Optimal treatment of intermediate-risk prostate carcinoma with radiotherapy

Clinical and translational issues


  • Alan M. Nichol M.D.,

    1. Department of Radiation Oncology, University of Toronto and the Princess Margaret Hospital-University Health Network, Toronto, Ontario, Canada
    Current affiliation:
    1. Department of Radiation Oncology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Padraig Warde M.B.B.S.,

    1. Department of Radiation Oncology, University of Toronto and the Princess Margaret Hospital-University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Robert G. Bristow M.D., Ph.D.

    Corresponding author
    1. Department of Radiation Oncology, University of Toronto and the Princess Margaret Hospital-University Health Network, Toronto, Ontario, Canada
    • Department of Radiation Oncology, Princess Margaret Hospital-University Health Network, 610 University Avenue, Toronto, Ontario M5G 2M9, Canada
    Search for more papers by this author
    • Robert G. Bristow is a Canadian Cancer Society Research Scientist.

    • Fax: (416) 946-4586


The clinical heterogeneity of intermediate-risk prostate carcinoma presents a challenge to urologic oncology in terms of prognosis and management. There is controversy regarding whether patients with intermediate-risk prostate carcinoma should be treated with dose-escalated external beam radiotherapy (EBRT) (e.g., doses > 74 gray [Gy]), or conventional-dose EBRT (e.g., doses < 74 Gy) combined with androgen deprivation (AD). Data for this review were identified through searches for articles in MEDLINE and in conference proceedings, indexed from 1966 to 2004. Currently, the intermediate-risk prostate carcinoma grouping is defined on the basis of prostate-specific antigen (PSA), tumor classification (T classification), and Gleason score. Emerging evidence suggests that additional prognostic information may be derived from the percentage of positive core needle biopsies at the time of diagnosis and/or from the pretreatment PSA doubling time. Novel prognostic biomarkers include protein expression relating to cell cycle control, cell death, DNA repair, and intracellular signal transduction. Preclinical data support dose escalation or combined AD with radiation as a means to increase prostate carcinoma cell kill. There is Level I evidence that patients with intermediate-risk prostate carcinoma benefit from dose-escalated EBRT or AD plus conventional-dose EBRT. However, clinical evidence is lacking to support the uniform use of AD plus dose-escalated EBRT. Patients in the intermediate-risk group should be entered into well designed, randomized clinical trials of dose-escalated EBRT and AD with sufficient power to address biochemical failure and cause-specific survival endpoints. These studies should be stratified by novel prognostic markers and accompanied by strong translational endpoints to address clinical heterogeneity and to allow for individualized treatment. Cancer 2005. © 2005 American Cancer Society.

In 2004, the American Cancer Society estimated that 230,110 men would be diagnosed with prostate carcinoma and that 29,900 would die of the disease (available from URL: http://www.cancer.org [accessed January 10, 2005]). Recent data from the Cancer of the Prostate Strategic Urological Research Endeavor (CaPSURE) registry indicate that there has been a stage migration of prostate carcinoma over the last 15 years, when prostate-specific antigen (PSA) screening became widely available in the U.S.1 The CaPSURE data base uses the risk stratification for prostate carcinoma defined by D'Amico et al.2 In their scheme, intermediate-risk prostate carcinoma is defined as clinical T1–T2 disease, Gleason score < 8, and PSA ≤ 20 ng/mL with at least 1 of the following adverse factors present: clinical T2b disease, PSA > 10 ng/mL, or Gleason score = 7. According to CaPSURE, between 1989 and 2002, the proportion of patients presenting with high-risk disease (clinical T3–T4 disease, or Gleason score = 8–10, or PSA > 20 ng/mL) decreased from 41% to 15%, and the proportion of patients presenting with low-risk disease (clinical T1c–T2a disease, Gleason score = 2–6, and PSA ≤ 10 ng/mL) increased from 31% to 47%.3 Therefore, despite the introduction of PSA screening, nearly one-third of all new patients with prostate carcinoma continue to present with intermediate-risk disease.

Treatment approaches for low-risk prostate carcinoma include watchful waiting, hormone therapy (e.g., androgen deprivation [AD]), radical prostatectomy, brachytherapy, or external beam radiotherapy (EBRT), depending on tumor and patient characteristics. A number of nonrandomized, retrospective studies have shown equipoise between surgical, brachytherapy, and EBRT approaches for low-risk disease.2, 4 High-risk disease now is approached routinely with combined hormono-radiotherapy based on randomized studies that have shown improvements in overall survival.5, 6 Between 1990 and 2000, the use of neoadjuvant androgen deprivation (NAD) and EBRT rose from 4.9% to 73.5% in the intermediate-risk group, despite the absence of any randomized clinical trial (RCT) for this risk group.7

There is considerable controversy regarding the optimal treatment of patients with intermediate-risk prostatic carcinoma in relation to the role of dose-escalated EBRT (e.g., total dose > 74 gray [Gy]) versus conventional-dose EBRT (e.g., total dose < 74 Gy) and AD. To address this controversy, we have reviewed the categorization and prognostication of intermediate-risk prostate carcinoma. In this report, we also discuss new molecular and cellular biomarkers that may triage intermediate-risk patients further into subgroups for refined prognostication and treatment. Finally, we critically appraise the available pre-clinical and clinical evidence for dose-escalated EBRT alone or in combination with AD for this heterogeneous group of patients.

