- Email: [email protected]
What does postradiotherapy PSA nadir tell us about freedom from PSA failure and progression-free survival in patients with low and intermediate-risk localized prostate cancer?
ADULT UROLOGY CME ARTICLE
WHAT DOES POSTRADIOTHERAPY PSA NADIR TELL US ABOUT FREEDOM FROM PSA FAILURE AND PROGRESSION-FREE SURVIVAL IN PATIENTS WITH LOW AND INTERMEDIATE-RISK LOCALIZED PROSTATE CANCER? K. D. DEWITT, H. M. SANDLER, V. WEINBERG, P. W. MCLAUGHLIN,
AND
M. ROACH III
ABSTRACT Objectives. To determine whether the post-external beam radiotherapy (RT) prostate-specific antigen nadir (nPSA) improves our ability to predict freedom from PSA failure, progression-free survival (PFS), and overall survival. Controversy regarding the importance of nPSA after external beam RT as a prognostic indicator for patients with localized prostate cancer has continued. Methods. This analysis was based on the data from 748 patients with low and intermediate-risk localized prostate cancer treated with external beam RT alone. Patients were categorized by nPSA quartile groups with cutpoints of less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or greater. Both univariate and multivariate analyses were used to determine the significance of nPSA on PSA failure (American Society for Therapeutic Radiology Oncology consensus definition), PFS (death after PSA failure), and overall survival (death from any cause). Results. Freedom from PSA failure was strongly associated with nadir quartile groups (P ⬍0.0001). PFS was also significantly different statistically among nadir quartile groups (P ⫽ 0.02). No statistically significant difference was found in overall survival associated with nPSA at this point. Conclusions. nPSA is a strong independent predictor of freedom from PSA failure and PFS in patients with low and intermediate-risk localized prostate cancer treated with RT alone. Longer follow-up and larger patient numbers are required to confirm these observations. UROLOGY 62: 492–496, 2003. © 2003 Elsevier Inc.
G
reat controversy exists regarding the value of the postradiotherapy (RT) prostate-specific antigen nadir (nPSA) as a predictor of clinical outcome among patients with localized prostate cancer. Although numerous studies have established the importance of PSA monitoring after external beam RT (EBRT),1–5 no consensus has been reached as to whether the nPSA is significant in This work was funded in part by generous unrestricted support from Roy Howard. From the Departments of Radiation Oncology, Medical Oncology, and Urology, Comprehensive Cancer Center, University of California, San Francisco, School of Medicine, San Francisco, California; and Department of Radiation Oncology, University of Michigan School of Medicine, Ann Arbor, Michigan Reprint requests: Mack Roach III, M.D., Department of Radiation Oncology, University of California, San Francisco, School of Medicine, 1600 Divisadero Street, Box 1708, Suite H-1031, San Francisco, CA 94143-1708 Submitted: December 6, 2002, accepted (with revisions): April 10, 2003
492
© 2003 ELSEVIER INC. ALL RIGHTS RESERVED
terms of progression-free survival (PFS) or overall survival (OS).6 Several investigators have suggested varying nPSA cutpoints as significant predictors of freedom from PSA failure (FFPF). Critz et al.7–10 used 0.2 and 0.5 ng/mL, Hanks et al.11 and Horwitz et al.12 used an nPSA of 1.5 ng/mL to predict biochemical treatment failure, and Perez et al.6 used an nPSA of 1.0 ng/mL. A consensus as to which of these cutpoints holds the most predictive value has not been met. Moreover, it has not been established whether similar nPSA values predict FFPF for different treatment modalities such as brachytherapy and EBRT. Recent data from Hanlon13 and Hanks et al.11 have suggested a strong association between posttreatment nPSA and distant failure, as well as death from prostate cancer. This relationship between the post-treatment biochemical profile and PFS is critical in establishing the significance of the nPSA 0090-4295/03/$30.00 doi:10.1016/S0090-4295(03)00460-6
as a prognostic indicator, because biochemical failure has not been proven as a surrogate for any measurable clinical outcome. In this report, we analyzed the significance of nPSA in predicting FFPF and PFS in patients treated with EBRT for low and intermediate-risk prostate cancer. Because it is rare to die of prostate cancer in the absence of biochemical failure, establishing an nPSA as a marker for FFPF and PFS might enable us to identify, at an early point, a group of patients for whom early aggressive salvage intervention and applicable clinical trials would be most efficacious.14 MATERIAL AND METHODS This study included 924 patients treated with EBRT alone (no hormonal therapy allowed) between 1987 and 1998, with 95% treated before 1997, for whom complete data for staging was available, including pretreatment PSA (pPSA), Gleason score, and T stage. The average patient age was 70.6 years (range 49 to 90). A minimum of 24 months of PSA follow-up was available for each censored patient. The selection of patients to be included in this analysis was based on the risk of disease outside the locoregional field. Thus, our analysis included patients with low and intermediate-risk disease for whom EBRT alone was a reasonable treatment option.15 To eliminate the potential impact of micrometastatic disease on nPSA, 125 patients at high risk of subclinical metastatic disease as determined by the presenting disease features were excluded from this data analysis. This exclusion included patients with Gleason score 8 to 10 and Stage T2c bulky disease or Stage T3 disease; Stage T2c-T3 with Gleason score 7 and pPSA level greater than 20 ng/mL; and Stage T12a,b with Gleason score 8 to 10 