Long-term survival after radical prostatectomy versus external-beam radiotherapy for patients with high-risk prostate cancer


  • See editorial on pages 2830-2, this issue.



The long-term survival of patients with high-risk prostate cancer was compared after radical prostatectomy (RRP) and after external beam radiation therapy (EBRT) with or without adjuvant androgen-deprivation therapy (ADT).


In total, 1238 patients underwent RRP, and 609 patients received with EBRT (344 received EBRT plus ADT, and 265 received EBRT alone) between 1988 and 2004 who had a pretreatment prostate-specific antigen (PSA) level ≥ 20 ng/mL, a biopsy Gleason score between 8 and 10, or clinical tumor classification ≥ T3. The median follow-up was 10.2 years, 6.0 years, and 7.2 years after RRP, EBRT plus ADT, and EBRT alone, respectively. The impact of treatment modality on systemic progression, cancer-specific survival, and overall survival was evaluated using multivariate Cox proportional hazard regression analysis and a competing risk-regression model.


The 10-year cancer-specific survival rate was 92%, 92%, and 88% after RRP, EBRT plus ADT, and EBRT alone, respectively (P = .06). After adjustment for case mix, no significant differences in the risks of systemic progression (hazard ratio [HR], 0.78; 95% confidence interval [CI], 0.51-1.18; P = .23) or prostate cancer death (HR, 1.14; 95% CI, 0.68-1.91; P = .61) were observed between patients who received EBRT plus ADT and patients who underwent RRP. The risk of all-cause mortality, however, was greater after EBRT plus ADT than after RRP (HR, 1.60; 95% CI, 1.25-2.05; P = .0002).


RRP alone and EBRT plus ADT provided similar long-term cancer control for patients with high-risk prostate cancer. The authors concluded that continued investigation into the differing impact of treatments on quality-of-life and noncancer mortality will be necessary to determine the optimal management approach for these patients. Cancer 2011. © 2011 American Cancer Society.

To assist with patient counseling and guide treatment selection, the National Comprehensive Cancer Network (NCCN) has recommended risk stratification of patients with newly diagnosed prostate cancer according to prostate-specific antigen (PSA) level, biopsy Gleason score, and clinical stage.1 Although the widespread use of PSA testing has altered the clinical and demographic characteristics of men with newly diagnosed prostate cancer,2 men with what is characterized as high-risk disease continue to be encountered. Indeed, the management of these patients represents 1 of the most significant current challenges in prostate cancer treatment, because the optimal therapeutic strategy remains to be established.

It is noteworthy that several studies have reported effective long-term cancer control for men with locally advanced prostate cancer who received treatment with external-beam radiotherapy (EBRT), particularly when combined with adjuvant androgen-deprivation therapy (ADT).3-8 At the same time, the role of surgery in the form of radical prostatectomy (RRP) for patients with high-risk disease also has continued to be evaluated at select centers; in fact, durable survival outcomes after RRP for high-risk tumors have been reported.9, 10 Therefore, given the lack of randomized trials comparing the efficacy of different therapeutic options for prostate cancer, the treatment of these patients continues to be based largely on individual physician experience and biases.11

It is worth noting that the retrospective series to date that have compared outcomes after surgery and radiation for high-risk tumors have demonstrated widely disparate results, with several groups reporting improved outcomes after RRP12-17; others reporting better results after radiation18, 19; and a few,20-22 including a small prospective trial,23 noting equivalent efficacy. However, those studies involved different definitions of high-risk prostate cancer, evaluated disparate outcome measures, and included a relatively limited number of patients, often with short-term follow-up. Furthermore, it has been noted that men who undergo RRP are younger and healthier in terms of comorbidity than men who receive radiation treatment,11, 24 differences that may further obscure the ability of such studies to establish the impact of treatment modality on outcome.

