Claude Schulman, Clinic Edith Cavell, Rue Edith Cavell 32, 1180 Brussels, Belgium. e-mail: firstname.lastname@example.org
In many patients with prostate cancer, androgen deprivation therapy (ADT) is administered over prolonged periods of time. The benefits of long-term ADT in patients with advanced disease are well established and, more recently, studies have shown that long-term adjuvant ADT used in combination with radiotherapy improves survival in patients with earlier stages of disease. Nevertheless, clinicians should remain aware of the potential long-term side effects of ADT and the strategies that can be used to manage or prevent long-term complications.
One such strategy is intermittent androgen deprivation (IAD), in which patients receive cycles of ADT, the duration of which is usually determined by PSA levels. Accumulating data indicate that this approach improves the tolerability of ADT (particularly sexual dysfunction) and patients' quality of life, without compromising clinical outcomes (progression and survival). Indeed, the latest European Association of Urology guidelines state that IAD should no longer be considered investigational. Nevertheless, some questions remain unanswered, including: who are the most suitable patients for IAD and what are the optimal PSA levels for stopping and restarting treatment?
Osteoporosis (and the resultant increased risk of fractures) is a well-recognized complication of long-term ADT. Bone mineral density should be measured before and during long-term ADT and patients advised to make appropriate lifestyle changes to help preserve bone health. Pharmacological intervention is also an option. Denosumab (an NF-κB ligand inhibitor) significantly reduces ADT-induced bone loss and the risk of fractures in patients with non-metastatic disease. In those whose disease has metastasized, zoledronate and denosumab are licensed to prevent skeletal-related events and a large randomized study has shown that denosumab is more effective than zoledronate in this setting.
For many years, androgen deprivation therapy (ADT) has been the mainstay of treatment for advanced prostate cancer. Treatment is typically continued life-long and many patients will receive therapy over prolonged periods. More recently, long-term adjuvant ADT with radiotherapy has become the standard of care in those with high-risk localized or locally advanced disease . Although the benefits of long-term treatment in these settings are well established (see Schröder et al. in this supplement), patients may also have to live with ADT-related side effects for many years.
The side effects of ADT are a consequence of induced sex-steroid deficiency and vary in their degree of morbidity and effects on patients' quality of life (QoL) [2–5]. Commonly recognized short-term side effects of ADT include hot flushes, loss of libido or impotence, and fatigue . There are also numerous long-term complications associated with ADT including osteoporosis and skeletal complications, altered body composition and worsening insulin resistance, metabolic syndrome, diabetes mellitus and potentially increased cardiovascular risk, arterial stiffness, and cognitive decline [6,7]. This well-known side-effect profile of ADT may impact negatively on patients' QoL  and any decision to use ADT should take into account both the clinical benefits and potential short- and long-term side effects of treatment, and include measures to prevent and/or manage ADT-related complications. This article will review some of the current strategies for managing the side effects associated with ADT, particularly the use of intermittent androgen deprivation (IAD) and approaches to minimize the potential long-term complications of ADT on bone health.
USE OF IAD TO IMPROVE TOLERABILITY OF ADT
RATIONALE FOR IAD
IAD involves alternating androgen deprivation with treatment cessation to allow testosterone recovery between treatment periods and has been investigated as an alternative clinical strategy to continuous androgen deprivation (CAD). In general, IAD involves the administration of ADT for a fixed interval, until castration-induced apoptosis is maximal and tumour regression is induced; ADT is then discontinued before the development of the androgen-independent phenotype . The on- and off-treatment periods are then repeated. In practice, discontinuation and re-initiation are triggered by the attainment of pre-specified PSA levels, although actual levels vary between studies, as do the medication(s) used.
The aims of IAD are to minimize treatment-related side effects and improve patient QoL compared with CAD, without compromising the overall efficacy of treatment [1,9,10]. Theoretically, IAD also has the potential to delay progression to castrate-resistant disease. Supporting the rationale for IAD, preclinical studies have shown that castration-resistant status is associated with adaptive stem-cell survival mechanisms activated by androgen deprivation , and IAD restores androgen sensitivity and delays the time to progression to castration resistance [12,13]. A meta-analysis of phase II trials involving 1446 patients with prostate cancer has shown that 39% of time was spent off treatment after a median of two IAD cycles , while a similar analysis of a phase III trial revealed that patients spent 49% of time off therapy after 15 cycles of IAD . Therefore, IAD also has the potential to offer cost savings vs CAD (assuming outcomes are similar/better) by reducing medication use and the management costs associated with the side effects of continuous therapy .
