- Email: [email protected]
Survival results in patients with screen-detected prostate cancer versus physician-referred patients treated with radical prostatectomy: Early results
Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
Original article
Survival results in patients with screen-detected prostate cancer versus physician-referred patients treated with radical prostatectomy: Early results夡 Kimberly A. Roehl, M.P.H.a, Scott E. Eggener, M.D.b, Stacy Loeb, M.D.b, Norm D. Smith, M.D.b, Jo Ann V. Antenor, M.P.H.c, William J. Catalona, M.D.b,* a
b
Department of Psychiatry, Washington University School of Medicine, St. Louis, MO 63105, USA Department of Urology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USA c Department of Neurology, Washington University School of Medicine, St. Louis, MO 63105, USA Received 17 August 2005; received in revised form 29 November 2005; accepted 30 November 2005
Abstract Objective: Screening using a standardized protocol may improve outcomes of patients undergoing treatment for prostate cancer. We compared the 7- year progression-free survival rates after radical retropubic prostatectomy in patients whose prostate cancer was detected through a formal screening program with those of patients referred for treatment by other physicians who did not use a standardized screening/referral protocol. Methods: A single surgeon (W.J.C.) performed radical retropubic prostatectomy in 3,177 consecutive patients between 1989 and 2003. Of these patients, 464 had cancer detected in a screening study, and 2,713 were referred from outside institutions. We compared the screened and referred cohorts for age at surgery, clinical stage, pathologic stage, Gleason sum, preoperative prostate-specific antigen (PSA) levels, and adjuvant radiation therapy. Kaplan-Meier product limit estimates were used to calculate 7-year progression-free probabilities, and Cox proportional hazards models were used to determine the clinical and pathologic parameters associated with cancer progression in each group. Results: The overall 7-year progression-free survival rates were 83% for the screened patients compared with 77% for the referred patients (P ⫽ 0.002). Preoperative PSA, Gleason sum, clinical stage, pathologic stage, and adjuvant radiotherapy were all significantly associated with cancer progression. There was a significantly higher proportion of referred patients with a preoperative PSA ⱖ10, Gleason sum ⱖ7, and nonorgan-confined disease. Conclusions: Patients with screened-detected prostate cancer have more favorable clinical and pathologic features, and 7-year progression-free survival rates than referred patients. On multivariate analysis, including other clinical variables, screening status was a significant independent predictor of biochemical outcome. © 2006 Elsevier Inc. All rights reserved. Keywords: Prostate; Prostatic neoplasms; Prostate-specific antigen; Screening; Radical prostatectomy
1. Introduction The principal goal of an early detection program for prostate cancer is to detect clinically relevant tumors while they are still curable. To examine whether current screening protocols might accomplish this objective, we compared patients whose cancer was diagnosed through a prostate夡 Supported by a grant from Beckman Coulter Incorporated, Fullerton, CA, and by the Urological Research Foundation. * Corresponding author. Tel.: ⫹1-312-695-4471; fax: ⫹1-312-6951482. E-mail address: [email protected] (W.J. Catalona).
1078-1439/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.urolonc.2005.11.039
specific antigen (PSA) and digital rectal examination (DRE)-based screening study to those whose cancer was detected during routine clinical practice. van der CruijsenKoeter et al. [1] recently reported that both clinical stage and Gleason grade were significantly more favorable in the screening arm of the European Randomized Study of Screening for Prostate Cancer; however, they did not address the association between screening status and subsequent treatment outcomes. Pending the results of randomized clinical trials, there is increasing evidence suggesting that screening leads to improved outcomes [2– 6]. Based on cases diagnosed between 1992 and 1995, approximately 90% of new prostate cancer
466
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
cases are expected to be local or regional at diagnosis. Furthermore, prostate cancer mortality rates continue to decline in the Surveillance, Epidemiology and End Results database. In the present study, we sought to determine whether screening for prostate cancer might be responsible, at least in part, for these encouraging trends. For this purpose, we compared pathologic tumor features, progressionfree, cancer-specific, and all-cause survival rates after radical retropubic prostatectomy (RRP) between men who participated in a standardized screening program with those in referred patients who did not. In addition, we included multivariate models to distinguish which preoperative clinical or pathologic parameters were independent predictors of cancer progression.
2. Materials and methods From January 1989 to February 2003, a single surgeon (W.J.C.) performed RRP in 3177 consecutive men using a standardized technique [7]. These patients comprised 2 separate cohorts. The first included 464 (15%) whose prostate cancer was detected through a screening study. The study protocol involved PSA testing and DRE at 6 or 12-month intervals, as previously described [8]. A biopsy was recommended for a suspicious DRE or PSA level higher than 4.0 ng/ml (until May 1995) or 2.5 ng/ml (after May 1995). The second cohort consisted of 2713 (85%) men who were diagnosed with prostate cancer by other physicians and either requested or were referred for RRP. Indication for biopsy in these men was not standardized and may have included PSA, DRE, or evaluation for other problems. Clinical stage was classified as either localized (stage cT1 or cT2) or advanced (stage ⱖcT3). Pathologic stage was considered organ-confined if the cancer was confined to the prostate with clear (noncancerous) surgical margins (stage pT1 R0 or pT2 R0). Any specimen with extraprostatic tumor extension, positive (cancerous) surgical margins, seminal vesicle invasion, or lymph node metastases was classified as pathologically advanced disease (stage pT2 R1, pT3a/b, N1 or higher). Postoperatively, both cohorts underwent PSA testing every 6 months and DRE yearly. Biochemical progression was defined as a detectable PSA level (⬎0.2 ng/ml) confirmed by repeat measurements.
known to be living at the most recent date of follow-up and used the log-rank test to compare the strata. For multivariable analyses, we used the Cox proportional hazards model to determine whether study entry status (screened or referred) was associated with risk of cancer progression, accounting for pretreatment variables, such as age at diagnosis (as a continuous variable), preoperative PSA, biopsy Gleason score (ⱕ6 vs. 7 vs. 8 –10), and clinical stage (T1 vs. T2 vs. T3). Adjusted relative risks and their respective 95% confidence intervals (CIs) and P values are reported for the full model. We also sought to determine whether study entry status was associated with risk of cancer progression, accounting for pathologic variables, such as surgical Gleason score (ⱕ6 vs. 7 vs. 8 –10), pathologic stage (pT1/2 R0 vs. pT2 R1 / T3a/b vs. pT3c/N1), and use of adjuvant radiotherapy. All statistical analyses were performed using Statistical Analysis Systems (version 8.2; SAS Institute, Inc., Cary, NC) and SPSS 10.0 for Windows (SPSS Inc., Chicago, IL).