Most recently, Zelefsky and colleagues compared the survival of patients with cT1 through cT3b prostate cancer who were treated at a single center with IMRT and RRP.15 After controlling for clinicopathologic variables, RRP was associated with a reduced risk of metastases and cancer-specific mortality.15 However, only 409 patients (17% of the overall cohort) with high-risk tumors were included in the study, which accounted for 19 deaths from prostate cancer at a follow-up of 5 years.15 At the same time, Cooperberg et al, in a primarily community-based dataset, likewise demonstrated decreased cancer-specific mortality after surgery versus radiation for patients with clinically localized disease.16 Again, however, only 6.5% of patients who underwent RRP and only 24% of patients who received EBRT in that series were in the highest risk category, and the median follow-up (3.9 years after RRP and 4.5 years after radiation) was relatively short.16

Here, we compared the outcomes after RRP and EBRT for patients who were classified with high-risk prostate cancer according to NCCN criteria.1 Patients were treated at 1 of 2 high-volume centers in a contemporary period. With a robust dataset and long-term follow-up, we report the clinically relevant endpoints of systemic progression (SP), death from prostate cancer, and overall mortality, controlling for case mix and patient variables.


Study Population

After each center's institutional review board approval was obtained, we reviewed the Mayo Clinic Prostatectomy Registry and the Fox Chase Cancer Center Radiation Oncology Database to identify 1847 patients with high-risk prostate cancer who received definitive local therapy in the form of RRP (Mayo Clinic) or EBRT (Fox Chase Cancer Center) between 1988 and 2004. High-risk disease was defined according to NCCN guidelines1 as PSA ≥ 20 ng/mL, or clinical stage ≥ T3N0M0, or a biopsy Gleason score from 8 to 10. Tumors were classified according to the 2002 American Joint Committee on Cancer classification,25 and the Gleason system was used for grading. No patient had clinical evidence of pelvic lymph node disease on cross-sectional imaging or had evidence of distant metastases on bone scan.

Surgical Treatment

In total, 1238 patients underwent RRP for high-risk cancers. Patient demographics are presented in Table 1. Surgical procedures were performed by different surgeons using standardized techniques. No patient received ADT or radiation before RRP. All men who were included in this analysis were treated with an open retropubic approach. Adjuvant therapy was defined as treatment received within 90 days of RRP and was given at the discretion of the treating physician. Medical hormone-deprivation therapy generally was intended to be life-long. However, given the retrospective nature of this study, it is uncertain whether patients discontinued treatment after a period of ADT. In total, 503 patients (40.6%) received adjuvant therapy after RRP, of whom 367 (29.6%) received ADT, 85 (6.9%) received EBRT, and 51 (4.1%) received both. Salvage treatment was defined as secondary therapy that was initiated > 90 days after RRP and likewise was administered per individual physician. At last follow-up, 253 men (20.4%) had received salvage EBRT, and 415 men (33.5%) had received salvage ADT. The median time from RRP to salvage treatment was 2.7 years (range, 0.2-15.4 years) for salvage EBRT and 10.3 years (range, 1.4-20.2 years) for salvage ADT.

Table 1. Patient Demographics
 Treatment: No. of Patients (%) 
FeatureRRP, n=1238EBRT+ADT, n=344EBRT Alone, n=265P
  1. RRP indicates radical retropubic prostatectomy; EBRT, external-beam radiation therapy; ADT, androgen-deprivation therapy; IQR, interquartile range; PSA, prostate-specific antigen; AJCC, American Joint Commission on Cancer; cT, clinical tumor classification; NA, not available; IMRT, intensity-modulated radiation therapy.

