J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 25 NUMBER 21 JULY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Prediction of Prostate Cancer for Patients Receiving Finasteride: Results From the Prostate Cancer Prevention Trial Ian M. Thompson, Donna Pauler Ankerst, Chen Chi, Phyllis J. Goodman, Catherine M. Tangen, Scott M. Lippman, M. Scott Lucia, Howard L. Parnes, and Charles A. Coltman Jr From the Department of Urology, University of Texas Health Sciences Center at San Antonio; Southwest Oncology Group, San Antonio; Department of Clinical Cancer Prevention, The University of Texas M.D. Anderson Cancer Center, Houston, TX; The Fred Hutchinson Cancer Research Center, Seattle, WA; Department of Pathology, University of Colorado Health Sciences Center, Denver, CO; and the Department of Cancer Prevention, National Cancer Institute, Washington, DC. Submitted June 10, 2006; accepted December 18, Supported in part by PHS Cooperative Agreement Grants No. CA37429, CA35178, CA45808, and 5UO1CA from the National Cancer Institute, Department of Health and Human Services. Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Address reprint requests to Ian M. Thompson, MD, Department of Urology, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Dr, San Antonio, TX 78229; thompsoni@uthscsa.edu. A B S T R A C T Purpose Using data from men in the finasteride group of the Prostate Cancer Prevention Trial (PCPT), we evaluated the impact of prostate-specific antigen (PSA) and other risk factors on the risk of prostate cancer. Methods Four thousand four hundred forty men in the finasteride group of the PCPT underwent prostate biopsy, had at least one PSA and a digital rectal exam (DRE) during the year before biopsy, had at least two PSA values from the 3 years before biopsy, and were on finasteride at the time of PSA evaluation. Logistic regression was conducted using the variables age, race, family history of prostate cancer, PSA, PSA velocity, and DRE adjusting for history of prior prostate biopsy. Results Six hundred forty-nine (14.6%) of 4,440 men were diagnosed with prostate cancer; 250 had Gleason 7 or higher cancer. Factors associated with an increased risk of prostate cancer included high PSA value and a rising PSA (24.9% risk for PSA value of 1.0 ng/ml and 24.8% risk for a rising PSA), family history of prostate cancer, abnormal DRE result, African American race, and older age. Factors associated with an increased risk of Gleason 7 or higher grade prostate cancer included PSA, abnormal DRE, and older age. A prior negative biopsy was associated with decreased risk of prostate cancer and high-grade prostate cancer. Conclusion Risk factors for prostate cancer on biopsy for men receiving finasteride include PSA, DRE, age, race, family history, and history of a prior negative biopsy. With the exception of the approximate reduction of PSA by half with finasteride, the impact of these risk factors is similar to men who do not receive finasteride. J Clin Oncol 25: by American Society of Clinical Oncology 2007 by American Society of Clinical Oncology X/07/ /$20.00 DOI: /JCO INTRODUCTION Prostate biopsy in a man undergoing regular prostate cancer screening is commonly prompted by an elevated prostate-specific antigen (PSA), generally of levels higher than 4.0 ng/ml, or an abnormal digital rectal examination (DRE). The Prostate Cancer Prevention Trial (PCPT) results demonstrated that PSA is not a dichotomous (normal/abnormal) marker of prostate cancer, but reflects a range of risk of the disease. 1 We recently demonstrated that other risk factors (including age, family history of prostate cancer in a first-degree relative, ethnicity/race, and a history of a prior negative prostate biopsy) provide independent predictive information for the risk of cancer or of high-grade disease, and more intensive prostate cancer screening may now be augmented using these risk factors. 2 Despite a 25% reduction in prostate cancer prevalence with finasteride in the PCPT, an increase in high-grade disease diminished initial interest in this strategy. 3 Recent data suggest that an improved sensitivity of PSA with finasteride for detection of cancer and high-grade disease may have played a role in this phenomenon, and an increased interest in chemoprevention has emerged. 2,4-6a As a result of this interest and because of the substantial number of men who are receiving finasteride for obstructive voiding symptoms who are also being monitored for prostate cancer with PSA, an investigation of the relationship between the standard risk factors and prostate cancer under finasteride use becomes relevant for clinical practice. 3 Finasteride causes a fall in PSA by approximately 50%, with a further fall over time, depending on presence or absence of cancer. 7 The drug also causes a reduction in prostate volume by 25%. 3 These factors affect interpretation of PSA 3076

