A Phase I trial of intravenous melphalan, paclitaxel, and cisplatin plus granulocyte-colony stimulating factor in patients with suboptimal advanced epithelial ovarian carcinoma or peritoneal carcinoma

Abstract

BACKGROUND

The primary purpose of this study was to establish the maximum tolerated dose (MTD) of intravenous melphalan in combination with paclitaxel and cisplatin plus granulocyte-colony stimulating factor (G-CSF) in patients with suboptimal advanced epithelial ovarian carcinoma or primary peritoneal carcinoma.

METHODS

Patients with suboptimal (>2 cm residual tumor) Stage III or Stage IV epithelial ovarian carcinoma or peritoneal carcinoma were eligible for this Phase I study. In the first stage of the study, the doses of paclitaxel and cisplatin were fixed at 135 mg/m² and 75 mg/ m², respectively, and the dose of intravenous melphalan was escalated in consecutive cohorts of 3–6 patients depending on toxicity. The planned dose escalation levels of melphalan were 6 mg/m², 10 mg/m², and 14 mg/m². In the second stage of the study, the doses of cisplatin and melphalan were fixed at 75 mg/m² and the MTD level, respectively, and the dose of paclitaxel was escalated. The planned dose escalation levels of paclitaxel were 150 mg/m², 175 mg/m², 200 mg/m², 225 mg/m², and 250 mg/m². G-CSF was administered for 12–19 days with each cycle, and cycles were repeated every 4 weeks for a total of 6 cycles. Other end points included clinical or surgical response, progression free survival, and survival.

RESULTS

Between January 1993 and May 1996, 34 women with untreated advanced stage epithelial ovarian carcinoma or primary peritoneal carcinoma were treated with 192 cycles of therapy. The MTD of melphalan was 10 mg/m², with the dose-limiting toxicity being thrombocytopenia. Paclitaxel was escalated to a dose level of 200 mg/m² with a toxicity rate of < 33%. The clinical response rate was 80% in 29 patients with measurable disease. Of 11 patients who underwent second-look surgery, 5 (45%) had a surgical pathologic complete response. The median progression free survival was 16.8 months and the median survival was 32.8 months.

CONCLUSIONS

The combination of intravenous melphalan, paclitaxel, and cisplatin was found to have acceptable toxicity and good activity. A Phase II study of this combination appears to be warranted. Cancer 1999;86:2291–300. © 1999 American Cancer Society.

Effective treatment for patients with advanced epithelial ovarian carcinoma remains a challenge for the oncologist. It has been approximately 20 years since the introduction of cisplatin into first-line combination chemotherapy regimens for advanced stage patients. During that interval, the 5-year survival rate has not risen above 15–20%. Despite objective response rates as high as 90%, most patients develop progressive tumor and die of their disease; the median survival rate ranges from 18 months to 36 months.1–8

On the basis of the results of a prospective randomized study of the Gynecologic Oncology Group (GOG) in which the new combination of paclitaxel and cisplatin was compared with the standard combination of cyclophosphamide and cisplatin in patients with suboptimal advanced epithelial ovarian carcinoma, most experts currently believe that a chemotherapy regimen including paclitaxel and a platinum drug represents optimal treatment for such women.9 In the GOG trial, the paclitaxel combination improved the response rate (73% vs. 60%), the median progression free survival (18 months vs. 14 months), and the median survival (38 months vs. 24 months). However, mature survival data from which to draw any conclusions regarding long term survival are not yet available from that study.

Nevertheless, the findings of the GOG trial are encouraging. While we await long term follow-up of the patients in this study, an attractive strategy is to build on the foundation of the combination of paclitaxel and a platinum drug with the hope of achieving even better survival. Among the many agents that might be incorporated into a paclitaxel-cisplatin combination is melphalan. Oral melphalan has been used extensively as a single agent in the treatment of patients with advanced epithelial ovarian carcinoma for over 3 decades.10–12 Several investigators also have reported the use of intravenous melphalan, either as a single agent or in combination with other drugs, for women with ovarian carcinoma.13–16 In the past several years, however, interest in melphalan has waned because of reports of its leukemogenicity.17–19 These studies indicate that the leukemogenic potential of melphalan is related directly to duration of use and cumulative dose.

