GYNECOLOGIC ONCOLOGY ARTICLE NO.

64, 8087 (1997)

GO964529

A Randomized Trial of Paracentesis plus Intraperitoneal Tumor Necrosis Factor-a versus Paracentesis Alone in Patients with Symptomatic Ascites from Recurrent Ovarian Carcinoma*
H. W. HIRTE,1 D. MILLER,2 K. TONKIN,3 B. FINDLAY,4 V. CAPSTICK,5 J. MURPHY,6 R. BUCKMAN,7 J. CARMICHAEL,8 M. LEVINE,9 AND W. HILL10
1 Ontario Cancer Treatment and Research Foundation, Hamilton Regional Cancer Centre, Hamilton, Ontario; 2Cancer Control Agency of British Columbia, Vancouver, British Columbia; 3Ontario Cancer Treatment and Research Foundation, London Regional Cancer Centre, London, Ontario; 4 Hotel Dieu Hospital, St. Catharines, Ontario; 5Cross Cancer Institute, Edmonton, Alberta; 6Toronto Hospital, Toronto, Ontario; 7 Ontario Cancer Treatment and Research Foundation, TorontoSunnybrook Regional Cancer Centre, North York, Ontario; 8 Ontario Cancer Treatment and Research Foundation, Kingston Regional Cancer Centre, Kingston, Ontario; 9 Ontario Clinical Oncology Group, Hamilton Ontario; and 10Knoll Pharma Inc., Markham, Ontario, Canada

Received March 6, 1996

Purpose: Previous phase I and II studies of intraperitoneal recombinant human tumor necrosis factor-a (rhTNF-a) suggested a high degree of efcacy in reducing or eliminating ascitic uid. To more accurately determine the efcacy of this agent, the role of paracentesis versus paracentesis plus intraperitoneal rhTNF-a was studied in a randomized trial. Patients and methods: Thirty-nine patients with symptomatic ascites with a volume of 1000 ml from recurrent epithelial ovarian carcinoma or primary peritoneal carcinoma, which was refractory to standard therapy, were randomized either to receive 0.06 mg/m2 rhTNF-a (Knoll, Canada) (the dose determined optimal from phase I and II studies) intraperitoneally after drainage of uid or to receive drainage alone. A maximum of three treatments were given at weekly intervals. Eighteen patients were randomized to receive rhTNF-a. Results: None of 18 evaluable rhTNF-a patients had either a complete response (CR) (no clinical evidence of ascites and 400 ml of uid on ultrasound) or a partial response (PR) (asymptomatic ascites and 1000 ml of uid on ultrasound). There were no CRs or PRs in the 17 evaluable patients who received drainage alone. The intraperitoneal infusion of rhTNF-a was generally well tolerated. Moderate to severe toxicity consisted of pain/discomfort in 42.1%, fever/ chills in 36.9%, nausea/vomiting in 10.5%, edema in 10.5%, and hypotension in 5.3% of patients receiving rhTNF-a. Conclusion: rhTNFa, as given in this study, was not effective in preventing recurrence of ascites in this patient population. 1997 Academic Press

platinum-resistant, further chemotherapy is usually ineffective and regular paracenteses are often required to control the symptoms of ascites accumulation. However, in two nonrandomized studies in ovarian carcinoma patients with malignant ascites refractory to conventional therapy, recombinant human tumor necrosis factor-a (rhTNF-a) infused intraperitoneally resulted in complete disappearance of ascites in 31 of 32 patients, usually with 1 to 3 infusions of rhTNF-a [1, 2]. To evaluate the efcacy of intraperitoneal (ip) rhTNFa in such patients, a randomized study comparing paracentesis alone to paracentesis plus ip rhTNF-a was performed.
PATIENTS AND METHODS

INTRODUCTION

Despite optimal surgery and chemotherapy, the majority of patients with advanced epithelial ovarian carcinoma develop recurrent disease, often complicated by the development of malignant ascites. In patients whose tumors are
* Supported by Knoll Pharma Inc.
0090-8258/97 $25.00 Copyright 1997 by Academic Press All rights of reproduction in any form reserved.