2 Predicting Prostate Cancer With Finasteride and could affect interpretation of DRE in men receiving finasteride, thereby altering the decision to perform prostate biopsy and the risk of a diagnosis of prostate cancer. We herein characterize the effect of risk factors for prostate cancer under long-term finasteride use. METHODS The PCPT randomly assigned 18,882 men ages 55 or older with a normal DRE and PSA level of less than 3 ng/ml to either finasteride or placebo for 7 years. 3 A PSA test and DRE were performed annually. At annual visits, study participants were recommended to undergo prostate biopsy if DRE was abnormal or if PSA value for men receiving placebo exceeded 4.0 ng/ml. In the finasteride group, to achieve similar rates of biopsy recommendation in the two study groups, PSA values were initially doubled. However, because of a continued decline of PSA in this group, after 4 years of participation, finasteride PSA values were multiplied by 2.3. At the end of 7 years, all men not previously diagnosed with prostate cancer were requested to undergo an end-of-study biopsy. The PCPT was approved by institutional review boards at all sites. This analysis included all participants in the finasteride group who underwent prostate biopsy at any of the seven annual visits, including the endof-study required biopsy, and who had a PSA and DRE within 1 year of the biopsy. An additional PSA measurement during the 3 years before the biopsy was also required to compute PSA velocity. PSA measurements were excluded if they were taken while the participant was off finasteride, and as a result, the biopsies associated with the PSA and DRE tests used in this analysis were typically performed while the participant was still on finasteride. For participants with multiple biopsies, the last biopsy was used so the effect of a prior negative biopsy may be assessed. Similar qualitative results were obtained when the first biopsy was used instead of the last. Family history was coded as 1 if brother, father, or son had prostate cancer, otherwise it was coded as 0; current DRE result was coded as 0 for negative or normal and 1 for positive or abnormal, indicating suspicion for cancer; and prior biopsy was coded as 0 for no prior biopsy and 1 for one or more prior biopsies all negative for prostate cancer. Two indicators for race and ethnicity were considered, African American (1 for African American and 0 for anyone else) and Hispanic (1 for Hispanic ethnicity and 0 for anyone else). The value for age was the participant s age at biopsy. PSA was transformed to the natural logarithm log (PSA) for use in all models because the transformation improved the goodness of fit of models to the observed data. All PSA values within 3 years before biopsy were used to compute PSA velocity, defined as the slope of log PSA per year as obtained by linear regression. For the case in which only two PSA measurements were available and linear regression could not be computed, velocity was defined as the ratio of the change in log PSA to the change in time between PSA measurements. Prostate cancer was defined as any indication of prostate cancer of any grade on biopsy, and high-grade disease was defined as prostate cancer with Gleason grade 7 or higher. Gleason score was assigned centrally by the study pathologist. The 2 tests were used for comparison of characteristics between participants included and excluded in the analysis. Univariate logistic regression was used for associations between single risk factors, either as a factor variable or continuous regressor, and multiple logistic regression was used for multivariable risk associations. All statistical tests and P values were for null hypotheses versus two-sided alternatives. For both the prostate cancer and high-grade multivariable risk analysis, a stepwise Akaike Information Criterion (AIC) modeling strategy was used to search through the space of risk models containing all possible subsets of main effects and two-way interactions between the risk factors: log PSA, log PSA velocity, DRE, age, family history, race, ethnicity, and prior biopsy. The AIC objective function equals minus twice the log likelihood plus 2 times the number of parameters in the model. 