A Phase II Multicenter, Randomized, Double-Blind, Safety Trial Assessing the Pharmacokinetics, Pharmacodynamics, and Efficacy of Oral 2-Methoxyestradiol Capsules in Hormone-Refractory Prostate Cancer

The need for new therapies for prostate cancer is paramount. Prostate cancer is the most common male malignancy in the United States with ∼200,000 new cases of prostate cancer with ∼32,000 deaths per year (1). Standard initial therapy for metastatic disease consists of androgen ablation and patients will become refractory to hormonal therapy after a median time of ∼24 months (2). Only recently has any therapy been shown to prolong overall survival. Specifically, randomized phase III trials of docetaxel-based chemotherapy have shown a modest survival benefit when compared with mitoxantrone plus prednisone (3, 4). Clearly, new therapies are required.

Angiogenesis, the formation of new microvessels from existing vasculature, is fundamental to progressive tumor growth. In the absence of neovascularization, tumor growth is limited to 1 to 2 mm³. Over the last two decades, substantial laboratory and indirect clinical evidence has accumulated to support the central role of angiogenesis in prostate cancer progression. Therefore, the inhibition of angiogenesis may be a viable strategy for the treatment of hormone-refractory prostate cancer. The supporting data include the following. Increased microvessel density in prostate cancer is associated with a shorter time to recurrence and higher stage after a radical prostatectomy as well as a shorter time to recurrence after radiation therapy (5–7). Vascular endothelial growth factor (VEGF) is a potent and specific stimulator of endothelial cell proliferation and angiogenesis (8), and VEGF expression, measured in prostate cancer cytosolic extracts, correlates with microvessel density and is associated with relapse after a radical prostatectomy (9). Furthermore, VEGF is found at higher levels in the plasma of patients with metastatic prostate cancer than in the plasma of patients with localized disease or in the plasma of healthy controls (10). These data are consistent with findings from other cancers in which VEGF has been elevated in urine and plasma and is typically detected at higher levels in cancer patients than in the healthy volunteers (11). Basic fibroblast growth factor (bFGF) is also produced by prostate cancers and is found at increased levels in the serum of patients with prostate cancer when compared with levels in men without prostate cancer. Inhibition of angiogenesis may therefore be a viable strategy for the treatment of prostate cancer.

2-Methoxyestradiol (2ME₂) is an estrogen metabolite that is naturally formed in vivo by the sequential hydroxylation and O-methylation of estradiol at the 2-position. 2ME₂ binds very poorly to the estrogen receptor (0.05% of estradiol binding; ref. 12) and does not exhibit direct estrogenic activity in many test systems. Several investigations have provided encouraging results for the use of 2ME₂ in the treatment of cancer. Preclinical studies have indicated that 2ME₂ is both antiproliferative by acting directly on the tumor cell compartment and antiangiogenic. In vitro, 2ME₂ inhibits proliferation of many cancer and endothelial cell lines in the submicromolar to low micromolar range and the activity is independent of the estrogen responsiveness of the cell line. For example, the IC₅₀ value for the hormone-independent prostate cancer cell line, DU-145, was 1.8 μmol/L. The inhibition of proliferation is believed to result primarily from the induction of apoptosis possibly through the activation of p53 or death receptor 5 (13) as well as inhibition of hypoxia-inducible factor-1α, a key angiogenic transcription factor (14). Additionally, the rate of tubulin polymerization or depolymerization is inhibited by 2ME₂ in certain cell lines (15). In vivo, oral administration of 2ME₂ is effective in xenograft and metastatic disease models with no significant toxicity (16, 17). A concomitant reduction in tumor vasculature has also been observed in 2ME₂-treated animals (18). The antiangiogenic activity of 2ME₂ has been shown in vivo in corneal micropocket (16), Matrigel plug assays (19), and chick chorioallantoic model systems (20).

