Androgen-deprivation remains the cornerstone of treatment for patients with hormone-sensitive advanced prostate cancer [1]. The combination of hormonal suppression and the radiotherapy is able to significantly decrease the recurrences and the mortality of patients affected by locally advanced prostate cancer. Substantial toxicities were also identified when hormone therapy was used.[2–11]. A recently conducted meta-analysis has demonstrated an overall absolute benefit of 7.5-10% in favor of the addition of the hormone therapy to in terms of biochemical failure and clinical progression free survival [12].

The optimal duration of androgen ablation therapy is still controversial despite a growing number of prognostic factors that have been shown to predict disease outcome [13]. In all prospective randomized trials, long term hormonal therapy (more than 2 years) improves survival in patients with locally high risk (Gleason score ≥ 8, median PSA ≥ 20, stage ≥ T3) advanced prostate cancer. This strategy, however, is also associated with an increased risk of morbidities, such as osteoporosis, diabetes, depression, dislipidemia and abdominal obesity [14, 15].

The value of short-term hormonal treatment in patients with intermediate risk disease (Gleason score ≥ 7, median PSA ≥ 10, stage ≥ T2c) has also been detected [16, 17], although its role in high-risk disease is still controversial. Nevertheless, short courses of hormonal suppression have been shown to be very effective in downstaging localized disease. Surgical series have provided some histopathological confirmation, with significant decreases in margin-positive rates in prostatectomy after neoadjuvant hormonal therapy [18, 19].

In order to determine if difference exists between a shorter and a longer hormone therapy (HT) in combination with radiotherapy, a literature based meta-analysis was conducted.

The analysis was conducted following 4 steps: definition of the outcomes (definition of the question the analysis was designed to answer), definition of the trial selection criteria, definition of the search strategy, and a detailed description of the statistical methods used [20, 21].

Although the number of studies is small to produce reliable estimates, the data presented herein strengthen the role of the hormone-therapy duration in patients candidates to receive androgen suppression in association with radiotherapy for locally advanced prostate cancer (Table 2 Figures 2, 3).

Hormonal suppression in patients affected by locally advanced prostate cancer candidates to receive exclusive radiotherapy is able to improve outcome, as a series of trials recently demonstrated; according to the results of our previous meta-analysis, hormonal suppression plus radiotherapy significantly decreases recurrence and mortality of patients with localized prostate cancer. The magnitude of this survival benefit ranges from 4.9% to 5.5% for OS and CSS, respectively, which translates into 20 and 18 NNT. Local and distant relapse were significantly decreased by hormonal treatment, by 36% and 28%, without significantly affect toxicity [38].

Although the use of adjuvant hormonal therapy has been shown in randomized trials to improve outcomes, two issues still need to be clarified: 1) the optimal hormone therapy duration and 2) the impact of the radiation dose on outcome and local control. Within this context, the current meta-analysis was aimed to assess if difference exist between a shorter and longer androgen suppression, according to the trialist' definition. The pooled results of the considered trials show that a longer treatment approach improves all outcomes when compared to a shorter one (Table 2 Figure 2, 3). In particular, BF is significantly reduced when hormone-therapy is administered in a longer fashion, with an overall absolute difference of 10.1%, which corresponds to only 9-10 patients needed to be treated for one to benefit (Table 2). Although the meta-regression analysis could be not valid given the small number of gathered trials, the benefits present across all outcomes are provided in favour of a longer strategy, regardless of any of the selected negative predictors, suggesting an overall effect independent by any clinico-pathological feature. Conversely to the these predictors, the lenght of treatment seems to significantly affect all outcomes, suggesting the even more independent value of a longer approach (Table 3). Although this interpretation should be softened, given the limitations related to a meta-regression analysis, a further individual patient data meta-analysis could eventually clarify if a particular subset of patients would better benefit of such approach. Moreover, these conclusion should be driven only when an homogenous risk classes classification was likely to be prospectively adopted in all analyzed trials. Indeed, the significant heterogeneity in the selected primary outcome can easily reflect that the magnitude of the reported benefit is different across trials, and this can likely be a clear effect of the different patient' selection.

The adopted meta-analytic method did not allow to extract patients' subgroups data from those trials where either low, intermediate and high risk patients were accrued together. For example, no advantage for longer duration was found in those trials enrolling low risk patients [7, 39] (Figure 2, 3). Conversely, the positive effect of long term hormonal duration can be partially driven by the inclusion of those 2 trials including patients with node-positive disease [34, 36]. So, we are not able to actually exclude a 'carry-over' effect due to the inclusion of such trials with these characteristics. The mixture of such confounders (such as different cut-offs in the Gleason score) in trial selection does not allow to derive clear and definitive conclusions.