CLINICAL EVIDENCE SUPPORTING IAD
A recent systematic review has evaluated evidence from 19 key phase II IAD trials and interim data from 8 randomized phase III IAD trials using GnRH agonists with or without anti-androgens in >2000 patients with prostate cancer . In general, ADT was well tolerated in the phase II studies, with a decrease in adverse events and improvement in QoL during the off-treatment period. More recently, IAD has been compared with CAD in several randomized phase III trials, based on the safety and feasibility of the approach in the smaller studies. Phase III trials have involved patients with locally advanced, recurrent and/or metastatic prostate cancer; a combination of a GnRH agonist and an anti-androgen was used in all but one study  (Table 1). The power of these phase III studies to detect differences in long-term outcomes has been questioned , but accumulating data, particularly from the larger trials, indicate that IAD has similar efficacy to CAD for overall survival with similar or possibly longer time to progression [15,24–26].
Table 1. Characteristics of phase III studies comparing intermittent and continuous androgen deprivation
Number of patients randomized to IAD or CAD after initial ON period;
b based primarily on patient numbers and duration of follow-up;
c interim data;
d reported for most recent analysis;
data from analysis at a median of 51 months' follow-up ;
f information taken from clinicaltrials.gov (NCT00003653);
numerically greater for CAD (24 vs 18 months for IAD) but P-value not specified;
number of patients recruited into study for initial ON period; + endpoint improved for the IAD vs CAD arms; = similar results obtained in IAD vs CAD arms; AA, anti-androgen; ADT, androgen deprivation therapy; CAD, continuous androgen deprivation; CPA, cyproterone acetate; GnRHa, GnRH agonist; IAD, intermittent androgen deprivation; NA, not applicable; NR, not reported; OFF, off treatment; ON, on treatment; OS, overall survival; PSA, prostate-specific antigen; RP, radical prostatectomy; SEUG, South European Uroncological Group; SWOG, Southwest Oncology Group; TID, thrice daily.
Phase III data supporting IAD: SWOG JPR.7 and SEUG 9401 trials
Recently, two large-scale phase III trials have shown that IAD appears to have a beneficial effect on treatment-related side effects and QoL vs CAD, with no negative impact on overall survival and time to progression. In the Southwest Oncology Group (SWOG) JPR.7 trial – which is the largest phase III study with the longest follow-up (83 months) in this setting – median overall survival was similar in the IAD and CAD arms (8.8 vs 9.1 years, respectively; hazard ratio [HR] 1.02, non-inferiority P= 0.009) . Similarly, in the South European Uroncological Group (SEUG) 9401 phase III trial, overall survival was comparable between IAD and CAD (HR 0.99, P= 0.84) at a median follow-up of 51 months (Figure 1) , and at an extended median follow-up of 57 months (HR 0.96, P= 0.61) . There were slightly more cancer-related deaths with IAD vs CAD in SWOG JPR.7  and SEUG 9401 . However, this was balanced by a slightly higher rate of non-cancer-related deaths in the CAD arm in SWOG JPR.7  and an excess of cardiovascular deaths and deaths from other causes in the CAD arm at the extended follow-up of SEUG 9401 . Time to progression was significantly improved with IAD vs CAD in SWOG JPR.7 (HR 0.80, P= 0.024) , but no significant between-group difference was observed for time to progression in SEUG 9401 (HR 0.81, P= 0.11) .
The tolerability profile of IAD was generally better than CAD in the two studies, particularly for sexual function, and QoL benefits were also observed with IAD. In SWOG JPR.7, there were fewer hot flushes in the IAD group vs CAD, but no other differences in side effects were observed, including myocardial events and osteoporotic fractures . Patients treated with IAD had significantly improved QoL vs CAD (P≤ 0.01) in several domains including: physical function, fatigue, urinary problems, hot flushes, desire for sexual activity and erectile function . In SEUG 9401, several side effects appeared to be more frequent with CAD vs IAD, although this only reached statistical significance for hot flushes (P < 0.01) . Patients who received IAD reported greater sexual activity than patients on CAD: 28% and 10% of patients reported sexual activity at 15 months after randomization, respectively (P < 0.01).
