- Email: [email protected]
Estimating the impact on prostate cancer mortality of incorporating prostate-specific antigen testing into screening
ADULT UROLOGY CME ARTICLE
ESTIMATING THE IMPACT ON PROSTATE CANCER MORTALITY OF INCORPORATING PROSTATE-SPECIFIC ANTIGEN TESTING INTO SCREENING ANTHONY V. D’AMICO, RICHARD WHITTINGTON, S. BRUCE MALKOWICZ, ANDREW A. RENSHAW, JOHN E. TOMASZEWSKI, CHRISTY BENTLEY, DELRAY SCHULTZ, SEAN ROCHA, ALAN WEIN, AND JEROME P. RICHIE
ABSTRACT Objectives. Whether early detection using prostate-specific antigen (PSA) and digital rectal examination (DRE) compared with DRE alone will reduce prostate cancer mortality awaits the results of ongoing prospective randomized trials. However, the impact that early detection could have on prostate cancerspecific survival can be estimated by assuming that PSA failure after radical prostatectomy (RP) will translate into death from prostate cancer. Methods. The study population consisted of 1274 men with clinically localized prostate cancer who underwent RP in Boston, Massachusetts or Philadelphia, Pennsylvania between 1989 and 2000 and had a preoperative PSA level greater than 4 but not more than 10 ng/mL. The primary endpoint was actuarial freedom from PSA failure (defined as PSA outcome). Results. The relative risk of PSA failure after RP for patients diagnosed with a PSA of greater than 4 to 5, 5 to 6, 6 to 7, or 7 to 8 ng/mL compared with greater than 8 up to 10 ng/mL was 0.3 (95% confidence interval [CI] 0.2 to 0.5), 0.5 (95% CI 0.4 to 0.8), 0.6 (95% CI 0.4 to 0.9), or 0.9 (95% CI 0.6 to 1.3), respectively. On the basis of the estimates of the 5-year PSA outcome, patients with a biopsy Gleason score of 5 or 6 (781 of 1274; 61%) consistently benefited from RP performed when the PSA at diagnosis was greater than 4 to 7 ng/mL compared with greater than 8 to 10 ng/mL (93% versus 78%, P ⬍0.0001). A benefit to early detection was not found for the vast majority (266 of 312; 88%) of patients who had a biopsy Gleason score of 7 or higher. Conclusions. Early detection using both PSA and DRE-based screening may benefit men who present with biopsy Gleason score 5 or 6 prostate cancer and a PSA level greater than 4 to 7 ng/mL compared with greater than 8 up to 10 ng/mL. This finding awaits validation from ongoing prospective randomized trials. UROLOGY 58: 406–410, 2001. © 2001, Elsevier Science Inc.
T
he preoperative prostate-specific antigen (PSA) level has been shown to predict the time to postoperative PSA failure.1,2 The time to postoperative PSA failure has been shown to predict for the time to distant failure, which in turn predicts From the Department of Radiation Oncology, Brigham and Women’s Hospital and Dana Farber Cancer Institute, Boston, Massachusetts; Departments of Radiation Oncology, Urology, and Pathology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania; Department of Mathematics, University of Millersville, Millersville, Pennsylvania; and Departments of Pathology and Urology, Brigham and Women’s Hospital, Boston, Massachusetts Reprint requests: Anthony V. D’Amico, M.D., Ph.D., Department of Radiation Oncology, Brigham and Women’s Hospital, 75 Francis Street, L-2 Level, Boston, MA 02215 Submitted: March 15, 2001, accepted: April 19, 2001
406
© 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
for the time to death from metastatic prostate cancer.3 Taken together, these associations suggest that a higher PSA level at diagnosis would lead to lower rates of cancer control and possibly an increased risk of dying of metastatic prostate cancer. However, only the results of ongoing prospective randomized screening studies4 – 6 evaluating the use of PSA and digital rectal examination (DRE) compared with DRE alone for screening will determine whether a reduction in prostate cancer mortality will be achieved. The goal of this study was to estimate the possible reduction in prostate cancer mortality that could be attributed to early detection through the use of both PSA and DRE by using PSA failure after RP as a surrogate for death from prostate cancer. In 0090-4295/01/$20.00 PII S0090-4295(01)01204-3
TABLE I. Pretreatment clinical characteristics of 1274 study patients stratified by the prostatespecific antigen level at diagnosis PSA PSA PSA PSA PSA >4–5 ng/mL P >5–6 ng/mL P >6–7 ng/mL P >7–8 ng/mL P >8–10 ng/mL Clinical Characteristic (n ⴝ 237) Value (n ⴝ 306) Value (n ⴝ 262) Value (n ⴝ 185) Value (n ⴝ 284) Biopsy Gleason 2–4 Biopsy Gleason 5–6 Biopsy Gleason 7 Biopsy Gleason 8–10 1992 AJCC T1c 1992 AJCC T2a 1992 AJCC T2b 1992 AJCC T2c
38 145 46 8 121 76 24 16
0.83
0.78
31 186 75 14 153 88 40 25
0.04
0.99
38 157 52 15 131 95 17 19
0.52
0.08
30 111 34 10 82 72 19 12
0.77
0.24
44 182 47 11 141 85 35 23
KEY: PSA ⫽ prostate-specific antigen; AJCC ⫽ American Joint Commission on Cancer. Chi-square P value evaluates the comparison of the proportion of patients with the given pretreatment clinical characteristic distribution who had a PSA at diagnosis as stated compared with a PSA at diagnosis of ⬎8 –10 ng/mL.