Thirty-nine patients with histologically proven epithelial malignancies of the ovary with symptomatic ascites due to recurrent disease were studied. They had an estimated ascites volume of greater than 1000 ml and had received at least one course of platinum-based chemotherapy and also fullled the following eligibility criteria: (1) life expectancy over 3 months, (2) age 18 years or older, (3) Karnofsky performance status 60, (4) no more than two different previous chemotherapy regimens, (5) no radiation therapy, major surgery, or any form of chemotherapy or immunotherapy within the previous month, (6) serum creatinine 166 mmol/liter, (7) serum bilirubin, SGOT, and SGPT 2.5 times upper limit of normal, (8) hemoglobin 95 g/liter, leukocytes 3.5 1 109/liter, and platelets 100 1 109/liter, (9) no evidence of other severe nonmalignant disease, (10) no evidence of bowel obstruction, (11) no loculation of ascites or mucinous ascites, and (12) were able to give informed written consent. Glucocorticosteroids and antithrombotic or thrombolytic drugs could not be administered during the course of the
80

TUMOR NECROSIS FACTOR-a IN OVARIAN CARCINOMA

81

study and patients on diuretics or nonsteroidal anti-inammatory medications had to be on a stable dose of these medications for at least 3 weeks prior to entering the study. A nonrandomized open label protocol was instituted for patients randomized to the control arm who subsequently developed progressive ascites. These patients were subsequently treated according to the protocol outlined for the patients randomized to the rhTNF-a arm of the randomized protocol. There were a total of 16 patients entered on the open label protocol. Drug Dosage and Administration Patients with clinically symptomatic ascites fullling the entry criteria were randomly assigned to one of two different treatment modalities (see Fig. 1): Group A, paracentesis followed by ip administration of rhTNF-a; and Group B, paracentesis alone. Patients weight and abdominal circumference were noted and a catheter was placed into the abdominal cavity (by ultrasonic guidance if required). Ascites was drained as completely as possible (either by gravity or using a vacuum bottle), and the volume drained was recorded. The volume of residual ascites was then estimated by ultrasonography using the recommendations of Goldberg [3] (see ultrasound response criteria below). Patients assigned to Group A then received an infusion of rhTNF-a via the catheter used for drainage. Lyophilized rhTNF-a was provided by Knoll Pharma Inc. (Markham, Ontario). The contents of a rhTNF-a vial were diluted to 1.0 ml in sterile water for injection and the patients specic dose (0.06 mg rhTNFa/m2 BSA) was further diluted in 1000 ml normal saline containing 0.5% (i.e., 25 ml of 20%) human serum albumin to prevent nonspecic adhesion of rhTNF-a to the infusion system. The solution was infused over 2 hr. In order to ensure a uniform dispersion of rhTNF-a solution, the infusion took place with patients in the supine position and patients were turned from side to side every 30 min for 2 hr postinfusion. Group B patients received only paracentesis. Patients were hospitalized on the day of treatment and remained under clinical surveillance for at least 6 hr after the start of therapy. As prophylaxis against the known ulike side effects of rhTNF-a, oral acetaminophen (650 mg) was administered 20 min prior to treatment. Nausea and/or vomiting were controlled by the therapeutic use of commonly used antiemetics (excluding glucocorticosteroids). Hypotension was treated with normal saline infusions or inotropic drugs, when deemed necessary. Patients were treated at weekly intervals up to a maximum of three treatments (see Fig. 1). One week after the rst treatment (Day 8), patients were reexamined. Patients were examined for evidence of ascites both clinically and by ultrasound. Any patient with uid present on ultrasound had the uid drained. Residual uid was then estimated by ultrasound. The drained volume plus the estimated residual vol-

