Presented in part at the 2006 Annual Meeting of the American Society of Clinical Oncology, Atlanta, Georgia, June 2–6, 2006.
The presence of multiple determinants of aggressive cancer biology may impact prostate cancer-specific mortality (PCSM) rates compared with fewer factors. The authors estimated PCSM after radiation therapy with short-course androgen suppression therapy (RT+AST) or radical prostatectomy (RP) in men with clinically localized, intermediate-risk to high-risk prostate cancer.
The study cohort included 3240 men treated from 1981 to 2002 with RT with 6 months of AST (n = 550) or RP (n = 2690) for localized prostate cancer with at least 1 risk factor (prostate-specific antigen [PSA] >10 ng/mL, biopsy Gleason score 7–10, or clinical tumor category T2b or T2c). Competing risks regression analyses were used to determine whether the number of risk factors present was associated with time to PCSM.
Men with all 3 risk factors had significantly shorter time to PCSM after RT+AST (adjusted hazards ratio [HR] of 9.3; 95% confidence interval [95% CI], 1.9–44.5 [PGray = .005]) or RP (adjusted HR of 6.3; 95% CI, 3.2–12.2 [PGray < .001]) when compared with men with any 1 or 2 risk factors. The 7-year estimates of PCSM for men having 1, 2, or 3 risk factors were 0.83% (95% CI, 0.27–1.4%), 2.6% (95% CI, 1.0–4.2%), and 12.6% (95% CI, 7.1–18.1%), respectively.
External beam radiation therapy (RT) combined with short-term androgen suppression therapy (AST) is a standard nonoperative treatment for men with intermediate-risk to high-risk prostate cancer, whereas radical prostatectomy alone remains the standard surgical option. Despite treatment, a significant proportion of these men will experience prostate-specific antigen (PSA)-defined failure and cancer-specific death, indicating a need for more aggressive initial therapy.
To identify patients who are at an increased risk of experiencing PSA failure after RT or RP, investigators have developed prognostic algorithms that utilize risk factors based on pretreatment PSA level, biopsy Gleason score, and clinical tumor stage.1–4 However, PSA-defined failure may not accurately reflect the likelihood of progression to metastatic disease and prostate cancer-specific death because of competing causes of mortality and the long natural history of some types of PSA failure.5, 6
Consequently, investigators have performed studies to demonstrate that risk stratification systems can also predict for time to prostate cancer-specific mortality (PCSM) after RT or RP alone,7, 8 but to our knowledge no such studies have been performed in men treated with RT+AST. The purpose of the current study was to estimate PCSM rates after RT+AST or RP in men with clinically localized, intermediate-risk to high-risk prostate cancer and to investigate whether the number of risk factors present is significantly associated with the time to PCSM.
MATERIALS AND METHODS
Patient Selection and Treatment
The study cohort included 3240 men who were treated between 1981 and 2002 with either RT combined with short-course AST (n = 550) or RP alone (n = 2690) for clinically localized prostate cancer with at least 1 risk factor defining intermediate-risk to high-risk disease (pretreatment serum PSA level >10 ng/mL, biopsy Gleason score of 7 or higher, or 2002 American Joint Committee on Cancer [AJCC] clinical tumor category T2b or T2c). The baseline clinical, treatment, and follow-up data for these patients were compiled from the multi-institutional databases maintained by the Cancer of the Prostate Strategic Urologic Research Endeavor,9, 10 the Center for Prostate Disease Research, and the Brigham and Women's Hospital/Dana-Farber Cancer Institute. The study was performed with the approval of the Institutional Review Board (IRB) at each of the individual participating institutions, and each study patient signed an IRB-approved informed consent form.
All men in the RT+AST cohort were initially treated with 6 months of androgen suppression therapy (2 months before, 2 months during, and 2 months after RT) in the form of oral bicalutamide (given at a dose of 50 mg every day) or flutamide (at a dose of 250 mg 3 times a day) in combination with injections of goserelin (at a dose of 3.6 mg/month) or leuprolide (at a dose of 7.5 mg/month) in either 1-month or 3-month formulations. Men who could not tolerate the entire course of RT or AST were excluded from this analysis. Men in the RP cohort were not permitted to have received preoperative AST or adjuvant RT.
