- © 1999 by American Society of Clinical Oncology
Vinblastine Versus Vinblastine Plus Oral Estramustine Phosphate for Patients With Hormone-Refractory Prostate Cancer: A Hoosier Oncology Group and Fox Chase Network Phase III Trial
- Gary Hudes,
- Lawrence Einhorn,
- Eric Ross,
- Andrew Balsham,
- Patrick Loehrer,
- Harry Ramsey,
- John Sprandio,
- Michael Entmacher,
- William Dugan,
- Rafat Ansari,
- Frank Monaco,
- Mark Hanna and
- Bruce Roth
- From the The Fox Chase Network and Division of Biostatistics, Fox Chase Cancer Center, Philadelphia, PA; and Hoosier Oncology Group, Walther Cancer Institute, and Division of Biostatistics, Indiana University, Indianapolis, IN.
- Address reprint requests to Gary Hudes, MD, Fox Chase Cancer Center, 7701 Burholme Ave, Philadelphia, PA 19111.
Abstract
PURPOSE: To compare vinblastine versus the combination of vinblastine plus estramustine as treatment for patients with hormone-refractory prostate cancer (HRPC).
PATIENTS AND METHODS: A total of 201 patients with metastatic prostate cancer, progressive after hormonal therapy and antiandrogen withdrawal (if prior antiandrogen treatment), were randomized to receive vinblastine (V) 4 mg/m2 by intravenous bolus weekly for 6 weeks followed by 2 weeks off, either alone or together with estramustine phosphate (EM-V) 600 mg/m2 PO days 1 through 42, repeated every 8 weeks. Of 193 eligible patients, 98 received V, and 95 received EM-V.
RESULTS: Overall survival trended in favor of EM-V but was not significantly different as determined by Kaplan-Meier analysis (P = .08). Median survival was 11.9 months for EM-V and 9.2 months for V. EM-V was superior to V for secondary end points of time to progression (P < .001, stratified log rank test; median 3.7 v 2.2 months, respectively) and for proportion of patients with ≥ 50% prostate-specific antigen (PSA) decline sustained for at least 3 monthly measurements (25.2% v 3.2%, respectively; P < .0001). Granulocytopenia was significantly less for EM-V compared with V (grade 2, 3, and 4 = 7%, 7%, and 1% v 27%, 18% and 9%, respectively; P < .0001); however, grade 2 or worse nausea (26% v 7%, respectively; P = .0002) and extremity edema (22% v 8%, respectively; P = .005) were more frequent for EM-V.
CONCLUSION: Although overall survival was not significantly greater for the combination, EM-V was superior to V for time to progression and PSA improvement. These results encourage further study of estramustine-based antimicrotubule drug combinations in HRPC.
AS THE LEADING CAUSE of noncutaneous malignancy and the second leading cause of cancer death in men, adenocarcinoma of the prostate has attracted increasing attention in the past decade. Although prostatic carcinoma is diagnosed earlier because of the widespread use of serum prostate-specific antigen (PSA), the mortality rate from this tumor remains relatively constant at approximately 40,000 deaths per year.1 Virtually all deaths are a consequence of metastatic disease most often involving skeletal sites. Although symptomatic improvement and control of metastatic disease is obtained in most patients after androgen ablation, disease progression occurs at a median of 12 to 18 months. Secondary remissions with additional hormonal maneuvers are less frequent and usually of shorter duration. Chemotherapy with cytotoxic agents has not extended the median survival of approximately 10 months for patients with hormone-refractory metastatic prostate cancer. Consequently, identification of active agents and combinations will be necessary for progress against both “hormone-sensitive” and hormone-refractory disease. Effective therapies must target androgen-independent tumor cells.
