Regulators, clinicians, ethicists, industry leaders, and patient advocates gathered at the FDA–University of Maryland CERSI ADEPT 10 workshop to address a persistent unfortunate reality: neonatal drug development remains structurally misaligned with both science and policy. Rare disease frameworks may offer a path forward if sponsors are willing to rethink evidence, risk, and trial design.
Note: The following is an overview of the 10-hour workshop. All information came directly from that meeting. The report was written in collaboration with generative AI, with specific direction, resources, editing, and review provided by me.
Why Neonates Through a Rare Disease Lens?
The ADEPT 10 workshop — Addressing Challenges in Neonatal Product Development – Leveraging Rare Disease Frameworks — focused on drugs and biologics only (not devices), and on neonates, not fetal therapies. Why?
Because our youngest, most vulnerable patients pose drug development challenges that are dramatically different from older infants, children, and adults. The purpose of ADEPT 10 workshop was to bring together stakeholders from both the rare disease and neonatology communities to “share experiences, lessons learned, and ideas about how we can move forward in these two challenging product development areas,” said Anne Massaro, MD, Office of Pediatric Therapeutics (OPT), FDA.
Why the emphasis on rare disease? Many neonatal conditions are, in effect, rare diseases and should be developed as such.
Although clinicians may not instinctively categorize common NICU conditions as “rare,” Melissa Lestini, MD, neonatologist in OPT, reminded the audience that under statutory definition, rare disease means affecting fewer than 200,000 persons in the United States. Many neonatal conditions — bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), severe retinopathy of prematurity — fall well below that threshold.
The implication: If neonatal diseases are structurally rare, then rare disease development tools — regulatory incentives, adaptive design logic, patient-centered endpoints — should be actively leveraged.
1,300 Pediatric Labeling Changes, Less Than 15% in Neonates
While FDA recently surpassed 1,300 pediatric labeling changes under BPCA and PREA, “less than 15% of those have included studies in neonates or indications for neonates,” Lestini said.
That gap persists despite nearly 25 years of pediatric legislation.
Structural drivers include:
- Small population sizes
- Rapid developmental physiology and ontogeny
- Organ maturation variability
- Limited validated outcome measures
- Ethical safeguards for vulnerable populations
- Long-term follow-up requirements
- Limited commercial incentive
For R&D leaders who understand pediatric drug development, this list is familiar. But in neonates, these challenges compound. Drug metabolism is not simply scaled-down adult pharmacology. Enzymatic expression, receptor density, organ function, and circulatory transitions shift dynamically in early life.
As Lestini emphasized, extrapolation is often constrained in neonates because “rapid physiologic changes and organ maturation” limit assumptions that disease course and drug response mirror older populations.
Regulatory signal: expect FDA scrutiny on PK/PD modeling, dosing rationale, and developmental safety margins in neonates to be more stringent than in older pediatric cohorts.
Legislative Tools and a Structural Constraint
Next, the speakers later surfaced a lesser-discussed tension in pediatric policy.
Under the Pediatric Research Equity Act (PREA), sponsors must study new drugs in pediatric populations if there is pathophysiologic overlap with adults. But orphan-designated diseases are exempt from PREA requirements.
“Rare diseases that affect both adult and neonatal populations may not see the benefit of legislative mandates like PREA unless this orphan exemption is changed in the future,” Lestini said.
This creates a paradox: diseases rare enough to qualify for orphan incentives may simultaneously be exempt from pediatric study mandates.
For neonatal sponsors, that means:
- BPCA (voluntary, exclusivity-based incentive) remains critical.
- Orphan Drug Act incentives can help.
- Rare Pediatric Disease Priority Review Vouchers (RPD PRV) have been granted — but neonatal-specific designations represent only 6% of total RPD designations.
- No novel drugs have been approved for neonatal-specific conditions since surfactants in the 1990s.
The policy infrastructure exists, yet neonatal therapy development remains sparse. That’s a gap well worth addressing.
