As listed in Tables 1 and 2, six phase III randomized trials were identified in the systematic review of the evidence.21-27 These trials spanned the years from 2000 to 2014 (a 14-year period), and none compared similar interventions. The primary outcome for all trials was therapeutic efficacy, although it was framed in a variety of ways, such as rPFS,23 OS,23 time to PSA progression,24 time to progression in general,25 and median duration of response.25,27
Upon review of the available evidence, the Expert Panel concluded that the majority of the evidence was insufficient to inform evidence-based recommendations and that formal expert consensus would be needed to help inform clinical opinions. Results of the consensus ratings can be found in the Data Supplement 8: Consensus Panel review results.
Quality Assessment of the Literature
The modest number of randomly assigned patients in the majority of the identified trials created obstacles with respect to determination of the true efficacy or generalizability of the findings. An additional challenge with the data identified by the systematic review was the lack of similar treatment arms; no two trials included the same comparisons. The largest trial included 1,717 participants.28 However, most of the remaining trials included fewer than 140 patients per comparison arm.24-27,29 The trial of abiraterone acetate in chemotherapy-naïve patients reported significant PFS results, which led to the trial being stopped early.21 The enzalutamide trial in chemotherapy-naïve patients also reported a PFS advantage and was stopped early.24 Patients were offered crossover in both studies.
Table 1. Characteristics of Phase III Randomized Trials
Abbreviations: AAWD, antiandrogen withdrawal; CAB, combined androgen blockade; DES, diethylstilbestrol; Dex, dexamethasone; LHRH, luteinizing hormone–releasing hormone; NR, not reported; PO, orally.
*Patients were enrolled before AAWD; in Dawson,27 only 15 such patients were enrolled before protocol change.
†On the basis of reported use of opioid analgesics.
‡Patients were offered the choice of continuing with LHRH; approximately 47% continued.
§On the basis of reported European Cooperative Oncology Group performance status.
‖Patients who received CAB rather than intermittent androgen blockade.
¶Patients were required to continue use of initial LHRH.
#Continuation of initial LHRH NR.
Table 2. Results of Phase III Randomized Trials
Abbreviations: AAWD, antiandrogen withdrawal; CR, complete response; DES, diethylstilbestrol; Dex, dexamethasone; LHRH, luteinizing hormone–releasing hormone; NR, not reported; OS, overall survival; PFS, progression-free survival; PO, orally; PR, partial response; PSA, prostate-specific antigen; SD, stable disease.
*Patients were enrolled before AAWD; only 15 such patients were enrolled before protocol change.
Research Question 1
Should a castrate state be maintained in patients who develop CRPC?
PCO 1. For men who develop CRPC despite castrate levels of testosterone:
Patients should be maintained in a castrate state indefinitely. This PCO is based on indirect scientific evidence and current understandings of disease progression mechanisms in prostate cancer. A discussion with patients about the limited nature of available scientific evidence and the balance among potential harms, benefits, costs, and patient preferences is essential when planning treatment.
A castrate state should be maintained through orchiectomy or pharmacologic castration (eg, luteinizing hormone–releasing hormone [LHRH] agonists/antagonists, antiandrogens).
Literature review and analysis. No randomized controlled trials (RCTs) met the sample size inclusion criteria. Within the supplementary literature, one small RCT suggested a cost and potential cause-specific survival advantage for intermittent versus continuous androgen blockade in men who develop CRPC (who have not had an orchiectomy), but the study was not adequately powered.30 However, retrospective post hoc analyses of a prospective series reported that eugonadal or superphysiologic levels of testosterone are associated with a risk of progression and death in men with CRPC.29 Multiple adverse effects and harms were reported with ADT, including hot flushes, fatigue, impotence, gynecomastia, loss of libido, osteoporosis, and a risk for metabolic syndrome.31-33
Clinical interpretation. Maintenance of a castrate state through orchiectomy or pharmacologic castration in patients who develop CRPC despite castrate levels of testosterone is suggested, which is supported by current understandings of disease progression mechanisms34 and agrees with published guidelines.35-37 RCTs are needed, such as the ongoing German SPARE trial of abiraterone acetate plus LHRH therapy versus abiraterone acetate–sparing LHRH therapy in chemotherapy-naïve patients with progressive CRPC, to measure the clinical benefit of continued ADT (LHRH therapy) during second-line hormonal therapy (ClinicalTrial.gov identifier NCT02077634).