3. Results 3.1. Clinical characteristics Table 1 shows the demographics of our study population. Mean age at diagnosis was higher in the screened group than the referred group (64.3 ⫾ 6.5 vs. 60.4 ⫾ 7.3 years, P ⬍ 0.0001). Race was also significantly different between groups. Although 94% of the patients were white, there was a higher proportion of black, Hispanic, or Asian patients in the referred than in the screened population (7% vs. 5%, respectively, P ⫽ 0.005). The 2 cohorts also differed in other clinical parameters. The median preoperative PSA was lower in the screened population (5.1 ⫾ 7.0 vs. 6.0 ⫾ 6.8 ng/ml, P ⬍ 0.0001). There were more screened patients diagnosed with a PSA value of 2.6 – 4.0 ng/ml (24% vs. 9%, P ⫽ 0.001) and significantly fewer screened patients with a PSA value of ⬎10.0 ng/ml (13% vs. 18%, P ⫽ 0.001). There were more screened patients with a biopsy Gleason sum of ⱕ6 (85% vs. 79%, P ⫽ 0.02). Although there were more PSA-detected cancers (clinical T1c) in the referred patients (57% vs. 51%, P ⫽ 0.001), there was no significant difference in the proportion of clinically localized cancers between the groups.
2.1. Statistical analysis 3.2. Pathologic characteristics To compare clinical and pathologic parameters between groups, chi-square, Armitage chi-square, Fisher exact test, or t-tests were used for normally distributed, binomial proportions or categorical data, and the Wilcoxon rank sum test was used for nonparametric continuous variables. KaplanMeier product limit estimates were used to compare cancer progression and mortality rates by screening status. For both cancer-specific and overall survival, we censored patients
There were significantly fewer patients with Gleason sum ⱖ7 in the screened group (26%) than in the referred group (40%) (P ⬍ 0.0001) (Table 2). Patients in the screened group were more likely to have organ-confined disease than those in the referred group (75% vs. 68%, P ⫽ 0.02). However, there was no significant difference between the groups in the proportion of men electing adjuvant ra-
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
467
Table 1 Clinical characteristics for screened versus referred patients No. patents screened (%) Age at diagnosis (yrs) ⬍50 50–59 60–69 ⱖ70 Race White Black Other Preoperative PSAb ⬍2.6 2.6–4.0 4.1–10.0 ⬎10.0 Biopsy Gleason sumc 2–6 7 8–10 Clinical stage cT1a/b cT1c cT2 cT3 Median follow-up (range)
P Valuea
No. patents referred (%)
8 (2) 102 (22) 251 (54) 103 (22)
195 (7) 1013 (37) 1204 (45) 301 (11)
⬍0.0001
443 (95) 19 (4) 2 (1)
2532 (93) 76 (3) 104 (4)
0.005
38 (8) 108 (24) 253 (55) 58 (13)
168 (6) 239 (9) 1773 (66) 491 (19)
⬍0.0001
388 (85) 48 (11) 20 (4)
2084 (79) 458 (17) 111 (4)
0.002
50 (2) 1533 (57) 1094 (40) 30 (1) 45 mos (0–158)
0.02
5 (1) 236 (51) 220 (47) 2 (1) 79.5 mos (0–160)
⬍0.0001
a
Armitage chi-square P values show differences by group. Preoperative PSA unavailable for 49 subjects. c Biopsy Gleason sum unavailable for 68 subjects. b
diotherapy (4% in the screened group vs. 7% in the referred group, P ⫽ 0.08). Pathologic stage was significantly associated with biochemical progression (P ⬍ 0.0001).
No. patents screened (%)
No. patents referred (%)
P Valuea
337 (74) 96 (21) 25 (5)
1597 (60) 898 (34) 177 (7)
⬍0.0001
346 (75) 95 (21) 21 (4) 20 (4)
1844 (68) 714 (27) 135 (5) 178 (7)
0.02
time was significantly longer in the screened patients (79.5 ⫾ 44.5 vs. 45 ⫾ 40.2 months, P ⬍ 0.0001). The 7-year progression-free survival rates were 83% (95% CI 78% to 87%) for the screened group and 77% (95% CI 75% to 79%) for the referred group (P ⫽ 0.002) (Fig. 1). On univariate analysis, age at diagnosis, screening status (screened vs. referred), PSA, Gleason score, and clinical stage were all significantly associated with biochemical progression. Table 3 shows the multivariate Cox proportional hazards model for the prediction of biochemical progression. Age at diagnosis, screening status, preoperative PSA, biopsy Gleason score, and clinical stage were all significant independent predictors of cancer progression. Table 4 shows the multivariate Cox proportional hazards model for the prediction of biochemical progression based on pathologic features. Pathologic Gleason score, pathologic stage, and adjuvant radiotherapy were all significant independent predictors of cancer progression; however, screening status did not remain significant when considering pathologic features. The year of diagnosis was not significant and did not change the results of the multivariate model.
0.08
3.4. All-cause and prostate cancer-specific survival
3.3. Cancer progression and multivariate models for predicting cancer progression Overall, 14.9% (69 of 464) of screened patients have had cancer progression compared to 15.6% (407 of 2713) of referred patients (P ⫽ 0.9). However, the median follow-up Table 2 Pathologic characteristics for screened versus referred patients
Gleason sumb 2–6 7 8–10 Pathologic stage pT1/2 R0 pT2 R1, pT3a/b pT3c/N1 Adjuvant radiotherapy a
Armitage chi-square, chi-square, or Wilcoxon rank sum P value shows differences by group. b Gleason sum unavailable for 47 subjects.
Of the 3177 men in the study, 169 (5%) died of any cause, and 19 (1%) died of prostate cancer. All-cause and
468
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
1.0 .9 .8
Survival Distribution
.7 .6 .5
---
Screened Referred
40
50
.4 .3 .2 .1 0.0 0
10
20
30
60
70
80
90
Number of Months Fig. 1. Seven-year biochemical progression-free survival for screened (83%) versus referred (77%) patients (P ⫽ 0.002). There was a 14.9% rate of prostate cancer that recurred in the screened population versus 15.6% in the referred population. (Color version of figure is available online.)
cancer-specific survival were similar between the 2 groups (P ⫽ 0.4 and P ⫽ 0.5, respectively). The 7-year all-cause survival was 95% in the screened group and 92% in the referred group, and the 7-year cancer-specific survival was 99% in both groups.
