Single-use evidence: designing a post-marketing commitment that can do more than tick the box

The strategic value of a post-marketing commitment is decided before a single patient enrols. Not at readout, when the data lands and someone in medical affairs asks whether it might support a label extension. At protocol design, months earlier, when a required study gets scoped to answer exactly one regulatory question and nothing else.

That study is what I call "single-use evidence." It does its one job, satisfies the letter of the commitment, closes the file, and then it is structurally incapable of doing a second, because the population is too narrow, there is no comparator, and the endpoints stop at safety. By the time anyone wants more from it, the protocol has set like poured concrete, and you are back at the table negotiating a fresh study from scratch.

Same architecture, deployed twice

When Janssen ran CANVAS (NCT01032629), the cardiovascular safety trial for canagliflozin, the primary endpoint was three-point MACE in a placebo-controlled population of 4,330 patients, standard cardiovascular-outcomes fare [2]. But the protocol also carried renal secondary endpoints from the start: albuminuria progression, change in eGFR. When the renal signal proved worth chasing, CREDENCE (NCT02065791) could pick up the same adjudicated, placebo-controlled, hard-outcome architecture and aim it at a renal population of 4,401 patients [3]. CREDENCE is publicly documented as supporting canagliflozin's 2019 label expansion in diabetic kidney disease. Same architecture, deployed twice: once to discharge an obligation, once to open a market.

A word on how to read that. I am not claiming Janssen scoped CANVAS with CREDENCE already in mind; no public source traces that intent, and I will not pretend one does. What the pair shows is narrower and more useful: rigour plus a second layer of endpoints leaves a door open that a minimal study bricks up.

PSOLAR (NCT00508547) makes the same point from the registry side. This psoriasis-biologic safety registry has enrolled 15,849 patients, and it captures PASI, DLQI and EuroQOL at every visit [4], none of which you need to answer "is this drug safe." That surplus is what makes the dataset reusable for effectiveness and value questions across the biologic class later.

The levers here are nameable, and there are only a handful of them: endpoints, comparator, adjudication, population breadth, and what you collect beyond the safety minimum. None of that is luck. Those choices, made at scoping, decide whether the data ever gets a second life.

Why the thinnest study is the rational default

If reuse comes down to a handful of choices, why do so many commitments get built without them? Under real conditions, the narrowest study is simply the sensible-looking call. That is worth diagnosing properly, not scolding outright.

A required study is a study you are obligated to deliver, on a clock, often while the rest of the organisation has already pivoted to launch, and every endpoint you add, every eligibility criterion you loosen, every comparator you introduce piles cost, recruitment risk and time onto something you have no choice about finishing. So teams scope to the letter, and the evidence that this is the norm is not subtle. In Li's cohort, only 29.3% of studies specified a registry as their data source, and 44.7% specified nothing [1]. Wallach and colleagues (2018), looking at PMRs at first approval between 2009 and 2012, found the median commitment description ran to 44 words, and 30% were too thin even to tell whether the study was progressing [7]. Wallach and colleagues (2019) found that only 14.8% of post-marketing commitments outlined any clinical trial at all [8].

A commitment can be fulfilled, filed on time, closed, and still throw off nothing you can use again. The sharpest evidence for this sits in a 2024 analysis of dose-related commitments [9]. Among fulfilled dose-related PMRs and PMCs in 2016–2022, only 30% (3 of 10) produced a label change, down from 42.9% (6 of 14) in the earlier 2010–2015 cohort. These are the studies that finished. Most of them still changed nothing about the product.

It gets easier to fool yourself when the reporting metrics look healthy. FDA's own performance reporting shows roughly 70% of open PMRs and 80% of PMCs submitted on time in FY2024 [10]. That is on-time annual reporting. It is a different and far kinder number than actual completion: of the FY2018 PMR cohort, only about 35% (112 of 316) were completed six years later [10]. On-time paperwork is not a finished study, and a finished study is not a reusable one. Three different bars  routinely collapsed into one green tick.

Three questions before the post-marketing commitment protocol locks

1. What does this commitment legally have to prove? The floor. The specific risk, the paediatric gap, the accelerated-approval effect you are required to verify. Get this exactly right, and no wider.
2. What would it need to also prove to be reusable for a label expansion, a new indication, or a payer and HTA dossier? The second layer. This is the deliberate part, and it is where most of the strategic value is won or lost.
3. Does the current draft protocol actually collect that second layer, or only the first? Read your own protocol as if you were the medical affairs lead two years from now, asking it a question it was never designed to answer.