In a retrospective study of 116 patients with optimal advanced stage epithelial ovarian carcinoma who underwent treatment at our institution between 1978 and 1988 on three consecutive prospective trials, patients received either the combination of oral melphalan and cisplatin (in the first trial) or the combination of cyclophosphamide and cisplatin (in the second and third trials).20 In a multivariate analysis, the use of melphalan versus cyclophosphamide was a significant predictor of progression free survival (P = 0.0002) and survival (P = 0.04). These findings suggested to us that melphalan may be a more active alkylating agent than cyclophosphamide against epithelial ovarian carcinoma when combined with cisplatin. Therefore, a Phase I trial combining melphalan with paclitaxel and cisplatin seemed to be worthwhile. The primary purpose of this study was to establish the maximum tolerated dose (MTD) of intravenous melphalan in combination with paclitaxel and cisplatin plus granulocyte-colony stimulating factor (G-CSF) in patients with suboptimal advanced epithelial ovarian carcinoma or primary peritoneal carcinoma.

MATERIALS AND METHODS

Patient Eligibility

Patients entered onto this study were required to have one of the following: 1) histologically confirmed primary epithelial ovarian carcinoma, Grade 2 or 3 and Stage III or IV; or 2) histologically confirmed primary peritoneal carcinoma. All patients must have had appropriate initial surgery, suboptimal residual tumor (>2 cm) or disease beyond the peritoneal cavity, and no prior cytotoxic chemotherapy. Patients with either measurable disease (by physical examination or by imaging techniques) or nonmeasurable disease were evaluated. Every effort was made to document measurable disease.

Patients underwent initial surgery at other hospitals or at the M. D. Anderson Cancer Center. For those patients who underwent surgery at M. D. Anderson, the postsurgical residual tumor was documented carefully in terms of the size of the masses (measurement in centimeters of the greatest dimension) and the extent of the disease (involvement of other organs or peritoneal surfaces). The greatest dimension of residual tumor and the extent of disease for those patients who underwent surgery on at other hospitals were determined by discussing the findings with the operating surgeon, reviewing the dictated operative summary, and performing careful physical examination and imaging studies prior to the initiation of chemotherapy.

Pathologic confirmation of each patient's tumor was required before beginning chemotherapy. Histologic typing and grading were performed by the staff of the Department of Pathology. The World Health Organization classification system of histologic typing21 and the pattern-grading system of histologic gradingg22 were used.

Other eligibility criteria were as follows: adequate bone marrow function (white blood count ≥ 3000/μL, granulocyte count ≥ 1500/μL, and platelet count ≥ 100,000/μL); adequate renal function (serum creatinine ≤ 1.4 mg%); adequate hepatic function (serum bilirubin and serum glutamic oxaloacetic transaminase [SGOT] levels = twice normal); a life expectancy of at least 12 weeks; a Zubrod performance status of 0–2;23 and age >18 years and <65 years. Exclusion criteria were as follows: a diagnosis of borderline (low malignant potential) or Grade 1 ovarian tumor; optimal (≤2 cm) residual disease after surgery (unless disease extended beyond the peritoneal cavity); prior chemotherapy or radiotherapy; evidence of severe infection or gastrointestinal bleeding; history of cardiac arrhythmia or current use of antiarrhythmic medication; history of congestive heart failure, unstable angina, or myocardial infarction within the past 6 months; unclassified ovarian carcinoma or carcinoma of questionable ovarian or peritoneal origin; Zubrod performance score ≥ 3; and another malignancy (prior or concomitant, with the exception of nonmelanoma skin carcinoma). All patients who entered the trial gave their written, informed consent. The protocol was approved by the institutional Clinical Research Committee and the Institutional Review Board.

Pretreatment Evaluation

Baseline examinations included a complete history and physical examination with documentation of all measurable disease (by physical examination or imaging studies). Signs and symptoms and performance status also were documented. Laboratory studies included a complete blood count, differential count, platelet count, chemical survey, SGOT, CA 125, serum magnesium, urinalysis, chest radiograph, electrocardiogram, and computed tomography (CT) scan of the abdomen and pelvis.