Table 3 Cox proportional hazards model for prediction of cancer recurrence in 3177 men based on clinical characteristics
Screened versus referred Age at diagnosis PSA at diagnosis Biopsy Gleason score ⱕ6 7 8–10 Clinical stage T1 T2 T3
Adjusted relative risk (95% CI)
P Value
1.5 (1.1–1.9) 1.02ys (1.01–1.03) 1.03 (1.02–1.04)
0.004 0.01 ⬍0.0001
1.0 2.7 (2.2–3.4) 3.9 (2.9–5.3) 1.0 1.9 (1.5–2.3) 2.2 (1.3–3.8)
⬍0.0001 ⬍0.0001 ⬍0.0001 0.004
4. Discussion There is limited evidence concerning whether prostate cancer screening improves treatment outcomes. Because this is a critical issue, many groups have attempted to address this question using various study designs. First, there have been several case-control studies, retrospectively Table 4 Cox proportional hazards model for prediction of cancer recurrence in 3177 men based on pathologic characteristics Adjusted relative risk (95% CI) Screened versus referred Pathologic Gleason score ⱕ6 7 8–10 Pathologic stage pT1/2 R0 pT2 R1,pT3a/b pT3c/N1 Adjuvant radiotherapy
1.2 (0.9–1.6)
P Value 0.1
1.0 1.9 (1.5–2.4) 3.6 (2.7–4.7)
⬍0.0001 ⬍0.0001
1.0 3.0 (2.4–3.7) 6.2 (4.7–8.2) 0.7 (0.5–0.9)
⬍0.0001 ⬍0.0001 0.01
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
comparing men who died of prostate cancer to matched controls for their use of screening. Weinmann et al. [9] compared 769 men who died of prostate cancer between 1997 and 2001 to 929 matched controls within the same health maintenance organization. In a similar manner, Jacobsen et al. [10] compared 173 Olmsted County men who died of prostate cancer between 1976 and 1991 to agematched controls with a similar duration of medical records. Both of these studies found that men who died of prostate cancer were significantly less likely to have had DRE during the 10 years before death than controls during the same time. However, these studies both included a period before the widespread use of PSA, so neither study was able to make conclusions about its relative contribution. Furthermore, such case-control studies are limited by the potential for misclassification bias and the inability to determine causality. A second study design that has been used to evaluate the effects of screening is the analysis of population data. Hankey et al. [2] examined Surveillance, Epidemiology and End Results data from 1973 to 1995. They found several positive trends after the introduction of widespread PSA screening, such as a 17.9% annual decrease in the incidence of distant stage disease and 1.9% annual decrease in prostate cancerspecific mortality in white men, beginning in 1991. This trend of decreasing prostate cancer-specific mortality has continued to the present at a rate of about 4% per year through 2002, and, cumulatively, there has been more than a 20% reduction in the age-specific prostate cancer mortality rate in the United States [11]. Other ecologic studies have yielded conflicting results. Lu-Yao et al. [12] used data from Medicare claims records to compare the cancer-specific mortality rates between 2 geographic areas with the different use of screening. Although Seattle had considerably higher PSA testing rates, the adjusted rate ratio for prostate cancer mortality was similar to that in Connecticut (1.03, 95% CI 0.95– 1.11). Shaw et al. [13] used a model to estimate the decrease in cancer-specific mortality that would be expected if PSA screening had a survival benefit and found that the observed mortality decreases were less dramatic than predicted using the model. It is noteworthy that both of these studies only included men older than age 65, and it is likely that the greatest benefit for early prostate cancer detection would accrue to younger men. Perron et al. [14] used the Pearson product-moment correlation coefficients to compare the prostate cancer incidence from 1989 to 1993 in Quebec, with cancer-specific mortality from 1995 to 1999. Although a substantial decline in mortality was observed in nearly all age groups and regions, they did not find a strong inverse relationship between the incidence and mortality in each group (i.e., those groups with the highest increase in incidence, which they assumed was caused entirely by PSA screening, were not necessarily the same groups to have the highest reduction in prostate cancer mortality 6 years later). However, none of
469
these studies considered differences between the study groups in other baseline characteristics, such as clinical stage, Gleason grade, other risk factors, that are known have a considerable impact on prognosis for prostate cancer. Bartsch et al. [15] reported a significantly higher decrease in prostate cancer mortality after the introduction of free widespread PSA screening in Tyrol, Austria, than in the rest of Austria, where PSA screening was not so widely applied. A likely explanation for these discordant results is the complexity of analyzing incidence and mortality trends at the population level. Both parameters may be affected by the presence or absence of other confounding factors at any given time. Nevertheless, there are measures that can be used to help minimize the effects of confounders. For example, Etzioni et al. [16] generated a computer model in which they used estimations of the lead time associated with prostate cancer screening to help interpret incidence trends. They concluded that screening appears to “overdiagnose” a lower proportion of prostate cancer than was previously thought and that the majority of screendetected cancers would have surfaced clinically during the patient’s lifetime. In general, the reported estimates of the rate of prostate cancer overdiagnosis estimates are exaggerated because they are based largely on older patients, in whom overdiagnosis is more common because of their shorter life expectancy. Although Etzioni et al. [16] did cite a lower rate of overdiagnosis than other studies, their definition of overdiagnosis was different (i.e., cancer that would not have been diagnosed during lifetime) than that used in most other studies (i.e., cancer that would not have caused symptoms during lifetime.) Another method that can help elucidate the effects of prostate cancer screening is a cohort study in which screened and unscreened populations are followed over time to determine if any differences emerge. Using this type of study design, it was recently shown that participants in a formal screening protocol had more favorable clinical stage and biopsy Gleason grade [1]. In the present study, we carried the analysis to the next level and examined more long-term treatment outcomes, including pathologic stage, progression-free survival rate, and cancer-specific mortality rate. We found that the screened group had lower PSA levels at cancer detection, lower Gleason scores at RRP, and a higher proportion of patients with pathologically organconfined disease and clear surgical margins. Not only were the clinical and pathologic characteristics of the screened group more favorable than in the referred group, but also the screened patients had a significantly higher 7-year progression-free survival rate after RRP. This result occurred despite the screened group having substantially longer followup. Improved progression-free survival in the screened group was likely a result of earlier cancer detection. With longer follow-up, we anticipate the difference between the groups would increase, giving screened men an even larger progression-free survival advantage. Some might argue that
470
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