Estramustine (EM) is a nornitrogen mustard-estradiol conjugate, with both hormonal and nonhormonal effects in man. EM inhibits microtubule function by binding to tubulin2,3 and microtubule-associated proteins.4 As a single agent, EM has limited activity in patients with hormone-refractory prostate cancer (HRPC). In a contemporary phase II trial using PSA and other clinical criteria to gauge response, 14% of 44 patients with HRPC responded to oral EM 15 mg/kg daily.5
The antimicrotubule properties of EM, especially the unique interaction with microtubule-associated proteins, led to the hypothesis that synergistic antimicrotubule effects and cytotoxicity could be achieved by combining EM with other microtubule inhibitors.6 In support of the hypothesis, additive or greater antimicrotubule effects were observed in vitro with the combination of EM plus vinblastine (V).7 The lack of clinical myelotoxicity or neurotoxicity associated with EM and the differing patterns of resistance for EM and V in vitro8 provided further impetus for clinical trials of EM combined with V (EM-V).
A total of 83 patients were treated with the EM-V combination in three nonrandomized trials.9-11 Although the response criteria varied, a PSA decrease greater than 50% occurred in 42% of the patients, and objective responses were observed in six (31%) of 17 patients with measurable soft tissue disease. This interesting activity formed the basis for randomized trials conducted to determine if the two-drug combination was superior to the single agents.
Although there was considerable (and generally disappointing) experience with EM as a single agent, less was known about the activity of V in HRPC. In the only published phase II trial, Sella et al12 reported eight objective responses in 39 patients (20.5%; 95% confidence interval [CI], 6.3% to 34.7%) treated with V 1.5 to 1.7 mg/m2/d for 5 days by continuous intravenous infusion every 3 weeks. With this background, the Hoosier Oncology Group and the Fox Chase network initiated a phase III trial of V alone versus EM-V for patients with metastatic HRPC.
PATIENTS AND METHODS
Patient Selection
Eligibility required histologically proven adenocarcinoma of the prostate and disease progression after primary hormonal therapy. A trial of antiandrogen withdrawal for a minimum of 4 weeks with continued evidence of progression was required for patients taking antiandrogen therapy before registration. Because patients on this trial had not received bicalutamide or nilutamide, longer periods of withdrawal were not required. The presence of metastatic tumor with either bidimensionally measurable disease or assessable disease (eg, bone metastases on radionuclide bone scan) was required. Patients with an elevated tumor marker (eg, PSA) as the sole evidence of tumor were not eligible. For protocol entry, progressive disease was considered to be either increasing palpable or radiographic tumor (eg, bone metastases or soft tissue lesion), increasing PSA, or a combination of radiographic and PSA progression, with or without worsening disease symptoms. Other requirements were no prior chemotherapy; Eastern Cooperative Oncology Group (ECOG) performance status of 0, 1, or 2; adequate organ function (defined as WBC ≥ 4,000/μL, platelets ≥ 100,000/μL, serum creatinine ≥ 2.0 mg/dL, and bilirubin ≥ 2.0 mg/dL); no history of myocardial infarction, active angina pectoris, or uncontrolled congestive heart failure within 6 months of registration; no clinically apparent CNS metastases; and no history of deep venous thrombosis. Patients with bidimensionally measurable soft tissue metastases were eligible, regardless of serum PSA value. Serum PSA ≥ 20 ng/mL was required for patients without measurable disease. Radiotherapy had to be completed at least 4 weeks before enrollment, and strontium-89 therapy completed at least 12 weeks before enrollment. The institutional review board of each participating institution approved the study protocol. All patients were required to give written informed consent before enrollment, according to institutional and federal guidelines.
