizes HIIC and EPIC in the same patients trying to improve the results in this poor prognosis clinical entity (Yonemura et al, 2003).

The use of intraperitoneal chemotherapy just after the cytoreduction, but before the construction of the anastomosis; HIIC, is the most widely used method (Glehen et al, 2004). In addition to the advantages of the route of administration of the chemotherapeutic agent (intraperitoneal) and the association of hyperthermia, all adhesions are eliminated at this point in the surgery allowing the distribution of the fluid to the whole peritoneal cavity. Another major benefit of HIIC utilized after maximal cancer resection with peritonectomy is related to the reduction of tumor burden to a microscopic level. The cytotoxic drugs can eliminate free cancer cells before they implant deep in scar tissue and away from the effects of both systemic and intraperitoneal chemotherapy.

There are four techniques by which to perform the HIIC: closed, partially closed, open and by a peritoneal expander. In the closed technique the intestinal anastomoses, the fascia and skin of the abdominal wall are closed prior to the HIIC. In the partially closed technique the skin of the abdominal wall is closed with a running suture. After the chemotherapy treatment the abdominal skin is reopened and intestinal anastomoses performed. However, these methods do not guarantee a uniform drug and heat distribution as shown through radiological exams and dye studies (Hughes et al, 1992; Elias et al, 2000). The open technique or "Coliseum technique" (Figure 1) was described by Sugarbaker and is performed by elevating the skin edges of the midline abdominal incision on a self-retaining retractor (Sugarbaker, 1999). This method, in spite of greater heat-loss and the theoretical problem of environmental chemotherapy contamination, has the advantage of allowing the surgeon to manipulate the fluid inside the abdominal cavity in order to optimize heat and drug distribution. Yonemura et al. also described an open technique with the help of a peritoneal expander which, as the coliseum method, permits the surgeon to distribute the fluid (Yonemura, 2003). The peritoneal expander technique is performed through the closure of the proximal and distal portions of the abdominal incision, leaving a central gap for a plastic cylinder. Through this opening the surgeon's hand can manipulate the abdominal contents. The peritoneal expander is attached to a self-retaining retractor. This method was designed to accelerate the heating process inside the abdomen and diminish the heat loss (De Simone et al, 2003).

Some groups use HIIC and also EPIC in the same patient; only a few groups utilize EPIC alone as a preferential route of drug administration (Glehen et al, 2004). HIIC plus EPIC may have some disadvantages when compared with EPIC alone:

• Hyperthermia is usually not used in EPIC;
• Drug distribution depends on gravity;
• This technique is more associated with a greater incidence of wound healing complications.

Figure 1. The "Coliseum technique" for heated intraoperative intraperitoneal chemotherapy.

The delayed delivery of intraperitoneal chemotherapy produces the worst results (Sautner et al, 1994). In addition to the disadvantages presented by the closed technique, in this method the drugs are delivered after adhesion formation. This not only prevents treatment of the entire abdominal and pelvic cavity, but also allows viable tumor cells to become imbedded inside the scar of healing tissue. Trapped in scar tissue and under the stimulus of growth factors released by inflammatory cells, these cancer cells are in a sanctuary site and cannot be reached by chemotherapeutic drugs (Sugarbaker, 1990; Sugarbaker, 1996).

The duration of the peritoneal washing with chemotherapy solution varies greatly. In the HIIC techniques, this period varies between 60 to 90 minutes. The volume and flow employed are also differences encountered among the groups; 3 liters of fluid, with a flow of 2L/minute can be considered an average (Glehen et al, 2004).

In the EPIC technique the drug is administered as fast as possible and remains within the peritoneal cavity for 23 hours. An additional chemotherapy instillation occurs the next day. The EPIC usually is performed during the first five post-operative days. Changes in patient position facilitate drug distribution.