prostate cancer was “overdiagnosed” in both of our treatment cohorts, and probably more in the screened group, and that this may account for some of the high progression-free survival rates. However, all of our patients had a sufficient life expectancy to qualify as candidates for RRP, and it is not possible to determine with certainty that any given cancer will never develop the capacity to cause symptoms or death if left untreated. The low number of deaths in our population to date precludes any meaningful cause-specific or all-cause survival analysis at this time. Nevertheless, because a measurable postoperative PSA and, more specifically, the PSA doubling time have correlated with prostate cancer-specific mortality [17], we anticipate that longer follow-up will reveal mortality differences between the screened and referred patients. In a randomized, controlled trial comparing RRP to watchful waiting for early prostate cancer, Bill-Axelson et al. [18] reported that men treated surgically had a significant reduction in cancer-specific mortality at 10 years (P ⫽ 0.04), although the absolute mortality reduction was modest. However, the majority of the patients in their study had clinical stage T2 prostate cancer that was not detected through screening. Thus, it remains unclear how these results will compare to screen-detected prostate cancer. Several limitations of our study deserve mention. First, it is possible that the difference in progression-free survival between screened and referred patients was affected by lead time or length-time bias. Lead-time bias refers to the possibility that screening may appear to improve outcomes by detecting cancers earlier in the clinical evolution, when, in reality, the ultimate outcome is unchanged. Unfortunately, it is only possible to truly address the effect of lead-time bias in a randomized controlled trial. Length-time bias refers to the possibility that indolent cancers remain in the asymptomatic phase for a longer time, thus making it temporally more likely that they will be detected through screening than a rapidly progressive aggressive tumor. Although it is certainly true that the application of widespread screening has led to a stage migration, there is substantial evidence that the great majority of cancers thus detected have the features commonly associated with clinically significant tumors [19,20]. Nevertheless, evidence from randomized controlled trials is needed to confirm that our finding of a 7-year progression-free survival advantage is a result of screening. Another limitation of our study is the potential for selection bias. Because the cohorts were not randomized, it is possible that the referred patients were sent by their physicians for reasons that differentiate them from the general group of community men who participated in the screening study. Furthermore, misclassification bias is also a possibility because the rate of screening among the referred patients is unknown. Because of the wide availability of prostate cancer screening in the United States, it is likely that some of the referred patients were screened with periodic PSA measurements and/or DRE by their referring physicians. This shortcoming is not unique to our study; even the
ongoing randomized trials are confounded by a high percentage of “opportunistic screening” among controls [1]. By including such screened patients in the referred or control group, a true comparison of screened and unscreened populations cannot be determined. However, this limitation would underestimate any survival advantage, making our estimates of the difference between groups more conservative. We believe that in daily medical practice, physicians’ recommendations and patients’ inclinations toward screening are highly variable. In the first round of the Prostate, Lung, Colorectal and Ovarian Screening Trial, only 41% of men with a PSA level higher than the biopsy threshold underwent the recommended biopsy within 1 year [21]. The proportion of men complying with the biopsy recommendation was significantly different in men with a PSA level of 4 –7 ng/ml versus those with a PSA level of 7–10 ng/ml or higher. This result highlights the difficulty of designing a trial that will prove that screening for prostate cancer saves lives. Nevertheless, as the body of evidence from studies of all types continues to grow, we believe that the widespread use of routine screening will result in lower rates of cancer progression in men with clinically significant prostate cancer. 5. Conclusions We compared the pathologic tumor features and treatment outcomes after RRP between men whose cancer was detected as part of a screening study and a contemporary referred patient cohort. At 7 years after RRP, screened men had a progression-free survival advantage over referred patients (83% vs. 77%, P ⫽ 0.002). On multivariate analysis, including age, PSA, Gleason score, and clinical stage, the patient’s screening status (screened vs. referred) remained a significant independent predictor of biochemical progression. Additional follow-up is necessary to elucidate further the long-term survival benefits associated with screening for prostate cancer. References [1] van der Cruijsen-Koeter IW, Vis AN, Roobol MJ, et al. Comparison of screen detected and clinically diagnosed prostate cancer in the European randomized study of screening for prostate cancer, section Rotterdam. J Urol 2005;174:121–5. [2] Hankey BF, Feuer EJ, Clegg LX, et al. Cancer surveillance series: Interpreting trends in prostate cancer–Part I: Evidence of the effects of screening in recent prostate cancer incidence, mortality, and survival rates. J Natl Cancer Inst 1999;91:1017–24. [3] Tarone RE, Chu KC, Brawley OW. Implications of stage-specific survival rates in assessing recent declines in prostate cancer mortality rates. Epidemiology 2000;11:167–70. [4] Paquette EL, Sun L, Paquette LR, et al. Improved prostate cancerspecific survival and other disease parameters: Impact of prostatespecific antigen testing. Urology 2002;60:756 –9. [5] Jemal A, Murray T, Ward E, et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55:10 –30.
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471 [6] Nicholson PW, Harland SJ. Survival prospects after screen-detection of prostate cancer. BJU Int 2002;90:686 –93. [7] Walsh PC, Lepor H, Eggleston JC. Radical prostatectomy with preservation of sexual function: anatomical and pathological considerations. Prostate 1983;4:473– 85. [8] Smith DS, Humphrey PA, Catalona WJ. The early detection of prostate carcinoma with prostate specific antigen: The Washington University experience. Cancer 1997;80:1852– 6. [9] Weinmann S, Richert-Boe KE, Van Den Eeden SK, et al. Screening by prostate-specific antigen and digital rectal examination in relation to prostate cancer mortality: A case-control study. Epidemiology 2005;16:367–76. [10] Jacobsen SJ, Bergstralh EJ, Katusic SK, et al. Screening digital rectal examination and prostate cancer mortality: A population-based casecontrol study. Urology 1998;52:173–9. [11] American Cancer Society. Cancer facts and figures 2005. Atlanta, (GA): American Cancer Society, 2005. [12] Lu-Yao G, Albertsen PC, Stanford JL, et al. Natural experiment examining impact of aggressive screening and treatment on prostate cancer mortality in two fixed cohorts from Seattle area and Connecticut. BMJ 2002;325:740. [13] Shaw PA, Etzioni R, Zeliadt SB, et al. An ecologic study of prostatespecific antigen screening and prostate cancer mortality in nine geographic areas of the United States. Am J Epidemiol 2004;160:1059 – 69.
471
[14] Perron L, Moore L, Bairati I, et al. PSA screening and prostate cancer mortality. CMAJ 2002;166:586 –91. [15] Bartsch G, Horninger W, Klocker H, et al. Prostate cancer mortality after introduction of prostate-specific antigen mass screening in the Federal State of Tyrol, Austria. Urology 2001;58:417–24. [16] Etzioni R, Penson DF, Legler JM, et al. Overdiagnosis due to prostate-specific antigen screening: Lessons from U.S. prostate cancer incidence trends. J Natl Cancer Inst 2002;94:981–90. [17] D’Amico AV, Moul JW, Carroll PR, et al. Surrogate end point for prostate cancer-specific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst 2003;95:1376 – 83. [18] Bill-Axelson A, Holmberg L, Ruutu M, et al. Radical prostatectomy versus watchful waiting in early prostate cancer. N Engl J Med 2005;352:1977– 84. [19] Ohori M, Wheeler TM, Dunn JK, et al. The pathological features and prognosis of prostate cancer detectable with current diagnostic tests. J Urol 1994;152:1714 –20. [20] Krumholtz JS, Carvalhal GF, Ramos CG, et al. Prostate-specific antigen cutoff of 2.6 ng/mL for prostate cancer screening is associated with favorable pathologic tumor features. Urology 2002;60:469 –73. [21] Pinsky PF, Andriole GL, Kramer BS, et al. Prostate biopsy following a positive screen in the prostate, lung, colorectal and ovarian cancer screening trial. J Urol 2005;173:746 –50.
