Intermittent versus continuous androgen deprivation for locally advanced, recurrent or metastatic prostate cancer: a systematic review and meta-analysis

Background Prostate cancer is the most common cancer in older men in the United States (USA) and Western Europe. Androgen deprivation (AD) constitutes, in most cases, the first-line of treatment for these cases. The negative impact of CAD in quality of life, secondary to the adverse events of sustained hormone deprivation, plus the costs of this therapy, motivated the intermittent treatment approach. The objective of this study is to to perform a systematic review and meta-analysis of all randomized controlled trials that compared the efficacy and adverse events profile of intermittent versus continuous androgen deprivation for locally advanced, recurrent or metastatic hormone-sensitive prostate cancer. Methods Several databases were searched, including MEDLINE, EMBASE, LILACS, and CENTRAL. The endpoints were overall survival (OS), cancer-specific survival (CSS), time to progression (TTP) and adverse events. We performed a meta-analysis (MA) of the published data. The results were expressed as Hazard Ratio (HR) or Risk Ratio (RR), with their corresponding 95% Confidence Intervals (CI 95%). Results The final analysis included 13 trials comprising 6,419 patients with hormone-sensitive prostate cancer. TTP was similar in patients who received intermittent androgen deprivation (IAD) or continuous androgen deprivation (CAD) (fixed effect: HR = 1.04; CI 95% = 0.96 to 1.14; p = 0.3). OS and CSS were also similar in patients treated with IAD or CAD (OS: fixed effect: HR = 1.02; CI 95% = 0.95 to 1.09; p = 0.56 and CSS: fixed effect: HR = 1.06; CI 95% = 0.96 to 1.18; p = 0.26). Conclusion Overall survival was similar between IAD and CAD in patients with locally advanced, recurrent or metastatic hormone-sensitive prostate cancer. Data on CSS are weak and the benefits of IAD on this outcome remain uncertain. Impact in QoL was similar for both groups, however, sexual activity scores were higher and the incidence of hot flushes was lower in patients treated with IAD.


Background
Prostate cancer is the most common cancer in older men in the United Kingdom (UK), United States (USA) and Western Europe [1]. Despite its high incidence, the disease is often responsive to treatment even when metastatic and may be cured when localized. In patients with locally advanced tumors, recurrent or metastatic, the goals of therapy are to prolong survival, slow the progression of disease and preserve the quality of life [2].
Androgen deprivation (AD) constitutes, in most cases, the first-line of treatment for these cases. However, the deleterious effects of continuous androgen deprivation (CAD) are widely known and are related to some degree of deterioration in the quality of life. The most frequent symptoms resulting from AD include sexual dysfunction, fatigue, anemia, reduced muscle and bone mass, depression, abnormal lipid metabolism, cognitive dysfunction and development or worsening of metabolic syndrome [3,4]. The negative impact of CAD in quality of life, secondary to the adverse events of sustained hormone deprivation, plus the costs of this therapy, motivated the intermittent treatment approach.
In recent years, non-randomized studies (most of them with heterogeneous criteria for selection and clinical assessment) were performed to confirm the effectiveness of intermittent androgen deprivation (IAD) in patients with prostate cancer. Two systematic reviews [5,6] previously published analyzed the results of these nonrandomized studies and the authors suggested that the best candidates for intermittent androgen deprivation are patients with biochemical progression after prostatectomy or radiation, with no evidence of metastases and with mildly aggressive tumors. On the other hand, patients with large tumor volumes, positive lymph nodes and bone metastases, PSA >100 ng/ml or short PSA doubling time, would be best treated with continuous deprivation.
Irrespective of official guideline recommendations, IAD is a treatment option used worldwide by both urologists and oncologists even outside of clinical trials [5].
The 2008 UK National Institute for Health and Clinical Excellence (NICE) recommends that IAD may be offered to men with metastatic prostate cancer providing they are informed that there is no long-term evidence of its effectiveness [7].
The results of some randomized controlled trials (RCT) were statistically inconclusive [8] or controversial. Crook et al. [9,10] analyzed IAD versus CAD for prostate-specific antigen (PSA) elevation after radiotherapy, assessing overall survival (OS) in a non-inferiority randomized trial with 1,386. The results demonstrated that IAD was non-inferior to CAD in regard to OS (8.8 years versus 9.1 years respectively -hazard ratio for death, 1.02; CI 95% 0.86 to 1.21, p for non-inferiority = 0.009), besides providing potential benefits in aspects such as physical function, fatigue, urinary problems, hot flashes, libido, and erectile function. Furthermore, the authors pointed that time to hormone resistance was statistically significantly improved on the IAD arm (HR 0.80, 95% CI 0.67-0.98; p = 0.024).
Regarding time-to-progression or castration-resistant disease, however, Crook et al. [9,10] showed better results in favor of CAD, while the results of Salonen et al. [11,12] were favorable to IAD. Such contradictory results make it difficult to emit a strong scientific-based recommendation for or against IAD.
The objective of this study is to analyze all published randomized controlled trials (RCTs) that compared the efficacy and adverse events profile of IAD versus CAD for locally advanced, recurrent or metastatic hormonesensitive prostate cancer.