8 Models with minimal AIC were identified and among sets of models with statistically similar AICs, models with no interaction terms were selected for reporting. RESULTS Of the 9,423 PCPT participants randomly assigned to finasteride, 5,676 men had at least one biopsy during study, either for cause or as required at the end of study. Of these men, 357 were excluded from this analysis because no PSA or DRE result within a year of biopsy was available, 740 were excluded because they were off treatment when their PSA level was measured, and 139 men were excluded from the analysis because they did not have an additional PSA within 3 years of biopsy. Thus, 4,440 men in the finasteride group were included in the analysis. Characteristics of the 4,440 PCPT finasteride arm participants included in this risk association analysis are presented in Table 1 and contrasted against 4,983 finasteride arm participants of the PCPT excluded because of failure to meet the eligibility requirements of this analysis. The subgroup of participants used in this analysis represents a heavily screened group, measured by the increased PSA and DRE screens, and increased number of biopsies. However, in part due to the domination of the number of end-of-study biopsies relative to interim PSA or abnormal DRE-prompted biopsies of the 4,440 biopsy results used in this analysis, 3,563 (80.2%) were required end-of-study Table 1. Characteristics of Participants From the Finasteride Arm of the PCPT Included and Excluded From the Risk Analysis Characteristic Included in Risk Analysis (n 4,440) Excluded From Risk Analysis (n 4,983) No. % No. % Age at baseline, years , , , , , , or older Family history Negative 3, , Positive Race/ethnicity White 4, , African American Hispanic Other No. of PSA screens None , , , , No. of DRE measures None , , , , No. of biopsies , , , Years on finasteride , , , Abbreviations: PSA, prostate-specific antigen; DRE, digital rectal exam. P

3 Thompson et al biopsies and 877 (19.8%) were interim this differential has a minimum impact on results. Our previous analyses adjusting for population differences between participants included and excluded on the operating characteristics of PSA found no difference between adjusted and unadjusted estimates, which was driven by the required end-ofstudy biopsies of the PCPT. 9 The subgroup of participants used for analysis is a treatment-compliant group, measured by years on finasteride; 95.6% of the participants considered in this analysis were on finasteride 5 years or more by the time of their last prostate biopsy. Hence, results from the analysis are generally limited to men on finasteride 5 years or more. Men included tended to be younger and were more often white than those excluded, though the difference in demographic distribution is small and the statistical significance is likely driven by large sample sizes. For men with one or more previous biopsies in this analysis, the median time between the biopsy used for analysis and the previous biopsy was 3.1 years (range, 25 days to 6.6 years). Six hundred forty-nine (14.6%) of the finasteride group participants included in this analysis developed prostate cancer over the course of the trial, and of these, 250 (38.5% of prostate cancers and 5.6% of participants) had high-grade tumors. The median PSA level (measured within 1 year of biopsy) for the 4,440 participants in this analysis was 0.5 ng/ml (range, 0.3 to 111.0), which is depressed from the usual range as finasteride approximately halves PSA. Therefore, to avoid confusion in interpretation, PSA will be referred to as finasteride PSA (finpsa). The distribution of prostate cancer and high-grade disease cases by finpsa level is presented in Tables 2 and 3. The risk of prostate cancer starts at 6.2% for finpsa levels of less than 0.5 ng/ml and increases to 24.9% by the threshold finpsa level of 1.0 ng/ml. A finpsa level of 2.0 ng/ml correspondingly doubles prostate cancer risk to 50%. Similarly, the risk of high-grade disease steadily increases from 1% at finpsa levels of less than 0.5 ng/ml to 39.7% for finpsa levels exceeding 3.0 ng/ml. A steady or declining finpsa, measured by a velocity of less than or equal to 0, corresponds to a risk of prostate cancer of 7.5% and high-grade disease of 1.7%. A rising PSA, measured by a velocity of greater than 0, increases the risk of prostate cancer from 7.5% to 24.8% and high-grade risk from 1.7 to 11.3%. An abnormal DRE more than doubles the risk of prostate cancer (13.0% for normal DRE and 30.5% for abnormal) and nearly triples the risk of high-grade disease (4.7% for normal DRE and 14.9% for abnormal). A positive family history increases the risk of prostate cancer to 20.3% from 13.5% for a negative family history, and there is a significant increase in the risk of prostate cancer and high-grade disease for African Americans. Tables 2 and 3 also present unadjusted odds ratios (ORs) for each of the risk factors, which do not measure independent prognostic information Table 2. Distribution of Prostate Cancer by Risk