Defining Intermediate-Risk Prostate Carcinoma

Prognostic variables for PSA-based outcomes after treatment for localized prostate carcinoma can be defined in terms of biochemical freedom from survival (bFFS) or biochemical no evidence of disease (bNED) and have been studied within both univariate and multivariate models. Consequently, biochemical prognostic groupings are based on initial clinical tumor classification (T classification), pretreatment PSA level, and Gleason score8–10 (see Table 1). The biochemical outcome of the intermediate-risk subgroup is highly dependent on the selected definition of biochemical failure. This currently is defined by the American Society for Therapeutic Radiology and Oncology (ASTRO) consensus definition as three consecutive rises in the PSA level after treatment.11 Other biochemical failure definitions are being studied, but none had supplanted the ASTRO definition at the time of this writing.12

Table 1. Risk Group Definitions for Clinically Staged Patients with Prostate Carcinoma
Study/definitionLow stageIntermediate stageHigh stageVery high stage
  • NCCN: National Comprehensive Cancer Network; GS: Gleason score (range, 2–10); PSA: prostate-specific antigen (ng/mL); 2002 TNM classification: T2a: half of 1 lobe of the prostate, T2b: 1 lobe of the prostate, and T2c: both lobes of the prostate; 1997 TNM classification: T2a: 1 lobe of the prostate, and T2b: both lobes of the prostate; RTOG: Radiation Therapy Oncology Group; 1992 TNM classification: T2a: half of 1 lobe of the prostate, T2b: 1 lobe of the prostate, and T2c: both lobes of the prostate.

  • a

    Patients with multiple adverse factors may have been shifted into the next highest risk group.

  • b

    Bulky T2 areas (> 5 cm2 × 5 cm2) were grouped as T3.

Scherr et al., 200317 (NCCN; 2002 TNM classification)T1–T2a, and GS ≤ 6, and PSA < 10T2b–T2c,a or GS = 7, or PSA = 10–20T3a,a or GS = 8–10, or PSA > 20T3b–T4
Roach et al., 200315 (15–35% risk of positive pelvic lymph nodes) PSA < 30 and GS = 7–10 or PSA ≥ 30 and GS ≤ 6  
Lukka et al., 200118 (Canadian Consensus: 1997 TNM classification)T1a–T2a, and GS ≤ 6, and PSA < 10T2b, or GS = 7, or PSA = 10–20T3/T4, or GS = 8–10, or PSA > 20 
Roach et al., 200013 (RTOG trials)bT1–T2 and GS ≤ 6T1–T2 and GS = 7 or T3 and GS ≤ 6T1–T2 and GS = 8–10 or T3 and GS = 7T3 and GS = 8–10
Zelefsky et al., 199810T1–T2, and GS ≤ 6, and PSA < 10One of the following: T3, or GS ≥ 7, or PSA > 10Two of T3, or GS ≥ 7, or PSA > 10 
D'Amico et al., 19982 (1992 TNM classification)T1c–T2a, and GS ≤ 6, and PSA < 10T2b, or GS = 7, or PSA = 10–20T2c, or GS = 8–10, or PSA > 20 

Several authors and organizations have defined variably the interface between the intermediate and high-risk groupings (Table 1). For example, the initial risk groupings defined by Roach et al. were derived from a metaanalysis of the Radiation Therapy Oncology Group (RTOG) randomized trials that were conducted in the pre-PSA era.13, 14 An application of these risk groupings to a recent cohort of patients has validated them again in the modern PSA-era, in which a PSA value > 20 ng/mL confers a high risk of biochemical failure, lower progression-free survival, and lower overall survival.15 Also in 2003, Roach et al. defined an intermediate-risk group (e.g., patients with a 15–35% risk of lymph node involvement, as determined by the following formula: percent risk = 2/3*PSA + 10*[Gleason score = 6]) based on the results of the RTOG 94-13 study.16 Because this definition includes patients with clinical T3 disease and PSA values > 20 ng/mL, it would be classified otherwise as high-risk by a number of groups, including authors of the RTOG meta-analyses, the National Comprehensive Cancer Network (NCCN17), and the Canadian Consensus.18 A recent report suggested that the use of a single-factor prognostic model (e.g., clinical T2b disease, or PSA > 10 ng/mL, or Gleason score = 7) to define intermediate-risk disease created prognostic groups with greater internal consistency than a 2-factor model (e.g., any 2 of clinical T2b disease, PSA > 10 ng/mL, or Gleason score = 7).19 For the purpose of the current review, we have defined patients with intermediate-risk prostate carcinoma using the single-factor interpretation of the NCCN criteria,17 which is the as the same Canadian Consensus definition.18 We also have confined our discussion to the use of EBRT in intermediate-risk disease; however, it is recognized that some groups may utilize brachytherapy (another form of dose-escalation) as a component of therapy for patients with intermediate-risk disease.4

Prognostic Markers for Intermediate-Risk Prostate Carcinoma

Interrogating the relative slopes of EBRT dose-response curves (i.e., the PSA-based tumor control probability vs. the total radiotherapy dose) for prostate carcinoma leads to the conclusion that there is great heterogeneity within the intermediate-risk group.20, 21 Understanding this heterogeneity may allow for more effective triaging a priori of intermediate-risk patients into subgroups with varying probabilities of local control or development of distant metastases. This can be illustrated using prostate-specific nomograms, such as the Memorial Sloan-Kettering Prostogram (version 4.02; available from URL://www.nomograms.org [accessed May 27, 2004]). For example, 2 patients who receive 78 Gy (1 patient with T1c disease, Gleason score = 6, and PSA = 10.1 and the other patient with T2b disease, Gleason score = 7, and PSA = 19.9) would have 5-year PSA progression-free probabilities of 88% and 54%, respectively.22, 23 The well established prognostic factors of PSA, clinical T classification, and Gleason score determine < 50% of the variability of biochemical failure-free survival.24 Indeed, the clinical stage, as determined by digital rectal examination, historically has been an important prognostic variable; however, as more and more patients are diagnosed with nonpalpable disease, it has become less useful.25 Furthermore, the utility of computed tomography scanning and bone scans to address the heterogeneity within the intermediate-risk group is low.26 Clearly, better prognostic and predictive factors are required for this heterogeneous group of patients. A strong predictor of tumor radiocurability, independent of known prognostic factors, could lead to novel treatment strategies combined with radiotherapy or to a decision to abort radiotherapy altogether in favor of a radical prostatectomy.