Randomization and Treatment
Treatment was assigned in equal proportion using a stratified, permuted block randomization scheme, with central randomization through the Hoosier Oncology Group office. Patients were stratified according to ECOG performance score (0 or 1 v 2) and by presence or absence of bidimensionally measurable disease. Treatment with V alone consisted of V 4 mg/m2 body surface area given by intravenous bolus injection weekly for 6 consecutive weeks followed by 2 weeks off. Treatment with EM-V consisted of the identical dose and schedule of V plus EM 600 mg/m2 per day taken orally in two or three divided doses on days 1 through 42 of each 8-week cycle. The manufacturer provided estramustine phosphate (Emcyt; Pharmacia and UpJohn, Kalamazoo, MI) at no cost to patients randomized to EM-V. Patients were instructed to take Em 1 hour before or 2 hours after eating. Pill counts or diaries were not required to document compliance with EM. For both arms, treatment cycles were repeated every 8 weeks until disease progression. Patients who had not undergone orchiectomy were required to continue luteinizing hormone-releasing hormone agonist therapy while on study. Pre-study documentation of castrate-level serum testosterone was not required.
Doses of the drugs were modified each week for specific toxicities as follows: for WBC ≥ 3,000/μL and platelets ≥ 100,000μL, the full dose or V was given; for WBC 2,000 to 2,999/μL or platelets 75,000 to 99,000/μL, the dose of V was reduced by 50%; and for WBC less than 2,000/μL or platelets less than 75,000/μL, the dose of V was omitted for the week. The dose of EM was reduced by two capsules (280 mg) per day for vomiting or nausea that was unresponsive to oral antiemetics. For grade 3 or 4 nonhematologic toxicity caused by either drug, the dose of the offending agent was held until resolution of toxicity, and the drug was resumed at 50% of the starting dose for the remainder of protocol therapy.
Evaluations
Patients were evaluated with history and physical examination and toxicity assessment every 4 weeks. Blood counts were obtained weekly and PSA was measured every 4 weeks. Scans and x-rays were performed every 8 weeks to assess measurable disease, and bone scans were performed every 12 weeks. Pain and quality of life (QOL) were evaluated with the SF-36 Health Survey13 before the first, second, and sixth treatment cycles. Pain intensity and pain frequency were assessed on a 5-point scale. QOL evaluation times were selected to assess both early effects of treatment and long-term effects of continued therapy, particularly chronic toxicity that might adversely alter QOL.
Toxicity was graded according to the National Cancer Institute common toxicity criteria (Bethesda, MD). Standard solid tumor response14 criteria were used to monitor the effect of treatment on measurable soft tissue metastases. The following criteria were used to evaluate the effect of treatment on PSA: Partial response required 50% or greater decrease from baseline for three successive measurements obtained every 4 weeks; progression was defined as 50% or greater increase above baseline or the lowest value achieved for three consecutive measurements, with or without new symptomatic lesions on bone scan or increase in measurable disease. In addition to progression defined by PSA increase, patients with other signs or symptoms of cancer-related deterioration were removed from the study at the discretion of the treatment physician and were considered to have progressive disease. PSA stabilization was defined as neither partial response nor progression for at least 90 days. Overall survival and time to progression were determined for each patient from the time of registration.
Statistical Considerations
All analyses conducted were based on intention to treat. Baseline characteristics of patients in each treatment arm were summarized using descriptive statistics, and differences between the two arms were tested with Wilcoxon rank sum test or Fisher's exact test.
The primary end point of the trial was overall survival. Assuming a median survival of 10 months for the V arm, the trial had 80% power to detect a 50% increase in overall survival for EM-V patients with type I error of 0.05 (one-sided log-rank test), with planned accrual of 100 assessable patients per arm. Secondary end points were time-to-disease progression, response in measurable disease and PSA, and change in QOL measures. Progression-free survival and overall survival curves were estimated by the product-limit method.15 Survival duration was defined as time of registration until death or was censored on the date of last follow-up evaluation. In respect to time-to-progression measurement, individuals who died before progression occurred (n = 3, one on V, two on EM-V) were considered censored at the time of death. Comparisons of progression-free and overall survival distributions between treatment arms were conducted using one-sided stratified log-rank tests. The proportion of patients in each arm that achieved partial response in measurable disease, ≥ 50% decrease in PSA from baseline measurement, and improvement in pain were compared and analyzed with Fisher's exact test. All tests were conducted at a type I error of 0.05.