addition, it was assumed that patients undergoing screening with both PSA and DRE compared with DRE alone would be diagnosed at a lower PSA level. Specifically, this study analyzed the PSA outcome after RP for patients diagnosed and treated when the serum PSA level was greater than 4 ng/mL but not more than 8 ng/mL compared with greater than 8 and up to 10 ng/mL. MATERIAL AND METHODS PATIENT SELECTION, STAGING, AND TREATMENT Of 2127 men with PSA detected or clinically palpable but localized prostate cancer who underwent RP in Boston, Massachusetts or Philadelphia, Pennsylvania between 1989 and 2000, 1274 had a preoperative PSA level of more than 4 ng/mL but not more than 10 ng/mL and comprised the study population (Table I). The median age of the study population was 61 years (range 38 to 73). Patients underwent a staging evaluation as described in a previous publication.7 A single genitourinary pathologist reviewed the diagnostic biopsy specimens for all patients undergoing RP at each institution (Hospital of the University of Pennsylvania, n ⫽ 617; and Brigham and Women’s Hospital, n ⫽ 657). Patients who received adjuvant radiotherapy or neoadjuvant or adjuvant androgen suppression therapy were excluded.
FOLLOW-UP The median follow-up for patients treated at the Hospital of the University of Pennsylvania, the Brigham and Women’s Hospital, and all patients was 42 months (range 6 to 132), 40 months (range 6 to 130), and 40 months (range 6 to 132), respectively. Patients were seen 1 month after RP and then at 3-month intervals for 2 years, every 6 months for 5 years, and annually thereafter. At each follow-up visit, a serum PSA level was obtained before performing the DRE. All pretreatment PSA values were obtained within 2 months of the date of RP.
STATISTICAL ANALYSIS A Cox regression analysis8 evaluating the ability of the PSA level at diagnosis (greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL) to predict the time to PSA failure was performed. To assess for potential confounding due to an imbalance in the proportion of patients with a given biopsy Gleason score or 1992 American Joint CommisUROLOGY 58 (3), 2001
sion on Cancer (AJCC) T-category among the PSA groups being compared, a chi-square analysis evaluating these proportions was performed. The assumptions of the Cox model were tested and met. The relative risk of PSA failure and the 95% confidence interval were calculated based on the coefficients from the Cox regression model for each of the PSA groups compared. Significant differences in PSA outcome based on the PSA level at diagnosis were further analyzed using the traditional rankings of biopsy Gleason score 2 to 4, 5 to 6, 7, and 8 to 10. The estimates of PSA outcome were calculated using the Kaplan-Meier actuarial method9 and graphically displayed for illustration purposes. Pairwise comparisons were made using a log-rank test. PSA failure was defined as two consecutive detectable measurements obtained after an undetectable value (less than 0.2 ng/mL) with the time of PSA failure defined as the time of the first detectable measurement. Time 0 was defined as the date of diagnosis. If the PSA level never reached undetectable levels postoperatively (n ⫽ 2), PSA failure was defined as occurring at time 0.
RESULTS PROGNOSTIC FACTOR COMPARISON Table I lists the distribution for the pretreatment biopsy Gleason score and 1992 AJCC clinical Tcategory for each of the PSA groups evaluated in this study. With one exception, no significant differences (P ⱕ0.08) were found in the distribution of the biopsy Gleason score and 1992 AJCC clinical T-category identified for patients with a PSA greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 to 10 ng/mL. TIME TO PSA FAILURE ANALYSIS The results of the Cox regression analyses evaluating the ability of the PSA level at diagnosis (greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, and greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL) to predict the time to PSA failure are listed in Table II. Up to a threefold reduction in the risk of PSA 407
TABLE II. P values from Cox regression analyses and relative risk of PSA failure evaluating ability of PSA level at time of diagnosis to predict time to postoperative PSA failure PSA Level at Diagnosis (ng/mL) ⬎4.0–5.0 ⬎5.0–6.0 ⬎6.0–7.0 ⬎7.0–8.0 ⬎8.0–10.0
Median Follow-up (mo) 41 38 40 40 39
(6–130) (6–130) (6–132) (6–132) (6–132)
P Value
RR (95% CI)
⬍0.0001 0.004 0.02 0.5 —
0.3 (0.2–0.5) 0.5 (0.4–0.8) 0.6 (0.4–0.9) 0.9 (0.6–1.3) 1.0*
KEY: PSA ⫽ prostate-specific antigen; RR ⫽ relative risk; CI ⫽ confidence interval. Numbers in parentheses are the range. * All PSA groups compared with ⬎8.0 –10.0 ng/mL.