There are a number of promising methods for improving our ability to stratify and select the appropriate treatment for patients with intermediate-risk prostate carcinoma. For example, information obtained from systematic prostate biopsies can identify patients with a greater volume of disease and, thus, a greater risk of PSA failure, metastatic burden, and prostate carcinoma-specific mortality (PCSM).27–29 The percentage of positive biopsies (i.e., ≥ 50% diagnostic cores involved with malignancy) is an independent prognostic variable that may triage patients within the intermediate-risk group into relatively favorable or unfavorable risk groups.27, 29 Results from more recent analyses suggest that the pretherapy PSA doubling time, or a high proliferative index (as measured by Ki-67 staining) can predict the risk for distant metastases and death in prostate carcinoma patients who are treated with radical intent.30–32 Other approaches to the determination of pretreatment local or systemic tumor bulk include magnetic resonance spectroscopy of metabolically active tumor cells within the prostate33 and the use of the polymerase chain reaction to detect potentially metastatic prostate carcinoma cells in the bloodstream. The latter assay, or one similar to it, could guide the use of systemic therapy in addition to radiotherapy.34

A number of intrinsic, pretreatment molecular biomarkers also may predict tumor cell radioresistance, extracapsular disease, and/or the presence of metastases (see Fig. 1). These include markers of intratumoral hypoxia, genetic stability, cell proliferation, or cell death.35, 36 Intratumoral hypoxia is viewed as a negative prognostic factor capable of generating aggressive prostate tumor cell variants that have increased capacity for chemoresistance and radioresistance, androgen independence, and/or metastases.37–41 Using a polarographic pO2 electrode, groups at Princess Margaret Hospital and at the Fox Chase Cancer Center have shown that intermediate-risk patients are heterogeneous for intratumoral hypoxia.42, 43 Furthermore, the level of hypoxia may predict for intrinsic radioresistance after brachytherapy or EBRT.43, 44

Figure 1.

Radiotherapy induces damage within the DNA, the cytoplasm, and the outer plasma membrane of prostate carcinoma cells. This elicits a series of intracellular signal-transduction cascades leading to G1, S-phase, and G2 cell cycle arrests and attempted DNA repair. Initial sensing of cellular damage also activates the epidermal growth factor receptor (EGFR) pathway, including downstream RAS and phosphatase and tensin homolog (PTEN)-AKT signaling. Mutations in the genes involved in these pathways or increased levels of intratumoral hypoxia can lead to relative tumor cell resistance and decreased tumor cell kill. These radioresistance pathways are targets for a number of novel molecular agents that may be used in combination with radiotherapy. For prostate carcinoma cells that sustain lethal amounts of damage, tumor cell death occurs in a dose-responsive manner through terminal growth arrest, mitotic catastrophe, and possibly, apoptosis. Androgen deprivation likely increases radiation-induced cell death in vivo by augmenting terminal growth arrest. For patients with subclinical metastatic disease, androgen deprivation also may arrest the growth of systemic metastases terminally or transiently and alter rates of disease progression. To our knowledge, there is no assay currently available that directly determines the killing of prostate carcinoma clonogens in a quantitative manner. Instead, clinical surrogates for cell kill include postradiotherapy prostate-specific antigen (PSA) values and their nadir and/or the presence of active disease within prostate biopsies taken 2–3 years after radiotherapy and stained for biomarkers of cell proliferation. SA-βgal: senescence-associated β-galactosidase.

Gene or protein expression relating to tumor cell apoptosis also has been tested as a predictive assay for radiocurability. Pollack et al.45 have altered expression of the antiapoptotic protein BCL-2 or the proapoptotic protein BAX as independent biomarkers predicting for bNED after radiotherapy alone. However, staining for p53, BAX, or BCL-2 as potential predictive factors of radiocurability remains controversial.35, 46 This may be due to small clinical sample sizes, differences in the quantitation or timing of immunohistochemical endpoints, and variable clinical treatment parameters. Alternate mechanisms of prostate cell death (e.g., mitotic catastrophe or terminal growth arrest) (see Fig. 1) also could confound analyses that are focused solely on apoptotic endpoints.47, 48

Other potential prognostic biomarkers relate to cell proliferation (e.g., proliferating cell nuclear antigen and Ki-6745, 49) or terminal cell growth arrest (e.g., p53, p21WAF, p16INK4a, senescence-associated β-galactosidase).35, 48 The terminal arrest proteins are activated by lethal DNA damage during radiotherapy and can correlate to clonogenic cell kill.48 Indeed, the loss of p16INK4a expression within pretreatment prostate biopsies predicted for an increased risk of local failure and distant metastatic disease in patients with locally advanced prostate carcinoma who were treated within the RTOG 86-10 trial.50