RESULTS
Between March 1993 and June 1995, 201 patients were registered at 19 Hoosier Oncology Group Institutions and 12 Fox Chase Network Institutions. A total of eight patients (4%) that registered were subsequently found to be ineligible (three on the V arm and five on the EM-V arm), which left 98 and 95 eligible patients on the two arms. The reasons for ineligibility were CNS metastases (two patients), ECOG performance status 3 (three patients), and failure to meet laboratory criteria for entry (three patients). The number of patients in each arm who completed at least one 8-week cycle of treatment was similar, 75 (77.3%) for V and 71 (74.7%) for EM-V. As listed in Table 1, patients in each group had similar median age, performance status, baseline hemoglobin, lactate dehydrogenase, and PSA values. The number of patients in each arm with measurable disease was also similar. Of the 193 eligible patients, 110 were enrolled after first-line hormonal therapy. Of the remaining 83 patients, 70 had second-line and 13 had third-line hormonal treatment (excluding flutamide withdrawal) before enrollment.
Patient Characteristics by Treatment Arm
Survival and Time to Progression
Overall survival was the primary end point of the trial. Median survival for the V arm was 9.2 months compared with 11.9 months for the EM-V arm. Although the survival function trended better for the EM-V arm compared with the V arm (Fig 1), this difference did not reach statistical significance (P = .08). In contrast, the time-to-progression curves (Fig 2) are significantly different (P < .0004) and favor EM-V. Median time to progression for EM-V was 3.7 months versus 2.2 months for V alone.
Kaplan-Meier survival curves.
Kaplan-Meier time-to-progression curves.
Increasing bone metastases was the reason for progressive disease in 44 patients (23.2%) at a median of 3.1 months from registration. Progression in bidimensionally measurable disease defined disease progression in 24 patients (12.6%) at a median of 2.1 months. Worsening symptoms, such as pain or decline in performance status considered to be poor subjective response, occurred in 41 patients (21.6%), and discontinuation of therapy was required at a median of 2.1 months. In addition, 51 patients (26.9%) stopped protocol treatment at their request or at the discretion of the treating physician at a median of 3.2 months. Disease progression classified solely by PSA increase, as defined by the protocol criteria, occurred in 29 patients (15.3%) at a median of 3.2 months from registration.
To investigate possible antigonadotrophic effects of EM metabolites in nonorchiectomized patients, some of whom may have had residual serum testosterone above the surgically castrate level, we compared time to progression and overall survival by treatment arm and orchiectomy status. For EM-V, median survival durations for orchiectomy (n = 65) and nonorchiectomy patients (n = 30) were 14.5 months and 10.0 months, respectively; and the survival function estimates for orchiectomy and nonorchiectomy patients differed significantly (P = .005, two-sided stratified log rank test) in favor of orchiectomy. For V alone, the median survivals for surgically castrated patients (n = 57) and medically castrated patients (n = 41) were 9.6 months and 7.8 months, respectively, with no difference between the estimated overall survival distributions (P = .869). Considering only patients who had orchiectomy, the overall survival function estimates were significantly different in favor of EM-V (P = .0439). In contrast to overall survival, treatment effect on time to progression was not influenced by orchiectomy.
Measurable Disease
A total of 63 patients had bidimensionally measurable soft tissue metastases, 33 on V and 30 on EM-V. There were no complete responses. A total of eight partial responses were documented in measurable disease, two in the V arm (6%; 95% CI, 0.0% to 14.1%) and six in the combination arm (20.0%; 95% CI, 6% to 35.4%). In addition, one minor response, which consisted of a greater than 25% but less than 50% reduction in the size of indicator lesions, occurred in each of the arms. The difference in partial response proportions between the two arms did not reach statistical significance (P = .131, two-tailed Fisher's exact test).