was greater than 4 to 7 ng/mL compared with greater than 8 to 10 ng/mL. COMMENT
FIGURE 1. Estimate of PSA failure-free (biochemical no evidence of disease) survival for patients stratified by the PSA level at diagnosis. Numbers at risk are shown annually above the time axis. Log-rank P values: PSA greater than 4 to 5 ng/mL versus greater than 8 to 10 ng/mL, P ⬍0.0001; PSA greater than 5 to 6 ng/mL versus greater than 8 to 10 ng/mL, P ⫽ 0.006; PSA greater than 6 to 7 ng/mL versus greater than 8 to 10 ng/mL, P ⫽ 0.03.
failure (P ⱕ0.02) was noted for patients diagnosed when the PSA level was greater than 4 but not more than 7 ng/mL compared with greater than 8 and up to 10 ng/mL (Fig. 1). PSA OUTCOME BY GLEASON SCORE Comparisons of the estimates of PSA outcome after RP for patients diagnosed when the PSA was greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL stratified by the biopsy Gleason score are shown in Table III. On the basis of these comparisons, patients with a biopsy Gleason score of 5 or 6 consistently benefited from RP performed when the PSA at diagnosis 408
The measurement of an annual PSA level has previously been shown to detect prostate cancer on average 4 to 6 years sooner than DRE,10 and the detection of prostate cancer at lower PSA levels has been suggested to improve the likelihood of finding curable disease.11 In addition, in populations in which annual PSA testing has been adopted, higher rates of organ-confined disease are found at RP,12 and metastatic disease at presentation is almost nonexistent.13 However, the impact of early detection through the use of annual PSA testing on prostate cancer mortality is currently unknown and awaits the results of ongoing prospective randomized screening studies.4 – 6 This study was designed to provide early insight regarding the impact that screening with both PSA and DRE may have on prostate cancer-specific mortality. The results suggested that for men with Gleason score 5 or 6 prostate cancer, a significant reduction in the relative risk of PSA failure after RP occurred as a result of early detection. Specifically, when men were diagnosed and treated at PSA levels less than 7 but greater than 4 ng/mL compared with greater than 8 to 10 ng/mL, the relative risk of PSA failure decreased by up to a factor of 3. This translated into a 15% improvement in PSA failurefree survival by 5 years (93% versus 78%, P ⬍0.0001) after RP. This study did not find a benefit to early detection for patients with biopsy Gleason 7 to 10 prostate cancer, with the possible exception of men presenting with a PSA between 4 and 5 ng/mL and biopsy Gleason score 7. However, given the small number (n ⫽ 46) of patients in this subgroup, longer follow-up and larger numbers will be needed to see whether the benefit in PSA outcome is maintained. A possible explanation for the lack UROLOGY 58 (3), 2001
TABLE III. Log-rank P values for comparisons of estimates of PSA outcome after RP for study patients with a PSA level at diagnosis as shown compared with a PSA >8–10 ng/mL stratified by biopsy Gleason score Biopsy Gleason Score 2–4 5–6 7 8–10
PSA Level at Diagnosis (ng/mL) >4–5
>5–6
>6–7
>7–8
0.11 0.004* 0.003* 0.90
0.09 0.0002* 0.21 0.18
0.29 0.02* 0.30 0.76
0.07 0.99 0.55 0.66
KEY: PSA ⫽ prostate-specific antigen; RP ⫽ radical prostatectomy. * Significant.
of benefit in the vast majority (266 of 312; 88%) of patients with high-grade cancer is that these patients comprised only 25% (312 of 1274) of the entire study population and many of these patients may have had occult micrometastatic disease at the time of RP.3 Both of these factors will reduce the power of this study to detect a difference in PSA outcome from early detection even if one exists. Similarly, no benefit was noted for patients with biopsy Gleason 2 to 4 prostate cancer. However, patients with well-differentiated prostate cancer will take the longest to see a benefit from early detection because the time to failure is protracted.14 The primary limitation of this study was that it was a retrospective, matched-cohort analysis based on the PSA outcome of patients treated at an academic institution and not a prospective, randomized, community-based screening trial whose endpoint was prostate cancer-specific mortality. However, previous comparisons15,16 of men undergoing RP have suggested that the pathologic characteristics of men in a screened and referral-based population are similar. Nonetheless, a selection bias may still exist. The assumptions warrant further discussion. First, the study’s primary endpoint was the PSA outcome, which presumed that all patients who sustained PSA failure would, given enough time, die of metastatic prostate cancer and none would be salvaged with additional therapy. A study3 on the natural history of prostate cancer after RP reported that the median interval to the development of metastatic disease after PSA failure was 8 years, with an additional 5 years on average beyond distant disease until death from prostate cancer. Therefore, the assumption that PSA failure will translate into death from prostate cancer is more likely to be true in general for younger patients without significant comorbid illness at the time of RP. Given that the median age of the patients in this study was 61 years and all the men were otherwise healthy surgical candidates with at least a UROLOGY 58 (3), 2001