Treatment Plan

After giving informed consent and undergoing pretreatment evaluation, eligible patients were admitted to the hospital for the first of six planned cycles of combination chemotherapy. Paclitaxel was administered on Day 1 of each cycle by infusion over 24 hours in 1 L 5% dextrose followed by melphalan by infusion over 1 hour in 100 mL normal saline. Immediately after completion of the melphalan infusion, intravenous (IV) hydration, consisting of 1 L normal saline plus 20 mEq KCl and 2 gm MgSO₄, was administered over 4 hours. Immediately after completion of the IV hydration, cisplatin mixed in 1 L half-normal saline plus 50 gm mannitol was infused over 4 hours. The starting doses were paclitaxel 135 mg/m², melphalan 6 mg/m², and cisplatin 75 mg/m². Cycles were repeated every 28 days.

Patients were premedicated with dexamethasone 20 mg orally 12 hours and 6 hours prior to the anticipated initiation of the paclitaxel infusion. Thirty minutes before the paclitaxel infusion, patients were premedicated further with diphenhydramine 50 mg IV and cimetidine 300 mg IV. Approximately 15 minutes prior to the cisplatin infusion, patients were premedicated with an antiemetic regimen consisting of ondansetron 0.15 mg/kg, which was repeated every 3–4 hours for an additional 2 doses. Other antiemetics (e.g., dexamethasone, diphenhydramine) were used during paclitaxel or melphalan administration, as necessary.

G-CSF was administered after every cycle of chemotherapy at a dose of 5 μg/kg per day subcutaneously, starting approximately 24 hours after completion of the cisplatin infusion and continuing through Day 22 provided that the postnadir neutrophil count was ≥10,000/μL. Patients who failed to achieve an absolute granulocyte count of ≥10,000/μL by Day 22 continued on G-CSF until this level was reached or exceeded. Patients who had a white blood count ≥15,000/μL after the nadir discontinued G-CSF during that treatment cycle; G-CSF was resumed in that cycle only if the white blood count fell to <4000/μL. After the initial dose level was completed, the protocol was revised to include G-CSF at the same dose, starting approximately 24 hours after completion of the cisplatin infusion and continuing for a total of 12 days provided that the postnadir neutrophil count was ≥10,000/μL.

After the completion of 6 cycles of chemotherapy, patients who were clinically free of disease, as determined by a normal serum CA 125 level and a normal CT scan of the abdomen and pelvis, were candidates for second-look laparotomy and were counseled about the surgery. Second-look surgery was to be performed within 8 weeks after the completion of chemotherapy.

Three patients were entered at each dose level. In the first stage of the study, only the dose of melphalan was escalated. The planned dose escalation scheme for melphalan was as follows: −1, 4 mg/m²; 0, 6 mg/m²; +1, 10 mg/m²; +2, 14 mg/m²; and +3, 16 mg/m². All patients at each dose level were observed for at least 3 weeks before proceeding to the next dose level. Six patients were to be entered at any dose level that produced dose-limiting toxicity in a single patient. Dose-limiting toxicity was assessed in each cycle of therapy and was defined as Grade 3 or 4 nonhematologic toxicity, Grade 2 nonhematologic toxicity that failed to reverse by Day 29, Grade 3 or 4 neutropenia ≥7 days in duration, neutropenic fever, or Grade 4 thrombocytopenia. If a total of 3 of 6 patients in a cohort developed Grade 3 neutropenia of ≥7 days' duration or if 2 of 6 patients developed Grade 4 neutropenia of ≥7 days' duration, neutropenic fever, or Grade 4 thrombocytopenia at a given dose level, then the immediately preceding dose level was considered the MTD. At least 6 patients were to be studied at the MTD level.

