Duration of androgen deprivation therapy with postoperative radiotherapy for prostate cancer: a comparison of long-course versus short-course androgen deprivation therapy in the RADICALS-HD randomised trial.
Parker CC, Kynaston H, Cook AD, Clarke NW, Catton CN, Cross WR, Petersen PM, Persad RA, Pugh CA, Saad F, Logue J, Payne H, Bower LC, Brawley C, Rauchenberger M, Barkati M, Bottomley DM, Brasso K, Chung HT, Chung PWM, Conroy R, Falconer A, Ford V, Goh CL, Heath CM, James ND, Kim-Sing C, Kodavatiganti R, Malone SC, Morris SL, Nabid A, Ong AD, Raman R, Rodda S, Wells P, Worlding J, Parulekar WR, Parmar MKB, Sydes MR, RADICALS investigators
- DOI
- 10.1016/S0140-6736(24)00549-X
- Record issued
- 2026-08-16
- Engine
- 7.39.0
- Exported
- 2026-09-20
Prepared by Alpha1. This document is confidential: it is intended for the recipient it was shared with and must not be redistributed. The live record at alpha1science.com/verify/4ac5e671-67ec-41af-8602-ea78f6cc8315 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern ×3−1.5★
- CitationsUnresolved reference−0.25★
- LinksLink inconsistent with the paper−0.25★
- 01Declared data/code link does not hold what the paper says
Resolves to a parked / unavailable page ("Page Not Found | UCL Innovative Clinical Trials Unit") — the resource is not actually there.
“http://www.ctu.mrc.ac.uk/our_research/datasharing/application_process”
This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.
This is a well-conducted phase 3 randomised controlled trial with rigorous design, clear reporting, and appropriate statistical methods. The paper is transparent about its limitations and provides a data sharing statement. Minor reporting gaps include the lack of explicit statistical software identification and race/ethnicity data, plus a few copyedit issues in tables.
Both reviewers independently scored all dimensions and agreed on all statuses; no divergence to reconcile. The study type is interventional (phase 3 RCT). Non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification covered only a subset of tests (4 recomputed consistently); other statistics remain unverified. The citation check flagged one reference not found in registry.
Numerical inconsistencies
1 finding · worst lowValues that contradict each other or are impossible for the stated sample: recomputed p-values and test statistics, GRIM/GRIMMER checks on summary numbers, percentages against their own counts, totals against their parts, and estimates against their own confidence intervals.
- Internal contradictions in the reported numbersAssessed
Recomputed 4 tests: 4 consistent, 0 inconsistent; 2 recomputed directly from the reported test statistics, 2 via agent-written checks.
- CONSISTENTreported p = .029 · recomputed p = .030Recomputed HR 0.773 (95% CI 0.612–0.975), reported p=0.029
“HR 0.773 [95% CI 0.612–0.975]; p=0.029”
Taken as given: 0.612–0.975 is a two-sided 95% confidence interval for the HR of 0.773, not a range, an IQR, or a different interval level; the HR is a RATIO measure, so the interval is symmetric on the log scale; p=0.029 is the p for THIS estimate, not for another comparison in the same sentenceMethod: back the two-tailed p out of the log-scale CI width and compare it against the printed pHow we recomputed it: pCI(0.773, 0.612, 0.975, 1) - CONSISTENTreported p = .002 · recomputed p = .003Recomputed HR 0.634 (95% CI 0.471–0.853), reported p=0.0024
“HR 0.634 [95% CI 0.471–0.853]; p=0.0024”
Taken as given: 0.471–0.853 is a two-sided 95% confidence interval for the HR of 0.634, not a range, an IQR, or a different interval level; the HR is a RATIO measure, so the interval is symmetric on the log scale; p=0.0024 is the p for THIS estimate, not for another comparison in the same sentenceMethod: back the two-tailed p out of the log-scale CI width and compare it against the printed pHow we recomputed it: pCI(0.634, 0.471, 0.853, 1) - CONSISTENTreported p = .029 · recomputed p = .030Reviewers 1, 2Primary metastasis-free survival HR p-value
“Metastasis-free survival was improved in patients allocated to long-course ADT compared with short-course ADT (HR 0·773 [95% CI 0·612–0·975]; p=0·029; , ).”