ume were added to give the total ascites volume. Patients with drainage not performable due to technical reasons had their total uid estimated by ultrasound. All patients from whom any uid could be drained received their second treatment according to their randomization group. Those patients with no drainable ascites proceeded to follow-up and were reexamined 3 weeks after their assessment (Day 29) (see Fig. 1). This procedure was repeated 1 week following the second treatment (Day 15). All patients from whom any uid could be drained received their third treatment according to their treatment allocation. Patients with no drainable ascites proceeded to follow-up and were reexamined 3 weeks after their assessment (Day 36) (see Fig. 1). Patients having received a third treatment were reexamined 4 weeks after this treatment (Day 43) (see Fig. 1). All patients were reexamined at the end of their follow-up period (4 weeks after their last treatment). Again, patients were examined for evidence of ascites, both clinically and by ultrasound. Patients with uid present on ultrasound were drained and the residual uid volume was estimated by ultrasound. Patients in whom drainage was not performable for technical reasons had their total uid volume estimated by ultrasound. Patients were then assigned to the response or failure group according to the denitions given below. A clinical assessment was conducted by the investigator and a second physician not involved in the trial. Patients not responding to therapy after three treatments or with a relapse within their follow-up period could receive other therapy at the investigators discretion once evidence of treatment failure had been documented. Otherwise, no additional chemotherapy was allowed until the end of the follow-up period. Patients were clinically reevaluated for signs of delayed response or relapse of ascites at 3 and 6 months after their last cycle of therapy. Evaluation of Efcacy To be evaluable for efcacy, patients were examined 4 weeks after the last treatment or at the time of relapse of the ascites. Patients who were not evaluable at 4 weeks after last treatment (i.e., death or loss to follow-up) and with no information about a relapse were regarded as lost cases and evaluated separately. The time between the last treatment to 4 weeks after the last treatment was dened as the followup period. Response Criteria (i) Clinical response. (1) Complete responseClinical (CRc): Lack of accumulation of ascites at the end of the follow-up period as determined by clinical observation. (2) Partial responseClinical (PRc): Minimal uid recurrence, at the end of the follow-up period as determined by clinical observation (asymptomatic ascites).

82

HIRTE ET AL.

FIG. 1

Treatment schema.

In order to minimize bias, a clinical assessment (abdominal girth measurement, assessment for the presence of ascites and estimation of ascites uid volume) was conducted at baseline and at the end of the follow-up period (or at time of relapse) by a second physician who was unaware of the treatment received by the patient. In the event of a discrepancy between the trial investigator and the second physician, the assessment of the second physician was used to determine response. (ii) Ultrasound response. Ultrasound examinations were performed in accordance with the following guidelines: (a) A transverse scan through the cranial sections of the liver to

detect subdiaphragmatic uid, (b) longitudinal scans of the liver through the left lobe and through the mid and lateral portions of the right lobe, (c) inspection of the abdominal cavity by three transverse scans across the abdomen (approximately 5 cm above, across, and approximately 5 cm below the diaphragm), and (d) scanning of the space of Douglas (standard technique or intravaginal ultrasound). The sonographic criteria used to determine ascites volume were as described by Goldberg [3]: (a) uid accumulation was termed as none if there were no signs of uid in the abdominal cavity; (b) uid accumulation was termed as minimal ascites (1400 ml) if there were occasional thin layers

TUMOR NECROSIS FACTOR-a IN OVARIAN CARCINOMA

83

of uid between single bowel loops or small uid collections in the space of Douglas; (c) uid accumulation was termed as minor ascites (4011000 ml) if there were thin layers of uid around the liver (below the diaphragm and the interior edge) or frequently occurring layers of uid between bowel loops; and (d) uid accumulation was termed as major (1000 ml) if there was massive uid collection around the liver or gross collections of abdominal uid with oating bowel loops. Ultrasound responses were dened as follows. (1) Complete responseUltrasound (CRu): No or minimal accumulation of ascites (400 ml) at the end of the follow-up period as determined by drainage whenever medically feasible, and residual volume estimated by ultrasound. When drainage was not medically feasible, total volume was determined by ultrasound alone. (2) Partial responseUltrasound (PRu): Minor uid recurrence (1000 ml) at the end of the followup period as determined by drainage whenever medically feasible, and residual volume estimated by ultrasound. When drainage was not medically feasible, the total volume was determined by ultrasound alone. The ultrasound evaluation was conducted by a designated ultrasonographer at each center. Whenever possible, this individual was unaware of the treatment received by the patient. Denition of Response A responder was dened as a patient who met one of the following criteria: (a) A complete response required both a clinical complete response as well as an ultrasound complete response. (b) A partial response required (i) a complete clinical response and a partial ultrasound response (CRc / PRu) or (ii) a partial clinical response and a complete ultrasound response (PRc / CRu) or (iii) a partial clinical response and a partial ultrasound response (PRc / PRu). Denition of Relapse Relapse was dened as the occurrence of symptomatic ascites requiring aspiration or uid volume greater than 1000 ml as determined by drainage and/or ultrasound, which occurred after the last scheduled treatment but prior to the end of the 4-week follow-up period. Denition of Treatment Failure Any of the following criteria constituted treatment failure: (a) Patients who relapsed prior to the end of the 4-week follow-up period, (b) patients who, at the end of the 4-week follow-up period, did not meet the denition of a responder, or (c) patients who, at anytime, did not respond to therapy and were removed from the study because of adverse reactions.

