Fax: (617) 726-4899
Rosiglitazone versus placebo for men with prostate carcinoma and a rising serum prostate-specific antigen level after radical prostatectomy and/or radiation therapy†
Article first published online: 13 AUG 2004
Copyright © 2004 American Cancer Society
Volume 101, Issue 7, pages 1569–1574, 1 October 2004
How to Cite
Smith, M. R., Manola, J., Kaufman, D. S., George, D., Oh, W. K., Mueller, E., Slovin, S., Spiegelman, B., Small, E. and Kantoff, P. W. (2004), Rosiglitazone versus placebo for men with prostate carcinoma and a rising serum prostate-specific antigen level after radical prostatectomy and/or radiation therapy. Cancer, 101: 1569–1574. doi: 10.1002/cncr.20493
Presented in part as Abstract 1588 at the 39th annual meeting of the American Society of Clinical Oncology, Chicago, Illinois, May 31–June 3, 2003.
- Issue published online: 17 SEP 2004
- Article first published online: 13 AUG 2004
- Manuscript Accepted: 10 JUN 2004
- Manuscript Revised: 8 JUN 2004
- Manuscript Received: 15 APR 2004
- National Institutes of Health. Grant Number: 5 P50 CA90381-02
- Prostate Cancer Foundation
- SmithKline Beecham Pharmaceuticals
- prostate carcinoma;
- prostate-specific antigen;
- peroxisome proliferator activated receptor γ;
The objective of this study was to assess the biologic activity of rosiglitazone, a peroxisome proliferator-activated receptor γ agonist that has been approved to treat type 2 diabetes, in men with recurrent prostate carcinoma using change in prostate specific antigen (PSA) doubling time (PSADT) as the primary outcome variable.
Men with histologically confirmed prostate carcinoma, no recent hormone therapy, a rising serum PSA level after radical prostatectomy and/or radiation therapy, and no radiographic evidence of metastases were assigned randomly to receive either oral rosiglitazone (4 mg twice daily) or placebo. The treatment was continued until the men developed disease progression or adverse effects. A positive outcome was defined as a posttreatment PSADT > 150% the baseline PSADT and no new metastases.
One hundred six men were enrolled. The median treatment duration was 315 days for men in the placebo group and 338 days for men in the rosiglitazone group (P = 0.28). Forty percent of men in the in the placebo group and 38% of men in the rosiglitazone group had a posttreatment PSADT > 150% of the baseline PSADT and no new metastases (P = 1.00). In exploratory analyses, the rate of a positive outcome remained higher than expected in the placebo group, even when a positive outcome was redefined using more stringent criteria. The time to disease progression was similar between the groups.
Rosiglitazone did not increase PSADT or prolong the time to disease progression more than placebo in men with a rising PSA level after radical prostatectomy and/or radiation therapy. The unexpected discordance between baseline and posttreatment PSADT in the placebo group reinforced the importance of randomized controlled trials in this setting. Cancer 2004. © 2004 American Cancer Society.
Prostate carcinoma is the most common solid tumor in men worldwide. In 2004, there will be approximately 230,110 new cases of prostate carcinoma and 29,900 prostate carcinoma deaths in the United States.1
Approximately 80% of newly diagnosed prostate carcinoma is localized clinically, and most men with these early-stage prostate carcinomas undergo radical prostatectomy or receive radiation therapy. Approximately 33–50% of patients experience disease recurrence after surgery or radiation therapy.
A rising serum prostate specific antigen (PSA) level after radical prostatectomy or radiation therapy provides an early indication of recurrent disease. A serial rise in the serum PSA level after surgery or radiation therapy, however, typically predates clinically or radiographically detectable metastatic disease by many years without additional treatment. In a retrospective series of men with a rising PSA levels after radical prostatectomy, for example, the median time to metastasis was 8 years from the time of PSA elevation.2 Three years after initial postoperative PSA elevation, only 27% of patients had radiographically detectable metastases; after 7 years, only 48% of patients had metastatic disease.
