TREATMENT WITH FINASTERIDE FOLLOWING RADICAL PROSTATECTOMY FOR PROSTATE CANCER G. ANDRIOLE, M.D. M. LIEBER, M.D. J. SMITH, M.D. M. SOLOWAY, M.D. E SCHROEDER, M.D. D. KADMON, M.D. J. DEKERNION, M.D.
J. RAJFER, M.D. R. BOAKE, M.D. D. CRAWFORD, M.D. E. RAMSEY, M.D. J. PERREAULT, M.D. J. TRACHTENBERG, M.D.
From Merck Research Laboratories,
Y. FRADET, M.D. N. BLOCK, M.D. R. MIDDLETON, M.D. J. NG D. FERGUSON G. GORMLEY, M.D
Rahway, New Jersey*
The objective of this study was to evaluate the effect of finasteride (10 mg/d) or placebo on serum prostate-specific antigen (PSA) and recurrence rates in men with detectable PSA levels after radical prostatectomy. Methods, A total of 120 men, 48 to 89 years old, previously treated with radical prostatectomy for prostate cancer within the past 10 years, with serum PSA levels between 0.6 and 10.0 ng/mL, with no evidence of skeletal metastasis on bone scan, and with no previous androgen deprivation therapy, were treated with 10 mg finasteride or placebo in a double-blind fashion for 12 months. After the first year, all patients were treated with finasteride for an additional 12 months. Primary endpoints were serum PSA levels and recurrence rates defined as positive bone scan or positive biopsy. Results. Patients treated with finasteride had a delayed increase in serum PSA compared with placebo of approximately 9 months in the first year and 14 months by the end of the second year. Patients with baseline PSA levels lessthen 1.O ng/mL had no significant increase in serum PSA during the 2 years of treatment. Fewer recurrences were observed in the finasteride group, but these differences were not statistically significant. Finasteride was well tolerated, and side effects were balanced between treatment groups. Conclusions. The results of this study indicate that treatment with finasteride delays but does not prevent the rise in serum PSA observed in untreated patients with detectable PSA levels after radical prostatectomy. The reduction in local and distant recurrences in the finasteride group suggests that the effect on PSA reflects a direct effect on tumor growth without affecting the initial responseto subsequent hormonal therapy. These data require confirmation by studies that are longer and larger, focused on demonstrating significant differences in progression rates and survival before the use of finasteride can be considered as an option for men with detectable PSA levels after radical prostatectomy. ABSTRACT-Objectives.
*Authors’ ajjliations are: G.A., Washington University School of Medicine, St. Louis; M.L., Mayo Clinic, Rochester; J.S., Vanderbilt University, Nashville; MS., University of Miami, Miami; F.S., Erasmus University, Rotterdam; D.K., Baylor College of Medicine, Houston; J.D., UCLA School of Medicine, Los Angeles; J.R., Harbor-UCLA Medical Center; R.B., University of Alberta, Alberta; D.C., University of Colorado, Denver; E.R., University of Manitoba, Manitoba; J.P. Hopital Saint-Luc, Quebec; J.T., Toronto General Hospital, Ontario; Y.F., Lirval University, Quebec; N.B., University of Miami, Miami; R.M., University of Utah Medical Center, Salt Lake City; J.N., D.F., and G.G., Merck Research Laboratories, Rahway. Submitted (Rapid Communication): September 29, 1994, accepted (with revisions): November 8, 1994
45, NUMBER 3
Radical prostatectomy is a common treatment for prostate cancer, with more than 100,bOO procedures conducted worldwide each year. If at surgery the cancer is confined to the prostate gland, the procedure offers the potential to be curative.‘,2 Following radical prostatectomy, serum prostate-specific antigen (PSA) levels decrease to low levels and are often undetectable. Serum PSA levels after radical prostatectomy significantly correlate with prognosis. 3-6 Patients who persistently have undetectable levels of PSA after surgery have a good prognosis, whereas patients with detectable 491
PSA levels that are rising over time often develop evidence of clinical progression. Unfortunately, because the length of time between elevations in PSA levels and objective signs of progression is highly variable, the decision of when to treat with conventional androgen deprivation or radiation therapy is difficult, because each option has limiting side effects. Although total androgen ablation is an effective treatment for advanced disease, its use as adjuvant therapy prior to objective relapse has been limited given the physical and psychological effects of castration in a setting where progression may not occur for many months or years, if at all. Radiation therapy is limited to the treatment of residual disease in the area around the excised prostate, but often fails to prevent progression due to microscopic extension beyond the field of radiation.6,7 An alternative approach for the patient with microscopic residual disease, detected by an elevated PSA level in the absence of clinical progression, is treatment with a