and pPSA greater than 20 ng/mL. Patients with these presenting features have been shown to have a high risk of micrometastatic disease in several analyses.15 Locoregional RT would be expected to fail in all patients with metastatic disease regardless of an nPSA. Thirty-seven patients were excluded because of incomplete PSA follow-up, and two were excluded because of hormonal intervention. Additionally, 12 patients were excluded because of clinical failure (local or distant) within 1 year of completing EBRT, because it was assumed that these patients had existing, but undetected, metastatic disease at the time of treatment. The median follow-up was 5 years after RT completion, with a minimal follow-up of 2 years. PSA failure was defined according to the American Society for Therapeutic Radiology Oncology definition requiring three consecutive rising PSA measurements.16 The date of failure was the midpoint between the value before the three rising PSA values and the first of the three rising values. The duration of FFPF was the difference between the end of EBRT and the date of failure or the date of the last PSA measurement if no failure occurred. Because competing causes of death are likely to obscure the impact of nPSA, particularly in an elderly population, PFS was presented as a “worst-case” estimate of outcome and defined as death after PSA failure. The rationale for this definition was that PFS defined in this way would be more relevant to a younger population of men not at high risk of death from comorbid conditions. This definition has been used in prior studies that established pPSA as an important prognostic indicator in men with low and intermediate-risk prostate cancer treated with RT alone.17 Local or metastatic disease was not required for death to be considered as prostate cancer related. UROLOGY 62 (3), 2003
The duration of PFS was measured from the end of EBRT until death or the date last known to be alive. The nPSA was the lowest recorded PSA value after EBRT before failure in the event this occurred. If no biochemical failure occurred, the nPSA was defined as the lowest PSA after RT completion. If the PSA only increased and the patient had at least 1 year of follow-up, the last pre-EBRT PSA was recorded as the nadir value, with the time to nadir equal to 0. Because all high-risk patients were excluded, this occurred in only 3 patients. Patients were analyzed by nPSA quartile group (less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or greater). This approach of having cutpoints at quartile intervals was chosen to have an unbiased comparison of the relationship between the nPSA and FFPF/PFS. Distributions of pretreatment features among the four nadir groups were compared using a chi-square statistic for categorical variables (eg, Gleason score), analysis of variance methods for continuous variables (eg, age), and the nonparametric Kruskal-Wallis test to compare distributions (eg, PSA). Post hoc tests were used to identify which groups differed when overall significant differences occurred. The probability of FFPF, PFS, and OS was calculated using the Kaplan-Meier product limit method. Comparisons of distributions, including sequential pairwise comparisons between quartile groups, were performed using the log-rank test. Multivariate analyses to identify independent predictors of outcome were carried out using Cox’s proportional hazard model and the log likelihood test.
RESULTS A summary of the pretreatment patient characteristics is shown in Table I. Because patients with a high risk of metastatic disease were excluded from this analysis, only a minority of patients had T3 lesions, Gleason score 8 to 10, or Radiation Therapy Oncology Group (RTOG) risk group 3 within each quartile. However, no statistically significant difference was found in the distribution of T stage, Gleason score, or RTOG risk group among nadir quartile groups. The pPSA, percentage of change in PSA level, and time to nPSA differed significantly among the nadir quartile groups (all with P ⬍0.0001). FFPF BY NADIR QUARTILE GROUPS A statistically significant difference in FFPF was seen by nadir quartile group (P ⬍0.0001; Fig. 1). The 5-year FFPF estimate was 83%, 72%, 58%, and 33% for nPSA quartile group of less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or more, respectively. Sequential pairwise comparisons using the log-rank test between quartile groups were all significantly different, with those having nPSA levels of 1.2 ng/mL or greater having the poorest outcome (P ⫽ 0.03, 0.01, and ⬍0.0001). Although there is a progression between nPSA and FFPF whereby lower nPSA levels result in better biochemical outcomes, the cutoff of 1.2 ng/mL marks a very significant difference in prognosis in terms of FFPF. 493
TABLE I. Comparison of groups nPSA <0.3 ng/mL nPSA 0.3–<0.6 ng/mL nPSA 0.6–<1.2 ng/mL nPSA >1.2 ng/mL P Value n pPSA nPSA Change in PSA (%) Time to nPSA (mo) T stage (%) T1 T2 T3 Gleason score (%) 2–6 7 8–10 RTOG risk group (%) R1 R2 R3 Mean Dmax (Gy)
n ⫽ 175 6.9 0.1 97 26.5
n ⫽ 177 8.9 0.4 94 24.5
n ⫽ 223 11.8 0.8 92 25.6
n ⫽ 173 18.0 1.8 85 18.2
⬍0.0001 ⬍0.0001 ⬍0.0001
27 63 10
19 77 4
31 62 4
21 68 11
0.01
66 30 4
75 20 5
72 25 3
66 30 4
0.29
56 34 10 72.2
58 31 11 71.3
58 35 7 71.7
50 40 10 71.0
0.49 0.03
KEY: nPSA ⫽ nadir prostate-specific antigen; pPSA ⫽ pretreatment PSA; RTOG ⫽ Radiation Therapy Oncology Group; Dmax ⫽ maximal dose.
FIGURE 1. FFPF by nPSA quartile groups. Sequential pairwise comparisons using the log-rank test between quartile groups were all significantly different: less than 0.3 versus 0.3 to less than 0.6 ng/mL, P ⫽ 0.03; 0.3 to less than 0.6 versus 0.6 to less than 1.2 ng/mL, P ⫽ 0.01; 0.6 to less than 1.2 versus 1.2 ng/mL or greater, P ⬍0.0001.