Public Health Burden: Rare Does Not Mean Small Impact
Connecting neonatal disorders to the Global Burden of Disease framework puts things into unsettling perspective.
For neonates, nearly all disability-adjusted life years (DALYs) are attributable to years of life lost — early mortality.
Neonatal disorders rank among the leading global causes of disease burden — surpassed in 2021 only by COVID-19 and essentially tied with ischemic heart disease.
What this means: small patient numbers do not equate to small societal impact, particularly when lifetime costs of prematurity exceed $25 billion annually in the United States.
Rare Disease Trial Failure: Rethinking Study Design
Marshall Summar, MD, founder of Uncommon Cures and longtime rare disease investigator, shifted the lens from neonates specifically to structural rare disease trial failures.
Orphanet data suggest that 80–90% of rare diseases affect fewer than 150 patients.
Why do these trials fail? For the same reason many clinical trials fail.
“In rare disease, about 55% to 60% of the time, it’s actually recruitment and retention — which actually comes back to study design.”
The traditional double-blind, placebo-controlled design increases Type II error risk in small cohorts. Summar argued that rare disease sponsors often face a greater danger of false negatives than false positives.
“The bigger risk in rare disease clinical research… is actually the false negative,” he said.
This critique is directly relevant to neonates. If populations are small and heterogeneity high, classical powering assumptions may not hold. Sponsors who rigidly apply traditional paradigms risk underpowered, inconclusive trials, even when the science is strong.
Ethical Architecture: Subpart D and the Moral Dimension of Risk
The afternoon session pivoted to ethics; specifically, 21 CFR 50 Subpart D, the additional safeguards for children in clinical investigations.
Dahlia Feltman, MD, physician ethicist at FDA and neonatologist, reminded attendees that pediatric research approval hinges on two central regulatory pathways:
- 50.52 — prospect of direct benefit, with risk justified by anticipated benefit.
- 50.53 — no prospect of direct benefit, risk limited to a minor increase over minimal risk.
She underscored that children are a vulnerable population because they cannot provide informed consent themselves. But she also highlighted a pendulum swing: from unethical inclusion of children to inappropriate exclusion.
The regulatory framework attempts to balance protection with access. Crucially, Feltman emphasized that when evidence is limited — as is often the case in rare disease and neonates — benefit-risk evaluation becomes complex.
“We may have clinical data… or we might be able to extrapolate… But sometimes we only have non-clinical studies such as animal or in vitro data.”
That sets the stage for the central theme of Panel A.
Panel A: When Doing Nothing Is the Only Certainty
Allyson Berent, DVM, DACVIM, chief science officer of FAST (Foundation for Angelman Syndrome Therapeutics), offered one of the most resonant reframes of the day.
“In severe and debilitating disorders with no approved therapies, the baseline risk of doing nothing is actually the most certain thing that we know.”
She shared a perspective often underweighted in regulatory discourse: families understand uncertainty — they live it. They distinguish between uncertain risk and certain disease progression.
“Uncertainty should never be confused with recklessness,” she said.
This is not a call for lax standards. Berent advocated rigorous translational science, biomarker development, safety monitoring, and structured dose-escalation.
But she argued that informed consent discussions often overemphasize theoretical risks while underrepresenting the lived certainty of decline.
For sponsors, the implication is operational:
- Engage families early.
- Develop endpoints that reflect meaningful functional change.
- Be transparent about both uncertainty and disease trajectory.
Clinical Judgment, Not Statistical Worship
Robert “Skip” Nelson, MD, executive director of pediatric drug development at Johnson & Johnson and former FDA senior pediatric ethicist, grounded the discussion in regulatory philosophy.
“The tolerable degree of uncertainty is a clinical judgment,” he said, referencing ICH E11A.
Nelson emphasized that Subpart D was intentionally framed around how responsible parents make decisions: weighing severity, alternatives, and urgency.
Risk assessment, he argued, is “fundamentally a moral judgment,” not merely a statistical one.