S-a-reductase inhibitor such as finasteride. Finasteride inhibits the conversion of testosterone to the more potent androgen dihydrotestosterone (DHT) by the type 2 isoenzyme of 5a-reductase found in the prostate. By lowering DHT but not testosterone (T), finasteride has been shown to produce a suppressive effect on the growth of the prostate in men with benign prostatic hyperplasia (BPH) without producing significant side effects.sT9 Several studies have demonstrated a suppressive effect of 5a-reductase inhibition on some but not all prostate cancer cell lines.lOJr However, the relevance of these studies to the in vivo biology of prostate cancer in man remains to be established. More relevant are the results of a pilot study in men with Stage D prostate cancer treated with finasteride or placebo for 12 weeks.12 In this study a statistically significant reduction in serum PSA occurred within 6 weeks, which was different from the placebotreated group (P ~0.05). No significant side effects were observed and disease progression was not accelerated. The data with finasteride therefore suggest that treatment of men with low volume disease after radical prostatectomy may prevent or significantly delay the PSA level from rising due to a suppressive effect on tumor growth, thus preventing or delaying clinical relapse while maintaining optimal quality of life until such time as total androgen ablation may be required. Important to the success of this approach is the demonstration that the subsequent effectiveness of conventional hormonal therapies is not compromised by this treatment strategy 492
PATIENTS A total of 120 men between 48 and 89 years of age, previously treated with radical prostatectomy for prostate cancer within the past 10 years, with serum PSA levels between 0.6 and 10.0 ng/mL, with no evidence of skeletal metastasis on bone scan, and with no previous androgen deprivation therapy, were enrolled in this multicenter study The time between surgery and treatment with finasteride varied, as did the time between postsurgical detection of PSA and treatment, and the men participating in this trial were very heterogeneous. Radiation therapy at least 6 months prior to entry or evidence of lymph node invasion at the time of surgery did not exclude the patient. Patients with a mass palpable on rectal examination were required to have a negative biopsy prior to entry. Patients with significant concurrent illnesses, treatment with other investigational drugs, or PSA levels continuing to decline from previous surgery or radiation therapy were excluded. All clinical centers obtained institutional review board approval, and all patients gave written informed consent. Each patient who met all of the inclusion criteria and none of the exclusion criteria was randomized in a double-blind fashion to receive 10 mg of finasteride or matching placebo once daily for 12 months. The lo-mg dose was selected based on the PSA response in a pilot study observed in men with Stage D disease.12 After 12 months of treatment, all patients were offered participation in an unblinded 12-month extension and received 10 mg of finasteride if they consented. Patients were discontinued from the study if their bone scan became positive for metastasis or if they developed other signs of clinical progression. As a precaution, during the first 12 months of treatment while PSA levels were blinded, patients were to be discontinued from the study if serum PSA levels increased by 10 ng/mL above the baseline value obtained on entry into the study regardless of whether objective progression could be documented. During the second year of the study (months 13 to 24), while PSA levels were not blinded, patients were to be discontinued if PSA levels increased to twice the month 12 value. During the study, patients were evaluated with frequent PSA levels, rectal evaluations, periodic bone scans, electrocardiograms, chest radiograph, physical examination, and serum chemistry measurements. LABORATORYEVALUATIONS
Blood samples for serum PSA measurements were obtained in the morning and prior to digital
U [email protected]
45, NLIMBEB 3
of the study
Finasteride (10 md Placebo Number randomized (n) 54 66 Mean age (yr) 67.8 67.0 Time since surgery (yr) 2.8 + 2.5 2.9 f 2.7 Number of men with Gleason score 2 to 7 40 45 8to 10 10 12 Pathologic stage B 25 22 L 23 36 7 6 D, Baseline PSA 0.6 to 1 .O 10 13 1.1 to 3.9 30 36 4.0 to 10.0 14 17 DHT (ngidl) * 36.5 f 15.8 35.5 * 19.5 Testosterone (ng/dL) * 426 & 165 423 f 137 LH (U/L]* 15.3 * 7.4 14.5 + 6.0 FSH (U/L]* 8.7 + 8.7 8.6 + 6.5 PSA (ng/mL) * 2.2 + 2.9 2.1 f 2.6 *The normal rangefor theseparametersis O.S jollows: dihydrotestosterone (DHT; 25 to 75 @IL), testcxterone(280 to1100 mIU/mLJ,luteinizing hormone(LH; 2 to 20 mILJ/mL), jollicle-stimulating hormone(FSH; 2 to 20 mILJ/mL), prostatespecificantigen (PSA;the normal rangeof 0 to 4 ng/mLis notapplicableto men following prostatectomy).