Men who experienced PSA failure after initial therapy with RT+AST or RP were managed according to standard practice. In general, men initially treated with RT+AST who experienced PSA failure (defined as a serum PSA level >1.0 ng/mL and rising) received salvage, life-long hormonal therapy. Men who underwent salvage local therapy after RT+AST were excluded from this study. Men initially managed with RP who experienced PSA failure (defined as a postoperative serum PSA level >0.2 ng/mL and rising) typically received salvage, prostatic fossa RT at the time of first postoperative PSA failure, followed by life-long hormonal therapy at the time of second PSA failure.
The time at which salvage hormonal therapy was initiated was determined by the treating physicians. When available as imaging modalities, computed tomography (CT) of the pelvis and bone scan were negative for metastatic disease at the time of initiation of salvage therapy. Information regarding whether hormonal therapy was given in an intermittent or continuous manner was not available for analysis.
As part of their staging evaluations, all men underwent a history and physical examination; digital rectal examination (DRE); serum PSA; and transrectal, ultrasound-guided core needle biopsy of the prostate gland with Gleason score histologic grading.11 Men whose cancer was diagnosed after a transurethral resection of the prostate were excluded. The clinical stage was determined from the DRE findings according to the 2002 AJCC staging system.12 PSA measurements were made using the Hybritech (San Diego, CA), Tosoh (Foster City, CA), or Abbott (Chicago, IL) assays. Men typically underwent pretreatment staging evaluation with CT of the pelvis and bone scan when these studies were available. No men with evidence of bone metastases or lymph node involvement were included in this analysis.
The median follow-up times for the living 550 men treated with RT+AST and the living 2690 men treated with RP were 4.6 years (interquartile range [IQR], 2.1–7.1 years) and 4.2 years (IQR, 2.4–5.7 years), respectively, using the date of surgery or the RT start date as time zero. Men generally had a serum PSA measurement and DRE performed every 3 months for 2 years, then every 6 months for 3 additional years, and then annually. At the time of last follow-up, 99 (18%) of the RT+AST-managed men and 825 (31%) of the RP-managed men had sustained a first PSA failure. No patient died as a result of prostate cancer before PSA failure. Overall, there were 62 prostate cancer-specific deaths, including 11 (2%) in the RT+AST group and 51 (1.9%) in the RP group. The determination of the causes of death was made using death certificates.
Descriptive statistics were used to analyze the baseline patient characteristics. The chi-square metric was used to compare distributions of pretreatment PSA, biopsy Gleason score, tumor category, and age between men treated with RT+AST and those treated with RP. The Wilcoxon 2-sample test was used to assess for differences in the distributions of continuous variables between the 2 groups.
Competing risks methodology was employed for the analyses in this study to account for competing causes of mortality. A competing risks regression analysis13 was performed to evaluate the association between each risk factor (pretreatment PSA >10 ng/mL, biopsy Gleason score of 7 or higher, or 2002 AJCC clinical tumor category T2b or T2c) and the time to PCSM for the entire study cohort. A separate competing risks regression analysis was performed to determine whether age as a continuous variable or the number of risk factors present was associated with the time to PCSM for the entire study cohort, for the RT+AST-managed men, and for the RP-managed men. Time zero was defined as the date of initiation of RT and the date of surgery for the RT+AST and RP cohorts, respectively.
The adjusted hazard ratios (HRs) of PCSM were estimated on the basis of the coefficients from the competing risks regression models13 and are reported with the associated 95% confidence intervals (95% CIs). Estimates of PCSM after initial therapy were calculated using the cumulative incidence method14 and were graphically displayed, and 95% CIs were calculated for select, timepoint estimates of PCSM for men having any 1, any 2, or all 3 risk factors. Comparisons of the cumulative incidence estimates of PCSM after initial therapy were made using Gray's K-sample tests that account for competing causes of mortality.15 The Bonferroni correction16 was applied to adjust for multiple comparisons (ie, a significant P value was defined as P < .05/n, in which n is the number of comparisons).
Descriptive Statistics of the Pretreatment Clinical Characteristics
The baseline pretreatment clinical characteristics of the study population stratified by initial treatment modality are summarized in Table 1. Men treated with RT+AST were older (Pχ2 < .001) and presented with higher pretreatment PSA levels (Pχ2 < .001) and biopsy Gleason scores (Pχ2 < .001) compared with men treated with RP. The median ages of the RT+AST- and RP-managed men at the time of initial therapy were 71.3 years (IQR, 66.0–74.9 years) and 64.0 years (IQR, 59.2–68.2 years), respectively (PWilcoxon < .001). The median pretreatment PSA levels were 11.7 ng/mL (IQR, 7.3–19.5 ng/mL) and 8.6 ng/mL (IQR, 5.2–14.0 ng/mL) for the RT+AST- and RP-managed cohorts, respectively (PWilcoxon < .001).
