Malignant ascites affects approximately 10% of patients with gastric cancer (GC), and poses significant difficulties for both patients and clinicians. In addition to the dismal general condition of affected patients and the diversity of associated complications such as jaundice and ileus, problems in assessing scattered tumors have hampered the expansion of clinical trials for this condition. However, the accumulation of reported studies is starting to indicate that the weak response to treatment in GC patients with malignant ascites is more relevant to their poor prognosis rather than to the ascites volume at diagnosis. Therefore, precise assessment of initial state of ascites, repetitive evaluation of treatment efficacy, selection of suitable treatment, and swift transition to other treatment options as needed are paramount to maximizing patient benefit. Accurately determining ascites volume is the crucial first step in clinically treating a patient with malignant ascites. Ultrasonography is commonly used to identify the existence of ascites, and several methods have been proposed to estimate ascites volume. Reportedly, the sum of the depth of ascites at five points (named “five-point method”) on three panels of computed tomography images is well correlated to the actual ascites volume and/or abdominal girth. This method is already suited to repetitive assessment due to its convenience compared to the conventional volume rendering method. Meanwhile, a new concept, “Clinical Benefit Response in GC (CBR-GC)”, was recently introduced to measure the efficacy of chemotherapy for malignant ascites of GC. CBR-GC is a simple and reliable patient-oriented evaluation system based on changes in performance status and ascites, and is expected to become an important clinical endpoint in future clinical trials. The principal of treatment for GC patients with ascites is palliation and prevention of ascites-related symptoms. The treatment options are various, including a standard treatment based on the available guidelines, cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (HIPEC), laparoscopic HIPEC alone, intravenous chemotherapy, intraperitoneal chemotherapy, and molecular targeting therapy. Although each treatment option is valid, further research is imperative to establish the optimal choice for each patient.

Malignant ascites caused by gastric cancer (GC) is an accumulation of excess fluid within the abdominal cavity associated with serious clinical problems. First, it is one of the late manifestations of GC, and thus is often accompanied by a severely impaired patient condition[1-3]. A retrospective study of 119 patients with malignant ascites due to GC revealed that 31% of these patients were classed as having as Eastern Cooperative Oncology Group performance status (ECOG-PS) of 3 or more[2]. Generally, such patients have difficulties in receiving standard treatment and show dismal prognosis, with a reported median survival time of 4.6 mo when treated with 5-fluorouracil alone or 5-fluorouracil plus methotrexate[3]. Consequently, these GC patients are often excluded from clinical trials[1] and attract only limited attention with respect to management strategies, thus further veiling the condition and possible treatment of ascites development.

Adequately assessing both ascites volume and treatment efficacy is another clinical problem for patients with malignant ascites due to GC[4]. Because it is difficult to measure the exact nature and extent of disseminated tumors on radiological examination[5], the evaluation of treatment efficacy has empirically relied on changes in the ascites volume. In clinical trials, each protocol had arbitrarily defined the response of ascites to treatment or applied ambiguous definitions of the condition, further complicating inter-trial comparisons. Indeed, the lack of a “standard” method to evaluate treatment efficacy for these patients in daily clinical practice urgently demands the development of a reliable assessment framework.

Despite these documented difficulties, pioneers in this field have successfully conducted several phase II studies or retrospectively reported precious results for specific treatment options, and based on these results, patients and clinicians now have expanded treatment options[6]. For instance, long-term survival was achieved after combination therapy of surgical treatment and chemotherapy among selected patients with ascites and GC, although these reports presented limited patient numbers[7-9]. However, the data has not yet been integrated and selecting the most suitable treatment remains a great burden for both patients and clinicians.

In this manuscript, we first review the relevant literature to elucidate the incidence of ascites development among patients with GC. Then, we introduce recently reported methods to measure ascites volume by computed tomography (CT), and explain a new concept for evaluating treatment efficacy based on patient-oriented parameters. Finally, we discuss each treatment option with respect to future directions.

