- Email: [email protected]
Time to Prostate Specific Antigen Recurrence After Radical Prostatectomy and Risk of Prostate Cancer Specific Mortality
Time to Prostate Specific Antigen Recurrence After Radical Prostatectomy and Risk of Prostate Cancer Specific Mortality Stephen J. Freedland,* Elizabeth B. Humphreys, Leslie A. Mangold, Mario Eisenberger and Alan W. Partin From the Departments of Urology and Oncology, The James Buchanan Brady Urological Institute, Johns Hopkins Medicine, Baltimore, Maryland
Purpose: In patients treated with radical prostatectomy who have biochemical recurrence we have previously reported that time from surgery to biochemical recurrence and postoperative prostate specific antigen doubling time are significantly related to the risk of prostate cancer death. We performed a more thorough examination of the association of time from surgery to biochemical recurrence and the risk of prostate cancer death. Materials and Methods: We retrospectively studied the records of 379 patients treated with radical prostatectomy between 1982 and 2000 who had had biochemical recurrence. We examined the association of time from surgery to prostate specific antigen recurrence and prostate specific antigen doubling time, and the risk of prostate cancer death using the Spearman correlation and Cox proportional hazards regression, respectively. Results: Longer time from surgery to prostate specific antigen recurrence was associated with a slower prostate specific antigen doubling time (Spearman r ⫽ 0.36, p ⬍0.001) and a decreased risk of prostate cancer death (RR 0.76, 95% CI 0.66 to 0.88, p ⬍0.001). The 15-year actuarial prostate cancer specific survival rate after biochemical recurrence in patients with recurrence at 3 years or less was 41% (95% CI 29 to 53) compared to 87% (95% CI 75 to 93) in patients with recurrence more than 3 years after radical prostatectomy. On multivariate analysis a shorter time from surgery to prostate specific antigen recurrence was associated with an increased risk of prostate cancer death (3 or less vs more than 3 years, RR 2.70, 95% CI 1.37 to 5.31, p ⫽ 0.004). Conclusions: Earlier prostate specific antigen recurrence is associated with an increased risk of prostate cancer death. These data suggest that perhaps time to prostate specific antigen recurrence may be a reasonable intermediate end point in patients treated with radical prostatectomy, although this must be validated in other studies. Key Words: prostate, prostatic neoplasms, prostatectomy, prostate-specific antigen, mortality
he common use of RP has resulted in large databases containing data on many patients treated with RP, which have provided valuable insight into prostate cancer biology. After prostatectomy essentially all PSA producing cells are removed. Therefore, slight increases in PSA, eg 0.2 or greater,1 or 0.4 or greater2 ng/ml, are used to indicate cancer recurrence and often as intermediate study end points. Recent data suggested that post-recurrence PSADT may be a surrogate for PCSM.3 Similarly we4 and others5 have found that a short PSADT after biochemical recurrence is strongly associated with PCSM. In light of these data does time to biochemical recurrence remain an
T
important variable? If so, is it sufficiently important to justify using it as a study end point? We have previously studied 379 patients treated with RP at our institution with biochemical recurrence who had PSADT data available and found that, after controlling for PSADT, time to biochemical recurrence was significantly associated with PCSM.4 In the current study using the same cohort with slightly longer followup we performed more thorough analysis of the association between time to PSA recurrence, and postoperative PSADT and PCSM to assess whether time to PSA recurrence is a reasonable intermediate end point in patients treated with RP.
Submitted for publication October 14, 2005. Study received approval from the Institutional Review Board, Johns Hopkins University. Supported by National Institute of Health/National Cancer Institute-SPORE Grant P50CA58236, The Prostate Cancer Foundation, Department of Defense Prostate Cancer Research Program PC030666 and an American Foundation for Urological Disease/ American Urological Association Education and Research Scholarship Award. Views and opinions of, and endorsements by the author(s) do not reflect those of the United States Army or Department of Defense. * Correspondence and current address: Department of Surgery, Box 3850, Duke University Medical Center, Durham, North Carolina 27710 (telephone: 919-668-8361; FAX: 919-668-7093; e-mail: [email protected]).
MATERIALS AND METHODS
0022-5347/06/1764-1404/0 THE JOURNAL OF UROLOGY® Copyright © 2006 by AMERICAN UROLOGICAL ASSOCIATION
Patient Population The Institutional Review Board at Johns Hopkins University approved this study and, when required, written informed consent was obtained from all subjects. We evaluated data on 5,096 patients with prostate adenocarcinoma treated with RP at our institution from April 1982 to October 2000 who had followup data available. During a mean followup ⫾ SD of 6.3 ⫾ 4.5 years (median 5) 997 patients (20%) experienced biochemical recurrence, defined as a single postoperative PSA of 0.2 ng/ml or greater.1 Of these 997
1404
Vol. 176, 1404-1408, October 2006 Printed in U.S.A. DOI:10.1016/j.juro.2006.06.017
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH patients 411 had data available that included 2 or more PSA values separated by 3 months or more within 2 years after biochemical recurrence and who did not receive adjuvant radiation or hormonal therapy before recurrence. Two patients who received preoperative radiation and 10 who received hormonal therapy were excluded. Another 20 patients were excluded who received salvage radiation with a durable PSA response of more than 2 years, and who were considered to have only local recurrence and to have been cured by surgery plus radiation. Another 22 patients who underwent salvage radiation but did not achieve a durable PSA response were considered to have distant failure and they were included. This resulted in 379 patients. Prostate cancer death was defined as death in any patient with metastasis that showed progression following hormonal therapy or death in any patient not treated with hormonal therapy but with widespread metastases without another obvious cause of death. Because of the limited number of nonprostate cancer deaths (18) in this relatively healthy cohort, PCSM approached all cause mortality and, therefore, data are only reported for PCSM.