Taken as given: The HR is a ratio (log-scale).; The 95% CI is two-sided.; The p-value is two-tailed.Method: Recomputed p-value from the reported HR and 95% CI using the normal approximation for the log hazard ratio.How we recomputed it: pCI(0.773, 0.612, 0.975, 1) - CONSISTENTreported p = .002 · recomputed p = .003Reviewers 1, 2Freedom from distant metastasis HR p-value
“Freedom from distant metastasis was improved in the long-course ADT group compared with the short-course ADT group (HR 0·634 [95% CI 0·471–0·853]; p=0·0024).”
Taken as given: The HR is a ratio (log-scale).; The 95% CI is two-sided.; The p-value is two-tailed.Method: Recomputed p-value from the reported HR and 95% CI using the normal approximation for the log hazard ratio.How we recomputed it: pCI(0.634, 0.471, 0.853, 1)
- lowinternal contradictionTable 2 lists the long-course ADT group as n=743, while Table 1 and the text state n=762. This may be due to missing data, but the discrepancy is not explained.
“Long-course ADT (n=743)”
Table 2Find in source - lowinternal contradictionIn Table 1, the percentage for Denmark in the short-course ADT column is listed as 95%, which is clearly a typographical error (should be ~5%).
“Denmark | 35 (95%) | 29 (4%) | 64 (4%)”
Table 1Find in source
Overstated conclusions
None foundConclusions that reach past what the paper's own results support — including a significance claim that no longer holds when the statistic is recomputed, and efficacy resting on an unvalidated surrogate endpoint.
Checked — nothing surfaced.
7 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewer 1Long-course ADT improves metastasis-free survival compared with short-course ADT.The primary outcome shows a statistically significant improvement with HR 0.773 (95% CI 0.612-0.975, p=0.029).Evidence: Primary outcome result: HR 0.773, p=0.029.
“Compared with adding 6 months of ADT, adding 24 months of ADT improved metastasis-free survival in people receiving postoperative radiotherapy.”
AbstractFind in source - supportedReviewer 1The benefit in metastasis-free survival did not translate into overall survival improvement.Overall survival HR 0.880 (95% CI 0.663-1.169, p=0.38) shows no significant difference.Evidence: Overall survival result: HR 0.880, p=0.38.
“We found no evidence of a benefit to overall survival with long-course ADT.”
Results ¶4Find in source - supportedReviewer 1Long-course ADT is associated with higher toxicity.Grade 3+ toxicity was higher in the long-course group (19% vs 14%, p=0.025).Evidence: Toxicity result: 142 (19%) vs 105 (14%), p=0.025.
Toxicity of grade 3 or higher was reported for 105 (14%) of 753 participants in the short-course ADT group and 142 (19%) of 757 participants in the long-course ADT group (p=0·025).
Abstractreviewer’s wording - supportedReviewers 1, 2The treatment effect did not differ by pre-radiotherapy PSA level or comorbidity.Interaction p-values were 0.99 and 0.67, respectively, indicating no significant heterogeneity.Evidence: Subgroup analyses: interaction p=0.99 for PSA, p=0.67 for Charlson score.
“The metastasis-free survival treatment effect did not differ meaningfully in either of the two prespecified subgroup analyses, pre-radiotherapy PSA level (interaction p=0·99) or Charlson Comorbidity Index score (interaction p=0·67; ).”
Results ¶3Find in source - supportedReviewer 2Long-course ADT (24 months) improves metastasis-free survival compared with short-course ADT (6 months) in patients receiving postoperative radiotherapy.The primary outcome analysis shows a statistically significant improvement in metastasis-free survival with long-course ADT.Evidence: HR 0.773 (95% CI 0.612-0.975), p=0.029, with 10-year MFS 78.1% vs 71.9%.