Research Question 2
In chemotherapy-naïve patients who develop CRPC but have no radiographic evidence of metastases (M0 CRPC), should second-line hormonal therapies be used? If so, what agents or specific sequence of agents should be offered?
Literature review and analysis. No phase III randomized trials have evaluated the association between second-line hormonal therapies and clinical outcomes in patients with CRPC and no radiographic evidence of metastases (M0 CRPC). However, evidence from supplementary literature has suggested associations between PSA absolute value and rate of rise with clinical outcomes in this population.3 No evidence provides guidance about the optimal order of second-line hormonal therapies for patients with M0 CRPC.
In 201 patients with M0 CRPC, Smith et al3 found that baseline PSA levels > 10 ng/mL and higher PSA velocities were independently associated with shorter time to first bone metastasis, OS, and metastasis-free survival. A PSA doubling time of < 6 months was associated with significantly (P = .001) shorter bone metastasis–free survival compared with doubling times of 6 to 19 months and > 19 months. After multivariable adjustment, Gleason grade was not significantly associated with any of these outcomes. Second-line hormonal therapies that lower PSA values or slow the rate of rise may be reasonable for patients with castration-resistant disease and a baseline elevated PSA or a rapid PSA doubling time or velocity. (Online calculators for determining PSA doubling time or velocity can be found at nomograms.mskcc.org/prostate/psadoublingtime.aspx or www.asure.ca [PSA Calculator Tool tab].) Such men are at the greatest risk of developing metastatic disease and may, therefore, benefit from additional antitumor therapy, but this has not been prospectively demonstrated in studies to date. Age and life expectancy should be taken into consideration. Older patients with short life expectancies and a high risk of developing metastatic disease may not be optimal candidates for second-line hormonal therapies.
Within the supplemental literature was one related phase II trial (STRIVE) that compared enzalutamide (160 mg/day) to the antiandrogen bicalutamide (50 mg/day) for safety and efficacy among chemotherapy-naïve men with asymptomatic or mildly symptomatic disease (M0N0/1, n = 139; M1N1, n = 257) despite primary ADT.38 For the M0 population, although PFS significantly favored enzalutamide (hazard ratio [HR], 0.24; 95% CI, 0.14 to 0.42), median PFS was not reached. PSA response, a secondary end point, was significantly greater (P < .001) for enzalutamide versus bicalutamide, irrespective of the definition of complete response (PSA decline ≥ 50% [or 90%] from baseline). (The M1 population will be discussed under Research Question 3.) Although an important phase II study, STRIVE was not designed to compare OS among patients with clinically defined CRPC. Phase III trials are needed.
With regard to the use of corticosteroid monotherapy for men with M0 CRPC, no phase III studies were identified. In the supplemental literature, one underpowered (n = 82) single-center phase II study compared PSA response rates among chemotherapy-na¨ıve men with M0 disease randomly assigned to dexamethasone versus prednisolone,39 but the results were inconclusive. Dexamethasone may be more active than prednisolone in M0 CRPC, but trials with larger sample sizes are needed.
Clinical interpretation. Clinicians face many challenges when treating patients with CRPC, particularly those who are chemotherapy-naïve with no evidence of radiographic metastases (M0 CRPC). The absence of clinical trial data on the topic leads to uncertainty among clinicians about optimal treatment and sequencing. Cost-effectiveness has not yet been demonstrated for second-line hormonal therapy in this population. Only for patients at high risk of developing metastases did the panel feel comfortable with providing guidance in the absence of high-quality data. Sequencing was particularly troublesome given the variety of agents fairly recently approved for use by the FDA in the chemotherapynaïve population and patients still in clinical trials.40
Research Question 3
In chemotherapy-naïve patients who develop CRPC and have radiographic evidence of metastases but minimal symptoms (M1a/ M1s CRPC), should second-line hormonal therapies be used? If so, what agents are recommended?