Nevertheless, even taking into account all the drawbacks in correlation and meta-regression estimations of cumulating few trials, the absolute difference in favour of the longer HT (and the lower NNT) strongly justifies a prolonged treatment strategy. Moreover, the meta-regression analysis did also show a significant correlation between the length of the longer strategy and the effect in the reduction of BF, LR and DM and the improvement of CSS (Table 3 Figure 2). The latest conclusion allows to carefully speculate upon a 'treat-until-progression' strategy, at least in the context of a randomized trial.

Recent reports suggest that the risk of cardio metabolic problems with long-term castration deprivation therapy could counteract the benefits of hormone therapy, although this has also been questioned. Long term androgen suppression can reduce the quality of life and increase the risk of fatal myocardial infarction, fractures, and the development of a metabolic syndrome. In the context of neoadjuvant approaches, it has been recently reported that hormone-therapy is associated with a significant increase of all-cause mortality in those patients with a cardiac comorbidity, but not in those who did not show any history of cardiovascular disease [40]. This represent a crucial step-forward in the selection of those patient who could be considered as appropriate candidates for a longer hormonal suppression.

The radiotherapy component of the combined treatment is crucial: cancer-specific and overall mortality rates at 10 years are significantly lower with the combined treatment in respect to androgen suppression alone. Moreover, the survival benefit seem to depend on the duration of homone-therapy, and was also reported by the Early Prostate Cancer Programme using adjuvant antiandrogen treatment as well. In studies with short or intermediate androgen deprivation of, survival prolongation has only been reported in subgroups.

At the start of these trials, the standard radiation dose to the prostate was 70 Gy or less. With the advent of intensity-modulated and image-guided radiotherapy techniques, radiation doses of 78 Gy or higher are now possible, and randomised studies have shown that biochemical relapse-free survival improves with high radiation doses. As a consequence, the overall survival benefit might further increase further with higher safely radiation doses [41–44]. Clearly, open issues remain whether the duration of the hormonal suppression could eventually affect the outcome when radiation therapy is administered with higher dosages, and what should be irradiated (prostate or the whole pelvis).

The general shared consensus is that high risk patients would better benefit from two to three years of adjuvant androgen deprivation, while those with intermediate' risk prostate cancer from six months of adjuvant androgen deprivation plus radiotherapy delivered with a dose lower or equal to 70Gy (for doses higher than 70 Gy the appropriate duration of adjuvant androgen deprivation is still unknown) [16, 17]. According to the trial recently published by Bolla et al, the effect of short-term and long-term androgen suppression upon five-year mortality is likely to be maintained at ten years [39, 45], whereas the short-term benefit may actually be no longer effective. The intriguing factor that can help in the daily decision between one strategy over the other is the relative weight of adverse events and the adverse effects on quality of life between the two groups of patients; indeed, although no significant difference in the overall quality of life did emerge between the two groups in the Bolla et al trial (p = 0.37), a clinically meaningful difference was present for hot flushes, sexual interest, and sexual activity [34]. Unfortunately, our analysis could investigate neither the pooled toxicities nor the quality of life data into a cumulative fashion, given the lack of a complete reporting in the original trials.

Although the treatment duration seems to significantly impact upon outcome, it should be acknowledged that other factors may have a critical role, or may be more important: indeed, in the recent update of the Crook et al study, the more relevant weight for benefit seems related more to the early biochemical response to hormonal treatment before radiotherapy, rather than its overall duration [46]. For these reasons, a critical view of all data as shown in our analysis (taking into account all the limitations related to the adopted approach and methodology) is recommended.

Taking into account the current data available in literature, and the results reported herein, the significant differences in outcome in favour of a longer hormonal treatment duration do support the adoption of such a strategy for patients affected by locally advanced prostate cancer, although several limitations should be considered given the small number of trials included. Nevertheless, a careful patient selection upon the presence/absence of cardiovascular and/or metabolic comorbidities should be adopted, in order to avoid significan late-term toxicities. On the contrary, a wide adoption of a longer administration regardless of the risk stratification (notwithstanding the results of the meta-regression analysis) seems in some way not enough mature, in absence of more reliable data coming from an individual patient data meta-analysis.