Effect of Treatment on Serum PSA
An additional secondary end point was PSA response to treatment (Table 2). Compared with treatment with V alone, a significantly greater proportion of patients treated with EM-V had ≥ 50% decrease of PSA that was sustained for at least three successive monthly measurements (3.2% v 25%, respectively; P < .0001). PSA decreases of 75% or greater were uncommon in both arms, occurring in 10.3% and 2.1% of patients treated with EM-V and V, respectively (P = .028). After tumor progression, 12 patients initially treated with V received the EM-V combination. From the measurement preceding the start of EM-V, three of these patients achieved a 50% or greater decrease in PSA.
V Versus EM-V: Effects on PSA
Pain and QOL
Compliance for pain and QOL assessments was poor. QOL evaluations were completed before (baseline) and after cycle 1 of treatment for only 45 (23%) of the 146 patients who completed one cycle of treatment. Of the 45 patients with pain and QOL data, 36 patients had pain at baseline (20 of 23 for V and 16 of 22 for EM-V). For pain frequency, eight (50%) of 16 patients treated with EM-V versus two (10%) of 20 patients treated with V alone improved by at least 1 point on a 5-point scale after cycle 1 (8 weeks) of treatment (P = .011, Fisher's exact test). For pain intensity, seven (35%) out of 20 patients treated with V versus two (12.5%) of 16 treated with EM-V improved by at least 1 point after one cycle of treatment (P = .24). The results of this analysis must be interpreted with caution because an additional 54 patients, 27 in each arm, reported baseline pain but did not complete the follow-up QOL assessments. Interpretation of these data is further limited by lack of narcotic diaries to correlate with the pain scores. In other measures of QOL, the mean scores for fatigue, measures of social function, and physical function were not significantly changed from baseline for patients treated with either V or EM-V.
Toxicity
Hematologic toxicity was chiefly granulocytopenia (Table 3). Unexpectedly, patients on the V arm experienced significantly more frequent and severe granulocytopenia. Granulocytopenia of grade 2, 3, or 4 occurred in 27, 18, and nine patients, respectively, receiving V alone compared with only seven, seven, and one patient, respectively, treated with EM-V. The difference in the proportion of patients in each group who experienced grade 2 or higher toxicity was significant (P < .0001 Fisher's exact test). The higher incidence of granulocytopenia in the V arm led to a greater percentage of patients who required reduction of one or more doses of V during the first two cycles (71.4% for V v 29.5% for EM-V). However, infectious complications were similar in the two arms, eight for V alone and seven in patients treated with EM-V. There were no grade 4 infections. Thrombocytopenia was infrequent, with only three patients (two on V and one on EM-V) experiencing grade 2 or higher toxicity.
Granulocytopenia by Treatment Arm
Nonhematologic toxicities are summarized in Table 4. Nausea and leg edema of grade 2 or higher were significantly more common in patients receiving EM-V (P = .0001 and P = .028, respectively), whereas constipation was significantly more common in patients receiving V alone (P = .049). Other neurotoxicity and fatigue occurred with equal frequency on the two treatment arms. Intolerable nausea led to reduction of EM dosage during the initial two cycles of treatment for 28 patients on the EM-V arm.
Nonhematologic Toxicity by Treatment Arm
Serious cardiovascular toxicity was uncommon in both arms. Three EM-V patients had deep venous thrombosis in a lower extremity. Three EM-V patients developed congestive heart failure (two grade 3 and one grade 4), compared with one V patient (grade 4). Other cardiovascular adverse events in the EM-V arm were grade 1 hypotension, grade 1 palpitations, and grade 1 atrial fibrillation in one patient each. In the V arm, four additional patients experienced grade 1 cardiac events, which consisted of palpitations, atrial flutter, atrial tachycardia, and hypotension in one patient each.