Types of studies
We included RCTs with parallel design that compared the use of intermittent versus continuous androgen deprivation.

Types of participants
The selected studies included patients with locally advanced, recurrent or metastatic hormone-sensitive prostate cancer. For MEDLINE, we used the search strategy methodology for randomized controlled trials [13] recommended by the Cochrane Collaboration [14]. For EMBASE, adaptations of this same strategy were used [13], and for LILACS, we used the search strategy methodology reported by Castro et al. [15]. An additional search on the SCI database was also performed to retrieve articles that were cited on the included studies. The specific terms relevant to this review were added to the overall search strategy methodology for each database.

Search strategy for identification of studies
The overall search strategy was: #1: prostate cancer, #2: intermittent, #3: androgen deprivation and #4: Randomized Controlled Trial. Searches in electronic databases combined the terms: #1 AND #2 AND #3 AND #4.
The search was not limited by date, language or specific outcome.

Critical evaluation of the selected studies
All the references retrieved by the search strategies had their title and abstract evaluated by two of the researchers. Every reference with the least indication of fulfilling the inclusion criteria was listed as pre-selected. The complete article of all pre-selected references was retrieved. The articles were analyzed by two different researchers and included or excluded according to the previously reported criteria. The excluded trials and the reason of their exclusion are listed in this article. Data was extracted from all the included trials.
Details regarding the main methodology characteristics empirically linked to bias [16] were extracted with the methodological validity of each selected trial assessed by two reviewers (T.E.A.B and O.C). Particular attention was given to some items such as: the generation and concealment of the sequence of randomization; blinding; application of intention-to-treat analysis; sample size pre-definition; loss of follow-up description; adverse events reports; if the trial was performed in multiple center or a single center; and the sponsorship.

Data extraction
Two independent reviewers extracted the data. The name of the first author and year of publication were used to identify the study. All data were extracted directly from the text or calculated from the available information when necessary. The data of all trials were based on the intention-to-treat principle, so they compared all patients allocated in one treatment with all those allocated in the other.
The primary endpoint was overall survival (OS). The OS was calculated from the date of randomization to the date of death, with data censored at the last known date that the patient was alive.
Other clinical outcomes were also evaluated: Cancer-specific survival (CSS): the cancer-specific survival was calculated from the date of randomization to the date of death from prostate cancer or a complication of cancer treatment; Time to progression (TTP) or castration-resistant disease: defined as increases in PSA level or evidence of new clinical disease while the patient was receiving androgen-deprivation therapy and testosterone was at castration level; Differences in the quality of life (QoL): hot flashes, desire for sexual activity, urinary symptoms, depression and gynecomastia; Died because of cardiovascular events.