Improved tolerability for IAD vs CAD was also reported in another large phase III study, SEUG 9901 . The incidence of side effects was significantly lower in the IAD arm (P < 0.0001), particularly for hot flushes (7% vs 23%), gynaecomastia (10% vs 33%) and headaches (5% vs 12%) . In addition, significantly fewer patients in the IAD group experienced a decline in sexual function.
The tolerability and QoL data from SWOG JPR.7, SEUG 9401 and SEUG 9901 are supported by several other analyses, which suggest better tolerability and, possibly, QoL with IAD than CAD. Using data from five randomized studies that included 1382 patients, IAD was associated with slightly fewer adverse events and the incidence of impotence was significantly reduced (P= 0.008) . In a more recent, systematic review of phase II studies, a decrease in adverse events was noted during the off-treatment period . Recovery of sexual function in particular was highlighted in several studies . In those studies for which QoL was reported, three reported benefits for IAD vs CAD, while two studies reported no differences . In this context, the results of the phase III SWOG 9346 trial [NCT00002651] involving >1500 men are eagerly awaited, as the primary outcomes are QoL measures.
USING PSA TO PREDICT TREATMENT SUCCESS WITH IAD
A meta-analysis of 10 phase II IAD trials (n= 1446) has shown that the initial PSA level, the PSA nadir following treatment, and the PSA threshold for restarting treatment are all important predictors of outcome in patients receiving IAD . The predictive value of PSA on survival after treatment induction has also been demonstrated in the phase III SWOG 9346 trial . In this study of patients with new metastatic disease, those with a PSA level of ≤4 ng/mL after 7 months of treatment had one-quarter of the risk of dying compared with those with a PSA level of >4 ng/mL (HR 0.26, P < 0.0001) (Figure 2). These results suggest that IAD should be considered in patients who respond to ADT with a decline in PSA levels to ≤4 ng/mL. If the PSA level does not drop to ≤4 ng/mL after 6–9 months of initial ADT, IAD should not be considered. In another study of IAD in patients with metastatic disease, there was a trend for reduced risk of progression with a PSA nadir of ≤0.2 ng/mL vs >0.2–4 ng/mL .
In a recent analysis of two studies that evaluated the combination of radiotherapy and 6 months' ADT in men with localized or locally advanced prostate cancer, PSA nadir and PSA post-treatment levels of >0.5 ng/mL were associated with an increased risk of prostate cancer-specific mortality . It was concluded, therefore, that PSA levels could be used to select patients who may need more prolonged treatment: men with PSA levels of >0.5 ng/mL at the end of treatment should be considered for long-term androgen suppression, while those with a PSA nadir of >0.5 ng/mL should be considered for inclusion in randomized trials investigating the use of drugs that have extended survival in castration-resistant metastatic prostate cancer .
THE FUTURE OF IAD
Available evidence from phase II and III trials has led to the endorsement of IAD by the European Association of Urology (EAU). The 2012 EAU guidelines on prostate cancer state “IAD is currently widely offered to patients with prostate cancer in various clinical settings, and its status should no longer be regarded as investigational”. The guidelines recommend that the initial (induction) cycle should last between 6 and 9 months. Criteria for stopping treatment are: a PSA level <4 ng/mL in metastatic disease, or <0.5 ng/mL in relapsing disease. Treatment should be resumed if the patient experiences clinical progression, or if PSA levels rise to 4–10 ng/mL in non-metastatic situations, or 10–15 ng/mL in patients with metastatic disease. However, the optimal PSA thresholds at which ADT must be stopped or resumed are empirical. Therefore, further clarification is required to define the initial treatment period and the PSA triggers for stopping and restarting therapy. Further investigation to establish the appropriate selection of patients for IAD is also warranted. As most of the large phase III studies are not yet mature, definitive evidence of the non-inferiority of IAD vs CAD in terms of overall survival and progression, and superiority of safety, tolerability and QoL endpoints is also needed. In addition, it is currently unclear if a GnRH agonist may be used alone, as published experiences are based on GnRH agonist plus anti-androgen therapy. The guidelines state that GnRH antagonists may be a valid alternative if clear results are obtained from randomized trials.