If the dose-limiting toxicity that determined the initial MTD was platelet toxicity, then, in a second stage of the study, the dose of paclitaxel was to be escalated while the doses of melphalan and cisplatin were kept constant. The planned dose escalation scheme for paclitaxel was as follows: 0, 135 mg/m²; +1, 150 mg/m²; +2, 175 mg/m²; +3, 200 mg/m²; +4, 225 mg/m²; and +5, 250 mg/m². The dose-limiting toxicities were the same as those for the initial portion of the study. Although the study was begun with the idea of using the same criteria for MTD of paclitaxel that were used for the MTD of melphalan, the study was revised after discussions with our biostatisticians (J. J. Lee and P. F. Thall) to determine the MTD of paclitaxel when combined with the established dose of melphalan and cisplatin 75 mg/m², using the continuous reassessment method with target toxicity of 33%.23, 24

Evaluation During Study

All eligible patients were evaluated for response and toxicity. All toxicities encountered during the study were evaluated according to the GOG Common Toxicity Criteria and recorded.25

During treatment, patients were followed twice weekly with a complete blood count, differential count, and platelet count. Before each cycle, a history and physical examination, tumor measurements, response data, performance status, toxicity notation, complete blood count, differential count, platelet count, chemical survey, SGOT, serum magnesium, and serum CA 125 were performed. If a CT scan of the abdomen and pelvis was the sole documentation of measurable disease, then it was repeated just prior to Cycle 4 (midway through the planned treatment duration). Other studies, such as chest radiographs, electrocardiograms, and audiograms, were performed at any time during the study if they were indicated clinically. For all patients who were clinically disease free on physical examination after 6 cycles of chemotherapy, a serum CA 125 level and CT of the abdomen and pelvis were obtained approximately 4 weeks after the start of Cycle 6. Patients who were clinically free of disease at the completion of the planned therapy (as documented by a normal physical examination, normal CT scan of the abdomen and pelvis, and serum CA 125 level < 100 mU/mL) were offered second-look laparotomy to determine disease status. Second-look laparotomy was to be performed within 8 weeks of the completion of therapy.

Palpable lesions were measured in centimeters before each cycle of therapy. The greatest dimension of the tumor and its perpendicular dimension were measured for bidimensionally measured lesions. Size was reported as the product of these dimensions. Measurements were made and recorded by the attending physician and the oncology research nurse. An estimate of overall objective response was made and recorded at each visit.

A clinical complete response (CR) was defined as the disappearance of all clinical evidence of tumor for a minimum of 4 weeks. In addition, patients must have had a normal serum CA 125 level to be considered complete responders. A partial response (PR) was defined as a decrease ≥ 50% in the sum of the products of the greatest dimensions of measured lesions for a minimum of 4 weeks; no simultaneous increase in the size of any lesion or the appearance of new lesions was allowed. Stable disease (SD) was defined as a change in measurable disease too small to meet the requirements for PR or disease progression and the appearance of no new lesions for at least 8 weeks with no worsening of symptoms. Progressive disease (PD) was defined as an increase of at least 25% in the size of any measured lesion or the appearance of any new lesions.

A pathologic complete response (PCR) was defined as no histologically viable tumor present in any of the biopsy specimens and cytologic washings performed at second-look surgery. A pathologic partial response (PPR) was defined as a decrease ≥ 50% in the sum of the greatest perpendicular dimensions of all measured lesions when comparing the operative findings at second-look surgery with known, clinically detectable tumor (by physical examination or imaging study) for patients with measurable disease or with residual tumor (by operative note from the primary surgery) for patients with nonmeasurable disease; no simultaneous increase in the size of any lesion or the appearance of any new lesions was allowed in this category.

Dose Modifications

A white blood count ≥ 3000/μL, a neutrophil count ≥ 1500/μL, a platelet count ≥ 100,000/μL, and a recovery of nonhematologic toxicity to Grade 1 or less were required for retreatment. In the case of any patient who experienced a delay of >42 days between chemotherapy doses, melphalan was to be discontinued, and the patient was to continue receiving the combination of paclitaxel and cisplatin.

No reduction in cisplatin dose was made for any degree of hematologic toxicity. Grade 4 granulocyte toxicity or Grade 3 platelet toxicity required a 1 dose level reduction of melphalan. In the next cycle, the dose of melphalan was escalated back to the starting dose except in patients with persistent Grade 3 or 4 hematologic toxicity. Anemia was not an indication for dose reduction. In the second portion of the study, once the +1 dose level of paclitaxel was reached, the dose reduction criteria for hematologic toxicity changed. For Grade 3 platelet toxicity, the melphalan dose was to be reduced by 1 level; for Grade 4 platelet toxicity or febrile neutropenia, both the melphalan dose and the paclitaxel dose were to be reduced by 1 level.