Age at diagnosis: Median [IQR}, y66.0 [60.4-70.3]68.8 [62.9-73.7]69.3 [65.0-73.9]<.0001
Year of treatment   <.0001
 1988-1993433 (35)57 (16.6)134 (50.6) 
 1994-1998431 (34.8)132 (38.4)80 (30.2) 
 1999-2004374 (30.2)155 (45)51 (19.2) 
Pretreatment PSA: Median [IQR], ng/mL20.5 [7.7-30.1]17.4 [8.1-34.0]22.0 [10.2-33.0].02
2002 AJCC tumor classification   <.0001
 cT1a-cT1c265 (21.4)61 (17.8)69 (26) 
 cT2562 (45.4)136 (39.5)114 (43) 
 cT3-cT4411 (33.2)147 (42.7)82 (31) 
Biopsy Gleason score   <.0001
 ≤6471 (38)75 (21.8)138 (52.1) 
 7303 (24.5)105 (30.5)74 (27.9) 
 8-10464 (37.5)164 (47.7)53 (20) 
Type of EBRT   <.0001
 ConventionalNA5 (1.5)40 (15.1) 
 ConformalNA238 (69.2)204 (77) 
 IMRTNA101 (29.3)21 (7.9) 
Charlson comorbidity index   .12
 0 230 (66.9)183 (69.1) 
 1 89 (25.9)55 (20.8) 
 ≥2 25 (7.2)27 (10.1) 


Clinical characteristics of the 609 patients who received EBRT for high-risk cancers are listed in Table 1. Comorbidity status for these patients was assessed using the Charlson comorbidity score.26 Radiation technique at the Fox Chase Cancer Center evolved over the study period from conventional EBRT to 3-dimensional conformal radiotherapy (3DCRT), and, more recently, to intensity-modulated radiation therapy (IMRT). The techniques for treatment planning and delivery of each of these methods have been published previously.27, 28 Overall, 564 of 609 patients (92.6%) received either conformal or IMRT. The median radiation dose received was 72 Gray (Gy) (range, 5040-7900 Gy) and was administered in 2-Gy fractions. Pelvic lymph nodes consistently were included within the radiation portal for these patients. Patients were considered to have received adjuvant ADT with EBRT if they initiated any form of ADT within 6 months of the date of starting treatment with EBRT. In total, 344 of 609 patients (56.5%) patients who received EBRT also received adjuvant ADT. The median duration of adjuvant ADT in this group was 22.8 months (range, 1-108 months). Among these patients, 57 (16.6%) subsequently received salvage ADT at a median of 3.0 years (range, 0.3-9.2 years) after initial treatment.

Outcome Assessment

Post-treatment assessments, including physical examination and serum PSA measurement, were done quarterly for the initial 2 years, semiannually for an additional 2 years, and annually thereafter. SP involved demonstrable metastases on radionucleotide bone scan or on biopsies outside of the prostate/prostatic bed. Vital status was identified from death certificates or physician correspondence. For patients who were followed elsewhere, the outcomes were monitored annually by correspondence.

Statistical Considerations

Comparison of patient clinicopathologic variables between the treatment groups was performed using chi-square and Kruskal-Wallis tests, as appropriate. Survival was measured from the day of surgery or the start of EBRT, as appropriate. Survival was estimated using the Kaplan-Meier method and was compared with the log-rank test. Patients were censored at last follow-up or death if the endpoint of interest had not been attained. Cox proportional hazard regression models were used to analyze the impact of treatment approach on SP and survival, with adjustment for patient age, pretreatment PSA, and biopsy Gleason score (treated as continuous variables), as well as clinical tumor classification (treated as a categorical variable). Two additional methods were used to control for the baseline clinicopathologic differences, which have been documented in patients with prostate cancer who undergo surgery versus EBRT2, 22: 1) the analysis was restricted to include only EBRT patients with a Charlson comorbidity index of 0 or 1, and 2) a competing risk-regression model was used to account for competing causes of mortality.29

All tests were 2-sided, and P values ≤ .05 were considered significant. Statistical analyses were done using the SAS software package (version 9.1.3; SAS Institute, Cary, NC).


Consistent with previously noted demographic trends in prostate cancer treatment,11, 24 in the current study, the men who received radiation were significantly older than the men who underwent surgery (P < .0001) (Table 1), whereas patients in the EBRT plus ADT cohort also had significantly higher biopsy Gleason scores (P < .0001) and more advanced tumor classification (P < .0001) than patients in the RRP cohort. The median total pretreatment PSA level was highest in the EBRT-alone cohort (P = .02).