Original article
Survival results in patients with screen-detected prostate cancer versus physician-referred patients treated with radical prostatectomy: Early results夡 Kimberly A. Roehl, M.P.H.a, Scott E. Eggener, M.D.b, Stacy Loeb, M.D.b, Norm D. Smith, M.D.b, Jo Ann V. Antenor, M.P.H.c, William J. Catalona, M.D.b,* a
b
Department of Psychiatry, Washington University School of Medicine, St. Louis, MO 63105, USA Department of Urology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USA c Department of Neurology, Washington University School of Medicine, St. Louis, MO 63105, USA Received 17 August 2005; received in revised form 29 November 2005; accepted 30 November 2005
Abstract Objective: Screening using a standardized protocol may improve outcomes of patients undergoing treatment for prostate cancer. We compared the 7- year progression-free survival rates after radical retropubic prostatectomy in patients whose prostate cancer was detected through a formal screening program with those of patients referred for treatment by other physicians who did not use a standardized screening/referral protocol. Methods: A single surgeon (W.J.C.) performed radical retropubic prostatectomy in 3,177 consecutive patients between 1989 and 2003. Of these patients, 464 had cancer detected in a screening study, and 2,713 were referred from outside institutions. We compared the screened and referred cohorts for age at surgery, clinical stage, pathologic stage, Gleason sum, preoperative prostate-specific antigen (PSA) levels, and adjuvant radiation therapy. Kaplan-Meier product limit estimates were used to calculate 7-year progression-free probabilities, and Cox proportional hazards models were used to determine the clinical and pathologic parameters associated with cancer progression in each group. Results: The overall 7-year progression-free survival rates were 83% for the screened patients compared with 77% for the referred patients (P ⫽ 0.002). Preoperative PSA, Gleason sum, clinical stage, pathologic stage, and adjuvant radiotherapy were all significantly associated with cancer progression. There was a significantly higher proportion of referred patients with a preoperative PSA ⱖ10, Gleason sum ⱖ7, and nonorgan-confined disease. Conclusions: Patients with screened-detected prostate cancer have more favorable clinical and pathologic features, and 7-year progression-free survival rates than referred patients. On multivariate analysis, including other clinical variables, screening status was a significant independent predictor of biochemical outcome. © 2006 Elsevier Inc. All rights reserved. Keywords: Prostate; Prostatic neoplasms; Prostate-specific antigen; Screening; Radical prostatectomy
1. Introduction The principal goal of an early detection program for prostate cancer is to detect clinically relevant tumors while they are still curable. To examine whether current screening protocols might accomplish this objective, we compared patients whose cancer was diagnosed through a prostate夡 Supported by a grant from Beckman Coulter Incorporated, Fullerton, CA, and by the Urological Research Foundation. * Corresponding author. Tel.: ⫹1-312-695-4471; fax: ⫹1-312-6951482. E-mail address: [email protected] (W.J. Catalona).
1078-1439/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.urolonc.2005.11.039
specific antigen (PSA) and digital rectal examination (DRE)-based screening study to those whose cancer was detected during routine clinical practice. van der CruijsenKoeter et al. [1] recently reported that both clinical stage and Gleason grade were significantly more favorable in the screening arm of the European Randomized Study of Screening for Prostate Cancer; however, they did not address the association between screening status and subsequent treatment outcomes. Pending the results of randomized clinical trials, there is increasing evidence suggesting that screening leads to improved outcomes [2– 6]. Based on cases diagnosed between 1992 and 1995, approximately 90% of new prostate cancer
466
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
cases are expected to be local or regional at diagnosis. Furthermore, prostate cancer mortality rates continue to decline in the Surveillance, Epidemiology and End Results database. In the present study, we sought to determine whether screening for prostate cancer might be responsible, at least in part, for these encouraging trends. For this purpose, we compared pathologic tumor features, progressionfree, cancer-specific, and all-cause survival rates after radical retropubic prostatectomy (RRP) between men who participated in a standardized screening program with those in referred patients who did not. In addition, we included multivariate models to distinguish which preoperative clinical or pathologic parameters were independent predictors of cancer progression.
2. Materials and methods From January 1989 to February 2003, a single surgeon (W.J.C.) performed RRP in 3177 consecutive men using a standardized technique [7]. These patients comprised 2 separate cohorts. The first included 464 (15%) whose prostate cancer was detected through a screening study. The study protocol involved PSA testing and DRE at 6 or 12-month intervals, as previously described [8]. A biopsy was recommended for a suspicious DRE or PSA level higher than 4.0 ng/ml (until May 1995) or 2.5 ng/ml (after May 1995). The second cohort consisted of 2713 (85%) men who were diagnosed with prostate cancer by other physicians and either requested or were referred for RRP. Indication for biopsy in these men was not standardized and may have included PSA, DRE, or evaluation for other problems. Clinical stage was classified as either localized (stage cT1 or cT2) or advanced (stage ⱖcT3). Pathologic stage was considered organ-confined if the cancer was confined to the prostate with clear (noncancerous) surgical margins (stage pT1 R0 or pT2 R0). Any specimen with extraprostatic tumor extension, positive (cancerous) surgical margins, seminal vesicle invasion, or lymph node metastases was classified as pathologically advanced disease (stage pT2 R1, pT3a/b, N1 or higher). Postoperatively, both cohorts underwent PSA testing every 6 months and DRE yearly. Biochemical progression was defined as a detectable PSA level (⬎0.2 ng/ml) confirmed by repeat measurements.
known to be living at the most recent date of follow-up and used the log-rank test to compare the strata. For multivariable analyses, we used the Cox proportional hazards model to determine whether study entry status (screened or referred) was associated with risk of cancer progression, accounting for pretreatment variables, such as age at diagnosis (as a continuous variable), preoperative PSA, biopsy Gleason score (ⱕ6 vs. 7 vs. 8 –10), and clinical stage (T1 vs. T2 vs. T3). Adjusted relative risks and their respective 95% confidence intervals (CIs) and P values are reported for the full model. We also sought to determine whether study entry status was associated with risk of cancer progression, accounting for pathologic variables, such as surgical Gleason score (ⱕ6 vs. 7 vs. 8 –10), pathologic stage (pT1/2 R0 vs. pT2 R1 / T3a/b vs. pT3c/N1), and use of adjuvant radiotherapy. All statistical analyses were performed using Statistical Analysis Systems (version 8.2; SAS Institute, Inc., Cary, NC) and SPSS 10.0 for Windows (SPSS Inc., Chicago, IL).
