A Phase 3 trial closes, the primary endpoint is met, and the sponsor files. But the question no one answered is the one the FDA reviewer will eventually ask: how long does the benefit actually last, and what happens to patients after the protocol window closes? A JAMA analysis published this week frames that gap precisely, arguing that extension studies address two distinct and irreplaceable functions: assessing durability of treatment benefit and detecting delayed adverse events that randomized controlled trial timelines structurally cannot capture.

Both functions matter operationally. And sponsors who treat extension studies as optional add-ons, rather than designed protocol phases, are setting themselves up for a painful conversation with regulators after approval.

What the JAMA Signal Actually Says

The JAMA authors identify two compounding problems with trial design as commonly practiced. First, most pivotal trials are powered for a primary endpoint measured at a fixed timepoint, not for sustained remission or functional durability over years. Second, rare or delayed adverse events, particularly in immunology, oncology, and neurology, may not manifest within the trial’s observation window at all. The extension study is the most direct mechanism for closing both gaps simultaneously.

That framing aligns with what ICH E1, adopted at Step 4 on October 27, 1994, has required for over three decades: drugs intended for long-term treatment of non-life-threatening conditions must include adequate population exposure data to characterize safety at scale. ICH E1 specifies minimum exposure thresholds, but the guideline is explicit that those thresholds represent a floor, not a ceiling. For chronic disease indications, the expectation of regulators on both sides of the Atlantic is that long-term safety data will follow the pivotal submission, not precede it.

What the JAMA analysis adds is a sharper clinical operations framing: the extension study is not a post-market commitment sponsors negotiate away. It is a protocol design decision that must be made before the pivotal trial enrolls its first patient.

The scale of the problem becomes concrete when you look at how infrequently extension studies are actually structured. A 2021 analysis in Clinical and Translational Science reviewed 216,944 trials registered on ClinicalTrials.gov between 2009 and 2019 and found only 1,115 classified as uncontrolled extension studies. That is a prevalence below 0.6%. For a design element that ICH E1 and FDA guidance treat as foundational to chronic disease drug development, that number is startling.

Who Carries the Operational Burden

Checkpoint inhibitor programs illustrate what happens when extension data arrives too late. A FAERS database analysis covering January 2011 to December 2023 identified 3,738 delayed immune-related adverse event cases out of 147,854 total cases reviewed. Those signals emerged years after initial approval, through passive pharmacovigilance, rather than through a designed extension study. The difference matters: passive surveillance catches events after harm, while a protocol-driven extension study can characterize incidence rates prospectively, build a safety narrative the label can actually use, and give the sponsor a defensible dataset if the FDA comes back with questions.

The retention math compounds the difficulty. Research on open-label extension (OLE) enrollment shows that an average of 74% of RCT participants enroll in a subsequent OLE, but only 57% of those who enroll complete it. Apply that attrition to a pivotal trial of 500 patients: roughly 370 enter the extension, and only 211 finish. For rare disease programs or indications with high dropout during the main trial, those numbers get worse. Sponsors who have not pre-specified their extension sample size requirements, retention strategies, and estimand framework before the pivotal trial begins will find themselves negotiating with an inadequate dataset.

Gene therapy programs face the most demanding version of this problem. FDA’s guidance on long-term follow-up after human gene therapy product administration explicitly requires that LTFU observations extend the scheduled assessments of pivotal studies, in some cases for many years post-dosing. The design and consent architecture for that kind of observation window cannot be retrofitted after a Phase 3 protocol is locked. It has to be built in from the start, with sites, IRBs, and patients aligned to a commitment that extends well past any individual trial team’s tenure on the program.

What Clinical Ops Leaders Must Do Now

If you are designing a protocol for a chronic disease, oncology, or gene therapy indication today, the extension study is not a Phase 4 discussion. It belongs in the protocol synopsis alongside the primary endpoint, with a pre-specified estimand, a retention budget, and a site selection criterion that accounts for multi-year follow-up capacity. The FDA’s posture on post-market commitments is tightening, and the difference between a commitment the agency negotiates into your approval letter and a dataset you control at filing is the difference between a clean label and a constrained one.

The EMA operates under a parallel framework for extension data, with requirements codified under 21 CFR Part 99 governing conditions for study completion extensions on supplemental applications. For sponsors running global development programs, reconciling FDA and EMA requirements early, before the pivotal protocol is finalized, determines whether one extension study serves both agencies or whether you are running two.

The next signal to watch is the FDA’s advisory committee calendar through early 2027. Several chronic disease and immunology NDAs with limited long-term data are approaching PDUFA dates, and AdCom panels in these areas have grown increasingly pointed in their questions about durability endpoints and post-approval safety architecture. The sponsor who arrives at that table with a designed extension study already enrolling is positioned differently from the one still negotiating what long-term data will look like.

References

  1. JAMA, “Beyond the Trial, What Do Extension Studies Add to Randomized Clinical Trials?”
  2. ICH E1, “The Extent of Population Exposure to Assess Clinical Safety for Drugs Intended for Long-Term Treatment of Non-Life-Threatening Conditions”
  3. Clinical and Translational Science, “Uncontrolled Extensions of Clinical Trials and the Use of External Controls, Scoping Opportunities and Methods” (2021)
  4. FAERS Analysis, Delayed Immune-Related Adverse Events in Checkpoint Inhibitor Studies (2011–2023)
  5. ResearchGate, “Open-Label Extension Studies: Interpretation and Statistical Issues”
  6. FDA, “Long Term Follow-up After Administration of Human Gene Therapy Products: Guidance for Industry”
  7. eCFR, 21 CFR Part 99, Subpart D, Section 99.303: FDA Extension of Time for Completing Studies
Website | + posts

Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.