This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lotan, Y. Articles by Lippman, S. M. Search for Related Content PubMed PubMed Citation Articles by Lotan, Y. Articles by Lippman, S. M.

Implications of the Prostate Cancer Prevention Trial: A Decision Analysis Model of Survival Outcomes

From the Department of Urology, University of Texas Southwestern Medical Center, Dallas; and Departments of Biostatistics and Applied Mathematics, and Clinical Cancer Prevention, University of Texas M.D. Anderson Cancer Center, Houston, TX

Address reprint requests to Yair Lotan, MD, Department of Urology, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-9110; e-mail: Yair.Lotan{at}UTSouthwestern.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
PURPOSE: To assess the estimated effect of finasteride prevention of prostate cancer on overall survival.

METHODS: Data for our decision tree model came from men in the two arms (finasteride or placebo) of the Prostate Cancer Prevention Trial (PCPT) and from clinically localized prostate cancer patients studied for long-term survival outcomes. Our model compared survival outcomes for men treated with finasteride or placebo. Prostate cancer rates were based on the 7-year period prevalence of prostate cancer detected in the PCPT; survival probabilities were abstracted from the long-term outcome studies. We assessed variability in the PCPT and long-term survival studies to test the variability of our model.

RESULTS: Survival advantages for a finasteride-treated (v those not treated with finasteride) population include gains of 1.7 months in 15-year cause-specific survival (assuming finasteride-altered Gleason scores and prostate cancer prevalence rates in the PCPT), of up to 3 months for cancers treated conservatively or surgically (assuming finasteride does not alter Gleason scores), and of 0.35 months (assuming the rate of cancers detected by for-cause biopsies in the PCPT), which increased to 1.7 months when assuming a 30% rate of biopsy-detected cancer in the PCPT placebo group. Model-variability analyses support several survival benefits associated with finasteride (eg, the uniform benefits assuming finasteride does not alter Gleason scores) but question certain others (eg, in 15-year recurrence-free survivals assuming finasteride does alter Gleason scores).

CONCLUSION: Finasteride can impart survival benefits according to our model, especially when we assume that finasteride does not alter Gleason scores.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
The landmark Prostate Cancer Prevention Trial (PCPT)¹ found that finasteride caused a 25% relative risk reduction in the period prevalence of prostate cancer (v placebo; 18% v 24%; P < .001). However, a greater proportion of tumors in the finasteride group (37%) than in the placebo group (22%) were higher grade (Gleason score 7; P < .001). To determine the implications of the PCPT finasteride results of fewer cancers overall but potentially a higher number of aggressive cancers, we developed decision analysis models capable of comparing projected survival rates for individuals who received finasteride prevention with survival rates for individuals who received placebo. These model projections are based on the Gleason score distributions and prostate cancer incidence and prevalence rates found in the PCPT integrated with published survival outcomes for patients diagnosed with clinically localized prostate cancer.

The effect on survival is an important component in determining the utility of a preventive strategy. Although it is intuitively sensible that a phase III reduction in cancer rates would be beneficial, the numerical benefits may be minimal when effects on the entire pool of potential male patients are evaluated. Childhood vaccines against measles, rubella, and pertussis increase life expectancy by about 0.1 month each.² Screening for cervical cancer every 3 years from age 20 will increase a woman’s life expectancy by 3 months,³ and tamoxifen used for prevention of breast cancer can increase the life span by 2.3, 1.4, and 0.9 months for patients initiating tamoxifen at ages 35, 50, and 60 years, respectively.⁴ Evaluating finasteride effects on survival will be an important aspect of determining its use as a preventive agent.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
We used DATA 3.5 (Treeage Software, Williamstown, MA) to develop a decision analysis model (Fig 1; Appendix) comparing the survival outcomes within a 15-year period of finasteride-treated men with survival outcomes of men not treated with finasteride: the model assumes that all men met the PCPT entry criteria ( 55 years old, prostate-specific antigen [PSA] of 3.0 ng/mL, and normal digital rectal examination [DRE]). The probability of cancer detection was abstracted from the PCPT 7-year prostate cancer prevalence data (18% and 24% for the finasteride and placebo groups, respectively).¹ PCPT Gleason score distributions also were used in constructing the model (Table 1). Model outcomes represent survival in months based on the presence or absence of cancer and Gleason score. We assumed that patients without cancer would survive the entire study period of 15 years. This assumption was made because most chemopreventive studies likely would target men whose baseline life expectancy is greater than 15 years.