Takeaway: FDA deliberations are not purely algorithmic. They are contextual, and dialogic, and human. Sponsors who treat regulatory engagement as a checkbox exercise miss the deliberative nature of pediatric review.
Off-Label Reality in the NICU
Matthew Rysavy, MD, PhD, neonatologist and clinical trialist, highlighted a well-known fact in pediatric medicine:
“The majority of the treatments that we give [children] are off-label.”
Neonatology, only 50 years old as a subspecialty, evolved from adult medicine, using adult ventilators and drugs, and scaling down as needed.
This historical reality complicates modern ethics debates. Standard of care in NICUs frequently involves therapies never formally studied in neonates.
Thus, when evaluating investigational products, comparisons to “usual care” must acknowledge that usual care itself often rests on limited neonatal-specific evidence.
For sponsors, this reframes comparator debates. In some contexts, the investigational therapy may be better characterized relative to off-label precedent rather than labeled standard.
Top Takeaways from Day 1
Day 1 of ADEPT 10 brought the following points home:
- Neonatal diseases structurally resemble rare diseases.
- Legislative incentives exist but are underutilized in neonatal-specific innovation.
- Trial failure in rare disease is often design-driven, not science-driven.
- Benefit–risk in pediatric contexts is deliberative and morally grounded.
- Families weigh certainty of decline against uncertainty of intervention.
- Off-label precedent complicates comparator ethics.
For drug and biologic developers, the message is clear: neonatal development requires neonatal-specific paradigms.
From Constraint to Strategy: Innovative Trial Designs in Small Populations
The second day opened with a pragmatic question: If neonates and many pediatric conditions function like rare diseases (or actually are rare diseases), how should sponsors design trials accordingly?
Rebecca Chiu, PhD, from CDER’s Office of Biostatistics, framed the discussion by attempting to define innovative trial design, a study type the FDA is warming up to. It typically refers to studies that don’t follow the traditional randomized controlled trial model. Basket and adaptive trials are examples.
“There’s actually no fixed definition or universally agreed upon definition of what makes a trial design innovative,” she said. “Typically, what we consider as innovative tends to change over time as our methods become more familiar or more widely used.”
Rare diseases — and many neonatal conditions — face barriers that make conventional trials more difficult to conduct, including:
- Extremely limited patient numbers
- Poorly characterized natural history
- Uncertain effect sizes
- Heterogeneous phenotypes
- Limited consensus on endpoints
Alternative trial designs have the flexibility to accommodate trials in very small, heterogeneous patient populations.
Bayesian Borrowing: Efficiency With Guardrails
Bayesian methods were highlighted as particularly relevant in small populations, where prior data can be leveraged.
“Bayesian approaches work by formally combining prior information with new trial data to obtain an updated understanding of the treatment effects,” Chiu said.
In a childhood-onset lupus example, pediatric data were combined post hoc with adult trial results using a robust mixture prior. When sufficient weight was assigned to adult evidence, the posterior credible interval excluded no treatment effect, supporting efficacy.
What this means for sponsors:
- Bayesian borrowing is acceptable — when scientifically justified.
- Extensive simulation is required.
- Sensitivity analyses must test assumptions about compatibility.
This is not a statistical shortcut. It is a structured framework for incorporating prior knowledge into small trials without inflating Type I error.
Adaptive Designs: Pre-Specified Flexibility
Chiu later explained the pros and cons of adaptive trial designs.
“Adaptive designs allow us to prospectively plan modifications based on accumulating data,” she said.
Types of adaptive trials include:
- Group sequential designs
- Sample size re-estimation
- Adaptive enrichment
- Response-adaptive randomization
- Predictive probability adjustments
An epilepsy example discussed under FDA’s Complex Innovative Design (CID) program combined Bayesian borrowing with interim predictive probability analyses to adapt sample size — reducing required enrollment in a rare pediatric population.
“These designs often require careful planning… extensive simulations and pre-specification to avoid increasing the chance of erroneous conclusions,” she said. Chiu also emphasized an FDA best practice: engage early. Adaptive trial designs require careful planning from the outset. Early discussions with the FDA can help identify potential risks before they become issues that potentially prolong the study timeline.