For toxicity > Grade 2 (other than symptomatically controlled bone pain) related to G-CSF, the drug was to be withheld until toxicity decreased to Grade 2. If toxicity > Grade 2 occurred, then, in subsequent cycles, the dose of G-CSF was to be reduced by 50%. If Grade 3 or 4 toxicity recurred, then G-CSF was to be discontinued. If a patient had recurrent local problems at the injection sites, then the G-CSF was to be divided and given twice daily.

Persistent elevation of serum creatinine to >1.8 mg% required withholding treatment until the creatinine level returned to the normal range. If the creatinine elevation persisted beyond 6 weeks after a previous dose, then cisplatin was to be discontinued.

Grade 2 peripheral neuropathy required a 50% reduction in cisplatin dose. Grade 3 or 4 peripheral neuropathy required discontinuation of cisplatin. Tinnitus required a 50% reduction in cisplatin dose. Symptomatic hearing loss required discontinuation of cisplatin.

Cardiovascular toxicity of any kind was to be evaluated by a cardiologist. Paclitaxel infusion was to be discontinued for a cardiac arrhythmia that showed evidence of atrioventricular node block. No adjustments were allowed for gastrointestinal toxicity. In patients who experienced severe hypersensitivity reactions to paclitaxel, they were to be rechallenged at the discretion of the study chairman. In such cases, premedication was to consist of dexamethasone 8 mg IV at 24 hours, 18 hours, 12 hours, and 6 hours prior to paclitaxel; cimetidine 300 mg IV 30 minutes prior to paclitaxel; and diphenhydramine 50 mg IV 30 minutes prior to paclitaxel. In addition, paclitaxel was to be administered in the usual volume but at one-quarter of the planned rate over the first 6 hours. If a severe reaction still developed, then patients were to go off study.

Criteria for Removal from Study

Patients were removed from the study for any one of the following reasons: evidence of disease progression after a minimum of two cycles of therapy; the development of unacceptable toxicity; or patient refusal or inability to comply with protocol requirements.

Statistical Considerations

Patient characteristics, toxicity, and response rates were summarized in tabular form. Logistic regression analysis was used to assess the effects of melphalan and paclitaxel doses on the probability of toxicity, defined as the proportion of cycles in which any dose-limiting toxicity was observed. Progression free survival and overall survival were estimated by using the Kaplan–Meier method.

RESULTS

Patient Characteristics

Between January, 1993 and May, 1996, 34 women with advanced stage epithelial ovarian carcinoma or peritoneal carcinoma were enrolled in this Phase I study. Thirty-three of the 34 patients underwent primary surgery prior to enrolling in the study. One patient's diagnosis was based on cytologic analysis of an effusion; she was not considered a good surgical candidate because of her debilitation from tumor and effusions, but subsequent second-look surgery confirmed the diagnosis of ovarian carcinoma. All patients were evaluable for toxicity and antitumor activity. Twenty-nine patients had measurable disease and were assessable for clinical response. Patient characteristics are presented in Table 1.

Table 1. Patient Characteristics (N = 34)

Characteristic	No.	%
FIGO: International Federation of Gynecology and Obstetrics.
Age (yrs)
Median	56	—
Range	41–71	—
Zubrod performance status
0	0	0
1	29	85
2	5	15
Disease type and FIGO stage
Ovarian carcinoma
Stage IIIc	13	38
Stage IV	14	41
Peritoneal carcinoma	7	21
Histologic type
Serous	18	53
Mixed epithelial	10	29
Undifferentiated	2	6
Unclassified adenocarcinoma	3	9
Clear cell	1	3
Measurable disease
Yes	29	85
No	5	15

Dose Escalation

A total of 192 cycles of chemotherapy were delivered to the 34 patients. The median number of cycles per patient was 6, with a range of 2–6 cycles. Table 2 presents information on the 6 dose levels along with the observed toxicity rates for each dose level.