View larger version (19K): [in this window] [in a new window]   Fig 1. Decision tree model. Let i = 0, 1 for placebo and finasteride groups, respectively; j = 0, 1, 2, or 3 for Gleason scores 2 to 4, 5 to 6, 7, 8 to 10, respectively; ij, proportion of patients with different Gleason sum scores. Expected survival in months is Yj based on varying survival rates for different Gleason scores from the literature (Table 2; mathematical notation and model derivation in Appendix).

We also used the decision analysis model to study the effects of different prostate cancer incidences, or varying probabilities of cancer, on survival in the study groups. Given that current cancer prevention studies attempt to identify cohorts at higher risk, we evaluated the potential benefit of finasteride in a population at a higher risk of developing prostate cancer. An additional analysis assumed that there were no differences in Gleason score (grade) distributions between the finasteride and placebo groups because the finasteride-associated increase in grade in the PCPT may be an artifact, as suggested by several studies discussed later.¹¹

We performed several additional analyses to evaluate the variability of our model in estimating survival outcomes. Initially we entered the overall relative risk reduction from the point estimate in the PCPT (24.8%) into our model to evaluate survival outcomes. We assessed the variability of the overall outcomes generated in this way by comparing them with outcomes calculated by entering the upper end (30.6%) and lower end (18.6%) of the 95% CI estimation¹ of the relative risk reduction into the model. We had to calculate the relative risk reduction conferred by finasteride in the subset of PCPT cancers detected for cause because the PCPT did not report this result. In addition, we applied the delta method to compute the variance of the point estimate of this relative risk reduction, and derived the 95% CI accordingly. The resulting point estimate was 23.5% (95% CI, 14.2% to 32.8%). We then entered this point estimate (23.5%) into our model to calculate the survival outcomes of men with cancers detected for cause, and assessed the variability of these outcomes by comparing them with outcomes generated in our model by entering the upper end (32.8%) and lower end (14.2%) of the 95% CI estimation of this point estimate. We also evaluated the variability of the model by assessing the survival outcomes generated by entering into the model the various estimates (including their SEs) for the Gleason score–associated survival rates reported for two of the long-term studies integrated within our model.^6,8

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
Table 3 lists the calculated survival outcomes (cause-specific, recurrence-free, or metastases-free outcomes) based on prostate cancer survival probabilities from the literature and on the Gleason score distribution and prostate cancer prevalence rates in the PCPT. The life-expectancy benefit for a finasteride-treated population (v those not treated with finasteride) in which the patients diagnosed with prostate cancer are treated conservatively is less than 1 month overall, with a slightly negative effect on the younger patients (–0.4 months in the group aged 55 to 59 years) and a positive effect in older patients (+0.8 months in the group aged 65 to 69 years). The benefit for a finasteride-treated population (v those not treated with finasteride) in which the patients diagnosed with prostate cancer are treated by radical prostatectomy is a 1.7-month advantage in 15-year cause-specific survival and a 0.7-month advantage in 15-year recurrence-specific survival (Fig 2). If the Gleason score distributions in the finasteride and control groups are assumed to be equivalent, then the finasteride group gains up to a 3-month cause-specific survival advantage in either the conservative or surgical treatment scenario (Fig 2).

View larger version (21K): [in this window] [in a new window]   Fig 2. Projected incremental survival differences of the Prostate Cancer Prevention Trial (PCPT) finasteride arm (v placebo; x-axis; assuming PCPT prostate cancer prevalence) modeled on historical data (y-axis). Bars reflect differences assuming finasteride does (blue) or does not (gray) alter Gleason scores. Error bars reflect estimates using the 95% CI of the risk reduction associated with finasteride in the PCPT.

In analyzing the subset of PCPT cancers detected by for-cause biopsies, the differences in overall survival outcomes between the finasteride-treated and control populations were smaller than were these differences in the analysis including all prostate cancers. The diminished differences are due to the lower incidences of cancer in the for-cause biopsy subset (4.6% and 6% in the finasteride and placebo groups, respectively). In the finasteride-treated population, in whom diagnosed prostate cancer is treated by radical prostatectomy, finasteride offered a 0.35-month survival advantage in 15-year cause-specific survival and a 0.09-month survival advantage in 15-year recurrence-specific survival (v control; Fig 3). In the finasteride-treated population in whom diagnosed prostate cancer was treated conservatively, finasteride had a minimal decrease in life expectancy in younger patients (0.2 months) and a 0.04-month benefit in older patients (v control; Fig 3). If the Gleason score distributions in the finasteride and control groups are assumed to be equivalent, there is up to a 0.7-month survival advantage for the finasteride-treated population. The variability study of the various survival outcomes of this subset of PCPT patients (diagnosed from for-cause biopsies) produced similar results to those of the variability study involving all PCPT prostate cancer patients (see the preceding paragraph; Fig 3).