TABLE 1 Baseline Patient Characteristics

Factor Mean age { SD Mean No. of days diagnosis to randomiziation { SD Mean Karnofsky { SD Mean chemotherapy courses { SD Mean previous paracenteses { SD Stage of disease (%) II III IV Unknown TNF 59.9 { 9.1 817.3 { 1238 71.9 { 15.3 2.1 { 0.8 3.8 { 2.6 0 17 1 0 (0.0) (94.4) (0.0) (0.0) Paracentesis 61.5 { 9.1 798.9 { 582 75.7 { 9.8 2.2 { 0.7 4.0 { 2.8 2 (9.5) 15 (71.4) 3 (14.3) 1 (4.8) P value 0.59 0.60 0.36 0.73 0.85 0.20

Toxicity Evaluation The WHO criteria for toxicity were used for the grading of severity of symptoms and abnormal ndings which were as follows: Grade 1 mild toxicity, Grade 2 moderate toxicity, Grade 3 severe toxicity, and Grade 4 lifethreatening toxicity. Quality of Life Assessment To assess the effect of treatment on quality of life, the EORTC Core Quality of Life Questionnaire was administered at baseline before the rst treatment, immediately prior to each additional treatment, and 4 weeks after the last treatment or within 4 weeks of relapse. Statistical Analysis Students t test was used to determine the signicance of differences for all comparisons, except for stage of disease, where a x2 test was used. Trial Execution The trial commenced in June 1992. Although it was determined a priori that a sample size of 128 patients (64 patients per arm) was required to detect at least a 30% difference in response rates between treatment arms (this was believed to be a clinically relevant difference), in the fall of 1993, after 39 patients had been entered on the study, investigators became concerned by the absence of any responses in the 18 patients treated with rhTNF-a. This observation was veried by the trial coordination center, and at a subsequent meeting of investigators in November 1993, it was decided to close the trial to further accrual.
RESULTS

A total of 39 patients were randomized: 18 to rhTNF-a and 21 patients to paracentesis alone. There were no signicant differences in baseline characteristics of patients between treatment groups (Table 1). All patients completed at

84

HIRTE ET AL.

least one treatment (paracentesis { injection of rhTNF-a), 15 patients on the treatment arm and 19 patients on the control arm completed at least two treatments, and 14 patients on the treatment arm and 17 on the control arm completed the maximum of three treatments. Four patients failed to complete the study and were deemed inevaluable. All were on the control arm. One patient had two treatments and then expired with no ultrasound or drainage performed prior to this. Two patients developed bowel obstruction; one required a colostomy after three treatments but no mention was made of whether ascites was present at that time and a second patient developed bowel obstruction after one treatment but there was no indication of whether ascites was present because no ultrasound or drainage was performed. A fourth patient had three treatments but was too ill to be examined for entry into the open label protocol and no information about ascites was obtained. One rhTNF-a patient had a complete clinical response; however, no ultrasound was performed and thus the patient did not meet the criteria for either a complete or partial response. The patient expired 10 days later. There were no responses in the other 17 patients randomized to the rhTNFa arm. There were no responses with respect to ascites accumulation in 17 evaluable control patients nor were there any responses in the 16 patients entered on the open label study after developing recurrent ascites in the control arm. No reduction in measurable solid tumor masses were noted. Because there were no documented responses in patients treated with rhTNF-a in the randomized study or the open label study, it was decided to close the study to further accrual at this point. There was a signicant difference in adverse events between treatment arms with 16 of 18 (88.9%) of patients on the treatment arm developing at least one adverse event, but only 3 of 21 (14.3%) patients on the control arm (P 0.0001). There were a total of 62 adverse events on the treatment arm (1.3 adverse events per treatment) and 4 in the control arm (0.070 adverse events per treatment). The most common adverse events in the treatment arm were fever/fatigue/chills (0.45 events per treatment), nausea/vomiting (0.36 events per treatment), and pain/discomfort at the paracentesis/injection site (0.41 events per treatment). Adverse events in the control arm were limited to fever/fatigue/ chills (0.036 events per treatment) and pain/discomfort at the paracentesis site (0.036 events per treatment). Most adverse events (95%) were mild or moderate in degree of severity (Table 2). There were no signicant differences in survival. Patients on the treatment arm survived a mean of 132.4 { 80.3 days after randomization, while patients on the control arm survived 145.8 { 161.9 days (P 0.58). Because there were no responses in patients treated with rhTNF-a, the quality of life data collected on these patients were not further analyzed.

