FDA's Single-Trial Default Reshapes Drug Development: What the February 2026 Policy Means for Sponsors and Sites
核心洞察
The FDA announced on February 19, 2026, that a single adequate and well-controlled clinical trial with confirmatory evidence is now the default standard for drug approval, replacing the two-trial doctrine in place since 1962.
The policy shift, formalized in an NEJM correspondence piece in July 2026, compresses development timelines and concentrates evidentiary burden onto a single pivotal trial, demanding deeper biomarker and mechanistic support.
Sponsors in rare disease, oncology, and CNS indications face immediate disruption, while sites must now absorb greater operational risk with compressed enrollment windows and heightened Phase 2 scrutiny.
On February 19, 2026, the FDA announced a structural shift in drug development that most sponsors have yet to fully operationalize: a single adequate and well-controlled clinical trial, paired with confirmatory evidence, is now the default standard for drug approval. The two-trial doctrine that has governed New Drug Application (NDA) and Biologics License Application (BLA) strategy since the Kefauver-Harris Amendment of October 10, 1962, is no longer the starting assumption.
The policy's significance was underscored by a correspondence piece in the New England Journal of Medicine, Volume 395, Issue 2, published July 9, 2026 — five months after the FDA's announcement. Its appearance in NEJM signals that the academic and regulatory communities are converging on this shift as settled doctrine, not as a pilot program. Sponsors still building two-trial programs as the default are, according to the analysis, "running an outdated playbook at significant cost to time and capital."
The Evidentiary Framework Replaces the Dogma
The Kefauver-Harris Amendment demanded proof of effectiveness, and over six decades the FDA interpreted this into a practical standard: two independent, adequate, and well-controlled investigations. The logic was reproducibility — one trial could be noise, two confirmed a signal. That reasoning, rooted in an era of large, homogeneous trial populations and limited biomarker stratification, makes less sense in today's environment of adaptive designs and molecularly defined cohorts.
The FDA's 2023 Draft Guidance on single pivotal trial pathways outlined what confirmatory evidence can encompass: robust mechanistic data, consistent results across subgroup analyses, corroborating biomarker readouts, and real-world data supporting the primary endpoint signal. The February 2026 policy formalized that guidance into the operating default. As the NEJM correspondence makes clear, advisory committees, peer reviewers, and agency statisticians will now calibrate their expectations accordingly. Sponsors submitting two-trial packages without explaining why two trials were necessary may face unanticipated scrutiny.
The precedent for single-trial approvals is not new but was previously treated as the exception. Amgen's LUMAKRAS (sotorasib), cleared in May 2021 as the first targeted therapy for KRAS G12C-mutated non-small cell lung cancer (搜索), reached approval on the basis of a single study under Breakthrough Therapy designation. The February 2026 policy removes that distinction: single-trial submission is now available as a default pathway regardless of designation status.
Who Faces Disruption First
Sponsors running Phase 2/3 programs in rare disease, oncology, and CNS indications — where enrollment is slow, patient populations are small, and the cost of a second confirmatory trial often exceeds commercial return — are most immediately affected. Between December 1992 and December 2021, 278 drugs were approved under the FDA's Accelerated Approval pathway, with 50% converting to traditional approval by the end of that period. Many relied on surrogate endpoints precisely because running two full confirmatory trials was operationally impossible. The single-trial default, combined with Accelerated Approval infrastructure, provides a cleaner path: one rigorous trial with a pre-specified primary endpoint, supported by mechanistic data and biomarker consistency.
The disruption cuts differently for large-molecule programs in crowded therapeutic categories. A sponsor developing a fourth-in-class PCSK9 inhibitor faces a different evidentiary bar. The confirmatory evidence standard assumes the single trial is demonstrably robust — pre-registered, properly powered, with transparent methodology. In categories where comparative effectiveness matters, a single trial without a head-to-head arm may not satisfy that bar even if it technically meets the regulatory threshold. The policy sets a floor, not a ceiling.