A significant effect according to nPSA was seen among patients with a pPSA level less than 10 ng/mL (P ⬍0.0001) and 10 ng/mL or greater (P ⬍0.0001), again showing that nPSA is an independent prognostic indicator for FFPF and not simply a reflection of the pPSA. Univariate analysis indicated that Gleason score (less than 7 versus 7 versus greater than 7), T stage (1 versus 2 versus 3), pPSA (less than 4 versus 4 to less than 10 versus 10 to less than 20 versus 20 ng/mL or greater), RTOG risk group (1 versus 2 versus 3), and maximal dose (69 Gy or less versus greater than 69 to 77 Gy versus greater than 77 Gy) were significant predictors of FFPF among these 494
FIGURE 2. PFS by nPSA quartile groups.
low and intermediate-risk patients. Age was found to be of borderline significance (P ⫽ 0.06) favoring patients 65 years old or younger. Multivariate analysis comparing nPSA, RTOG risk group, pPSA, and age using Cox’s proportional hazard model identified nPSA to be the most important predictor of FFPF (P ⬍0.0001). pPSA and RTOG risk group were also independent predictors of FFPF (P ⬍0.0001 and P ⫽ 0.002, respectively). The results were identical whether variables were considered as continuous or categorical features. PFS AND OS PFS was also found to be statistically significantly different among nadir quartile groups (P ⫽ 0.02; Fig. 2). The probability of death after biochemical failure was greater with an increase in UROLOGY 62 (3), 2003
nPSA. Multivariate analysis also indicated that nPSA was a strong predictor of PFS (P ⫽ 0.001). pPSA was of borderline significance as a predictor of PFS (P ⫽ 0.09). No statistically significant difference was found in OS with death from any cause associated with nPSA. In contrast, using the same potential factors in modeling biochemical failure, age at the time of treatment was a significant predictor of OS (P ⫽ 0.03) and pPSA was of borderline statistical significance (P ⫽ 0.07) for these low and intermediate-risk patients. COMMENT PSA has become commonly accepted in the monitoring of men with adenocarcinoma of the prostate.18 The pPSA level has been shown to be a significant predictor of post-therapeutic outcome after definitive RT for prostate cancer.1,3,4,17,19,20 The use of PSA after treatment for prostate cancer as a marker of outcome has been less well established. After prostatectomy, PSA rapidly falls off to undetectable or near undetectable levels in most patients. After RT, PSA declines more slowly and often does not reach undetectable levels. This has resulted in considerable debate, resulting in the consensus statement regarding guidelines for PSA levels after RT.2,6 –10,16,21 nPSA levels have been widely used as a predictor of biochemical success or failure. However, several investigators have used various absolute post-RT nPSA levels to report biochemical treatment failure, ranging from 0.2 to 4.0 ng/mL.6,8,10 –12,14,22 No consensus has been reached as to which nPSA value conveys the most prognostic significance. Recent data from Hanlon13 have suggested that nPSA is a significant prognostic indicator for both distant metastasis and death from prostate cancer. That study found nPSA to have predictive value for both distant metastasis and death from prostate cancer when analyzed using cutpoints of less than 1.0, 1.1 to 2.0, and greater than 2.0 ng/mL. Patients with an nPSA greater than 2.0 ng/mL had a significantly worse outcome compared with the other nPSA groups with regard to both outcome measures.13 In this report, we confirm the observations made by prior studies that the nPSA level is a predictor of biochemical treatment failure. Although it is important to define the relationship between nPSA and FFPF, there is a potential for statistical bias inherent in comparing high nPSA levels and PSA failure. The remaining question is whether an intermediate value between 0.3 and 1.2 ng/mL really has prognostic significance. Our data suggest that nPSA does hold prognostic significance for FFPF within this range, with a lower risk of biochemical failure associated with a lower nPSA. Longer folUROLOGY 62 (3), 2003
low-up and randomized prospective trials are necessary to confirm the trend seen in this report and others. This emphasizes the importance of establishing the relationship between nPSA and PFS. Our study addressed this issue and confirmed that nPSA is also a strong predictor of disease-free survival in low and intermediate-risk patients. The analysis of nPSA by quartile group was chosen to provide an objective way of measuring the significance of this continuous variable. By looking at nPSA as a categorical variable, we did not choose a cutpoint that unfavorably weighted the nPSA value in a multivariate analysis, leading to a potentially biased result. Our goal, therefore, was not to provide an absolute nPSA value as the consensus endpoint as in other studies,6,8,10 –12,14 but to show that nPSA is a significant predictor of FFPF, as well as disease-free survival. Our data do not support the previously published data by Critz et al.,10 which stated that all patients with an nPSA level greater than 1.0 ng/mL had biochemical evidence of recurrence at a minimum of 5 years of follow-up after RT completion. For the patients in our analysis, the 5-year Kaplan-Meier estimate of FFPF for patients achieving a minimal PSA greater than 1 ng/mL was 32%. This difference in outcome is likely explained by the different treatment methods between the two groups. All patients in the analysis by Critz et al.10 received permanent iodine-125 seed implantation in addition to EBRT, and the patients