Determination of PSADT PSADT was calculated by the natural log of 2 (0.693) divided by the slope of the linear regression line of natural log of all PSA values obtained within the first 2 years after biochemical recurrence.6 As previously described, PSADT was categorize as less than 3, 3.0 to 8.9 and 9.0 to 14.9 vs 15 or greater months.4 Seven patients with negative or zero PSADT (no PSA increase) were assigned to the 15 months or greater PSADT category. All PSA values were obtained after biochemical recurrence and before subsequent radiation or hormonal therapy. Therefore, subsequent therapy had no impact on PSADT calculation. In a subset of fewer than 100 patients frozen banked serum collected before the introduction of the PSA assay was used to measure PSA retrospectively.
Statistical Analysis The association between time to PSA recurrence and PSADT was examined using the Spearman correlation. The association between time to PSA recurrence and time from PSA recurrence to PCSM was examined using Cox proportional hazards regression. For multivariate analysis a forward stepwise Cox proportional hazards model was used with p ⬍0.15 determining which variables would be entered into the model at each step. Variables considered for entry were preoperative PSA, PSADT (categorical variable), pathological Gleason sum (7 or less vs 8 or greater), time from surgery to biochemical recurrence, patient age at RP (continuous), surgical margin status, extraprostatic extension, seminal vesicle invasion and lymph node metastasis. Time from surgery to PSA recurrence was examined as a continuous and as a categorical (3 or less vs more than 3 years) variable because we have previously found that this cutoff point provides the greatest likelihood chi-square ratio for estimating time to PCSM.4 All statistical analyses were performed using STATA® 8.0.
1405
TABLE 1. Preoperative clinical and pathological characteristics in 379 patients with PSA recurrence after radical prostatectomy No. pts Mean age at surgery ⫾ SD Preop PSA (ng/ml): Mean ⫾ SD Median No. biopsy Gleason score (%): 2–6 7 8–10 No. race (%): White Black Nonblack-nonwhite No. clinical stage (%): T1b T1c T2a T2b T2c T3 No. pathological Gleason score (%): 2–6 7 8–10 No. pos surgical margins (%) No. extracapsular extension (%) No. seminal vesicle invasion (%) No. pos lymph nodes (%)
379 59.7 ⫾ 6.2 14.4 ⫾ 15.0 10.6 173 (47) 146 (39) 53 (14) 360 (95) 14 (4) 5 (1) 6 (2) 73 (19) 130 (34) 119 (32) 29 (8) 20 (5) 54 (14) 195 (51) 130 (34) 178 (47) 337 (89) 143 (38) 124 (33)
RESULTS Patient Characteristics and Overall Survival Most patients had pathologically extraprostatic disease and high grade disease (table 1). Mean followup after surgery ⫾ SD in the 379 patients was 10.8 ⫾ 4.9 years (median 10, range 2 to 21). Mean time to biochemical recurrence was 3.5 ⫾ 3.0 years (median 2) and mean followup after biochemical recurrence was 7.3 ⫾ 3.9 years (median 7). During this time 76 of the 379 patients (20%) died of prostate cancer. Median prostate cancer specific survival in the study cohort of 379 patients with PSADT data available who experienced biochemical recurrence was 16 years. In this cohort 5, 10 and 15-year cause specific survival from the time of biochemical recurrence was 93% (95% CI 90 to 95), 75% (95% CI 68 to 80) and 53% (95% CI 42 to 64), respectively. Association Between Time to PSA Recurrence and PSADT There was a significant association between time to PSA recurrence and post-recurrence PSADT with later recurrence associated with slower PSADT (Spearman r ⫽ 0.36, p ⬍0.001). While overall median PSADT was 12.1 months, median PSADT was 9.2 months in patients with recurrence at 3 years or less compared to 17.8 months in patients with recurrence more than 3 years after surgery (fig. 1). Overall 6% of patients had a PSADT of less than 3 months. However, 19 of 244 patients (8%) with recurrence at 3 years or less had a PSADT of less than 3 months compared to only 4 of 135 (3%) with recurrence more than 3 years after surgery (table 2). Time to PSA Recurrence and Risk of PCSM Each 1-year delay in time to PSA recurrence was associated with a 24% decreased risk of PCSM on univariate analysis (RR 0.76, 95% CI 0.66 to 0.88, p ⬍0.001). The risk of PCSM appeared relatively stable in patients with recurrence up
1406
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH TABLE 3. Time from surgery to biochemical recurrence and risk of prostate cancer specific mortality in 379 men with biochemical recurrence after radical prostatectomy No. Prostate Ca Deaths
FIG. 1. Box plot shows association between time from surgery to PSA recurrence and post-recurrence doubling time with men with PSADT more than 100 months assigned value of 100 months.