“Compared with adding 6 months of ADT, adding 24 months of ADT improved metastasis-free survival in people receiving postoperative radiotherapy.”
AbstractFind in source - supportedReviewer 2The benefit of long-course ADT did not translate into an improvement in overall survival.The overall survival analysis showed no statistically significant difference.Evidence: HR 0.880 (95% CI 0.663-1.169), p=0.38.
“We found no evidence of a benefit to overall survival with long-course ADT.”
Results ¶3Find in source - supportedReviewer 2Long-course ADT should be offered to individuals who can accept the additional duration of adverse effects.This is a clinical recommendation based on the observed benefit and known harms, appropriately weighed.Evidence: The paper discusses the trade-off between MFS benefit and adverse effects.
“For individuals who can accept the additional duration of adverse effects, long-course ADT should be offered with postoperative radiotherapy.”
AbstractFind in source
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary outcome is metastasis-free survival, a composite of distant metastasis or death from any cause. Although metastasis-free survival is a surrogate for overall survival, the paper cites the ICECaP study as evidence that metastasis-free survival is a robust early outcome measure for disease-specific survival, and the trial was designed with this endpoint. The paper also reports overall survival as a secondary outcome, showing no significant benefit, but the primary claim is based on metastasis-free survival, which is a clinically meaningful endpoint and validated as a surrogate.
“Based on the evidence from ICECaP that metastasis-free survival can serve as a useful intermediate outcome measure, it remains plausible that long-course ADT will improve overall survival.”
- ADEQUATEEffect sizeThe primary effect is a hazard ratio of 0.773 (95% CI 0.612–0.975) for metastasis-free survival, with an absolute improvement in 10-year metastasis-free survival from 71.9% to 78.1% (6.2% absolute benefit). The number needed to treat is 16. This is a clinically meaningful difference, and the paper explicitly states that the benefit should be weighed against adverse effects, indicating clinical significance.
“These results indicate that, on average, 16 people need to be treated with long-course ADT for one of them to avoid an metastasis-free survival event within 10 years.”
Data authenticity concerns
1 finding · worst lowAn adversarial read for patterns associated with data that may not be genuine: results that look too clean, implausibly large effects, duplicated data or images, and methods that do not match the results reported.
- Other integrity concernAssessed
3 integrity concerns flagged (0 high).
- lowotherIn Table 1, the percentage for Denmark in the short-course ADT group is listed as 95%, which is likely a typo for 5%.
“Denmark | 35 (95%) | 29 (4%) | 64 (4%)”
Table 1Find in source
Reporting gaps
None foundRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
Checked — nothing surfaced.
The introduction cites three prior phase 3 trials (RTOG 0534, GETUG-AFU 16, RTOG 9601) and notes that none compared different durations of ADT in the postoperative setting. The hypothesis that long-course ADT would be more effective is explicitly stated. Limitations of prior work (e.g., lack of duration comparisons) are addressed by the trial design.
“Radiotherapy is also commonly used after radical prostatectomy, but only three phase 3 randomised controlled trials in this setting have assessed the addition of ADT, and none have compared different durations of ADT.”
“In developing the RADICALS-HD trial in 2006, we hypothesised that long-course ADT would be more effective than short-course ADT in people receiving postoperative radiotherapy.”
“In developing the RADICALS-HD trial in 2006, we hypothesised that long-course ADT would be more effective than short-course ADT in people receiving postoperative radiotherapy.”
“Given the results of a subgroup analysis of RTOG 9601 in 2020, we also wanted to assess any benefit from long-course versus short-course ADT with respect to comorbidity and PSA levels at the time of radiotherapy.”