The doses of 2ME₂ assessed in this dose-ranging phase II study were based on the results of the phase I study in patients with locally recurrent or metastatic breast cancer. The phase I study evaluated doses ranging from 200 to 1,000 mg/d (125-625 mg/m²/d). This dose range was selected based on results obtained from nonclinical efficacy studies in mice and safety studies in rats and dogs. In vivo studies in murine models have indicated that substantial antitumor and antiangiogenic activities were observed in the range of 75 to 150 mg/kg/d (225-450 mg/m²/d). To obtain efficacy, safety, pharmacokinetic, and pharmacodynamic data on 2ME₂ in men, we conducted a randomized, placebo-controlled phase II study.

This study employed a phase II, multicenter, randomized, double-blind placebo-controlled design that assessed the safety, pharmacokinetic, and pharmacodynamic characteristics and efficacy of 2ME₂. The study was conducted at Indiana University and the University of Wisconsin Comprehensive Cancer Center after local institutional review board approval at both locations. Patients were eligible for this protocol if they had documented hormone-refractory disease with increasing prostate-specific antigen (PSA) by Prostate-Specific Antigen Working Group criteria (21) and/or evidence of metastatic disease and had not previously received chemotherapy. Enrollment required progression on medical or surgical androgen ablation, but castrate levels of testosterone were not stipulated. One patient had a total testosterone of 90 ng/dL and all other patients had levels of <30 ng/dL. If PSA was the only evidence of progressive disease, then it had to be >10 ng/mL. If a subject was receiving antiandrogen therapy, progression (at least 25% PSA increase) had to be shown after discontinuing this class of agent for at least 4 weeks for flutamide and 6 weeks for bicalutamide. Liver transaminases had to be less than twice the upper limit of normal, bilirubin <1.5 mg/dL, creatinine <1.8 mg/dL, total WBC count >3,000/mm³, platelets >100,000/mm³, international normalized ratio of prothrombin time ≤1.2, and activated partial thromboplastin time within 2 seconds of the upper limit of normal. A Karnofsky performance status of ≥70 and a stable pain management regimen for at least 2 weeks were required. Patients were also excluded if they had suffered a myocardial infarction within the preceding 3 months, uncontrolled angina in the last 3 months, or uncontrolled congestive heart failure. Major surgery within 21 days of starting 2ME₂, administration and coadministration of PC-SPES, or herbal medication containing saw palmetto also precluded enrollment.

Patients were randomized into one of two cohorts, with each cohort consisting of 16 patients. All patients in both cohorts were evaluated for safety as well as the effect of 2ME₂ on changes in tumor response and PSA levels. One cohort received 400 mg 2ME₂ and the other 1,200 mg 2ME₂ administered as a single daily oral dose in a double-blind manner. 2ME₂ was delivered as a capsule formulation, with each unit dose containing 200 mg 2ME₂ as a micronized crystalline solid. Randomization was done at each site and drug was dispensed from the representative Investigational Drug Services, with patients instructed to take two capsules from one bottle and four from a second bottle. In the 400 mg dose cohort, the second bottle had four placebo capsules. Each cycle of therapy consisted of 28 days of study drug administration. Patients were treated at the same dose level until the appearance of significant treatment-emergent toxicities or disease progression. Evidence of disease progression was defined as the appearance of new lesions >1 cm in size, unidimensional or bidimensional tumor measurements increasing >50%, or increase in PSA according to Prostate-Specific Antigen Working Group definitions (21) or symptomatic progression. Patients were not required to be removed from study due solely to PSA progression if the investigator felt the patient was potentially benefiting (e.g., slower PSA velocity).

Tumor response was evaluated after every cycle by using physical examination and patient interview to monitor for symptomatic progressive disease. Radiographic assessment of tumor response was done after every two cycles for the duration of therapy by using bone scan, chest X-ray, or computed tomography of the abdomen and pelvis (when disease was present at baseline).

Measurable disease. A complete response was defined as the complete disappearance of all clinically detectable disease measured by physical examination and/or radiographic studies for a period of at least 4 weeks. A partial remission was defined as a >50% decrease in the sum of the products of the two longest perpendicular diameters of all measurable lesions for a period of at least 4 weeks without an increase of >25% in the size of any area known to contain malignant disease and without the appearance of any new areas of malignancy. Progressive disease was defined as an increase of at least 25% in the size of measurable lesions.