in our study received EBRT alone. This suggests that the target nPSA is different for these two different treatment modalities. The results of our multivariate analysis of nPSA, pPSA, RTOG risk group, and age demonstrated that a relationship exists between nPSA and death from prostate cancer. Although our analysis using quartile group cutpoints did not show pPSA to be an independent predictor of PFS, it is possible that higher cutpoints or longer follow-up may yield positive results in this regard. Multivariate analysis did not show nPSA to be an independent prognostic variable for OS. Although this may have been a result of advanced age and competing comorbidities in the patient cohort, longer follow-up is required to establish the relationship between nPSA and OS definitively. CONCLUSIONS Multivariate analysis indicated nPSA after EBRT is strongly associated with FFPF and PFS in patients with low and intermediate-risk prostate cancer. An nPSA level of 1.2 ng/mL or greater strongly correlated with a worse prognosis regarding FFPF and PFS. The results of this study showed that nPSA is a prognostic indicator for biochemical failure and PFS. Longer follow-up and larger patient 495
numbers are required to confirm these observations. REFERENCES 1. Kaplan ID, Cox RS, and Bagshaw MA: Prostate specific antigen after external beam radiotherapy for prostatic cancer: followup. J Urol 149: 519 –522, 1993. 2. Kuban DA, el-Mahdi AM, and Schellhammer PF: Prostate-specific antigen for pretreatment prediction and posttreatment evaluation of outcome after definitive irradiation for prostate cancer. Int J Radiat Oncol Biol Phys 32: 307–316, 1995. 3. Pisansky TM, Cha SS, Earle JD, et al: Prostate-specific antigen as a pretherapy prognostic factor in patients treated with radiation therapy for clinically localized prostate cancer. J Clin Oncol 11: 2158 –2166, 1993. 4. Zagars GK: Serum PSA as a tumor marker for patients undergoing definitive radiation therapy. Urol Clin North Am 20: 737–747, 1993. 5. Zietman AL, Coen JJ, Shipley WU, et al: Radical radiation therapy in the management of prostatic adenocarcinoma: the initial prostate specific antigen value as a predictor of treatment outcome. J Urol 151: 640 –645, 1994. 6. Perez CA, Michalski JM, and Lockett MA: Chemical disease-free survival in localized carcinoma of prostate treated with external beam irradiation: comparison of American Society for Therapeutic Radiology and Oncology consensus or 1 ng/mL as endpoint. Int J Radiat Oncol Biol Phys 49: 1287– 1296, 2001. 7. Critz FA, Levinson AK, Williams WH, et al: Prostatespecific antigen nadir: the optimum level after irradiation for prostate cancer. J Clin Oncol 14: 2893–2900, 1996. 8. Critz FA, Levinson K, Williams WH, et al: Prostatespecific antigen nadir of 0.5 ng/mL or less defines disease freedom for surgically staged men irradiated for prostate cancer. Urology 49: 668 –672, 1997. 9. Critz FA, Levinson AK, Williams WH, et al: The PSA nadir that indicates potential cure after radiotherapy for prostate cancer. Urology 49: 322–326, 1997. 10. Critz FA, Williams WH, Holladay CT, et al: Post-treatment PSA or ⫽ 0.2 ng/mL defines disease freedom after radiotherapy for prostate cancer using modern techniques. Urology 54: 968 –971, 1999.
496
11. Hanks GE, Lee WR, and Schultheiss TE: Clinical and biochemical evidence of control of prostate cancer at 5 years after external beam radiation. J Urol 154(2 Pt 1): 456 –459, 1995. 12. Horwitz EM, Vicini FA, Ziaja EL, et al: An analysis of clinical and treatment related prognostic factors on outcome using biochemical control as an end-point in patients with prostate cancer treated with external beam irradiation. Radiother Oncol 44: 223–228, 1997. 13. Hanlon A: Posttreatment prostate-specific antigen nadir highly predictive of distant failure and death from prostate cancer. Int J Radiat Oncol Biol Phys 53: 297–303, 2002. 14. Lee WR, Hanlon AL, and Hanks GE: Prostate specific antigen nadir following external beam radiation therapy for clinically localized prostate cancer: the relationship between nadir level and disease-free survival. J Urol 156(2 Pt 1): 450 – 453, 1996. 15. Roach M III, Lu J, Pilepich MV, et al: Predicting longterm survival and the need for hormonal therapy: a metaanalysis of RTOG prostate cancer trials. Int J Radiat Oncol Biol Phys 47: 617–627, 2000. 16. Panel ASTRO Consensus Statement: Guidelines for PSA following radiation therapy. Int J Radiat Oncol Biol Phys 37: 1035–1041, 1997. 17. Roach M, Weinberg V, McLaughlin P, et al: Serum prostate specific antigen (PSA) and survival following external beam radiotherapy of the prostate. Urology 61: 730 –735, 2003. 18. von Eschenbach A, Ho R, Murphy GP, et al: American Cancer Society guidelines for the early detection of prostate cancer: update, June 10, 1997. Cancer 80: 1805–1807, 1997. 19. Russell KJ, Dunatov C, Hafermann MD, et al: Prostate specific antigen in the management of patients with localized adenocarcinoma of the prostate treated with primary radiation therapy. J Urol 146: 1046 –1052, 1991. 20. Zagars GK: Prostate specific antigen as an outcome variable for T1 and T2 prostate cancer treated by radiation therapy. J Urol 152(5 Pt 2): 1786 –1791, 1994. 21. Geist RW: Reference range for prostate-specific antigen levels after external beam radiation therapy for adenocarcinoma of the prostate. Urology 45: 1016 –1021, 1995. 22. Zietman AL, Tibbs MK, Dallow KC, et al: Use of PSA nadir to predict subsequent biochemical outcome following external beam radiation therapy for T1-2 adenocarcinoma of the prostate. Radiother Oncol 40: 159 –162, 1996.
