A re-assessment of the role of combined androgen blockade for advanced prostate cancer


L. Klotz, Division of Urology, Sunnybrook and Women's College Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario, M4N 3M5, Canada.
e-mail: Laurence.klotz@sw.ca


Combined androgen blockade is a controversial topic, which has arguments both for and against. It is revisited by the authors of this mini-review, with a full discussion on the benefits and cautions with this approach. A wide range of other issues is also addressed in this section: bilateral testicular cancer, male-factor infertility, and buccal mucosa urethroplasty. All of these are of interest to general urologists, as well as to those with a more specific area of interest.


cyproterone acetate


Prostate Cancer Trialists’ Collaborative Group


hazard ratio.


The addition of an antiandrogen to surgical or medical castration, termed combined therapy, was first described in 1979. Twenty-five years later, much debate still surrounds the benefits of combined therapy compared with castration alone. Here we address the question ‘Does combined therapy have a role in the current treatment of metastatic prostate cancer?’. We review pertinent study data, particularly that relating to the nonsteroidal antiandrogen bicalutamide and present an analysis combining historical trial data that provides an estimate of the benefit of combined therapy using bicalutamide compared with castration alone.


The testes produce most of the serum testosterone and the adrenal glands produce the remaining androgens [1]. After castration (whether surgical or medical), the adrenal glands continue to produce the androgens, androsterone and dehydroepiandrosterone. These are converted to testosterone in the peripheral tissues and in the prostate gland (Fig. 1). After castration minimal amounts of testosterone may persist, derived from the adrenal and residual testicular secretion.

Figure 1.

The rationale for combination therapy consisting of castration plus an antiandrogen. ACTH, adrenocorticotropic hormone; CRH, corticotropin-releasing hormone; EGF, epidermal growth factor; IGF-1, insulin-like growth factor-1; IL-6, interleukin-6; KGF, keratinocyte growth factor; T, testosterone.

It is well established that antiandrogens act through competition with the testosterone metabolite dihydrotestosterone and other androgens for binding sites on the androgen receptors in the prostate cell nucleus. Their blockade of the androgen receptor activation by nonsteroidal growth factors, cytokines, and other nonligand dependent activators is less well recognized [2]. This latter mechanism is important in the androgen-depleted environment and implies that the antiandrogens should be considered more accurately as ‘androgen receptor antagonists’. The combination of medical or surgical castration plus an antiandrogen blocks the action of androgens produced by both the testes and adrenal glands. Androgen-dependent genes responsible for prostate cell function and division are activated by the androgen receptor (Fig. 2). Therefore, inhibiting receptor activation (either by androgen or ligand independent activators) by antiandrogens promotes apoptosis and inhibits prostate cancer growth.

Figure 2.

Functional regions of the androgen receptor. AF, activation function; Zn, zinc.



Both steroidal antiandrogens, e.g. cyproterone acetate (CPA) and chlormadinone acetate, and the nonsteroidal antiandrogens bicalutamide, flutamide and nilutamide, have been investigated in clinical trials as part of combined therapy for advanced prostate cancer. A survival disadvantage for combined therapy with the steroidal antiandrogen CPA compared with castration alone has been demonstrated in a subset analysis of data from the Prostate Cancer Trialists’ Collaborative Group (PCTCG) meta-analysis [3] (discussed in more detail later).


The properties of the three nonsteroidal antiandrogens differ in several important respects. One such property is their blockade of nonandrogenic activation of androgen receptors. These factors include cytokines, e.g. interleukin-6, growth factors such as IGF and epidermal growth factor, and signal transduction factors such as protein kinase A. These factors are capable of activating normal androgen receptors. In the presence of androgen-receptor mutations the androgen receptor may become ‘promiscuous’ and be activated by a much wider variety of ligands, cytokines and other molecules. The nonsteroidal antiandrogens differ in the degree to which they block androgen independent activation [4]. Flutamide has been shown to activate cells with specific single-point mutations of promiscuous androgen receptors identified from patients with androgen insensitivity syndrome [4]. Similarly, nilutamide caused transcription of a mutant androgen receptor extracted from a human metastatic cancer and with a point mutation in the same codon as the mutation in the androgen-independent cell line LNCaP, while bicalutamide maintained antagonistic action [2]. However, in a novel cell subline, LNCaP-abl, which has a hypersensitive proliferative response to androgen, bicalutamide showed agonistic effects on androgen-receptor transactivation activity and was unable to block androgen effects [5]. Flutamide also exerted stimulatory effects on androgen receptor activity that was 2.4–4 times greater in LNCaP-abl cells than in LNCaP cells. For bicalutamide, the induction of reporter gene activity was lower than that seen with flutamide. The stimulatory effects of bicalutamide on androgen receptor activity were 2–2.5-times greater in LNCaP-abl cells than in LNCaP cells.