Increases in serum PSA levels after radical prostatectomy or radiation therapy follow an exponential growth curve, and the relation between log PSA and time is linear.2–4 Log slope PSA, PSA doubling time (PSADT), Gleason grade, and the interval between local treatment and biochemical progression predict the probability and time to the development of distant metastatic disease and survival. The PSADT and the equivalent log slope PSA are better predictors of the probability and time to clinical disease recurrence than preoperative PSA, interval between local treatment and rising PSA, Gleason grade, or pathologic disease stage.2, 3 In men who had a rising PSA level after undergoing radical prostatectomy, a PSADT shorter than the median value of 10 months was the most significant predictor for progression to metastatic disease.2 Similarly, a short posttreatment PSADT after radiation therapy predicted progression to metastatic disease.3, 4 The PSA velocity was related significantly to survival in men with rising PSA levels after external beam radiation therapy.5
Men with an isolated PSA recurrence after local treatment may represent an ideal population for evaluation of novel therapies based on minimal disease state, indolent natural history, and a preference to avoid the adverse effects of androgen-deprivation therapy. The evaluation of novel agents in this setting, however, is hindered by the lack of convenient, validated endpoints. Survival or time to disease progression are impractical because of the long interval between the initial PSA increase and the development of metastases. In addition, the PSA response criteria commonly used in androgen-independent prostate carcinoma (> 50% posttreatment decline in PSA6) may overlook biologically and potentially clinically meaningful activity, particularly with cytostatic agents. Changes in PSADT may be more sensitive for detecting biologic activity than traditional PSA response criteria, although PSADT has not been evaluated adequately as an endpoint.
Peroxisome proliferator-activated receptor γ (PPARγ) is a member of the nuclear receptor superfamily of ligand-activated transcription factors.7 PPARγ is expressed at high levels in adipose tissue and plays a central role in adipocyte differentiation. The receptor also is expressed in other tissues, including intestine, liver, kidney, breast, and prostate. Several lines of evidence suggest that PPARγ contributes to the pathogenesis of human prostate carcinoma. First, PPARγ is expressed in the normal human prostate, in primary prostate carcinomas, and in human prostate carcinoma cell lines.8–10 Second, heterozygous deletion of the PPARγ gene is common in human primary prostate carcinomas.8 Third, PPARγ ligands, including 15d-PGJ2, troglitazone, and rosiglitazone, inhibit the proliferation of human prostate carcinoma cells in vitro.8–10 Fourth, troglitazone decreased the growth of human prostate carcinoma cells in immunodeficient mice.9 Fifth, troglitazone was associated with stable disease (but with only 1 PSA increase > 50%) in 12 men with androgen-dependent prostate carcinoma who were enrolled in a single-arm, Phase II study.8
Rosiglitazone is a synthetic PPARγ agonist that has been approved for use in type II diabetes mellitus. In this study, we prospectively compared the effects of rosiglitazone and placebo in men with a rising PSA levels after radical prostatectomy and/or radiation therapy. We evaluated changes in PSADT as a screen for biologic and clinical activity.
MATERIALS AND METHODS
Study participants were recruited at Dana Farber Cancer Institute, Massachusetts General Hospital, University of California–San Francisco, and Memorial Sloan-Kettering Cancer Center between December, 2000 and September, 2002. Participants had histologically confirmed adenocarcinoma of the prostate, biochemical disease progression after radical prostatectomy and/or radiation therapy (external beam radiation therapy and/or brachytherapy), and no radiographic evidence of metastases. Biochemical disease progression was defined as 3 rises in PSA level with each PSA determination at least 4 weeks apart and each PSA level > 0.2 ng/mL. Men who underwent radical prostatectomy had baseline PSA levels ≥ 2 ng/mL. Men who either received primary radiation therapy or underwent radical prostatectomy followed by radiation therapy had baseline PSA levels ≥ 2 ng/mL and > 150% postradiation nadir. Men with baseline PSADT > 24 months were excluded. One patient was enrolled with a baseline PSADT of 22 months based on all PSA levels that were available at registration. When additional PSA levels were submitted after randomization, his baseline PSADT was recalculated at 47 months. He was included in all analyses. Men who had received prior neoadjuvant or adjuvant hormone therapy were included if the interval between the completion of hormone therapy and study entry was > 1 year. Men who had a history of treatment with chemotherapy for prostate carcinoma or prior hormone therapy for recurrent prostate carcinoma were excluded. Men with a Cancer and Leukemia Group B performance status > 2, New York Heart Association Class 3 or 4 cardiac function, current treatment with insulin or an oral hypoglycemic agent, fasting blood glucose < 60 mg/dL, or serum alanine aminotransferase (ALT) > 1.5 times the upper limit of normal also were excluded.