through year 3. It was decreased in men with recurrence in year 4 and then it appeared relatively stable from years 5 to 8, after which there only 1 prostate cancer death was noted (table 3). The 15-year actuarial prostate cancer specific survival after biochemical recurrence in patients with recurrence at 3 years or less was 41% (95% CI 29 to 53) compared to 87% (95% CI 75 to 93) in patients with recurrence more than 3 years after surgery (fig. 2). On multivariate analysis only PSADT, pathological Gleason sum and time to PSA recurrence (as a continuous and a categorical variable) were significantly associated with time to PCSM (table 4). A total of 54 patients (14%) received hormonal therapy before metastatic disease developed, of whom 7 went on to metastatic disease, while 47 remained metastasis-free at last followup. After excluding these 54 men and adjusting for PSADT and pathological Gleason sum time to PSA recurrence remained associated with PCSM as a continuous variable (RR 0.87, 95% CI 0.75 to 1.01, p ⫽ 0.07) and as a categorical variable (3 or fewer vs more than 3 years RR 2.78, 95% CI 1.33 to 5.82, p ⫽ 0.01). DISCUSSION
Surgery—PSA Recurrence (yrs)
No. Pts
Observed
Expected
Observed/Expected Prostate Ca Death Ratio
1 2 3 4 5 6 7–8 9 or Greater
123 74 46 36 21 26 10 15
32 24 10 6 1 1 1 1
24.03 14.79 9.15 10.34 3.98 4.79 4.95 3.97
1.33 1.62 1.09 0.58 0.25 0.21 0.20 0.25
months to calculate PSADT. Therefore, what information can we give to patients at the time of recurrence? In addition, the use of time to PSA recurrence as an end point for study avoids the complexity involved in calculating PSADT, eg how many PSA measurements are required during what time period, etc. However, data linking early PSA recurrence to PCSM has largely been lacking. In the current study we found that early PSA recurrence was associated with shorter post-recurrence PSADT, which is a known poor prognostic factor.3,4,6 – 8 Moreover, even after adjusting for PSADT and pathological Gleason sum early biochemical recurrence was associated with an increased risk of PCSM. This suggests that early PSA recurrence may be a reasonable end point in studies of patients treated with RP. It is becoming increasingly clear that many patients who experience PSA recurrence have a prolonged course until metastasis6 or PCSM.4 To better risk stratify these patients, groups have attempted to identify characteristics that predict poor outcomes. Specifically several studies describe that post-recurrence PSADT is strongly prognostic for metastasis6 – 8 and PCSM.3,4 Indeed, in the current study PSADT was one of the strongest predictors of PCSM. However, clinical application of PSADT has limitations. 1) It requires several PSA values to calculate it. 2) One group suggested that PSADT calculations using low PSA values may not be accurate for predicting “eventual PSADT.”9 Indeed, we have previously found that using all PSA values within 2 years
Due to the sensitivity of PSA for the early detection of disease recurrence after RP, PSA recurrence can predate metastasis by years.6 At the time of PSA recurrence, eg PSA 0.2 ng/ml, PSADT calculations may not be accurate. Moreover, it requires several measurements during multiple
TABLE 2. Time from surgery to PSA recurrence and postoperative PSADT in 379 patients with PSA recurrence after radical prostatectomy Surgery– PSA Recurrence (yrs)
15.0 or Greater
9.0–14.9
3.0–8.9
Less than 3.0
1 2 3 4 5 6 7–8 9 or Greater
35 (28) 23 (31) 18 (39) 20 (56) 14 (67) 17 (65) 13 (54) 18 (64)
16 (13) 19 (25) 15 (33) 6 (17) 3 (14) 5 (19) 8 (33) 7 (25)
57 (46) 30 (40) 12 (26) 8 (22) 3 (14) 4 (15) 2 (8) 3 (11)
15 (12) 3 (4) 1 (2) 2 (6) 1 (5) 0 1 (4) 0
PSADT (mos)
FIG. 2. Kaplan-Meier 15-year actuarial prostate cancer specific survival curves after PSA recurrence in patients with PSA recurrence, stratified by PSA recurrence 3 or less vs more than 3 years after surgery (log rank p ⬍0.001).
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH TABLE 4. Multivariate Cox proportional hazards analysis of factors predicting time to prostate cancer specific death after PSA recurrence following radical prostatectomy RR
95% CI
p Value
Time from surgery to PSA recurrence as continuous variable Yrs from surgery to PSA recurrence 0.87 0.75–1.00 0.04 (continuous) PSA doubling time (vs 15.0 mos or greater): Less than 3 mos 21.49 8.78–52.57 ⬍0.001 3.0–8.9 Mos 9.23 4.23–20.15 ⬍0.001 9.0–14.9 Mos 3.28 1.37–7.83 0.01 Pathological Gleason sum (8 or 2.13 1.31–3.46 0.002 greater vs less than 8) Time from surgery to PSA recurrence as categorical variable Yrs from surgery to PSA recurrence 2.70 1.37–5.31 0.004 (3 or less vs greater than 3) PSA doubling time (vs 15.0 mos or greater): Less than 3 mos 22.97 9.45–55.81 ⬍0.001 3.0–8.9 Mos 8.82 4.05–19.20 ⬍0.001 9.0–14.9 Mos 2.92 1.22–6.99 0.02 Pathological Gleason sum (8 or 2.22 1.37–3.61 0.001 greater vs less than 8)
after recurrence rather than the first 2 PSA values only provided better risk assessment for predicting metastasis.6 This suggests that perhaps multiple PSA values during an extended period may be needed to accurately measure PSADT. 3) Given that salvage radiation is most beneficial when given at lower PSA values,10 many patients today are treated with salvage therapy before accurate PSADT measurement is ever available. Given these potential limitations to calculating and using PSADT for risk assessment, we examined an alternative end point in patients treated with RP, that is time to PSA recurrence. We have previously reported outcomes in this cohort,4 although the current study contains a slightly longer followup (mean 10.8 vs 10.3 years after surgery) and more prostate cancer deaths (76 vs 66). In our prior study, time to PSA recurrence, which was significantly related to PCSM, was examined as 3 or less vs more than 3 years. This was done to facilitate the construction of easy to use tables assessing the risk of PCSM. However, we hypothesized that the association between time to PSA recurrence and PCSM would be on a continuum, rather than have a discreet 3 year cutoff. Therefore, in the current study we performed more thorough examination of the association between time to PSA recurrence and PCSM. We found a steady lengthening of PSADT and a decrease in the risk of PCSM as time to PSA recurrence increased. Specifically for each 1-year delay in biochemical recurrence the risk of prostate cancer death decreased 24%. Thus, although in our data a cutoff of 3 or less vs more than 3 years provided the best PCSM stratification, time to PSA recurrence should really be viewed as a continuum of risk. This is an important observation that was not presented in our initial description of this cohort.4 In the current study the best single cutoff point for stratifying patients at low vs high risk was PSA recurrence 3 or less vs more than 3 years. This cutoff differs slightly from that in an earlier study including some patients in the current cohort.6 In that study 2 or less vs more than 2 years provided the best risk stratification. We explored whether this slight difference in results reflected the changing nature of our patients with time or differing patient populations. In the current study earlier year of surgery was significantly