3. Results 3.1. Clinical characteristics Table 1 shows the demographics of our study population. Mean age at diagnosis was higher in the screened group than the referred group (64.3 ⫾ 6.5 vs. 60.4 ⫾ 7.3 years, P ⬍ 0.0001). Race was also significantly different between groups. Although 94% of the patients were white, there was a higher proportion of black, Hispanic, or Asian patients in the referred than in the screened population (7% vs. 5%, respectively, P ⫽ 0.005). The 2 cohorts also differed in other clinical parameters. The median preoperative PSA was lower in the screened population (5.1 ⫾ 7.0 vs. 6.0 ⫾ 6.8 ng/ml, P ⬍ 0.0001). There were more screened patients diagnosed with a PSA value of 2.6 – 4.0 ng/ml (24% vs. 9%, P ⫽ 0.001) and significantly fewer screened patients with a PSA value of ⬎10.0 ng/ml (13% vs. 18%, P ⫽ 0.001). There were more screened patients with a biopsy Gleason sum of ⱕ6 (85% vs. 79%, P ⫽ 0.02). Although there were more PSA-detected cancers (clinical T1c) in the referred patients (57% vs. 51%, P ⫽ 0.001), there was no significant difference in the proportion of clinically localized cancers between the groups.
2.1. Statistical analysis 3.2. Pathologic characteristics To compare clinical and pathologic parameters between groups, chi-square, Armitage chi-square, Fisher exact test, or t-tests were used for normally distributed, binomial proportions or categorical data, and the Wilcoxon rank sum test was used for nonparametric continuous variables. KaplanMeier product limit estimates were used to compare cancer progression and mortality rates by screening status. For both cancer-specific and overall survival, we censored patients
There were significantly fewer patients with Gleason sum ⱖ7 in the screened group (26%) than in the referred group (40%) (P ⬍ 0.0001) (Table 2). Patients in the screened group were more likely to have organ-confined disease than those in the referred group (75% vs. 68%, P ⫽ 0.02). However, there was no significant difference between the groups in the proportion of men electing adjuvant ra-
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
467
Table 1 Clinical characteristics for screened versus referred patients No. patents screened (%) Age at diagnosis (yrs) ⬍50 50–59 60–69 ⱖ70 Race White Black Other Preoperative PSAb ⬍2.6 2.6–4.0 4.1–10.0 ⬎10.0 Biopsy Gleason sumc 2–6 7 8–10 Clinical stage cT1a/b cT1c cT2 cT3 Median follow-up (range)
P Valuea
No. patents referred (%)
8 (2) 102 (22) 251 (54) 103 (22)
195 (7) 1013 (37) 1204 (45) 301 (11)
⬍0.0001
443 (95) 19 (4) 2 (1)
2532 (93) 76 (3) 104 (4)
0.005
38 (8) 108 (24) 253 (55) 58 (13)
168 (6) 239 (9) 1773 (66) 491 (19)
⬍0.0001
388 (85) 48 (11) 20 (4)
2084 (79) 458 (17) 111 (4)
0.002
50 (2) 1533 (57) 1094 (40) 30 (1) 45 mos (0–158)
0.02
5 (1) 236 (51) 220 (47) 2 (1) 79.5 mos (0–160)
⬍0.0001
a
Armitage chi-square P values show differences by group. Preoperative PSA unavailable for 49 subjects. c Biopsy Gleason sum unavailable for 68 subjects. b
diotherapy (4% in the screened group vs. 7% in the referred group, P ⫽ 0.08). Pathologic stage was significantly associated with biochemical progression (P ⬍ 0.0001).
No. patents screened (%)
No. patents referred (%)
P Valuea
337 (74) 96 (21) 25 (5)
1597 (60) 898 (34) 177 (7)
⬍0.0001
346 (75) 95 (21) 21 (4) 20 (4)
1844 (68) 714 (27) 135 (5) 178 (7)
0.02
time was significantly longer in the screened patients (79.5 ⫾ 44.5 vs. 45 ⫾ 40.2 months, P ⬍ 0.0001). The 7-year progression-free survival rates were 83% (95% CI 78% to 87%) for the screened group and 77% (95% CI 75% to 79%) for the referred group (P ⫽ 0.002) (Fig. 1). On univariate analysis, age at diagnosis, screening status (screened vs. referred), PSA, Gleason score, and clinical stage were all significantly associated with biochemical progression. Table 3 shows the multivariate Cox proportional hazards model for the prediction of biochemical progression. Age at diagnosis, screening status, preoperative PSA, biopsy Gleason score, and clinical stage were all significant independent predictors of cancer progression. Table 4 shows the multivariate Cox proportional hazards model for the prediction of biochemical progression based on pathologic features. Pathologic Gleason score, pathologic stage, and adjuvant radiotherapy were all significant independent predictors of cancer progression; however, screening status did not remain significant when considering pathologic features. The year of diagnosis was not significant and did not change the results of the multivariate model.
0.08
3.4. All-cause and prostate cancer-specific survival
3.3. Cancer progression and multivariate models for predicting cancer progression Overall, 14.9% (69 of 464) of screened patients have had cancer progression compared to 15.6% (407 of 2713) of referred patients (P ⫽ 0.9). However, the median follow-up Table 2 Pathologic characteristics for screened versus referred patients
Gleason sumb 2–6 7 8–10 Pathologic stage pT1/2 R0 pT2 R1, pT3a/b pT3c/N1 Adjuvant radiotherapy a
Armitage chi-square, chi-square, or Wilcoxon rank sum P value shows differences by group. b Gleason sum unavailable for 47 subjects.
Of the 3177 men in the study, 169 (5%) died of any cause, and 19 (1%) died of prostate cancer. All-cause and
468
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
1.0 .9 .8
Survival Distribution
.7 .6 .5
---
Screened Referred
40
50
.4 .3 .2 .1 0.0 0
10
20
30
60
70
80
90
Number of Months Fig. 1. Seven-year biochemical progression-free survival for screened (83%) versus referred (77%) patients (P ⫽ 0.002). There was a 14.9% rate of prostate cancer that recurred in the screened population versus 15.6% in the referred population. (Color version of figure is available online.)
cancer-specific survival were similar between the 2 groups (P ⫽ 0.4 and P ⫽ 0.5, respectively). The 7-year all-cause survival was 95% in the screened group and 92% in the referred group, and the 7-year cancer-specific survival was 99% in both groups.