Analysis and presentation of results
Data analysis was performed using the Review Manager 5.1.2 statistical package (Cochrane Collaboration Software) [17].
To estimate the absolute gain in OS, cancer-specific survival and time to progression, the meta-analytic survival curves were calculated as suggested by Parmar et al. [18]. A pooled estimate of the HR was computed by a fixed-effect model according to the inverse-variance method [21]. Thus, for effectiveness or adverse events, an HR or RR >1 favors the standard arm (continuous treatment) whereas an HR or RR <1 favors the intermittent treatment.
If statistic heterogeneity was found in the metaanalysis, an additional analysis was performed, using the random-effects model described by DerSimonian and Laird [22] that provides a more conservative analysis.
To assess the possibility of publication bias, the funnel plot test described by Egger et al. was performed [23]. When the pooled results were significant, the number of patients needed to treat (NNT or NNH) to cause or to prevent one event was calculated by pooling absolute risk differences in the trials included in this metaanalysis [24][25][26]. For all analyses, a forest plot was generated to display the results.

Results
The diagram represents the flow of identification and inclusion of trials, as recommended by the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) statement [27] (Figure 1).
Overall, 73 references were identified and screened. Twenty-nine studies were selected and retrieved for full-text analysis. Of these studies, 16 were excluded for various reasons described on Table 1.

Characteristics of included studies
Thirteen randomized trials comprising 6,419 patients with hormone-sensitive prostate cancer were included in this analysis. More than 2,800 patients (44%) had metastatic disease and 1,595 patients (25%) had PSA progression after prostatectomy (n = 209) or radiotherapy (n = 1,386). The number of patients in these trials varied from 43 to 1,535, but only 5 involved >500 patients. The average age of patients was 70 years.
Full details of trial design are not available for all studies: several reports are available only in abstract form [48,49,56].
The criteria for stopping and re-start of therapy in groups treated with IAD were similar, however not uniform (Table 2). Average follow-up time in these studies was of >2 years. The induction period varied, but most patients used IAD for a period of approximately 3-6 months (Table 3).
More details on the treatment modality, follow-up, number of patients and primary endpoint in the 13 trials included in this analysis are summarized in Tables 2  and 3.
The efficacy analysis was summarized in Table 4.
One study [58], on re-introduction of treatment in IAD group, used a hormone therapy scheme (cyproterone 300 mg/d) that was different from the one used in the CAD group (LHRHa: triptoreline 11.25 mg plus cyproterone 200 mg/d). When OS analysis was performed without this study, the results remained similar between IAD and CAD (fixed effect: HR = 1.04; CI 95% = 0.97 to 1.12; p = 0.25; with no heterogeneity: Chi 2 = 7.08, df = 6, P = 0.31, I 2 = 15%).
Mortality secondary to cardiovascular events was report in only 4 studies [9][10][11][12]44,45,58]. Results of this analysis   were favorable to the group treated with IAD (fixed effect: RR = 0.80; CI 95% = 0.67 to 0.94; p = 0.007; NNH = 33) however with high heterogeneity levels (Chi 2 = 6.67, df = 3 [P = 0.08]; I 2 = 55%). According to funnel plot analysis [23] the possibility of publication bias was low for all of the outcomes. When funnel plot shows asymmetry, there is a possibility of bias and despite its limmitations, this method is widely used by authors.