Table 1. Baseline Characteristics of the 3240 Patient Study Cohort Stratified by Treatment
No. of patients (%)* or value (Interquartile range)
RP (n = 2690)
RT + AST (n = 550)
RP indicates radical prostatectomy; RT, radiation therapy; AST, androgen suppression therapy; PSA, prostate-specific antigen; AJCC, American Joint Committee on Cancer.
Percentages may not sum to 100% because of rounding error.
Competing Risks Regression Analyses and Estimates of PCSM
A competing risks regression analysis was performed to assess whether each risk factor was significantly associated with the time to PCSM for the entire study cohort. As shown in Table 2, a pretreatment PSA >10 ng/mL (adjusted hazards ratio [HR] of 2.1; 95% CI, 1.2–3.7 [PGray = .006]), a biopsy Gleason score of 7 or higher (adjusted HR of 3.2; 95% CI, 1.9–5.3 [PGray < .001]), and a clinical tumor category of T2b or T2c (adjusted HR of 2.9; 95% CI, 1.6–5.1 [PGray < .001]) were each independently associated with the time to PCSM after initial treatment.
Table 2. Results of Competing Risks Regression Analysis Assessing the Association of Each Individual Risk Factor with Time to Prostate Cancer-Specific Mortality after Treatment for All Patients (n = 3240)
Adjusted HR (95% CI)
HR indicates hazards ratio; 95% CI, 95% confidence interval; PSA, prostate-specific antigen; AJCC, American Joint Committee on Cancer.
PSA >10 ng/mL
Biopsy Gleason score 7 or higher
2002 AJCC tumor category T2b or T2c
A separate competing risks regression analysis was then performed to evaluate whether the number of risk factors present was significantly associated with the time to PCSM. As shown in Table 3, after adjusting for age, the presence of all 3 risk factors was found to be significantly associated with a shorter time to PCSM for all men (adjusted HR of 7.1; 95% CI, 3.9–12.9 [PGray < .001]), whereas age as a continuous variable (PGray = .27) and the presence of any 2 risk factors (PGray = .15) were not. The presence of all 3 risk factors was also associated with shorter time to PCSM for the RT+AST-managed men (adjusted HR of 9.3; 95% CI, 1.9–44.5 [PGray = .005]) and for the RP-managed men (adjusted HR of 6.3; 95% CI, 3.2–12.2 [PGray < .001]).
Table 3. Results of Competing Risks Regression Analysis Assessing Whether Age and the Number of Risk Factors Present Are Associated with Time to Prostate Cancer-Specific Mortality
As illustrated in Figure 1, after adjusting for multiple comparisons, men with all 3 risk factors had significantly shorter time to PCSM after initial therapy compared with men with any single (PGray < .001) or any 2 risk factors (PGray < .001). There was no statistically significant difference noted in the time to PCSM after initial therapy between men with any single risk factor and men with any 2 risk factors (PGray = .14). The 7-year estimates of PCSM for men having 1, 2, or 3 risk factors were 0.83% (95% CI, 0.27–1.4%), 2.6% (95% CI,: 1.0–4.2%), and 12.6% (95% CI, 7.1–18.1%), respectively.
The estimated rates of PCSM after RT+AST and RP stratified by the number of risk factors present are illustrated in Figures 2 and 3, respectively. Compared with men with any single risk factor (PGray ≤ .001) or any 2 risk factors (PGray < .004), men with all 3 risk factors had significantly shorter time to PCSM after RT+AST or RP. There was no statistically significant difference in time to PCSM after RT+AST or RP between men with any single risk factor and men with any 2 risk factors (PGray ≤ .67). The 7-year estimates of PCSM after RT+AST for men with 1, 2, or 3 risk factors were 0.74% (95% CI, 0.0–2.2%), 2.5% (95% CI, 0.0–6.0%), and 14.0% (95% CI, 1.3–26.7%), respectively. The corresponding estimates of PCSM were 0.83% (95% CI, 0.24–1.4%), 2.6% (95% CI, 0.88–4.3%), and 11.8% (95% CI, 5.7–18.0%) for men managed with RP.