rectal examination. Values were obtained using the two-site immunometric assay produced by Hybritech (Tandem-R). To minimize variability, the baseline PSA level for each patient was derived from the average of three values obtained at 7-day intervals prior to starting the blinded medication. Each subsequent value was compared to the baseline value to calculate the change from baseline. The Hybritech assay used has a sensitivity of 0.2 ng/mL and an interassay variation of 3.8%. Serum testosterone, luteinizing hormone (LH), DHT, and follicle-stimulating hormone (FSH) levels were measured at Nichols Laboratory in California. All radiographic evaluations were conducted and interpreted at the individual sites. Bone scan evidence of metastasis was confirmed by skeletal radiographs. STA~STICALANALYSIS All laboratory measurements were evaluated using an analysis of variance (ANOVA) approach based on the rank-transformed values of change from baseline and an all-patients-treated approach. Of the 120 patients entered into the study, 71 had data through at least month 18 and were therefore included in the analysis of 2-year data. If a patient had missing data, the last available value was used in the analysis. Assumptions of the ANOVA were tested for normality by the Shapiro-Wilk test and for homogeneity of variance by the Levene test.13J4 Within group tests were made at each visit using
/MARCH 1995 I VOLUME 45, NUMBER3
the Wilcoxon signed rank test. The proportion of patients with clinical progression or PSA level increased more than 10 ng/mL above the baseline value was compared using Fisher’s exact test. Subgroup analysis was undertaken using nonparametric analysis of variance to compare treatment effects on PSA based on: Gleason score (2 to 7 versus 8 to lo), baseline PSA (0.6 to 1.0 ng/mL versus 1.1 to 3.9 ng/mL versus 4 to 10 ng/mL), and stage of cancer (B, C, or D) at the time of radical prostatectomy. Summary statistics are presented as median f SE unless otherwise noted.15 RESULTS The baseline characteristics of the 120 men randomized in this study are outlined in Table I. There were no significant differences between treatment groups among the baseline characteristics. Blinded randomization resulted in a slight imbalance in the number of men assigned to finasteride (n = 54) versus placebo (n = 66). Of the 120 men who entered the study, 10 men in the finasteride group and 16 in the placebo group did not complete the study. In addition, 10 men who completed the study elected not to enter the open extension. Table II summarizes the number of men who completed the first year of the study and entered the open extension. After 2 months of treatment, serum DHT levels were decreased by 62 -c 3% from baseline in the finasteride treatment group (P ~0.001 versus placebo) and remained suppressed throughout the study. During the 2 years on finasteride, serum testosterone levels increased 20%, serum LH increased 25%, and FSH increased by about 15%. During the first year of treatment, PSA levels in the placebo group slowly and progressively increased from a baseline median of 2.01 to 3.8 ng/mL at month 12 (Fig. 1A). In the finasteride group, PSA levels were maintained at or below baseline for the first 6 months and then increased above baseline at a rate slower than the placebo group during the second 6 months. The pattern of PSA change in the subset of patients participating in the extension (months 13 to 24) was similar to the first year data except for the clear but transient decrease in PSA, which occurred when the placebo patients were switched to finasteride at month 12 (Fig. 1B). Extrapolation of the PSA value at month 24 in the group continuously treated with finasteride onto the PSA curve of the placebo-treated patients indicates that after 24 months of treatment with finasteride PSA levels are similar to the untreated group after about 8 months, suggesting a 14-month delay in 493
Patients entered Patients discontinued due to recurrence Patients discontinued without recurrence to PSA elevations Patients discontinued due to adverse experience or noncompliance Patients included in PSA analysis *Only
extension. +Two addknud
with prostate-specific patients
Months 0 to 12 (Double Blind) 120 9t 9
Months 13 to 24 (Open Extension) 84 13 6
(PSA) data through
at least month
74’ 18 were analyzed
as part uf the
but wwe not discontinued.
2.0 d LI)
o Placebo (N=66) n Finasteride 10 m g (N=54)
Baseline -1 .o
o Placebo (N=40) a Finastetide (N=24)
* p en OS vs ~hcebo:
: ** p so.01
Month ** p
A FIGURE 1. Serum PSA (median change from baseline) with 10 mg finasteride m. l ) or placebo (0).