PCO 3. After first-line hormonal treatment failure and a discussion with chemotherapy-naïve patients about potential harms, benefits, costs, and patient preferences,
Literature review and analysis. Three phase III RCTs identified in the systematic review provide the evidence base to inform this PCO.21,23,25 An RCT (COU-AA-302) of abiraterone acetate plus prednisone administered in chemotherapy-naïve men with primarily asymptomatic metastatic CRPC resulted in a statistically significant rPFS benefit compared with placebo and prednisone (median rPFS, 16.5 v 8.3 months; HR, 0.53; 95% CI, 0.45 to 0.62; P < .001). Time to opiate use, chemotherapy initiation, performance status deterioration, and PSA progression also were significantly longer in the abiraterone acetate arm (P < .01).21 After a median follow-up of 49.2 months, abiraterone acetate plus prednisone significantly prolonged OS (median, 34.7 v 30.3 months; HR, 0.81; 95% CI, 0.70 to 0.93; P = .0033) with an acceptable toxicity profile.41 Similar OS and rPFS benefits for abiraterone acetate plus prednisone versus prednisone alone were seen among men age ≥ 75 years.42
An RCT23 (PREVAIL) compared enzalutamide (160 mg oral) versus placebo administered in chemotherapy-naïve men with cytologically confirmed adenocarcinoma of the prostate with documented asymptomatic or mildly symptomatic metastases who had PSA progression, radiographic progression, or both in soft tissue or bone, despite receipt of LHRH analog therapy or orchiectomy. The trial was stopped early as a result of significantly improved survival results for patients administered enzalutamide, with an 81% reduction in the risk of radiographic progression or death at 12 months (HR, 0.19; 95% CI, 0.15 to 0.23; P < .001) and a 29% reduction in the risk of death at 18 months (HR, 0.71; 95% CI, 0.60 to 0.84; P < .001) as well as significantly improved time to initiation of chemotherapy, reduction in risk of first skeletal event, time to PSA progression, and response rate combined with an acceptable toxicity profile. Similar OS and rPFS benefits for enzalutamide were seen among men age ≥ 75 years.43 With respect to patient-reported outcomes,44 median time to deterioration in Functional Assessment of Cancer Therapy–Prostate total score was significantly longer for patients administered enzalutamide (11.3 months; 95% CI, 11.1 to 13.9 months) than placebo (5.6 months; 95% CI, 5.5 to 5.6 months; HR, 0.62; 95% CI, 0.54 to 0.72; P < .001). A significantly greater proportion of patients administered enzalutamide (v placebo) reported clinically meaningful improvements in the Functional Assessment of Cancer Therapy–Prostate total score (40% v 23%), the EuroQual Group Health Questionnaire utility index (28% v 16%), and the visual analog scale (27% v 18%; all P < .001).
In an open-label extended analysis of 787 of the 1,717 patients enrolled in the PREVAIL study, rPFS (as a post hoc analysis only) and OS were revisited after the prespecified number of deaths for the final analysis (n = 784) was reached.45 With the inclusion of data from 5 months postcrossover for the placebo group, the median follow-up was 31 months. By this point, 52% of the original 872 patients in the enzalutamide arm and 81% of the original 845 in the placebo arm had received subsequent antineoplastic therapies (chemotherapy, abiraterone acetate, sipuleucel-T, or radium-223 dichloride) known to affect survival. Similar statistics were not provided for patients in the open-label extended analysis only. Nevertheless, patients who had been treated with enzalutamide had a 23% reduced risk of death compared with those treated with placebo (35.3 v 31.3 months; HR, 0.77; 95% CI, 0.67 to 0.88; P < .001). In the post hoc analysis, enzalutamide reduced the risk of radiographic progression or death by 68% compared with placebo (20.0 v 5.4 months; HR, 0.32; 95% CI, 0.28 to 0.37; P < .001).
