Dose-escalated radiation therapy for intermediate-risk prostate cancer

Patient selection for androgen deprivation therapy using percentage of positive cores


  • Presented at the American Society for Therapeutic Radiology and Oncology 50th Annual Meeting, Boston, Massachusetts, September 2008.



Randomized trials supported the use of androgen deprivation therapy (ADT) with radiation therapy (RT) for intermediate-risk prostate cancer. However, the value of concurrent ADT was less certain with dose-escalated RT. Better methods of stratifying patients in this risk group may help select patients who are most likely to benefit.


A total of 238 men with intermediate-risk (prostate specific antigen [PSA] 10-20, Gleason 7, or stage T2b-c) adenocarcinoma of the prostate were treated with external beam RT between 1989 and 2006. Patients had Gleason≤6 (39%) or 7 (61%) tumors; median PSA was 10.5 ng/mL. A median of 37.5% of biopsy cores were positive from a median of 9 biopsy cores sampled. The median RT dose was 74 Gy to the prostate. A total of 112 patients (47%) received neoadjuvant and concurrent ADT (median, 4 months). Median follow-up period was 49 months.


The freedom from biochemical failure (FFBF, nadir + 2 definition) was 93% at 3 years, 86% at 4 years, and 80% at 5 years. On univariate analysis, the only factor associated with FFBF was percentage of positive cores (PPC, P = .0340). The prognostic value of PPC≥50 was not evident in patients receiving ADT (FFBF at 4 years 90% vs 91%, P = .3015). For patients not receiving ADT, the impact of PPC≥50 (FFBF at 4 years 76% vs 93%, P = .0844) was more pronounced. On multivariate analysis, PPC (P = .0388) was significantly associated with FFBF, whereas Gleason sum, ADT, RT dose, PSA, and T-stage were not.


After dose-escalated external beam RT, intermediate-risk prostate cancer patients with PPC≥50 had the highest risk for biochemical failure and may be most likely to derive a benefit from ADT. Cancer 2009. © 2009 American Cancer Society.

Outcomes with surgery or radiation therapy (RT) for nonmetastatic prostate cancer vary by risk category1 as defined by T-stage, PSA level, and Gleason sum. In men with higher risk features (eg, stage≥T2b, PSA>10, or Gleason sum≥7), several randomized studies support the use of androgen deprivation therapy (ADT) in combination with RT. This form of combined therapy has been shown to improve cause-specific2, 3 and overall survival,4, 5 related to the ability of ADT to favorably impact the rate of local and distant control. However, patients in these trials were treated with standard radiation doses ≤70 Gy. Acknowledging that there exist dose-response data for local control of prostate cancer,6 the benefit of concurrent ADT may be less apparent in patients treated with dose-escalated, image-guided RT.7-10 Patients with intermediate-risk disease (stage T2b-c, PSA 10-20, or Gleason sum 7)11 and a relatively lower risk for disease recurrence, therefore, may be candidates to receive dose-escalated RT rather than standard-dose RT with short-term ADT. This approach would spare some men from the perhaps unnecessary risks of ADT.12-16

Further risk stratification within this heterogeneous17 intermediate-risk group would be useful to assess which patients are most likely to fail therapy. Several reports have demonstrated the prognostic effect of percentage of positive cores (PPC) for risk stratification in cohorts of patients treated with either prostatectomy18-22 or radiation therapy.23-30 The goal of this study was to review biochemical outcomes for patients treated with primary RT for intermediate-risk prostate cancer and to identify which pretreatment risk factors were associated with failure. Specific attention was given to the prognostic impact of PPC. By studying the impact of this risk factor for the entire cohort and for subsets of patients receiving ADT or dose-escalated RT, we hoped to gain insight on whether PPC might be a useful tool to identify patients most likely to benefit from dose-escalated RT with concurrent ADT.


This study included patients with nonmetastatic adenocarcinoma of the prostate treated with primary external beam RT at the University of Chicago Pritzker School of Medicine between 1989 and 2006. No patient had prior radical prostatectomy or brachytherapy boost with external beam RT. After exclusion of patients with low-risk (PSA<10, clinical stage T1-T2a, and Gleason score ≤6) or high-risk (PSA>20, clinical stage T3-4, or Gleason score ≥8) prostate cancer, 238 patients with intermediate-risk prostate cancer were identified. Patient data, including demographic, treatment, and follow-up information, were prospectively entered into a departmental database. Review of these outcomes was performed with approval from the hospital's institutional review board.