There are differential effects of the nonsteroidal antiandrogens in their interaction with androgen receptor co-suppressors and co-activators. For example, bicalutamide has been shown to activate the nuclear androgen receptor co-suppressor N-CoR and inhibit the co-activator SRC-1. Both these effects would result in inhibition of cell growth signals by activated androgen receptors. However, the effect of flutamide in this system is much more muted [6].

These studies emphasize the important biological differences between the nonsteroidal antiandrogens in the androgen-depleted environment and suggest that bicalutamide may be superior to flutamide and nilutamide in delaying androgen-independent progression. Other key differences between the nonsteroidal antiandrogens are their androgen receptor binding affinities and potencies, clinical efficacies and tolerability profiles.

In vitro studies show that the binding affinity of bicalutamide for the human and rat prostate androgen receptor is 2–4 times greater than that of flutamide and twice that of nilutamide. Moreover, bicalutamide showed greater potency than flutamide in reducing intact rat ventral prostate and seminal vesicle weights [7].

In a randomized, double-blind direct comparison trial in 813 men with metastatic prostate cancer [8] there was a trend to improved overall survival with bicalutamide plus an LHRH agonist than with flutamide plus an LHRH agonist (median survival 180 vs 148 weeks), although the difference did not achieve statitical significance (hazard ratio, HR, 0.87; 95% CI 0.72–1.05; P = 0.15). This is the only trial comparing two nonsteroidal antiandrogens as components of combined therapy.

The nonsteroidal antiandrogens also have different tolerability profiles. Flutamide is associated with diarrhoea [9,10] and liver toxicity [10]. High rates of visual disturbances and alcohol intolerance have been reported with nilutamide [10]. Bicalutamide is better tolerated [1,10]. In the trial by Schellhammer et al.[8] the incidence of diarrhoea was statistically significantly lower with bicalutamide plus an LHRH agonist than with flutamide plus an LHRH agonist (12% vs 26%; P < 0.001). In that study, haematuria was the only adverse event to occur significantly more frequently with bicalutamide plus agonist than with flutamide plus agonist (12% vs 6%; P = 0.007); in most cases this was mild to moderate, unrelated to treatment (96%), and did not lead to treatment withdrawal. Overall, the incidence of withdrawal from therapy due to an adverse event was lower with bicalutamide plus agonist (10%) than with flutamide plus agonist (16%).


Evidence for the effectiveness of combined therapy compared with castration alone comes from individual trials, meta-analyses of trial results and, in the case of combined therapy using bicalutamide, from the latest analysis of historical trial data.


Crawford et al.[11] reported the first large controlled trial (603 men) to show an advantage of combined therapy over castration alone for treating metastatic prostate cancer. There were significant improvements in progression-free (median 16.5 vs 13.9 months; P = 0.039) and overall (median 35.6 vs 28.3 months; P = 0.035) survival with leuprolide plus flutamide compared with leuprolide plus placebo.

The largest randomized trial of combined therapy vs. monotherapy conducted to date studied bilateral orchidectomy plus either flutamide or placebo in 1387 patients with metastatic prostate cancer [9]. There was a trend for improved survival (a survival benefit of ≈ 10%) with combined therapy over castration alone, but this was not statistically significant (HR for risk of death with flutamide vs placebo, 0.91, 90% CI 0.81–1.01; P = 0.14). The trial was powered to detect the 25% difference that had been observed in the study by Crawford et al.[11]. At progression, the treatment arm was unblinded and patients who had been on placebo could then receive open-label flutamide; therefore, the trial actually compared initial with delayed combined therapy.

To date, over 31 randomized, long-term (treatment for > 1 year) trials in > 8000 men with advanced prostate cancer have investigated the effectiveness and tolerability of combined therapy, with variable results [3]. Differences between therapies are unlikely to be detected during the early follow-up, as most deaths within 1–1.5 years from diagnosis of advanced cancer are likely to be from causes other than cancer. This significantly reduces the power to detect differences in any analysis/study with a relatively short follow-up.


In an attempt to better understand this large volume of trial data there have been several meta-analyses [3,12,13]. The benefits and limitations of the various meta-analyses of studies comparing combined therapy with castration alone were reviewed previously [14]. The most complete assessment of evidence is the meta-analysis published in 2000 by the PCTCG, which assessed data from 27 trials (including a total of 8275 patients) and incorporated individual patient data [3].

The PCTCG meta-analysis found an overall trend for improved overall survival in patients treated with combined therapy compared with castration alone, although this was not statistically significant (HR 0.958; sem 0.026; P = 0.11, two-sided) (Fig. 3); the survival differences were not apparent before 2 years of follow-up. When steroidal and nonsteroidal antiandrogens were disaggregated, combined therapy with the nonsteroidal antiandrogens flutamide and nilutamide was associated with a statistically significant 8% decrease in the risk of death over castration alone (95% CI 3–13; P = 0.005, two-sided); this translated into a 2.9% absolute improvement in 5-year survival (Fig. 3) [14]. However, combined therapy with the steroidal antiandrogen CPA was associated with a statistically significant 13% increase in the risk of death compared with castration alone (95% CI 0–27; P = 0.04, two-sided), and this equated to a 2.8% reduction in 5-year survival.