View larger version (26K): [in this window] [in a new window]   Fig 3. Projected incremental survival differences of the Prostate Cancer Prevention Trial (PCPT) finasteride arm (v placebo; x-axis; assuming PCPT for-cause biopsy-detected prostate cancer prevalence) modeled on historical data (y-axis). Bars reflect differences assuming finasteride does (blue) or does not (gray) alter Gleason scores. Error bars reflect estimates using the 95% CI of the risk reduction associated with finasteride in the PCPT.

View larger version (22K): [in this window] [in a new window]   Fig 4. Survival differences of men using finasteride (assuming Prostate Cancer Prevention Trial [PCPT] for-cause detected cancer rate of finasteride) v men not using finasteride and with assumed 10% to 30% cancer rates, modeled on historical 15-year survival data,6,9 which are reflected in the y-axis. Bars show differences assuming finasteride does (blue) or does not (gray) alter Gleason scores. Error bars reflect estimates using the 95% CI of the risk reduction associated with finasteride in the PCPT.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

Our analysis, which used large studies available in the literature that stratified survival by Gleason scores, was designed to evaluate the effects of these confounding factors on survival. We first evaluated survival outcomes based on all men who developed prostate cancer, or on the 7-year period prevalence of prostate cancer, in the PCPT. The benefit for a finasteride-treated population, in whom diagnosed prostate cancer is treated by radical prostatectomy, is a 1.7-month advantage in 15-year cause-specific survival and 0.7-month advantage in 15-year recurrence-specific survival (v patients not receiving finasteride; Table 3; Fig 2). Older patients (65 to 69 years old), in whom diagnosed prostate cancer is treated conservatively, also had a 0.8-month survival advantage in the finasteride-treated population, but younger patients (55 to 59 years old) do not benefit from finasteride. This is due to the higher cause-specific mortality from prostate cancer at 15 years in this patient group when treated conservatively.

A major issue regarding the PCPT is the clinical relevance of the detected cancers. In the editorial accompanying the PCPT primary article, Scardino¹⁴ noted that several screening trials for prostate cancer detected less than 10% cancer rates,^14-16 versus the 18% and 24% rates in the finasteride and placebo groups, respectively, of the PCPT. The published studies we used for long-term survival analyses included patients with prostate cancer detected for cause. Therefore, we evaluated the subset of PCPT patients whose cancers also were detected for cause (in addition to the overall group of PCPT prostate cancer patients). This subset represented a significantly smaller proportion of the PCPT population (4.6% of the finasteride and 6% of the placebo arm) but a similar finasteride-related reduction (23.5%) in cancer detection. The risks of higher Gleason scores (7 to 10) were 48% v 29% in the finasteride and placebo groups, respectively. In the analysis assuming this difference in Gleason scores, the finasteride-treated population in whom diagnosed prostate cancer is treated by radical prostatectomy gained a 0.35-month advantage in 15-year cause-specific survival and had a 0.09-month advantage in 15-year recurrence-specific survival (Fig 3). This advantage would be even smaller if the analysis were based on 2001 Surveillance, Epidemiology, and End Results prostate-cancer incidence data—443 patient cases per 100,000 men ages 55 to 64 years old (www.SEER.cancer.gov), which amounts to a 3.1% prevalence during 7 years.

The PCPT population was at relatively low risk for prostate cancer (normal DRE and PSA of 3.0 ng/mL or lower), and there is a strong impetus to identify populations at higher risk. Finasteride offers significantly greater survival advantages with increasing incidences of prostate cancer, assuming the drug causes an equivalent risk reduction in patients at higher risk (Fig 4). Although tools are not currently available to predict with certainty which males will develop prostate cancer, there are groups at higher risk for prostate cancer, such as men with a positive family history, black men, and men with a PSA between 2.5 and 4 ng/mL.^14,17,18 If our ability to predict prostate cancer improves in the future (eg, through validated molecular risk markers), finasteride may become a more attractive preventive agent.