Serologic disease. A partial serologic response was at least a 50% decrease in PSA for three consecutive measurements 4 weeks apart (i.e., a sustained response for 2 months). Serologic progressive disease was defined as a 50% increase in PSA from the lowest level recorded and was confirmed 4 weeks later on the study. Stable serologic disease was scored if neither of these variables were met and the patient in consideration did not have measurable disease progression or increase pain or palliative radiation. Time to progression was defined as the period measured from the initiation of therapy until at least a 50% increase in PSA level confirmed at least 4 weeks later, taking as reference the lowest PSA level recorded since the treatment started. The date of the first recorded increase was deemed the date of progression.

Patients were enrolled until 16 patients were treated in each arm. If two or more of these 16 patients showed a tumor response (including PSA response), then an amendment was to be made to allow an additional 16 patients to be enrolled in that arm. If six or more of the 32 patients treated at one dose level had a tumor response, it was considered that this treatment arm would warrant further study. This study had a 95% probability of stopping early for any given arm if the treatment was ineffective (tumor response rate of ≤5%). With a two-stage design, each treatment arm of this study would have at least 80% power to accept the alternative hypotheses and reject the null hypothesis at a 0.05 α level.

Prostate-specific antigen velocity before and after treatment with 2-methoxyestradiol. Regression analyses were done post hoc to estimate the PSA rate of increase (velocity) for each patient as stabilization of disease was noted on study. There were patients who had PSA declines, although there were no PSA declines meeting Prostate-Specific Antigen Working Group definition of “response.” A regression line was fitted through the PSA measurements obtained on each subject before and after initiation of therapy. A “change point” regression model was employed to test whether the slope of the regression before and after initiation of therapy (possible change point) changed for each subject. A random-effects model was used where the initial PSA level (intercept) and the rate of change, the slope (i.e., PSA velocity), before and after therapy were allowed to vary by subject. The dose (400 or 1,200 mg qd) was also incorporated into the model as was an interaction term between PSA slope and dose level to test whether dose was related to rate of PSA increase. The mathematical equation of the regression model is: where y_ij is the log PSA level of subject i at time point t_jj (measured as weeks from enrollment), t_0i is the week when treatment with 2ME₂ was initiated for this subject, (t_ij − t_0j)⁻ = max[0,(t_ij − t_0j)] counts the number of weeks before treatment initiation, (t_ij − t_0j)⁻ = min[0,(t_ij − t_0j)] counts the number of weeks after treatment initiation, and X_i is the dose level assignment for subject i (1: 1,200 mg, 0: 400 mg).

The following questions were tested through this model: Is the slope before and after treatment different from 0? Slopes that are substantially (i.e., significantly in the statistical sense) higher than 0 indicated PSA increases compared with baseline, whereas negative slopes are associated with PSA decreases. The second question was: Is there a statistically significant treatment effect (measured by the dose-by-week interaction)? A positive interaction effect would imply an increase in PSA velocity in the higher compared with the lower dose, whereas a negative interaction effect would indicate that the PSA velocity is lower in the higher dose compared with the lower dose. Both questions were tested via the t test of whether the regression line slope is equal to 0. The third question was: Are the slopes of PSA increases within each treatment arm different before and after treatment initiation? Appropriate contrast equations were constructed and tested via the F test.

Analysis of duration of therapy with 2-methoxyestradiol. We focused our analyses on the first two cycles of therapy because almost all subjects received at least 8 weeks of treatment. Analyses beyond this point become tenuous as subjects were withdrawn from the study for reasons related to the study outcome in a nonrandom fashion. This has potentially serious and unpredictable effects on the study outcome. For example, if the slope of PSA increase is steeper in the less effective treatment group, subjects may be removed earlier and at lower PSA levels compared with subjects treated with the more effective treatment (which may result in apparently slower PSA increases). Thus, the difference in PSA levels between the two treatments may be attenuated or even reversed in favor of the less effective therapy, inducing serious bias in the analyses.

To assess whether the duration of therapy and time to PSA progression with the low and high doses of 2ME₂ were equal between the two groups, we carried out a Kaplan-Meier estimation of the distribution of therapy durations. We compared these between the two groups using the log-rank test.