10-year life expectancy lends credence to this assumption. Nevertheless, although the age and health of the patient population studied increased the likelihood that the PSA outcome may eventually be shown to act as a surrogate for survival in these men, until conclusively shown, the possibility for lead time and length time bias exists. Next, the assumption that no effective salvage therapy exists for patients who sustain PSA failure after RP is likely overstated. Although no prospective randomized trials have been reported on this issue, retrospective series17,18 have suggested a possible benefit in some patients for postoperative radiotherapy, although the group that may benefit appears to be small. Finally, the reference group that represented the population whose diagnosis was made based on the DRE alone was selected to have a PSA level greater than 8 not more than 10 ng/mL at diagnosis. However, many patients who are diagnosed with prostate cancer based on the DRE alone will have a PSA level greater than 10 ng/mL4,5,11 at diagnosis. Therefore, the results of this study may underestimate the potential reduction in prostate cancer mortality, given the conservative reference group selected to represent the unscreened patients. It will be nearly a decade before the results of the randomized screening studies4 – 6 that will determine the impact that the use of annual PSA testing in addition to DRE will have on prostate cancer mortality become available. In the interim, the guidelines set forth by the American Cancer Society recommend that obtaining an annual PSA level in addition to performing the DRE should be discussed19 with all men who have at least a 10-year life expectancy beginning at age 50, or at age 45 in high-risk populations. Given the number of other issues that require assessment during an annual visit to the primary care physician’s office and limited time, the issue of PSA testing needs clarification. Therefore, although no conclusions can be drawn from a nonrandomized comparison, the data presented provide physicians with informa409
tion that may be used to counsel patients with regard to the potential clinical utility of PSA testing. In particular, the possibility exists that for the vast majority of men who present today with newly diagnosed prostate cancer (biopsy Gleason score 5 or 6) and at least a 10-year life expectancy, early detection using PSA testing in addition to an annual DRE may lead to a decrease in prostate cancer mortality. REFERENCES 1. Walsh PC, Partin AW, and Epstein JI: Cancer control and quality of life following anatomical radical retropubic prostatectomy: results at 10 years. J Urol 152: 1831–1836, 1994. 2. D’Amico AV, Whittington R, Malkowicz SB, et al: A multivariate analysis of clinical and pathological factors which predict for prostate-specific antigen failure after radical prostatectomy after prostate cancer. J Urol 154: 131–138, 1995. 3. Pound CR, Partin AW, Eisenberger MA, et al: Natural history of progression after PSA elevation following radical prostatectomy. JAMA 281: 1591–1596, 1999. 4. Schroder FH, van der Maas P, Beemsterboer P, et al, for the Rotterdam section of the European Randomized Study of Screening for Prostate Cancer: Evaluation of the digital rectal examination as a screening test for prostate cancer. J Natl Cancer Inst 90: 1817–1823, 1998. 5. Maattanen L, Auvinen A, Stenman UH, et al: European randomized study of prostate cancer screening: first-year results of the Finnish trial. Br J Urol 79: 1210 –1214, 1999. 6. Etzioni R, Legler JM, Feuer EJ, et al: Cancer surveillance series: interpreting trends in prostate cancer—part III: quantifying the link between population prostate-specific antigen testing and recent declines in prostate cancer mortality. J Natl Cancer Inst 16: 1033–1039, 1999. 7. D’Amico AV, Whittington R, Malkowicz SB, et al: Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA 280: 969 –974, 1998.
410
8. Cox DR: Regression models and life tables. J R Stat Soc B 34: 187–189, 1972. 9. Kaplan EL, and Meier P: Non-parametric estimation from incomplete observations. J Am Stat Assoc 53: 457–500, 1958. 10. Carter HB, and Pearson JD: PSA and the natural course of prostate cancer, in Schroder FH (Ed): Recent Advances in Prostate Cancer and BPH. New York, Parthenon, 1997, pp 187– 193. 11. Carter HB, Epstein JI, and Partin AW: Influence of age and prostate-specific antigen on the chance of curable prostate cancer among men with nonpalpable disease. Urology 53: 126 –130, 1999. 12. Catalona WJ, Richie JP, Ahmann FR, et al: Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol 151: 1283–1290, 1994. 13. Labrie F, Candas B, Cusan L, et al: Diagnosis of advanced or noncurable prostate cancer can be practically eliminated by prostate-specific antigen. Urology 47: 212–217, 1996. 14. Albertsen PC, Hanley JA, Gleason DF, et al: Competing risk analysis of men aged 55 to 74 years at diagnosis managed conservatively for clinically localized prostate cancer. JAMA 280: 975–980, 1998. 15. Epstein JI, Walsh PC, Carmichael M, et al: Pathologic and clinical findings to predict tumor extent of non-palpable (stage T1c) prostate cancer. JAMA 271: 368 –374, 1994. 16. Humphrey PA, Keetch DW, Smith DS, et al: Prospective characterization of pathological features of prostatic carcinoma detected via serum prostate-specific antigen based screening. J Urol 155: 816 – 820, 1996. 17. Cadeddu JA, Partin AW, Deweese TL, et al: Long term results of radiation therapy for prostate cancer recurrence following radical prostatectomy. J Urol 159: 173–178, 1998. 18. Pisansky TM, Kozelsky TF, Myers RP, et al: Radiotherapy for isolated serum prostate specific antigen elevation after prostatectomy for prostate cancer. J Urol 163: 845– 850, 2000. 19. Concato J: Prostate-specific antigen: a useful screening test? Cancer J Sci Am 2: 188 –192, 2000.