Discussion
Currently, many oncology guidelines do not recommend the use of IAD for patients with metastatic hormonesensitive prostate cancer [36,59]. But it is important to note that these guidelines date back to 2007. Guidelines from the European Association of Urology (EUA) state that IAD is now widely offered to patients with prostate cancer and should no longer be regarded as investigational [60,61].
In this present meta-analysis, we included a larger number of studies. When HRs were not directly reported in the original study, they were estimated indirectly by using the reported number of events and the corresponding P value for the log-rank statistics, or by reading survival curves as suggested by Parmar et al. [18]. To reduce reading errors, original survival curves were digitalized and enlarged, and data extraction was based on reading electronic coordinates for each point of interest, as described by DeLaurentiis in another meta-analysis [64].
In the final OS and CSS analyses we included 8 and 7 studies respectively. The induction time of androgen deprivation was, in general, 3-6 months. Two studies were available only in abstract form [48,49,56]. Overall, goserelin was the most used LHRHa, followed by leuprorelin. Only two studies [48,49,53] did not used luteinizing hormonereleasing hormone agonist (LHRHa) or orchiectomy. These studies used cyproterone acetate (CPA).
In this meta-analysis, randomized studies included heterogeneous groups of patients (locally advanced, metastatic or recurrent). It was not possible to perform the proper  analysis of subgroups of patients (i.e. by Gleason score, initial PSA levels, minimal or extensive metastatic disease) because results for such sets were not published in the studies. As to overall survival, our results reinforce the equivalent efficacy of CAD and IAD, regardless of previous treatments.
Apropos of CSS, despite the overall analysis demonstrating only a trend towards treatment with CAD, the results must be interpreted with caution. As mentioned before, in one of the studies [11,12] the baseline PSA levels were lower on the IAD arm, suggesting that these patients had less extensive disease. Once this trial was excluded from  Table 5 Significant differences in quality of life (QoL) Study the analysis, results favored the group treated with CAD; therefore we cannot exclude the possibility that IAD may carry a higher risk of death by cancer.
Recently, a large prospective Danish study, including over 30,000 patients, investigated the relationship between androgen deprivation therapy (ADT) and cardiovascular diseases such as myocardial infarction (MI) and stroke in men with prostate cancer [65]. The authors found that patients treated with medical endocrine therapy had an increased risk for MI and stroke with adjusted HRs of 1.31 (95% confidence interval [CI], 1.16-1.49) and 1.19 (95% CI, 1.06-1.35), respectively, compared with nonusers of ADT. However, androgen deprivation secondary to orchiectomy did no increase the risks for MI (HR: 0.90; 95% CI, 0.83-1.29) or stroke (HR: 1.11; 95% CI, 0.90-1.36). Nevertheless, the conclusions might have been affected, by the lack of information on prognostic lifestyles.
In this meta-analysis, mortality secondary to cardiovascular diseases, despite the heterogeneity found and the paucity of studies reporting this outcome, was favorable to the group treated with IAD. This aspect might have influenced the similar results achieved in OS.
In the event of biochemical recurrence after prostatectomy, when androgen deprivation is indicated, the best candidates for IAD are patients who reach PSA values <0.5 ng/mL after the induction time, according to a randomized study that was considered pure (i.e which included only patients with recurrence after prostatectomy) [1,46,47]. For biochemical recurrences after radiotherapy, the best candidates for IAD are patients who reach PSA levels <4 ng/mL after the induction time [9,10]. Only in these situations (recurrence after radiotherapy or prostatectomy), TTP seems to be favorable to patients receiving CAD. In locally advanced and/or metastatic tumors, the best candidates for IAD are patients who reach PSA values <4 ng/mL after induction time.
Although TTP analysis was performed in this metaanalysis, the results should be evaluated with caution since criteria for progression were different among the included studies.
Patients with a lower burden of disease (i.e. less extensive diseases) and without co-morbidities may be the ones achieving most benefits with IAD treatment.
The implementation of guidelines to assess, monitor and reduce known risk factors for cardiovascular disease may influence the results of overall survival in patients treated with CAD.
Regarding quality of life, differences were not clinically significant in most studies. But it was observed that sexual activity scores seemed to be higher in patients treated with IAD, and the incidence of hot flushes was lower.
Furthermore, it is important to note that IAD modality offers an economic benefit with reduction of pharmaceutical costs during "off-therapy" [62]. Economic analysis is not the aim of this investigation. The studies included failed to report the costs of both modalities (intermittent and continuous). Niraula et al. [62], analyzing only drug costs, estimated that the median cost savings with IAD would be 48%. Hering et al. [54], also based only on the average drug costs, demonstrated that the cost of treatment in IAD arm was approximately 50% lower that in the CAD, over 48 months.

Conclusions
Overall survival was similar between IAD and CAD in patients with locally advanced, recurrent or metastatic hormone-sensitive prostate cancer. Data on CSS are weak and the benefits of IAD on this outcome remain uncertain. Impact in QoL was similar for both groups, however, sexual activity scores were higher and the incidence of hot flushes was lower in patients treated with IAD.