Table 2. Treatment Schedule per Dose Level

Dose level	No. of patients	Dose (mg/m^2)			Total no. of cycles	Observed toxicity (%)
		Paclitaxel	Melphalan	Cisplatin
1	6	135	6	75	36	2.8
2	6	135	10	75	34	5.6
3	4	135	14	75	22	20.8
4	6	150	10	75	32	15.6
5	6	175	10	75	36	22.2
6	6	200	10	75	32	13.9

Toxicity

No treatment-related deaths occurred in this study. Of the 192 cycles delivered, dose reductions were required in 51 cycles (27%): Dose reductions of melphalan were necessary in 37 cycles, dose reductions of cisplatin were necessary in 11 cycles, and dose reductions of paclitaxel were necessary in 9 cycles. Treatment delays of 1 week because of leukopenia occurred in 2 cycles.

Hematologic toxicity is detailed in Table 3. The nadir granulocyte count typically occurred on Days 8–10 of each cycle, with a median of 9 days to reach the nadir granulocyte count. Grade 3 or 4 neutropenia, as noted above, was a moderate problem, but the incidence of neutropenic fever was relatively low. Grade 3 or 4 thrombocytopenia did occur with higher doses of melphalan. Anemia was not a significant problem in this study population, with only 1 patient requiring transfusion of 2 units of packed cells in 1 cycle.

Table 3. Hematologic Toxicity (Gynecologic Oncology Group Criteria) per Cycle in 192 Assessable Cycles (N = 34)

Dose level	No. of cycles	Grade 3 neutropenia	%	Grade 4 neutropenia	%	Neutropenic fever	%	Grade 3 thrombocytopenia	%	Grade 4 thrombocytopenia	%
1	36	7	19	13	36	1	3	3	8	0	0
2	34	15	44	9	26	0	0	6	18	1	3
3	22	4	18	5	23	3	14	6	27	3	14
4	32	3	9	15	47	2	6	2	6	3	9
5	36	7	19	18	50	6	17	9	47	3	8
6	32	5	16	17	16	3	25	10	31	1	9

Nonhematologic toxicity is detailed in Table 4. In most cases, nausea and vomiting was nonexistent or mild. Grade 2 or 3 peripheral neuropathy was noted in only 4 cycles (2%). Tinnitus was reported in 4 patients in 57 cycles (4%).

Table 4. Nonhematologic Toxicity (Gynecologic Oncology Group Criteria) per Cycle in 192 Assessable Cycles

Toxicity	Grade
	0	1	2	3	4
Nausea and emesis	136	50	5	1	0
Peripheral neuropathy	152	36	2	2	0
Hearing	185	0	7	0	0
Skin rash	190	1	1	0	0
Allergy	191	0	1	0	0
Mucositis	191	1	0	0	0

The MTD of melphalan was reached at a dose level of 10 mg/m², with 2 of 4 patients experiencing Grade 4 thrombocytopenia in 3 cycles at the next higher dose level of 14 mg/m². In addition, at the 14 mg/m² dose level, Grade 3 thrombocytopenia was observed in all 4 patients during 6 cycles.

Using the standard criteria used to establish the MTD of melphalan, the MTD of paclitaxel was established at a dose level of 135 mg/m², with 2 of 6 patients experiencing neutropenic fever in 2 cycles at the next higher dose level of 150 mg/m². However, the study was continued with escalation of the paclitaxel dose, because we had revised the protocol to employ the continuous reassessment method for this second stage. The fact that the group of patients who received 10 mg/m² melphalan plus 200 mg/m² paclitaxel (10 of 200 patients) had a lower observed toxicity rate (13.9%) than the group of patients that received 10 mg/m² melphalan plus 175 mg/m² paclitaxel (10 of 175 patients; 22.2%) is an artifact of small sample size and random variation, because, in general, a higher chemotherapy dose is associated with greater toxicity. The study was terminated prior to reaching the target toxicity of 33% based on the continuous reassessment method because of a high rate of neurotoxicity observed in a concurrent study of cisplatin in combination with escalating doses of paclitaxel above 200 mg/m².