The study was a prospective, randomized, placebo-controlled clinical trial. At a baseline visit, patients underwent a physical examination, radionuclide bone scan, complete blood count, and determination of serum ALT, serum PSA, and fasting blood glucose levels. Eligible patients were assigned randomly to treatment with rosiglitazone (Avandia®; SmithKline Beecham Pharmaceuticals, Collegeville, PA) (4 mg orally twice daily) or placebo. Patients and study personnel were blinded to treatment assignments. Other therapies for prostate carcinoma were not permitted during the study.
Serum PSA levels were measured every 4 weeks. To reflect clinical practice, PSA testing was performed in real time at local clinical laboratories. Bone scans were repeated every 12 months. Study treatment continued until disease progression, expected Grade ≥ 3 toxicity (according to the National Cancer Institiute's Common Toxicity Criteria), or any unexpected treatment-related toxicity ≥ Grade 3.
The Institutional Review Boards at each of the participating institutions approved the study. All patients provided written informed consent. The study sponsors played no role in the study design; in the collection, analysis, or interpretation of data; or in the writing of this report.
PSADT was calculated by natural log of 2 (0.693) divided by the slope of the relation between the natural log of PSA and time.2 Slope was calculated by linear regression. The baseline (pretreatment) PSADT was calculated using the three PSA increases required for study entry, all other PSA values that were obtained during the interval between the first and third PSA increase, and the PSA value immediately preceding the first PSA increase. The posttreatment PSADT was calculated using PSA measurements that were obtained at baseline and monthly for the first 6 months of treatment. For patients who discontinued treatment prior to 6 months, all available PSA measurements prior to discontinuation of treatment were used to calculate PSADT. Patients who either completed ≥ 3 months of treatment or developed progressive disease at any time after the start of treatment, also were considered evaluable.
A positive PSADT outcome was defined as either a posttreatment PSADT > 150% of the baseline PSADT or a negative posttreatment PSADT (i.e., declining PSA) and no new metastases. Disease progression was defined as metastatic disease, new disease-related symptoms, or PSA ≥ 200% of the baseline value. The time to disease progression was defined as the interval between randomization and disease progression. Patients who died from any cause were censored at the time of their last known PSA level < 200% of baseline. Patients who discontinued treatment prior to disease progression were censored at their last PSA measurement.
The primary study objective was to compare the proportion of men with positive PSADT outcomes between the groups. The study was designed to distinguish between a 30% positive outcome rate among evaluable, rosiglitazone-treated men and a 5% positive outcome rate among evaluable, placebo-treated men. The design provided 86% power using a two-sided Fisher exact test with a Type I error of 5%. All eligible patients were included in the analyses.
The secondary study objective was to compare the time to disease progression between the groups. The design provided 83% power to detect a doubling in the median time to disease progression, from 10 months to 20 months, assuming 24 months of accrual and an additional 24 months of follow-up and using a two-sided log-rank test with a Type I error of 5%.
A Fisher exact test was used to compare differences in baseline characteristics between the groups. T tests were used to compare PSA slopes and PSADT between groups. T tests also were used to compare posttreatment changes in PSA slope between the groups. The log-rank test was used to test for differences in the time to disease progression. The Mehta exact test for ordered categorical data11 was used to compare differences in toxicity between the groups.