Factors and Odds Ratios From Univariate and Multivariate Models Factor No. of Participants Prostate Cancer Univariate Multivariate No. % OR P OR P Finasteride PSA, ng/ml , , Log finasteride PSA, log ng/ml (continuous) log ng/ml/yr 0.0 2, , log ng/ml (continuous) DRE Normal 4, Abnormal Family history Negative 3, Positive Age at biopsy African American No 4, Yes Prior negative biopsy No 3, Yes Abbreviations: OR, odds ratio; PSA, prostate-specific antigen; DRE, digital rectal exam. Denotes the reference value for the logistic regression. The logarithm of PSA treated as a continuous risk factor in logistic regression; OR corresponds to increase in odds for a unit increase in log PSA. Log PSA velocity treated as a continuous risk factor in logistic regression; OR corresponds to increase in odds for a unit increase in log PSA velocity JOURNAL OF CLINICAL ONCOLOGY

4 Predicting Prostate Cancer With Finasteride Table 3. Distribution of High-Grade Disease by Risk Factors and Odds Ratios From Univariate and Multivariate Models Factor No. of Participants High-Grade Disease Univariate Multivariate No. % OR P OR P Finasteride PSA (ng/ml) , , Log finasteride PSA, log ng/ ml (continuous) log ng/ml/year 0.0 2, , log ng/ml (continuous) DRE Normal 4, Abnormal Family history Negative 3, Positive Age at biopsy African American No 4, Yes Prior negative biopsy No 3, Yes Abbreviations: OR, odds ratio; PSA, prostate-specific antigen; DRE, digital rectal exam. Denotes the reference value for the logistic regression. The logarithm of PSA treated as a continuous risk factor in logistic regression; OR corresponds to increase in odds for a unit increase in log PSA. Log PSA velocity treated as a continuous risk factor in logistic regression; OR corresponds to increase in odds for a unit increase in log PSA velocity. for each accounting for known risks, but rather the effect of each risk factor in isolation of other known risks, and adjusted ORs from the multivariable models, which combine all the significant risk factors, for risk of prostate cancer and high-grade disease. For calculation of adjusted risk of prostate cancer using a multivariable analysis, logarithm of finpsa (OR, 3.48; 95% CI, 3.08 to 3.92; P.001), DRE result (OR, 2.77; 95% CI, 2.14 to 3.57: P.001), family history of prostate cancer (OR, 1.45; 95% CI, 1.16 to 1.82; P.001), age (OR, 1.02; 95% CI, 1.00 to 1.04; P.02), African American race (OR, 1.79; 95% CI, 1.16 to 2.75; P.01), and prior negative biopsy (OR, 0.66; 95% CI, 0.51 to 0.86; P.002) all remained statistically significant. For calculation of the risk of high-grade disease using a multivariable analysis, logarithm of finpsa (OR, 4.86; 95% CI, 4.06 to 5.82; P.001), DRE result (OR, 3.06; 95% CI, 2.16 to 4.33; P.001), age (OR, 1.04; 95% CI, 1.02 to 1.07; P.001), and prior negative biopsy (OR, 0.63; 95% CI, 0.43 to 0.92; P.02) remained statistically significant, but family history Table 4. Risk Factors in the Multivariable Risk Models for Participants No Finasteride 3 Placebo v Finasteride 5 Years Risk Factor OR for Prostate Cancer No Finasteride 3 Placebo Finasteride 5 Years OR for High-Grade Disease No Finasteride 3 Placebo Finasteride 5 Years Log PSA DRE Family history NS NS Prior negative biopsy Age NS African American NS NS Abbreviations: OR, odds ratio; PSA, prostate-specific antigen; DRE, digital rectal exam; NS, not significant. Taken from the Prostate Cancer Prevention Trial placebo arm study

5 Thompson et al and African American race were no longer statistically significant (P.05). Table 4 summarizes which risk factors remained significant and their magnitudes in the multivariable risk models for prostate cancer and high-grade disease for this analysis of 4,440 PCPT participants on finasteride contrasted with the analysis of 5,519 PCPT participants on the placebo arm previously reported in Thompson et al. 2 The effects of PSA, DRE, family history, and prior negative biopsy remain similar across finasteride and placebo for prediction of prostate cancer and high-grade disease. The effect of age is similar between finasteride and placebo for high-grade disease. Unlike for the placebo arm, age remained a statistically significant predictor for prostate cancer on finasteride. However, the OR approached 1.0. In the placebo analysis, African Americans had statistically significant higher risk for highgrade disease but not for prostate cancer, yet in the finasteride arm, the situation was reversed. This could be due to an enhanced effect of finasteride against high-grade cancers in African Americans, but the small numbers of African Americans in this analysis (N 152) prohibits any determination. DISCUSSION A man who undergoes annual DRE and PSA determinations for the early diagnosis of prostate cancer should currently expect not to be told that his PSA is either normal or abnormal, but should expect an evaluation of his level of risk; both for prostate cancer and for highgrade prostate cancer. We have recently developed a prostate cancer risk calculator to help in this assessment. 