Randomisation was centralised using minimisation with a random element, stratified by key prognostic factors. The trial was open-label, which is stated. A power calculation was provided (80% power to detect HR 0.72). Inclusion/exclusion criteria were pre-specified. The analysis followed intention-to-treat. Outlier handling is not applicable in this context; missing data are addressed through the ITT principle and censoring.
“Randomisation was achieved centrally by the method of minimisation with a random element, stratified by Gleason score, positive margins, radiotherapy timing, planned radiotherapy schedule, and planned ADT type.”
“Based on 300 metastasis-free survival events from the 1523 participants, this revised design had 80% power with two-sided 5% α to detect an increase in 10-year metastasis-free survival from 75% to 81% (HR 0·72).”
“The allocated treatment was open label.”
“Randomisation was achieved centrally by the method of minimisation with a random element, stratified by Gleason score, positive margins, radiotherapy timing, planned radiotherapy schedule, and planned ADT type.”
“Based on 300 metastasis-free survival events from the 1523 participants, this revised design had 80% power with two-sided 5% α to detect an increase in 10-year metastasis-free survival from 75% to 81% (HR 0·72).”
“The allocated treatment was open label.”
Age, PSA, Gleason score, T stage, and other clinical characteristics are reported in Table 1. Sex is not explicitly reported but is implicit (prostate cancer). Since the study is single-sex by nature, a justification is not required. Demographics include age and country; race/ethnicity were not collected, which is acknowledged as a limitation.
“Age, years | 65 (60–69) | 65 (61–69) | 65 (60–69)”
“Data on race and ethnicity were not collected.”
“Data on race and ethnicity were not collected.”
The paper states 'Appropriate ethical review was in place for each participating country (appendix p 8)' and 'All participants gave written informed consent.' The appendix is referenced for details, which is acceptable. Regulatory compliance is implied through the trial registration and adherence to ethical standards.
“Appropriate ethical review was in place for each participating country (appendix p 8).”
“All participants gave written informed consent.”
“Appropriate ethical review was in place for each participating country (appendix p 8). All participants gave written informed consent.”
The ADT regimens (LHRH agonist, bicalutamide, degarelix) are described with dosing and schedule. The statistical software is not explicitly named in the text, but the analysis methods are standard. Since this is a drug trial, the investigational product is the key resource and is adequately identified.
“ADT was given with local choice of a subcutaneous gonadotrophin-releasing hormone analogue, supplemented by 3 weeks of an oral anti-androgen started 1 week before the first gonadotrophin-releasing hormone analogue administration.”
“ADT was given with local choice of a subcutaneous gonadotrophin-releasing hormone analogue, supplemented by 3 weeks of an oral anti-androgen started 1 week before the first gonadotrophin-releasing hormone analogue administration.”
The log-rank test and Cox regression are named. Proportional hazards assumption was tested. Exact p-values are reported (e.g., p=0.029). Effect sizes with 95% CIs are provided. The statistical software is not named, but this is a minor omission. Data presentation includes Kaplan-Meier curves and forest plots.
“The statistical significance of differences between groups was evaluated with the log-rank test, stratified by randomisation minimisation factors.”
“Metastasis-free survival was improved in patients allocated to long-course ADT compared with short-course ADT (HR 0·773 [95% CI 0·612–0·975]; p=0·029; , ).”
“The statistical significance of differences between groups was evaluated with the log-rank test, stratified by randomisation minimisation factors. Effect estimates were obtained from Cox regression models, also stratified by randomisation minimisation factors.”
“Metastasis-free survival was improved in patients allocated to long-course ADT compared with short-course ADT (HR 0·773 [95% CI 0·612–0·975]; p=0·029; , ).”
The data sharing section describes a controlled access approach with a formal application process and contact email. This is adequate for patient-level data. No code was generated, so code sharing is not applicable.
“The RADICALS trial data are held at the MRC Clinical Trials Unit at University College London (UCL), which encourages optimal use of data by using a controlled access approach to data sharing.”