The fitted regression model, therefore, was logit [prob(toxicity)] = −5.71 + 0.22 M + 0.0094 P, in which prob(toxicity) is the probability of toxicity, M is the melphalan dose, and P is the paclitaxel dose. Figure 1 shows the contours of the curve for constant predicted prob(toxicity) = 0.05, 0.10, 0.15, and 0.20. Goodness-of-fit analyses indicated that this model provided a reasonable fit for these data (residual deviance = 40.1 with 31 degrees of freedom).

Figure 1. Contour plot of the probability of toxicity (ptox) of melphalan and paclitaxel doses. The points indicate the observed toxicity rate at each dose combination, and the lines correspond to the predicted probability of toxicity under the logistic model.

Antitumor Activity

At the completion of chemotherapy, 20 of 34 patients (59%) were clinically disease free, as indicated by a normal physical examination, CT scan of the abdomen and pelvis, and serum CA 125 level. Of the 29 patients who had clinically measurable disease prior to starting chemotherapy, 15 patients (52%) had a CCR, 8 patients (28%) had a CPR, and 6 patients (21%) had no response.

Eleven patients (32%) who were candidates for second-look surgery consented to and underwent the procedure. Five patients (45%) had a PCR, and 6 patients (55%) had a PPR. Of the 6 partial responders, 2 had only persistent microscopic tumor, and 4 had persistent small macroscopic tumors. When all 34 patients in the study are included in the analysis—those who underwent second-look surgery and those who did not–the PCR rate was 15%.

The median progression free survival was 16.8 months for the entire study group, and the median survival was 32.8 months. At the time of the analysis, 9 patients (26%) were alive without evidence of disease, 6 patients (18%) were alive with disease, and 19 patients (56%) had died of disease.

DISCUSSION

For almost 2 decades, platinum-based combination chemotherapy has been the standard postoperative treatment for patients with advanced epithelial ovarian carcinoma. Recently reported studies of cyclophosphamide in combination with a platinum drug (cisplatin, carboplatin, or both) in patients with suboptimal (>1 or 2 cm), Stage III and IV epithelial ovarian carcinoma indicate a median progression free survival rates of 12–18 months and a median survival rate of 17–27 months.26–29 There is no evidence that the 5-year survival rate for such patients has exceeded 15–20%.

In the mid-1990s, the GOG's randomized trial firmly established the combination of paclitaxel and a platinum drug as standard postoperative treatment for patients with advanced epithelial ovarian carcinoma in the United States.9 Ongoing clinical trials are attempting to identify the optimal paclitaxel dose, the optimal duration of infusion of paclitaxel, and the relative toxicity and efficacy of cisplatin compared with carboplatin. Building on the foundation of previous studies with paclitaxel and cisplatin, the current study was initiated in an effort to develop a new combination chemotherapy regimen that possibly may have acceptable toxicity and enhanced activity for patients with advanced epithelial ovarian carcinoma. We also included patients with peritoneal carcinoma, because several studies, including our own, have suggested that this malignancy is similar to advanced epithelial ovarian carcinoma in terms of chemoresponsiveness and survival.30–33 We chose to add melphalan to this combination principally because a previous retrospective study of our experience suggested that melphalan may be a more active alkylating agent than cyclophosphamide against epithelial ovarian carcinoma when combined with cisplatin.20 In addition, our group had extensive experience with melphalan, and a parenteral formulation of melphalan became more accessible at the time we developed the trial. We were concerned about the leukemogenicity of melphalan but theorized that the potential advantages might offset the risk of subsequent leukemia by limiting the exposure to melphalan to no more than 6 cycles of treatment. Previous studies have indicated that the risk of nonlymphocytic leukemia from melphalan is related directly to the duration of exposure to the drug and the total cumulative dose received.18, 19

Other groups have pursued a strategy similar to ours by adding other drugs to the nucleus of paclitaxel and cisplatin. Kohn and colleagues have reported their experience with the combination of dose-intensive paclitaxel, cisplatin, and cyclophosphamide plus G-CSF.34 In a Phase I–II study, they treated 36 patients with advanced epithelial ovarian carcinoma, most of whom had bulky residual disease. The MTDs were 250 mg/m² for paclitaxel, 75 mg/m² for cisplatin, and 750 mg/m² for cyclophosphamide. The median survival was 26.5 months. The GOG currently is conducting two Phase I studies in patients with untreated ovarian carcinoma: In one study, topotecan is added to the combination of paclitaxel and cisplatin plus G-CSF; in the other study, oral etoposide is added to the combination of paclitaxel and carboplatin.