In the supplemental literature, two related phase II trials (TERRAIN and STRIVE) compared enzalutamide (160 mg/day) to the antiandrogen bicalutamide (50 mg/day) for safety and efficacy among chemotherapy-naïve men with asymptomatic or mildly symptomatic progressive disease during treatment with ADT.38,46 As mentioned under Research Question 2, STRIVE included patients with either M0N0/1 (n = 139) or M1N1 (n = 257) disease.38 For the asymptomatic or mildly symptomatic M1 population, median PFS was significantly longer for enzalutamide (16.5 months) versus bicalutamide (5.5 months; HR, 0.24; 95% CI, 0.17 to 0.34). Patients with M1 disease treated with enzalutamide also had significantly greater PSA response (P < .001) irrespective of the definition of complete response (PSA decline ≥ 50% [or 90%] from baseline). Unlike STRIVE, TERRAIN randomly assigned only patients with M1 disease and radiographically confirmed metastases (n = 184 enzalutamide; n = 191 bicalutamide) but found similar results for the M1 population. Median PFS was significantly longer for enzalutamide (15.7 months) than for bicalutamide (5.8 months; HR, 0.44; 95% CI, 0.34 to 0.57; P < .001).
Although both are important phase II studies, neither STRIVE nor TERRAIN was designed to compare OS among patients with clinically defined CRPC. Thus, the question of whether earlier treatment with enzalutamide improves survival compared with the current practice of later treatment cannot be answered, but the similarity in results for PFS between the two studies is encouraging.
The remaining phase III randomized trials included a mix of asymptomatic and symptomatic patients24,25,27 or all symptomatic patients.26 No significant differences in survival outcomes were reported between treatment groups. However, Small et al25 found that patients randomly assigned to AAWD and ketoconazole (AAWD/K) experienced higher rates of PSA decline ≥ 50% (27% v 11%; P < .001) and objective response (20% v 2%; P = .02) compared with those who underwent AAWD alone.25 Of patients randomly assigned to AAWD who later had ketoconazole, the total PSA response rate was similar to those who received immediate AAWD/K, whereas the objective response rate was lower in those who received sequential therapy compared with immediate AAWD/K. The 11% PSA response results with AAWD alone varied from prior phase I and II studies that reported it as high as 40%. This lower rate may reflect shorter patient exposure to antiandrogens than in earlier reports.47 In contrast, the 20% PSA response rate detected in the ketoconazole intervention arm is in line with a study by Trump et al48 of 38 patients with CRPC and radiographic metastases treated with high-dose ketoconazole (400 mg three times a day) plus hydrocortisone wherein an objective response was observed in 17% of evaluable patients.
One additional phase III trial was identified in the systematic review of orteronel plus prednisone versus placebo among chemotherapy-naïve men with metastatic CRPC. The study does not inform our recommendations because of a lack of improvement in OS and a high adverse event rate (46%). Orteronel is no longer under development for treatment of metastatic CRPC.49
In the trials of prednisone versus flutamide,26 high- versus low-dose megestrol acetate,27 and diethylstilbestrol versus bicalutamide (single-facility phase II trial),29 no meaningful objective differences in outcomes were detected between treatment groups. Three members of the Consensus Panel reported the use of high-dose bicalutamide in this setting, but data suggest possible excess mortality associated with this dose in a related context.50
No evidence provides guidance about the optimal order of second-line hormonal therapies for patients with M1 CRPC. In the trial by Ryan et al,21 significant PFS and OS advantages and delay in clinical decline were detected in favor of abiraterone/prednisone compared with prednisone alone. A PSA response was seen in 62% of patients in the abiraterone treatment arm. The Beer et al23 trial of enzalutamide versus placebo, which reported early significant rPFS and OS advantages, found a PSA decline of > 50% in 78% of men in the enzalutamide arm. A similar PSA response also has been reported in the phase III randomized trial that compared dexamethasone and aspirin with either immediate or delayed diethylstilbestrol.24
AAWD/K produced greater PSA and objective responses than AAWD alone but no differences in OS, and 21% of patients experienced a grade 3 and 4 adverse event.25 Because ketoconazole usually is given with low-dose corticosteroids, this may influence PSA response. In the control arm of Ryan et al,21 PSA response was seen in 24% of patients who received prednisone alone. In the Nakabayashi et al51 retrospective review of 138 patients started on low-dose ketoconazole (200 mg three times a day), 28% had a PSA response. Fifty-five patients (40%) subsequently received high-dose ketoconazole (400 mg three times a day); 13% had an additional PSA response (P value not reported). In general, high-dose ketoconazole was associated with a greater risk of adverse effects, and six patients (11%) discontinued therapy as a result of worsening or new adverse effects from high-dose therapy. For patients who could not tolerate high-dose ketoconazole therapy, low-dose ketoconazole had similar efficacy.51
ASCO issued a systemic therapy guideline in 20149 that supports the use of immunotherapy (sipuleucel-T)11 or chemotherapy (docetaxel and prednisone) in men with metastatic CRPC. The use of radium-223 was recommended for men with bone metastases.52 Consult that guideline for the full recommendations.
