Purpose: The aims of this study were to determine the dose and schedule of i.p. cisplatin with i.p. gemcitabine in patients with persistent disease at second-look assessment, the toxicity of this regimen, and the time to treatment failure and overall survival.
Experimental Design: We performed a Phase I/II evaluation of i.p. cisplatin at 75 mg/m2 on day 1 with planned gemcitabine at 500, 750, 1000, or 1250 mg/m2 i.p. on days 1, 8, and 15 on a 28-day schedule for four courses. Eligible patients completed surgical cytoreduction followed by adjuvant platinum-based chemotherapy. They had second-look assessment showing microscopic or macroscopic (≤1 cm) disease, followed by i.p. port placement.
Results: The Phase I dose-limiting toxicity was grade 3 thrombocytopenia at day 15 on dose level 1 (n = 5). The protocol was amended, and the Phase II portion accrued to 30 patients, who were given i.p. cisplatin (75 mg/m2) on day 1 and gemcitabine at 500 mg/m2 on days 1 and 8 on a 21-day schedule for four courses. Nine patients were removed from the study: one each for hypersensitivity, cellulitis, and i.p. port malfunction; two for progression of disease; and four for renal toxicity. Other toxicities included grade 3 nausea (7%) and transient grade 3 neuropathy (3%). Grade 1 or 2 neuropathy was frequently seen (80%). Five patients (17%) returned to the operating room at a median of 6 months (range, 1–20 months) after i.p. therapy for evaluation of abdominal pain; two patients had recurrence, and all had areas of fibrous tissue with encasement of the bowel. In two patients, the fibrous tissue was causing partial bowel obstruction. No other patients had symptoms prompting surgical exploration. Pharmacokinetic (PK) studies showed a median area under the curve (AUC) i.p. of 3041 h·μm (range, 676-5702 h·μm) and AUC in plasma of 4.0 h·μm (range, 0.92–8.2 h·μm) reached between 120 and 240 min; the pharmacological advantage was 759-fold (range, 217-1415-fold) for i.p. versus plasma drug levels. The mean residence time of gemcitabine with i.p. administration was 4.7 h. The median time to progression of the intent to treat population was 15.93 months (95% confidence interval, 9.13–25.9 months), with a median overall survival of 43.5 months [95% confidence interval, (34.66-∞)]. No statistical differences were seen with respect to overall survival if patients were grouped in terms of optimal debulking or not (median not reached versus 34.8 months, respectively; P = 0.16) or whether visible disease was present or not at the start of i.p. therapy (34.8 versus 47.7 months; P = 0.47). With regard to time to treatment failure, a statistical difference favored patients with optimal versus nonoptimal primary debulking (25.2 versus 10.2 months, respectively; P = 0.03).
Conclusions: The median time to treatment failure and overall survival of 15.9 months and 43.5 months, respectively, are consistent with our historical data in patients receiving i.p. platinum-based regimens for consolidation. The fibrotic changes seen in explored patients suggest local toxicity of this combination. The absolute benefit of i.p. consolidation requires randomized trials to assess efficacy.
The majority of patients with ovarian cancer are in clinical remission after primary surgical debulking and adjuvant platinum- and taxane-based chemotherapy (1) . Despite the initial response, 70–90% of patients relapse at a median of 18–24 months, indicating that most patients have persistent microscopic or macroscopic small-volume disease (2) . Attempts at improving outcome after primary therapy have included additional systemic chemotherapy with the same or different agents (3 , 4) or with regional therapy via the i.p. approach. A variety of novel biological and immune-directed therapies are also now being studied in this patient group.
The propensity of relapsed ovarian cancer to remain confined to the peritoneal cavity makes it suitable for the investigation of i.p. consolidation therapy. The strategy of using i.p. delivery to improve the pharmacokinetic advantage by increasing concentration and exposure time of drug to tumor has been studied extensively. The ability to bypass poorly developed vasculature supporting small-volume disease makes this particularly attractive to evaluate in the consolidation setting (5 , 6) . The only prospective randomized study evaluating i.p. therapy used it as a part of the primary treatment regimen, and the results are not directly applicable to the consolidation setting. Nonrandomized Phase II data have suggested a benefit of i.p. therapy in patients with a negative second look compared with historical controls (7) . Other published data for i.p. consolidation therapy consist of a series of Phase II trials in which patients with small-volume (≤1 cm) residual disease treated with i.p. platinum combinations generally showed surgically defined complete response rates of 25–35%. These Phase II data have suggested a clinically meaningful impact on survival compared with historical controls in those patients achieving a complete response after i.p. treatment, with a median survival in small-volume responders of 40 months versus 10 months for nonresponders (P = 0.009; Ref. 8 ). A large retrospective review from our institution recently described the relapse characteristics of a heterogeneous group of patients treated with a variety of platinum-containing i.p. regimens largely used in the consolidation setting that can serve as a benchmark for patients in this study (9) .