1407
associated with longer time to recurrence (Spearman r ⫽ ⫺0.438, p ⬍0.001), likely reflecting that in men treated in the pre-PSA era only those with PSA data available could be included in the analysis. Consequently men with early recurrence would necessarily be excluded from this study. Likewise in men treated in the most recent years not enough time has elapsed to capture all PSA recurrences and allow enough time for measuring PSADT. Thus, only those men with early recurrence would have been included. Given the limited followup and long natural history of prostate cancer, it is currently too premature to examine whether time to recurrence remains significantly prognostic in men treated in recent years. Analogous to the current study, researchers from the Mayo Clinic found a significant association between early PSA recurrence after RP and a short PSADT.8 However, despite this association and the fact that short PSADT strongly predicted clinical progression in that series, early PSA recurrence was surprisingly not associated with clinical progression. This may reflect the increased use of early salvages therapy in men with early recurrence and censoring of these patients relative to lower rates of salvage therapy in men with delayed recurrence in that series. In contrast, in the current study there was no significant association between early PSA recurrence and early hormonal therapy (chi-square p ⫽ 0.32). Alternative explanations for different findings between the prior8 and the current study are different time to PSA recurrence cutoff points and different patient populations. After validation in other studies the finding that time to PSA recurrence is significantly related to PCSM would have important clinical implications. For instance, if the link between time to PSA recurrence and PCSM can be firmly enough established, this would potentially justify the design of adjuvant clinical trials using time to PSA recurrence as an end point. Given the long natural history of PSA recurrence, having an earlier end point to use in these trials would be extremely helpful and it would speed the process of bringing potential lifesaving drugs to the market. However, the current findings must first be validated in other series, ideally in the context of a prospective clinical trial. In the current study only 54 men received early hormonal therapy. It is unlikely that these men dramatically influenced our findings since excluding these men did not materially change our results. The current study, in which most patients did not receive early hormonal therapy, allows a unique insight into the natural history of prostate cancer. However, because hormonal therapy can delay metastasis,11,12 many patients today receive hormonal therapy before the development of metastasis. It is debatable whether early hormonal therapy in patients with biochemical recurrence after RP delays PCSM.13 If early hormonal therapy dramatically improves survival,14 the estimates generated in the current study may underestimate the likelihood of survival. PSA and PSADT are time dependent factors. We calculated PSADT while assuming that PSA increased exponentially during the first 2 years after recurrence. Although PSADT may not have been calculable in some men until the end of year 2 because no second PSA value was available until then, calculated PSADT at that point was presumed to be PSADT at the time of biochemical recurrence. However, it
1408
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH
is certainly plausible that in the long term PSADT values may not be stable.
5.
CONCLUSIONS 6.
In 379 patients who experienced biochemical recurrence after RP, shorter time from surgery to PSA recurrence was associated with more rapid post-recurrence PSADT and an increased risk of PCSM. Patients with recurrence within 3 years after surgery were at particularly high risk for PCSM. Even after controlling for PSADT and pathological Gleason sum, a shorter time from surgery to PSA recurrence was associated with an increased risk of PCSM. Time from surgery to PSA recurrence may be a reasonable intermediate end point in patients treated with RP.
7.
8.
9.
Abbreviations and Acronyms PCSM PSA PSADT RP
⫽ ⫽ ⫽ ⫽
prostate cancer specific mortality prostate specific antigen PSA doubling time radical prostatectomy
10.
11.
REFERENCES 1.
Freedland, S. J., Sutter, M. E., Dorey, F. and Aronson, W. J.: Defining the ideal cutpoint for determining PSA recurrence after radical prostatectomy. Urology, 61: 365, 2003 2. Amling, C. L., Bergstralh, E. J., Blute, M. L., Slezak, J. M. and Zincke, H.: Defining prostate specific antigen progression after radical prostatectomy: what is the most appropriate cut point? J Urol, 165: 1146, 2001 3. D’Amico, A. V., Moul, J. W., Carroll, P. R., Sun, L., Lubeck, D. and Chen, M. H.: Surrogate end point for prostate cancerspecific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst, 95: 1376, 2003 4. Freedland, S. J., Humphreys, E. B., Mangold, L. A., Eisenberger, M., Dorey, F. J., Walsh, P. C. et al: Risk of prostate cancer-specific mortality following biochemical recurrence after radical prostatectomy. JAMA, 294: 433, 2005
12.
13. 14.
Albertsen, P. C., Hanley, J. A., Penson, D. F. and Fine, J.: Validation of increasing prostate specific antigen as a predictor of prostate cancer death after treatment of localized prostate cancer with surgery or radiation. J Urol, 171: 2221, 2004 Pound, C. R., Partin, A. W., Eisenberger, M. A., Chan, D. W., Pearson, J. D. and Walsh, P. C.: Natural history of progression after PSA elevation following radical prostatectomy. JAMA, 281: 1591, 1999 Okotie, O. T., Aronson, W. J., Wieder, J. A., Liao, Y., Dorey, F., DeKernion, J. B. et al: Predictors of metastatic disease in men with biochemical failure following radical prostatectomy. J Urol, 171: 2260, 2004 Ward, J. F., Blute, M. L., Slezak, J., Bergstralh, E. J. and Zincke, H.: The long-term clinical impact of biochemical recurrence of prostate cancer 5 or more years after radical prostatectomy. J Urol, 170: 1872, 2003 Soergel, T. M., Koch, M. O., Foster, R. S., Bihrle, R., Wahle, G., Gardner, T. et al: Accuracy of predicting long-term prostate specific antigen outcome based on early prostate specific antigen recurrence results after radical prostatectomy. J Urol, 166: 2198, 2001 Stephenson, A. J., Shariat, S. F., Zelefsky, M. J., Kattan, M. W., Butler, E. B., Teh, B. S. et al: Salvage radiotherapy for recurrent prostate cancer after radical prostatectomy. JAMA, 291: 1325, 2004 Moul, J. W., Wu, H., Sun, L., McLeod, D. G., Amling, C., Donahue, T. et al: Early versus delayed hormonal therapy for prostate specific antigen only recurrence of prostate cancer after radical prostatectomy. J Urol, 171: 1141, 2004 Iversen, P., Tammela, T. L., Vaage, S., Lukkarinen, O., Lodding, P., Bull-Njaa, T. et al: A randomised comparison of bicalutamide (‘Casodex’) 150 mg versus placebo as immediate therapy either alone or as adjuvant to standard care for early non-metastatic prostate cancer. First report from the Scandinavian Prostatic Cancer Group Study No. 6. Eur Urol, 42: 204, 2002 Crawford, E. D.: Early versus late hormonal therapy: debating the issues. Urology, 61: 8, 2003 Messing, E. M., Manola, J., Sarosdy, M., Wilding, G., Crawford, E. D. and Trump, D.: Immediate hormonal therapy compared with observation after radical prostatectomy and pelvic lymphadenectomy in men with node-positive prostate cancer. N Engl J Med, 341: 1781, 1999