The levers you pull to close that gap are the same ones the good examples used, plus timing:

• Endpoints: the effectiveness, quality-of-life or hard-outcome measures CANVAS and PSOLAR carried beyond their safety floor. A pre-specified secondary endpoint costs little at design and is ruinously expensive to bolt on later.
• Comparator: active, placebo, or none. Pfizer's ORAL Surveillance (NCT02092467) was a well-powered, active-comparator trial in 4,372 patients, exactly the design that enables reuse, and it reshaped tofacitinib's label with a boxed warning, not an indication [11]. Rigour buys optionality. It does not choose the direction.
• Data source: registry, claims, EHR. Of the RWE actually used in 218 FDA labelling expansions, Deng and colleagues (2025) found 75.0% drew on EHR data and only 4.6% on registries, 87.5% using a cohort design [12]. That tells you where the field sits, not which source gets reused more; no published breakdown links data-source type to reuse rate, so do not let a vendor sell you one. Whichever you pick has to clear the regulatory-grade bar to be reusable at all.
• Population breadth: indication-matched versus adjacent, US-only versus global enrolment. Every widening helps reuse and hurts feasibility. A judgement, not a formula.
• Timing: get the study genuinely underway early. Tibau and colleagues (2025) found indications with a confirmatory trial already ongoing at accelerated approval converted to full approval in a median 2.78 years, against 5.59 for those that had not started (p<0.001) [13]. The honest limit: starting earlier fixes timing, not rigour. FDA's January 2025 draft guidance on when a confirmatory trial counts as "underway" pushes this conversation to around the End-of-Phase-2 meeting [14].

One more lever sits with the regulators themselves, and the Europeans state it more plainly than the US framework does. EMA's post-authorisation efficacy studies guidance is explicit that reuse is the point; its Q&A says: "The MAH should consider whether the final results have an impact on the marketing authorisation... Relevant results of the study will be included in the SmPC" [15]. And FDA's March 2023 draft oncology guidance describes the same idea as a "one-trial" approach: one trial built to support accelerated approval and, with longer follow-up, serve as its own confirmatory evidence [16]. Both agencies are describing architecture built to do more than one job.

"Isn't the second layer just scope creep on the company's dime?"

For a resource-lean biotech already behind on a required study, the narrowest protocol is the responsible choice. That's not optional, whatever the budget looks like. You are obligated to deliver it, and regulatory requirment gives you no discretion to slow it down for your own convenience. Every endpoint, every broadened eligibility band, every active comparator adds cost, recruitment risk and months to a study whose only firm deadline belongs to the regulator. Building in "optionality" can look a lot like gold-plating a compliance exercise with money the company does not have. On a bet that may never pay out.

That argument is right about the cost and wrong about the timing of it. The marginal cost of the second layer at design time, a handful of extra fields on the CRF, one pre-specified secondary, a slightly wider eligibility band, is small set against the cost of negotiating and running a whole new study eighteen months later because the first one could not answer the question. The second layer should be the critical 10%, the minimum viable evidence that unlocks the most optionality, not a licence to boil the ocean. That discipline is exactly what a lean IEP is for when resources are tight.

That said, I have to concede the harder point, because the evidence forces it. Good design is necessary and it is not sufficient. AMAG's PROLONG trial (NCT01004029) for Makena was a rigorous, dual-primary, 1,740-patient randomised study, and it simply failed to confirm the original accelerated-approval effect; Makena was withdrawn in 2023 [17]. ORAL Surveillance produced a boxed warning. You cannot design your way to a positive result, and no amount of second-layer instrumentation manufactures a signal that was never in the drug. Good design buys you the option to reuse. It does not promise you will want to.

The floor is not the ceiling

The commitment letter tells you the floor. It never tells you the ceiling, and the ceiling is set the day the protocol locks, by you, whether or not you were thinking about it at the time.

So before you sign off a PMR or PMC protocol, run the three questions: what it must prove, what would make it reusable, and whether the draft in front of you collects that second layer or only the first. Do it while the protocol is still concrete you can pour, not concrete that has already set.

And do it inside your integrated evidence plan, not bolted on after submission. A post-marketing commitment is a late chapter of the same evidence story the IEP has been telling since first-in-human. It is the design decision that most cleanly carries a biotech from clinical development into medical affairs, and the one that quietly decides whether an FDA obligation becomes an asset or a receipt. The requirement is the same either way. What you do with the protocol is not.