Both gene expression profiling (e.g., DNA microarrays) and protein expression profiling (e.g., serum or tumor proteomics) soon may “fingerprint” those patients who harbor resistant disease on the basis of these genetic and/or microenvironmental factors. The recent use of DNA microarrays associated with robust bioinformatics has identified unique prognostic gene clusters and specific biomarkers that are independent of PSA, T classification, and Gleason score.51, 52 For example, the genes hepsin, MUC1, and AZGP1 may be important new markers that can differentiate between the least aggressive and most aggressive prostate carcinomas.53, 54

Other studies using comparative genomic hybridization have characterized unique molecular identifiers associated with Gleason score pattern 3 or 4.55, 56 This may be useful in discerning the relative prognosis for patients who have tumors with a Gleason score of 7 within the intermediate-risk category (e.g., Gleason score 3 + 4 vs. 4 + 3), because it has been considered traditionally that tumors with a higher component of the 4 pattern indicate a worse prognosis.57, 58 This information may help distinguish between truly aggressive tumors and the effect of “Gleason score shift” over the last decade. The latter shift or “creep” is the observed phenomenon of migration of Gleason scores upward associated with improved prognosis, most probably due to increased rates of PSA screening and altered pathologic scoring approaches.59–62 Improved biologic prognostic factors, thus, are required to prevent aggressive treatment based on a small component of Gleason score = 4 (e.g., 3 + 4) that may not be associated with an adverse prognosis.

Clinical Radiotherapy and Intermediate-Risk Prostate Carcinoma

The success or failure of radical radiotherapy depends on the daily proportionate killing of tumor clonogens, which make up < 1% of the total population of cells within a tumor.63 Both preclinical and clinical data support a dose-response relation for prostate carcinoma, in which increased killing of tumor clonogens leads to increased local control.32, 64–67 Currently, local control of prostate carcinoma after radical radiotherapy is inferred from postradiotherapy PSA kinetics and/or PSA nadir values or by the absence of active tumor within postradiotherapy biopsies.30, 48, 68, 69 We now review the use of these endpoints and others to appraise critically the use of dose-escalated EBRT or conventional EBRT in combination with AD as therapeutic options for patients with intermediate-risk prostate carcinoma.

Clinical Outcomes after Dose Escalation for Intermediate-Risk Prostate Carcinoma

Three comprehensive, systematic reviews of dose escalation for patients with localized prostate carcinoma have been completed.70–72 These reviews summarize the results from cohort studies and describe multiple sources of bias inherent in the literature. Three important biases arise from the use of higher radiotherapy doses as technology improves over time. Regarding the first bias, many cohort studies report bFFS using the ASTRO consensus definition. This definition back-dates failure to the midpoint between the nadir PSA and the first of three consecutive PSA rises.11 Because of this back-dating, the ASTRO consensus definition favors cohorts with shorter follow-up.12, 73, 74 In addition, it has been shown on multivariate analysis that “treatment year” (a surrogate for stage migration) is a significant predictor of bFFS.75 Because the radiation dose tends to track with treatment year, this stage migration represents another bias that favors the patients with shorter follow-up. The third bias relates to Gleason score shift migration (see above). Pollack and colleagues have shown, using a matched-pair analysis, that upgrading of Gleason scores occurred over the period 1992–1997 in an early and late cohort treated with conformal radiotherapy. Over this period, increasing numbers of tumors were upgraded to Gleason scores ≥ 6–8. This led to an apparent improvement in biochemical outcome for the most recent cohort across all Gleason score groups.59 With these caveats, for the endpoint of bFFS, Vicini et al. identified improved outcomes in patients with intermediate-risk prostate carcinoma who were treated with dose-escalated EBRT.70 Brundage et al., as part of the Ontario Practice Guidelines process, also found considerable evidence for the use of dose-escalated EBRT (e.g., doses > 74 Gy) in the intermediate-risk group.71 Nilsson et al., using the principles adopted by the Swedish Council of Technology Assessment in Health Care, reached the same conclusion.72

Kuban et al. reported the bFFS of a multiinstitutional cohort of 4839 patients with T1 and T2 prostate carcinoma who were treated with EBRT alone and were followed for a minimum of 5 years.76 Their cohort included 2190 patients with intermediate-risk disease. With a median follow-up of 6.3 years, the authors observed a significant improvement in PSA outcomes for patients who received doses > 72 Gy compared with patients who received lesser doses (e.g. 70% vs. 55% 5-year bFFS; P < 0.0001). Those authors verified that differences in neither the duration of follow-up nor the treatment year had an impact on their finding that a higher total radiation dose improved biochemical outcome.

The only definitive method for testing the hypothesis of an EBRT dose-response relation in prostate carcinoma is through prospective RCTs. In the RCT of dose escalation reported by Pollack et al., the patients with intermediate-risk prostate carcinoma (defined as PSA > 10 ng/mL) who were treated with a dose of 78 Gy had improved 5-year bFFS outcomes compared with patients who were treated with a dose of 70 Gy (62% vs. 43% 5-year bFFS).77 In the recently reported RCT by Zietman et al. comparing 70.2 Gy equivalent (GyE) (a combination of photons and protons) with 79.2 GyE in patients with low-risk and intermediate-risk disease, there was a significant benefit from dose escalation in the intermediate-risk group (78.7% vs. 60.5% bFFS; note that the majority of patients in the study had low-risk disease with Gleason scores ≤ 6 and PSA < 10 ng/mL).67 Within the next few years, other RCT data will become available to elucidate the benefit of dose-escalation in EBRT for patients with intermediate-risk prostate carcinoma.78, 79