Figure 3.

Overall survival data as HRs for combined therapy compared with castration alone from published meta-analyses. Figure reproduced with permission from [4]. Derived from information in [3];†Relative risks using data from [3]; CPA, cyproterone acetate; FLUT, flutamide; NILUT, nilutamide; NSAA, nonsteroidal antiandrogen; NSAA(LH), relative risks (log HR) from P values; NSAA(PH), relative risks (proportional hazards) from discrete proportional hazards model.

In the PCTCG meta-analysis [3] the results appeared to be independent of patient age, disease stage or whether surgical or medical castration was used. Although the results of the trials differed, such variation could be expected by chance (the test of treatment- by-trial interaction was not significant, P > 0.1).


As bicalutamide was not available when most combined vs monotherapy studies were conducted, no data on this agent were available for inclusion in the meta-analysis discussed above. There is no direct comparison of combined therapy using bicalutamide 50 mg with castration alone. A double-blind randomized trial is currently ongoing in Japan of bicalutamide 80 mg (the registered dose in Japan) combined with an LHRH agonist vs an agonist plus placebo in 205 men with advanced prostate cancer. Preliminary data at a median follow-up of 15 months indicate that combined therapy with bicalutamide has significant benefits over LHRH agonist monotherapy; there was an improvement in PSA normalization rate (79.4% vs 38.6%; P < 0.001); reduction of the risk of treatment failure (median time to failure 22.1 vs 15.6 months; P = 0.038) and progression (median time not reached; P = 0.015); and improvement in quality of life (P < 0.001). There were too few events to assess overall survival [15].

As described earlier, bicalutamide was compared in a large trial with another nonsteroidal antiandrogen, flutamide, in the combined therapy setting (HR 0.87; 95% CI 0.72–1.05) [8]. At the time this trial was designed a direct comparison of combined therapy using bicalutamide with castration alone was considered unethical, as combined therapy was considered standard care and superior to monotherapy. To understand the role of bicalutamide in combined therapy we used data from the Schellhammer trial [8] in conjunction with the PCTCG meta-analysis data [3] for flutamide plus castration vs castration alone (HR 0.92, 95% CI 0.86–0.98) to calculate an estimate of the likely benefit of bicalutamide combined therapy vs castration alone (Fig. 4).

Figure 4.

HRs and 95% CI for overall survival for bicalutamide plus castration vs flutamide plus castration, flutamide plus castration vs castration alone, and for bicalutamide plus castration vs castration alone.

Conventional wisdom dictates that data from different trials should not be directly compared. In fact , a methodologically validated technique has evolved to integrate the results of trials which share a common arm (in this case, flutamide as MAB) but differ in the alternate arm. Typically, this is employed where a placebo arm may no longer be feasible or ethical. Stringent criteria regarding comparability of the patient populations must be met. The PCTCG trials and the Schellhammer trial, both of which enrolled D2 patients from the pre-PSA era, appear to meet these criteria. Rothmann et al.[16] applied the method to estimate the effect of capecitabine relative to 5-fluorouracil alone for metastatic colorectal cancer, by combining the results from a trial comparing capecitabine with 5-fluorouracil plus leucovorin and a meta-analysis of trials comparing 5-fluorouracil plus leucovorin to 5-fluorouracil alone. Fisher et al.[17] described a similar application to estimate the effect of clopidogrel relative to placebo in patients with myocardial infarction, ischaemic stroke or symptomatic peripheral arterial disease. Results from an active-controlled trial comparing clopidogrel with aspirin were used with data from 40 trials of aspirin vs placebo to obtain an estimate of the effect of clopidogrel vs placebo.

Using these methods the HR for combined therapy using bicalutamide vs castration alone is estimated by multiplying the HR for bicalutamide combined therapy vs flutamide combined therapy with the HR for flutamide combined therapy vs castration alone (i.e. 0.87 × 0.92 = 0.80, see Appendix).