Several important points are raised by this analysis. The small benefit imparted by finasteride chemoprevention on survival based on PCPT results is primarily due to most men not developing prostate cancer, whether they take finasteride or not. In addition, most prostate cancers are Gleason sum 5 to 6 and have good survival regardless of treatment (Table 2). The small absolute difference in higher Gleason scores between the PCPT arms amounted to 1.4 patients of 100 entered onto the study. A 25% reduction in cancer detection attributed to finasteride decreases the absolute number of patients diagnosed with prostate cancer by six patients for every 100 treated. Disease prevalence and the mortality from a disease play critical roles in determining the potential benefit of a prevention strategy in terms of life expectancy in the general population. A careful consideration by Wright and Weinstein¹⁹ of gains in life expectancy from medical interventions found that a gain of 1 month due to a preventive strategy aimed at the general population signals an important intervention. Therefore, our calculated benefit from finasteride chemoprevention may be important. On the basis of prostate cancer prevalence rates and comparative data from two cohorts of surgically treated patients, the finasteride-treatment advantage in 15-year cause-specific survival was 1.7 months and the advantage in 10-year metastases-free survival was 2 months (Fig 2). Such a large benefit was not seen with conservative therapy in younger men. One must also consider that applying this strategy to older men with less than 15 years of life expectancy may not yield a survival advantage because of competing causes of mortality. Although the survival benefit for patients diagnosed by for-cause biopsies was 0.35 months, this benefit increased to 1.7 months when assuming an increased cancer incidence of 30%, which is a realistic incidence of future high-risk cohorts selected by molecular risk markers.

Examining other common prevention strategies places the present findings in perspective. A program of physical exercise begun at age 35 increases life expectancy by 6.2 months.²⁰ Complete smoking cessation at age 35 increases life expectancy by 9 months.²¹ Childhood vaccines against measles, rubella, and pertussis, however, increase life expectancy by about 0.1 month each.² Screening for cervical cancer every 3 years from age 20 will increase a woman’s life expectancy by 3 months.³

Survival outcome modeling also has been applied to the results of the sister trial of the PCPT, the Breast Cancer Prevention Trial (BCPT), which found that tamoxifen decreased the risk of invasive breast cancer by 50% in women with risks of breast cancer that were higher than average.²² This finding led to the approval of tamoxifen by the US Food and Drug Administration for primary prevention of invasive breast cancer in high-risk women.²³ Survival outcome modeling of BCPT results found that tamoxifen could increase survival by 2.3, 1.4, and 0.9 months for patients initiating tamoxifen at ages 35, 50, and 60 years, respectively.⁴ The survival advantage increased with increasing Gail-model risk but did not exceed 6.7 months.⁴

The maturity of pharmacologic cancer prevention is reflected by its two large-scale definitive trials, the PCPT and BCPT, with positive primary end point results of reduced cancer prevalence or incidence. This maturity highlights the importance of survival-outcome modeling to the interpretation of large-scale cancer prevention study. Primary mortality end points are extremely difficult if not impossible to design into these trials, which therefore need survival-outcome modeling to examine their impact on mortality and advanced-disease outcomes.

Our analysis demonstrates that the survival advantage of finasteride chemoprevention increases when assuming no difference in Gleason-score distribution between the finasteride and placebo groups. This finding may be important because previous studies suggest that the increase in prostate cancer Gleason score or grade in the finasteride arm of the PCPT may be an artifact. The earlier studies indicate that finasteride has no effect on the histologic characterization of cancer in needle-core samples,²⁴ and that Gleason grading is difficult to assess in hormonally treated patients.^11,12,25 The randomized, double-blind Proscar Long-Term Efficacy and Safety Study found no detectable difference between the Gleason scores of cancers detected in the finasteride or placebo group.¹¹ Civantos et al¹² compared prostate cancer specimens from five patients taking finasteride for 3 to 24 months before radical prostatectomy with specimens from 60 age- and stage-matched patients who had no finasteride before radical prostatectomy and from 113 patients who had leuprolide acetate therapy before radical prostatectomy. Finasteride increased Gleason scores in association with apoptosis and led to a reduction in size and collapse of cancerous glands. The authors recommended that "no Gleason grading or modified Gleason adjusting for the above changes should be done; otherwise the report would indicate a worse grade tumor than it is." They also stated that, "The clinician should inform the pathologist that the patient is receiving finasteride since tumor recognition and Gleason grade are affected."¹² The International Consultation on Prostate Intraepithelial Neoplasia and Pathologic Staging of Prostate Carcinoma (Workgroup 5) made a similar recommendation regarding the diagnosis of prostate cancer in needle biopsies: "Gleason grading of hormone-treated biopsies is fraught with difficulty and should not be undertaken."²⁵ Therefore, Gleason grade after finasteride or any other form of androgen-deprivation therapy may not be a reliable predictor of cancer behavior. This Gleason score–hormone therapy issue is important because Gleason score distribution assumptions in our study had a strong impact on finasteride outcome. When assuming an equal Gleason distribution in the finasteride group and the group without finasteride, finasteride produced 2 to 3 more months of 15-year cause-specific survival (v that associated with the different distributions reported for the PCPT; Fig 2).