Purpose: In patients treated with radical prostatectomy who have biochemical recurrence we have previously reported that time from surgery to biochemical recurrence and postoperative prostate specific antigen doubling time are significantly related to the risk of prostate cancer death. We performed a more thorough examination of the association of time from surgery to biochemical recurrence and the risk of prostate cancer death. Materials and Methods: We retrospectively studied the records of 379 patients treated with radical prostatectomy between 1982 and 2000 who had had biochemical recurrence. We examined the association of time from surgery to prostate specific antigen recurrence and prostate specific antigen doubling time, and the risk of prostate cancer death using the Spearman correlation and Cox proportional hazards regression, respectively. Results: Longer time from surgery to prostate specific antigen recurrence was associated with a slower prostate specific antigen doubling time (Spearman r ⫽ 0.36, p ⬍0.001) and a decreased risk of prostate cancer death (RR 0.76, 95% CI 0.66 to 0.88, p ⬍0.001). The 15-year actuarial prostate cancer specific survival rate after biochemical recurrence in patients with recurrence at 3 years or less was 41% (95% CI 29 to 53) compared to 87% (95% CI 75 to 93) in patients with recurrence more than 3 years after radical prostatectomy. On multivariate analysis a shorter time from surgery to prostate specific antigen recurrence was associated with an increased risk of prostate cancer death (3 or less vs more than 3 years, RR 2.70, 95% CI 1.37 to 5.31, p ⫽ 0.004). Conclusions: Earlier prostate specific antigen recurrence is associated with an increased risk of prostate cancer death. These data suggest that perhaps time to prostate specific antigen recurrence may be a reasonable intermediate end point in patients treated with radical prostatectomy, although this must be validated in other studies. Key Words: prostate, prostatic neoplasms, prostatectomy, prostate-specific antigen, mortality
he common use of RP has resulted in large databases containing data on many patients treated with RP, which have provided valuable insight into prostate cancer biology. After prostatectomy essentially all PSA producing cells are removed. Therefore, slight increases in PSA, eg 0.2 or greater,1 or 0.4 or greater2 ng/ml, are used to indicate cancer recurrence and often as intermediate study end points. Recent data suggested that post-recurrence PSADT may be a surrogate for PCSM.3 Similarly we4 and others5 have found that a short PSADT after biochemical recurrence is strongly associated with PCSM. In light of these data does time to biochemical recurrence remain an
T
important variable? If so, is it sufficiently important to justify using it as a study end point? We have previously studied 379 patients treated with RP at our institution with biochemical recurrence who had PSADT data available and found that, after controlling for PSADT, time to biochemical recurrence was significantly associated with PCSM.4 In the current study using the same cohort with slightly longer followup we performed more thorough analysis of the association between time to PSA recurrence, and postoperative PSADT and PCSM to assess whether time to PSA recurrence is a reasonable intermediate end point in patients treated with RP.
Submitted for publication October 14, 2005. Study received approval from the Institutional Review Board, Johns Hopkins University. Supported by National Institute of Health/National Cancer Institute-SPORE Grant P50CA58236, The Prostate Cancer Foundation, Department of Defense Prostate Cancer Research Program PC030666 and an American Foundation for Urological Disease/ American Urological Association Education and Research Scholarship Award. Views and opinions of, and endorsements by the author(s) do not reflect those of the United States Army or Department of Defense. * Correspondence and current address: Department of Surgery, Box 3850, Duke University Medical Center, Durham, North Carolina 27710 (telephone: 919-668-8361; FAX: 919-668-7093; e-mail: [email protected]).
MATERIALS AND METHODS
0022-5347/06/1764-1404/0 THE JOURNAL OF UROLOGY® Copyright © 2006 by AMERICAN UROLOGICAL ASSOCIATION
Patient Population The Institutional Review Board at Johns Hopkins University approved this study and, when required, written informed consent was obtained from all subjects. We evaluated data on 5,096 patients with prostate adenocarcinoma treated with RP at our institution from April 1982 to October 2000 who had followup data available. During a mean followup ⫾ SD of 6.3 ⫾ 4.5 years (median 5) 997 patients (20%) experienced biochemical recurrence, defined as a single postoperative PSA of 0.2 ng/ml or greater.1 Of these 997
1404
Vol. 176, 1404-1408, October 2006 Printed in U.S.A. DOI:10.1016/j.juro.2006.06.017
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH patients 411 had data available that included 2 or more PSA values separated by 3 months or more within 2 years after biochemical recurrence and who did not receive adjuvant radiation or hormonal therapy before recurrence. Two patients who received preoperative radiation and 10 who received hormonal therapy were excluded. Another 20 patients were excluded who received salvage radiation with a durable PSA response of more than 2 years, and who were considered to have only local recurrence and to have been cured by surgery plus radiation. Another 22 patients who underwent salvage radiation but did not achieve a durable PSA response were considered to have distant failure and they were included. This resulted in 379 patients. Prostate cancer death was defined as death in any patient with metastasis that showed progression following hormonal therapy or death in any patient not treated with hormonal therapy but with widespread metastases without another obvious cause of death. Because of the limited number of nonprostate cancer deaths (18) in this relatively healthy cohort, PCSM approached all cause mortality and, therefore, data are only reported for PCSM.
Determination of PSADT PSADT was calculated by the natural log of 2 (0.693) divided by the slope of the linear regression line of natural log of all PSA values obtained within the first 2 years after biochemical recurrence.6 As previously described, PSADT was categorize as less than 3, 3.0 to 8.9 and 9.0 to 14.9 vs 15 or greater months.4 Seven patients with negative or zero PSADT (no PSA increase) were assigned to the 15 months or greater PSADT category. All PSA values were obtained after biochemical recurrence and before subsequent radiation or hormonal therapy. Therefore, subsequent therapy had no impact on PSADT calculation. In a subset of fewer than 100 patients frozen banked serum collected before the introduction of the PSA assay was used to measure PSA retrospectively.