PDUFA VII, signed into law on September 30, 2022, included explicit commitments to advance complex adaptive and Bayesian trial designs — exactly the tools that make a single pivotal trial credible. Sponsors who have invested in adaptive Phase 2/3 designs, Bayesian interim analyses, and pre-specified subgroup analyses that function as internal replication are positioned to move faster. Those still running fixed-sample, two-arm trials with traditional frequentist analyses are not.
Operational Friction at the Site Level
The gap between sponsor strategy and site execution is where the February 2026 shift creates its sharpest operational friction. The single-trial default compresses every downstream timeline into one trial that must carry the full evidentiary weight. For CNS programs, where screen failure rates already run 40 to 60 percent on highly specific populations, every delayed startup day now costs twice as much program risk as it did under a two-trial architecture.
Lecanemab's path to traditional approval illustrates the operational demands. The FDA's traditional approval of lecanemab (Leqembi) on July 6, 2023, was supported by data from a single Phase 3 randomized controlled trial, CLARITY AD, which enrolled 1,795 participants across a narrowly defined early Alzheimer's population. Sites were required to execute amyloid (搜索) confirmation via PET or CSF at screening — a procedure that adds cost, calendar time, and a meaningful screen failure rate. Under the February 2026 framework, every CNS program targeting a biomarker-confirmed population will face that same intensity without a second trial to absorb enrollment shortfalls.
The budget implications are substantial. Research from the Tufts Center for the Study of Drug Development found that a single protocol amendment in a typical Phase 3 trial adds an average of three months and $535,000 in direct costs. Under the single-trial default, any mid-course amendment becomes a significant event at a site network already running at full enrollment pressure. Contract language is already shifting: amendment clauses that previously triggered routine budget reconciliations are now structured with performance holdbacks tied to enrollment pace.
Phase 2 Rigor Becomes a Site Readiness Test
The single-trial default does not reduce the evidential burden — it concentrates it. Sponsors need their Phase 2 data to carry far more weight than historically required. The FDA's March 2024 revised draft guidance on Early Alzheimer's Disease (搜索) drug development makes clear that endpoint selection and biomarker validation in Phase 2 are prerequisites for a credible Phase 3 submission. For site operations, the quality of data from a 60-patient Phase 2 cohort is now a direct input into whether a sponsor will invest in a full single-trial Phase 3 at those same sites.
ICH E6(R3), formally adopted by the FDA on September 8, 2025, incorporates risk-based monitoring frameworks that make it operationally straightforward for sponsors to deploy intensive remote oversight across both phases. Sites that built documentation discipline into Phase 2 — ALCOA-C-compliant source records, TMF completeness above 95 percent at interim milestones, PI-level protocol deviation review within 48 hours — are the sites getting Phase 3 activation calls. Those that did not are being scored out of networks they expected to stay in.
The Protocol Decision That Cannot Be Deferred
For sponsors currently designing a Phase 3 protocol with a two-trial confirmatory strategy, the February 2026 policy requires an active choice. Defaulting to two trials now carries an implicit cost: consuming budget, time, and patient exposure to generate a second dataset the agency no longer requires as a baseline. That may be defensible in high-prevalence conditions where enrollment is feasible or where two trials strengthen reimbursement negotiations — but it must be a documented decision, surfaced in Type B meeting requests, not an inherited assumption from 1962.
The practical directive for clinical operations leaders: before the next IND or pre-NDA meeting, audit the confirmatory evidence package as if the single trial is already complete. Can biomarker data stand alone as mechanistic support? Are subgroup analyses pre-specified with adequate power? Has the statistical analysis plan been reviewed against the 2023 Draft Guidance's confirmatory evidence standards? The FDA's new default means the agency will ask those questions at submission, and the answer cannot be "we were planning to run a second trial."
The 2023 Draft Guidance has not yet been finalized. When it publishes in the Federal Register, every sponsor with an active IND will need to reopen their confirmatory evidence strategy — and those who have already built adaptive, biomarker-anchored protocols will have a substantial head start.