UROLOGY 62 (3), 2003
WHAT DOES POSTRADIOTHERAPY PSA NADIR TELL US ABOUT FREEDOM FROM PSA FAILURE AND PROGRESSION-FREE SURVIVAL IN PATIENTS WITH LOW AND INTERMEDIATE-RISK LOCALIZED PROSTATE CANCER? K. D. DEWITT, H. M. SANDLER, V. WEINBERG, P. W. MCLAUGHLIN,
AND
M. ROACH III
ABSTRACT Objectives. To determine whether the post-external beam radiotherapy (RT) prostate-specific antigen nadir (nPSA) improves our ability to predict freedom from PSA failure, progression-free survival (PFS), and overall survival. Controversy regarding the importance of nPSA after external beam RT as a prognostic indicator for patients with localized prostate cancer has continued. Methods. This analysis was based on the data from 748 patients with low and intermediate-risk localized prostate cancer treated with external beam RT alone. Patients were categorized by nPSA quartile groups with cutpoints of less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or greater. Both univariate and multivariate analyses were used to determine the significance of nPSA on PSA failure (American Society for Therapeutic Radiology Oncology consensus definition), PFS (death after PSA failure), and overall survival (death from any cause). Results. Freedom from PSA failure was strongly associated with nadir quartile groups (P ⬍0.0001). PFS was also significantly different statistically among nadir quartile groups (P ⫽ 0.02). No statistically significant difference was found in overall survival associated with nPSA at this point. Conclusions. nPSA is a strong independent predictor of freedom from PSA failure and PFS in patients with low and intermediate-risk localized prostate cancer treated with RT alone. Longer follow-up and larger patient numbers are required to confirm these observations. UROLOGY 62: 492–496, 2003. © 2003 Elsevier Inc.
G
reat controversy exists regarding the value of the postradiotherapy (RT) prostate-specific antigen nadir (nPSA) as a predictor of clinical outcome among patients with localized prostate cancer. Although numerous studies have established the importance of PSA monitoring after external beam RT (EBRT),1–5 no consensus has been reached as to whether the nPSA is significant in This work was funded in part by generous unrestricted support from Roy Howard. From the Departments of Radiation Oncology, Medical Oncology, and Urology, Comprehensive Cancer Center, University of California, San Francisco, School of Medicine, San Francisco, California; and Department of Radiation Oncology, University of Michigan School of Medicine, Ann Arbor, Michigan Reprint requests: Mack Roach III, M.D., Department of Radiation Oncology, University of California, San Francisco, School of Medicine, 1600 Divisadero Street, Box 1708, Suite H-1031, San Francisco, CA 94143-1708 Submitted: December 6, 2002, accepted (with revisions): April 10, 2003
492
© 2003 ELSEVIER INC. ALL RIGHTS RESERVED
terms of progression-free survival (PFS) or overall survival (OS).6 Several investigators have suggested varying nPSA cutpoints as significant predictors of freedom from PSA failure (FFPF). Critz et al.7–10 used 0.2 and 0.5 ng/mL, Hanks et al.11 and Horwitz et al.12 used an nPSA of 1.5 ng/mL to predict biochemical treatment failure, and Perez et al.6 used an nPSA of 1.0 ng/mL. A consensus as to which of these cutpoints holds the most predictive value has not been met. Moreover, it has not been established whether similar nPSA values predict FFPF for different treatment modalities such as brachytherapy and EBRT. Recent data from Hanlon13 and Hanks et al.11 have suggested a strong association between posttreatment nPSA and distant failure, as well as death from prostate cancer. This relationship between the post-treatment biochemical profile and PFS is critical in establishing the significance of the nPSA 0090-4295/03/$30.00 doi:10.1016/S0090-4295(03)00460-6
as a prognostic indicator, because biochemical failure has not been proven as a surrogate for any measurable clinical outcome. In this report, we analyzed the significance of nPSA in predicting FFPF and PFS in patients treated with EBRT for low and intermediate-risk prostate cancer. Because it is rare to die of prostate cancer in the absence of biochemical failure, establishing an nPSA as a marker for FFPF and PFS might enable us to identify, at an early point, a group of patients for whom early aggressive salvage intervention and applicable clinical trials would be most efficacious.14 MATERIAL AND METHODS This study included 924 patients treated with EBRT alone (no hormonal therapy allowed) between 1987 and 1998, with 95% treated before 1997, for whom complete data for staging was available, including pretreatment PSA (pPSA), Gleason score, and T stage. The average patient age was 70.6 years (range 49 to 90). A minimum of 24 months of PSA follow-up was available for each censored patient. The selection of patients to be included in this analysis was based on the risk of disease outside the locoregional field. Thus, our analysis included patients with low and intermediate-risk disease for whom EBRT alone was a reasonable treatment option.15 To eliminate the potential impact of micrometastatic disease on nPSA, 125 patients at high risk of subclinical metastatic disease as determined by the presenting disease features were excluded from this data analysis. This exclusion included patients with Gleason score 8 to 10 and Stage T2c bulky disease or Stage T3 disease; Stage T2c-T3 with Gleason score 7 and pPSA level greater than 20 ng/mL; and Stage T12a,b with Gleason score 8 to 10 and pPSA greater than 20 ng/mL. Patients with these presenting features have been shown to have a high risk of