DISCUSSION

Phase I and II studies of intraperitoneal rhTNF-a suggested substantial efcacy in controlling malignant ascites due to recurrent ovarian carcinoma [1, 2]. The phase I trial indicated that 0.08 mg/m2 rhTNF-a was both safe and effective in the treatment of malignant ascites [1]. This dose was subsequently chosen for a phase II multicenter trial [2]. Twenty of 20 ovarian carcinoma patients treated had a response. The present study was performed to conrm this activity in a randomized setting in a similar population of ovarian carcinoma patients using the same dose of rhTNFa. However, in our study, intraperitoneal rhTNF-a showed no evidence of efcacy in the patients studied. In contrast, in the phase I and II studies using the same source of rhTNFa, responses were seen in 30 of 31 patients with ascites due to recurrent ovarian carcinoma [1, 2]. It is not clear why rhTNF-a showed no efcacy in our study. In the phase I and II studies, patient selection and response criteria were similar and the dose of rhTNF-a ranged from 20 to 350 mg/m2 [1, 2]. The dose of rhTNF-a used in our study (0.06 mg/m2) had equivalent activity to 0.08 mg/m2 used in the phase II study. Although the lot of rhTNF-a used in our study was different from that in the phase I and II studies, the rhTNF-a used in our study showed appropriate TNF-a content and biological activity as measured by the L-929 cytotoxicity assay (4). Thus the bioactivity of the rhTNF-a used in the study did not appear to have inuenced the result of the study. In a nude mouse model of intraperitoneally xenografted human ovarian carcinoma, rhTNF-a in doses of 1 to 5 1 104 units/mouse/day prevented the development of ascitic disease when intraperitoneal therapy started the day after tumor cell injection. If treatment was delayed for 7 days, only the highest doses of rhTNF-a showed any activity (5). There are no other animal or clinical data available indicating whether there is a doseresponse relationship for the administration of intraperitoneal TNF-a and ascites resolution. Because of the high degree of efcacy of the 0.08 mg/m2 dose in the phase II study, there was no provision to escalate the dose of rhTNF-a in the initial trial design, and given the complete lack of activity of the 0.06 mg/m2 dose in the rst 39 patients treated in our study (which had been effective in all 20 ovarian carcinoma patients treated in the phase II study), the dose of rhTNF-a was not escalated. In the fall of 1993, a decision was made to terminate patient accrual to this study based on an interim analysis of outcome. This analysis was not a planned interim analysis but was performed in response to concerns raised by study investigators. The decision to terminate accrual to any study is a difcult one. However, the study steering committee believed that because no efcacy had been demonstrated in the rst 18 patients randomized to the rhTNF-a arm of the study, it would be unethical to continue patient accrual.

TUMOR NECROSIS FACTOR-a IN OVARIAN CARCINOMA

85

TABLE 2 Severity of Adverse Events

TNF patients Adverse event treatment 0.234 0.192 0.021 0.277 0.085 0.000 0.064 0.298 0.043 0.000 0.021 0.000 0.000 0.043 0.000 0.000 0.000 0.000 0.021 0.021 0.000 1.319 Paracentesis patients Adverse event treatment 0.018 0.018 0.000 0.000 0.000 0.000 0.000 0.018 0.018 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.070