Patient characteristics are included in Table 1. Of 238 patients, 151 had information available regarding biopsy cores. A median of 9 cores was biopsied, and a median of 3 biopsy cores were positive. PPC was defined as the number of cores with any amount of cancer divided by the number of total cores. The median PPC was 37.5% (range, 8%-100%). In 80 patients, information was available regarding the amount of tumor seen by individual needle core. For these cases, a calculation of percentage of total tumor involvement for all cores was performed. For example, a patient with a 12-core biopsy showing 50% tumor involvement in 6 cores would have (50%)*(6 of 12) + (0%)*(6 of 12) = 25% tumor involvement. The median tumor involvement for all cores was 9%. Biopsies performed elsewhere were reviewed at this institution before treatment by a genitourinary pathologist. Patients were treated with a median RT dose of 74 Gy, at 1.8-2 Gy/fraction (range, 62-76.4 Gy). Forty-six patients (19%) received doses <70 Gy. Intensity-modulated RT was used in 121 patients (51%), with planning target volume margins of 6-10 mm beyond the clinical target volume. Only 6 patients were treated with initial whole pelvic fields to cover lymph nodes at risk. The use of ADT was at the discretion of the treating physician over this time period. ADT was administered in 112 patients (47%) for a median 4 months duration, given 2 months neoadjuvantly and 2 months concurrently with RT. Only 7 patients (6%) received ADT for more than 6 months.

Table 1. Patient Characteristics (N=238)
CharacteristicsNo. (%)
  1. RT indicates radiation therapy; PSA, prostate antigen

Median age, y70 [range, 52-83 y]
 Caucasian94 (39)
 African American133 (56)
 Other/unknown11 (5)
Median pre-RT PSA level, ng/mL10.5 [range, 0.9-19.9 ng/mL]
PSA level, ng/mL 
 0-411 (5)
 4-10104 (44)
 10-20123 (52)
Clinical T stage 
 T1142 (60)
 T2a53 (22)
 T2b31 (13)
 T2c12 (5)
Gleason sum 
 2-693 (39)
 7145 (61)
Primary Gleason score, n=196 
 24 (3)
 3116 (77)
 431 (21)
% Cores positive, n=151 
 0-3353 (35)
 34-4944 (29)
 50-10054 (36)
% Tumor involvement, n=80 
 0-528 (35)
 5-2033 (41)
 20-10019 (24)

Median follow-up, defined as the time from RT completion to the date of the last PSA test, was 49 months. Median potential follow-up, defined as the time from RT completion to date of study analysis, was 83 months (range, 24-225 months). Freedom from biochemical failure (FFBF) was defined according to the nadir + 2 definition.31 Differences between groups were tested with univariate analysis using chi-square analysis. Kaplan-Meier curves for FFBF were generated, and survival comparisons were made with the log-rank test. Subset analysis was performed on a subset of patients treated with RT doses ≥74 Gy, given the hypothesis that the impact of ADT may be minimized by higher doses that would favorably impact local control. Such an analysis of patients treated with higher doses thereby might be expected to have different statistical associations between prognostic variables and biochemical outcome. Multivariate analysis was performed with proportional hazards analysis, using prespecified categorical explanatory variables (clinical stage, Gleason score, PSA, dose, ADT, PPC) stratified by the median value.


Use of ADT

Univariate analysis was performed to test whether any subgroup of patients was more likely to be treated with ADT. Patients with higher Gleason sum were not more likely to receive ADT (50% vs 42%, P = .2041), nor were patients with PPC ≥ 37.5% (57% vs 47%, P = .2388), patients with ≥T2b disease (42% vs 48%, P = .4495), or patients treated with RT at dose ≥74 Gy (52% vs 46%, P = .1041). However, patients with higher PSA levels (≥10) were more likely to receive ADT (57% vs 37%, P = .0016), as were patients with a primary Gleason score of 4 rather than a primary score of 3 (81% vs 43%, P = .0001).

Biochemical Outcome

For all patients, FFBF at 3, 4, and 5 years was 93%, 86%, and 80%, respectively. Results of univariate analyses are shown in Table 2, part A.

Table 2. Biochemical Failure Analyses
  1. FFBF-4y indicates freedom from biochemical failure at 4 years; PPC, percent positive cores; ADT, androgen deprivation therapy; RT, radiation therapy; CI, confidence interval; PSA, prostate specific antigen.