Statistical Analysis The association between time to PSA recurrence and PSADT was examined using the Spearman correlation. The association between time to PSA recurrence and time from PSA recurrence to PCSM was examined using Cox proportional hazards regression. For multivariate analysis a forward stepwise Cox proportional hazards model was used with p ⬍0.15 determining which variables would be entered into the model at each step. Variables considered for entry were preoperative PSA, PSADT (categorical variable), pathological Gleason sum (7 or less vs 8 or greater), time from surgery to biochemical recurrence, patient age at RP (continuous), surgical margin status, extraprostatic extension, seminal vesicle invasion and lymph node metastasis. Time from surgery to PSA recurrence was examined as a continuous and as a categorical (3 or less vs more than 3 years) variable because we have previously found that this cutoff point provides the greatest likelihood chi-square ratio for estimating time to PCSM.4 All statistical analyses were performed using STATA® 8.0.
1405
TABLE 1. Preoperative clinical and pathological characteristics in 379 patients with PSA recurrence after radical prostatectomy No. pts Mean age at surgery ⫾ SD Preop PSA (ng/ml): Mean ⫾ SD Median No. biopsy Gleason score (%): 2–6 7 8–10 No. race (%): White Black Nonblack-nonwhite No. clinical stage (%): T1b T1c T2a T2b T2c T3 No. pathological Gleason score (%): 2–6 7 8–10 No. pos surgical margins (%) No. extracapsular extension (%) No. seminal vesicle invasion (%) No. pos lymph nodes (%)
379 59.7 ⫾ 6.2 14.4 ⫾ 15.0 10.6 173 (47) 146 (39) 53 (14) 360 (95) 14 (4) 5 (1) 6 (2) 73 (19) 130 (34) 119 (32) 29 (8) 20 (5) 54 (14) 195 (51) 130 (34) 178 (47) 337 (89) 143 (38) 124 (33)
RESULTS Patient Characteristics and Overall Survival Most patients had pathologically extraprostatic disease and high grade disease (table 1). Mean followup after surgery ⫾ SD in the 379 patients was 10.8 ⫾ 4.9 years (median 10, range 2 to 21). Mean time to biochemical recurrence was 3.5 ⫾ 3.0 years (median 2) and mean followup after biochemical recurrence was 7.3 ⫾ 3.9 years (median 7). During this time 76 of the 379 patients (20%) died of prostate cancer. Median prostate cancer specific survival in the study cohort of 379 patients with PSADT data available who experienced biochemical recurrence was 16 years. In this cohort 5, 10 and 15-year cause specific survival from the time of biochemical recurrence was 93% (95% CI 90 to 95), 75% (95% CI 68 to 80) and 53% (95% CI 42 to 64), respectively. Association Between Time to PSA Recurrence and PSADT There was a significant association between time to PSA recurrence and post-recurrence PSADT with later recurrence associated with slower PSADT (Spearman r ⫽ 0.36, p ⬍0.001). While overall median PSADT was 12.1 months, median PSADT was 9.2 months in patients with recurrence at 3 years or less compared to 17.8 months in patients with recurrence more than 3 years after surgery (fig. 1). Overall 6% of patients had a PSADT of less than 3 months. However, 19 of 244 patients (8%) with recurrence at 3 years or less had a PSADT of less than 3 months compared to only 4 of 135 (3%) with recurrence more than 3 years after surgery (table 2). Time to PSA Recurrence and Risk of PCSM Each 1-year delay in time to PSA recurrence was associated with a 24% decreased risk of PCSM on univariate analysis (RR 0.76, 95% CI 0.66 to 0.88, p ⬍0.001). The risk of PCSM appeared relatively stable in patients with recurrence up
1406
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH TABLE 3. Time from surgery to biochemical recurrence and risk of prostate cancer specific mortality in 379 men with biochemical recurrence after radical prostatectomy No. Prostate Ca Deaths
FIG. 1. Box plot shows association between time from surgery to PSA recurrence and post-recurrence doubling time with men with PSADT more than 100 months assigned value of 100 months.
through year 3. It was decreased in men with recurrence in year 4 and then it appeared relatively stable from years 5 to 8, after which there only 1 prostate cancer death was noted (table 3). The 15-year actuarial prostate cancer specific survival after biochemical recurrence in patients with recurrence at 3 years or less was 41% (95% CI 29 to 53) compared to 87% (95% CI 75 to 93) in patients with recurrence more than 3 years after surgery (fig. 2). On multivariate analysis only PSADT, pathological Gleason sum and time to PSA recurrence (as a continuous and a categorical variable) were significantly associated with time to PCSM (table 4). A total of 54 patients (14%) received hormonal therapy before metastatic disease developed, of whom 7 went on to metastatic disease, while 47 remained metastasis-free at last followup. After excluding these 54 men and adjusting for PSADT and pathological Gleason sum time to PSA recurrence remained associated with PCSM as a continuous variable (RR 0.87, 95% CI 0.75 to 1.01, p ⫽ 0.07) and as a categorical variable (3 or fewer vs more than 3 years RR 2.78, 95% CI 1.33 to 5.82, p ⫽ 0.01). DISCUSSION
Surgery—PSA Recurrence (yrs)
No. Pts
Observed
Expected
Observed/Expected Prostate Ca Death Ratio
1 2 3 4 5 6 7–8 9 or Greater
123 74 46 36 21 26 10 15
32 24 10 6 1 1 1 1
24.03 14.79 9.15 10.34 3.98 4.79 4.95 3.97
1.33 1.62 1.09 0.58 0.25 0.21 0.20 0.25
months to calculate PSADT. Therefore, what information can we give to patients at the time of recurrence? In addition, the use of time to PSA recurrence as an end point for study avoids the complexity involved in calculating PSADT, eg how many PSA measurements are required during what time period, etc. However, data linking early PSA recurrence to PCSM has largely been lacking. In the current study we found that early PSA recurrence was associated with shorter post-recurrence PSADT, which is a known poor prognostic factor.3,4,6 – 8 Moreover, even after adjusting for PSADT and pathological Gleason sum early biochemical recurrence was associated with an increased risk of PCSM. This suggests that early PSA recurrence may be a reasonable end point in studies of patients treated with RP. It is becoming increasingly clear that many patients who experience PSA recurrence have a prolonged course until metastasis6 or PCSM.4 To better risk stratify these patients, groups have attempted to identify characteristics that predict poor outcomes. Specifically several studies describe that post-recurrence PSADT is strongly prognostic for metastasis6 – 8 and PCSM.3,4 Indeed, in the current study PSADT was one of the strongest predictors of PCSM. However, clinical application of PSADT has limitations. 1) It requires several PSA values to calculate it. 2) One group suggested that PSADT calculations using low PSA values may not be accurate for predicting “eventual PSADT.”9 Indeed, we have previously found that using all PSA values within 2 years
Due to the sensitivity of PSA for the early detection of disease recurrence after RP, PSA recurrence can predate metastasis by years.6 At the time of PSA recurrence, eg PSA 0.2 ng/ml, PSADT calculations may not be accurate. Moreover, it requires several measurements during multiple
TABLE 2. Time from surgery to PSA recurrence and postoperative PSADT in 379 patients with PSA recurrence after radical prostatectomy Surgery– PSA Recurrence (yrs)
15.0 or Greater
9.0–14.9
3.0–8.9
Less than 3.0
1 2 3 4 5 6 7–8 9 or Greater