All analyses were carried out according to the intent-to-treat principle, so all patient results were analyzed according to initial treatment (dose) group randomization status despite treatment adjustments or interruptions for toxicity reasons. Statistical tests were carried out at the 95% significance level.

Blood samples were collected at baseline and on day 1 of all cycles to measure changes in the levels of PSA, sex hormone-binding globulin (SHBG), dihydroepiandrosterone, dihydroepiandrosterone sulfate, dihydrotestosterone, and total and free testosterone levels. The measurements were done in Clinical Laboratory Improvement Amendments–certified laboratories at Indiana University and the University of Wisconsin Comprehensive Cancer Center. Blood and urine samples were also collected to monitor changes in the levels of the angiogenic proteins VEGF (plasma and urine), bFGF (plasma and urine), and vascular cell adhesion molecule (serum only).

VEGF in EDTA plasma was measured by a commercially available 96-well plate quantitative sandwich immunoassay (Quantikine human VEGF, R&D Systems, Minneapolis, MN) with a standard curve ranging from 31.2 to 500 pg/mL. At the time of assaying, all samples and standards were brought to room temperature and prepared on the plate as recommended by the manufacturer. The plate was read at 450 nm using a Molecular Devices (Sunnyvale, CA) SpectraMax 190 plate reader. The standard curve was constructed by plotting VEGF concentration versus absorbance. A linear scale gave the best fit for this assay. The reported limit of sensitivity is 9.0 pg/mL in plasma.

The assay was validated in the laboratories at the University of Wisconsin Comprehensive Cancer Center, showing a mean r² of 0.999 (range, 0.997-1.00) over 5 months. Within-day variability was assessed with triplicate determinations of each standard curve, with a coefficient of variation (CV) of <10% over the concentration range. Triplicate determinations of VEGF in six subject plasma samples run on a single plate had a mean CV of 7.2% (range, 4.4-11.0%) over the concentration range 36 to 418 pg/mL, and variability did not vary with concentration. Between-day variability was assessed with five-assay standard curve assessments over a 6-month period. The mean CV ranged from 8.2% for the 500 pg/mL standard to 11.5% for the 31.2 pg/mL standard, and the mean CV was <10% over the entire concentration range.

bFGF in EDTA plasma was measured by a commercially available 96-well plate quantitative sandwich immunoassay (Quantikine human VEGF) with a standard curve ranging from 1 to 64 pg/mL. At the time of assay, all samples and standards were brought to room temperature and prepared on the plate as recommended by the manufacturer. The plate was read at 490 nm using a Molecular Devices SpectraMax 190 plate reader. The standard curve was constructed by plotting bFGF concentration versus absorbance. A log scale gave the best fits for this assay. The reported limit of sensitivity is 0.22 pg/mL in plasma. The assay was validated in the laboratories at the University of Wisconsin Comprehensive Cancer Center, showing a mean r² of 0.996 (range, 0.995-0.999) over 5 months. Within-day variability was assessed with triplicate determinations of each standard curve, with a CV of <10% over the concentration range. Triplicate determinations of bFGF in six subject plasma samples run on a single plate had a mean CV of 4.7% (range, 0.5-8.9%) over the concentration range 2 to 60 pg/mL, and variability did not vary with concentration. Between-day variability was assessed with five-assay standard curve assessments over a 6-month period. The mean CV ranged from 34.9% for the 64 pg/mL standard to 47.4% for the 1 pg/mL standard, and the mean CV was <10% over the entire concentration range.

Patients were seen weekly for the first 4 weeks and then on day 1 of each 4-week cycle. The occurrence of grade ≥3 nonhematologic or grade 4 hematologic treatment-related toxicity according to the National Cancer Institute Common Toxicity Criteria version 2.0 that did not resolve within 14 days mandated that the patient be taken off study. If the toxicity resolved in 14 days, then a 50% dose reduction was instituted and the patient was allowed to stay on study.