A. Univariate Analyses,n=238FFBF-4yP
PPC, ≥50 vs <50 82% vs 92%.0340
Clinical T stage, ≤T2a vs ≥T2b 89% vs 75%.1763
ADT, yes vs no 88% vs 85%.3402
RT dose, ≥74 Gy vs < 74 Gy 91% vs 82%.3432
Primary Gleason score, 4 vs ≤3 93% vs 89%.5662
PSA level, >10.5 vs ≤10.5 86% vs 86%.6163
Gleason sum, ≤6 vs 7 87% vs 85%.7843
B. FFBF-4y PPC by ADT, n=151ADTNo ADTP
C. Multivariate Analysis, n=238 Risk Ratio (95% CI)P
PPC, <50 vs ≥50 0.62 (0.39-0.98).0388
Gleason sum, ≤6 vs 7 0.73 (0.41-1.33).3023
ADT, yes vs no 0.86 (0.55-1.33).4924
RT dose, < 74 Gy vs ≥74 Gy 0.92 (0.55-1.52).7302
PSA level, ≤10.5 vs >10.5 0.96 (0.55-1.72).8957
Clinical T stage, ≤T2a vs ≥T2b 1.01 (0.60-1.80).9686

Patients treated with ADT did not appear to have a different biochemical outcome compared with patients not treated with ADT (FFBF at 4 years [FFBF-4y] 88% vs 85%, P = .3402). ADT use was not associated with FFBF-4y for patients treated with RT doses <74 Gy (86% vs 81%, P = .1373) or patients treated with RT doses ≥ 74 Gy (90% vs 91%, P = .5133).

On univariate analysis, PPC was the only tested prognostic variable associated with biochemical outcome. Meanwhile, percentage of tumor involvement for all cores was not associated with biochemical outcome; FFBF-4y was 90% for tumor involvement >9%, versus 93% for tumor involvement ≤9%, P = .7469. Patients with PPC ≥ 50% had lower biochemical control rates (FFBF-4y 82% vs 92%, P = .0340). These statistically significant differences held whether the patients were analyzed using a stratification point of 33% or 37.5% for PPC. FFBF-4y for patients with PPC ≤ 33%, 33% to 50%, or ≥50% were 95%, 89%, and 82%, respectively (P = .0678). FFBF outcomes according to PPC are shown in Figure 1. Subset analysis revealed a decreased association between PPC and biochemical outcome for patients treated with ADT (Table 2, part B). In these patients (n = 75), PPC≥50% was not associated with FFBF-4y (90% vs 91%, P = .3015). However, the prognostic impact of PPC≥50% was greater in patients not treated with ADT (n = 76, FFBF-4y 76% vs 93%, P = .0844). Similar statistical associations were apparent when using a stratification point of 33% of 37.5% for PPC.

Figure 1.

Freedom from biochemical failure (n = 238) for patients with percentage of positive cores (PPC) <50% (red), compared with patients with percentage of positive cores ≥50% (green), P = .0340.

All univariate analyses were repeated in a subgroup analysis for patients treated with RT dose ≥74 Gy (n = 127). For this subset of patients, PPC (FFBF-4y 80% for ≥50 vs 94% for <50, P = .0188) and Gleason sum (FFBF-4y 87% for Gleason 7 vs 100% for Gleason ≤6, P = .0406) were associated with biochemical outcome. PSA level (P = .2280), Gleason primary score (P = .2861), T stage (P = .3331), and ADT (P = .5133) were not associated with outcome. In patients treated with ADT and RT dose ≥74 Gy (n = 51), PPC ≥ 50% was not associated with FFBF-4y (95% vs 92%, P = .6989). Patients treated with RT dose ≥74 Gy but without ADT (n = 48) who had PPC ≥ 50% did have lower biochemical control rates than patients with PPC < 50% (FFBF-4y 70% vs 96%, P = .0056).

Results of multivariate analysis are shown in Table 2, part C (n = 238). Upon controlling for possible confounding among all prognostic factors, only PPC was associated with biochemical outcome. The statistical conclusions were unchanged on repeating the multivariate analysis on a subset of patients with a minimum of 36 months of follow-up, to minimize potential bias from patients with shorter follow-up period (data not shown).

Other Clinical Endpoints

Five patients in this intermediate-risk cohort developed distant metastasis at a median time of 22 months (range, 2-30 months). Two patients died from prostate cancer, at 57 and 173 months. Because of the low numbers of these events, analysis was not performed according to distant metastasis or cause-specific survival. Overall survival at 4 and 5 years was 94% and 92%, respectively. Multivariate analysis was repeated for overall survival. No tested explanatory variables, including T-stage (P = .2247), Gleason sum (P = .2201), RT dose (P = .2510), PSA (P = .3729), PPC (P = .3813), and ADT (P = .6593), were associated with survival.


