Cautious Optimism About FDA’s One Pivotal Trial Policy

Cautious Optimism About FDA’s One Pivotal Trial Policy

By Deborah Borfitz

April 14, 2026 | Earlier this year, the U.S. Food and Drug Administration (FDA) made a single pivotal clinical trial the default requirement for getting a medicine to market in lieu of the traditional two-trial mandate. It was a controversial move designed to reduce clinical development costs for drug sponsors, but it comes with the peril of potential project failure if study results aren’t rigorously defensible—or if companies fail to invest heavily in the quality of that single trial using readily accessible tools for data monitoring, artificial intelligence (AI), and biosimulation.

The predicament has not escaped the attention of technology providers like Certara and eClinical Solutions that are now actively seeking to derisk these high-stakes trials. Certara brings biosimulation software and expert consulting to the table for predicting drug behavior and optimizing trial design, while eClinical Solutions offers a clinical data cloud platform and biometrics services to digitize and manage the flow of trial data.

The one-trial approach has long been in vogue for oncology and rare disease indications that require faster development pathways, notes Venu Mallarapu, chief transformation and AI officer at eClinical Solutions. That it is now being applied to all trials means the margin for error is much smaller while expectations for data quality, rigor, and operational perfection have increased substantially.

The FDA announced that one pivotal trial will now be the rule rather than the exception in an article that was published earlier this year in The New England Journal of Medicine (DOI: 10.1056/NEJMsb2517623). “This is one of the most significant changes in regulatory policy in a very long time, and I very much applaud it,” says Piet van der Graaf, PharmD, Ph.D., senior vice president of applied biosimulation at Certara, professor of systems pharmacology at Leiden University (Netherlands), and professor of pediatrics at Cincinnati Children’s Hospital Medical Center. “I would think most, if not everyone, in the industry would agree with me.”

Reactions to the new policy have been mixed, according to Mallarapu. While some expect it to accelerate the pace of bringing new medicines to market, the “contrarian view” is that it lowers safety standards and risks introducing unreliable drugs by relying on a weakened evidence base.

Both perspectives are valid, he adds, and the reason why sponsors need to “take stock of how things are managed internally” to identify their optimal strategy. Risk-based quality management (RBQM), a major specialty and focus area for eClinical Solutions, was “designed for just this kind of scenario.”

Revisions to ICH E8 (General Considerations for Clinical Studies), finalized in 2021, also align well with RBQM adoption. The international guideline advocates for a shift away from a rigid, checklist-based oversight approach to a flexible and proactive risk-based one focused on quality by design.

RBQM helps in building the process, systems, and teams for operating under the new regulatory framework, says Mallarapu. While one pivotal trial is a new standard, it is not a mandate, and the FDA may require more than a single trial if it is deemed necessary.

‘Vulnerable at the Seams’

The “RBQM problem,” referring to the stubbornly slow and uneven adoption rate of the approach, has yet to be solved despite over a decade of regulatory encouragement and proven benefits, Mallarapu says. Progress has been hindered by legacy processes, lack of skills, and technology adoption barriers.

“Many organizations have the framework on paper, but ... operational infrastructure doesn’t support and give context to be able to defend some of their risk-based decisions,” he continues. That’s another tappable opportunity, especially for sponsors looking to have AI agents to take over much of the work of surfacing trends across sites, flagging protocol deviations, and documenting that decisions being made are consistently right “so that they are not reacting to problems but preventing them.”

Real-time monitoring and data checks are the key to success when moving to a single pivotal study, and the context in which eClinical Solutions is having conversations with its customers, says Mallarapu. The single pivotal trial design amplifies all the risks clinical studies have traditionally faced and made them “more vulnerable at the seams between data systems.”

The biggest vulnerability, in his view, is fragmented data. Clinical operations, RBQM, safety, and site management data typically live in different systems within pharma companies, so they “lose the ability to connect the dots and do so in real time.” That’s a problem in a single trial model where one unmanaged protocol deviation pattern can provide sufficient grounds for regulatory refusal.

This is to say nothing about the “human bandwidth problem” when it comes to the data deluge in clinical trials. A recent study by the Tufts Center for the Study of Drug Development and TransCelerate BioPharma indicates that close to 6 million data points are being collected in phase 3 protocols.

AI agents are needed to quickly triage some of the data that’s being collected to flag what’s important for prioritization and give study teams the “ability and cognitive space to make decisions in near-real time if not real time,” Mallarapu says. “The companies most at risk are those that look at RBQM as a compliance checkbox rather than a discipline that requires integrated operations and access to data systems and teams ... automated around decision points [using AI].”

Maturing RBQM

The level of AI adoption in RBQM depends on the appetite of sponsors for the required technology investment and to what degree leadership views AI as a strategic necessity, says Mallarapu. His perspective is that AI is particularly useful for this kind of centralized oversight and “even more so in the context of this single pivotal study model.”

Rather than “automation for the sake of automation,” AI-powered RBQM is about giving human experts “the right signal at the right time so that they can make the right decisions that can make or break the study,” he says. The relevant capabilities of AI are fourfold—descriptive (insights into what’s happening), prediction (helping teams foresee what’s likely to happen), generative (providing insights for decision-making), and agentic (acting autonomously within defined guardrails)—and study sponsors are variably at one of these stages.