UROLOGY 58 (3), 2001
ESTIMATING THE IMPACT ON PROSTATE CANCER MORTALITY OF INCORPORATING PROSTATE-SPECIFIC ANTIGEN TESTING INTO SCREENING ANTHONY V. D’AMICO, RICHARD WHITTINGTON, S. BRUCE MALKOWICZ, ANDREW A. RENSHAW, JOHN E. TOMASZEWSKI, CHRISTY BENTLEY, DELRAY SCHULTZ, SEAN ROCHA, ALAN WEIN, AND JEROME P. RICHIE
ABSTRACT Objectives. Whether early detection using prostate-specific antigen (PSA) and digital rectal examination (DRE) compared with DRE alone will reduce prostate cancer mortality awaits the results of ongoing prospective randomized trials. However, the impact that early detection could have on prostate cancerspecific survival can be estimated by assuming that PSA failure after radical prostatectomy (RP) will translate into death from prostate cancer. Methods. The study population consisted of 1274 men with clinically localized prostate cancer who underwent RP in Boston, Massachusetts or Philadelphia, Pennsylvania between 1989 and 2000 and had a preoperative PSA level greater than 4 but not more than 10 ng/mL. The primary endpoint was actuarial freedom from PSA failure (defined as PSA outcome). Results. The relative risk of PSA failure after RP for patients diagnosed with a PSA of greater than 4 to 5, 5 to 6, 6 to 7, or 7 to 8 ng/mL compared with greater than 8 up to 10 ng/mL was 0.3 (95% confidence interval [CI] 0.2 to 0.5), 0.5 (95% CI 0.4 to 0.8), 0.6 (95% CI 0.4 to 0.9), or 0.9 (95% CI 0.6 to 1.3), respectively. On the basis of the estimates of the 5-year PSA outcome, patients with a biopsy Gleason score of 5 or 6 (781 of 1274; 61%) consistently benefited from RP performed when the PSA at diagnosis was greater than 4 to 7 ng/mL compared with greater than 8 to 10 ng/mL (93% versus 78%, P ⬍0.0001). A benefit to early detection was not found for the vast majority (266 of 312; 88%) of patients who had a biopsy Gleason score of 7 or higher. Conclusions. Early detection using both PSA and DRE-based screening may benefit men who present with biopsy Gleason score 5 or 6 prostate cancer and a PSA level greater than 4 to 7 ng/mL compared with greater than 8 up to 10 ng/mL. This finding awaits validation from ongoing prospective randomized trials. UROLOGY 58: 406–410, 2001. © 2001, Elsevier Science Inc.
T
he preoperative prostate-specific antigen (PSA) level has been shown to predict the time to postoperative PSA failure.1,2 The time to postoperative PSA failure has been shown to predict for the time to distant failure, which in turn predicts From the Department of Radiation Oncology, Brigham and Women’s Hospital and Dana Farber Cancer Institute, Boston, Massachusetts; Departments of Radiation Oncology, Urology, and Pathology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania; Department of Mathematics, University of Millersville, Millersville, Pennsylvania; and Departments of Pathology and Urology, Brigham and Women’s Hospital, Boston, Massachusetts Reprint requests: Anthony V. D’Amico, M.D., Ph.D., Department of Radiation Oncology, Brigham and Women’s Hospital, 75 Francis Street, L-2 Level, Boston, MA 02215 Submitted: March 15, 2001, accepted: April 19, 2001
406
© 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
for the time to death from metastatic prostate cancer.3 Taken together, these associations suggest that a higher PSA level at diagnosis would lead to lower rates of cancer control and possibly an increased risk of dying of metastatic prostate cancer. However, only the results of ongoing prospective randomized screening studies4 – 6 evaluating the use of PSA and digital rectal examination (DRE) compared with DRE alone for screening will determine whether a reduction in prostate cancer mortality will be achieved. The goal of this study was to estimate the possible reduction in prostate cancer mortality that could be attributed to early detection through the use of both PSA and DRE by using PSA failure after RP as a surrogate for death from prostate cancer. In 0090-4295/01/$20.00 PII S0090-4295(01)01204-3
TABLE I. Pretreatment clinical characteristics of 1274 study patients stratified by the prostatespecific antigen level at diagnosis PSA PSA PSA PSA PSA >4–5 ng/mL P >5–6 ng/mL P >6–7 ng/mL P >7–8 ng/mL P >8–10 ng/mL Clinical Characteristic (n ⴝ 237) Value (n ⴝ 306) Value (n ⴝ 262) Value (n ⴝ 185) Value (n ⴝ 284) Biopsy Gleason 2–4 Biopsy Gleason 5–6 Biopsy Gleason 7 Biopsy Gleason 8–10 1992 AJCC T1c 1992 AJCC T2a 1992 AJCC T2b 1992 AJCC T2c
38 145 46 8 121 76 24 16
0.83
0.78
31 186 75 14 153 88 40 25
0.04
0.99
38 157 52 15 131 95 17 19
0.52
0.08
30 111 34 10 82 72 19 12
0.77
0.24
44 182 47 11 141 85 35 23
KEY: PSA ⫽ prostate-specific antigen; AJCC ⫽ American Joint Commission on Cancer. Chi-square P value evaluates the comparison of the proportion of patients with the given pretreatment clinical characteristic distribution who had a PSA at diagnosis as stated compared with a PSA at diagnosis of ⬎8 –10 ng/mL.
