This review found that androgen suppressant therapy with radiotherapy was associated with significant survival benefits in patients with prostate cancer who had high risks of relapse or development of metastatic disease. Significant variation across the trials for the principal results and uncertainty about the impact of bias make the reliability of the authors' conclusions unclear.

To compare the combination of distinct hormone therapy modalities with radiotherapy alone in patients with localised or locally advanced prostate cancer.

PubMed, EMBASE, LILACS, The Cochrane Library and ClinicalTrials.gov were searched for relevant studies in English, Spanish or Portuguese to October 2011; search terms were reported.

Randomised controlled trials in patients with localised (cT1-2) or locally advanced (T3-4 N0-2 MO) prostate carcinoma that compared radiotherapy alone with radiotherapy plus androgen suppression administered prior to, during or after radiotherapy were eligible for inclusion. Studies were excluded where anti-androgenic therapy was administered in both treatment and control groups. Androgenic suppression could include orchiectomy, luteinising hormone-releasing hormone analogues, peripheral antiandrogens or oestrogenic therapy. The outcomes were overall survival, disease-free survival and toxicity.

The included studies were published between 1988 and 2011. Androgenic suppressive treatment comprised goserelin (3.6mg/month continuously or for three to six months) leuprolide (7.5mg/month for four to six months), diethylstilbestrol (25mg oral daily, administered continuously) bicalutamide (150mg oral daily, at the discretion of the investigator) and orchiectomy. Flutamide was administered with goserelin or leuprolide in some studies. Treatment durations varied widely from three months to permanent or continuous treatment. Median follow-up ranged from five years to 14.5 years. Where stated, radiotherapy doses ranged from 60 to 70 Gray units.

Two reviewers performed the study selection independently.

Methodological quality was evaluated by two reviewers using criteria described by Sterne et al. for randomisation, allocation concealment, blinding, description of drop-outs, use of intention-to-treat analyses and declared sources of funding.

Data were extracted to calculate hazard ratios (HR) for time-to-event outcomes and odds ratios (OR) for dichotomous outcomes and 95% confidence intervals (CI) for the outcomes. Where data were incomplete or missing, the reviewers calculated estimates using described numbers of events and corresponding p-values for log-rank statistics or by transcription of survival curves.

Pooled hazard ratios, odds ratios and 95% CIs for the summary estimates were calculated using a random-effects model. Statistical heterogeneity was evaluated using Χ² and Ι². Potential sources of heterogeneity were investigated when heterogeneity was statistically significant. The risk of publication bias was evaluated through visual appraisal of funnel plots. Subgroup analyses examined the influence of androgen suppression modality hormonal blockade and duration of therapy on the results.

Ten RCTs (6,555 patients, range 82 to 1,979) were included in the review. Randomisation and allocation concealment were described in six trials. Blinding was used in one trial. Losses to follow-up were described in nine trials. Sample sizes were stated in eight trials. Intention-to-treat analyses were used in six trials. Where stated, trials were either industry-sponsored, publicly funded or used both industry and public funding. The authors reported substantial heterogeneity across the results of the trials for overall survival and presented results from subgroup analyses.

There were statistically significant benefits of androgenic suppressive treatment for overall survival in the subgroups: radiotherapy with goserelin treatment (HR 0.72, 95% CI 0.60 to 0.87; five RCTs; Ι²=50%), central hormone blockage (two studies of goserelin and one study of orchiectomy) (HR 0.61, 95% CI 0.47 to 0.81; three RCTs; Ι²=15%), complete hormonal blockade (HR 0.79, 95% CI 0.69 to 0.90; five RCTs; Ι²=34%), studies of treatment duration six months or less (HR 0.79, 95% CI 0.69 to 0.90; five RCTs; Ι²=34%) and studies with treatment duration of more than one year (HR 0.61, 95% CI 0.47 to 0.81; three RCTs; Ι²=15%). No significant differences in overall survival were observed with orchiectomy and radiotherapy compared to radiotherapy (one RCT) or bicalutamide or estrogenic therapy without luteinising hormone-releasing hormone analogues.

Androgenic suppressive treatment with radiotherapy conferred statistically significant benefits in disease-free survival for each subgroup of treatment modality: goserelin (HR 0.53, 95% CI 0.43 to 0.65; five RCTs; Ι²=74%), goserelin and leuprolide with flutamide (HR 0.48, 95% CI 0.30 to 0.77; two RCTs; Ι²=68%), leuprolide and flutamide (HR 0.55, 95% CI 0.34 to 0.90; one RCT) and bicalutamide (HR 0.61, 95% CI 0.52 to 0.72; one study). Significant benefits in disease-free survival were seen for treatment durations of less than six months (HR 0.57, 95% CI 0.50 to 0.65; six RCTs; Ι²=31%) and of more than one year (HR 0.43, 95% CI 0.38 to 0.50; three RCTs; Ι²=0%).

There was insufficient data to undertake a meta-analysis of adverse events; the authors stated that analyses of individual trials showed no evidence of increased toxicity with androgenic suppressive treatment with or without radiotherapy.

Visual appraisals of funnel plots showed no evidence of publication bias for the primary outcome of overall survival.

The results of the review supported use of androgen suppressive treatment with goserelin monotherapy for patients with prostate cancer who were at high risk of recurrence or metastatic cancer.

The review addressed a clear question. Inclusion criteria were defined and reproducible. Appropriate databases were searched for relevant studies. The restriction to studies published in particular languages meant there was some risk of language bias. Steps were taken to minimise errors and biases for study selection of studies and assessment of methodological quality but were not stated for data extraction.

The results were presented in subgroups; this was appropriate because of clinical and statistical heterogeneity between the trials. Methodological quality (bias) was assessed and summarised but the authors did not use the results of this assessment to perform sensitivity analyses or contextualize the results. The principal results were subject to substantial heterogeneity.

The heterogeneity of the principal results and uncertainty about the impact of bias make the reliability of the authors conclusions uncertain.

Practice: The authors stated that use of goserelin for more than one year was associated with the most beneficial survival outcomes and should be considered to be the standard treatment for patients at higher risk of relapse or developing metastatic cancer.

Research: The authors stated that further research should determine the role of androgen suppressive therapy and the risk of metabolic syndrome and osteoporosis that has been documented in observational studies.

Part-funded by AstraZeneca.