Toxicity Associated with Dose-Escalated Radiotherapy

The normal tissue toxicity associated with dose-escalated EBRT also has been reviewed.80 Significantly worse rectal and bladder toxicity was reported by both physicians and patients for the 78-Gy cohort arm compared with the 70-Gy arm in the RCT by Pollack et al.77, 81 However, this largely has historic interest, because these patients were treated with suboptimal radiation techniques compared with to today's standards. With appropriate attention to dose-volume constraints for the bladder and rectum, and with the use of intensity-modulated radiotherapy (IMRT), doses of > 80 Gy can be given safely to patients with prostate carcinoma. For example, at a median follow-up of 3 years, Zelefsky et al. found that RTOG Grade 3 rectal toxicity and bladder toxicity occurred in 0.8% and 0.6% of patients, respectively.82, 83 However, caution should be exercised in interpreting toxicity data with short follow-up. Late effects could continue to increase beyond the 5-year mark, and there is great interest in standardizing the approach to both the recording and analysis of late toxicity at periods > 5–10 years.84, 85 Indeed, analysis of late toxicity in the second decade after combined photon and proton prostate radiotherapy to 77.4 Gy suggests that, although severe gastrointestinal toxicity is rare, genitourinary morbidity continues to develop well into the second decade associated with high rates of posttreatment hematuria.86

Clinical Outcomes after Combined Hormonoradiotherapy for Intermediate-Risk Prostate Carcinoma.

An understanding of the interaction between AD and radiotherapy during prostate carcinoma cell kill may provide unique insights into the future role of AD as a modifier of tumor cell radiosensitivity. Initially, it was believed that the preclinical radiosensitization of prostate xenografts observed with AD and large, single radiation doses was secondary to increased radiation-induced apoptosis.87 However, this was not confirmed when AD was combined with more clinically relevant fractionated irradiation protocols. Instead, the radiosensitization in vivo was correlated with increased tumor cell growth arrest.35, 48, 88–90 In their preclinical studies of AD plus radiation, Kaminski et al.91 suggested that AD may lead to delayed tumor regrowth through both increased clonogen cell kill (see Fig. 1) and/or a reduced growth rate of surviving clonogens.

These preclinical data support the clinical use of AD plus EBRT in patients who harbor micrometastatic disease at the time of diagnosis (in whom the metastatic cells may be arrested permanently or significantly by AD) or in patients with radioresistant tumors that may be sensitized by adjunctive or concurrent AD. A review of RCTs and subgroup analyses in which AD was given concurrent with EBRT versus AD given in an adjuvant setting suggests that concurrent AD-EBRT synergistically improves local control in the prostate and pelvis. This is consistent with a unique biologic effect different from that of adjuvant AD given post-EBRT, because the latter would impact solely on systemic disease and would affect overall survival and not local control.16, 92

The RCT data presented in Table 2 show that improvement in local control and/or disease-free or overall survival can be achieved when AD is combined with conventional-dose EBRT.5, 6, 29, 93–96 The trials with the greatest numbers of patients did not address combined-modality therapy in the intermediate-risk group. However, D'Amico et al.97 recently reported improved survival in 206 patients who were treated with 70 Gy plus 6 months of AD compared with patients who received 70 Gy alone (88% vs. 78%; P < 0.04). Approximately 80% of those patients could be classified with intermediate-risk disease. A number of concerns have been raised concerning the trial, however, including the fact that the difference in survival was based on only six prostate carcinoma-specific deaths in the control arm, compared with no prostate carcinoma-specific deaths in the experimental arm.98, 99 Nonetheless, another study93 showed improved PSA-based outcomes using 64 Gy plus AD. In that RCT, 70% of patients could be classified with intermediate-risk disease. More recently, Dearnaley et al.100 reported a nonsignificant trend toward improved freedom from PSA failure in patients who were treated with 74 Gy, rather than 64 Gy, after 3–6 months of neoadjuvant AD. The vast majority of patients in both arms in that trial had World Health Organization Grade 2 pathology (68–77%; 5–15% of patients had Grade 3 pathology), and 74–83% of patients had T2 or T3 tumors and median baseline PSA values of 14–15 ng/mL. In that study, the 5-year actuarial control rates were 71% versus 59%, respectively (P = 0.10); no survival data were reported at the time of the current report.

Table 2. Trials of Radiotherapy Alone versus Combined Radiotherapy and Androgen Deprivation
D'Amico et al., 200497Laverdiere et al., 200493RTOG 8531 (Pilepich et al., 20036 and Lawton et al., 200494)RTOG 8610 (Pilepich et al., 200195)EORTC 22863 (Bolla et al., 2002 and Bolla et al., 199796)
  • RTOG: Radiation Therapy Oncology Group; EORTC: European Organization for Research and Treatment of Cancer; GS: Gleason score; Gy: gray; NAD: neoadjuvant androgen deprivation; CAD: concurrent androgen deprivation; AAD: adjuvant androgen deprivation; IAD: indefinite androgen deprivation; N/A: not available in reference.

  • a

    PSA: prostate specific antigen (ng/ml).