DISCUSSION
The results of this randomized trial demonstrate differences in time to progression and improvements in PSA that favor the combination of V and EM. Despite these differences in secondary outcome measures, overall survival for patients on the two treatments was not significantly different. With a sample size adequate to detect a 50% or greater difference in survival, this trial was not designed to detect smaller differences that might also be considered clinically meaningful. For example, the power of statistically detecting a 25% difference in overall survival (the approximate difference observed in median survival), assuming a one-sided test with the available sample sizes, was estimated to be only 51% using the methods of Schoenfeld and Richter.16
As predicted from previous trials, a greater proportion of patients who received EM-V experienced nausea and fluid retention; the latter usually consisted of mild to moderate peripheral edema. The incidence of other cardiovascular toxicity was not significantly different between the two arms. An unexpected observation was the striking difference in granulocytopenia, with grade 3 or 4 toxicity in 27% of patients receiving V alone compared with only 7% of patients receiving EM-V. The mechanism of this apparent protective effect of EM on V-induced granulocytopenia is unclear. Patients receiving single-agent EM often demonstrate increased WBC and granulocyte counts,17 an observation that is consistent with increased mobilization or production of granulocytes. Alternatively, a pharmacologic interaction of EM and V, which might have caused reduced plasma concentrations or cumulative exposure to V, cannot be excluded.
EM has both hormonal and nonhormonal cytotoxic effects in vivo.18 Approximately 20% of the administered dose of EM is metabolized to estrone and estradiol, resulting in concentrations sufficient to lower serum testosterone to castrate levels.19 The hormonal effects of EM may be particularly relevant for patients with incomplete suppression of testosterone. Approximately 60% of patients in each arm of this trial had bilateral orchiectomy, and the remaining patients continued on luteinizing hormone-releasing hormone agonist or diethylstilbestrol (two patients) during protocol treatment. Although it is unlikely that significant elevations of serum testosterone were common in either group, we did not require measurement of serum testosterone to document castrate levels. To examine the potential effects of EM metabolites on outcomes in the EM-V arm, we compared time to progression and overall survival in medically and surgically castrated patients. In this retrospective analysis, time to progression was not affected by orchiectomy, whereas an interaction of orchiectomy and treatment was observed for overall survival. However, the greater survival for EM-V compared with V in the orchiectomy subgroup rather than in nonorchiectomy patients is evidence that further lowering of testosterone was not an important factor on treatment outcome. Because estrogens may be active by mechanisms other than antigonadotrophic action, we cannot definitely exclude the possibility that estrogenic metabolites of EM contributed to the outcomes favoring EM-V.
The present trial and three other recently completed phase III trials20-22 in patients with HRPC demonstrate that achieving pain relief, measurable disease response, and reduction of PSA may not translate into major improvement in overall survival (eg, a 50% increase) for an entire treatment group. Although it is possible that crossover of patients to the more active arm in each of these trials could obscure differences in overall survival between two treatments, the lack of a major increase in survival is more likely because of the modest antitumor activity of all of the therapies used in these trials. Future phase III trials that use overall survival as a primary end point should have adequate statistical power to detect smaller differences in overall survival.
Although the combination of EM-V was not shown to be significantly better than V alone in overall survival, the superior results of the combination in time to progression and PSA encourage further clinical development of EM-based combination antimicrotubule therapy in HRPC. Preliminary results obtained with paclitaxel-EM23 and docetaxel-EM24,25 combinations suggest activity superior to that of the EM-V combination. Thus, the optimal combination and schedule for EM-based antimicrotubule therapy has yet to be determined.
There is increasing evidence that EM and other tubulin/microtubule-binding agents block mitotic progression by a common mechanism of inhibiting the dynamic instability of spindle microtubules.26 For EM and V, inhibition of microtubule dynamics in vitro by the combination is greater than additive, considering the inhibitory effects of each drug alone.27 It is likely that other antimicrotubule drug combinations will demonstrate additive or greater inhibition of microtubule dynamics and clinical antitumor activity. The encouraging clinical results, including the present randomized trial, support further study of antimicrotubule drug combinations to produce more effective treatment for HRPC.
- Received September 15, 1998.
- Accepted June 14, 1999.