The studies abstracted for our decision-tree model did not stratify by PSA, suggesting several limiting implications for our present analysis.^5,7-9 First, PCPT eligibility criteria selected for men with PSA less than 3 ng/mL and likely led to detecting cancers at an earlier stage than would occur in a population with higher PSA levels mixed in at baseline. This feature of the PCPT creates a lead-time bias resulting in longer survival for PCPT patients diagnosed with prostate cancer, but this bias would affect both arms of our model. Second, it is likely that outcomes of treating high-grade cancers detected in the PCPT population would be better than would outcomes of high-grade cancers treated in the general population, potentially benefiting the finasteride chemoprevention population. The absolute difference in the number of patients with Gleason sum score 8 to 10 is only one in 100, however, and thus any survival advantage for early detection of a high-grade cancer would not affect the overall model outcomes significantly. It also is arguable that patients with higher Gleason scores but low PSA values (< 4 ng/mL) will have better long-term outcomes than will those included in the model. Patients with a T1c cancer, PSA 2.6 to 4.0, and Gleason 8 to 10, however, have a predicted 40% risk of extraprostatic extension at the time of prostatectomy.²⁶ The 15-year surgical cause-specific survival used in the model for cancers classified as Gleason 7 to 10 was 71%, which is a favorable outcome for such cancers and likely is realistic for higher Gleason score cancers detected early.⁶ Another limitation of the present study is that our model can predict survival only within a 15-year period because there are no relevant studies with Gleason score distribution that had more than a 15-year follow-up.

There are several possible sources of variability within this model, including the variability in the overall prostate cancer risk reduction associated with finasteride in the PCPT. We performed several analyses to evaluate this PCPT-related source of variability. We evaluated the upper and lower ends of the 95% CI (18.6% to 30.6%) in addition to the point estimate of the risk reduction (24.8%) from the PCPT, as indicated by the error bars in Figure 2. We performed similar analyses to test the variability in survival outcomes based on the finasteride-associated reduction in the risk of the subset of cancers detected by for-cause biopsies, as indicated by the error bars in Figures 3 and 4. The variability analyses of outcomes, assuming that finasteride did not affect Gleason score distribution, supported a finasteride-associated benefit in all outcomes. The variability analyses of outcomes assuming that finasteride did affect Gleason score distribution supported certain finasteride-associated survival benefits (eg, in 15-year cause-specific survival in surgically treated patients [Fig 2]) and questioned certain others. Our variability analyses based on Gleason score–associated survival rates from two of the model’s other, long-term studies^6,8 indicated the limited variability (less than 0.3 months) in certain major survival outcomes (eg, 15-year cause-specific and 10-year metastasis-free survival) related to the overall prostate cancer risk reduction in the PCPT (Fig 2). Furthermore, we assessed the variability in our model by comparing the same end point (10-year metastases-free survival) in two different studies, and found a similar survival advantage in both studies (Figs 2 and 3).^6,8 Regardless of the assumed prostate cancer incidence (10%, 20%, or 30%; Fig 4), there is a large difference in variability between the cause-specific and recurrence-free survival outcomes, which may be due to the increased risk of developing a PSA recurrence (v the risk of cause-specific mortality) in the populations whose data were entered into the model.