The pharmacokinetics characteristics of intraperitoneal administration of chemotherapeutic agents are of great importance in the choice of drugs. Selected chemotherapy agents have a prolonged retention within the peritoneal cavity. In 1978 Dedrick and colleagues demonstrated that the peritoneal clearance of a drug is inversely proportional to its molecular weight. Further investigation confirmed that certain agents delivered into the peritoneal cavity will maintain higher levels inside the abdomen compared to the plasma levels. This finding is known as the peritoneal-plasma barrier. This is an inaccurate nomenclature; even after extensive stripping of the peritoneal surfaces there are no large changes in the clearance of the drugs. There is an increase in the barrier with complete healing of the abdomen and scar tissue accumulation after peritonectomy procedures (Sugarbaker et al, 1990).

The peritoneal-plasma barrier has an important practical utility since it maintains a high concentration of the drug inside the peritoneal cavity with a lower plasma concentration. This diminishes the systemic toxicities that accompanies cancer chemotherapy treatment. From this data the ideal drug to be used should have the following characteristics: high molecular weight, fast systemic clearance and water solubility. For use as HIIC augmentation by hyperthermia and non-cell cycle specific cytotoxicity is desired.

A large number of chemotherapeutic agents have been used for the treatment of peritoneal surface malignancies arising from gastrointestinal tumors. Mitomycin C is the most frequently used agent. The synergism between hyperthermia and mitomycin C has been demonstrated in the laboratory. Clinically, it has been found to be effective in the control of cancer lines with low growth rates and high chemotherapy resistance such as the mucinous appendiceal adenocarcinomas (Sugarbaker et al, 1999).

Five-fluorouracil is an ideal agent to be used in EPIC, but not in HIIC, since the 5-fluorouracil does not show augmentation with hyperthermia and its cytotoxicity is cell-cycle related (Storm, 1989).

Doxorubicin has proven to be an excellent drug to be used in HIIC. In a rat model, when used at 43^oC, this drug showed a marked deep penetration without increasing plasma concentrations (Jacquet et al, 1998). These positive characteristics make this drug an appropriate choice in many tumors. At high doses and repeated treatments, peritoneal sclerosis is a limiting factor in dose escalation.

Synergism between cisplatin and hyperthermia has been shown in several clinical trials. In animal models this finding was considered to be a consequence of higher and selective uptake of the drug by the cancer cells (Los et al, 1993). Thus, cisplatin clearly deserves special attention from investigators for use in HIIC.

Oxaliplatin, used for several years in Europe, but only recently approved in the United States, was the subject of a pharmacokinetics study by Elias and colleagues (Elias et al, 2002). Their report showed that this agent achieved high tumor and intraperitoneal concentrations without toxic plasma levels. Since this drug has shown an impact in the systemic treatment of advanced colorectal cancer, it must be explored in the treatment of carcinomatosis secondary to colorectal tumors.

When confronted in the operating room with peritoneal seeding from gastrointestinal cancer, the surgeon must make a decision regarding the possible risks and benefits of a definitive treatment versus supportive care. Although this combined treatment has proven to be the only possibility of cure for patients presenting with peritoneal carcinomatosis, one cannot forget the morbidity, mortality and cost that accompany the combined treatment. With the combined treatment strategy a proportion of patients will be free of disease and a second group will experience a better outcome with longer survival. However, as in most gastrointestinal cancer treatments, there will be a group of patients that will be treated with minimal benefit. The disease will continue to progress despite the treatment provided; sometimes these patients will have a worse clinical course due to the inappropriately excessive treatment. In an attempt to offer a suitable treatment for individual patients quantitative prognostic indicators have been developed.

The peritoneal surface malignancy literature supports three pre-operative quantitative prognostic indicators: tumor histopathology, radiological features and prior surgical score. In the operating room after complete exploration of the abdomen and pelvis, the surgeon has another prognostic indicator that is an index to measure the tumor burden within the peritoneal surfaces. This is called the peritoneal cancer index (PCI). Unfortunately, the most reliable prognostic indicator to date can be assessed only after the surgical procedure. The completeness of cytoreduction score indicates the amount of tumor that could not be surgically resected.