Data showing the incidence of ascites secondary to GC are scarce and glancing. The development of malignant ascites, an end-stage manifestation of GC, requisitely depends on the tumor stage at diagnosis of primary lesions. Thus, in countries where GC is diagnosed at an earlier stage through validated screening programs[10,11], the incidence of malignant ascites could be relatively low. Contrarily, in countries where less attention is paid to GC due to its lower morbidity, diagnosis of the disease is often delayed until symptoms develop. Therefore, the available literature covers a range of incidence rates of ascites development due to GC across various countries and study periods (Table 1).

Specifically, a retrospective study analyzing more than 7000 patients who underwent gastrectomy in a single Japanese institution from 1960 to 1988 found that 14.2% of the patients developed peritoneal recurrence[10]. Similarly, a Japanese nationwide study in 2009 suggested that 9.9% of 13002 patients who underwent gastrectomy in 2002 died from peritoneal involvement of GC during the 5-year follow-up period[12], while further studies also suggested that approximately 40% of consecutive patients with peritoneal dissemination also showed malignant ascites[13,14]. Thus, these previous data indicated that approximately 4%-5% of all patients undergoing gastrectomy would be subsequently diagnosed with malignant ascites.

More direct evidence on the incidence of ascites comes from a retrospective analysis in a hospital serving a single, well-defined area of Norway[15]. The authors analyzed 354 patients with clear chart descriptions and identified 6.2% (22/354) as having ascites at the diagnosis of GC. A similar, larger scale retrospective study from China identified ascites in 2.6% of GC patients at the time of initial diagnosis, and in 3.7% of patients thereafter[16]. In these Chinese patients, the diagnosis of malignant ascites was confined to cytology-positive cases, therefore the real prevalence of ascites due to GC would be somewhat higher than that reported[2].

Malignant ascites due to GC is often accompanied by other symptoms related to peritoneal dissemination, a particular recurrence mode of GC[17,18] that can cause obstruction of the gastrointestinal tract, bile duct, and ureter. In addition, half of the patients with peritoneal recurrence have concomitant recurrence sites[17], including lymph nodes, liver, and lung, necessitating the systemic evaluation of each site using imaging modalities. We thus consider that diversity among the accompanying symptoms and conditions of GC patients with malignant ascites could induce factors that hinder clear depiction of these patients in the literature and consequently, could obstruct the establishment of reliable guidelines.

Based on the available data, we estimate that 3%-6% of patients with GC has ascites to some extent at the initial presentation of cancer. Eventually, 10%-15% of those patients treated by curative resection will develop peritoneal recurrence and approximately half of them are likely to develop ascites. Thus, 8%-13.5% of the total number of patients diagnosed with GC will have accompanying malignant ascites. However, the incidence of “massive” malignant ascites that defies conventional treatment remains obscure.

Ascites develops from an imbalance between the production and drainage of peritoneal fluid[19-22]. In adults, the serous membrane covers nearly 2 m² of the peritoneal surface[23], and the cavity typically contains 50-100 mL of fluid that turns over at the rate of 5 mL/24 h[24]. Peritoneal fluid is generated by the transudation of plasma from peritoneal capillaries[19], and it serves to lubricate the serous membrane. The fluid eventually drains into the lymphatic system via open-ended channels (named stomata) and then into the systemic circulation through the right thoracic duct[24].

Among multiple factors, increased vascular permeability due to vascular endothelial growth factor (VEGF) is considered an important driver of increased ascites production. Zebrowski et al[25] demonstrated markedly elevated VEGF levels in malignant ascites including from GC. They further showed augmented permeability of endothelial cells in vitro when cultured with malignant ascites. Another important factor in the development of ascites is matrix metalloproteinase-9 and -2, a key enzyme in tumor cell metastasis to distant organs due to its role in breaking down the tissue matrix. Reportedly, matrix metalloproteinase enhances the release of biologically active VEGF in a time- and dose-dependent manner, and might thus by a key regulator of VEGF in ascites production[26]. In fact, inhibition of matrix metalloproteinase significantly suppressed tumor growth in a rodent model of metastasis[27]. Several other factors could also play a role in the development of malignant ascites during cancer (Table 2), and each could be a potential target of prevention and treatment.