No. of patients with intermediate-risk disease (definition)a∼ 163/206 (PSA = 10–20 or GS = 7)∼112/161 (T2 clinical stage)0/9770/4560/415
Treatment arms70 Gy vs. 70 Gy + 2 mos each NAD, CAD, and AAD1) 64 Gy vs. 2) 64 Gy + 3 mo NAD vs. 3) 64 Gy + 3 mo NAD and 6 mo CAD and AAD65–70 Gy vs. 65–70 Gy + IAD65–70 Gy vs. 65–70 Gy + 4 mo NAD and CAD70 Gy vs. 70 Gy + 3 yr CAD and AAD
Time of reported endpoints5 yrs7 yrs10 yrs8 yrs5 yrs
Local controlN/AN/A61% vs. 77% P < 0.000158% vs. 70% P < 0.0277% vs. 97% P < 0.001
Biochemical control (endpoint)66% vs. 79% P = 0.001 (No AD for salvage)1) 42% vs. 2) 66% P = 0.009; 1) 42% vs. 3) 69% P = 0.003 (PSA < 1.5)9% vs. 30% P < 0.0001 (PSA < 1.5)3% vs. 16% P < 0.0001 (PSA < 1.5)45% vs. 76% P < 0.0001 (PSA < 1.5)
Disease-specific mortality93% vs. 100% P = 0.02N/A17% vs. 22% P = 0.00523% vs. 32% P = 0.056% vs. 21% P = 0.0001
Overall survival88% vs. 78% P = 0.04N/A38% vs. 53% P = 0.00444% vs. 53% P < 0.162% vs. 78% P = 0.0002

Despite these recent data, two issues confound current decision making with regard to the role of AD in patients with intermediate-risk prostate carcinoma. The first issue is that the largest randomized trials that have shown a benefit to adjunctive AD (either neoadjuvant or adjuvant) combined with EBRT largely have been completed in patients with high-risk disease. Thus, the conclusions from those trials may not be applicable to all patients with intermediate-risk disease. The second issue is that many of these trials were completed in the era of conventional-dose EBRT (e.g., doses < 74 Gy), when long-term bFFS rates with EBRT at these lower doses alone were approximately 40% in patients with intermediate-risk disease.101, 102

Nonrandomized, single-institution series have reported clinical outcome data regarding the role of AD in addition to dose-escalated EBRT for patients with intermediate-risk prostate carcinoma. Kupelian et al.103 reported on the treatment outcomes in a cohort of 1041 consecutively treated patients with T1–T2 prostate carcinoma who were treated either with radical prostatectomy, EBRT, and brachytherapy (permanent seed implantation) or with combined brachytherapy and EBRT. Seven hundred eighty-five patients were treated with EBRT (484 patients received ≤ 72 Gy and 301 patients received > 72 Gy), and 143 of those patients were given neoadjuvant AD for ≤ 6 months. Although AD was found to be a significant predictor of biochemical outcome on univariate analysis for the entire patient cohort, when the group of patients who received ≤ 72 Gy was excluded, it was no longer significant (P = 0.91). Zelefsky et al. also reported in their cohort of 772 patients (89% with T1–T2 disease; treated with IMRT to a median dose of 81 Gy) that AD appeared to have no influence on bFFS (median follow-up, 24 months).82 Furthermore, a lack of benefit (and a possible detrimental effect) of short-course AD on 5-year metastasis-free survival or cause-specific survival was reported by Martinez et al. in a large retrospective review of 1260 patients who were treated with combined EBRT and brachytherapy.104 Although these studies do not represent RCT data, their uniform findings suggest that the addition of AD to dose-escalated EBRT may not be required to optimize biochemical outcomes for patients intermediate-risk prostate carcinoma.

The recent RCT by D'Amico et al. is promising. Compared with the randomized trials of dose-escalation alone that had similar control arms of 64–70 Gy, only the study by D'Amico et al. showed a survival benefit; and their study also had a greater number of patients with intermediate-risk and high-risk disease.29 Further data showing similar survival benefits will be required before uniformly recommending AD plus conventional-dose EBRT versus dose-escalated EBRT for the intermediate-risk group. However, it is conceivable that selected patients may benefit from short-term AD plus dose-escalated EBRT if their tumors have adverse features that reflect local radioresistance and/or increased systemic spread.

Toxicity of Adjunctive AD

Long-term AD is the treatment of choice for patients with high-risk prostate carcinoma on the basis of randomized trials, which show a survival advantage.5, 6, 105 The physiological side effects of long-term AD are well known and include anemia, sarcopeniam and osteoporosis.106–108 More relevant to the treatment of intermediate-risk prostate carcinoma is the toxicity of AD administration for a period of 3–6 months. Hot flashes occur in approximately 80% of men.109 Decreased libido, erectile dysfunction, and fatigue also are experienced in the majority of treated men.110, 111 Reversible depression and cognitive impairment also have been linked to short-term use of hormones, but the correlation is not consistent across all studies.104, 106

When patients were treated with a luteinizing hormone-releasing hormone agonist and an antiandrogen as complete AD (CAD) therapy, a decline in hemoglobin (Hgb) of > 1.0 g/dL was observed in 75% of patients after 2 months.112 D'Amico et al. studied biochemical outcomes in a cohort of 110, mostly intermediate-risk patients (and some high-risk patients) treated with 6 months of CAD and radiotherapy.113 Based on values taken prior to and 1 month after AD, it was found that patients who had a drop in Hgb > 1 g/dL had significantly poorer biochemical outcomes (median follow-up, 20 months).113 This finding is somewhat unexpected, because patients with high-risk prostate carcinoma have improved biochemical outcomes with long-term AD (Table 2). However, the Hgb effect is similar to that observed during radical radiotherapy studies in patients with cervical carcinoma114 and head and neck carcinoma.115, 116 It is possible that the AD, which improves survival by controlling metastatic disease in high-risk patients, impairs biochemical control by reducing local control in intermediate-risk patients. This awaits a formal subgroup analysis of the recent RCT of 6 months of CAD and EBRT versus EBRT alone.29 In the meantime, these results suggest that caution is warranted when AD and EBRT are used for the treatment of patients with intermediate-risk prostate carcinoma outside of the context of a clinical trial.