Clinical interpretation. For chemotherapy-naïve patients who develop CRPC and have radiographic evidence of metastases, two second-line hormonal therapy options are supported by strong clinical trial evidence and are well tolerated. Abiraterone acetate plus prednisone extends rPFS and OS in addition to a variety of secondary end points, such as median time to opiate use, chemotherapy initiation, performance status deterioration, and PSA progression.
According to the manufacturer’s warnings and precautions,53 abiraterone acetate should be used with caution in patients with a history of cardiovascular disease. Drug safety was not established in patients with a left-ventricular ejection fraction < 50% or with New York Heart Association class II to IV disease. Abiraterone acetate can cause hypertension, hypokalemia, and fluid retention. Low risks of adrenocortical insufficiency or hepatotoxicity also are associated with abiraterone acetate use. A low risk of seizure associated with enzalutamide use exists54; however, among chemotherapy-naïve patients, the risk (0.1%) was similar between those who received enzalutamide and those who received placebo. Posterior reversible encephalopathy syndrome also has been associated with enzalutamide use, which required discontinuation of the drug.
According to the 2014 ASCO systemic therapy guideline for men with metastatic CRPC,9 other treatment options include immunotherapy (sipuleucel-T) or chemotherapy (docetaxel and prednisone). The systemic therapy guideline specifically recommends radium-223 for men with bone metastases.52 If none of the aforementioned hormonal therapy, immunotherapy, or chemotherapy options can be tolerated and/or accessed, other antiandrogens, prednisone, and ketoconazole/hydrocortisone may be offered. Enrollment in a clinical trial is always an option. The goal of treatment is symptom relief with extension and quality of life and deferral of chemotherapy for as long as possible. Palliative care should not be overlooked, particularly for patients who exhibit symptoms or decreased quality of life.20
Research Question 4
How often should patients with CRPC undergo PSA monitoring?
PCO 4. No evidence provides guidance about the optimal frequency of PSA monitoring before starting second-line hormonal therapy or after treatment has begun.
Literature review and analysis. Because no data inform this question, the Expert Panel relied on clinical experience, training, and judgment to formulate this PCO. Consideration was given to the inconvenience and anxiety introduced by more-frequent PSA testing versus potential harms that result from delayed recognition of a rapid PSA doubling time.
Clinical interpretation. Various PSA metrics are available and under evaluation for use in monitoring disease progression among patients with CRPC. Although some studies suggest that PSA doubling time is prognostic for OS, specifically among chemotherapy-naïve patients with metastatic CRPC, none of the available metrics or emerging biomarkers (eg, circulating tumor cells, androgen receptor splice variants, cancer stem cells) are as yet approved to serve as a surrogate metric for OS in clinical trials.55,56
Research Question 5
What imaging modalities are appropriate for patients with CRPC?