Adverse event Fever/fatigue/chills

Severity Mild Moderate Severe Mild Moderate Severe Mild Moderate Severe Mild Moderate Severe Mild Moderate Severe Mild Moderate Severe Mild Moderate Severe

n 11 9 1 13 4 0 3 14 2 0 1 0 0 2 0 0 0 0 1 1 0 62

n 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 4

P value

0.001

Nausea/vomiting

0.0001

Pain/discomfort

0.0001

Hypotension

0.45

Edema

0.20

Other GI

1.0

Other

0.20

Total: All events

The management of ascites due to recurrent ovarian carcinoma is problematic. Paracentesis alone is the technique most commonly used to palliate symptoms of ascites. Peritoneal shunts have been used to palliate refractory ascites in ovarian carcinoma patients with limited success and have been associated with high complication rates including shunt blockage and thrombotic and embolic complications related to the shunt [69]. A variety of other agents have been administered intraperitoneally, in an attempt to control malignant ascites. These include radioisotopes (as well as radiolabeled monoclonal antibody) [1012], chemotherapeutic agents (bleomycin) [1315], and biological response modiers (Cornebacterium parvum, interferon-b, interferon-a, and interleukin-2 with or without LAK cells) [1622]. Although reduction in ascites accumulation was achieved in 33 to 87% of patients, responses were often only of short duration. In addition, these studies were either phase I or II in design and the true test of any therapy is in a randomized setting. The development of ascites in patients with ovarian carcinoma contributes signicantly to the morbidity of this disease. Mean survival after the discovery of ascites is 3040 weeks. Although early work emphasized the importance of lymphatic obstruction by tumor cells (23), recently new insight has been shed on the pathophysiology of malignant ascites. Secreted products of tumor cells may act in a para-

crine fashion by eliciting uid leakage from the neighboring host vasculature [24, 25]. Factors which have been demonstrated to mediate such effects include vascular permeability factor (VPF) [26], epidermal growth factor (EGF), and transforming growth factor-a (TGF-a) [27]. In addition, the neovasculature of a growing tumor is thought to be intrinsically leaky, thereby providing an additional site for uid extravasation from the vascular space [28]. Signicantly elevated concentrations of VPF have been demonstrated in malignant ascites of patients with ovarian carcinoma [29] and the malignant tissue appears to be one of the sources of this activity [30]. In an animal model of ascites-producing tumors, VPF accumulated in the vessels of the tumor and the peritoneal surfaces [31, 32]. VPF may also be a critical factor in supporting the widespread intraperitoneal growth of ovarian carcinoma by allowing the induction of neovascularization and stroma formation of metastatic deposits [33]. Indeed, injection of anti-VPF antibodies in nude mice transplanted with human tumors led to tumor regression and decreased density of the tumor-supporting vasculature [34]. EGF and TGF-a have also been shown to induce the production of ascites when injected intraperitoneally in a murine model [27]. The mechanism by which EGF or TGFa stimulates ascites formation is not clear. Indomethacin, a blocker of prostaglandin (PG) synthesis, inhibited the effect