Month *** p so.oot
PSA = 3.1 @ml 10
for 12 months
(IA) and 24 months
Month + P
FIGURE 2. Serum PSA (median change from baseline) in patients with baseline PSA values of 0.6 to 1.O ng/mL after treatment for 12 months (2A) or 24 months (2B) with IO-mg finasteride R, 0) or placebo (0).
the increase in PSA. This effect may be due to loss of patients with recurrence. Several subset analyses were conducted to explore the possibility that some groups of patients might respond better to finasteride than others. As such, these analyses must be considered hypothesis generating rather than hypothesis testing. Analysis of PSA based on Gleason score (2 to 7 versus 8 to 10) or pathologic stage (B, C, or D) at the time of surgery indicated similar trends to
those observed in the all-patients-treated analysis. Placebo-treated patients had slow consistent increases in PSA until switched to finasteride and finasteride-treated patients had limited or no increases for approximately 6 months followed by slow increases. One subgroup analysis demonstrated a different pattern of response to treatment. Patients with baseline serum PSA levels between 0.6 and 1.0 ng/mL at randomization treated with finasteride for 12 months (Fig. 2A) and 24
20 15 3 g 2 2
10 5 0
GL.SCORE 7 PSA INCREASE
Serum PSA (ng/mL) (heavy shaded area),
13to24 Total *Placebo switched to finmteride
in 4 individual patients (A, B, C, D) treated or medical/surgical castration (no shade).
Patients with recurrence of prostate cancer
5/41 (12%) 9
Placebo* 7/66 8143
at month 12.
months (Fig. 2B) had no significant increase in PSA over baseline for the entire study. Those patients treated with placebo had an increase in PSA for 12 months, which returned to baseline when the patient was switched to finasteride. During the course of the study, a number of patients were found to have objective evidence of local or distant recurrence of prostate cancer. Table III summarizes the number of men in each treatment group based on recurrences that occurred during the first year (placebo controlled) and during the second year (all men on finasteride). Of the 54 men in the study who were treated with finasteride, 13 have subsequently received medical or surgical castration. In these men, the change in serum PSA in response to castration was determined as an indicator of hormonal responsiveness. The median length of treatment with finasteride was 13 months (range, 2 to 28 months) and the mean reduction in serum PSA was 84 * 8%. Nine of the 13 men had greater than 90% re-
U [email protected]
I MARCH 1995 I VOLUME 45, NUMBER 3
duction in PSA, 3 of the 13 had greater than a 60% reduction, and 1 of the 13 had no decrease. Figure 3 (A-D) shows individual PSA responses to castration following treatment with finasteride. The patient shown in Figure 3 had a slow increase in PSA while on finasteride for 24 months. He discontinued the study at month 24 and was treated with an LH-releasing hormone (LHRH) analogue resulting in a rapid decrease in serum PSA. The patient shown in Figure 3B had a very slow increase in serum PSA on finasteride for 12 months. When finasteride was discontinued, his PSA level increased rapidly until castration, which returned PSA to very low levels. The patient shown in Figure 3C, who developed metastatic disease, had a rise in PSA on placebo for 12 months followed by a rapidly increasing PSA on finasteride in the open extension. Treatment was discontinued at month 24, although by protocol this should have occurred at month 20. Treatment with an LHRH analogue produced the expected effect on PSA. The patient shown in Figure 3D also developed metastatic disease and had a rapid rise in PSA while on finasteride. Castration briefly slowed the rise, but only temporarily. Treatment with finasteride was generally well tolerated during the study During the first year, adverse clinical experiences were reported in 65% of finasteride-treated patients and 82% of placebotreated patients. The most common complaints were respiratory disorders, digestive disorders, and 495
urogenital disorders, with equal or fewer patients in the finasteride group for each category Ten men discontinued the study due to an adverse clinical event (5 on finasteride, 5 on placebo), mainly due to recurrence of prostate cancer. Sexually related adverse experiences were rare and consisted of breast tenderness or mild enlargement (gynecomastia) in 3 men on finasteride, decreased libido in 2 men on placebo and 1 man on finasteride, and hot flushes in 1 man on placebo. Laboratory adverse experiences were rare and consisted of increased liver function tests in 4 men on placebo but none on finasteride, increased PSA in 8 men on placebo and 5 on finasteride, and hematuria in 1 man in each treatment group. COMMENT Men who have undergone a radical prostatectomy for prostate cancer and subsequently are found to have detectable serum PSA levels are at high risk of developing objective clinical relapse. Recently, 5-a-reductase inhibitors have been shown to have a suppressive effect on prostate cancer cell lines”Jl and an acceptable safety profile for early therapy. 