At a median follow-up after RRP of 10.2 years (interquartile range, 6.6-14.0 years), 192 patients who underwent surgery experienced SP, and 404 died, including 115 patients who died of prostate cancer. The median follow-up after EBRT plus ADT and after EBRT alone was 6.0 years (interquartile range, 4.2-8.7 years) and 7.3 years (interquartile range, 4.5-9.6 years), respectively. Twenty-seven patients who received EBRT plus ADT and 35 patients who received EBRT alone experienced SP, and there were 90 and 108 deaths, respectively, including 19 and 25 patients who died from prostate cancer, respectively.

The estimated 10-year probability of freedom from SP did not differ significantly between the treatment arms (85% for RRP patients, 88% for EBRT plus ADT patients, and 81% for EBRT-alone patients; P = .24) (Fig. 1A). Likewise, the 10-year cancer-specific survival was equivalent for patients who underwent surgery (92%) and patients who received EBRT plus ADT (92%), and it was modestly better than for patients who received EBRT alone (88%; P = .06) (Fig. 1B). The 10-year overall survival rate, however, improved significantly for the patients who underwent RRP (77%) compared with the patients who received either EBRT plus ADT (67%) or EBRT alone (52%; P < .001) (Fig. 1C).

Figure 1.

These charts illustrate the outcome of patients with high-risk prostate cancer who underwent radical prostatectomy (RRP), received external-beam radiation therapy (EBRT) plus androgen-deprivation therapy (ADT), or received EBRT alone. Shown are (A) systemic progression-free survival, (B) cancer-specific survival, and (C) overall survival.

Table 2 lists the factors that were predictive of postoperative survival in the multivariate model, including the type of treatment. After controlling for patient age, year of treatment, pretreatment PSA, clinical T-classification, and biopsy Gleason score, patients who received EBRT alone had a significantly increased risk of SP (hazard ratio [HR], 1.53; 95% confidence interval [CI], 1.05-2.23; P = .03), death from prostate cancer (HR, 2.14; 95% CI, 1.35-3.39; P = .001), and overall mortality (HR, 2.04; 95% CI, 1.62-2.56; P < .0001) compared with patients who underwent RRP. EBRT plus ADT, conversely, demonstrated cancer control similar to that achieved by RRP, with an HR of 0.78 (95% CI, 0.51-1.18; P = .23) for SP and 1.14 (95% CI, 0.68-1.91; P = .61) for cancer-specific mortality. Nevertheless, receipt of EBRT plus ADT was associated with a significantly increased risk of all-cause mortality compared with surgery (HR, 1.60; 95% CI, 1.25-2.05; P = .0002).

Table 2. Multivariate Cox Regression Analysis of Factors Associated With Outcome in Patients With High-Risk Prostate Cancer
 Systemic ProgressionDeath From Prostate CancerDeath From Any Cause
VariableHR95% CIPHR95% CIPHR95% CIP
  • HR indicates hazard ratio; CI, confidence interval; PSA, prostate-specific antigen; cT, clinical tumor classification; EBRT, external-beam radiation therapy; RRP, radical retropubic prostatectomy; ADT, androgen-deprivation therapy.

  • a

    According to the 2002 American Joint Committee on Cancer classification system.

Year of treatment0.960.93-
Log2 PSA1.161.05-1.28.0021.171.03-
Biopsy Gleason score1.441.29-1.62<.00011.601.39-1.84<.00011.241.14-1.34<.0001
Tumor classification (cT2 vs cT1c)a1.741.12-
Tumor classification (cT3/cT4 vs cT1c)a2.421.54-3.79.00012.551.39-4.70.0031.200.93-1.56.17
Treatment (EBRT vs RRP)1.531.05-<.0001
Treatment (EBRT/ADT vs RRP)0.780.51-