The biologic behavior of the peritoneal surface malignancy is a major determinant of prognosis in these patients. Pancreatic adenocarcinomas with peritoneal seeding exemplify one clinical situation in which combined treatment is of no known value due the extremely aggressive behavior of this neoplasm. Gastric cancer patients represent another major challenge to the combined treatment since only those presenting small volume of peritoneal surface disease seems to have an improved life expectancy. Colon cancer patients are in an intermediate position. Even though these patients have a moderate amount of disease, if a complete cytoreduction is achieved, improvement in life expectancy may be observed. On the other hand there are the non-invasive tumors such as pseudomyxoma peritonei arising from an appendiceal adenoma. These tumors may have extensive intraperitoneal accumulation without local invasion, lymphatic infiltration or hematogenous dissemination. Thus, the use of peritonectomy procedures, even in bulky tumors, can eradicate all macroscopic disease and when combined with perioperative intraperitoneal chemotherapy leads to long-term survival.

The occurrence of prior abdominal surgeries greatly influences the likelihood of complete cytoreduction. During surgery, cancer cells free in the peritoneal space have a tendency to adhere to raw surfaces such as sites of surgical dissection. These entrapped tumor cells may now grow along retroperitoneal structures such as the ureters and the vena cava. Abraided and surgically traumatized small bowel surfaces also are favored sites for cancer cells to implant. This deeper non-anatomical cancerous dissemination caused by prior abdominal surgery represents a major cause of incomplete cytoreduction and thus of limited survival. The "Prior Surgical Score" (Table 1) was developed to quantitate the previous surgical trauma and to estimate the likelihood of a complete resection (Jacquet and Sugarbaker, 1996). This scoring system has proved to have a good correlation with prognosis in patients with mesothelioma (Sebbag et al, 2000) and pseudomyxoma peritonei (Sugarbaker and Chang, 1999) treated by cytoreductive surgery and intraperitoneal chemotherapy. Also, primary colon cancer with carcinomatosis may have a better prognosis that carcinomatosis with recurrent disease (Pestieau et al, 2000). Recommendations regarding primary gastrointestinal cancer operations need to change in order to minimize the prior surgical score.

One must conclude that if unexpected carcinomatosis is found, modification of the first surgical intervention is required. For example, if mucinous carcinomatosis arising from a ruptured appendiceal neoplasm is observed in a patient thought to have appendicitis, a minimal dissection should be undertaken. Beyond appendectomy, the surgeon is advised to aspirate the free mucoid fluid and to perform generous biopsies to plan a future definitive combined treatment (Gonzalez and Sugarbaker, 2004). A right colectomy should not be performed to avoid deep entrapment of tumor cells in the groove between the psoas muscle and vena cava. A more common situation is patients presenting with obstructing colonic malignancies with peritoneal seeding. If the patient has a good performance status, the only procedures indicated are a decompressing ostomy and biopsies.

Table 1. Prior surgical score (PSS)

Only the debilitated patient that will not be a candidate for the combined approach should undergo definitive resection.

Pre-operative CT of chest, abdomen and pelvis are performed in invasive tumors such as gastric and colorectal cancer patients to exclude systemic disease. Unfortunately, there is no reliable radiological examination able to predict the intraperitoneal tumor burden or its distribution in invasive tumors. This lack of accuracy occurs as a result of the pattern of the cancerous growth in the peritoneal cavity. In contrast with solid organs such as the liver in which metastasis grow as nodules, aggressive cancerous implants spread along the contours of the peritoneal surfaces. This means that a negative pre-operative CT may have little value in quantitating carcinomatosis in invasive tumors (Jacquet et al, 1993).

In contrast there are non-invasive malignancies for which CT represents a reliable tool to predict the success of a complete cytoreduction. In a retrospective analyses Jacquet et al, (1995) compared patients with mucinous tumors with complete and incomplete cytoreduction. The authors identified two radiological findings that were correlated with incomplete resection:

• Signs of segmental small bowel obstruction
• Tumor masses of 5 cm or greater diameter associated with the jejunum or the upper ileum or their mesentery.