“The RADICALS trial data are held at the MRC Clinical Trials Unit at University College London (UCL), which encourages optimal use of data by using a controlled access approach to data sharing. Requests for data can be made at any time and can be initiated by email to mrcctu. datareleaserequest@ucl.ac.uk or via our website.”
The trial is registered with ISRCTN and ClinicalTrials.gov. Methods are comprehensive. Limitations are explicitly discussed. Conclusions are proportional. Funding sources and conflicts of interest are declared. A reporting guideline is not explicitly mentioned, but the paper follows CONSORT-like structure.
“This study is registered with the ISRCTN registry, ISRCTN40814031, and with ClinicalTrials.gov (https://clinicaltrials.gov/) , NCT00541047 .”
“RADICALS-HD also has several limitations.”
“This study is registered with the ISRCTN registry, ISRCTN40814031, and with ClinicalTrials.gov (https://clinicaltrials.gov/) , NCT00541047 .”
“RADICALS-HD also has several limitations. Based on the recent results from the RADICALS-RT trial , and the ARTISTIC meta-analysis, postoperative radiotherapy is now typically given in the salvage, rather than the adjuvant, setting.”
Registered (3 IDs: ClinicalTrials.gov, ISRCTN, PROSPERO). No reporting guideline cited.
Broken references and links
2 findings · worst highReferences checked against Crossref, OpenAlex and Retraction Watch for retractions and resolvability, plus declared data and code links probed for whether they resolve to content matching the paper.
- Data/code link contradicts the paperRecomputed
- References not resolvable to a published paperRecomputed
Checked 17 references by DOI: 16 verified — 1 DOI unresolved.
- UNRESOLVED10.5281/zenodo.6586525RADICALS trial statistical analysis planCited DOI does not resolve to any Crossref record.
3 data/code links checked; 3 live; 1 resolved to content that does not match the paper.
- datahttp://www.ctu.mrc.ac.uk/our_research/datasharingLIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttp://www.ctu.mrc.ac.uk/our_research/datasharing/application_processLIVEINCONSISTENTHTTP 200Resolves to a parked / unavailable page ("Page Not Found | UCL Innovative Clinical Trials Unit") — the resource is not actually there.
- datahttps://www.mrcctu.ucl.ac.uk/media/1811/radicals-protocol-version-60-14-dec-2018_signed.pdfLIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
4 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 4 minor suggestions below.
4 copyedit issues flagged: mostly consistency, typo, clarity.
- MINORtypoTable 1, Country row“Denmark | 35 (95%) | 29 (4%) | 64 (4%)”→ Change '95%' to '5%' for the short-course ADT group.The percentage for Denmark in the short-course group appears to be a typo; 35/761 is approximately 5%, not 95%.
- MINORconsistencyTable 2, Long-course ADT n“Long-course ADT (n=743)”→ Verify whether the n should be 762 (as in Table 1) or 743 (as in Table 2).Table 1 lists n=762 for long-course ADT, but Table 2 lists n=743. This discrepancy may be due to missing data, but it should be clarified.
- MINORconsistencyTable 1, Country row“Denmark | 35 (95%) | 29 (4%) | 64 (4%)”→ The percentage for Denmark in the short-course column appears to be a typo; should likely be 5%.35/761 is approximately 4.6%, so 95% is clearly a typographical error.
- MINORclarityResults, paragraph 1“Of these 1523 participants, 1197 had been randomly allocated between only these two groups and 492 had been allocated to one of these groups as part of the RADICALS-HD three-way randomisation that also included no ADT.”→ Clarify that 1197 + 492 = 1689, which exceeds 1523, because some participants were counted in both groups? Or clarify the overlap.The numbers do not sum to the total, but the paper explains that some participants were part of a three-way randomisation, so the overlap is expected.
The published work is robust and well-reported. An informed reader should weigh the minor reporting gaps (statistical software not named, race/ethnicity not collected) and the table inconsistencies (Denmark percentage typo, n discrepancy) as minor issues that do not undermine the conclusions. No erratum is urgently required, but correcting the table typos and clarifying the n discrepancy would improve clarity.