addition, it was assumed that patients undergoing screening with both PSA and DRE compared with DRE alone would be diagnosed at a lower PSA level. Specifically, this study analyzed the PSA outcome after RP for patients diagnosed and treated when the serum PSA level was greater than 4 ng/mL but not more than 8 ng/mL compared with greater than 8 and up to 10 ng/mL. MATERIAL AND METHODS PATIENT SELECTION, STAGING, AND TREATMENT Of 2127 men with PSA detected or clinically palpable but localized prostate cancer who underwent RP in Boston, Massachusetts or Philadelphia, Pennsylvania between 1989 and 2000, 1274 had a preoperative PSA level of more than 4 ng/mL but not more than 10 ng/mL and comprised the study population (Table I). The median age of the study population was 61 years (range 38 to 73). Patients underwent a staging evaluation as described in a previous publication.7 A single genitourinary pathologist reviewed the diagnostic biopsy specimens for all patients undergoing RP at each institution (Hospital of the University of Pennsylvania, n ⫽ 617; and Brigham and Women’s Hospital, n ⫽ 657). Patients who received adjuvant radiotherapy or neoadjuvant or adjuvant androgen suppression therapy were excluded.
FOLLOW-UP The median follow-up for patients treated at the Hospital of the University of Pennsylvania, the Brigham and Women’s Hospital, and all patients was 42 months (range 6 to 132), 40 months (range 6 to 130), and 40 months (range 6 to 132), respectively. Patients were seen 1 month after RP and then at 3-month intervals for 2 years, every 6 months for 5 years, and annually thereafter. At each follow-up visit, a serum PSA level was obtained before performing the DRE. All pretreatment PSA values were obtained within 2 months of the date of RP.
STATISTICAL ANALYSIS A Cox regression analysis8 evaluating the ability of the PSA level at diagnosis (greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL) to predict the time to PSA failure was performed. To assess for potential confounding due to an imbalance in the proportion of patients with a given biopsy Gleason score or 1992 American Joint CommisUROLOGY 58 (3), 2001
sion on Cancer (AJCC) T-category among the PSA groups being compared, a chi-square analysis evaluating these proportions was performed. The assumptions of the Cox model were tested and met. The relative risk of PSA failure and the 95% confidence interval were calculated based on the coefficients from the Cox regression model for each of the PSA groups compared. Significant differences in PSA outcome based on the PSA level at diagnosis were further analyzed using the traditional rankings of biopsy Gleason score 2 to 4, 5 to 6, 7, and 8 to 10. The estimates of PSA outcome were calculated using the Kaplan-Meier actuarial method9 and graphically displayed for illustration purposes. Pairwise comparisons were made using a log-rank test. PSA failure was defined as two consecutive detectable measurements obtained after an undetectable value (less than 0.2 ng/mL) with the time of PSA failure defined as the time of the first detectable measurement. Time 0 was defined as the date of diagnosis. If the PSA level never reached undetectable levels postoperatively (n ⫽ 2), PSA failure was defined as occurring at time 0.
RESULTS PROGNOSTIC FACTOR COMPARISON Table I lists the distribution for the pretreatment biopsy Gleason score and 1992 AJCC clinical Tcategory for each of the PSA groups evaluated in this study. With one exception, no significant differences (P ⱕ0.08) were found in the distribution of the biopsy Gleason score and 1992 AJCC clinical T-category identified for patients with a PSA greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 to 10 ng/mL. TIME TO PSA FAILURE ANALYSIS The results of the Cox regression analyses evaluating the ability of the PSA level at diagnosis (greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, and greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL) to predict the time to PSA failure are listed in Table II. Up to a threefold reduction in the risk of PSA 407
TABLE II. P values from Cox regression analyses and relative risk of PSA failure evaluating ability of PSA level at time of diagnosis to predict time to postoperative PSA failure PSA Level at Diagnosis (ng/mL) ⬎4.0–5.0 ⬎5.0–6.0 ⬎6.0–7.0 ⬎7.0–8.0 ⬎8.0–10.0
Median Follow-up (mo) 41 38 40 40 39
(6–130) (6–130) (6–132) (6–132) (6–132)
P Value
RR (95% CI)
⬍0.0001 0.004 0.02 0.5 —
0.3 (0.2–0.5) 0.5 (0.4–0.8) 0.6 (0.4–0.9) 0.9 (0.6–1.3) 1.0*
KEY: PSA ⫽ prostate-specific antigen; RR ⫽ relative risk; CI ⫽ confidence interval. Numbers in parentheses are the range. * All PSA groups compared with ⬎8.0 –10.0 ng/mL.
was greater than 4 to 7 ng/mL compared with greater than 8 to 10 ng/mL. COMMENT
FIGURE 1. Estimate of PSA failure-free (biochemical no evidence of disease) survival for patients stratified by the PSA level at diagnosis. Numbers at risk are shown annually above the time axis. Log-rank P values: PSA greater than 4 to 5 ng/mL versus greater than 8 to 10 ng/mL, P ⬍0.0001; PSA greater than 5 to 6 ng/mL versus greater than 8 to 10 ng/mL, P ⫽ 0.006; PSA greater than 6 to 7 ng/mL versus greater than 8 to 10 ng/mL, P ⫽ 0.03.
