EORTC 30846.

Methods	Design: randomised controlled trial Setting: multicentric (27 institutions) Recruiting period: February 1986 to November 1998 Sample size: 302 recruited, 234 randomised patients Follow‐up (months): median 13 years
Participants	Population description: patients with lymph‐node‐positive (pN1‐3) cancer without local treatment of the primary tumour Inclusion criteria: men with locally confined or locally advanced PCa (category T2‐T3) and histologically or cytologically confirmed lymph node metastases N1‐3 (TNM 1972) but not N4 disease (computed tomography (CT) scan was mandatory) no previous treatment other than lymph node dissection or lymph node biopsy no evidence of further metastatic disease (assessed by bone scan and CT scan) World Health Organization (WHO) performance status (PS) of 0 to 2 Exclusion criteria: not reported Tumour stage: T2‐3, N1‐3, M0 Previous treatment: no previous treatment other than lymph node dissection or lymph node biopsy Number randomised: 234 patients (Early ADT: 119; included in analysis 119. Deferred ADT: 115; included in analysis 115) Withdrawals and exclusions: no exclusions Subgroup measured: not reported Subgroup reported: not reported Age: median immediate endocrine treatment: 66.6 years (range: 52.2 to 76.8 years) median delayed endocrine treatment: 64.3 years (range: 46 to 79.2 years) Baseline imbalances: The 2 groups were well balanced except for small differences for some factors: the median age was 66.6 years for the EET arm and 64.3 years for the DET arm. In the EET arm, 29.4% of the tumours were poorly differentiated (WHO grade 3) versus 33.9% in the DET arm, but T3‐4 (TNM 1972) were seen in 68.1% of the patients in the EET arm versus 62.6% in the DET arm.
Interventions	Early ADT (intervention group): route of administration: s.c. and p.o. or surgical intervention frequency, dose: 3.6 mg of Zoladex (AstraZeneca, London, UK) given s.c. every 4 wk and cryptoterone acetate (CPA) 50 mg given orally 3 times per day for the first 4 weeks of treatment, or orchiectomy number of patients randomised: 119 patients Deferred ADT (control group): route of administration: s.c. and p.o. or surgical intervention frequency, dose: 3.6 mg of Zoladex (AstraZeneca, London, UK) given s.c. every 4 weeks and cryptoterone acetate (CPA) 50 mg given orally 3 times per day for the first 4 weeks of treatment, or orchiectomy definition of deferred ADT: endocrine treatments were identical to the early ADT arm, same treatment indicated upon clinical progression or upon subjective progression, based on a rise of serum‐prostate‐specific antigen (PSA) or an increase in the T category or prostatic volume number of patients randomised: 115 patients
Outcomes	Primary outcome(s): overall survival (defined as the time of randomisation to the date of death) Secondary outcome(s): cancer‐related mortality and non‐cancer mortality
Funding sources	This publication was supported by grants 2U10 CA 11488‐16 through 5U10CA011488‐38 from the National Cancer Institute (Bethesda, Maryland, USA) This study was also sponsored by KWF Kankerbestrijding. These sponsors contributed to the design and conduct of the study; data collection, management, analysis, and interpretation; and preparation, review, and approval of the manuscript
Declaration of interest	The author certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (e.g. employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: none.
Notes	The trial is underpowered to reach its goal of showing non‐inferiority ("The trial was designed to prove non‐inferiority of deferred ADT to early ADT [...] Three hundred twenty patients were considered required [...] Since the trial was launched, the conception of what might be called equivalence or non‐inferiority has evolved and now allows only much smaller survival losses and smaller false error rates [...], so that the original sample size calculation would now be considered unethical. Furthermore, the recruitment was difficult, so that not even the originally planned 320 evaluable men were recruited. Reliable information concerning the treatment modalities, which were applied at the investigators' discretion at the time of progression under endocrine treatment, is not available. However, yearly follow‐up indicates that 50% of the patients in the delayed and early group, respectively, continued the same treatment as per protocol after they reached the end of the protocol treatment.")
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Information from publication: Information was not reported. Comment: We assume that randomisation was performed adequately at the EORTC Data Centre. However, information was not reported and there is therefore unclear risk of bias.
Allocation concealment (selection bias)	Low risk	Information from publication: "centrally". Comment: Randomisation was performed centrally at the EORTC Data Centre.
Blinding of participants and personnel (performance bias) Time to death of any cause	Unclear risk	Information from publication: There was no blinding. Comment: Time to death of any cause was measured and reported. It might be conceivable that even objective outcomes are influenced by lack of blinding. We finally judge that there is an unclear risk of bias.
Blinding of participants and personnel (performance bias) All other outcomes	High risk	Information from publication: There was no blinding. Comment: Time to death from prostate cancer and few adverse events were measured and reported. We judge that subjective outcomes are likely to be influenced by lack of blinding.
Blinding of outcome assessment (detection bias) Time to death from any cause	Low risk	Information from publication: There was no blinding of outcome assessment (or it was not reported). Comment: Time to death of any cause was measured and reported. Blinding of outcome assessment could have been expected. However, we judge that it is not likely that outcome assessment for objective outcomes is influenced by lack of blinding.
Blinding of outcome assessment (detection bias) All other outcomes	High risk	Information from publication: There was no blinding of outcome assessment (or it was not reported). Comment: Time to death from prostate cancer and few adverse events were measured and reported. Blinding of outcome assessment could have been expected. We judge that outcome assessment of subjective outcomes is likely to be influenced by lack of blinding.
Incomplete outcome data (attrition bias) Oncological outcomes (Time‐to‐death of any cause, Time‐to‐disease progression, Time‐to‐death from prostate cancer)	Low risk	All patients randomised were included in the analysis for time to death of any cause and time to death from prostate cancer.
Incomplete outcome data (attrition bias) Adverse events (Serious and other adverse events)	Unclear risk	Not applicable (outcome not measured/reported). Only deaths due to cardiovascular events or infection were reported.
Incomplete outcome data (attrition bias) Quality of life	Unclear risk	Not applicable (outcome not measured/reported).
Selective reporting (reporting bias)	High risk	There was no assessment of adverse events (except death due to cardiovascular events or infection) but it could have been expected or adverse events were measured but not reported. Data for the predefined outcome 'Time to clinical progression' were evaluated but not reported.
Other bias	Unclear risk	We identified no other sources of bias.