The PCPT has confronted physicians with the dilemma of whether or not to offer patients the option of initiating finasteride therapy. Treating 100 patients may prevent six prostate cancers (prevalence), but not without financial treatment costs and potential finasteride side effects. Potential sexual side effects, although reversible, may discourage younger men from taking finasteride, even though this group likely has the most to gain from prostate cancer prevention. This study evaluated the effects of finasteride on survival, but many other clinically important issues are beyond the scope of this analysis, including the beneficial effect of finasteride on urinary symptoms, decrease in number of patients requiring prostate cancer treatment, sexual side effects, and cost. With good treatment outcomes possible for higher grade cancers detected early, there is a potential 1.7-month advantage in 15-year cause specific survival associated with finasteride, assuming PCPT baseline eligibility criteria and prostate cancer prevalence (Table 3). The analysis including prostate cancers detected by for-cause biopsies indicates a 0.35-month survival benefit for the finasteride-treated (v control) population comprising surgically treated prostate cancer patients (Fig 3). Selecting patients with higher (than in the PCPT) risks of prostate cancer can result in finasteride survival benefits of 1.7 months (Fig 4). Compared with many recognized prevention strategies, finasteride offers a similar survival advantage and may merit use for the prevention of prostate cancer.

	Appendix

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
Mathematical details of model construction. Figure 1 is a decision tree model that calculates survival for a population of men treated with placebo or finasteride by varying cancer rates, Gleason score distribution, and survival rates.

Let i = 0, 1 for placebo and finasteride groups, respectively and j = 0, 1, 2, or 3 for Gleason scores 2 to 4, 5 to 6, 7, or 8 to 10, respectively. Then the expected survival in months for treatment i, out of a maximum possible M months is

where P_i is the probability of cancer detection in treatment group i, _ij is the proportion of patients in treatment group i with Gleason sum score j, Y_j is the expected survival in months based on varying survival rates for different Gleason scores from the literature (Table 2), and M is the maximum months of follow-up. M = 180 in the 15-year follow-up group and M = 120 in 10-year follow-up group.

	Authors’ Disclosures of Potential Conflicts of Interest

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Consultant/advisory role: Claus G. Roehrborn, Merck; Scott M. Lippman, Merck. For a detailed description of this category, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.

	NOTES

 
Supported in part by grant No. CA16672 (University of Texas M.D. Anderson Cancer Center Support Grant) from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services.

Authors’ disclosures of potential conflicts of interest are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION Appendix Authors’ Disclosures of... REFERENCES

 
1. Thompson IM, Goodman PJ, Tangen CM, et al: The influence of finasteride on the development of prostate cancer. N Engl J Med 349:215-224, 2003[Abstract/Free Full Text]

2. White CC, Koplan JP, Orenstein WA: Benefits, risks and costs of immunization for measles, mumps and rubella. Am J Public Health 75:739-744, 1985[Abstract/Free Full Text]

3. Eddy DM: Screening for cervical cancer. Ann Intern Med 113:214-226, 1990

4. Hershman D, Sundararajan V, Jacobson JS, et al: Outcomes of tamoxifen chemoprevention for breast cancer in very high-risk women: A cost-effectiveness analysis. J Clin Oncol 20:9-16, 2002[Abstract/Free Full Text]

5. Albertsen PC, Hanley JA, Gleason DF, et al: Competing risk analysis of men aged 55 to 74years at diagnosis managed conservatively for clinically localized prostate cancer. JAMA 280:975-980, 1998[Abstract/Free Full Text]

6. Zincke H, Oesterling JE, Blute ML, et al: Long-term (15 years) results after radical prostatectomy for clinically localized (stage T2c or lower) prostate cancer. J Urol 152:1850-1857, 1994[Medline]

7. Chodak GW, Thisted RA, Gerber GS, et al: Results of conservative management of clinically localized prostate cancer. N Engl J Med 330:242-248, 1994[Abstract/Free Full Text]

8. Gerber GS, Thisted RA, Scardino PT, et al: Results of radical prostatectomy in men with clinically localized prostate cancer. JAMA 276:615-619, 1996[Abstract]

9. Han M, Partin AW, Pound CR, et al: Long-term biochemical disease-free and cancer-specific survival following anatomic radical retropubic prostatectomy: The 15-year Johns Hopkins experience. Urol Clin North Am 28:555-565, 2001[CrossRef][Medline]

10. Pound CR, Partin AW, Eisenberger MA, et al: Natural history of progression after PSA elevation following radical prostatectomy. JAMA 281:1591-1597, 1999[Abstract/Free Full Text]