The pharmacokinetic variables of 2ME₂ in plasma were to be determined from the first eight patients of both cohorts on days 1 and 28 of cycle 1. The goal was to define the plasma concentration-time profile of 2ME₂ and to assess whether the pharmacokinetic profile of 2ME₂ is linear between the two doses. This analysis was also done to establish the dose that provides a potentially effective level of systemic exposure to the drug as indicated by preclinical studies, to determine whether significant drug accumulation occurs during repeated daily dosing, and to determine the intrapatient and interpatient variabilities in the peak and trough plasma concentrations of the drug, both within and between successive weeks of therapy. Blood samples for pharmacokinetic analysis were collected at the following times on day 1 of cycles 1 and 2 (28-day cycles): before dosing, 30, 60, and 90 minutes, and at the 2-, 4-, 8-, 12-, 16-, and 24-hour time point after dosing. Samples were also taken before dosing on days 8, 15, and 22. Pharmacokinetic analyses included plasma area under the curve, elimination half-life (t_1/2), t_max, C_max, C_min, estimated steady-state plasma concentration (C_ss), total clearance (C_l), and volume of distribution (V_d). The concentration versus time curves for 2ME₂ were adjusted for endogenous levels before calculating these variables. Plasma concentrations of 2ME₂ and 2-methoxyestrone were determined using a validated gas chromatographic/mass spectrometric method using trideuterated 2ME₂ and 2-methoxyestrone as internal standards.

The pharmacokinetic variables of both dose levels and changes in the pharmacodynamic associations were to be correlated with responses. However, this was not done because there were no serologic responses.

To ensure accuracy of data collected, an external clinical research organization did 100% source data verification on all subjects enrolled and the data were collected centrally. In addition, communication between sites and the sponsor was facilitated by regular scheduled teleconferences with a discussion of adverse events and study progress.

A total of 33 patients were accrued to this study and were representative of a hormone-refractory prostate cancer population (Table 1). The patients had a median age of 74 years (range, 57-84 years) and a median Karnofsky performance status of 90 (range, 70-100), with a median on study PSA of 122.40 ng/mL (range, 11-2,500 ng/mL). All but one patient had radiographic evidence of metastatic disease. Bone-only disease was evident in 20 patients; measurable disease with or without bone metastases was seen in 12 patients. The median number of hormonal therapy manipulations was 2 (range, 1-4 agents). Patients remained on study for a median of 16 weeks (range, 4-48 weeks).

Response and time to progression. No patients had a serologic response per Prostate-Specific Antigen Working Group criteria, nor were there any measurable disease responses. However, post hoc evidence of activity as defined by alterations in PSA velocity was observed. A total of eight patients had a decline in PSA of >20% (21-44%), seven were observed in the 1,200 mg group, and one in the 400 mg group. On study, 28 of 33 subjects showed a >1.5-fold increase in PSA. Sixteen patients did not have further data on study after the initial increase in PSA was observed as they were removed from study at this time and a confirmatory PSA was not available. The median time to PSA progression (unconfirmed) for the 400 mg group was 67 days, whereas it was 109 days in the 1,200 mg group (P = 0.094; Fig. 1A).

Prostate-specific antigen velocity. The average monthly PSA increase before therapy for the entire group was 0.06 ng/mL/mo. The pre-therapy PSA velocity in the 400 mg group did not differ from the 1,200 mg group (P = 0.12). The study showed that the 1,200 mg but not the 400 mg dose altered the PSA velocity and induced a “broken-arrow phenomenon” (22) as seen in Fig. 2. The average monthly increase in PSA level after therapy for the 400 mg group was 0.047 ng/mL/mo, which was significantly higher than 0 (P = 0.006), whereas the average monthly increase in PSA level for the 1,200 mg dose in the first 2 months was 0.004 ng/mL/mo, which was not statistically different from 0 (P = 0.801), implying that PSA levels in the high-dose group were stable at least in the first two cycles of therapy. There was a statistically significant dose-related (interaction) effect (P = 0.037). In other words, the 1,200 mg dose significantly slowed PSA velocity over the first two cycles of therapy compared with the 400 mg group. In addition, the PSA velocity after therapy was not significantly different compared with PSA velocity before therapy in the low-dose group (P = 0.277) but was significantly different (lower) in the high-dose group (P < 0.001). It should be noted that the difference in PSA velocity between the two groups was lost when analysis was carried out beyond the second treatment cycle.