Objectively evaluating the nature and volume of ascites is the first step in treating patients with ascites and peritoneal dissemination of GC. Cytology should be always considered at the initial evaluation because a positive result is diagnostic, while increased levels of carcinoembryonic antigen in ascites suggests the pathological accumulation of peritoneal fluid due to GC[2]. However, it should be noted that the sensitivity of these two measures is relatively low[2], and negative results warrant an integrated evaluation of the ascites based on a range of clinical data.

Although the relationship between ascites volume at diagnosis and prognosis remains controversial[14,28], a weak response of ascites to the anti-cancer treatment is well correlated with poor prognosis[1,14], suggesting that frequent and repetitive volume assessment is particularly important for decision making concerning continuation or withdrawal of the ongoing treatment.

In this field, the use of endoscopic ultrasonography is increasingly reported[29,30], because of its excellent ability to detect subtle ascites. This modality would therefore be especially beneficial for predicting prognosis in patients with GC, based on patients with ascites apparently having poorer outcomes than those patients without ascites[16,31-33]; however, the technique requires considerable expertise and is invasive for patients. Abdominal ultrasonography is often used in the emergency room and daily clinical practice due to its convenience, and thus is likely to first detect ascites. Recently, ultrasonography technologies allowed development of much smaller devices, which will eventually eliminate unnecessary further confirmatory examination with invasive modalities[34].

Indeed, as early as 1996 Inadomi et al[35] developed a protocol to measure ascites volume with ultrasonography, in patients with portal hypertension. They regarded the ascites in the abdominal cavity as a fluid retained in the base of a large sphere, and developed an algorithm accordingly using two variables: the ascites depth, defined as the distance to the deepest point of the ascites, and abdominal circumference. The calculated value proved to be well correlated with the ascites volume determined by distribution (dilution) of radiolabeled tracer. However, this method had the remaining problem that patients needed to remain in a prone position on their hands and knees for 10 min, and it is not clear whether it could be used in patients with postoperative intraperitoneal adhesions.

More recently, Irshad et al[36] reported notable findings of correlation between the smallest depth of ascites on ultrasonography and the subsequent drained volume of ascites. They conducted 60 paracenteses in 29 patients after evaluating the length between the most superficial bowel loop and the abdominal wall. They found that the length measured by ultrasonography was well correlated to the amount of drained fluid, and concluded that ultrasonography could successfully estimate the ascites volume. Establishment of a validated volume-measuring method based on ultrasonography would obviously be a great benefit for patients who require frequent monitoring, and thus further development of these initial studies are eagerly awaited.

The recent development of multi-detector CT permits small amounts of ascites to be detected[31,32] and thus imaged by reconstruction of three-dimensional imaging. Although a volume-rendering algorithm applied to such imaging would enable accurate assessment of ascites volume, the procedure requires an appreciable amount of time regardless of the operator expertise. To reduce the burden, Oriuchi et al[4] developed a very simple method to estimate ascites volume using the horizontal plane of CT imaging. In this technique, ascites depth was measured at five points on three CT images and the sum of measurements for each patient was multiplied by 200 (Figure 1). The estimated ascites volume was then correlated with the volume calculated with 3D-CT in patients having > 300 mL of ascites. The authors reported that this protocol was reliable even in patients with a history of surgical intervention that might cause changes of their ascites distribution due to adhesion. The accuracy of this method was confirmed by two following studies: one for assessing the ascites due to GC[1] and one for assessing ascites due to a perforated peptic ulcer of the upper gastrointestinal tract[37].