The synergy (10, 11, 12) and clinical activity of systemic cisplatin and gemcitabine in patients with ovarian cancer are well described both in the chemotherapy-naive and relapse populations (13 , 14) . Several characteristics of gemcitabine have suggested that it is a reasonable candidate for exploration in the i.p. setting. These characteristics include (a) the structural similarity to 1-β-d-arabinofuranosylcytosine, which has been successfully administered i.p.; (b) a dose-limiting toxicity that has largely been related to plasma levels; (c) i.p. administration in animal models without toxicity; and (d) data suggesting that prolonged exposure achieved with i.p. administration would maintain gemcitabine levels above the threshold for nucleotide incorporation into cells, which has been correlated with cytotoxic activity (15 , 16) .
In this Phase I/II study, we sought to define the maximum tolerated dose of i.p. gemcitabine with i.p. cisplatin given in consolidation in patients with microscopic or macroscopic (≤1 cm) disease after primary treatment of ovarian cancer. The purposes of the Phase II portion were to determine the time to treatment failure and overall survival of the treated population and, by comparison with historical data, to determine whether additional prospective studies of this combined i.p. approach were warranted.
PATIENTS AND METHODS
Eligible patients had histologically documented epithelial ovarian, fallopian tube, or peritoneal carcinoma after initial cytoreductive surgery and chemotherapy with a platinum-containing regimen followed by normalization of CA-125. All patients underwent second-look surgical reassessment (laparotomy or laparoscopy) and had confirmation of residual disease (microscopic or gross ≤1.0 cm) at the end of the procedure. Patients had adequate hematological (WBC count ≥3,000 cells/mm3; platelets ≥100,000 cells/mm3), renal (serum creatinine <1.6 mg/dl), and hepatic (bilirubin, aspartate aminotransferase, and alkaline phosphatase less than three times the upper limit of normal) function. Karnofsky performance status was >60%. Eligible patients had a functioning s.c.-implanted i.p. catheter. Patients were excluded if residual disease was >1 cm, neuropathy was grade 2 or higher, or if they had any contraindication to i.p. therapy, such as intra-abdominal infection or widespread adhesions. The Memorial Sloan-Kettering Cancer Center Institutional Review Board approved the protocol, and informed consent was obtained.
Treatment Plan and Pharmacokinetic Sampling.
Cisplatin (i.p.) was given at a fixed dose of 75 mg/m2 via the i.p. catheter on day 1. Gemcitabine (i.p.) was initiated in dose level 1 at 500 mg/m2 on days 1 and 8 and was originally planned to be administered on day 15 of a 28-day cycle. The i.p. chemotherapy was administered in 1 liter of normal saline. An additional 1 liter of normal saline was given as tolerated by gravity feed. The i.v. hydration for cisplatin followed standard institutional protocols. Three patients were to be enrolled at each dose level in the Phase I portion, with planned escalation of gemcitabine from 500 mg/m2 to 750, 1000, and 1250 mg/m2, respectively, with subsequent cohorts if dose-limiting toxicity was not present. Each patient was to receive four cycles of therapy (one cycle = 28 days). For retreatment, on day 1 of each cycle patients were required to have an absolute neutrophil count ≥1,000 cells/mm3 and platelets ≥100,000 cells/mm3 for cisplatin and gemcitabine administration. On days 8 and 15 of each cycle, patients were required to have an absolute neutrophil count ≥500 cells/mm3 and platelets ≥50,000 cells/mm3 for gemcitabine administration. Patients with progression to grade 3 neuropathy or grade 2 abdominal pain were removed from the study. If one patient had toxicity higher than grade 2, an additional three patients were accrued to the respective dose level. If two or more patients were seen with toxicity higher than grade 2 for a given dose level, this was considered a dose-limiting toxicity and further accrual to that cohort did not proceed. The protocol would accrue a total of 30 patients at the determined Phase II dose level. History and physical examination, complete blood cell counts, and a comprehensive panel, including creatinine and magnesium, were performed at each visit before treatment. CA-125 levels were measured with each course.