11. Andriole GL, Guess HA, Epstein JI, et al: Treatment with finasteride preserves usefulness of prostate-specific antigen in the detection of prostate cancer: Results of a randomized, double-blind, placebo-controlled clinical trial—PLESS Study Group, Proscar Long-term Efficacy and Safety Study. Urology 52:195-202, 1998[CrossRef][Medline]

12. Civantos F, Watson RB, Pinto JE, et al: Finasteride effect on prostatic hyperplasia and prostate cancer: A comparative clinicopathologic study of radical prostatectomies. J Urol Path 6:1-13, 1997

13. Stephenson RA: Prostate cancer trends in the era of prostate-specific antigen: An update of incidence, mortality, and clinical factors from the SEER database. Urol Clin North Am 29:173-181, 2002[CrossRef][Medline]

14. Scardino PT: The prevention of prostate cancer: The dilemma continues. N Engl J Med 349:297-299, 2003[Free Full Text]

15. Smith DS, Humphrey PA, Catalona WJ: The early detection of prostate carcinoma with prostate specific antigen: The Washington University experience. Cancer 80:1852-1856, 1997[CrossRef][Medline]

16. Hoedemaeker RF, van der Kwast TH, Boer R, et al: Pathologic features of prostate cancer found at population-based screening with a four-year interval. J Natl Cancer Inst 93:1153-1158, 2001[Abstract/Free Full Text]

17. Catalona WJ, Antenor JA, Roehl KA, et al: Screening for prostate cancer in high risk populations. J Urol 168:1980-1984, 2002[CrossRef][Medline]

18. Catalona WJ, Ramos CG, Carvalhal GF, et al: Lowering PSA cutoffs to enhance detection of curable prostate cancer. Urology 55:791-795, 2000[CrossRef][Medline]

19. Wright JC, Weinstein MC: Gains in life expectancy from medical interventions: Standardizing data on outcomes. N Engl J Med 339:380-386, 1998[Abstract/Free Full Text]

20. Hatziandreu EI, Koplan JP, Weinstein MC, et al: A cost-effectiveness analysis of exercise as a health promotion activity. Am J Public Health 78:1417-1421, 1988[Abstract/Free Full Text]

21. Tsevat J, Weinstein MC, Williams LW, et al: Expected gains in life expectancy from various coronary heart disease risk factor modifications. Circulation 83:1194-1201, 1991[Abstract/Free Full Text]

22. Fisher B, Costantino JP, Wickerham DL, et al: Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 90:1371-1388, 1998[Abstract/Free Full Text]

23. Grann VR, Sundararajan V, Jacobson JS, et al: Decision analysis of tamoxifen for the prevention of invasive breast cancer. Cancer J 6:169-178, 2000[Medline]

24. Yang XJ, Lecksell K, Short K, et al: Does long-term finasteride therapy affect the histologic features of benign prostatic tissue and prostate cancer on needle biopsy? PLESS Study Group: Proscar Long-Term Efficacy and Safety Study. Urology 53:696-700, 1999[CrossRef][Medline]

25. Algaba F, Epstein JI, Aldape HC, et al: Assessment of prostate carcinoma in core needle biopsy: Definition of minimal criteria for the diagnosis of cancer in biopsy material. Cancer 78:376-381, 1996[CrossRef][Medline]

26. Partin AW, Mangold LA, Lamm DM, et al: Contemporary update of prostate cancer staging nomograms (Partin tables) for the new millennium. Urology 58:843-848, 2001[CrossRef][Medline]

Submitted March 27, 2004; accepted December 13, 2004.

This article has been cited by other articles:

				R. S. Svatek, J. J. Lee, C. G. Roehrborn, S. M. Lippman, and Y. Lotan The cost of prostate cancer chemoprevention: a decision analysis model. Cancer Epidemiol. Biomarkers Prev., August 1, 2006; 15(8): 1485 - 1489. [Abstract] [Full Text] [PDF]

				I. M. Thompson, C. M. Tangen, E. A. Klein, and S. M. Lippman Phase III Prostate Cancer Prevention Trials: Are the Costs Justified? J. Clin. Oncol., November 10, 2005; 23(32): 8161 - 8164. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lotan, Y. Articles by Lippman, S. M. Search for Related Content PubMed PubMed Citation Articles by Lotan, Y. Articles by Lippman, S. M.