86

HIRTE ET AL. 6. Cheung, D. K., and Raaf, J. H. Selection of patients with malignant ascites for a peritoneovenous shunt, Cancer 50, 12041209 (1982). 7. Lacy, J. H., Wieman, T. J., and Shively, E. H. Management of malignant ascites, Surg. Gynecol. Obstet. 159, 397412 (1984). 8. Souter, R. G., Wells, C., Tarin, D., and Kettlewell, M. D. Surgical and pathological complications associated with peritoneovenous shunts in management of malignant ascites, Cancer 55, 19731975 (1985). 9. Edney, J. A., Hill, A., and Armstrong, D. Peritoneovenous shunts palliate malignant ascites, Am. J. Surg. 158, 598601 (1989). 10. Ariel, I. M., Oropeza, R., and Pack, G. T. Intracavitary administration of radioactive isotopes in the control of effusions due to cancer, Cancer 19, 10961102 (1966). 11. Jackson, G. L., and Blosser, N. M. Intracavitary chromic phosphate (32P) colloidal suspension therapy, Cancer 48, 25962598 (1981). 12. Buckman, R., De Angelis, C., Shaw, P., Covens, A., Osborne, R., Kerr, I., Reed, R., Michaels, H., Woo, M., Reilly, R., Law, J., Baumal, R., Groves, E., and Marks, A. Intraperitoneal therapy of malignant ascites associated with carcinoma of ovary and breast using radioiodinated monoclonal antibody 2G3, Gynecol. Oncol. 47, 102109 (1992). 13. Bitran, J. D. Intraperitoneal bleomycin: Pharmacokinetics and results of a phase II trial, Cancer 56, 24202423 (1985). 14. Ostrowski, M. J. An assessment of the long-term results of controlling the reaccumulation of malignant effusions using intracavitary bleomycin, Cancer 57, 721727 (1986). 15. Paladine, W., Cunningham, T. J., Sponzo, R., Donavan, M., Olson, K., and Horton, J. Intracavitary bleomycin in the management of malignant effusions, Cancer 38, 19031908 (1976). 16. Mantovani, A., Sessa, C., Peri, G., Allavena, P., Introna, M., Polentarutti, N., and Mangioni, C. Intraperitoneal administration of corynebacterium parvum in patients with ascitic ovarian tumors resistant to chemotherapy: Effects on cytotoxicity of tumor-associated macrophages and NK cells, Int. J. Cancer 27, 437446 (1981). 17. Rambaldi, A., Introna, M., Colotta, F., Landolfo, S., Colombo, N., Mangioni, C., and Mantovani, A. Intraperitoneal administration of interferon b in ovarian cancer patients, Cancer 56, 294301 (1985). 18. Gebbia, V., Russo, A., Gebbia, N., Valenza, R., Testa, A., Palmeri, S., and Rausa, L. Intracavitary beta-interferon for the management of pleural and/or abdominal effusions in patients with advanced cancer refractory to chemotherapy, In Vivo 5, 579582 (1991). 19. Bezwoda, W. R., Seymour, L., and Dansey, R. Intraperitoneal recombinant interferon-alpha 2b for recurrent malignant ascites due to ovarian cancer, Cancer 64, 10291033 (1989). 20. Stuart, G. C. E., Nation, J. C., Snider, D. D., and Thunberg, P. Intraperitoneal interferon in the management of malignant ascites, Cancer 71, 20272030 (1993). 21. Lissoni, P., Barni, S., Ardizzoia, A., Paolorossi, F., Tisi, E., Crispino, S., and Tancini, G. Intracavitary administration of interleukin-2 as palliative therapy for neoplastic effusions, Tumori 78, 118120 (1992). 22. Li, D-J., Wang, Y-R., Tan, X-Y., Wang, H-Z., Yao, X-D., and Ba, D-N. A new approach to the treatment of malignant effusion, Chin. Med. J. 103, 9981002 (1990). 23. Holm-Nielson, P. Pathogenesis of ascites in peritoneal carcinomatosis, Acta Pathol. Microbiol. Scand. 33, 1021 (1953). 24. Hirabayashi, K., and Graham, J. Genesis of ascites in ovarian cancer, Am. J. Obstet. Gynecol. 106, 492497 (1970). 25. Garrison, R. N., Heuser, L. S., and Galloway, R. H. Mechanisms of malignant ascites production, J. Surg. Res. 42, 126132 (1987). 26. Senger, D. R., Galli, S. J., Dvorak, A. M., Peruzzi, C. A., Harvey, V. S., and Dvorak, H. F. Tumor cells secrete a vascular permeability factor