Blood samples for pharmacokinetic analysis were collected at 0.5, 1, 2, 4, 6, and 24 h after completion of drug instillation. Peritoneal fluid for pharmacokinetic analysis was obtained by discarding the initial 10 ml of fluid withdrawn from the i.p. catheter. An additional 7 ml were obtained at 0.5, 1, 2, 4, 6, and 24 h after completion of drug instillation with the addition of additives according to the method of Freeman et al. (17) . If the i.p. catheter was not functioning for fluid withdrawal, no samples were obtained. Patients were evaluated by physical examination, CA-125 measurements, and a computed tomography scan every 3 months after completion of treatment.
Standard dose escalation criteria were used. If at any dose level two or more patients developed toxicity higher than grade 2, the dose was not escalated. The maximum tolerated dose is the dose level at which at most one of three or two of six toxicities are observed. A minimum of 3 patients and maximum of 24 patients could be enrolled in the Phase I portion. In the Phase II portion, a total of 30 patients were to be studied at the determined Phase II dose. The endpoints of the study were to assess the tolerated dose and safety of i.p. gemcitabine when given with i.p. cisplatin (Phase I) and to determine time to radiographic progression and overall survival (Phase II). Overall survival and time to progression distributions were estimated using Kaplan–Meier curves, and the prognostic effect of categorical covariates was assessed using the log-rank test.
Between June 1998 and July 2000, 30 patients were entered in the study. The patient characteristics are outlined in Table 1⇓ . The median age was 55 years (range, 22–76 years), with a median Karnofsky performance status of 90%. Patients predominantly had ovarian cancer (83%), with 10% and 7% having fallopian tube and peritoneal cancer, respectively. The majority of patients were stage III, with papillary serous histology. Patients were equally split between those with optimal (50%) and suboptimal (50%) debulking. Most patients had macroscopic disease ≤1 cm (73%) at the start of i.p. therapy, with 7% having only microscopic disease. All patients received platinum- and taxane-based therapy as primary treatment with responding disease.
Phase I Results.
The first three patients experienced grade 3 asymptomatic thrombocytopenia at day 21, which required treatment delay. The cohort was expanded according to the protocol with similar findings in the fourth and fifth patients. Day 21 thrombocytopenia was considered to represent dose-limiting toxicity. The protocol was amended, and the schedule was changed to 75 mg/m2 i.p. cisplatin on day 1, with i.p. gemcitabine at 500 mg/m2 on days 1 and 8 of a 21-day schedule. Hence, additional dose escalation was not performed. The modified schedule allowed for the accrual of 30 patients without difficulty.
The median number of cycles received was four, as planned. All 30 patients (each treated with the 21-day schedule and initial dose cohort) received at least one treatment. Nine patients were removed from the study: one each for platinum hypersensitivity, gemcitabine-related cellulitis, and i.p. port malfunction; two for disease progression; and four for renal toxicity (Table 2)⇓ . Abdominal discomfort was common but not dose limiting (grade 1, eight patients; grade 3, one patient). The remaining spectrum of toxicities is outlined in Table 3⇓ and includes the toxicities typically reported with cisplatin. Grade 3 nausea occurred in 7% of patients, with most patients reasonably controlled with standard antiemetics. Grade 1 or 2 neuropathy was reported in 80% of the patients, with most patients having residual disease from primary therapy. Only one patient reported transient grade 3 neuropathy. Four patients had grade 2 or 3 renal toxicity, prompting their removal from the study by the investigator. No other unexpected toxicities occurred.
Five patients (17%) had returned to the operating room at the time of this report (median follow-up, 24 months) after completion of i.p. therapy for the evaluation of abdominal pain, as outlined in Table 4⇓ . The median number of months after i.p. therapy for surgery was 6 (range, 1–20 months). Of the five patients, two had radiographic evidence of partial obstruction, and one had free air after a routine colonoscopy. Two patients were explored for abdominal pain with tenderness but had no radiographic abnormalities. Two patients had pathological evidence of recurrence, whereas the remaining three had no recurrent cancer. Each of the five patients was noted as having multiple adhesions, with areas of fibrous tissue forming a capsular encasement. Adhesiolysis was performed when appropriate.