The GnRH antagonist, abarelix, was initially licensed to treat advanced prostate cancer , but is no longer marketed in the USA, although it is still available in Germany. Degarelix, the first of a new generation of GnRH antagonists, is licensed in both the USA and the EU for the treatment of hormone-dependent advanced prostate cancer. Degarelix produces immediate blockade of GnRH receptors, leading to rapid suppression of testosterone and PSA levels, whilst avoiding the testosterone surge (flare) associated with GnRH agonists [35–37]. The rapid PSA suppression observed with degarelix makes it a suitable candidate for IAD.
Two phase III trials are currently investigating IAD with degarelix in prostate cancer: CS29 and CS37. The European CS29 trial (NCT00801242) is an open-label, uncontrolled, non-comparative, two-cycle trial of degarelix 240/80 mg in patients requiring ADT. Following an initial 7-month treatment period, patients with PSA levels <4 ng/mL stop therapy until PSA reaches >4 ng/mL (maximum 24 months), at which point degarelix is re-started. The primary endpoint is time to PSA >4 ng/mL in the off-treatment phase and secondary endpoints include time to testosterone >50 ng/dL and >220 ng/dL in the off-treatment phase, time to PSA >2 ng/mL in the off-treatment phase, QoL and safety/tolerability. The US CS37 trial (NCT00928434) is a randomized, controlled trial comparing IAD with degarelix vs CAD with degarelix or leuprolide in patients with biochemical failure after localized therapy. Patients are being randomized (1:1:1) to one of three treatment arms: degarelix 240/80 mg for 14 months, leuprolide for 14 months, or degarelix 240/80 mg for 7 months followed by a 7-month off-treatment period. Only patients with PSA ≤2 ng/mL after the first 7 months of treatment are eligible to continue in the trial. The primary endpoint is the proportion of patients with PSA levels ≤4.0 ng/mL on intermittent degarelix vs CAD (degarelix and leuprolide combined) at 14 months. Secondary endpoints include QoL, sexual function, and testosterone and PSA levels. Results from these trials are eagerly awaited.
MAINTAINING BONE HEALTH DURING ADT
Despite its advantages, not all patients will be suitable for IAD and even those who do benefit from this strategy will eventually require CAD for optimal testosterone control. Additional strategies are, therefore, required to help manage the side effects of long-term treatment with ADT, one of which is reduced bone mineral density (BMD).
IMPACT OF ADT ON BONE TURNOVER
Osteoporosis is a disease characterized by a deterioration of the bone architecture and a reduction in BMD, leading to an increased risk of fracture. It is relatively common in men, with the lifetime risk of fracture estimated to be between 13 and 25% . The major causes of osteoporosis in men include alcohol abuse, chronic glucocorticoid therapy and hypogonadism. Both prostate cancer itself and its treatment with ADT can have a substantial impact on bone health . ADT causes severe hypogonadism and results in a progressive loss of BMD and an increased risk of fractures in patients with prostate cancer [40–45]. The accelerated bone loss is likely to be related to increased bone turnover [40,46]; changes in skeletal sensitivity to the bone-resorbing effects of parathyroid hormone might also contribute to GnRH agonist-induced BMD loss .
ADT using either orchidectomy or GnRH agonists with or without anti-androgens has consistently been shown to decrease BMD in patients with prostate cancer [40–42,48–50]. The decrease in BMD is most dramatic in the first year after ADT initiation , but there is further loss with continued treatment . Consequently, the development of osteoporosis appears to increase steadily with duration of ADT. A study involving patients with prostate cancer treated with long-term ADT (≥2 years' therapy; n= 266) showed that the prevalence of osteoporosis was 42.9%, 49.2%, 59.5%, 65.7% and 80.6% after 2, 3, 6, 8 and ≥10 years of ADT, respectively . Moreover, analysis of the Surveillance, Epidemiology and End Results Medicare database (n= 50613) found that 19.4% of patients with prostate cancer treated with ADT experienced a fracture within 5 years of diagnosis compared with a fracture rate of 12.6% in those not receiving ADT, with the risk of fracture increasing with ADT duration . Skeletal fracture has been shown to be an independent negative predictor of overall survival in patients with prostate cancer . Therefore, strategies to preserve bone health and prevent or reduce the risk of fracture are of great importance for this patient population.