The association of both findings in the same patient indicated a likelihood of less than 5% to achieve a complete cytoreduction. Masses along the terminal ileum are common findings and can be resected with the cecum. These poor prognosis findings suggest a more advanced disease with an invasive component. Also, the nodules associated with the small bowel surfaces reflect the failure of the redistribution phenomenon, usually a result of previous surgical trauma or from end-stage of the disease.

The extent of carcinomatosis has proven to be directly related to the likelihood of a complete cytoreduction and thus with survival. With the intent to quantitate the extent of intraperitoneal disease Jacquet and Sugarbaker described the PCI.

The PCI (Figure 2) is a score obtained in the operating room with surgical exploration after the release of adhesions (Jacquet and Sugarbaker, 1996). The peritoneal space is divided into 13 abdominopelvic regions as seen in Table 2. Then, the largest tumor nodule within each anatomical location is measured. A value that goes from 0 to 3 is given to each of the 13 regions according to the size of the largest nodule found in the region. The number of nodules in each area is not counted but only the size of the largest nodule is registered. Then the PCI is calculated by the sum of the 13 values recorded; the PCI will range from 0 to 39 (13x3).

The PCI has two distinct roles. First, it provides an objective parameter for the surgeon that can be used in the decision-making process regarding the choice of the most suitable treatment for an individual patient. Second, the

Figure 2. The "Peritoneal Cancer Index" for staging peritoneal malignancies.

Table 2. Description of the anatomic structures included in each of the 13 abdominopelvic regions used to calculate the Peritoneal Cancer Index (PCI).

PCI is a tool that facilitates standardization of patients among researches at different institutions for better communication and evaluation of the results obtained with the combined treatment.

Despite the detailed information provided by the PCI, two important caveats are necessary. There are some patients with low PCI but with very guarded prognosis. This usually occurs in aggressive tumors with spread restricted to vital and unresectable anatomic sites such as the porta hepatis, the ureters, or the small bowel.

A second caveat regarding PCI concerns patients with non-invasive tumors that even in the presence of high PCI score can achieve a complete cytoreduction. Pseudomyxoma peritonei patients may have a large PCI but should still have definitive treatment. This finding can be explained in part due to the redistribution phenomenon and small bowel sparing that occurs in non-invasive mucus-producing tumors such as pseudomyxoma peritonei.

With these exceptions in mind, the PCI has prognostic implications in the survival in patients with colorectal, gastric and non-mucinous appendiceal tumors submitted to the combined treatment (Pestieau and Sugarbaker, 2000; Yonemura et al, 2003; Mahteme and Sugarbaker, 2004).

Other scores have been described such as the Gilly peritoneal carcinomatosis staging (Gilly et al, 1994), the carcinomatosis staging of the Japanese Research Society for Gastric Cancer (Fujimoto et al, 1997) or the Dutch simplified peritoneal carcinomatosis assessment (van der Vange et al, 2000). These scoring systems may be simpler but lack the precision when compared with the PCI.

A profound determinant of survival is the volume of tumor remaining after cytoreductive surgery that will be treated with perioperative intraperitoneal chemotherapy. As described by experimental studies and confirmed in clinical trials, when there are large residual nodules intraperitoneal chemotherapy is unable to eradicate peritoneal surface disease. Some treatment centers suggest that residual large nodules present within the peritoneal cavity indicate that intraperitoneal chemotherapy should be avoided (Katz and Barone, 2003).

However, minute nodules may be successfully treated by intraperitoneal chemotherapy. This concept raises the question of what would be a complete cytoreduction. Sugarbaker described the completeness of cytoreduction score (CCS) to measure the amount of disease left behind (Sugarbaker and Jablonsky, 1995). The important information for the CCS is not the number of nodules remaining, but the size of the largest nodules. The cutoff value used in this score to separate complete from incomplete cytoreductions is dependent on the penetration achieved by the chemotherapy into the cancer nodules. Considering these concepts and the pharmacokinetics studies Sugarbaker describes the CCS for mucinous appendiceal malignancy as follows (Figure 3):

• CC-0: no visible tumor
• CC-1: no nodule larger than 0.25 cm
• CC-2: nodules between 0.25 cm and 2.5 cm
• CC-3: nodules larger than 2.5 cm or a layering of cancer at any site.