Literature review and analysis. No trials that compared the utility of various imaging modalities for monitoring CRPC were identified by the systematic review. For patients with prostate cancer in general, early results from the National Oncologic PET registry are encouraging for 18F-PET but are not yet definitive for the CRPC population because of a lack of data on past versus current ADT use, differentiation between initial and second-line hormonal therapy use, and PSA change. In the interim, clinical experience, training, and judgment were considered in formulating this PCO because of the lack of clinical trial data to inform the PCO. 18F-PET appears to have greater sensitivity over bone scans, but evidence among patients with CRPC is evolving, and the impact on clinical outcomes remains undetermined. Whole-body magnetic resonance imaging to detect oligometastatic disease and radiotracers and imaging agents such as c-11 choline, prostate-specific membrane antigen, and 18F-flucicovine currently are considered investigational.
Clinical interpretation. In the absence of clinical trial–based evidence, the panel considers bone scan and either computed tomography or magnetic resonance imaging of the abdomen and pelvis appropriate.
Research Question 6
How often should patients with CRPC undergo radiographic imaging or routine radiographic restaging?
Literature review and analysis. No evidence provides guidance about how often patients with CRPC should undergo radiographic imaging. With no data to inform the question, the Expert Panel relied on clinical experience, training, and judgment to formulate the PCO. In addition to the potential inconvenience to the patient, discomfort, bother, and a small risk of complications can be associated with contrast agents or radiotracers administered during these tests. Finally, costs and overuse in busy health care sectors were considered.57
Clinical interpretation. The appropriate frequency of radiographic imaging is variable and largely depends on symptoms. In the absence of symptoms or some other clinical reason, radiographic imaging is not recommended.
Appendix Figure A1 (online only) shows a patient treatment algorithm if second-line hormonal therapy is considered. The Data Supplement provides recommendations from other guidelines for treatment considerations beyond second-line hormonal therapy.
Few studies examined cost-effectiveness or the budgetary impact associated with second-line hormonal therapies for CRPC. The focus primarily is on patients with metastatic disease in the postchemotherapy setting. These cost studies are not directly comparable with one another because of differences in methodology and assumptions,58-61 and the generalizability of the results to the chemotherapy-naïve population is questionable given differences in survival and total treatment costs. The two studies that provided estimates of 30-day treatment costs on the basis of average wholesale prices or reimbursement for the postchemotherapy population suggested that abiraterone acetate is less expensive than enzalutamide.9,58 These basic results may be similar for the chemotherapy-naïve population.
The only study that directly examined cost-effectiveness for the asymptomatic, chemotherapy-naïve population found that neither abiraterone nor sipuleucel-T were cost-effective compared with prednisone on the basis of a willingness-to-pay threshold of $150,000 per quality-adjusted life-year.62 On the basis of findings from a survey of oncologists,63 the authors suggested that $378,000 (2013 US dollars) may be a more accurate threshold. Although sipuleucel-T remains cost-ineffective at this new threshold, theincremental cost-effectiveness ratio for abiraterone is nearly costeffective at $389,000 per quality-adjusted life-year.
To the best of our knowledge, no studies have assessed the cost of follow-up after progression to CRPC. The few studies that assessed the costs of follow-up for patients with prostate cancer focused on the first 5 years after receipt of curative intent therapy.64 Actual charges for 5 years of follow-up after curative therapy for prostate cancer varied by 7.3-fold.
Although most studies that addressed out-of-pocket (OOP) expenses for patients with prostate cancer generally focused on those with clinically localized disease, the findings of Jung et al65 in 2012 are relevant. This study found that few patients fully understand the likely OOP costs in advance of the treatment decision. Also informative is a survey of factors that influence physician decisions to prescribe flutamide in conjunction with complete androgen blockade for patients with metastatic prostate cancer, which found that OOP costs were the most important factor.66
OOP costs, the potential adverse effects of OOP costs (referred to as financial toxicity),67 and expected quality of life should be discussed with patients during the treatment decision-making process. Oncologists must continue to advocate for patient access to beneficial therapies while being responsible stewards of health care resources.