We also accounted for the baseline differences in patient age and comorbidity status among the treatment groups, factors that had an impact on competing causes of mortality for patients with prostate cancer, in 2 ways. First, we repeated our analyses and included only EBRT patients who had a Charlson comorbidity index of 0 or 1. With this restriction, we observed no appreciable change in our results (Fig. 2), such that the estimated SP-free survival once again did not differ significantly between the 3 treatment groups, patients who underwent RRP or received EBRT plus ADT had a similar cancer-specific survival, and patients who underwent RRP had the highest overall survival (77% at 10 years). Likewise, the associations of treatment modality with outcome did not change in the multivariate Cox model when only EBRT patients who had a Charlson score of 0 or 1 were included (data not shown). Next, we also performed a competing risk-regression analysis and observed once again that there was no significant difference in the risk of either SP (HR, 0.71; 95% CI, 0.47-1.08; P = .11) or death from prostate cancer (HR, 0.99; 95% CI, 0.60-1.64; P = .97) between patients who received EBRT plus ADT and patients who underwent RRP.

Figure 2.

These charts illustrate the outcome of patients with high-risk prostate cancer who underwent radical prostatectomy (RRP), received external-beam radiation therapy (EBRT) plus androgen-deprivation therapy (ADT), or received EBRT alone. The analysis was restricted to radiation therapy patients who had a Charlson comorbidity index of 0 or 1. Shown are (A) systemic progression-free survival, (B) cancer-specific survival, and (C) overall survival.


We evaluated 2 large institutional datasets of patients who were treated during the PSA era for high-risk prostate cancer as defined by common (NCCN) criteria. We observed that both surgery and EBRT plus ADT were associated with a 10-year disease-specific survival rate of 92%. EBRT alone was associated with decreased treatment efficacy compared with EBRT plus ADT. Moreover, when controlling for case mix and patient variables, patients who received EBRT plus ADT had a significantly increased risk of all-cause mortality compared with patients who underwent RRP.

The outcomes noted here for patients with high-risk prostate cancer who were managed with surgery and EBRT plus ADT are consistent with the results from previous clinical trials3-8 and retrospective case series.9, 10 Indeed, the 5-year mortality rate from prostate cancer of 3.2% for patients with locally advanced tumors who received EBRT and long-term ADT recently reported by Bolla and colleagues8 is nearly identical to our 4% rate death from prostate cancer at 5 years. Furthermore, our data demonstrating improved survival with EBRT plus ADT versus EBRT alone for high-risk cancers confirms previous reports that demonstrated a benefit of ADT with EBRT.3-6, 8 It also is important to note that the median radiation dose received by the patients in our study (72 Gy) and the median duration of ADT in our EBRT plus ADT group (22.8 months) were suboptimal according to modern practice standards. Indeed, it has been reported that patients with high-risk disease derive a cancer-specific survival benefit from higher doses of radiation (78 Gy)30 and longer courses of ADT (range, 24-36 months).8, 31 Nevertheless, the radiation doses that we used were similar to the doses (range, 65-70 Gy) that were used in several prospective, randomized trials evaluating EBRT for high-risk disease.3-8

The 10-year SP-free survival rate (85%) and the cancer-specific mortality rate (8%) after RRP in the current study parallel the outcomes reported by Yossepowitch et al,10 whereas our 15-year prostate cancer-specific mortality rate of 15% after RRP is nearly identical to the findings of Stephenson and colleagues.32 One potential benefit from surgery for patients with high-risk disease is the ability to obtain pathologic staging, which, as suggested previously,33 may guide the selective application of secondary therapies. For example, Meng et al observed that patients with high-risk prostate cancer who received radiation therapy were 3.5 times more likely to receive ADT than patients who underwent RRP.11 Because increasing data have emerged on the adverse consequences of ADT on quality of life34 and noncancer morbidity35 among men with prostate cancer, the ability to delay if not avoid ADT may represent a potential advantage to surgery for these patients. The median time from RRP to salvage ADT in our study was 10.3 years. Although the timing of salvage ADT was at the discretion of the surgeon and was not identified as an indicator of treatment success, the results suggest that high-risk patients who undergo surgery may have long intervals of ADT-free survival.