2 The calculator, based on the results of prostate biopsies of 5,519 men from the PCPT, including biopsies performed at the end of the study, which were performed regardless of PSA and DRE findings, is less affected by previous estimates, which were confounded by ascertainment bias. Additionally, the prospective collection of other risk factors allowed their incorporation into the estimate of risk. A man undergoing annual screening should also have the expectation that his physician is familiar with the results of the PCPT, demonstrating a 24.8% reduction in the risk of prostate cancer in men who received finasteride. 3 For a man who weighs the risks and benefits of the use of finasteride and opts to take it for prevention or who uses the drug for treatment of lower urinary tract symptoms from prostate enlargement, the interpretation of both DRE and PSA are confounded by the drug s results. This study helps physicians and patients understand how risk of prostate cancer is affected by these measures. As in our previous analysis of men in the placebo group of the PCPT, a group of risk factors had a profound impact on a man s risk of both prostate cancer and high-grade disease if he was receiving finasteride. PSA was highly related to the risk of prostate cancer, reaching a 24.9% risk for values of 1.0 to 1.5 ng/ml; the risk of high-grade cancer reached 29.6% for PSA values of 2.0 to 3.0 ng/ml. The change in PSA over time (PSA velocity) had a close relationship with risk of both cancer and high-grade disease with a steady or declining PSA value having a risk of cancer of 7.5% and of high-grade disease of 1.7% compared with risks of 24.8% and 11.3%, respectively, for men with rising PSA. However, as we found and subsequently described in detail in our analyses of the placebo arm of this trial, PSA velocity was not statistically significant in any multivariable risk model accounting for PSA. That is, PSA velocity added no independent diagnostic information to PSA for the prediction of prostate cancer in screened participants. 2 This observation, which conflicts with that of previous reports, could be explained by the high proportion of small, low-grade prostate cancers detected in PCPT because of end-of-study biopsies, but could also be related to previous studies failures to examine the correlation of PSA velocity and prostate cancer risk after accounting for the diagnostic effect of PSA, a highly-correlated measure. 10 It has been speculated that finasteride could affect the detection of prostate cancer by DRE. Our data demonstrate the significant impact of DRE in men receiving finasteride, as illustrated by a fold increased risk of cancer and a 3.06-fold increased risk of highgrade disease for an abnormal DRE. It is important to understand generalizability of these data. Men in the PCPT were thought to be at a low risk for biopsy-detected prostate cancer by enrollment PSA values 3.0 ng/ml and a normal DRE. The cohort was also predominantly white. Also, as men undergoing biopsy in this analysis either had an adjusted PSA of 4.0 ng/ml or greater, abnormal DRE, or underwent end-of-study biopsy, this estimate is optimal for men on finasteride with a value more than 2.0 ng/ml or who have been taking this medication for 7 years. How do these results integrate with previous findings and with clinical care? We hope that a man who seeks the counsel of his physician regarding prostate cancer risk would be educated regarding the potential benefits and risks of early detection. 11 If he opts for screening, initial assessment using the risk calculator previously described would assist the patient-physician discussion regarding appropriateness of prostate biopsy. 2 Thereafter or concurrently, patient education would be initiated regarding chemoprevention of prostate cancer with finasteride. For the individual who opts for finasteride chemoprevention, PSA should fall over the ensuing years. Although it is attractive to consider the degree of fall within the first year or two related to subsequent prostate cancer risk, the data in this analysis, most of which are from 5 or more years of finasteride treatment, do not allow precise early estimates of risk. Nonetheless, after 5 or more years of therapy, the man with a low risk of prostate cancer should have a very low absolute value of PSA ideally, less than 0.5 ng/ml, and it should not be increasing. This report shows that similar