Patients were kept on protocol therapy until clinical evidence of progression or investigator's or patient's decision and patients with a slowly rising PSA were not taken off as soon as it reached 50% above baseline. The investigators and patients were blinded and no toxicities emerged that could differentiate one group from the other. Post hoc analysis indicated that the median time on therapy was 61 days for the low-dose group and 126 days for the 1,200 mg dose group. This was significantly different by log-rank test (P = 0.024; Fig. 1B). The reasons patients were taken off protocol therapy are listed in Table 2. It is of note that in this double-blind design there was no significant difference in the number of patients taken off therapy due to an adverse event. Furthermore, although patients stayed on therapy longer in the higher-dose group, the number of patients taken off study due to disease progression was similar to that in the lower-dose group.

There were no hematologic toxicities of any grade (i.e., no myelosuppression) and there were no grade 4 nonhematologic toxicities on study. Two patients on the 400 mg dose had a grade 3 elevation of transaminase involving aspartate aminotransferase and/or alanine aminotransferase. In one of the cases, the abnormality resolved within 14 days and did not recur with a 50% dose reduction. One patient on the 1,200 mg dose who had a grade 2 elevation was continued on therapy and resolved without holding therapy. There were no apparent adverse events that suggested major estrogenic exposure as seen with diethylstilbesterol. Three patients experienced grade 1 breast discomfort (one in 400 mg and two in 1,200 mg group) and in only one case was it considered related to therapy. Two thrombotic events were observed in 33 (6%) patients, both were grade 3 and occurred in the 400 mg group. One occurred 8 days after being removed from study due to progressive disease and was assessed as unrelated. The other episode was of unknown relationship to therapy and the patient was taken off study at the same time with progressive disease. One episode of amaurosis fugax, grade 2, which occurred after being on therapy for 1 month with “unknown” relationship to the drug, was observed. The event resolved and no clot was seen and patient continued on therapy. The only cardiac event was one episode of atrial fibrillation considered not related to therapy. There were no episodes of myocardial or cerebrovascular insufficiency.

Ultimately, pharmacokinetic analyses were run for seven patients in the 1,200 mg cohort on day 1 of whom five had complete pharmacokinetic analysis on day 28. In the 400 mg cohort, six and five patients had complete analyses on days 1 and 28 of cycle 1, respectively. The plasma concentration attained by unconjugated 2ME₂ was in the low nanogram per milliliter range and the average levels increased nonproportionally when the dose was augmented from 400 to 1,200 mg/d (Table 3). Once this was noted, it was decided that completion of the analyses in all 16 patients was not beneficial. The metabolite analysis revealed significant oxidation at the 17-position of 2ME₂, resulting in 80% to 95% conversion to 2-methoxyestrone, which is at least 10 times less active than 2ME₂ (data not shown). There is also extensive conjugation, with 80% to 90% conversion to the conjugated forms of 2ME₂ and 2-methoxyestrone. The conjugated forms have no reported antiproliferative activity. A shift in the t_max from 1.2 hours at 400 mg to 6.7 hours with 1,200 mg dose is consistent with dissolution in the gastrointestinal tract being a rate-limiting process. Steady-state trough levels of unconjugated 2ME₂ were observed during the first cycle at approximately days 8 to 15 and remain roughly constant thereafter (data not shown). No correlation was observed between the plasma levels of 2ME₂ or its metabolites and the elevation in transaminases seen in the three patients with mild hepatic toxicity. There was an increase in the C_max, area under the curve, and clearance from days 1 to 28 for both 400 and 1,200 mg groups (all P < 0.05, t test; Table 3).

Steroids. There was no alteration (increase or decrease) of dihydroepiandrosterone, dihydroepiandrosterone sulfate, or dihydrotestosterone levels. However, there was a 2.5-fold increase of SHBG at days 28 and 56 for the 400 mg cohort and a 4-fold increase for the 1,200 mg cohort at days 28 and 56 (Table 4).