Patients with no tumor greater than 0.25 cm (CC-0 and CC-1) after surgery are in the group considered as complete cytoreduction since chemotherapy can penetrate these small nodules. However, clinical trials have shown that patients with CC-2 or CC-3 score have a uniformly poor prognosis (Sugarbaker and Chang, 1999; Loggie et al, 2000; Pestieau and Sugarbaker, 2000).

The CCS has proven to be the strongest quantitative prognostic indicator in gastrointestinal cancer patients with peritoneal seeding that were treated with the use of this combined approach independent of the site of origin. This result has been demonstrated by several

Figure 3. Completeness of cytoreduction score (CC score) after cytoreductive surgery.

groups involved in the management of peritoneal surface malignancies in appendiceal, colorectal and gastric malignancy (Loggie et al, 2000; Pestieau and Sugarbaker, 2000; Elias et al, 2001; Yonemura et al, 2003; Pilati et al, 2003; Glehen et al, 2004).

One must consider that the CCS should be individualized according to the site of origin of the primary cancer, considering that the penetration of the chemotherapy inside of a tumor nodule is dependent on the intrinsic interstitial pressure of each malignancy. Thus, for more aggressive neoplasms such as gastric cancer that produce hard peritoneal surface nodules, the cutoff value may be 1 mm or less. More studies are needed to establish precise CCS for the different gastrointestinal cancer sites.

Appendiceal malignancies correspond to approximately 1% of all tumors arising from the large bowel. Three distinct histopathologic tumors have been described as primary neoplasms from the appendix: carcinoid, adenocarcinoma and tumors with features from both, designated adenocarcinoid tumors. Carcinoid tumors account for approximately two-thirds of all appendiceal malignancies, however carcinomatosis from this primary site is exceptional with only one case reported in the literature (Vasseur et al, 1996).

A small percentage of carcinoid tumors have mucus producing epithelial cells associated. This uncommon combined feature is known as adenocarcinoid and represents less than 5% of all appendiceal tumors. Adenocarcinoid tumors of the appendix spread frequently to the peritoneum (Aizawa et al, 2003; Mahteme and Sugarbaker, 2004). Carcinoid syndrome in these patients is usually not seen. The limited survival observed in these patients show that their prognosis is more dependent on the progression of malignant epithelial cells than on the carcinoid component. A recent study in the literature regarding the use of this combined approach for the treatment of patients with peritoneal carcinomatosis arising from adenocarcinoid tumors was reported by Mahteme and Sugarbaker. From 810 patients with peritoneal malignancy of appendiceal origin treated by cytoreductive surgery and intraperitoneal chemotherapy, 22 patients (2.7%) had adenocarcinoid. The treatment regimens evolved during the 22 years of the study. Seven patients treated early in this experience received only EPIC, while thirteen patients received HIIC mitomycin C at the end of the surgical procedure plus EPIC (5-fluorouracil). Carcinoid syndrome was noted in only one patient. Survival rates at 2 and 5 years were 39% and 25% respectively. The PCI and CCS were significant determinants of survival, but the extent of previous surgery was not (Mahteme and Sugarbaker, 2004).

The epithelial tumors of the appendix can be divided into mucus-producing neoplasms and intestinal adenocarcinomas. The last subtype represents approximately 10% of the appendiceal malignancies. Its biologic behavior is similar to colorectal tumors and carcinomatosis from this non-mucinous tumor should be treated as a colonic adenocarcinoma.