Patients with nonmetastatic prostate cancer can be adequately treated with primary RT, with or without ADT.32 The role of ADT has been established by randomized data from patients with intermediate- or high-risk features treated with external beam RT.2-5 However, these studies were designed using conventional or three-dimensional conformal RT, in which dose escalation was limited by late toxicity of treatment, particularly to the rectum.33 Intensity-modulated RT allows for better sparing of adjacent normal tissues and can be used to safely escalate dose to the target volume.34 Dose-escalation studies have demonstrated better rates of biochemical, local, and distant control in comparison to standard doses of RT33, 35-38; with long-term follow-up, dose-escalated RT may also improve cause-specific survival.36 Given the favorable impact of dose-escalated RT on tumor control, it is questionable whether patients will have the same magnitude of benefit from the concurrent use of ADT. This issue is of particular concern for patients with intermediate-risk prostate cancer, whose risk of micrometastatic disease is likely to be lower than for patients with high-risk prostate cancer. Several retrospective reports have been unable to demonstrate any favorable impact on biochemical control using ADT in the setting of dose-escalated therapy.7-10 Because of the concern for the potential long-term side effects of ADT,12-16 we certainly must be judicious in our means of patient selection for ADT.

This study suggests that PPC can be used to stratify biochemical outcome for patients with intermediate-risk prostate cancer who are treated with external beam RT. PPC was the only significant prognostic factor for biochemical outcome among the covariates analyzed in univariate and multivariate analysis. Patients with PPC≤33% had an FFBF-4y of 95%, whereas patients with PPC≥50% had an FFBF-4y of 82%. Subset analysis, although limited by fewer patient numbers and events, suggested that ADT was more beneficial for patients with higher PPC than for patients with lower PPC. Men with PPC≥50 not treated with ADT had the lowest FFBF, which was 76% at 4 years, and 55% by 5 years. Meanwhile, for the entire cohort, ADT did not appear to influence outcome, which is in agreement with other reports involving dose-escalated RT to patients with intermediate-risk features.7-10

The use of PPC to further stratify patients beyond the standard classification with PSA, T stage, and Gleason score has been reported by several groups. In patients treated with prostatectomy, higher PPC has been associated with locally aggressive disease39 and with poorer biochemical control.18-22 Cohorts of patients treated with standard doses of external beam RT23, 24, 27-29 and brachytherapy25, 26, 30 have also been studied with regard to PPC. Both standard-dose external beam and brachytherapy series have demonstrated associations between PPC and biochemical outcome. Of note, the magnitude of this association appears to be decreased for patients who are treated with the highest RT doses (ie, treated with the most effective implant).26 In a series of intermediate- to high-risk patients treated with brachytherapy and supplemental external beam RT, Nurani et al demonstrated that PPC was significantly associated with biochemical outcome in a multivariate analysis for all patients, although this association was minimized for patients with a high quality implant, defined as D90>100% (P = .15). A recent multi-institutional analysis of 3264 men treated with various doses of external beam RT similarly demonstrated that PPC was associated with biochemical outcome, in a manner that was dose-dependent; the predictive behavior of PPC was strongest for doses <66 Gy.40 On the basis of these results, one might conclude that PPC serves as a surrogate for tumor volume and/or aggressive biology, that primarily is a measure of ability to achieve local control, and perhaps to a lesser degree is a measure of micrometastatic potential. In the current series, PPC had prognostic value for the intermediate-risk group of patients, even for men treated with doses ≥74 Gy. As such, PPC may be a means of identifying intermediate-risk patients most likely to fail dose-escalated external beam RT, and thereby benefit from ADT. It remains to be seen with further follow-up whether the expected gain from ADT in this risk category of patients would be from enhancement of the local effect of RT, or on potential treatment of micrometastatic disease.

Results from this study should be interpreted as preliminary, and with limitations inherent to a retrospective analysis. There was no standardized method of prostate biopsy over the time period of study, and not all patients in the study had this information available. Subset analyses further reduced numbers available for analysis, and in turn, limited the power to detect any statistically significant factors associated with outcome. The era of RT did span several years and, therefore, also spanned a dose range, although the majority were treated with doses greater than standard dose RT (≥70 Gy). Meanwhile, the use and timing of ADT was not controlled for in this risk category of patients, and there were noted imbalances in the administration of ADT based on risk factors. Multivariate analysis was performed to attempt to control for potential interactions among all variables. Short of prospective or randomized data, longer follow-up and larger patient numbers are warranted to substantiate the associations demonstrated here.

In conclusion, PPC was the only prognostic factor found to be associated with biochemical outcome in this cohort of patients treated with external beam RT for intermediate-risk prostate cancer. Patients who had PPC ≥ 50% and were treated with RT alone had inferior outcomes to patients with lower PPC or patients treated with combination ADT. This study suggests that PPC can be used to discern which patients treated with dose-escalated external beam RT are most likely to experience failure, and that patients with PPC ≥ 50% may be the most likely to benefit from concurrent RT and ADT.

Conflict of Interest Disclosures

Dr. Jani is a Georgia Cancer Coalition Distinguished Cancer Clinician/Scholar and received support from this program.