At baseline, 32 of 33 patients had a testosterone level <50 ng/dL and the other patient's testosterone was 90 ng/dL. One patient refused ongoing androgen suppression, and after one cycle, his free and total testosterone rose from 1.5 and 19 ng/dL to 6 and 185 ng/dL, respectively. With restarting androgen ablation, the free and total testosterone fell to 1.3 and 19 ng/dL and the patient's disease responded. This patient was in the 400 mg group and was removed from study due to rapid progression, which was ultimately due to androgen-mediated growth. This indicates that 2ME₂ does not suppress testosterone production and therefore differentiates it from diethylstilbesterol.

There was no alteration in levels of free testosterone. However, evidence of assay interference emerged at one site. Specifically, there was an apparent increase in the average total testosterone of 2.3- and 4.6-fold at days 28 and 56, respectively, for the 400 mg dose. The corresponding apparent increases for the 1,200 mg dose were 4.7- and 4.5-fold. Only three patients at this site had an apparent increase of testosterone above 100 ng/dL. At the second site, there was no alteration of total testosterone levels at either dose or either time point. It was concluded that the most likely cause for this observation was assay interference at one of the sites.

Angiogenesis markers. Plasma and urine VEGF and bFGF and serum vascular cell adhesion molecule were analyzed on day 1 of cycle 1 and then on day 28 of all cycles administered. Neither dose influenced vascular cell adhesion molecule, VEGF, or bFGF levels in either plasma or urine.

Three subjects experienced hepatotoxicity and the relationship between hepatotoxicity and angiogenesis factors was evaluated. Two of two individuals with hepatotoxicity and still on study had plasma bFGF levels corrected for platelets as shown in Fig. 3 on D84. This compares with 3 of 18 individuals without hepatotoxicity being “outliers” at the same time point.

This novel dose-finding, double-blind, randomized, placebo-controlled trial has shown that 2ME₂ is a novel well-tolerated agent with only minimal transient hepatic toxicity in a minority of men with hormone-refractory prostate cancer. There were no confirmed serologic or measurable disease responses. The randomized, double-blind, placebo-controlled design provided the opportunity to perform an unbiased post hoc analysis to evaluate whether dose affected PSA velocity and time on therapy. As such, this analysis provides hypothesis-generating data that the 1,200 mg dose had a greater effect on the PSA velocity and time to PSA progression, which was associated with a longer time on therapy. The double-blind design removed investigator bias as the reason for the longer time on therapy and allows the contention that this agent was potentially associated with a clinical benefit. It should be reiterated that patients were kept on study until the physician or patient felt there was no clinical benefit. In some cases, patients came off for rapidly rising PSA (i.e., increase in PSA >50% of baseline) and in others because of symptomatic progression. It is also of note that seven of the eight PSA declines >20% were seen in the 1,200 mg group.

Recently, much attention has been directed at validating PSA changes as a surrogate of drug activity (22). It is recognized that none of these measures have been widely accepted as validated surrogates. Nonetheless, in attempt to quantify whether one or both doses had an effect on the rate of disease progression, we analyzed the PSA velocity before and after therapy. This study showed the 1,200 mg but not the 400 mg dose induced a broken-arrow phenomenon. Specifically, there was a significant decrease in PSA velocity after therapy compared with before therapy (P < 0.001) with the 1,200 mg dose. In contrast, the P indicated no significant change in PSA velocity after therapy versus before therapy with the 400 mg dose (P = 0.277). When analyzed in another manner, it was noted that the average monthly increase in PSA level for the low dose was 0.047 ng/mL/mo, which was significantly higher than 0 (P = 0.006). As such, the 400 mg dose did not reduce the PSA velocity. In contrast, the average monthly increase in PSA level for the high dose in the first 2 months was 0.004 ng/mL/mo. This was not statistically different from 0 (P = 0.801) and indicated that the drug decreased the PSA velocity to almost 0 and delayed disease progression as defined by PSA criterion. In sum, these analyses indicate the high dose significantly slowed the PSA velocity and showed a significant dose-related effect (P = 0.037). It is of note that other investigators have employed a similar design. The most notable example was a study of rosiglitazone in men with a rising PSA level after radical prostatectomy and/or radiation therapy. This study employed a randomized, placebo-controlled design and reported that the drug did not prolong the PSA doubling time or time to disease progression. The authors also noted an unexpected discordance between baseline and post-treatment PSA doubling times in the placebo group in this hormone-naive population. Although our population was hormone refractory, the data from this article reinforce the importance of randomized, blinded, controlled trials when looking at rates of increase of PSA and time to progression (23).

