The negative Phase 3 trial the commercial team wants to park
A 612-patient Phase 3 randomised trial of an oral JAK inhibitor in moderate-to-severe rheumatoid arthritis has just been unblinded. The primary endpoint, ACR20 at week 12, was not met: 47.2% versus 44.8%, p=0.41. One pre-specified subgroup - patients with baseline CRP above 8 mg/L, about 38% of the sample - showed 58% versus 41% with a nominal interaction p=0.03; there was no multiplicity control across the roughly 14 subgroup comparisons. The trial was prospectively registered and the primary completion date was 13 months ago; no results have been posted to the registry. At a commercial review meeting the marketing director asks that the primary manuscript be held until after an FDA advisory committee nine months away, that the CRP subgroup be the headline of any publication, and that the team skip congress presentation in favour of one high-impact journal submission. The medical affairs lead says she does not see a problem with the sequencing. You are the publication manager and must reply in writing within 48 hours.
1. How do you respond to the request to delay the primary manuscript, and what obligations are triggered by the trial's status?
Read the worked answer
Rule and facts first. GPP2022 and the ICMJE recommendations both treat public disclosure of trial results as an obligation owed to participants and to the evidence base, not a commercial option. Suppressing or deferring a negative result for marketing reasons is a breach regardless of the journal target.
The delay cannot be agreed. The stated rationale - protecting the asset ahead of an advisory committee - is not a defensible scientific reason for non-disclosure. A time-limited hold tied to a regulatory milestone is exactly the kind of arrangement GPP2022 prohibits, and journal editors increasingly check registry entries before accepting a trial report.
A separate, harder problem. Primary completion was 13 months ago and no results are posted. Where the trial falls within the applicable reporting requirements (for example the US requirement to post results generally within one year of primary completion), the company is already overdue and this is a compliance issue, not a publication preference. Flag it in writing to regulatory affairs and compliance the same day.
What you propose instead. (1) Post summary results to the registry within an agreed short window. (2) Submit the primary results manuscript now, with an honest title and abstract. (3) Build a separate reactive plan for the advisory committee - a congress presentation, a plain-language summary, and Q&A materials - that does not depend on withholding data. (4) Document the request, your reasoning, and the decision in the publication decision log, and escalate to the publication steering committee.
Tone. Give the marketing director an alternative rather than a flat refusal: the evidence can still be shaped well, disclosed on time, and contextualised, but not buried.
2. The CRP subgroup shows a numeric difference with a nominal interaction p=0.03. How should it be handled in the manuscript and in any derivative publication?
Read the worked answer
What the data support. A pre-specified subgroup with a nominal interaction p-value and no multiplicity control across roughly 14 comparisons is hypothesis-generating at best. The overall trial is negative, and the subgroup effect is not established.
Structural requirements for the manuscript. The title and abstract must reflect the primary, negative result. The subgroup belongs in the pre-specified secondary/exploratory section, reported with the interaction p-value, the subgroup n, the absence of multiplicity adjustment, and a plain statement that the finding requires independent confirmation. Do not move the subgroup into the primary endpoint position, do not rewrite the objective as if CRP selection were the question, and do not let the conclusion say the drug 'was effective in' that group.
Why the headline proposal fails. Leading with the subgroup would misrepresent the study's pre-specified question, invite a post-hoc-reframing criticism at peer review, and - if it were the only disclosure - effectively hide the negative primary result. That is spin, and it is one of the specific failures GPP2022 and journal reporting standards target.
A separate subgroup paper is possible, with conditions. It should follow, not replace, the primary publication; be labelled exploratory in the title; state the total number of subgroup analyses performed; and include a limitations paragraph on multiplicity.
Practical step. Circulate the statistical analysis plan and the full subgroup table to all authors before drafting, and have the trial statistician sign off on the interaction analysis. If the interaction claim is fragile, say so in the text rather than in a reviewer response.
3. What governance and documentation would you put in place over the next month so this does not recur?
Read the worked answer
Immediate written record. A dated decision log entry covering the request, the standards cited, the decision, who was consulted, and the escalation path. This protects the organisation and the individual if the question resurfaces.
Committee route. Take both issues to the publication steering committee with a standing agenda item on 'disclosure risk' - non-publication, delayed publication, and outcome-driven reframing. Membership should include medical affairs, regulatory, biostatistics, legal/compliance and a publication professional with a documented right to escalate.
Process controls. (1) A registry and results-posting tracker with owners and due dates, reviewed monthly, so overdue postings surface before commercial discussions begin. (2) A publication plan that records, for every study, the commitment to publish regardless of result direction, signed at study start rather than at manuscript stage. (3) A pre-specified rule that subgroup-led manuscripts require statistician sign-off and a multiplicity statement.
Capability building. Short training for the commercial and medical affairs teams on what GPP2022 does and does not permit, using anonymised internal examples rather than generic slides. Commercial colleagues often do not know that the advisory committee timetable is not a legitimate disclosure gate.
Metrics. Track plan-versus-actual delivery of primary results manuscripts, time from database lock to submission, and registry posting timeliness. Report quarterly to governance; a metric nobody reviews is decoration.