35 (28) 23 (31) 18 (39) 20 (56) 14 (67) 17 (65) 13 (54) 18 (64)
16 (13) 19 (25) 15 (33) 6 (17) 3 (14) 5 (19) 8 (33) 7 (25)
57 (46) 30 (40) 12 (26) 8 (22) 3 (14) 4 (15) 2 (8) 3 (11)
15 (12) 3 (4) 1 (2) 2 (6) 1 (5) 0 1 (4) 0
PSADT (mos)
FIG. 2. Kaplan-Meier 15-year actuarial prostate cancer specific survival curves after PSA recurrence in patients with PSA recurrence, stratified by PSA recurrence 3 or less vs more than 3 years after surgery (log rank p ⬍0.001).
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH TABLE 4. Multivariate Cox proportional hazards analysis of factors predicting time to prostate cancer specific death after PSA recurrence following radical prostatectomy RR
95% CI
p Value
Time from surgery to PSA recurrence as continuous variable Yrs from surgery to PSA recurrence 0.87 0.75–1.00 0.04 (continuous) PSA doubling time (vs 15.0 mos or greater): Less than 3 mos 21.49 8.78–52.57 ⬍0.001 3.0–8.9 Mos 9.23 4.23–20.15 ⬍0.001 9.0–14.9 Mos 3.28 1.37–7.83 0.01 Pathological Gleason sum (8 or 2.13 1.31–3.46 0.002 greater vs less than 8) Time from surgery to PSA recurrence as categorical variable Yrs from surgery to PSA recurrence 2.70 1.37–5.31 0.004 (3 or less vs greater than 3) PSA doubling time (vs 15.0 mos or greater): Less than 3 mos 22.97 9.45–55.81 ⬍0.001 3.0–8.9 Mos 8.82 4.05–19.20 ⬍0.001 9.0–14.9 Mos 2.92 1.22–6.99 0.02 Pathological Gleason sum (8 or 2.22 1.37–3.61 0.001 greater vs less than 8)
after recurrence rather than the first 2 PSA values only provided better risk assessment for predicting metastasis.6 This suggests that perhaps multiple PSA values during an extended period may be needed to accurately measure PSADT. 3) Given that salvage radiation is most beneficial when given at lower PSA values,10 many patients today are treated with salvage therapy before accurate PSADT measurement is ever available. Given these potential limitations to calculating and using PSADT for risk assessment, we examined an alternative end point in patients treated with RP, that is time to PSA recurrence. We have previously reported outcomes in this cohort,4 although the current study contains a slightly longer followup (mean 10.8 vs 10.3 years after surgery) and more prostate cancer deaths (76 vs 66). In our prior study, time to PSA recurrence, which was significantly related to PCSM, was examined as 3 or less vs more than 3 years. This was done to facilitate the construction of easy to use tables assessing the risk of PCSM. However, we hypothesized that the association between time to PSA recurrence and PCSM would be on a continuum, rather than have a discreet 3 year cutoff. Therefore, in the current study we performed more thorough examination of the association between time to PSA recurrence and PCSM. We found a steady lengthening of PSADT and a decrease in the risk of PCSM as time to PSA recurrence increased. Specifically for each 1-year delay in biochemical recurrence the risk of prostate cancer death decreased 24%. Thus, although in our data a cutoff of 3 or less vs more than 3 years provided the best PCSM stratification, time to PSA recurrence should really be viewed as a continuum of risk. This is an important observation that was not presented in our initial description of this cohort.4 In the current study the best single cutoff point for stratifying patients at low vs high risk was PSA recurrence 3 or less vs more than 3 years. This cutoff differs slightly from that in an earlier study including some patients in the current cohort.6 In that study 2 or less vs more than 2 years provided the best risk stratification. We explored whether this slight difference in results reflected the changing nature of our patients with time or differing patient populations. In the current study earlier year of surgery was significantly
1407
associated with longer time to recurrence (Spearman r ⫽ ⫺0.438, p ⬍0.001), likely reflecting that in men treated in the pre-PSA era only those with PSA data available could be included in the analysis. Consequently men with early recurrence would necessarily be excluded from this study. Likewise in men treated in the most recent years not enough time has elapsed to capture all PSA recurrences and allow enough time for measuring PSADT. Thus, only those men with early recurrence would have been included. Given the limited followup and long natural history of prostate cancer, it is currently too premature to examine whether time to recurrence remains significantly prognostic in men treated in recent years. Analogous to the current study, researchers from the Mayo Clinic found a significant association between early PSA recurrence after RP and a short PSADT.8 However, despite this association and the fact that short PSADT strongly predicted clinical progression in that series, early PSA recurrence was surprisingly not associated with clinical progression. This may reflect the increased use of early salvages therapy in men with early recurrence and censoring of these patients relative to lower rates of salvage therapy in men with delayed recurrence in that series. In contrast, in the current study there was no significant association between early PSA recurrence and early hormonal therapy (chi-square p ⫽ 0.32). Alternative explanations for different findings between the prior8 and the current study are different time to PSA recurrence cutoff points and different patient populations. After validation in other studies the finding that time to PSA recurrence is significantly related to PCSM would have important clinical implications. For instance, if the link between time to PSA recurrence and PCSM can be firmly enough established, this would potentially justify the design of adjuvant clinical trials using time to PSA recurrence as an end point. Given the long natural history of PSA recurrence, having an earlier end point to use in these trials would be extremely helpful and it would speed the process of bringing potential lifesaving drugs to the market. However, the current findings must first be validated in other series, ideally in the context of a prospective clinical trial. In the current study only 54 men received early hormonal therapy. It is unlikely that these men dramatically influenced our findings since excluding these men did not materially change our results. The current study, in which most patients did not receive early hormonal therapy, allows a unique insight into the natural history of prostate cancer. However, because hormonal therapy can delay metastasis,11,12 many patients today receive hormonal therapy before the development of metastasis. It is debatable whether early hormonal therapy in patients with biochemical recurrence after RP delays PCSM.13 If early hormonal therapy dramatically improves survival,14 the estimates generated in the current study may underestimate the likelihood of survival. PSA and PSADT are time dependent factors. We calculated PSADT while assuming that PSA increased exponentially during the first 2 years after recurrence. Although PSADT may not have been calculable in some men until the end of year 2 because no second PSA value was available until then, calculated PSADT at that point was presumed to be PSADT at the time of biochemical recurrence. However, it
1408
TIME TO PROSTATE SPECIFIC ANTIGEN RECURRENCE AND PROSTATE CANCER DEATH
is certainly plausible that in the long term PSADT values may not be stable.