To add significance to our PSA velocity findings, we observed that the decrease in PSA velocity with the 1,200 mg dose was associated with a longer time to PSA progression and the patients in the 1,200 mg group received protocol therapy twice as long. It should also be recalled this was a double-blind, placebo-controlled study and thus removed investigator bias as a confounding variable. It is recognized that these data are hypothesis generating; however, on the whole, they support the contention that 1,200 mg/d 2ME₂ capsules are biologically active and have some evidence of clinical activity with minimal to no toxicity. We also acknowledge that the PSA levels in the lower-dose group were numerically but not significantly higher and that this may be a confounding factor.

However, improvements in drug formulation of 2ME₂ are required. Specifically, the oral bioavailability was apparently dissolution rate limited and resulted in the nonlinearity of the pharmacokinetics. Moreover, the rapid conversion to 2-methoxyestrone and 85% conjugation resulted in low plasma levels. Reformulation of the drug has improved the bioavailability and has shown increased activity in relevant pharmacokinetic and efficacy models. These new formulations are being evaluated at the University of Wisconsin Comprehensive Cancer Center and Indiana University in two separate phase I trials.

In spite of pharmacokinetic and bioavailability limitations, it is interesting to note that the 1,200 mg dose resulted in more evidence of anticancer activity (i.e., the post hoc evaluation showing an alteration in PSA velocity) and a greater increase in SHBG. The lack of correlation between the pharmacokinetic and pharmacodynamic data may be due to accumulation of the lipophilic agent in tissues as has been seen with nicotine (24, 25). Moreover, the lack of suppression of testosterone in the patient who refused ongoing androgen ablation and the toxicity profile differentiates this agent from diethylstilbesterol. The increase in the SHBG is intriguing. This is a plasma glycoprotein that binds to certain steroids. SHBG, in turn, binds to a specific receptor on cell membranes. The role of SHBG in steroid-induced signal transduction is poorly understood. However, there is evidence that SHBG may mediate the activity of certain agents that induce tumor regressions in patients with prostate cancer. Diethylstilbestrol diphosphate, an estrogen prodrug, is one such agent (26). It is also possible that 2ME₂ may exert some of its effects through SHBG in addition to its effects on death receptor 5 (13). There is accumulating evidence that SHBG may have anticancer properties by impairing hormone-induced signal transduction or down-regulation of the level of sex hormones (27–29).

The exploratory analysis of angiogenesis factors (VEGF, vascular cell adhesion molecule, and bFGF) revealed that there was no change with drug administration. This is not surprising as 2ME₂ is not expected to directly influence the transcription of either of these genes (14), and changes in bFGF and VEGF are more likely to be markers of disease regression or progression. As clinical activity of 2ME₂ in this study was modest, and the majority of patients experienced stable disease, stable levels of angiogenesis factors were expected and observed. However, the intriguing observation that plasma bFGF was possibly associated with hepatotoxicity is worthy of further exploration. These results are consistent with considerable in vitro evidence, suggesting a role for both bFGF and VEGF in liver repair and regeneration. In vitro evaluations have shown that bFGF is generated in response to liver injury due to wounds (30), radiation (31), and chemicals (32) and that administration of bFGF can enhance liver regeneration and repair (33). Based on our review of the literature, these data are the first to show a possible relationship in vivo between drug-induced liver injury and bFGF. This tentative finding supports further study of bFGF as an early marker of drug-induced hepatotoxicity.

In conclusion, 2ME₂ is well tolerated with evidence of anticancer activity at the 1,200 mg/d dose. It should be reiterated the evidence of anticancer activity was a dose-response effect on PSA velocity. Reformulation to overcome the low bioavailability holds the promise of producing a more active drug. Moreover, the novel trial design permitted an unbiased evaluation of dose effect and is one that may be required for drugs that stabilize disease rather than induce remissions.