failure (P ⱕ0.02) was noted for patients diagnosed when the PSA level was greater than 4 but not more than 7 ng/mL compared with greater than 8 and up to 10 ng/mL (Fig. 1). PSA OUTCOME BY GLEASON SCORE Comparisons of the estimates of PSA outcome after RP for patients diagnosed when the PSA was greater than 4 to 5, greater than 5 to 6, greater than 6 to 7, or greater than 7 to 8 ng/mL compared with greater than 8 but less than 10 ng/mL stratified by the biopsy Gleason score are shown in Table III. On the basis of these comparisons, patients with a biopsy Gleason score of 5 or 6 consistently benefited from RP performed when the PSA at diagnosis 408
The measurement of an annual PSA level has previously been shown to detect prostate cancer on average 4 to 6 years sooner than DRE,10 and the detection of prostate cancer at lower PSA levels has been suggested to improve the likelihood of finding curable disease.11 In addition, in populations in which annual PSA testing has been adopted, higher rates of organ-confined disease are found at RP,12 and metastatic disease at presentation is almost nonexistent.13 However, the impact of early detection through the use of annual PSA testing on prostate cancer mortality is currently unknown and awaits the results of ongoing prospective randomized screening studies.4 – 6 This study was designed to provide early insight regarding the impact that screening with both PSA and DRE may have on prostate cancer-specific mortality. The results suggested that for men with Gleason score 5 or 6 prostate cancer, a significant reduction in the relative risk of PSA failure after RP occurred as a result of early detection. Specifically, when men were diagnosed and treated at PSA levels less than 7 but greater than 4 ng/mL compared with greater than 8 to 10 ng/mL, the relative risk of PSA failure decreased by up to a factor of 3. This translated into a 15% improvement in PSA failurefree survival by 5 years (93% versus 78%, P ⬍0.0001) after RP. This study did not find a benefit to early detection for patients with biopsy Gleason 7 to 10 prostate cancer, with the possible exception of men presenting with a PSA between 4 and 5 ng/mL and biopsy Gleason score 7. However, given the small number (n ⫽ 46) of patients in this subgroup, longer follow-up and larger numbers will be needed to see whether the benefit in PSA outcome is maintained. A possible explanation for the lack UROLOGY 58 (3), 2001
TABLE III. Log-rank P values for comparisons of estimates of PSA outcome after RP for study patients with a PSA level at diagnosis as shown compared with a PSA >8–10 ng/mL stratified by biopsy Gleason score Biopsy Gleason Score 2–4 5–6 7 8–10
PSA Level at Diagnosis (ng/mL) >4–5
>5–6
>6–7
>7–8
0.11 0.004* 0.003* 0.90
0.09 0.0002* 0.21 0.18
0.29 0.02* 0.30 0.76
0.07 0.99 0.55 0.66
KEY: PSA ⫽ prostate-specific antigen; RP ⫽ radical prostatectomy. * Significant.
of benefit in the vast majority (266 of 312; 88%) of patients with high-grade cancer is that these patients comprised only 25% (312 of 1274) of the entire study population and many of these patients may have had occult micrometastatic disease at the time of RP.3 Both of these factors will reduce the power of this study to detect a difference in PSA outcome from early detection even if one exists. Similarly, no benefit was noted for patients with biopsy Gleason 2 to 4 prostate cancer. However, patients with well-differentiated prostate cancer will take the longest to see a benefit from early detection because the time to failure is protracted.14 The primary limitation of this study was that it was a retrospective, matched-cohort analysis based on the PSA outcome of patients treated at an academic institution and not a prospective, randomized, community-based screening trial whose endpoint was prostate cancer-specific mortality. However, previous comparisons15,16 of men undergoing RP have suggested that the pathologic characteristics of men in a screened and referral-based population are similar. Nonetheless, a selection bias may still exist. The assumptions warrant further discussion. First, the study’s primary endpoint was the PSA outcome, which presumed that all patients who sustained PSA failure would, given enough time, die of metastatic prostate cancer and none would be salvaged with additional therapy. A study3 on the natural history of prostate cancer after RP reported that the median interval to the development of metastatic disease after PSA failure was 8 years, with an additional 5 years on average beyond distant disease until death from prostate cancer. Therefore, the assumption that PSA failure will translate into death from prostate cancer is more likely to be true in general for younger patients without significant comorbid illness at the time of RP. Given that the median age of the patients in this study was 61 years and all the men were otherwise healthy surgical candidates with at least a UROLOGY 58 (3), 2001