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Five-point method to measure ascites volume. Upper: Line between the bilateral antero-posterior mid-points of the abdominal wall is drawn at the plane of the root of the superior mesenteric artery. The distances between the inner surface of the right abdominal wall and the liver (A cm), and between the inner surface of the left abdominal wall and spleen (B cm) are obtained. When spleen is not observed on this plane, the distance between the left abdominal wall and margin to the ascites and internal organs are measured (B’ cm); Middle: The lower pole of the left kidney is observed on this plane. The sagittal line from the bilateral paracolic gutter, and between the bilateral antero-posterior midpoints of the abdominal wall is drawn. The distances C (cm) and D (cm) are thus obtained. Lower: A line between the anterior sides of the bilateral femoral artery is drawn. The distance between the inner surface of the abdomen (at the middle) and the line is obtained (E cm). The ascites volume is calculated by the equation of (A + B + C + D + E) × 200 (mL).

Of note, the horizontal plane of CT imaging at the level of the superior mesenteric artery occasionally does not depict the spleen, and ascites can occur in patients who undergo splenectomy with gastrectomy. In such cases, volume assessment could instead use the distance from the inner surface of the abdominal wall to the surface of an internal organ at a defined line (Figure 1). This altenative method is also reliable for assessing chronological changes in ascites volume (personal communication with Dr. Oriuchi). Despite the apparent limitation of only a small number of patients examined thus far and the lack of explicit evidence for using an alternative measuring parameter rather than the standard one of the distance between spleen and abdominal wall, this CT-based method is fascinating and warrants further exploration.

Because the peritoneal dissemination sparsely distributes in the abdominal cavity rather than forming a large mass, patients with malignant ascites often lacks measurable lesions. In addition to the ascites volume, the assessment of treatment efficacy in such cases has to be based on the integrated clinical information[38]. Standardized evaluation methods are required to select the patients who should continue the current treatment, and identify the patients who should receive other treatments because of disease progression. The candidate for the clinical endpoint of the treatment could be an improvement of quality of life (QOL) measured by established questionnaires. However, the assessment is often difficult to achieve due to its complexity; Badgwell et al[39] courageously demonstrated the difficulty of conducting self-reporting assessment of QOL using Functional Assessment of Cancer Therapy-General (FACT-G)[40]. Only 39% of the patients with incurable cancer and gastrointestinal obstruction completed the questions concerning quality of life and treatment satisfaction at one-month follow up. Based on our similar clinical experiences and observations, we noticed the need for new convenient method to evaluate the treatment efficacy for malignant ascites due to GC, and “Clinical Benefit Response in GC (CBR-GC)” was proposed[1,41].

CBR is an established palliative endpoint for gastrointestinal tract malignancy that is particularly applied in clinical trials for the treatment of pancreatic cancer[42-44]. CBR is based on three clinical factors: the change in pain, Karnofsky PS, and body weight, although because of the hierarchical structure of CBR, its use is slightly complicated and vulnerable to missing data[44]. In contrast, CBR-GC is a newly proposed concept for evaluating the response of malignant ascites due to GC to anti-cancer therapy. This evaluation system has two major components: (1) change in ECOG-PS; and (2) change in ascites evaluated using the patient-oriented method.

Change in ascites (response) with treatment (lateral axis of Figure 2) is determined by abdominal girth and intensity of palliation, which comprises the use of diuretics and paracentesis. When the abdominal girth decreases in volume and the intensity of palliation is not increased, the response of ascites is termed “Positive”. Meanwhile, either or both an increase in abdominal girth and increased intensity of treatment is classed as a “Negative” response to treatment. Comprehensive judgment is then based on the general assessment of ascites response and ECOG-PS. Positive CRB-GC is defined by an improvement of either or both ascites and ECOG-PS without deterioration of any parameter. If the CRB-GC is not positive, the status is determined as negative.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Clinical benefit response in gastric cancer. CBR-GC is defined by the ascites response to treatment (horizontal axis) and ECOG-PS (vertical axis). Response of ascites is judged by a combination of abdominal girth and treatment intensity. CBR-GC: Clinical benefit response in gastric cancer; ECOG-PS: Eastern Cooperative Oncology Group performance status.

When CRB-GC was used to evaluate the efficacy of paclitaxel treatment in patients with malignant ascites due to GC, 39.1% had a positive CRB-GC, suggesting that CRB-GC could serve as a prognostic predictor in such patients. The median survival time of patients with positive CRB-GC was 9.9 mo, while that of patients with negative CRB-GC was 3.6 mo (P value not shown).