One hundred six men were assigned randomly to receive either rosiglitazone or placebo. One man who was assigned to the rosiglitazone group was randomized in error; he did not receive study medication and was excluded from the analyses. All of the remaining 105 men are included in the analyses, including 1 man who had a baseline PSADT > 24 months and 2 men who elected not to start study treatment. Baseline characteristics, including age, type of prior treatment, Gleason score, and PSA, were similar between the groups (Table 1). The mean ± standard deviation baseline PSADT was 10.3 ± 7.8 months in the placebo group and 8.9 ± 5.2 months in the rosiglitazone group (P = 0.28).
|Characteristic||Rosiglitazone group (n = 52)||Placebo group (n = 53)|
|Mean ± SD||69 ± 9||68 ± 8|
|Gleason score (%)|
|Prior prostatectomy (%)||52||62|
|Prior radiation therapy (%)||75||83|
|Prior hormone therapy (%)||31||33|
|Mean ± SD||10.4 ± 11.2||9.2 ± 10.4|
|PSA slope (ng/mL per month)||0.11 ± 0.06||0.11 ± 0.09|
|Mean ± SD||8.9 ± 5.2||10.3 ± 7.8|
The median treatment duration was 315 days for men in the placebo group and 338 days for men in the rosiglitazone group (P = 0.28). Disease progression (new metastatic disease or a PSA level > 200% of the baseline level) was the most common reason for discontinuation of treatment in both groups. Figure 1 depicts the correlations between baseline PSADT and posttreatment PSADT for each group. Twenty-one of 53 men (40%) in the in the placebo group and 20 of 52 men (38%) in the rosiglitazone group had posttreatment PSADT > 150% of the baseline PSADT and no new metastases (P = 1.00) (Table 2).
|Positive PSADT outcomea||Placebo group||Rosiglitazone group||P value|
|No. (%)||No. (%)|
|Yes||20 (38)||21 (40)||1.00|
|No||30 (58)||31 (58)||—|
|Not evaluable||2 (4)||1 (2)||—|
|Total||52 (—)||53 (—)||—|
Because the rate of a positive PSADT outcome was markedly higher in the placebo group compared with the expected rate of 5%, we performed exploratory analyses using other changes in posttreatment doubling times to define a positive PSADT outcome. The rate of positive PSADT outcome remained higher than expected in the placebo group, even when a positive outcome was defined as a posttreatment PSADT > 200% or > 400% of the baseline PSADT (Table 3). In these exploratory analyses, the proportion of men with a positive PSADT outcome did not differ significantly between groups, regardless of the change in PSADT used to define a positive outcome.
|Posttreatment PSADT||Placebo group||Rosiglitazone group||P valuea|
|No. (%)||No. (%)|
|> 100% baseline PSADT||38 (73)||29 (55)||0.07|
|> 150% baseline PSADT||20 (38)||21 (40)||1.00|
|> 200% baseline PSADT||16 (31)||17 (32)||1.00|
|> 400% baseline PSADT||12 (23)||13 (25)||0.52|
Changes in PSA slope did not differ significantly between the groups (P = 0.38). It is noteworthy, however, that the mean PSA slope decreased by 5.4% in the placebo group (P = 0.16) and decreased by 33.6% in the rosiglitazone group (P = 0.002).
Time to Disease Progression and Survival
Disease progression was defined as metastatic disease, new disease-related symptoms, or a PSA level ≥ 200% of the baseline level. The median time to disease progression was 340 days for the placebo group and 377 days for the rosiglitazone group (P = 0.76) (Fig. 2). PSA ≥ 200% of the baseline value was the most common reason for declaring disease progression in the placebo group (62%) and in the rosiglitazone group (85%).
One man in the placebo group died from complications of a myocardial infarction. There were no other deaths in either group.
Rosiglitazone was tolerated well. There were no serious adverse events related to treatment in either group. Four men in the rosiglitazone group and three men in the placebo group withdrew from the study because of adverse effects. Consistent with the reported adverse effects of rosiglitazone in patients with type 2 diabetes,12 anemia and edema were more common in the rosiglitazone group compared with the placebo group.