of EGF, suggesting that PGs, especially PGE2 and PGI2 , which have been reported to increase vascular permeability, are involved in EGF-stimulated ascites formation [27]. However, the effect of indomethacin may also have been mediated by some compensatory or other unknown mechanism. Since both EGF and TGF-a are angiogenic mediators and induce discontinuity of the basement membrane, they may also induce the production of bloody ascites through the formation of fragile new blood vessels. The ascites from patients in the present study was not assessed for levels of EGF, TGF-a, or VPF. Although it is possible that EGF, TGF-a, or VPF may have inhibited the effects of the administered TNF-a, similar effects (i.e., lack of activity) should have been noted in the patients in the phase I and II studies. However, the rhTNF-a was highly effective in reducing ascites volumes in these earlier studies, suggesting that these other factors, which would have been expected to be present, were not able to abrogate the effects of rhTNF-a. Although soluble TNF receptors (s-TNF-R) have been shown to be present in malignant ascites [35], their role in affecting the response to rhTNF-a administered intraperitoneally is not clear. Levels of s-TNF-R in serum or ascites of patients in the present study were not determined. s-TNF-R would have been expected to be present in the ascites of patients treated in the phase I and II studies as well as in the ascites of patients in the present study. The fact that patients in the phase I and II studies had very high response rates, whereas there were no responses in the present study, suggested that s-TNF-R were not likely responsible for this variability in response to rhTNF-a. It would appear that ultimately approaches to reverse underlying pathophysiology of ascites generation, either by blocking production or interfering with the effects of factors such as VPF or by more effectively treating the underlying tumor, are most likely to be effective in controlling accumulation of malignant ascites.
REFERENCES
1. Raeth, U., Kaufmann, M., Schmid, H., Hofmann, J., Wiedenmann, B., Kist, A., Kempeni, J., Schlick, E., Bastert, G., Kommerell, B., and Maennel, D. Effect of intraperitoneal recombinant human tumor necrosis factor alpha on malignant ascites, Eur. J. Cancer 27, 121125 (1991). 2. Raeth, U., Schmid, H., Karck, U., Kempeni, J., Schlick, E., and Kaufmann, M. Phase II trial of recombinant human tumor necrosis factora (rHuTNF) in patients with malignant ascites from ovarian carcinomas and non-ovarian tumors with intraperitoneal spread, Proc. Am. Soc. Clin. Oncol. 10, 187 (1991). 3. Goldberg, B. Evaluation of ascites by ultrasound, Radiology 96, 15 22 (1970). 4. Knoll Technical Report 000981-02-D, on le, Knoll AG, Ludwigshafen, Germany. 5. Balkwill, F. R., Ward, B. G., Moodie, E., and Fiers, W. Therapeutic potential of tumor necrosis factor-a and g-interferon in experimental human ovarian cancer, Cancer Res. 47, 47554758 (1987).

TUMOR NECROSIS FACTOR-a IN OVARIAN CARCINOMA that promotes accumulation of ascites uid, Science 219, 983985 (1982). 27. Ohmura, E., Tsushima, T., Kamiya, Y., Okada, M., Onoda, N., Shizuma, K., and Demura, H. Epidermal growth factor and transforming growth factor a induce ascitic uid in mice, Cancer Res. 50, 4915 4917 (1990). 28. Heuser, L. S., Taylor, S. H., and Folkman, J. Prevention of carcinomatosis and bloody malignant ascites in the rat by an inhibitor of angiogenesis, J. Surg. Res. 36, 244250 (1984). 29. Yeo, K-T., Wang, H. H., Nagy, J. A., Sioussat, T. M., Ledbetter, S. R., Hoogewert, A. J., Zhou, Y., Masse, E. M., Senger, D. R., and Dvorak, H. F. Vascular permeability factor (vascular endothelial growth factor) in guinea pig and human tumor and inammatory effusions, Cancer Res. 53, 29122918 (1993). 30. Olson, T. A., Mohanraj, D., Carson, L. F., and Ramakrishnan, S. Vascular permeability factor gene expression in normal and neoplastic human ovaries, Cancer Res. 54, 276280 (1994). 31. Dvorak, H. F., Siousset, T. M., Brown, L. F., Berse, B., Nagy, J. A.,

87

32.

33.

34.

35.

Sotrel, A., Manseau, E. J., Van De Water, L., and Senger, D. R. Distribution of vascular permeability factor (vascular endothelial growth factor) in tumors: Concentration in tumor blood vessels, J. Exp. Med. 174, 12751278 (1991). Nagy, J. A. Immunochemical localization of vascular permeability factor (vascular endothelial growth factor) in ascites tumors: distribution in peritoneal wall microvasculature, J. Cell Biol. 115, 264a (1991). Nagy, J. A., Brown, L. F., Senger, D. R., Lanir, N., Van De Water, L., and Dvorak, H. F. Pathogenesis of tumor stroma generation: A critical role for leaky blood vessels and brin deposition, Biochim. Biophys. Acta 948, 305326 (1988). Kim, K. J., Li, B., Winer, J., Armanini, M., Gillett, N., Phillips, H. S., and Ferrara, N. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumor growth in vivo, Nature 362, 841844 (1993). Andus, T., Gross, V., Holstege, A., Ott, M., Weber, M., David, M., Gallati, H., Gerok, W., and Schoelmerich, J. High concentrations of soluble tumor necrosis factor receptors in ascites, Hepatology 16, 749 755 (1992).