8,g These observations suggest a potential role for finasteride in men with residual PSA levels after prostatectomy The purpose of this study was to determine if 10 mg of finasteride taken orally once a day would prevent or delay the recurrence of prostate cancer. To evaluate this, the primary endpoint was the change in serum PSA levels over time and the secondary endpoint was the number of men diagnosed with local or distant recurrence. A dose of 10 mg/d was chosen based on the results of a previous, shortterm study in men with Stage D prostate cancer which showed that a 10 mg-dose would suppress serum PSA by 15% to 20% over 6 weeks without producing significant toxicity.12 The 120 patients recruited for this study had undergone a radical prostatectomy and later were found to have serum PSA levels between 0.6 and 10.0 ng/mL but with negative bone scans. They were randomized to receive 10 mg of finasteride or placebo in a blinded fashion for 12 months followed by a 12-month open extension in which all men received 10 mg of finasteride. The men treated with placebo showed a progressive slow increase in serum PSA throughout the initial 12-month treatment period. In contrast, the men treated with finasteride had a small decrease from baseline in serum PSA for the first 6 months followed by a slow increase in PSA at a rate less than the placebo group. The overall effect was to delay the rise by about 9 months. Similar trends were
observed during the second 12 months, which appeared to extend the delay in PSA increase to about 14 months. At the end of 12 months of treatment, 11 men had been diagnosed with local or distant recurrence of prostate cancer. Seven of these men had been treated with placebo and 4 had been treated with finasteride. During the second year, 5 additional patients in the group continuously treated with finasteride had a recurrence and 8 patients in the group initially treated with placebo and later switched to finasteride had a recurrence. The observation that more recurrences occurred in the men treated with finasteride only during the second year (n = 8) compared to those who were treated during the first year (n = 4) may suggest that early treatment is better than later treatment. Subsequent treatment with orchidectomy or LHRH analogues demonstrated that the recurrent disease was hormonally responsive and similar to responses observed in hormonally intact patients who are castrated. The recurrence data suggest that the effect on serum PSA reflects an effect on tumor cell growth and not simply an effect on an androgen-regulated protein. Thus, in this patient population, changes in serum PSA while on finasteride continue to have biologic meaning. At the present time, data on survival is not available and therefore the long-term response remains to be demonstrated. To determine if some men are likely to respond better than others, a number of subset analyses were conducted. Gleason score (the sum of the two most common Gleason grades in the tumor) is recognized as an important predictor of outcome in men with prostate cancer, with higher scores having a poorer prognosis.16 As in the allpatients-treated analysis, those men with high and low Gleason scores receiving placebo had progressive increases in serum PSA over 12 months, whereas in the finasteride-treated patients, PSA changes similar to the group as a whole were seen. Another important predictor of outcome in men with prostate cancer is pathologic stage at the time of radical prostatectomy.16 In these groups, a significant increase in serum PSA occurred in placebo-treated patients regardless of pathologic stage. In the finasteride-treated patients, those with Stages B and C disease had no significant increases from baseline during the first 6 months similar to the overall analysis, whereas those with pathologic Stage D disease appeared to derive no benefit from treatment with finasteride. Analysis of PSA changes based on baseline PSA levels prior to therapy suggested important