micrometastatic disease in several analyses.15 Locoregional RT would be expected to fail in all patients with metastatic disease regardless of an nPSA. Thirty-seven patients were excluded because of incomplete PSA follow-up, and two were excluded because of hormonal intervention. Additionally, 12 patients were excluded because of clinical failure (local or distant) within 1 year of completing EBRT, because it was assumed that these patients had existing, but undetected, metastatic disease at the time of treatment. The median follow-up was 5 years after RT completion, with a minimal follow-up of 2 years. PSA failure was defined according to the American Society for Therapeutic Radiology Oncology definition requiring three consecutive rising PSA measurements.16 The date of failure was the midpoint between the value before the three rising PSA values and the first of the three rising values. The duration of FFPF was the difference between the end of EBRT and the date of failure or the date of the last PSA measurement if no failure occurred. Because competing causes of death are likely to obscure the impact of nPSA, particularly in an elderly population, PFS was presented as a “worst-case” estimate of outcome and defined as death after PSA failure. The rationale for this definition was that PFS defined in this way would be more relevant to a younger population of men not at high risk of death from comorbid conditions. This definition has been used in prior studies that established pPSA as an important prognostic indicator in men with low and intermediate-risk prostate cancer treated with RT alone.17 Local or metastatic disease was not required for death to be considered as prostate cancer related. UROLOGY 62 (3), 2003
The duration of PFS was measured from the end of EBRT until death or the date last known to be alive. The nPSA was the lowest recorded PSA value after EBRT before failure in the event this occurred. If no biochemical failure occurred, the nPSA was defined as the lowest PSA after RT completion. If the PSA only increased and the patient had at least 1 year of follow-up, the last pre-EBRT PSA was recorded as the nadir value, with the time to nadir equal to 0. Because all high-risk patients were excluded, this occurred in only 3 patients. Patients were analyzed by nPSA quartile group (less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or greater). This approach of having cutpoints at quartile intervals was chosen to have an unbiased comparison of the relationship between the nPSA and FFPF/PFS. Distributions of pretreatment features among the four nadir groups were compared using a chi-square statistic for categorical variables (eg, Gleason score), analysis of variance methods for continuous variables (eg, age), and the nonparametric Kruskal-Wallis test to compare distributions (eg, PSA). Post hoc tests were used to identify which groups differed when overall significant differences occurred. The probability of FFPF, PFS, and OS was calculated using the Kaplan-Meier product limit method. Comparisons of distributions, including sequential pairwise comparisons between quartile groups, were performed using the log-rank test. Multivariate analyses to identify independent predictors of outcome were carried out using Cox’s proportional hazard model and the log likelihood test.
RESULTS A summary of the pretreatment patient characteristics is shown in Table I. Because patients with a high risk of metastatic disease were excluded from this analysis, only a minority of patients had T3 lesions, Gleason score 8 to 10, or Radiation Therapy Oncology Group (RTOG) risk group 3 within each quartile. However, no statistically significant difference was found in the distribution of T stage, Gleason score, or RTOG risk group among nadir quartile groups. The pPSA, percentage of change in PSA level, and time to nPSA differed significantly among the nadir quartile groups (all with P ⬍0.0001). FFPF BY NADIR QUARTILE GROUPS A statistically significant difference in FFPF was seen by nadir quartile group (P ⬍0.0001; Fig. 1). The 5-year FFPF estimate was 83%, 72%, 58%, and 33% for nPSA quartile group of less than 0.3, 0.3 to less than 0.6, 0.6 to less than 1.2, and 1.2 ng/mL or more, respectively. Sequential pairwise comparisons using the log-rank test between quartile groups were all significantly different, with those having nPSA levels of 1.2 ng/mL or greater having the poorest outcome (P ⫽ 0.03, 0.01, and ⬍0.0001). Although there is a progression between nPSA and FFPF whereby lower nPSA levels result in better biochemical outcomes, the cutoff of 1.2 ng/mL marks a very significant difference in prognosis in terms of FFPF. 493
TABLE I. Comparison of groups nPSA <0.3 ng/mL nPSA 0.3–<0.6 ng/mL nPSA 0.6–<1.2 ng/mL nPSA >1.2 ng/mL P Value n pPSA nPSA Change in PSA (%) Time to nPSA (mo) T stage (%) T1 T2 T3 Gleason score (%) 2–6 7 8–10 RTOG risk group (%) R1 R2 R3 Mean Dmax (Gy)
n ⫽ 175 6.9 0.1 97 26.5
n ⫽ 177 8.9 0.4 94 24.5
n ⫽ 223 11.8 0.8 92 25.6
n ⫽ 173 18.0 1.8 85 18.2
⬍0.0001 ⬍0.0001 ⬍0.0001
27 63 10
19 77 4
31 62 4
21 68 11
0.01
66 30 4
75 20 5
72 25 3
66 30 4
0.29
56 34 10 72.2
58 31 11 71.3
58 35 7 71.7
50 40 10 71.0
0.49 0.03
KEY: nPSA ⫽ nadir prostate-specific antigen; pPSA ⫽ pretreatment PSA; RTOG ⫽ Radiation Therapy Oncology Group; Dmax ⫽ maximal dose.
FIGURE 1. FFPF by nPSA quartile groups. Sequential pairwise comparisons using the log-rank test between quartile groups were all significantly different: less than 0.3 versus 0.3 to less than 0.6 ng/mL, P ⫽ 0.03; 0.3 to less than 0.6 versus 0.6 to less than 1.2 ng/mL, P ⫽ 0.01; 0.6 to less than 1.2 versus 1.2 ng/mL or greater, P ⬍0.0001.