Table 3 Cox proportional hazards model for prediction of cancer recurrence in 3177 men based on clinical characteristics
Screened versus referred Age at diagnosis PSA at diagnosis Biopsy Gleason score ⱕ6 7 8–10 Clinical stage T1 T2 T3
Adjusted relative risk (95% CI)
P Value
1.5 (1.1–1.9) 1.02ys (1.01–1.03) 1.03 (1.02–1.04)
0.004 0.01 ⬍0.0001
1.0 2.7 (2.2–3.4) 3.9 (2.9–5.3) 1.0 1.9 (1.5–2.3) 2.2 (1.3–3.8)
⬍0.0001 ⬍0.0001 ⬍0.0001 0.004
4. Discussion There is limited evidence concerning whether prostate cancer screening improves treatment outcomes. Because this is a critical issue, many groups have attempted to address this question using various study designs. First, there have been several case-control studies, retrospectively Table 4 Cox proportional hazards model for prediction of cancer recurrence in 3177 men based on pathologic characteristics Adjusted relative risk (95% CI) Screened versus referred Pathologic Gleason score ⱕ6 7 8–10 Pathologic stage pT1/2 R0 pT2 R1,pT3a/b pT3c/N1 Adjuvant radiotherapy
1.2 (0.9–1.6)
P Value 0.1
1.0 1.9 (1.5–2.4) 3.6 (2.7–4.7)
⬍0.0001 ⬍0.0001
1.0 3.0 (2.4–3.7) 6.2 (4.7–8.2) 0.7 (0.5–0.9)
⬍0.0001 ⬍0.0001 0.01
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
comparing men who died of prostate cancer to matched controls for their use of screening. Weinmann et al. [9] compared 769 men who died of prostate cancer between 1997 and 2001 to 929 matched controls within the same health maintenance organization. In a similar manner, Jacobsen et al. [10] compared 173 Olmsted County men who died of prostate cancer between 1976 and 1991 to agematched controls with a similar duration of medical records. Both of these studies found that men who died of prostate cancer were significantly less likely to have had DRE during the 10 years before death than controls during the same time. However, these studies both included a period before the widespread use of PSA, so neither study was able to make conclusions about its relative contribution. Furthermore, such case-control studies are limited by the potential for misclassification bias and the inability to determine causality. A second study design that has been used to evaluate the effects of screening is the analysis of population data. Hankey et al. [2] examined Surveillance, Epidemiology and End Results data from 1973 to 1995. They found several positive trends after the introduction of widespread PSA screening, such as a 17.9% annual decrease in the incidence of distant stage disease and 1.9% annual decrease in prostate cancerspecific mortality in white men, beginning in 1991. This trend of decreasing prostate cancer-specific mortality has continued to the present at a rate of about 4% per year through 2002, and, cumulatively, there has been more than a 20% reduction in the age-specific prostate cancer mortality rate in the United States [11]. Other ecologic studies have yielded conflicting results. Lu-Yao et al. [12] used data from Medicare claims records to compare the cancer-specific mortality rates between 2 geographic areas with the different use of screening. Although Seattle had considerably higher PSA testing rates, the adjusted rate ratio for prostate cancer mortality was similar to that in Connecticut (1.03, 95% CI 0.95– 1.11). Shaw et al. [13] used a model to estimate the decrease in cancer-specific mortality that would be expected if PSA screening had a survival benefit and found that the observed mortality decreases were less dramatic than predicted using the model. It is noteworthy that both of these studies only included men older than age 65, and it is likely that the greatest benefit for early prostate cancer detection would accrue to younger men. Perron et al. [14] used the Pearson product-moment correlation coefficients to compare the prostate cancer incidence from 1989 to 1993 in Quebec, with cancer-specific mortality from 1995 to 1999. Although a substantial decline in mortality was observed in nearly all age groups and regions, they did not find a strong inverse relationship between the incidence and mortality in each group (i.e., those groups with the highest increase in incidence, which they assumed was caused entirely by PSA screening, were not necessarily the same groups to have the highest reduction in prostate cancer mortality 6 years later). However, none of
469
these studies considered differences between the study groups in other baseline characteristics, such as clinical stage, Gleason grade, other risk factors, that are known have a considerable impact on prognosis for prostate cancer. Bartsch et al. [15] reported a significantly higher decrease in prostate cancer mortality after the introduction of free widespread PSA screening in Tyrol, Austria, than in the rest of Austria, where PSA screening was not so widely applied. A likely explanation for these discordant results is the complexity of analyzing incidence and mortality trends at the population level. Both parameters may be affected by the presence or absence of other confounding factors at any given time. Nevertheless, there are measures that can be used to help minimize the effects of confounders. For example, Etzioni et al. [16] generated a computer model in which they used estimations of the lead time associated with prostate cancer screening to help interpret incidence trends. They concluded that screening appears to “overdiagnose” a lower proportion of prostate cancer than was previously thought and that the majority of screendetected cancers would have surfaced clinically during the patient’s lifetime. In general, the reported estimates of the rate of prostate cancer overdiagnosis estimates are exaggerated because they are based largely on older patients, in whom overdiagnosis is more common because of their shorter life expectancy. Although Etzioni et al. [16] did cite a lower rate of overdiagnosis than other studies, their definition of overdiagnosis was different (i.e., cancer that would not have been diagnosed during lifetime) than that used in most other studies (i.e., cancer that would not have caused symptoms during lifetime.) Another method that can help elucidate the effects of prostate cancer screening is a cohort study in which screened and unscreened populations are followed over time to determine if any differences emerge. Using this type of study design, it was recently shown that participants in a formal screening protocol had more favorable clinical stage and biopsy Gleason grade [1]. In the present study, we carried the analysis to the next level and examined more long-term treatment outcomes, including pathologic stage, progression-free survival rate, and cancer-specific mortality rate. We found that the screened group had lower PSA levels at cancer detection, lower Gleason scores at RRP, and a higher proportion of patients with pathologically organconfined disease and clear surgical margins. Not only were the clinical and pathologic characteristics of the screened group more favorable than in the referred group, but also the screened patients had a significantly higher 7-year progression-free survival rate after RRP. This result occurred despite the screened group having substantially longer followup. Improved progression-free survival in the screened group was likely a result of earlier cancer detection. With longer follow-up, we anticipate the difference between the groups would increase, giving screened men an even larger progression-free survival advantage. Some might argue that
470
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471