differences in response. Men with low baseline PSA levels had no increase in PSA over the 24 months of the study, suggesting that low-volume disease might be more responsive to treatment with a 5a-reductase inhibitor. The hormonal effects of finasteride in this study were similar to those observed in men with BPH. Serum DHT was decreased by 62% and serum testosterone was increased by 20% at month 12. LH was increased slightly but significantly, whereas FSH was not significantly affected.* As in previous studies, finasteride proved to be generally well tolerated with few side effects. Sexually related adverse effects were particularly limited compared with the effects observed with castration. In summary, this study demonstrated that in men at high risk for clinical relapse following radical prostatectomy as evidenced by the presence of residual serum PSA levels, treatment with 10 mg of finasteride delays the increase in PSA seen in untreated patients and then slows the subsequent rise leading to about a 9-month delay in PSA increase over 12 months and a 14-month delay over 2 years. It is important to note, however, that the 2-year data may represent the better responders in the study and therefore may be biased in favor of greater response. This effect can be observed in the PSA response between months 6 and 12 for the group treated for 12 months compared with those treated for 24 months. The overall effect on PSA suggests a tumor-suppressing effect by finasteride. This is supported by the reduction in local and distant recurrence of disease in the group treated with finasteride. Equally important is the fact that this effect occurs with minimum adverse effects that would be expected to provide a positive benefit-to-risk ratio. Subgroup analyses suggest that men with low baseline PSA levels might benefit the most from this approach. These results will require confirmation in larger studies over several years to determine if the observed effects translate into clinically meaningful reductions in recurrence rates and quality of life. Glenn Gormley, M.D. Merck Research Laboratories P.O. Box 2000 Rahway, NJ 07065 REFERENCES 1. Smith JA Jr., Hernandez AD, Wittwer CJ, Avent JM, Greenwood J, Hammond EH, and Middleton RG: Long-term
I MARCH 1995 I VOLUME 45, NUMBER 3
follow-up after radical prostatectomy. Identification of prognostic variables. Urol Clin North Am 18: 473476, 1991. 2. Frohmuller HG, and Wirth MP: Radical prostatectomy for carcinoma of the prostate: long-term results. Recent Results Cancer Res 126: 4349, 1993. 3. Frazier HA, Robertson JE, Humphrey PA, and Pulson DF: Is prostate-specific antigen of clinical importance in evaluating outcome after radical prostatectomy? J Urol 149: 516-518, 1993. 4. Abi-Aad AS, Macfarlane MT, Stein A, and Dekernion JB: Detection of local recurrence after radical prostatectomy by prostate-specific antigen and transrectal ultrasound. J Urol 147: 952-955, 1992. 5. Lightner DJ, Lange PH, Reddy PK, and Moore L: Prostate-specific antigen and local recurrence after radical prostatectomy. J Urol 144: 921-926, 1990. 6. Huson MA, and Catalona WJ: Effect of adjuvant radiation therapy on prostate-specific antigen following radical prostatectomy. J Urol 143: 1174-1177, 1990. 7. Morgan WR, Zincke H, Rainwater LM, Myers RP, and Klee GG: Prostate-specific antigen values after radical retropubic prostatectomy for adenocarcinoma of the prostate. Impact of adjuvant treatment (hormonal and radiation). J Urol 145: 319-323, 1991. 8. Gormley GJ, Stoner E, Bruskewitz R, Imperato-McGinley J, Walsh PC, McConnell JD, Andriole GL, Geller J, Bracken BR, Tenover JS, et al: The effect of finasteride in men with benign prostatic hyperplasia. N Engl J Med 327: 1185-1191, 1992. 9. Stoner E, and Members of the Finasteride Study Group: Three-year safety and efficacy data on the use of finasteride in the treatment of benign prostatic hyperplasia. Urology 43: 284-294, 1994. 10. Lamb JC, Levy MA, Johnson RK, and Isaacs JT: Response of rat and human prostatic cancers to the novel 5 alpha-reductase inhibitor, SK&F 105657. Prostate 21: 15-34, 1992. 11. Bolona M, Muzi P, Biordi L, Festuccia C, and Vicentini C: Antiandrogens and 5-alpha-reductase inhibition of the proliferation rate in PC3 and DU145 human prostatic cancer cell lines. Curr Ther Res 51: 799-813, 1992. 12. Presti JC Jr., Fair WR, Andriole G, Sogani PC, Seidmon EJ, Ferguson D, Ng J, and Gormley GJ: Multicenter, randomized, double-blind, placebo-controlled study to investigate the effect of finasteride (MK-906) on stage D prostate cancer. J Urol 148: 1201-1204, 1992. 13. Shapiro SS, and Wilk MB: An analysis of variance test for normality (complete samples). Biometrika 52: 591-611, 1965. 14. Levene H: Robust Tests for Equality of Variances. Contributions to Probability and Statistics. Stanford, Stanford University Press, 1960, pp 278-292. 15. Iglewicz B: Robust scale estimators and confidence intervals for location, in Hoaglin D, Mosteller F, and Tukey J (Eds): Understanding Robust and Exploratory Data Analysis. New York, John Wiley & Sons, 1983, p 424. 16. Humphrey PA, Frazier HA, Vollmer RT, and Pulson DF: Stratification of pathologic features in radical prostatectomy specimens that are predictive of elevated initial postoperative serum prostate-specific antigen levels. Cancer 71: 1821-1827, 1993.