Progression-Free Interval and Overall Survival.
The median time to progression of the intent-to-treat population was 15.93 months (95% confidence interval, 9.13–25.9 months), with a median overall survival of 43.5 months [95% confidence interval, ≥34.66 months (some patients were still alive at the time of publication; therefore, the interval is open-ended at present); see Figs. 1⇓ and 2⇓] . No statistical differences were seen with respect to overall survival if patients were grouped in terms of optimal debulking or not (median not reached versus 34.8 months; P = 0.16) or whether visible disease was present or not at the start of i.p. therapy (34.8 versus 47.7 months; P = 0.47). With regard to time to progression, a statistical difference favored patients with optimal versus nonoptimal primary debulking (25.2 versus 10.2 months; P = 0.03). No statistical difference was seen with regard to time to progression based on whether visible disease was present or not at the start of i.p. therapy (11.1 versus 22.0 months; P = 0.21), as seen in Table 5⇓ .
Previous studies have summarized the performance of i.p. catheters with regard to pharmacokinetic sampling and have reported that pericatheter sheaths can develop with a one-way valve effect, preventing aspiration of fluid in ∼50% of patients (18) . We were able to aspirate peritoneal fluid, yielding pharmacokinetic data, from 13 patients (43%), a percentage that is consistent with published reports. In these 13 patients, the median area under the curve for i.p. administration of gemcitabine was 3041 h·μm (range, 676-5702 h·μm), and the median plasma area under the curve was 4 h·μm (range, 0.95–7.5 h·μm), reached between 120 and 240 min (Table 6)⇓ . A mean pharmacological advantage of 791-fold (range, 217-1415-fold) was seen when we compared i.p. to plasma drug areas under the curves. The mean residence time for i.p. gemcitabine was 4.7 h. All patients (n = 13) had i.p. gemcitabine levels >20 μm at 2 h after instillation (above threshold for maximum triphosphate accumulation), and 10 patients (77%) still had levels >20 μm at 6 h.
The need to identify effective consolidation regimens for patients with ovarian cancer is illustrated by the consistent relapse rates of 70–90%, occurring at a median of 18–24 months (1 , 19) . The findings of the Gynecologic Oncology Group protocol 132 reported by Muggia et al. (2) , which evaluated paclitaxel or cisplatin or paclitaxel with cisplatin as primary treatment, was particularly informative in demonstrating persistent disease at second-look assessments in 69, 88, and 63% of patients, respectively. This would imply that to improve the outcome of patients with ovarian cancer, either primary therapy must be made more effective or treatment needs to be directed at this small-volume disease persisting after primary treatment, which has been termed “consolidation.”
The randomized data for the i.p. delivery of cytotoxic agents address its use as part of the primary treatment regimen. Recognizing the limitations and controversies surrounding three primary treatment studies, each has suggested an advantage for i.p. administration (20, 21, 22) . Nonrandomized Phase II data have suggested a benefit of i.p. therapy used as consolidation in patients with a negative second look compared with historical controls (7) . In the study by Barakat et al. (7) , patients with negative second looks received i.p. cisplatin with i.p. etoposide for a 12-week total course. Historically, patients with negative second looks have a 50% failure rate at a median of 24 months. In contrast, the median time to failure was not reached at 36 months of follow-up in the i.p.-treated patients. The four cycles of i.p. consolidation in this study were empirically based on providing a 12-week course of consolidation while minimizing cumulative toxicity. Furthermore, from a systemic standpoint, which is not directly comparable, older studies have shown no benefit for prolonged platinum administration when comparing 5 versus 10 cycles of administration (23) . Recently, a Southwest Oncology Group and Gynecologic Oncology Group study evaluating 3 versus 12 additional cycles of paclitaxel was stopped early and has been reported in abstract form as demonstrating a progression-free survival advantage of 28 versus 21 months (P = 0.0035) in favor of prolonged paclitaxel administration. These patients did not have second-look assessment, but one can estimate from the original demographics (stratified for optimal stage III, suboptimal stage III, and stage IV) that at least 30% of these patients (and likely more) would have a negative second look and, thus, would represent a better prognostic group than our patients, who all have disease and 73% of whom have macroscopic disease, preventing any direct comparison of time to treatment failure between the two groups (3) .