The carcinomatosis arising from mucus-producing appendiceal neoplasms are collectively grouped under the term "Pseudomyxoma peritonei" due to the characteristic mucoid ascites. However, these neoplasms do not have a uniform prognosis. Ronnett and colleagues after their study of 109 patients with pseudomyxoma peritonei treated by cytoreductive surgery and intraperitoneal chemotherapy developed a classification with clear prognostic value. According to this classification a minimally invasive and histologically bland pseudomyxoma peritonei occurs in patients with mucoid ascites as a result of ruptured appendiceal adenomas. This type of peritoneal surface malignancy is referred to as diffuse peritoneal adenomucinosis (DPAM). This group of patients presents an excellent prognosis since it is a disease that is restricted to the peritoneal cavity and thus if combined treatment is successfully completed it will lead to cure in 85% of patients. These authors showed, in contrast, an aggressive behavior of the mucus-producing adenocarcinomas referred to as peritoneal mucinous adenocarcinomas (PMCA). There is also a third subtype that is similar to DPAM but with isolated foci of PMCA, the "hybrid type" (Ronnett et al, 1995).

Sugarbaker et al, (1999) reported on the treatment of 385 patients with appendiceal mucinous malignancies with a mean follow-up of 37.6 months. A complete cytoreduction (CC-0 and CC-1) was obtained in 250 patients. They showed approximately 80% 5-year survival, while patients with CC-2 or CC-3 cytoreductions reached only a 20% 5-year survival (Figure 4). As expected, patients with DPAM did better than patients with PMCA or the hybrid subtype. The prior surgical score also showed to be an important statistically significant prognostic indicator. These favorable results are shared by other groups experienced in the use of the combined treatment (Table 3).

Appendiceal malignancies with peritoneal seeding are considered the paradigm for success of the combined treatment. This good outcome occurs as a result of four features that are unique for appendiceal tumors (Sugarbaker and Chang, 1999). First, these tumors demonstrate a wide spectrum of invasion, with the majority exhibiting a noninvasive histology. Second, the appendix has a tiny lumen that is obstructed early in the development of these tumors leading to perforation and release of epithelial cells into the peritoneal cavity early in the natural history of the disease. Symptomatic carcinomatosis will result in treatment prior to lymph node or hematogenous metastasis. Third, the texture of mucinous tumors is compatible with a deep penetration by the chemotherapeutic agent. Fourth, since these neoplasms are restricted to the peritoneal cavity, if the residual microscopic disease is sensitive to the intraperitoneal chemotherapy all the cancer may be eradicated and the patient will survive long term.

Figure 4. Influence of complete cytoreduction (CC-0 or CC-1) in the prognosis of patients with pseudomyxoma peritonei from appendiceal origin treated by peritonectomy procedures and perioperative intraperitoneal chemotherapy. Reproduced from Sugarbaker and Chang, 1999 with kind permission from Annals Surgical Oncology.

Table 3. Literature review of cytoreductive surgery and perioperative intraperitoneal chemotherapy as a treatment for mucinous appendiceal tumors with peritoneal dissemination

Esquivel and Sugarbaker showed, (2001) that patients with relapse of mucinous appendiceal malignancies should be treated again by the combined approach. If a complete cytoreduction is achieved at reoperation then the chance of long-term survival will be similar to the first intervention. Even a third or fourth reoperation may have palliative benefit in selected patients and prolong survival (Mohamed et al, 2003).

The pattern of recurrence after combined treatment using cytoreductive surgery plus EPIC was described (Zoetmulder and Sugarbaker, 1996). From 118 consecutive patients evaluated, only three exhibited metastatic disease. Nine patients developed pleural disease; however, in all of them the pleural cavity had been entered by dissection of the hemidiaphragm in the previous surgery. Two anatomic sites were the most common places for recurrence: small bowel surface and left sub-hepatic space along the lesser curvature of the stomach. Also it was noted that in 52% of the patients that recurred, disease was present within their abdominal scar. Of great importance was to the finding that 60% of all recurrences occurred within suture lines. These data reinforced the need to perform the intraperitoneal chemotherapy prior to the intestinal reconstruction and prior to abdominal closure (Zoetmulder and Sugarbaker, 1996).