10-year life expectancy lends credence to this assumption. Nevertheless, although the age and health of the patient population studied increased the likelihood that the PSA outcome may eventually be shown to act as a surrogate for survival in these men, until conclusively shown, the possibility for lead time and length time bias exists. Next, the assumption that no effective salvage therapy exists for patients who sustain PSA failure after RP is likely overstated. Although no prospective randomized trials have been reported on this issue, retrospective series17,18 have suggested a possible benefit in some patients for postoperative radiotherapy, although the group that may benefit appears to be small. Finally, the reference group that represented the population whose diagnosis was made based on the DRE alone was selected to have a PSA level greater than 8 not more than 10 ng/mL at diagnosis. However, many patients who are diagnosed with prostate cancer based on the DRE alone will have a PSA level greater than 10 ng/mL4,5,11 at diagnosis. Therefore, the results of this study may underestimate the potential reduction in prostate cancer mortality, given the conservative reference group selected to represent the unscreened patients. It will be nearly a decade before the results of the randomized screening studies4 – 6 that will determine the impact that the use of annual PSA testing in addition to DRE will have on prostate cancer mortality become available. In the interim, the guidelines set forth by the American Cancer Society recommend that obtaining an annual PSA level in addition to performing the DRE should be discussed19 with all men who have at least a 10-year life expectancy beginning at age 50, or at age 45 in high-risk populations. Given the number of other issues that require assessment during an annual visit to the primary care physician’s office and limited time, the issue of PSA testing needs clarification. Therefore, although no conclusions can be drawn from a nonrandomized comparison, the data presented provide physicians with informa409
tion that may be used to counsel patients with regard to the potential clinical utility of PSA testing. In particular, the possibility exists that for the vast majority of men who present today with newly diagnosed prostate cancer (biopsy Gleason score 5 or 6) and at least a 10-year life expectancy, early detection using PSA testing in addition to an annual DRE may lead to a decrease in prostate cancer mortality. REFERENCES 1. Walsh PC, Partin AW, and Epstein JI: Cancer control and quality of life following anatomical radical retropubic prostatectomy: results at 10 years. J Urol 152: 1831–1836, 1994. 2. D’Amico AV, Whittington R, Malkowicz SB, et al: A multivariate analysis of clinical and pathological factors which predict for prostate-specific antigen failure after radical prostatectomy after prostate cancer. J Urol 154: 131–138, 1995. 3. Pound CR, Partin AW, Eisenberger MA, et al: Natural history of progression after PSA elevation following radical prostatectomy. JAMA 281: 1591–1596, 1999. 4. Schroder FH, van der Maas P, Beemsterboer P, et al, for the Rotterdam section of the European Randomized Study of Screening for Prostate Cancer: Evaluation of the digital rectal examination as a screening test for prostate cancer. J Natl Cancer Inst 90: 1817–1823, 1998. 5. Maattanen L, Auvinen A, Stenman UH, et al: European randomized study of prostate cancer screening: first-year results of the Finnish trial. Br J Urol 79: 1210 –1214, 1999. 6. Etzioni R, Legler JM, Feuer EJ, et al: Cancer surveillance series: interpreting trends in prostate cancer—part III: quantifying the link between population prostate-specific antigen testing and recent declines in prostate cancer mortality. J Natl Cancer Inst 16: 1033–1039, 1999. 7. D’Amico AV, Whittington R, Malkowicz SB, et al: Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA 280: 969 –974, 1998.
410
8. Cox DR: Regression models and life tables. J R Stat Soc B 34: 187–189, 1972. 9. Kaplan EL, and Meier P: Non-parametric estimation from incomplete observations. J Am Stat Assoc 53: 457–500, 1958. 10. Carter HB, and Pearson JD: PSA and the natural course of prostate cancer, in Schroder FH (Ed): Recent Advances in Prostate Cancer and BPH. New York, Parthenon, 1997, pp 187– 193. 11. Carter HB, Epstein JI, and Partin AW: Influence of age and prostate-specific antigen on the chance of curable prostate cancer among men with nonpalpable disease. Urology 53: 126 –130, 1999. 12. Catalona WJ, Richie JP, Ahmann FR, et al: Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol 151: 1283–1290, 1994. 13. Labrie F, Candas B, Cusan L, et al: Diagnosis of advanced or noncurable prostate cancer can be practically eliminated by prostate-specific antigen. Urology 47: 212–217, 1996. 14. Albertsen PC, Hanley JA, Gleason DF, et al: Competing risk analysis of men aged 55 to 74 years at diagnosis managed conservatively for clinically localized prostate cancer. JAMA 280: 975–980, 1998. 15. Epstein JI, Walsh PC, Carmichael M, et al: Pathologic and clinical findings to predict tumor extent of non-palpable (stage T1c) prostate cancer. JAMA 271: 368 –374, 1994. 16. Humphrey PA, Keetch DW, Smith DS, et al: Prospective characterization of pathological features of prostatic carcinoma detected via serum prostate-specific antigen based screening. J Urol 155: 816 – 820, 1996. 17. Cadeddu JA, Partin AW, Deweese TL, et al: Long term results of radiation therapy for prostate cancer recurrence following radical prostatectomy. J Urol 159: 173–178, 1998. 18. Pisansky TM, Kozelsky TF, Myers RP, et al: Radiotherapy for isolated serum prostate specific antigen elevation after prostatectomy for prostate cancer. J Urol 163: 845– 850, 2000. 19. Concato J: Prostate-specific antigen: a useful screening test? Cancer J Sci Am 2: 188 –192, 2000.
UROLOGY 58 (3), 2001