A significant effect according to nPSA was seen among patients with a pPSA level less than 10 ng/mL (P ⬍0.0001) and 10 ng/mL or greater (P ⬍0.0001), again showing that nPSA is an independent prognostic indicator for FFPF and not simply a reflection of the pPSA. Univariate analysis indicated that Gleason score (less than 7 versus 7 versus greater than 7), T stage (1 versus 2 versus 3), pPSA (less than 4 versus 4 to less than 10 versus 10 to less than 20 versus 20 ng/mL or greater), RTOG risk group (1 versus 2 versus 3), and maximal dose (69 Gy or less versus greater than 69 to 77 Gy versus greater than 77 Gy) were significant predictors of FFPF among these 494
FIGURE 2. PFS by nPSA quartile groups.
low and intermediate-risk patients. Age was found to be of borderline significance (P ⫽ 0.06) favoring patients 65 years old or younger. Multivariate analysis comparing nPSA, RTOG risk group, pPSA, and age using Cox’s proportional hazard model identified nPSA to be the most important predictor of FFPF (P ⬍0.0001). pPSA and RTOG risk group were also independent predictors of FFPF (P ⬍0.0001 and P ⫽ 0.002, respectively). The results were identical whether variables were considered as continuous or categorical features. PFS AND OS PFS was also found to be statistically significantly different among nadir quartile groups (P ⫽ 0.02; Fig. 2). The probability of death after biochemical failure was greater with an increase in UROLOGY 62 (3), 2003
nPSA. Multivariate analysis also indicated that nPSA was a strong predictor of PFS (P ⫽ 0.001). pPSA was of borderline significance as a predictor of PFS (P ⫽ 0.09). No statistically significant difference was found in OS with death from any cause associated with nPSA. In contrast, using the same potential factors in modeling biochemical failure, age at the time of treatment was a significant predictor of OS (P ⫽ 0.03) and pPSA was of borderline statistical significance (P ⫽ 0.07) for these low and intermediate-risk patients. COMMENT PSA has become commonly accepted in the monitoring of men with adenocarcinoma of the prostate.18 The pPSA level has been shown to be a significant predictor of post-therapeutic outcome after definitive RT for prostate cancer.1,3,4,17,19,20 The use of PSA after treatment for prostate cancer as a marker of outcome has been less well established. After prostatectomy, PSA rapidly falls off to undetectable or near undetectable levels in most patients. After RT, PSA declines more slowly and often does not reach undetectable levels. This has resulted in considerable debate, resulting in the consensus statement regarding guidelines for PSA levels after RT.2,6 –10,16,21 nPSA levels have been widely used as a predictor of biochemical success or failure. However, several investigators have used various absolute post-RT nPSA levels to report biochemical treatment failure, ranging from 0.2 to 4.0 ng/mL.6,8,10 –12,14,22 No consensus has been reached as to which nPSA value conveys the most prognostic significance. Recent data from Hanlon13 have suggested that nPSA is a significant prognostic indicator for both distant metastasis and death from prostate cancer. That study found nPSA to have predictive value for both distant metastasis and death from prostate cancer when analyzed using cutpoints of less than 1.0, 1.1 to 2.0, and greater than 2.0 ng/mL. Patients with an nPSA greater than 2.0 ng/mL had a significantly worse outcome compared with the other nPSA groups with regard to both outcome measures.13 In this report, we confirm the observations made by prior studies that the nPSA level is a predictor of biochemical treatment failure. Although it is important to define the relationship between nPSA and FFPF, there is a potential for statistical bias inherent in comparing high nPSA levels and PSA failure. The remaining question is whether an intermediate value between 0.3 and 1.2 ng/mL really has prognostic significance. Our data suggest that nPSA does hold prognostic significance for FFPF within this range, with a lower risk of biochemical failure associated with a lower nPSA. Longer folUROLOGY 62 (3), 2003
low-up and randomized prospective trials are necessary to confirm the trend seen in this report and others. This emphasizes the importance of establishing the relationship between nPSA and PFS. Our study addressed this issue and confirmed that nPSA is also a strong predictor of disease-free survival in low and intermediate-risk patients. The analysis of nPSA by quartile group was chosen to provide an objective way of measuring the significance of this continuous variable. By looking at nPSA as a categorical variable, we did not choose a cutpoint that unfavorably weighted the nPSA value in a multivariate analysis, leading to a potentially biased result. Our goal, therefore, was not to provide an absolute nPSA value as the consensus endpoint as in other studies,6,8,10 –12,14 but to show that nPSA is a significant predictor of FFPF, as well as disease-free survival. Our data do not support the previously published data by Critz et al.,10 which stated that all patients with an nPSA level greater than 1.0 ng/mL had biochemical evidence of recurrence at a minimum of 5 years of follow-up after RT completion. For the patients in our analysis, the 5-year Kaplan-Meier estimate of FFPF for patients achieving a minimal PSA greater than 1 ng/mL was 32%. This difference in outcome is likely explained by the different treatment methods between the two groups. All patients in the analysis by Critz et al.10 received permanent iodine-125 seed implantation in addition to EBRT, and the patients in our study