On applying this analysis the balance of evidence suggests that there is a high probability (98.5%) that bicalutamide as part of combined therapy provides a survival advantage over castration alone. The HR is 0.80, indicating a 20% reduction in the risk of death, with a 95% CI of 0.66–0.98, indicating that this benefit could range from an absolute benefit of 2% to 34% (Fig. 4). These CIs are calculated from the combined sem, which was larger than the sem from either [8] or the PCTCG meta-analysis [3], and reflects a greater uncertainty when combining results (see Appendix). The key assumptions made when estimating the effect of bicalutamide are that the effect of flutamide in the trials in the PCTCG meta-analysis [3] was of a similar magnitude to that in the study population of Schellhammer et al.[8]. This would require there to be no important prognostic factors which were represented differently between the study populations (such as the extent of metastatic disease) upon which the size of the effect of bicalutamide relative to flutamide would differ, and that patients in the trials included in the PCTCG meta-analysis [3] were managed similarly to those in the comparative trial of bicalutamide and flutamide [8]. The main limitation of combining data across studies is that it is impossible to completely verify the above assumptions. However, the statistical consistency of the effect of flutamide in the PCTCG analysis [3] provides some reassurance of the validity of the assumptions.



When considering therapeutic options the clinician and patient must consider many factors. These include the patient's disease status, benefits of therapy, treatment side-effects, quality-of-life issues and cost. Individual trial and meta-analysis data suggest that there is a modest improvement in overall survival with combined therapy using nonsteroidal antiandrogens over castration alone in advanced disease, although these do not appear until after 2 years. An overall survival benefit, although modest, in the aged population where there is a high competing risk of death from other causes is noteworthy. The analysis reported here using historical data estimates the benefit of bicalutamide combination therapy as a 20% reduction in the risk of death (HR 0.8; 95% CI 0.66–0.98) over castration alone.

Compared with other cancer interventions the cost of combined therapy in advanced prostate cancer is reasonable [18]. Using Canadian drug costs and assuming a 4–7-month survival benefit with combined therapy in advanced prostate cancer, the estimated cost of combined therapy with bicalutamide per month of survival benefit is calculated as Can$437-$1107 [18]. The estimates of cost per month of survival gained are higher with other cancer therapies [18]. For example, when vinorelbine is added to cisplatinum for nonsmall-cell lung cancer (median 2-month survival benefit) the cost is calculated as Can$1241 per month of survival gained. For metastatic colorectal cancer, irinotecan added to 5-fluorouracil and leucovorin provides a 2–3-month survival benefit and the cost is calculated to be Can$11 214 per month of survival gained.


Trial data suggest that patients with minimal metastatic disease may have the greatest benefits from combined therapy. It may also be possible to use molecular markers, e.g. androgen-receptor gene amplification in tumours, to identify patients who would derive the greatest benefit from combined therapy. Further research is warranted to identify subsets of patients who would benefit most from combined therapy.

The timing of nonsteroidal antiandrogen administration in combined therapy is an important consideration. Can the benefit of combined therapy be obtained by initiating antiandrogens at the time of progression? To date, no trial comparing early vs delayed combined therapy has been conducted. PSA responses to antiandrogens given at the time of biochemical progression occur in about half of patients, but these responses are generally of short duration (median 3 months) [19]. A reasonable inference from these data is that this is not likely to translate into an equivalent survival benefit.


The use of combined therapy in prostate cancer management remains controversial. Such therapy with nonsteroidal antiandrogens offers a modest survival benefit compared with castration alone, which must be balanced against the potential for an increase in side-effects and a consequent adverse effect on the patient's quality of life. In this review we have placed the data from the trial comparing bicalutamide and flutamide in combined therapy into context with previously reported meta-analyses. An estimate of the benefit of combined therapy with bicalutamide suggests there is a high probability (estimated to be 98.5%) that bicalutamide 50 mg combined therapy provides a survival advantage over castration alone. The HR for survival in this analysis was 0.80, indicating a 20% reduction in the risk of death with combined therapy using bicalutamide over castration alone (with 95% CI indicating that the absolute survival benefit of 2–34%). When using combined therapy other properties of the nonsteroidal antiandrogens should also be considered, i.e. tolerability, binding affinity and the ability to block androgen-independent activation of the androgen receptor in the androgen-depleted environment.


Laurence Klotz and Paul Schellhammer are members of the Combination Therapy Advisory Group and group meetings are supported by AstraZeneca Pharmaceuticals.


L. Klotz is a study investigator for AstraZeneca and a paid member of the advisory board. P. Schellhammer is a study investigator for AstraZeneca. K. Carroll is an employee of AstraZeneca.


Combination of uncertainty in estimates of the likely benefit of bicalutamide combined therapy compared with castration alone. The formula for calculating the combined sem, where X1 represents the Schellhammer data [8] and X2 the PCTCG data [3]:

sem(X1 + X2) = √(sem(X1)2 + sem(X2)2)

sem   =   √(0.09592 + 0.034552)

=  0.1019

Calculation of 95% CI:

95% CI = mean ± 1.96 sem

Mean HR = 0.80 = 0.87 × 0.92

95% CI log scale = −0.223 ± 1.96 (0.1019)

=   −0.42, – 0.02

Exponentiate to original scale:

95% CI = 0.66, 0.98