STRATEGIES TO REDUCE BONE LOSS AND FRACTURES DURING ADT
Normal bone function requires a balance between osteoclast activity (bone resorption) and osteoblast activity (bone formation)  and therapies that inhibit osteoclast activity have been shown to alleviate the skeletal complications of ADT. Currently available treatment options include: bisphosphonates, selective oestrogen-receptor modulators (SERMs) and receptor activator of NF-κB ligand (RANKL) inhibitors.
Bisphosphonates inhibit the recruitment and differentiation of osteoclast precursors, and the attachment and survival of mature osteoclasts . Zoledronate is licensed for the prevention of skeletal-related events in patients with advanced malignancies and bone metastases. In a randomized, double-blind study in patients with hormone-refractory metastatic prostate cancer, zoledronate significantly reduced the incidence of skeletal-related events vs placebo (38% vs 49%; P= 0.028) . Several 1-year, randomized, controlled trials in prostate cancer have also demonstrated that alendronate , pamidronate  and zoledronate  significantly reduced ADT-related bone loss compared with placebo. However, unlike denosumab (see below), bisphosphonates are not licensed to treat ADT-induced bone loss in prostate cancer.
Selective oestrogen-receptor modulators
In vitro data suggest that SERMs inhibit osteoclast-mediated bone resorption . The SERM raloxifene has been shown to significantly increase BMD after 1 year compared with no raloxifene treatment in patients with prostate cancer receiving ADT (n= 48) . In a phase III trial of 1284 men with prostate cancer receiving ADT and at elevated fracture risk, toremifene significantly reduced fracture risk by 50% compared with placebo (P= 0.05) at 2 years . Toremifene also significantly increased BMD at the lumbar spine, hip and femoral neck (P < 0.0001 for all comparisons) and decreased bone turnover markers (P < 0.05) vs placebo. In the toremifene group, 17 patients (2.6%) had venous thromboembolic events, compared with 7 patients in the placebo group (1.1%); events were more common in those aged >80 years and in those who experienced prolonged immobilization. In a post-hoc analysis of men aged <80 years, the risk of thromboembolic events was 2.1% for toremifene vs 1.0% for placebo (P= 0.26) .
Receptor activator of NF-κB ligand inhibitors
Denosumab is a fully human monoclonal antibody that binds to, and blocks the activity of, RANKL, the main driver of osteoclast formation, function and survival . The protective effects of denosumab on bone loss and incidental vertebral fractures were investigated in a large, randomized, placebo-controlled, phase III trial (Hormone Ablation Bone Loss trial [HALT]) involving 1468 patients with prostate cancer receiving ADT and at elevated fracture risk . Mean lumbar spine BMD at 24 months was increased by 5.6% with denosumab, compared with a 1.0% loss with placebo (P < 0.001); significant between-group differences were seen as early as 1 month and were sustained through 36 months. BMD at the total hip, femoral neck and distal 1/3 radius was also significantly increased with denosumab vs placebo at all time points (1 month through 36 months; P≤ 0.001). Importantly, patients treated with denosumab had a decreased incidence of new vertebral fracture at 12, 24 and 36 months (Figure 3). The 3-year risk of new vertebral fractures was reduced by 62% with denosumab (P= 0.006 vs placebo). These benefits of denosumab were not associated with any increase in toxicity, as rates of adverse events were similar in both treatment groups. In a further subgroup analysis of this study (n= 309), denosumab significantly increased BMD at all measured skeletal sites for every subgroup analyzed, including those at greatest risk of fracture (ie older men and those with prevalent fractures) . Based on the results of the HALT trial, denosumab was licensed in both the USA and the EU for the treatment of bone loss associated with ADT in men with non-metastatic prostate cancer at increased risk of fractures.