The Future Relevant Clinical Endpoints for Future Clinical Trials

Although overall survival and PCSM remain the gold standards in determining the efficacy of new treatment approaches, the natural history of prostate carcinoma necessitates prolonged observation periods to identify these endpoints. PSA-based failure may not always be a reliable surrogate for PCSM; because, in some studies, up to 20% of patients who died after developing PSA failure died of nonprostate carcinoma-related causes.117 The use of “time to second PSA failure” (or development of hormone-refractory disease) has been suggested as a potential surrogate endpoint for PCSM in localized disease, but this observation has yet to be validated in multiple data sets.118

Reports of outcomes from trials of AD and EBRT that use a biochemical endpoint face several challenges. With neoadjuvant AD, it is difficult to choose a start point for follow-up that does not favor one arm over the other in the short term. In addition, there is a period of recovery of testosterone after cessation of the AD that may lead to a rising PSA profile. This is particularly problematic for the ASTRO consensus definition of bFFS and may lead falsely to the conclusion that neoadjuvant-treated patients fail at a higher rate. More recently, the groups of D'Amico et al. and Albertson et al. suggested that the posttreatment PSA doubling time can be a useful surrogate for PCSM.31, 117, 119 The rising testosterone profile seen in patients after treatment with AD likely would generate a falsely short PSA doubling time, rendering this measure difficult to interpret in AD-treated patients in the short term. Future definitions of biochemical failure that predict clinical failure better after radiotherapy may supercede the currently used ASTRO definition.120, 121 These include failure defined when the PSA is greater than a current nadir + 2 ng/mL or 3 ng/mL and dated at call (i.e., the failure date called when that criterion was met12, 122) or 2 consecutive rises of at least 0.5 ng/mL beyond the nadir and back-dated.123 Based on a recent, multiinstitutional, pooled analysis of 4839 patients, these latter definitions appear to have increased sensitivity and specificity relating to clinical and distant failure.120

Studies of neoadjuvant AD with surgery have shown significant reductions in the detection of positive surgical margins without a long-term improvement in bFFS. This observation supports the concept that AD alters the tissues and reduces the ability of pathologists to detect disease at the surgical margin without eliminating viable, unresected disease.124 Similarly, AD and/or EBRT therapy effects on tissue histology can confound biopsy interpretation and its role in predicting local control.49, 125, 126 Furthermore, patients are reluctant to undergo biopsies if their PSA profile is stable, and clinicians are reluctant to insist on biopsies due to perceived risks of infection, bleeding, or patient discomfort. This may introduce a substantial bias in the reporting of biopsy endpoints unless a large proportion of similarly treated patients agree to undergo posttreatment biopsies. Nonetheless, local control, as assessed by prostatic biopsies 2–3 years after radiotherapy, perhaps also stained for relevant biomarkers (e.g., p21WAF, p16INK4a, Ki-67), also may provide relevant early endpoints for future trials of AD plus dose-escalated EBRT.

Ongoing Trials in Intermediate-Risk Disease

A search of the MetaRegister of Controlled Trials (available from URL: www.controlled-trials.com [accessed August 15, 2004]) identified a number of ongoing studies that involve intermediate-risk patients (Table 3). The RTOG P-0126 and MRC trials are investigating dose-escalation using conventional fractionation schemes. The Princess Margaret Hospital PMH 99-07 and the EORTC 22991 studies may provide some answers regarding the usefulness of neoadjuvant hormone therapy in intermediate-risk patients treated with dose-escalated EBRT. Both the RTOG studies, 94-08 and 99-10, use conventional doses of EBRT and will not address the issue of the utility of AD in the setting of high-dose EBRT. The Memorial Sloan-Kettering Cancer Center study is comparing high-dose EBRT alone with the use of neoadjuvant AD and EBRT. In the latter study, if an equivalent PSA-based outcome is achieved in both arms, then the study may be extremely useful for the radiobiologic calculation of the radiation dose-equivalent cell kill achieved by the addition of AD.

Table 3. Unreported Trials of Dose Escalation and/or Androgen Deprivation that Included Patients with Intermediate-Risk Prostate Carcinoma from the MetaRegister of Controlled Trialsa
StudyEligibility criteriabTreatment armsSample size
  • RTOG: Radiation Therapy Oncology Group; PSA: prostate-specific antigen; Gy: gray; NAD: neoadjuvant androgen deprivation; CAD: concurrent androgen deprivation; EORTC: European Organization for Research and Treatment of Cancer; GS: Gleason score; AAD: adjuvant androgen deprivation; PMH: Princess Margaret Hospital; NAA: neoadjuvant antiandrogen; CAA: concurrent antiandrogen; MSKCC: Memorial Sloan-Kettering Cancer Center; MRC: Medical Research Council; FCCC: Fox-Chase Cancer Center.

  • a

    Available from URL: http:www.controlled-trials.com [accessed August 15, 2004].

  • b

    PSA: prostate specific antigen (ng/ml).