References

1. Li LY, Ramachandran R, Ross JS, Wallach JD. (2026). "Premarket and postmarket real-world evidence studies supporting U.S. Food and Drug Administration regulatory decision-making, 2016-2024." Clinical Trials. PMID: 41641793 / PMC12900038. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12900038/
2. Janssen Research & Development. "CANVAS (CANagliflozin cardioVascular Assessment Study)." ClinicalTrials.gov: NCT01032629. https://clinicaltrials.gov/study/NCT01032629
3. Janssen Research & Development. "CREDENCE (Evaluation of the Effects of Canagliflozin on Renal and Cardiovascular Outcomes in Participants With Diabetic Nephropathy)." ClinicalTrials.gov: NCT02065791. https://clinicaltrials.gov/study/NCT02065791
4. Janssen. "PSOLAR (Psoriasis Longitudinal Assessment and Registry)." ClinicalTrials.gov: NCT00508547. https://clinicaltrials.gov/study/NCT00508547
5. UCB Pharma / University of California, San Diego (OTIS). "Cimzia (certolizumab pegol) pregnancy exposure registry." ClinicalTrials.gov: NCT01797224. https://clinicaltrials.gov/study/NCT01797224
6. AstraZeneca. "Exenatide pregnancy registry." ClinicalTrials.gov: NCT00579150. https://clinicaltrials.gov/study/NCT00579150
7. Wallach et al. (2018). "Postmarket studies required by the US Food and Drug Administration for new drugs and biologics approved between 2009 and 2012: cross sectional analysis." BMJ. PMID: 29794072. https://pubmed.ncbi.nlm.nih.gov/29794072/
8. Wallach JD, Luxkaranayagam AT, Dhruva SS, Miller JE, Ross JS. (2019). "Postmarketing commitments for novel drugs and biologics approved by the US Food and Drug Administration: a cross-sectional analysis." BMC Medicine. PMID: 31203816 / PMC6572730. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6572730/
9. Gendy JM, Nomura N, Stuart JN, Blumenthal G. (2024). "US FDA's Dose Optimization Postmarketing Requirements and Commitments of Oncology Approvals and the Impact on Product Labels from 2010 to 2022: An Emerging Landscape from Traditional to Novel Therapies." Therapeutic Innovation & Regulatory Science. PMID: 38182940 / PMC10850176. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10850176/
10. FDA. "Report on the Performance of Drug and Biologics Firms in Conducting Postmarketing Requirements and Commitments" (FY2024): 70% of open PMRs and 80% of PMCs submitted on time; 35% (112 of 316) of the FY2018 PMR cohort completed. https://www.fda.gov/drugs/postmarketing-requirements-and-commitments-introduction/postmarketing-requirements-and-commitments-reports (figures independently corroborated by RAPS: https://www.raps.org/resource/fda-finds-most-postmarketing-requirements-commitments-progressing-on-schedule.html)
11. Pfizer. "ORAL Surveillance (Safety Study of Tofacitinib Versus Tumor Necrosis Factor (TNF) Inhibitor in Subjects With Rheumatoid Arthritis)." ClinicalTrials.gov: NCT02092467. https://clinicaltrials.gov/study/NCT02092467
12. Deng, Girman, Ritchey (2025). "Real-World Evidence in FDA Approvals for Labeling Expansion of Small Molecules and Biologics." Therapeutic Innovation & Regulatory Science. PMID: 40468096 / PMC12446098. https://pubmed.ncbi.nlm.nih.gov/40468096/
13. Tibau et al. (2025). "Factors in Time to Full Approval or Withdrawal for Anticancer Medicines Granted Accelerated Approval by the FDA." JAMA Network Open. PMID: 40136298. https://pubmed.ncbi.nlm.nih.gov/40136298/
14. FDA (DRAFT, January 2025). "Accelerated Approval and Considerations for Determining Whether a Confirmatory Trial is Underway." Federal Register, 7 January 2025. The guidance states that alignment on confirmatory-trial design should occur "as soon as practicable, and generally soon after the End-of-Phase 2 meeting." https://www.federalregister.gov/documents/2025/01/07/2024-31527/accelerated-approval-and-considerations-for-determining-whether-a-confirmatory-trial-is-underway
15. EMA (FINAL, effective 1 June 2017). "Post-authorisation efficacy studies (PAES): questions and answers." EMA/PDCO/CAT/CMDh/PRAC/CHMP/261500/2015 (legal basis Article 108a, Directive 2001/83/EC). Quotation verified verbatim against the EMA Q&A (Question 8). https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/pharmacovigilance-post-authorisation/post-authorisation-efficacy-studies-questions-answers
16. FDA (DRAFT, March 2023). "Clinical Trial Considerations to Support Accelerated Approval of Oncology Therapeutics." Federal Register, 27 March 2023; remains in draft. Describes a "one-trial" approach a single randomised trial supporting accelerated approval and, through longer follow-up, later serving as the confirmatory evidence. https://www.federalregister.gov/documents/2023/03/27/2023-05910/clinical-trial-considerations-to-support-accelerated-approval-of-oncology-therapeutics-draft
17. AMAG Pharmaceuticals. "PROLONG (Progestin's Role in Optimizing Neonatal Gestation)." ClinicalTrials.gov: NCT01004029. https://clinicaltrials.gov/study/NCT01004029
18. Mitra-Majumdar et al. (2022). "Analysis of supportive evidence and postmarketing requirements/commitments for drugs approved in 2020." JAMA Network Open. PMID: 35579897. https://pubmed.ncbi.nlm.nih.gov/35579897/