The natural history of peritoneal carcinomatosis arising from colorectal tumors is well known and marked by the lack of improvement with any treatment used. In 1989, Chu and colleagues evaluated 45 colorectal patients with peritoneal carcinomatosis and found a median survival of 6 months (Chu et al, 1989). The EVOCAPE-1 study reported a median survival of 6.9 months for these patients (Sadeghi et al, 2000). More recently, Jayne et al, (2002) from Singapore related a median survival of 7 months in patients with peritoneal carcinomatosis of colorectal origin. In these three studies, the 5-year survival was 0%. In general, the authors of phase II studies that employ the combined aggressive approach for the treatment of peritoneal seeding arising from colorectal malignancies, use these historical controls to evaluate their results. The similar median survival seen in these three studies, ranging from 6 to 7 months, the uniform absence of long-term survivors and consistent survival data over a decade of time suggest these studies a reliable historical control group.

Several phase II studies have demonstrated long-term survival in peritoneal carcinomatosis secondary to colorectal cancer when the combined treatment modality is used (Table 4). In 2004, a retrospective multicenter study reported the experience of 28 institutions used cytoreductive surgery and perioperative intraperitoneal chemotherapy (Glehen et al, 2004). The evaluation of 506 patients treated by the combined approach revealed an overall median survival of 32.4 months when complete cytoreduction was obtained. Despite differences in the protocols of the participating centers, the median survival achieved in this retrospective study was approximately 3 times higher than the historical controls. The majority of the researchers combine cytoreductive surgery with hyperthermic intraoperative intraperitoneal chemotherapy alone. In this study two hundred seventy-one patients underwent HIIC alone (53.5%), 123 underwent EPIC alone (24.3%) and 112 (22.2%) were submitted to both HIIC and EPIC treatments. The chemotherapeutic agents used were not similar among the centers and neither were the doses of the drugs employed. In addition, both open technique and the closed system were used to perform the intraperitoneal chemotherapy washing.

Even though a better outcome (never seen before in this group of patients) has been demonstrated by phase II studies, one could suggest that selection bias would be responsible for these good results. Only through phase III studies would this method be definitely validated as standard of care for these patients. In 2003, Verwaal and colleagues published the first prospective randomized trial regarding the combined approach for patients with carcinomatosis from colorectal cancer (Verwaal et al, 2003). In this study, 105 patients were randomly assigned to receive systemic chemotherapy (control group) or cytoreductive surgery plus HIIC followed by the same systemic chemotherapy employed in the control group. After a median follow-up of 21.6 months, the median survival was 12.6 months in the control group and 22.3 months in the experimental group (p=0.032). The mortality in the experimental group was 8%. This first clinical trial established the beneficial effect that the combined treatment might have in peritoneal carcinomatosis arising from colorectal cancer as compared to the standard of care.

Table 4. Results of phase II trials that combined aggressive cytoreduction with perioperative intraperitoneal chemotherapy for the treatment of patients with carcinomatosis of colorectal origin

However, many oncologists desire to have a second randomized trial. Unfortunately, previous experience has shown that the randomization of patients into this kind of study can be very difficult. In 1995, Elias designed a prospective randomized trial involving patients with colorectal peritoneal carcinomatosis. Systemic chemotherapy would be used in the control group to be compared to maximal cytoreduction plus EPIC. He reports that this trial was rapidly abandoned due to the great dissatisfaction with inclusion criteria in six of the first seven eligible patients; the patients were fully aware of the futility of the control arm of the study. Then, the design of the trial was modified. The control group would now receive maximal cytoreductive surgery and systemic chemotherapy while the experimental arm would be treated by cytoreductive surgery and EPIC plus similar systemic chemotherapy. However, during the 4 years of study only 35 patients were enrolled in this phase III randomized study in spite of the 90 patients that were necessary. There were enough eligible patients, but most of them refused to be randomized in the trial. They asked to be enrolled in the phase I-II trials involving HIIC (Elias and Pocard, 2003; Elias et al, 2004).