It is noteworthy that, although RRP and EBRT plus ADT were associated with nearly identical long-term cancer control, patients who received EBRT plus ADT had a significantly increased risk of all-cause mortality. One potential explanation for this finding may be an imbalance between the RRP cohort and the EBRT plus ADT cohort in terms of medical comorbidities and unmeasured confounding variables. We attempted to account for selection bias with multivariate regression models and a competing risk-regression analysis. Nevertheless, the retrospective nature of this study prohibited our ability to completely account for baseline differences in clinicopathologic demographics. Therefore, differences in age and comorbidity status may be responsible for the worse overall survival noted among patients who received with EBRT. In addition, comorbidity scores were not available for the surgical cohort, which is an acknowledged limitation of our current dataset, although a historic series from the Mayo Clinic noted that only 11% of patients who underwent RRP at that time had a Charlson score ≥ 2.36 Another explanation is that ADT had an adverse impact on men who received EBRT. ADT has been associated with an increased risk of cardiac death, particularly among men with coronary artery disease.37 Although the exact mechanism of this interaction is unknown, the metabolic effects of ADT may contribute to the comorbidities and risk factors for cardiac death, such as diabetes, hypercholesterolemia, and hypertension.38 Patient selection based on these medical comorbidities may have resulted in treatment with EBRT (vs RRP), and the addition of ADT may have potentiated the increased risk of cardiac mortality. However, it is important to note that long-term hormone use was not associated with an increased risk of cardiac toxicity in a large, phase 3, randomized trial of patients who received EBRT for locally advanced prostate cancer.31

Although a randomized trial from the pre-PSA era reported improved survival after surgery versus radiation,39 that study involved only 97 patients and has been criticized for its methodology. A more recent trial demonstrated equivalent efficacy between the treatments, although only 95 patients were included, and all men received adjuvant ADT.23 Given the lack of relevant outcome data from randomized trials comparing RRP and EBRT, currently, observational studies like ours remain the primary means to evaluate the relative efficacy of local therapies for high-risk prostate cancer. However, the retrospective comparative series to date have been limited by relatively small numbers of high-risk patients12, 19, 23 and the frequent use of biochemical failure (BF) as an outcome measure.18-22 Indeed, given the variability in definitions for BF after EBRT40, 41 and the difficulty with comparing BF across treatments,42 we chose to focus on the endpoints of distant metastases, cancer-specific survival, and all-cause mortality. Determining the impact of prostate cancer treatments on outcomes other than BF remains clinically important, because the natural history of PSA recurrence is variable and usually is prolonged, and BF does not always translate into systemic progression and prostate cancer death.43-45

We recognize that our study was limited by its retrospective nature and by the disparate number of patients who underwent RRP versus those who received EBRT. Moreover, although follow-up is considered in the life table method, the differing lengths of follow-up between the patients who underwent surgery and patients who received EBRT may have had an impact on our long-term comparisons. We also acknowledge the lack of a centralized pathology review. In addition, we acknowledge the significant heterogeneity of the kind of risk-group classification model that we used here, although it has not been observed that this model limits the ability to discern the impact of treatment modality on survival.15 A further limitation of such a nonrandomized study lies with the varied use of salvage therapies, specifically with regard to the choice, timing, and duration of treatment. However, we believe that the nonuniform application of secondary therapies reflect “real-world” clinical practice, in which standardized guidelines for the management of failure after primary treatment currently do not exist.

Despite these factors, we demonstrated that patients with high-risk prostate cancer can achieve long-term cancer control after definitive primary therapy with surgery or EBRT plus ADT. Indeed, the risk of prostate cancer-specific mortality at 10 years after primary treatment is relatively low (8%) in light of the aggressive features of these tumors. Continued investigation into the differing impact of treatments on quality-of-life measures34 and noncancer morbidity35 will be necessary to help determine the optimal management approach for these patients. In addition, further study to identify more effective systemic therapies that may be integrated with local treatments will be required to improve patient outcomes.


This work was supported by Grant P30 CA006927 (to Fox Chase Cancer Center) by a Mayo Clinic Prostate Specialized Program of Research Excellence (SPORE) grant (CA91956-09) from the National Cancer Institute.