Operationalizing AI-assisted RBQM to support smarter, more resilient trials with earlier signal detection necessitates governance structures that protect patient safety without slowing development, adds Mallarapu. In the case of safety, medical monitors need to review the integrated data coming from study sites and have the wherewithal to initiate corrective actions (e.g., re-training, increased monitoring, or protocol amendments) against root causes.

Building a mature RBQM framework is not about the “state of SOPs,” he says, but the “state of continuous, connected visibility across your trial.” Mallarapu described its four characteristics as integrated, with all data feeding into a single risk review; proactive, with safety signals surfacing before any deviations escalate; documented, so the decisions are automated by design and help companies be regulatorily compliant; and scalable, meaning it works across a portfolio of studies.

For customers of eClinical Solutions, this is enabled by a data repository and analytics platform branded elluminate, which functions as a modern clinical data lakehouse. AI agents flag risk signals across sites and act as an advisor to humans in the loop—data managers, medical monitors, and safety reviewers—so they can make “faster and more confident decisions.”

Regulatory Expectations

The FDA’s new default option for marketing approvals is “not novel per se,” points out Certara’s van der Graaf. The agency issued a guidance document in 2023 indicating that a single trial would suffice provided it demonstrated “substantial evidence” of effectiveness and was “adequate and well controlled.”

“What is significant about the announcement made recently by the FDA is that [the one-trial paradigm] is now going to be the default,” he adds. The nature of drug discovery and development has fundamentally changed because so much more is known today about the biology of disease and the safety, mechanism of action, and potential efficacy of drugs in humans to inform the design of clinical trials and “we don’t have to ignore all that knowledge.”

In prior decades, two clinical trials were needed to avoid “type 1” errors, which refer to false-positive findings where a new treatment is wrongly claimed to be effective, says van der Graaf. A single, well-controlled trial is now all that’s needed to “get the right answer” by leveraging model-informed drug development (MIDD) and biosimulation software to ensure it is adequate and well controlled to provide the substantial evidence regulators expect.

Certara has for many years now been using biosimulation and MIDD to that end, he continues. “We can simulate a clinical trial many, many times before we actually do the clinical trial, to [among other things] help us understand the number of patients that we need ... [and] the different doses that we need to test.”

Quantitative systems pharmacology (QSP) is van der Graaf’s specialty, which means he is a frequent user of the company’s Simcyp QSP biosimulation platform used to model complex diseases and predict drug responses. It is integrated with the Simcyp Physiologically Based Pharmacokinetic (PBPK) Simulator, which specializes in physiologically-based pharmacokinetic modeling for predicting drug behavior in diverse populations, to create digital twins—computer-simulated models of individual patients.

The technology enables sponsors to virtually mimic clinical trials enrolling people with different phenotypes and disease states to better understand the variability, for example in a patient population, he explains. Certara has had early success in making virtual trials part of the substantial evidence expectations of regulators by demonstrating the convergence of actual and digital twins in the understanding of disease and pharmacology.

“It is increasingly being accepted that biosimulation can substitute for actual clinical trials,” and many of the current case studies have used Certara’s virtual twin technology, van der Graaf notes. These examples include the prediction of drug-drug interactions on novel therapeutics as well as dose simulations in lieu of actual dose-finding studies.

In his own field, for instance, biosimulation and virtual patients have supported programs for T cell engagers in oncology to define the first-in-human starting dose, he reports. Case studied have also been published where biosimulation allowed phase 1 studies to start at significantly higher dose levels, meaning “patients are not being exposed to very low, almost homeopathic, doses.”

Growth Industry

As covered in a recently published case study, PBPK predictions from Certara’s Simcyp Simulator informed the drug product label for asciminib, a Novartis drug for chronic myeloid leukemia, to replace at least 10 clinical studies (Pharmaceutics, DOI: 10.3390/pharmaceutics17101266). This saved Novartis a lot of clinical development time and “tens of millions” of dollars in running human-populated trials, says van der Graaf.

This is but one example of the still largely untapped potential of biosimulation and modeling in clinical trials, he adds. Over the last decade, more than 100 drug labels have been informed by the Simcyp Simulator across therapeutic areas.

The technology is generally considered the most mature, widely adopted, and best-known PBPK modeling platform in the biopharmaceutical industry. It is consequently “very well-known and widely accepted by regulatory agencies across the world,” says van der Graaf, adding that Certara’s platform was qualified last August by the European Medicines Agency for predicting drug-drug interactions.

Biosimulation is a highly popular area and growth industry with a long list of far smaller players. “Almost every pharmaceutical company now has its own biosimulation group doing this, as do regulatory agencies,” he says, including the FDA’s Office of Clinical Pharmacology.

It is a core component of model-informed drug development, the importance of which was further elevated by the International Council for Harmonization with the adoption of its M15 guideline in January, says van der Graaf. The document sets out a unified framework for planning, evaluating, and submitting MIDD evidence to support regulatory decision-making.

This feeds into van der Graaf’s passion for patient-centric drug development. “At the end of the day, everything we do is to serve patients better, and I think the [one pivotal trial] policy of the FDA is very much in that spirit,” he says, and where biosimulation has a key role to play.