The possible limitation of CRB-GC is that subtle alleviation of ascites-related symptoms, which could be of real importance for patients, might be overlooked when solely based on these two factors of ascites response and ECOG-PS, when clinically, symptom improvement should signal continuation of the treatment to prevent deterioration in the patient’s condition. However, CRB-GC could become negative when both ascites and ECOG-PS are “stable”. Thus, further study is necessary to clarify whether this new evaluation system is adequately correlated with subtle symptom alleviation. As suggested by the authors, the relationship between improvement in patient quality of life and CRB-GC status should be addressed. Additionally, the abdominal girth seems precarious compared to the volume measurement by CT, and the threshold of decrease or increase in abdominal girth has to be clearly defined. Although, the authors found that the 5% of possible error did not change the primary results, the exact threshold of change in abdominal girth that determines “no change” should be verified.

Choosing a treatment option for GC with malignant ascites also relies on the patient’s general condition. Despite the presence of ascites, standard treatment as per the guidelines[45-47] should be the first choice when the patients are motivated enough and in a good general condition without insufficiency of vital organs. While many alternative treatment options have been reported, evidence-based guidelines for each remain to be developed.

Anti-cancer drugs administered into the peritoneal cavity penetrate the tumor nodules by passive diffusion. Because penetration depth is limited[64], such intraperitoneal administration is not sufficient to treat larger nodules, and combinations of intravenous and intraperitoneal administration of anti-tumor drugs have been attempted to treat such cases. Cisplatin is an effective agent to treat GC when it is administered intravenously. However, intraperitoneal administration of cisplatin is not a common practice due to its proven lack of benefit as adjuvant chemotherapy[65], probably due to immediate clearance from the peritoneal cavity. To prolong the effect of cisplatin within the abdominal cavity, a new drug delivery system has to be developed. In this context, the commonly applied agents are taxanes. For instance, a phase II randomized trial comparing intravenous and intraperitoneal paclitaxel administration has been implemented by Kodera et al[66] (UMIN000002957) under the supervision of Ministry of Health, Labor and Welfare as an advanced medical treatment project of the government.

For treatment of patients with malignant ascites due to GC, Kitayama et al[14] evaluated the efficacy of synchronous administration of intravenous and intraperitoneal paclitaxel, and oral administration of S-1. The study enrolled 33 patients with ascites due to GC, 9 of which had more than 2500 mL of ascites before treatment. After the initiation of the treatment, 70% of the cases showed > 50% reduction in ascites volume and an associated improvement in prognosis, with a median survival time of 455 d. Another study from the same group further showed that the combination treatment could be safely followed by curative resection (gastrectomy) in selected patients[7]. In such cases, the one-year survival rate was 82% with a median survival time of 26.4 mo. Meanwhile, the patients with refractory ascites against treatment had a median survival time of 12.1 mo. As discussed by the authors, the benefit of salvage gastrectomy remains unclear; however, such treatment is apparently beneficial for selected patient because 5-year survival could be achieved by this sequential treatment. The frequent adverse events for this regimen were neutropenia and leukopenia (25% of patients), while occlusion or infection of the access port is also conceivable with the intraperitoneal chemotherapy.

HIPEC after cytoreductive surgery has received increasing attention due to its efficacy in treating peritoneal dissemination of GC[67,68]. We speculate that many of the patients in these cited studies had some degree of malignant ascites; however, there was little focus on the efficacy of HIPEC with/without cytoreduction in patients with significant amounts of malignant ascites. Yang et al[8] explored the clinical benefit of this combination therapy in 28 patients with GC and heavy ascites or peritoneal carcinomatosis. Of these patients, 20 had ascites and positive cytology, and the detailed information of 12 patients with ascites was available. These patients died at 2, 3, 8, 8, 9.5, 9.5, 10.5, and 29.5 mo after surgery, and survival was confirmed at 3, 5, 9, and 19 mo, yielding a median survival time of 9.5 mo (95%CI: 7.6-11.4 mo). Although the beneficial effect of this treatment seems marginal for the majority of patients, long-term survival was observed and this is generally difficult to achieve with other modalities. Unfortunately, the study did not elucidate the preoperative factors significant for long-term survival, and thus patient selection would become an important issue to be solved. In addition, the related complications and even mortality associated with this combination regime might attenuate its attractiveness.