received EBRT alone. This suggests that the target nPSA is different for these two different treatment modalities. The results of our multivariate analysis of nPSA, pPSA, RTOG risk group, and age demonstrated that a relationship exists between nPSA and death from prostate cancer. Although our analysis using quartile group cutpoints did not show pPSA to be an independent predictor of PFS, it is possible that higher cutpoints or longer follow-up may yield positive results in this regard. Multivariate analysis did not show nPSA to be an independent prognostic variable for OS. Although this may have been a result of advanced age and competing comorbidities in the patient cohort, longer follow-up is required to establish the relationship between nPSA and OS definitively. CONCLUSIONS Multivariate analysis indicated nPSA after EBRT is strongly associated with FFPF and PFS in patients with low and intermediate-risk prostate cancer. An nPSA level of 1.2 ng/mL or greater strongly correlated with a worse prognosis regarding FFPF and PFS. The results of this study showed that nPSA is a prognostic indicator for biochemical failure and PFS. Longer follow-up and larger patient 495
numbers are required to confirm these observations. REFERENCES 1. Kaplan ID, Cox RS, and Bagshaw MA: Prostate specific antigen after external beam radiotherapy for prostatic cancer: followup. J Urol 149: 519 –522, 1993. 2. Kuban DA, el-Mahdi AM, and Schellhammer PF: Prostate-specific antigen for pretreatment prediction and posttreatment evaluation of outcome after definitive irradiation for prostate cancer. Int J Radiat Oncol Biol Phys 32: 307–316, 1995. 3. Pisansky TM, Cha SS, Earle JD, et al: Prostate-specific antigen as a pretherapy prognostic factor in patients treated with radiation therapy for clinically localized prostate cancer. J Clin Oncol 11: 2158 –2166, 1993. 4. Zagars GK: Serum PSA as a tumor marker for patients undergoing definitive radiation therapy. Urol Clin North Am 20: 737–747, 1993. 5. Zietman AL, Coen JJ, Shipley WU, et al: Radical radiation therapy in the management of prostatic adenocarcinoma: the initial prostate specific antigen value as a predictor of treatment outcome. J Urol 151: 640 –645, 1994. 6. Perez CA, Michalski JM, and Lockett MA: Chemical disease-free survival in localized carcinoma of prostate treated with external beam irradiation: comparison of American Society for Therapeutic Radiology and Oncology consensus or 1 ng/mL as endpoint. Int J Radiat Oncol Biol Phys 49: 1287– 1296, 2001. 7. Critz FA, Levinson AK, Williams WH, et al: Prostatespecific antigen nadir: the optimum level after irradiation for prostate cancer. J Clin Oncol 14: 2893–2900, 1996. 8. Critz FA, Levinson K, Williams WH, et al: Prostatespecific antigen nadir of 0.5 ng/mL or less defines disease freedom for surgically staged men irradiated for prostate cancer. Urology 49: 668 –672, 1997. 9. Critz FA, Levinson AK, Williams WH, et al: The PSA nadir that indicates potential cure after radiotherapy for prostate cancer. Urology 49: 322–326, 1997. 10. Critz FA, Williams WH, Holladay CT, et al: Post-treatment PSA or ⫽ 0.2 ng/mL defines disease freedom after radiotherapy for prostate cancer using modern techniques. Urology 54: 968 –971, 1999.
496
11. Hanks GE, Lee WR, and Schultheiss TE: Clinical and biochemical evidence of control of prostate cancer at 5 years after external beam radiation. J Urol 154(2 Pt 1): 456 –459, 1995. 12. Horwitz EM, Vicini FA, Ziaja EL, et al: An analysis of clinical and treatment related prognostic factors on outcome using biochemical control as an end-point in patients with prostate cancer treated with external beam irradiation. Radiother Oncol 44: 223–228, 1997. 13. Hanlon A: Posttreatment prostate-specific antigen nadir highly predictive of distant failure and death from prostate cancer. Int J Radiat Oncol Biol Phys 53: 297–303, 2002. 14. Lee WR, Hanlon AL, and Hanks GE: Prostate specific antigen nadir following external beam radiation therapy for clinically localized prostate cancer: the relationship between nadir level and disease-free survival. J Urol 156(2 Pt 1): 450 – 453, 1996. 15. Roach M III, Lu J, Pilepich MV, et al: Predicting longterm survival and the need for hormonal therapy: a metaanalysis of RTOG prostate cancer trials. Int J Radiat Oncol Biol Phys 47: 617–627, 2000. 16. Panel ASTRO Consensus Statement: Guidelines for PSA following radiation therapy. Int J Radiat Oncol Biol Phys 37: 1035–1041, 1997. 17. Roach M, Weinberg V, McLaughlin P, et al: Serum prostate specific antigen (PSA) and survival following external beam radiotherapy of the prostate. Urology 61: 730 –735, 2003. 18. von Eschenbach A, Ho R, Murphy GP, et al: American Cancer Society guidelines for the early detection of prostate cancer: update, June 10, 1997. Cancer 80: 1805–1807, 1997. 19. Russell KJ, Dunatov C, Hafermann MD, et al: Prostate specific antigen in the management of patients with localized adenocarcinoma of the prostate treated with primary radiation therapy. J Urol 146: 1046 –1052, 1991. 20. Zagars GK: Prostate specific antigen as an outcome variable for T1 and T2 prostate cancer treated by radiation therapy. J Urol 152(5 Pt 2): 1786 –1791, 1994. 21. Geist RW: Reference range for prostate-specific antigen levels after external beam radiation therapy for adenocarcinoma of the prostate. Urology 45: 1016 –1021, 1995. 22. Zietman AL, Tibbs MK, Dallow KC, et al: Use of PSA nadir to predict subsequent biochemical outcome following external beam radiation therapy for T1-2 adenocarcinoma of the prostate. Radiother Oncol 40: 159 –162, 1996.
UROLOGY 62 (3), 2003