Preclinical evidence suggests that PPARγ activation contributes to the pathogenesis of human prostate carcinoma. In this double-blind, randomized, controlled trial, however, rosiglitazone, a potent PPARγ agonist, did not increase PSADT or prolong the time to disease progression compared with placebo in men who had rising PSA levels after undergoing radical prostatectomy and/or receiving radiation therapy.
To the best of our knowledge, this is the first prospective study to use changes in PSA kinetics as the primary study outcome. Our results may have important implications about the use of PSA kinetics as a study endpoint. Thirty-eight percent of men in the placebo group had favorable outcomes, defined as posttreatment PSADT > 150% of the baseline PSADT. The higher than expected rate of positive PSADT outcomes persisted even when more stringent criteria (i.e., posttreatment PSADT > 200% or > 400% of the baseline PSADT) were used to define a positive outcome. The high rate of positive PSADT outcomes may reflect the limited precision of repeat PSADT assessments. Alternatively, the higher than expected rate of positive PSADT outcomes may be related to the placebo effect.13 The mean posttreatment PSA slope did not change significantly from baseline in the placebo group, suggesting that the higher than expected rate of positive PSADT outcomes was not due to temporal changes in tumor growth rate.
In a single-arm, Phase II study, granulocyte-macrophage colony-stimulating factor decreased PSA levels by > 50% in 3 of 29 men who had rising PSA levels after undergoing radical prostatectomy or receiving radiation therapy.14 Changes in PSADT were evaluated in exploratory analyses. The median PSADT increased from 8.4 months to 15.0 months, and 11 of 26 men had posttreatment PSADT > 150% of the baseline PSADT. In another single-arm, Phase II study, high-dose, weekly, oral calcitriol achieved no PSA decreases > 50% in 22 men who had rising PSA levels after undergoing radical prostatectomy or receiving radiation therapy, although 6 of 22 men had prolonged posttreatment PSADT.15 The similarity of these results with the PSADT changes in our placebo group raises the possibility that the observed changes reflect, in part, the variation in serial PSADT measurements rather than the biologic activity of investigational treatment. These observations also reinforce the need for a randomized, placebo-controlled study design when using changes in PSADT as a study outcome.
This study had limitations. Baseline PSADT was calculated using PSA values that were determined at irregular intervals, whereas posttreatment PSADT was calculated using monthly PSA values. For most patients, baseline PSADT was calculating using fewer PSA values than the posttreatment PSADT. Differences in the intervals between PSA measurements and the number of PSA values may have contributed to the variability between baseline and posttreatment PSADT in the placebo group. Additional studies using a mandated observation period prior to intervention or run-in period are needed to determine whether standardization of the intervals between PSA measurements and the number of PSA values will improve the precision of PSADT as a study outcome.
The current results do not diminish the potential value of changes in PSADT as an outcome variable for the early evaluation of novel therapeutic agents. In randomized studies of similar design, more active agents may demonstrate the value of PSA kinetics as a screen for biologic activity. Ultimately, larger studies to assess the relation between PSA kinetics and clinical outcomes will be required to evaluate the utility of posttreatment changes in PSADT in the clinical development of new drugs for the treatment of patients with prostate carcinoma.
PPARγ has been implicated in the pathogenesis of other human malignancies, including liposarcoma, colorectal carcinoma, gastric carcinoma, thyroid carcinoma, and breast carcinoma.16 Additional studies are needed to assess the efficacy of PPARγ agonists in other malignancies and to evaluate the safety and efficacy of PPARγ agonists in combination with other agents.
In summary, rosiglitazone did not increase PSADT or prolong the time to disease progression compared with placebo in men who had rising PSA levels after undergoing radical prostatectomy and/or receiving radiation therapy. The discordance between baseline and posttreatment PSADT in our placebo group suggests that caution is required when using changes in PSADT as an outcome in uncontrolled trials and reinforces the value of randomized, placebo-controlled trials in this setting.