The experience of this French group reflects a changing proactive behavior of patients who refuse to be randomized in a control group with well-known established poor prognosis. Future trials will probably be designed to standardize the most effective features of the combined approach instead of denying an effective treatment for the control group.

However, one should not assume that this combined treatment is suitable for all patients with peritoneal carcinomatosis arising from colorectal cancer. As in all surgical treatments for cancer it is important to exercise proper patient selection. Several quantitative prognostic indicators have been identified. For colorectal tumors there are two prognostic indicators that have shown to be reliable; the PCI and the CCS.

In patients with colorectal carcinomatosis, Pestieau and Sugarbaker reported that a PCI of 10 or less was associated with a five-year survival rate of 50%; a PCI between 10 and 20 carried a 5-year survival of 15%; a PCI greater than 20 resulted in no 5-year survivals (Pestieau and Sugarbaker, 2000). Elias in 2001, also using mitomycin C as the intraperitoneal chemotherapy agent, reported an improved prognosis when the PCI was lower than 16. Patients with PCI below 16 had a 3-year survival rate of 60.3% versus 32.5% in patients with PCI above this cutoff (Elias et al, 2001). More recently, the same French group using intraperitoneal oxaliplatin combined with systemic 5-fluorouracil and leucovorin used a PCI of 24 to separate patients with good prognosis from those with restricted long-term survival (Elias et al, 2002). These studies demonstrate that the PCI be used to separate patients who are likely to benefit from this aggressive combined treatment from those patients who should be treated with palliative intent. The numerical value for this PCI cutoff must be determined for a particular treatment regimen in a specific carcinomatosis disease state. In other words the critical PCI for colorectal cancer, gastric cancer and appendiceal cancer will all be different. Most likely, the more aggressive the cancer and the more heavily pretreated the patient, the lower the PCI that will be associated with long-term survivors.

The second quantitative prognostic indicator of predictive value in colorectal carcinomatosis in CCS. In the multicentric analysis performed by Glehen, the actuarial survival rate at 1-year, 3-years and 5-years for patients classified as CC-0 or CC-1 were 87%, 47% and 31%, respectively (Glehen et al, 2004). Shen and colleagues reported on 77 colorectal cancer patients treated by the combined approach. The 5-year survival rate for complete resection of all visible disease was 34% with a median survival of 28 months (Shen et al, 2004). In 2000, Sugarbaker reported the results of 104 patients with peritoneal seeding from colorectal cancer. The median overall survival for patients with complete cytoreduction was 24 months with a 30% 5-year survival, whereas patients with incomplete cytoreduction had a median survival of 12 months and 0% 5-year survival (Pestieau and Sugarbaker, 2000).

Other prognostic factors to be recognized in colorectal carcinomatosis patients were the use of neoadjuvant chemotherapy, lymph node involvement, presence of liver metastasis, poor histological differentiation, presence of signet ring cell type and location of the primary tumor in the rectum. The presence of these clinical features has been associated with poor prognosis (Gomez-Portilla et al, 1999; Marcus et al, 1999; Elias et al, 2001; Carmignani et al, 2004; Glehen et al, 2004; Knorr et al, 2004).

Among gastrointestinal cancers discussed in this paper, gastric cancer with peritoneal seeding deserves special attention because of its aggressive behavior and short median survival. In most gastric carcinomatosis patients this new combined treatment should first of all be considered a good palliation; in a small percentage of patients hope for cure may be realistic.

A review of multiple journal articles over the last 20 years supports palliative resections, even total gastrectomy if necessary, as the treatment of choice in selected patients with stage IV gastric cancer (Table 5). Complete removal of the primary cancer has a statistically significant improvement not only in survival, but also a better quality of life for these patients.

The rationale for palliative resection of the primary gastric cancer is to eliminate the serious complications that can result from the primary malignancy. These are most

Table 5. Journal articles since 1980 from English literature supporting palliative gastrectomy