RTOG 9408T1b–T2b and PSA ≤ 2075.6 Gy vs. 75.6 Gy with 4 mos of NAD and CAD1980
EORTC 22991T1b–c and either PSA = 10–50 or GS = 7; or T2a and PSA < 5070–78 Gy vs. 70–78 Gy with 6 mos of CAD and AAD800
PMH 9907T1b–T2b, PSA = 4–20 and GS = 7–10; or T1b–T2b, PSA = 10–20 and GS ≤ 675.6–79.8 Gy vs. 75.6–79.8 Gy with 3 mos of NAA and 2 mos of CAA378
RTOG 9910T1b–T4, GS = 2–6 and PSA 10–100; or T1b–T4, GS = 7 and PSA < 20; or T1b–T1c, GS = 8–10 and PSA < 2070.2 Gy with 2 mos of NAD and CAD vs. 70.2 Gy with 6 mos of NAD and 2 mos of CAD1540
MSKCCAt least 2 of: PSA > 10, GS > 7, or T486.4 Gy vs. 75.6 Gy with 2 yrs of NAD, CAD, and AAD400
RTOG P-0126T1b–T2b, GS = 2–6 and PSA = 10–20; or T1b–T2b, GS = 7 and PSA < 1570.2 Gy (39 fractions in 7.8 weeks) vs. 79.2 Gy (44 fractions in 8.8 weeks)1520
MRCT1b–T3a and PSA < 5064 Gy vs. 74 Gy800
FCCCT1b–T3c and 1 of: PSA > 10, GS ≥ 7, or Stage ≥ T2b76.0 Gy (36 fractions in 7.2 weeks) vs. 70.2 Gy (26 fractions in 5.2 weeks)300

The Fox Chase Cancer Center study is comparing moderate dose escalation with a hypofractionated regimen. Recent analyses of clinical results have suggested that the α/β ratio for prostate carcinoma is approximately 1.5 Gy (much lower than the typical value of 10 Gy for many other tumors) and, thus, is comparable to or lower than the ratio of the surrounding, late-responding rectal mucosa (i.e., an α/β ratio of approximately 3 Gy127). The lower α/β ratio for prostate carcinoma compared, with surrounding, late-responding, normal tissue, creates the potential for therapeutic gain. Indeed, recent reports of protocols that delivered 70 Gy with 2.5 Gy per fraction128, 129 or 50 Gy with 3.13 Gy per fraction130 are consistent with bNED rates of 60–80% and were associated with favorable toxicity profiles. This may result in future combined-modality trials using hypofractioned EBRT with or without AD.

Molecular-Based Protocols for Intermediate-Risk Prostate Carcinoma

Information regarding the molecular signaling pathways that are activated after ionizing radiation has increased exponentially over the last decade.131 A number of protein pathways have been associated with tumor radioresistance, including intratumoral hypoxia and the p53, RAS, epidermal growth factor receptor (EFGR), vascular endothelial growth factor receptor (VEGF), and phosphoinositide-3 kinase-phosphatase and tensin homolog (PTEN)/AKT pathways.35, 36, 132

The pretreatment assessment of biomarkers that represent these signaling pathways may lead to the appropriate choice of novel, molecular-targeted agents to use in combination with EBRT. For example, tumor cell kill as a result of terminal growth arrest explains the slow kinetics of decreasing PSA values and a final nadir occurring over a 12–16 month period after EBRT for prostate carcinoma.30, 48 In light of this finding, new treatment strategies that augment EBRT-induced terminal growth arrest in vivo may be possible clinically using histone deactylase inhibitors or retinoids.133 These agents also may radioprotect normal tissues134, 135 and currently are being tested prospectively in Phase I/II trials.136, 137

New molecular therapies that target p53, MDM2, BCL-2, BAX, PTEN/AKT, RAS, or clusterin pathways also hold promise for augmenting radiation-induced tumor cell kill and improving local control.46, 138–149 Improved outcomes also may be achieved with the use of hypoxia-targeted drugs in combination with EBRT. Preliminary data suggest that the neoadjuvant use of AD increases tumor oxygenation, which may be another mode of radiosensitization for this combination.150 Clinical interventions designed to improve oxygenation through direct hypoxic cell targeting or altered angiogenesis are underway in other tumor sites using a variety of agents (e.g., HIF1-α gene therapy; tirapazamine; VEGF inhibitors, such as SU5416 or SU6668; and cyclooxygenase-2 inhibitors) and could be used in hypoxic subgroups of patients with intermediate-risk disease.151–153 The recent success of the EGFR inhibitors (e.g., cetuximab) as radiosensitizers in head and neck carcinoma, leading to improved overall survival with minimal excess toxicity, serves as an excellent model for molecular-targeted EBRT approaches in future studies.140, 154

Conclusions and Recommended Management of Intermediate-Risk Patients with Radiotherapy

The heterogeneity of intermediate-risk prostate carcinoma presents a challenge to genitourinary oncology in terms of prognosis and optimal management. Although there is reasonable evidence that these patients benefit from dose-escalated EBRT or adjunctive hormone therapy with conventional EBRT, to our knowledge there is little evidence to date that these patients uniformly benefit from adjunctive AD when the total dose is > 74 Gy.

Currently, patients in the intermediate-risk group should be entered into well designed, RCTs of combined dose-escalation and AD of sufficient power to answer the important questions raised by nonrandomized studies. These trials also should investigate the optimal duration of hormone therapy. In addition, they should be stratified by new prognostic markers and should be accompanied by strong, correlative, scientific endpoints that can address risk-group heterogeneity and potential predictive factors for local or systemic recurrence.

With the advent of molecular profiling and the use of tissue arrays that facilitate the simultaneous study of thousands of genes or proteins within large patient cohorts, clinicians soon will be able to acquire individual pharmacogenomic and prognostic profiles for use in the clinical management of patients. This likely will lead to individualized treatment for patients with intermediate-risk prostate carcinoma and will stimulate novel combined-modality protocols to decrease local and distant failure in this diverse clinical prostate carcinoma population.


The authors thank Dr. Charles Catton for critical reading of this review and the members of the Princess Margaret Hospital-University Health Network Genito-Urinary Oncology Group for stimulating discussions.