The benefits of denosumab for preserving bone health go beyond preventing ADT-related bone loss. A large, randomized phase III trial of 1904 patients with castration-resistant prostate cancer and bone metastases compared denosumab with zoledronate for the prevention of skeletal-related events . Denosumab significantly delayed the median time to first skeletal-related event compared with zoledronate (20.7 vs 17.1 months, respectively; HR 0.82, P= 0.0002 for non-inferiority, P= 0.008 for superiority). Denosumab also significantly delayed the time to first and subsequent skeletal-related events compared with zoledronate (composite endpoint). Between treatment groups, no apparent differences in overall survival were observed and the overall incidence of adverse events was similar. Based on the results from this study and studies in breast cancer, other solid tumours and multiple myeloma, denosumab was approved in the USA and the EU for preventing skeletal-related events in patients with bone metastases from solid tumours.
Recently, denosumab became the first agent to demonstrate prevention of bone metastases in prostate cancer . The randomized, controlled, phase III 147 trial compared denosumab with placebo for delaying the development of bone metastases in 1432 men with metastatic castration-resistant prostate cancer (not spread to bone at baseline). Denosumab significantly increased bone metastasis-free survival by >4 months vs placebo (29.5 vs 25.2 months, respectively; HR 0.85; P= 0.028). Denosumab also significantly delayed the time to first bone metastases by 3.7 months vs placebo (HR 0.84; P= 0.032). Overall survival and adverse events were similar between treatment groups. These encouraging results suggest that denosumab may offer a potential new option for bone metastasis in castration-resistant prostate cancer. However, denosumab is not currently licensed for the prevention of bone metastases in prostate cancer.
RECOMMENDATIONS FOR MONITORING BONE HEALTH AND PREVENTING BONE LOSS IN MEN RECEIVING LONG-TERM ADT
The EAU guidelines on prostate cancer recommend that before initiating long-term ADT, a baseline evaluation of BMD should be performed by dual X-ray absorptiometry and, thereafter, regular BMD measurements should be based on the initial T-score . For an initial T-score <1.0, BMD should be measured every 2 years. Annual BMD assessments should be made if the T-score is between 1.0 and 2.5 in the absence of associated risk factors. A high risk for subsequent non-metastatic fracture (initial low BMD; T-score >2.5, or >1 if other risk factors are present) suggests the need for early use of preventive bone-loss therapy. To protect against non-metastatic bone fractures, increased exercise and calcium supplementation are recommended. Other lifestyle changes recommended before starting long-term ADT include cessation of smoking, decreased alcohol consumption and normalization of body mass index; some centres also advocate vitamin-D supplementation. As noted above, pharmacological intervention – using denosumab – is also an option to prevent ADT-induced bone loss in men with non-metastatic prostate cancer. In patients whose disease has progressed to a metastatic stage, zoledronate or denosumab may be used to prevent skeletal-related events; denosumab is more effective than bisphosphonates in this setting (see above).
In conclusion, the use of IAD for prostate cancer is no longer considered experimental by the EAU, but further work is required to define the optimal treatment parameters and to provide definitive evidence of the tolerability and QoL benefits vs CAD (assuming equal efficacy). Further investigation is needed to define patient selection, the criteria for stopping/restarting ADT and the optimal type of ADT. Indeed, GnRH antagonists may be a viable alternative to GnRH agonists for IAD and are currently under investigation in clinical trials. There is also the intriguing question of whether highly effective, well-tolerated agents may reduce or even negate the need for IAD given the questions that remain about its use.
For those patients receiving long-term ADT, associated bone loss can lead to an increased risk of fracture and these patients should be considered for drug therapy. Various agents have been shown to reduce ADT-induced bone loss, but only denosumab is currently licensed in this setting. For patients whose disease has progressed to the metastatic stage, denosumab is more effective than zoledronate for preventing skeletal-related events. More recent data also suggest that denosumab may prevent the development of bone metastases in patients with prostate cancer.
In conclusion, IAD and bone protection are two of the strategies that clinicians can use to swing the balance in favour of the considerable benefits associated with the long-term use of ADT in patients with prostate cancer.
Editorial assistance with this manuscript was provided by Nicky French, Bioscript Stirling, funded by Ferring Pharmaceuticals.