As an overall benchmark, our recently summarized experience of the long-term outcomes in a large series of patients with a variety of second-look findings treated with i.p. platinum-containing regimens can be used (9) . Patients with microscopically positive second looks had an overall survival of 57.6 months, and those with macroscopic disease but with disease of <1 cm (as in 73% of patients in our study) had an overall survival of 39.6 months. This range is consistent with the results in our study, with an overall survival of 43.6 months.
The efficacy of gemcitabine has been correlated with concentrations of gemcitabine triphosphate accumulation, which is related to plasma concentration (24) . No further increments in triphosphate accumulation have been seen with doses resulting in plasma concentrations >20 μm, suggesting enzymatic saturation at higher doses, and cytotoxicity has been seen in cell lines with concentrations as low as 1 μm. Plasma concentrations in our study were considerably below the threshold for maximum incorporation, ranging from 0.92 to 8.2 μm. Despite the low plasma concentrations, these levels were high enough to provide hematological toxicity, preventing dose escalation. Intraperitoneal concentrations remained above the threshold for maximum triphosphate accumulation a minimum of 2 h and a maximum of 6 h, which suggests that any benefit of i.p. administration might be duplicated with a 2-h fixed dose rate infusion (10 mg/m2/min) as suggested in multiple ongoing studies (25) . The benefit of fixed dose rate infusion, however, depends on an adequate blood supply that brings the agent to tumors, and we cannot exclude a therapeutic benefit by direct delivery of a drug to poorly vascularized tumor in certain settings. Patel et al. (26) allowed patients to serve as their own controls, comparing gemcitabine at 1000 mg/m2 given over the traditional 30 min versus a pharmacologically based fixed dose rate infusion. Patients with the prolonged infusion had a 1.4-fold increase in maximum intracellular 2′,2′-difluorodeoxycytidinetriphosphate concentration compared with those with short infusions (1.0–2.6-fold; P = 0.016). The potential clinical benefits of prolonged fixed dose rate infusions are under investigation. Moreover, the high levels of gemcitabine (>20 μm) represent no pharmacological advantage for i.p. administration.
The reported toxicities in our study are generally consistent with toxicities associated with cisplatin- and gemcitabine-based therapies. The thrombocytopenia seen on day 15 has prompted adoption of a day 1 and 8 administration during every 21-day schedule in other i.v. administration trials with cisplatin and gemcitabine (27) . Fixed dose rate i.v. schedules have been associated with more hematological toxicity, and although our systemic drug levels are low, they are clearly sufficient to cause dose-limiting myelosuppression in combination with cisplatin. We noted that the toxicity associated with cisplatin at 75 mg/m2 i.p. appears less than reported at 100 mg/m2 in the abstract of Armstrong et al. (22) , suggesting that if the i.p. approach is to be carried forward in any setting, the lower dose may be better tolerated. The five patients (17%) that came to exploratory surgery for evaluation of abdominal pain are noteworthy in that each had areas of fibrous encasement of the bowel requiring adhesiolysis. Two patients had recurrent disease, but three had adhesion formation and intermittent partial bowel obstruction, the likely source of their discomfort. No other patients have required intervention. No peritoneal inflammation was seen in the animal studies evaluating the preclinical pharmacokinetics of gemcitabine by i.p. administration (16 , 28) . Other agents evaluated for i.p. delivery, such as doxorubicin and mitoxantrone, have also caused this phenomenon (29 , 30) . Because no apparent advantage was seen with i.p. delivery of gemcitabine and because of the noted peritoneal fibrosis in some patients, we would recommend that any additional consolidation studies proceed with i.v. administration.
In summary, the median time to treatment failure and overall survival of 15.9 and 43.5 months, respectively, are consistent with historical data in second-look-positive patients receiving a variety of i.p. platinum-based regimens for consolidation. In addition, the fibrotic changes seen in explored patients and the pharmacokinetic profile, which may be duplicated with prolonged fixed dose rate infusion, suggest that future studies of gemcitabine should be performed with i.v. administration. Finally, the absolute benefit of i.p. consolidation requires randomized trials to assess efficacy. The prolonged survival in this patient population, with largely macroscopically positive second-look surgical assessments, is noteworthy.
Grant support: Grants CA-52477-10 Ov PPG and CA-89333-01A2 K23.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Paul Sabbatini, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021. Phone: (212) 639-6423; Fax: (212) 639-8865; E-mail:
- Received October 28, 2003.
- Revision received January 9, 2004.
- Accepted January 15, 2004.