prostate cancer was “overdiagnosed” in both of our treatment cohorts, and probably more in the screened group, and that this may account for some of the high progression-free survival rates. However, all of our patients had a sufficient life expectancy to qualify as candidates for RRP, and it is not possible to determine with certainty that any given cancer will never develop the capacity to cause symptoms or death if left untreated. The low number of deaths in our population to date precludes any meaningful cause-specific or all-cause survival analysis at this time. Nevertheless, because a measurable postoperative PSA and, more specifically, the PSA doubling time have correlated with prostate cancer-specific mortality [17], we anticipate that longer follow-up will reveal mortality differences between the screened and referred patients. In a randomized, controlled trial comparing RRP to watchful waiting for early prostate cancer, Bill-Axelson et al. [18] reported that men treated surgically had a significant reduction in cancer-specific mortality at 10 years (P ⫽ 0.04), although the absolute mortality reduction was modest. However, the majority of the patients in their study had clinical stage T2 prostate cancer that was not detected through screening. Thus, it remains unclear how these results will compare to screen-detected prostate cancer. Several limitations of our study deserve mention. First, it is possible that the difference in progression-free survival between screened and referred patients was affected by lead time or length-time bias. Lead-time bias refers to the possibility that screening may appear to improve outcomes by detecting cancers earlier in the clinical evolution, when, in reality, the ultimate outcome is unchanged. Unfortunately, it is only possible to truly address the effect of lead-time bias in a randomized controlled trial. Length-time bias refers to the possibility that indolent cancers remain in the asymptomatic phase for a longer time, thus making it temporally more likely that they will be detected through screening than a rapidly progressive aggressive tumor. Although it is certainly true that the application of widespread screening has led to a stage migration, there is substantial evidence that the great majority of cancers thus detected have the features commonly associated with clinically significant tumors [19,20]. Nevertheless, evidence from randomized controlled trials is needed to confirm that our finding of a 7-year progression-free survival advantage is a result of screening. Another limitation of our study is the potential for selection bias. Because the cohorts were not randomized, it is possible that the referred patients were sent by their physicians for reasons that differentiate them from the general group of community men who participated in the screening study. Furthermore, misclassification bias is also a possibility because the rate of screening among the referred patients is unknown. Because of the wide availability of prostate cancer screening in the United States, it is likely that some of the referred patients were screened with periodic PSA measurements and/or DRE by their referring physicians. This shortcoming is not unique to our study; even the
ongoing randomized trials are confounded by a high percentage of “opportunistic screening” among controls [1]. By including such screened patients in the referred or control group, a true comparison of screened and unscreened populations cannot be determined. However, this limitation would underestimate any survival advantage, making our estimates of the difference between groups more conservative. We believe that in daily medical practice, physicians’ recommendations and patients’ inclinations toward screening are highly variable. In the first round of the Prostate, Lung, Colorectal and Ovarian Screening Trial, only 41% of men with a PSA level higher than the biopsy threshold underwent the recommended biopsy within 1 year [21]. The proportion of men complying with the biopsy recommendation was significantly different in men with a PSA level of 4 –7 ng/ml versus those with a PSA level of 7–10 ng/ml or higher. This result highlights the difficulty of designing a trial that will prove that screening for prostate cancer saves lives. Nevertheless, as the body of evidence from studies of all types continues to grow, we believe that the widespread use of routine screening will result in lower rates of cancer progression in men with clinically significant prostate cancer. 5. Conclusions We compared the pathologic tumor features and treatment outcomes after RRP between men whose cancer was detected as part of a screening study and a contemporary referred patient cohort. At 7 years after RRP, screened men had a progression-free survival advantage over referred patients (83% vs. 77%, P ⫽ 0.002). On multivariate analysis, including age, PSA, Gleason score, and clinical stage, the patient’s screening status (screened vs. referred) remained a significant independent predictor of biochemical progression. Additional follow-up is necessary to elucidate further the long-term survival benefits associated with screening for prostate cancer. References [1] van der Cruijsen-Koeter IW, Vis AN, Roobol MJ, et al. Comparison of screen detected and clinically diagnosed prostate cancer in the European randomized study of screening for prostate cancer, section Rotterdam. J Urol 2005;174:121–5. [2] Hankey BF, Feuer EJ, Clegg LX, et al. Cancer surveillance series: Interpreting trends in prostate cancer–Part I: Evidence of the effects of screening in recent prostate cancer incidence, mortality, and survival rates. J Natl Cancer Inst 1999;91:1017–24. [3] Tarone RE, Chu KC, Brawley OW. Implications of stage-specific survival rates in assessing recent declines in prostate cancer mortality rates. Epidemiology 2000;11:167–70. [4] Paquette EL, Sun L, Paquette LR, et al. Improved prostate cancerspecific survival and other disease parameters: Impact of prostatespecific antigen testing. Urology 2002;60:756 –9. [5] Jemal A, Murray T, Ward E, et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55:10 –30.
K.A. Roehl et al. / Urologic Oncology: Seminars and Original Investigations 24 (2006) 465– 471 [6] Nicholson PW, Harland SJ. Survival prospects after screen-detection of prostate cancer. BJU Int 2002;90:686 –93. [7] Walsh PC, Lepor H, Eggleston JC. Radical prostatectomy with preservation of sexual function: anatomical and pathological considerations. Prostate 1983;4:473– 85. [8] Smith DS, Humphrey PA, Catalona WJ. The early detection of prostate carcinoma with prostate specific antigen: The Washington University experience. Cancer 1997;80:1852– 6. [9] Weinmann S, Richert-Boe KE, Van Den Eeden SK, et al. Screening by prostate-specific antigen and digital rectal examination in relation to prostate cancer mortality: A case-control study. Epidemiology 2005;16:367–76. [10] Jacobsen SJ, Bergstralh EJ, Katusic SK, et al. Screening digital rectal examination and prostate cancer mortality: A population-based casecontrol study. Urology 1998;52:173–9. [11] American Cancer Society. Cancer facts and figures 2005. Atlanta, (GA): American Cancer Society, 2005. [12] Lu-Yao G, Albertsen PC, Stanford JL, et al. Natural experiment examining impact of aggressive screening and treatment on prostate cancer mortality in two fixed cohorts from Seattle area and Connecticut. BMJ 2002;325:740. [13] Shaw PA, Etzioni R, Zeliadt SB, et al. An ecologic study of prostatespecific antigen screening and prostate cancer mortality in nine geographic areas of the United States. Am J Epidemiol 2004;160:1059 – 69.
471
[14] Perron L, Moore L, Bairati I, et al. PSA screening and prostate cancer mortality. CMAJ 2002;166:586 –91. [15] Bartsch G, Horninger W, Klocker H, et al. Prostate cancer mortality after introduction of prostate-specific antigen mass screening in the Federal State of Tyrol, Austria. Urology 2001;58:417–24. [16] Etzioni R, Penson DF, Legler JM, et al. Overdiagnosis due to prostate-specific antigen screening: Lessons from U.S. prostate cancer incidence trends. J Natl Cancer Inst 2002;94:981–90. [17] D’Amico AV, Moul JW, Carroll PR, et al. Surrogate end point for prostate cancer-specific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst 2003;95:1376 – 83. [18] Bill-Axelson A, Holmberg L, Ruutu M, et al. Radical prostatectomy versus watchful waiting in early prostate cancer. N Engl J Med 2005;352:1977– 84. [19] Ohori M, Wheeler TM, Dunn JK, et al. The pathological features and prognosis of prostate cancer detectable with current diagnostic tests. J Urol 1994;152:1714 –20. [20] Krumholtz JS, Carvalhal GF, Ramos CG, et al. Prostate-specific antigen cutoff of 2.6 ng/mL for prostate cancer screening is associated with favorable pathologic tumor features. Urology 2002;60:469 –73. [21] Pinsky PF, Andriole GL, Kramer BS, et al. Prostate biopsy following a positive screen in the prostate, lung, colorectal and ovarian cancer screening trial. J Urol 2005;173:746 –50.