Another of HIPEC demonstrated that only 14 out of 45 patients with peritoneal dissemination of GC could undergo optimal cytoreductive surgery, (meaning no visible tumor residue or only residual nodules < 2.5 mm). The authors performed HIPEC only in these 14 patients because the penetration of chemotherapeutic agents into tissue nodules is limited and they considered that performing HIPEC in patients with large residual nodules would not be beneficial[9]. Consequently, the optimal treatment group with cytoreduction and HIPEC showed longer median survival than cytoreduction only (median survival time of 18 mo vs 6 mo, P = 0.0007). They also found that the preoperative risk factors of incomplete cytoreduction were retention of ascites and preoperative malnutrition (prognostic nutrition index). Indeed, attempting to perform cytoreductive surgery with HIPEC on patients with GC and ascites is in and of itself a challenging task.

Because cytoreductive surgery and HIPEC could be too damaging for patients with GC and an already poor general condition, reduced invasiveness was attempted by using laparoscopic intraperitoneal hyperthermic chemotherapy. Reports of this modification are still scant and the survival benefit remains unclear[69-71]; however, the recurrence of ascites development was suppressed in the majority of studied cases and thus exploration of this treatment might be justified.

Molecular targeting therapies such as bevacizumab[72], cetuximab[73,74], panitumumab[75], and everolimus[76] have not shown significant survival benefit in patients with GC. Thus, these agents are unlikely to be used vigorously for malignant ascites secondary to GC. Trastuzumab is a humanized monoclonal antibody targeting human epidermal growth factor receptor type 2 (HER-2), and some clinical trials demonstrated the efficacy and safety of combination therapy with trastuzumab, cisplatin, and fluoropyrimidines[77,78]. Although the protocols used did not exclude the existence of ascites, data concerning ascites response was seldom described. An ongoing phase II trial focusing on patients with unmeasurable lesions of HER2-positve GC (UMIN000007941) may give some further indication for this treatment of malignant ascites.

Recently the clinical benefit of catumaxomab was reported. This chimeric antibody of mouse-derived anti-EpCAM Fab region and a rat antiCD3 Fab can recognize epithelial cell adhesion molecule-expressing tumor cells, CD-positive T cells, and Fcγ receptor-positive immune cells, and improved the quality of life in patients with ascites due to several kinds of malignancies compared to best supportive care[79,80]. Because this treatment did not prolong survival time, catumaxomab should be considered only in patients whose cancer is difficult to treat by conventional regimes. Meanwhile, the efficacy of combination therapy with catumaxomab and chemotherapy, which has minimum toxicity, such as monotherapy of taxanes, should be explored for severely deteriorated patients with GC and malignant ascites, as suggested by Imamoto et al[1].

Malignant ascites is a common manifestation of end-stage GC, affecting approximately 10% of patients. Because prognosis in these patients can be predicted during treatment by changes in the ascites volume, repetitive and objective evaluation of such volumes is critically important to maximize patient outcomes. Meanwhile, CBR-GC can assess the treatment efficacy based on changes in ECOG-PS and ascites due to anti-cancer treatments, and should be used for such assessments in future clinical trials.

For patients with malignant ascites due to GC, a guidelines-based standard treatment should be the first considered. In many cases, however, an alternative treatment has to be chosen from the wide range of options, based on the patient’s general condition, their understanding of the disease, and support from personal circumstances. Despite the progress in this field, the best supportive care is often the only thing that can be offered to patients suffering from end-stage GC. Thus, more understanding of this condition and development of evidence-based treatment strategies, together with new treatment options, are necessary.

We greatly thank Dr. Tsutomu Namikawa for kind advice in the preparation of this manuscript, and Ms. Rieko Yamaguchi for excellent data management.