In general, the combination of paclitaxel, cisplatin, and melphalan plus G-CSF was well tolerated. In the initial phase of the study, as expected, the MTD of melphalan was established at a dose of 10 mg/m², with thrombocytopenia being the dose-limiting toxicity. Dose levels of melphalan between 10 mg/m² and 14 mg/m² were not studied, and it is possible that we may have been able to achieve a slightly higher dose of melphalan, e.g., 12 mg/m², if we had tested for it.

Nadir neutropenia typically occurred on Days 8–10 of each cycle and was of very brief duration. The use of G-CSF allowed us to deliver acceptable doses of this combination without a high incidence of severe neutropenia or neutropenic fever. When we designed this trial, G-CSF had just recently become commercially available, and we had minimal experience with its use. At the initial dose level, G-CSF was administered for a total of 19 days, as noted above. Thereafter, the protocol was revised to include G-CSF administration for a total of 12 days on each cycle. Although it was not a focus of our study, it is quite possible that the duration of G-CSF administration could have been limited to as few as 7 days, which is our current standard practice when treating with the combination of paclitaxel and cisplatin.

Based on preliminary information available at the time this trial was designed, the dose of paclitaxel was escalated in the second stage. Sarosy et al.35 had observed a 50% response rate to paclitaxel at doses up to 250 mg/m² in patients with refractory ovarian carcinoma. In the second stage of the study, the MTD of paclitaxel, as defined using conventional criteria for defining the MTD, was reached at a dose level of 135 mg/m² based on granulocyte toxicity at the next higher dose level. However, on the basis of exploratory discussions with our biostatisticians regarding the use of the continuous reassessment method to identify a safe dose to use in a Phase II trial, we revised the protocol and continued the study with escalation of the paclitaxel doses.24, 36 Using this approach, the probability of dose-limiting toxicity was <25% at dose levels of paclitaxel up to 200 mg/m². Investigators conducting future Phase I studies of new drugs or combinations of drugs for the treatment of patients with ovarian carcinoma should consider employing the continuous reassessment method rather than the standard method to determine a safe dose to use in subsequent Phase II trials. However, this method was designed originally to be used for single-agent trials: Phase I studies of multidrug combinations introduce a new level of complexity.

Although antitumor activity was not the primary end point of this study, the combination of melphalan, paclitaxel, and cisplatin plus G-CSF showed good activity. Our study population was a very poor prognostic group; 18 of the 34 patients (53%) had either Stage IV ovarian carcinoma (n = 14 patients) or primary peritoneal carcinoma with a cytologically positive pleural effusion (n = 4 patients). Our median survival time of 32.8 months is very much in the range observed in the most recently reported GOG study.37 In that prospective randomized trial of patients with suboptimal Stage III and IV epithelial ovarian carcinoma, 615 evaluable patients were randomized to receive either single-agent cisplatin, single-agent paclitaxel, or the two-drug combination; median survival times in the three arms of the study were 30.2 months, 26.0 months, and 26.6 months, respectively. The findings of our study appear to warrant a Phase II study of this combination. In such a study, the recommended dose of melphalan would be 10 mg/m². The optimal dose of paclitaxel remains unclear. Unless new information indicates a definite benefit associated with high dose paclitaxel (in the range of 200–250 mg/m²), we would recommend a dose of 135–175 mg/m². Of course, any final recommendation of the addition of melphalan to the combination of paclitaxel and cisplatin plus G-CSF would require proof of a significant improvement in patient outcome, because this combination obviously would be associated with greater cost and potential toxicity.

There are now several drugs that have demonstrated activity in platinum-resistant disease: topotecan, gemcitabine, liposomal doxorubicin, oral etoposide, and vinorelbine. Most of these are already being tested in two-drug and three-drug combination regimens in patients with untreated advanced ovarian carcinoma. The combination of paclitaxel, cisplatin, and melphalan will be judged against these regimens in the design of future Phase III randomized trials.