External Controls and Baseline Controls: When Randomization Is Not Feasible
Externally controlled trials received careful attention, especially in neonatology, where placebo arms may be ethically untenable.
Externally controlled trials compare treated patients to historical or real-world controls and are “most convincing when we do see a large effect and usually on an objective endpoint.”
Chiu emphasized prerequisites:
- Strong natural history data
- Low likelihood of spontaneous improvement
- Large treatment effects
When asked about baseline (within-patient) controls, she responded:
“The considerations are actually similar… we really need to understand the natural course of the disease to ensure that whatever effect we are seeing can actually be attributed to the treatment itself.”
The underlying principle: causal attribution remains central. Flexibility does not eliminate epistemic responsibility.
Master Protocols and Data Sharing: Infrastructure as an Asset
Master protocols — basket, umbrella, and platform designs — were framed as tools to share infrastructure and potentially share controls.
“These designs can be particularly powerful in rare diseases because they can allow for sharing of infrastructure, and in some cases, allow for sharing of controls,” Chiu said.
But patient advocates highlighted a structural barrier: industry reluctance to share placebo or sham control data.
“There has to be a new infrastructure… some incentive for them to share that placebo control and sham control data,” one advocate argued.
This is a strategic point. Rare and neonatal programs may benefit from:
- Shared control arms
- Harmonized endpoints
- Common data elements
- Centralized registries
Absent incentives or policy scaffolding, however, such collaboration remains voluntary.
AI and Digital Health: From Data Density to Decision Support
Ryan McAdams, MD, neonatologist at the University of Wisconsin, shifted the lens to AI in the NICU.
“We’re overwhelmed with data,” he said. “Per patient, gigabytes of data.”
Neonates generate continuous physiologic streams — heart rate variability, respiratory patterns, labs, imaging — yet most models remain preclinical.
“Most of the studies right now are preclinical… they haven’t been generalized across hospitals, devices, and populations,” he said.
Examples such as heart rate variability monitoring for early sepsis detection and AI-based retinopathy screening illustrate potential.
Generalizability and reimbursement remain limiting factors.
Food for thought for sponsors:
- AI may enable earlier detection of progression.
- AI-derived biomarkers may become surrogate endpoints.
- Digital health technologies could help compress timelines.
Yet regulatory validation remains the limiting factor. Its evolving, thankfully.
EMA Perspective: Structured Pediatric Obligations
Ralph Bax, head of the Pediatric Medicines Office at EMA, provided the European lens on clinical trials.
He emphasized early interaction mechanisms, including the Innovation Task Force, academic briefings, and SME support as entry points for complex developments.
A critical structural difference the EMA and the FDA: under EU pediatric regulation, most products are not exempt from pediatric obligations.
The Pediatric Investigation Plan (PIP) is “mandatory… binding,” though modifiable, he said.
To address uncertainty in early planning, EMA has piloted a stepwise PIP approach allowing submission even when all elements are not fully defined.
Acceleration mechanisms include:
- PRIME (Priority Medicines)
- Accelerated assessment
- Conditional marketing authorization
- Marketing authorization under exceptional circumstances
For ultra-rare neonatal conditions, “marketing authorization under exceptional circumstances” may apply where full evidence “might never be expected.”
For global R&D teams, early FDA–EMA alignment on pediatric strategy is essential.
FDA Expedited Programs: Leveraging Synergy
Janet Maynard, MD, director of FDA’s Office of Rare Diseases, Pediatric, Urologic, and Reproductive Medicine, outlined the expedited FDA pathways:
- Priority Review
- Accelerated Approval
- Fast Track
- Breakthrough Therapy Designation
“These programs are really intended to facilitate and expedite the development and review of drugs and biologics that are for serious or life-threatening conditions that address unmet needs,” she said.
Nijat Bushkooj, MD, from CBER, highlighted Regenerative Medicine Advanced Therapy (RMAT) designation for cell and gene therapies.