A variety of risk stratification systems have been developed to categorize men with prostate cancer into risk groups for PSA failure after treatment with RT or RP based on the presence of known pretreatment prognostic factors.1–4 However, time to PSA failure may not accurately reflect time to PCSM given the long natural history of some types of PSA failure as well as competing causes of death in these men.5, 6 Furthermore, different definitions of PSA failure exist in the literature, and the administration of AST can result in the misclassification of failures that are biochemically defined.17 Therefore, risk stratification systems that predict for the endpoint of time to cancer-specific death may be more clinically useful.
Investigators have shown that risk stratification systems can predict for time to PCSM after RT or RP as monotherapies.7, 8 However, to our knowledge, no such studies have been performed in men treated with RT in combination with short-course AST. In the current study, we provided estimates of cancer-specific mortality after RT+AST or RP in a large cohort of men with localized, intermediate-risk to high-risk prostate cancer and demonstrated that a PSA >10 ng/mL, a biopsy Gleason score of 7 or higher, and a clinical tumor category of T2b or T2c are each risk factors that are independently associated with the time to PCSM. We also found that men with all 3 of these risk factors have significantly higher estimates of PCSM after RT+AST or RP when compared with men with any single risk factor or any 2 risk factors.
The clinical significance of this finding is that men with prostate cancer with a PSA >10 ng/mL, a biopsy Gleason score of 7 or higher, and clinical tumor category T2b or T2c disease are at the highest risk of cancer-specific death after standard treatment, suggesting that RT+AST or RP may be inadequate therapy for these men. Therefore, these men would be ideal candidates for enrollment in clinical trials investigating the benefit of adding systemic chemotherapy to RT+AST or RP.
Historically, systemic cytotoxic chemotherapy has been reserved for the treatment of men with advanced, hormone-refractory prostate cancer (HRPC) to provide palliation but without improvement in survival. However, 2 recent randomized studies18, 19 have shown that treatment with docetaxel-based chemotherapy is effective in prolonging survival and improving quality of life in men with metastatic HRPC when compared with standard treatment with mitoxantrone and prednisone. Consequently, there has been considerable interest in integrating docetaxel with local therapies either in the neoadjuvant or adjuvant settings to address microscopic, HRPC cells that may be present in sites beyond the reach of local therapy in men with clinically localized but high-risk disease.
The current study has some potential limitations that need to be considered. First, because prostate cancer deaths may occur up to and beyond 15 years after initial treatment, further follow-up will be needed to confirm our findings. With longer follow-up, there may be increased statistical power to assess whether the presence of any 2 risk factors might predict for a shorter time to PCSM as compared with the presence of any single risk factor, a finding that was not demonstrated in this study.
Second, the results of this study are only applicable to men with comparable clinical characteristics as the study cohort and who are treated in a similar manner (ie, RP followed by salvage RT and life-long hormonal therapy at first and second PSA failure, respectively, or RT+AST followed by life-long hormonal therapy at first PSA failure). Whether the use of a longer duration of AST in combination with RT or the administration of adjuvant RT after RP would affect time to PCSM remains unclear.
Third, information regarding the dose of radiation therapy used was not uniformly available, so an analysis of the impact of dose escalation on PCSM could not be performed. In 2 prospective, randomized studies of RT dose escalation in men with localized prostate cancer, men treated with higher doses had lower rates of PSA failure.20, 21 However, treatment with higher doses of RT has not yet been shown to be associated with improved cancer-specific or overall survival. Because the endpoint of our study was time to PCSM and not PSA failure, one would not expect RT dose to affect our results.
Finally, several studies suggest that there has been a trend among pathologists today to assign higher Gleason scores to prostate cancer histologic specimens when compared with pathologists in the past.22–24 Consequently, outcomes for men with prostate cancer treated in the modern era may appear to be artificially better than historical outcomes if no adjustment is made for this Gleason upgrading phenomenon. The results of the current study will need to be validated in a cohort of patients who have had their pathologic specimens reviewed by a modern pathologist.
In conclusion, for men with clinically localized prostate cancer, estimates of PCSM appear substantial for men with all 3 factors that define intermediate- to high-risk disease. Given the estimates of PCSM for men with all 3 risk factors noted in this study, these men are optimal candidates for enrollment in clinical trials investigating the benefit of adding systemic therapies such as docetaxel to RT+AST or RP.
Funding for the Cancer of the Prostate Strategic Urologic Research Endeavor and for the Center for Prostate Disease Research is provided by TAP Pharmaceuticals and the United States Army Medical Research and Material Command, respectively.