5.
CONCLUSIONS 6.
In 379 patients who experienced biochemical recurrence after RP, shorter time from surgery to PSA recurrence was associated with more rapid post-recurrence PSADT and an increased risk of PCSM. Patients with recurrence within 3 years after surgery were at particularly high risk for PCSM. Even after controlling for PSADT and pathological Gleason sum, a shorter time from surgery to PSA recurrence was associated with an increased risk of PCSM. Time from surgery to PSA recurrence may be a reasonable intermediate end point in patients treated with RP.
7.
8.
9.
Abbreviations and Acronyms PCSM PSA PSADT RP
⫽ ⫽ ⫽ ⫽
prostate cancer specific mortality prostate specific antigen PSA doubling time radical prostatectomy
10.
11.
REFERENCES 1.
Freedland, S. J., Sutter, M. E., Dorey, F. and Aronson, W. J.: Defining the ideal cutpoint for determining PSA recurrence after radical prostatectomy. Urology, 61: 365, 2003 2. Amling, C. L., Bergstralh, E. J., Blute, M. L., Slezak, J. M. and Zincke, H.: Defining prostate specific antigen progression after radical prostatectomy: what is the most appropriate cut point? J Urol, 165: 1146, 2001 3. D’Amico, A. V., Moul, J. W., Carroll, P. R., Sun, L., Lubeck, D. and Chen, M. H.: Surrogate end point for prostate cancerspecific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst, 95: 1376, 2003 4. Freedland, S. J., Humphreys, E. B., Mangold, L. A., Eisenberger, M., Dorey, F. J., Walsh, P. C. et al: Risk of prostate cancer-specific mortality following biochemical recurrence after radical prostatectomy. JAMA, 294: 433, 2005
12.
13. 14.
Albertsen, P. C., Hanley, J. A., Penson, D. F. and Fine, J.: Validation of increasing prostate specific antigen as a predictor of prostate cancer death after treatment of localized prostate cancer with surgery or radiation. J Urol, 171: 2221, 2004 Pound, C. R., Partin, A. W., Eisenberger, M. A., Chan, D. W., Pearson, J. D. and Walsh, P. C.: Natural history of progression after PSA elevation following radical prostatectomy. JAMA, 281: 1591, 1999 Okotie, O. T., Aronson, W. J., Wieder, J. A., Liao, Y., Dorey, F., DeKernion, J. B. et al: Predictors of metastatic disease in men with biochemical failure following radical prostatectomy. J Urol, 171: 2260, 2004 Ward, J. F., Blute, M. L., Slezak, J., Bergstralh, E. J. and Zincke, H.: The long-term clinical impact of biochemical recurrence of prostate cancer 5 or more years after radical prostatectomy. J Urol, 170: 1872, 2003 Soergel, T. M., Koch, M. O., Foster, R. S., Bihrle, R., Wahle, G., Gardner, T. et al: Accuracy of predicting long-term prostate specific antigen outcome based on early prostate specific antigen recurrence results after radical prostatectomy. J Urol, 166: 2198, 2001 Stephenson, A. J., Shariat, S. F., Zelefsky, M. J., Kattan, M. W., Butler, E. B., Teh, B. S. et al: Salvage radiotherapy for recurrent prostate cancer after radical prostatectomy. JAMA, 291: 1325, 2004 Moul, J. W., Wu, H., Sun, L., McLeod, D. G., Amling, C., Donahue, T. et al: Early versus delayed hormonal therapy for prostate specific antigen only recurrence of prostate cancer after radical prostatectomy. J Urol, 171: 1141, 2004 Iversen, P., Tammela, T. L., Vaage, S., Lukkarinen, O., Lodding, P., Bull-Njaa, T. et al: A randomised comparison of bicalutamide (‘Casodex’) 150 mg versus placebo as immediate therapy either alone or as adjuvant to standard care for early non-metastatic prostate cancer. First report from the Scandinavian Prostatic Cancer Group Study No. 6. Eur Urol, 42: 204, 2002 Crawford, E. D.: Early versus late hormonal therapy: debating the issues. Urology, 61: 8, 2003 Messing, E. M., Manola, J., Sarosdy, M., Wilding, G., Crawford, E. D. and Trump, D.: Immediate hormonal therapy compared with observation after radical prostatectomy and pelvic lymphadenectomy in men with node-positive prostate cancer. N Engl J Med, 341: 1781, 1999