Rare-disease–specific programs include:
- Rare Disease Endpoint Advancement (RDEA)
- START pilot program
- Rare Disease Evidence Principles (RDEP)
The RDEP program is particularly relevant for genetically defined diseases with extremely small populations.
The message: expedited programs are tools, but evidence expectations remain.
We’re Failing
Susan McCune, MD, a pediatrician and neonatologist and former OPD director, gave her unvarnished perspective.
“I’m not going to be maybe as positive as everyone else is because we’re failing, just so you know, in the neonatal space.”
“We’re failing in neonatology.”
“We’re failing those patients, because we’re not thinking about the etiology of their diseases.”
Neonatal disease does not simply overlap with adult disease, she explained. She described a “new Venn diagram” in which common neonatal conditions and severe genetic disorders manifest differently and require different study approaches.
In prematurity, the problem is ontologic. Preterm birth interrupts normal organ development. The fetus is adapted to a hypoxic intrauterine environment; the NICU is comparatively hyperoxic.
Treating bronchopulmonary dysplasia with adult pulmonary paradigms ignores that the preterm lung may still be in the canalicular stage of development — structurally and functionally distinct from adult lung pathology.
In genetic disease, the most severe phenotypes frequently present in the neonatal period, yet those are often the populations excluded from early trials.
The closing remarks from Lin Yao, director of the Division of Pediatric and Maternal Health in CDER, reinforced McCune’s point:
“We are failing in neonatology. We are failing in diseases of the premature.”
She framed that failure as the reason the workshop existed: to identify how rare disease frameworks can be leveraged to correct course.
The failure described is more misalignment than disregard. The misalignment includes:
- Applying adult disease models to developing organs
- Designing trials that exclude the most severe phenotypes
- Using endpoints that do not reflect developmental biology
- Accepting evidentiary standards optimized for large populations
The charge was not to lower standards.
It was to build standards that fit the biology.
Architecture can be redesigned. Ontogeny cannot.
Advocacy as Infrastructure: HIE as Case Study
The neonatal hypoxic ischemic encephalopathy (HIE) community gave a powerful advocacy perspective.
“In severe and debilitating disorders with no approved therapies, the baseline risk of doing nothing is actually the most certain thing that we know,” said Betsy Pilon.
HIE presents a paradox: technically rare under FDA thresholds, yet globally common.
Advocates emphasized:
- Early engagement with sponsors
- Registry development
- Longitudinal follow-up beyond NICU discharge
- Harmonized common data elements
- Biomarker qualification for presymptomatic intervention
The argument was not emotional — it was operational.
Families often serve as the connective tissue across fragmented neonatal and rare-disease ecosystems. They can:
- Improve endpoint relevance
- Reduce dropout
- Inform feasibility
- Identify clinically meaningful change
Regulators echoed that they routinely ask sponsors whether proposed endpoints align with patient input.
This reinforces much of the recent FDA guidance on patient-centric trials, in that patient engagement is a vital component of clinical trial design.
Top Takeaways from Day 2
- Design innovation must be pre-specified and simulation-supported.
- External controls are viable — under strict evidentiary conditions.
- AI and digital biomarkers may compress development timelines — once validated.
- EMA and FDA frameworks differ structurally; global alignment is essential.
- Expedited pathways are powerful — but contingent on robust data.
- Advocacy organizations function as longitudinal infrastructure in small populations.
Taken together, ADEPT 10 did not suggest that neonatal and rare disease development is impossible.
It suggested that it is possible — but sponsors must rethink:
- Evidence thresholds
- Trial architecture
- Regulatory engagement timing
- Data harmonization
- Patient partnership
The science is advancing. The policy tools exist. The incentives are imperfect but evolving.
The question now is whether development strategy will evolve quickly enough to match the urgency articulated across both days.
Are you a biopharma company or CRO developing pediatric therapies? Would you like to share your thoughts and approaches with the broader life sciences community? Get in touch to talk about developing thought leadership that showcases your expertise.