- 1.HIGHcopyeditIn Table 1, correct the Denmark percentage in the short-course ADT column from '95%' to '5%'.The current value is a clear typographical error (35/761 ≈ 5%), which could mislead readers.
- 2.HIGHcopyeditIn Table 2, clarify why the long-course ADT group is listed as n=743 while Table 1 and text state n=762; add a footnote explaining any missing data.The discrepancy between tables is confusing and could be interpreted as an error; a footnote will resolve it.
- 3.HIGHreportingIn the Methods, Statistical analysis section, explicitly name the statistical software (e.g., Stata, R) and version used.Both reviewers flagged this omission; naming the software improves reproducibility and transparency.
- 4.HIGHotherVerify the reference 'RADICALS trial statistical analysis plan' (DOI 10.5281/zenodo.6586525) that was not found in the registry; correct the DOI or remove the reference if it cannot be verified.An unresolved reference may be a fabrication signal; verifying it is essential for integrity.
- 5.MEDIUMreportingIn the Methods or Acknowledgments, state adherence to CONSORT guidelines for reporting randomised trials.Explicitly mentioning the reporting guideline strengthens transparency and is a common reviewer expectation.
- 6.MEDIUMreportingIn the Results or Limitations, add a sentence justifying the absence of race/ethnicity data or note it as a limitation more prominently.The paper already notes the absence, but a brief justification would address potential concerns about generalizability.
- 7.MEDIUMreportingIn the Methods, Study design and participants, provide the specific ethics committee names and protocol numbers in the main text rather than only in the appendix.Naming the ethics bodies in the main text strengthens the ethics reporting and is more transparent.
- 8.MEDIUMdata codeCheck the data sharing link that was flagged as inconsistent with the paper's content and correct or update it.A broken or inconsistent link undermines the data availability statement and reproducibility.
- 9.MEDIUMreportingProvide a direct link to the full statistical analysis plan (SAP) in the Methods or Data sharing section.A publicly accessible SAP enhances transparency and allows readers to verify pre-specified analyses.
- 10.LOWcopyeditIn Results, paragraph 1, clarify the overlap in participant counts (1197 + 492 = 1689 vs 1523) by explaining that some participants were part of the three-way randomisation.The current wording may confuse readers about the total number of participants; a brief clarification would help.
The star rating is the report’s one-glance summary. Every paper starts at 5★ and loses stars for the concrete problems the review finds — so a rating is never a vague average, it’s a running total you can read line by line under “How this rating was calculated.”
- Reporting — 8 dimensionseach dimension that fully fails−½★
- each dimension partially met−¼★
- Statistics · Integrity · Claimseach serious problem−1★
- each medium problem−½★
- Citationseach retracted or unverifiable reference−¼★
- Copyeditonly when the manuscript needs a full edit−½★
The rating never drops below 1★, and a demonstrable critical failure (an impossible statistic, a proven ethics violation) caps it at 1★ on its own — so the stars can never look healthy when the verdict is CRITICAL.
The rating draws on a panel of agents. Three independent Kaimen Rigor reviewers grade the eight dimensions below across several independent passes (the shown verdict is their majority vote — steadier than any single run), isolate the paper’s major claims and check its own evidence backs them, and flag integrity concerns. Alongside them, a citation agent resolves every reference against Crossref, OpenAlex, and Retraction Watch; a statistics agent recomputes reported tests; and rule-based checks verify that declared data/code links actually resolve. Full text is required — an abstract-only submission is not analyzed.
Graded against NIH, MDAR, ARRIVE 2.0, CONSORT, EQUATOR, and RRID guidelines. A dimension that doesn’t apply to the study type is skipped, never penalized.
This Kaimen Rigor review is model-assisted and is not a substitute for formal expert review. It complements human evaluation by surfacing potential methodological concerns — verify each finding against the source.