risk factors have a similar effect on the risk of prostate cancer and high-grade disease to that which we reported for equivalent men not on finasteride from the same trial, and hence, the same calculator can be used, approximately doubling the finpsa value observed; the area underneath the receiver operating characteristic curve equals 76.8% for prostate cancer and 85.5% for high-grade disease in this population. Extending results from analyses of the PCPT for use in the public will help physicians and their patients assess their current risk of disease and their risk of disease over time, whether or not they opt to use finasteride to reduce that risk. Physicians should communicate that PCPT risk estimates may be inflated due to inclusion of overdiagnosed cancers, but give an accurate picture of the high prevalence of a disease for which currently the field cannot reliably distinguish overdiagnosed from harmful cancers. Conversely, these results do reflect a sextant (6-core) biopsy, whereas the majority of biopsies today are 10 to 12 cores, which would be expected to lead to greater rates of detection. 12 A reasonable recommendation would be to use the estimates of high-grade prostate cancer from the risk calculator to help patients evaluate their risk of those tumors of greatest biologic aggressiveness and to use these risks to then evaluate their desire to undergo prostate biopsy JOURNAL OF CLINICAL ONCOLOGY

6 Predicting Prostate Cancer With Finasteride AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST Although all authors completed the disclosure declaration, the following authors or their immediate family members 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. For a detailed description of the disclosure categories, or for more information about ASCO s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Employment: N/A Leadership: N/A Consultant: Ian M. Thompson, Mission Pharmacal Stock: N/A Honoraria: Scott M. Lippman, Merck Research Funds: M. Scott Lucia, GlaxoSmithKline; Ian M. Thompson, Astra Zeneca Testimony: N/A Other: N/A AUTHOR CONTRIBUTIONS Conception and design: Ian M. Thompson, Catherine M. Tangen, Charles A. Coltman Jr Financial support: Charles A. Coltman Jr Administrative support: Ian M. Thompson, Charles A. Coltman Jr Collection and assembly of data: Phyllis J. Goodman, M. Scott Lucia Data analysis and interpretation: Donna Pauler Ankerst, Chen Chi, Phyllis J. Goodman, Catherine M. Tangen, Scott M. Lippman, M. Scott Lucia Manuscript writing: Ian M. Thompson, Donna Pauler Ankerst, Scott M. Lippman, Howard L. Parnes Final approval of manuscript: Ian M. Thompson, Donna Pauler Ankerst, Chen Chi, Phyllis J. Goodman, Catherine M. Tangen, Scott M. Lippman, M. Scott Lucia, Howard L. Parnes, Charles A. Coltman Jr REFERENCES 1. Thompson IM, Pauler DK, Goodman PJ, et al: Prevalence of prostate cancer among men with a prostate-specific antigen level or 4.0 ng per milliliter. N Engl J Med 350: , Thompson IM, Ankerst DP, Chi C, et al: Assessing prostate cancer risk: Results from the Prostate Cancer Prevention Trial. J Natl Cancer Inst 98: , Thompson IM, Goodman PJ, Tangen CM, et al: The influence of finasteride on the development of prostate cancer. N Engl J Med 349: , Unger JM, Thompson IM, LeBlanc M, et al: Estimated impact of the Prostate Cancer Prevention Trial on population mortality. Cancer 103: , Klein EA, Tangen CM, Goodman PJ, et al: Assessing benefit and risk in the prevention of prostate cancer: The Prostate Cancer Prevention Trial revisited. J Clin Oncol 23: , Lippman SM, Lee JJ: Reducing the risk of chemoprevention: Defining and targeting high risk 2005 AACR Cancer Res and Prevention Foundation Award Lecture. Cancer Res 66: , a. Thompson IM, Chi C, Ankerst DP, et al: Effect of finasteride on the sensitivity of PSA for detecting prostate cancer. JNCI 98: , Fetzioni RD, Howlader N, Shaw PA, et al: Long-term effects of finasteride on prostate specific antigen levels: Results from the Prostate Cancer Prevention Trial. J Urol 174: , Akaike H: A new look at the statistical model selection. IEEE Transactions on Automatic Control AC-19: , Thompson IM, Ankerst DP, Chi C, et al: Operating characteristics of prostate specific antigen in men with an initial PSA level of 3.0 ng/ml or lower. JAMA 294:66-70, Carter HB: Assessing risk: Does this patient have prostate cancer? J Natl Cancer Inst 98: , Chan EC, Vernon SW, O Donnell FT, et al: Informed consent for cancer screening with prostate-specific antigen: How well are men getting the message? Am J Public Health 93: , Levine MA, Ittman M, Melamed J, et al: Two consecutive sets of transrectal ultrasound guided sextant biopsies of the prostate for the detection of prostate cancer. J Urology 159: ,