This content introduces advanced statistical and regulatory considerations involved in designing rare disease clinical trials. It outlines best practices for innovative trial designs, endpoint strategies, and the use of real‑world or historical data to support regulatory acceptance.
Introduction
Designing clinical trials for rare diseases presents a unique set of statistical challenges that require thoughtful, tailored approaches. Unlike trials for common conditions, rare disease studies must contend with small, geographically dispersed patient populations, highly variable disease courses, and ethical constraints. The complexity of gene therapies adds another layer, with one-time interventions that require long follow-up periods and potential immune responses that can affect both efficacy and safety outcomes.
With extremely limited patient numbers, traditional statistical methods often fall short. Low enrollment reduces statistical power and robustness, while high inter-subject variability complicates interpretation. On the other hand, gene therapies often are more targeted to the disease pathophysiology and are expected to translate into substantial and durable treatment effect. Recognizing these challenges and opportunities, FDA guidance encourages sponsors to pursue innovative trial designs and engage early with regulators to align on expectations [1].
Best Practices for Trial Design
Within rare diseases, prevalence can still vary widely. When RCT, with novel adaptation to the product/indication context, is feasible, RCT is preferred for its strength in allowing valid causal inference. When RCT is not feasible, other study designs should strive to preserve valid causal inference and adequately address potential biases and quantify degree of uncertainty. Recommendation is to consult FDA guidance documents and early engagement with FDA.
To address the limitations of small patient populations, early and frequent engagement with the FDA is critical. This allows sponsors to set realistic expectations around novel designs, target population, data collection, endpoint selection, and sample size. Inclusion criteria should be carefully crafted to maximize effect size while maintaining subject safety, generalizability, balancing of benefits and risks, and accounting for disease heterogeneity. In some cases, longer observation periods or more frequent data collection within shorter timeframes can help increase the number of observations per patient, boosting statistical power without expanding cohort size. Yet, this must be balanced with the burden on the subject, particularly in pediatric populations. It is important to understand that increased sampling frequency or longer observation does not inherently improve statistical power. These benefits only occur when the biological trajectory changes sufficiently within the assessment window and the measurements are feasible and interpretable.
Additionally, Rare diseases may be designated an expedited review pathway depending on several factors). It is encouraged that the sponsor discusses their objective with the FDA and gets the regulatory framework approved. Evidence-based decisions about what is feasible in terms of rare disease drug clinical investigation enrollment depend on accurately estimated disease prevalence. The FDA recommends determining prevalence estimates for countries in which clinical investigation sites are being considered to inform available sample sizes. If prevalence estimates are anticipated to vary across countries, then, evaluation of the potential differences in prevalence estimates should be performed and presented to FDA.
Broader input into trial design by leveraging patient / family feedback, disease registries, patient advocacy groups, and investigator subject matter experts can help with justification of design decisions. For some indications, novel approaches can utilize real world-data / evidence to further minimize the need for a concurrent control, thus creating efficiencies and improving timelines for rare disease trials.
Sponsors may also consider integrating trial phases, such as combining Phases II and III, to maximize data collection from each patient, reduce the overall number of subjects to enroll, and potentially minimize timelines to achieving confirmatory data [2]. A well-designed seamless or controlled trial can allow early signal detection, dose refinement, and immediate transition into confirmatory evidence generation within a single protocol, without the downtime and inefficiencies of separate phases and protocols. Statistically, these designs rely on prespecified decision rules (e.g., go / no-go, dose selection, interim adaptability), alpha-spending plans, and simulation-based evaluation of operating characteristics to ensure Type I error control.
In diseases with heterogenous phenotypes, creating multiple cohorts based on distinct phenotypes may help ensure meaningful analysis and interpretation but may not always be feasible [2]. Novel approaches in which subjects can act as their own control, or where natural history studies are used to support comparator arm are another option within rare diseases with variable presentations. These adaptive and novel approaches should be considered to further explore treatment effect over time, in an efficient manner.
Factors such as feasibility, biological plausibility and strength of natural history comparators play a large role in rare diseases. There may be some history of FDA relying more heavily, on a case-by-case basis, on descriptive analyses, trajectory-based comparisons, and comparisons with external controls than on statistical hypothesis testing. It is recommended to discuss the overall plan for evidence generation with the FDA before initiating a pivotal trial.
Clinical Trial Endpoints
Selecting appropriate endpoints and determining the timing of their assessment are among the most critical decisions in rare disease trials. Endpoints must be sensitive, clinically meaningful, and feasible within the constraints of small sample sizes and heterogeneity of patient populations.
In cases where using clinical endpoints is not feasible because changes in symptoms and disease status occur too slowly to be measured in a clinical investigation of reasonable duration, surrogate endpoints may be considered [3]. These endpoints represent biological or functional measures that are not the ultimate clinical outcome but are strongly correlated with it.
Per BEST Glossary, surrogate endpoints fall into the following categories:
Validated surrogate – strongly predicts clinical benefit (e.g., HbA1c for diabetes). These are very typical biomarkers.
Reasonably likely surrogate – mechanistic rationale but limited data; may support accelerated approval (e.g., tumor shrinkage).
Candidate surrogate – under evaluation; not yet suitable for approval decisions.
Intermediate clinical endpoint – a clinical measure reasonably likely to predict benefit (e.g., irreversible morbidity components)[4].
Endpoint selection in rare disease clinical investigations needs to estimate the magnitude of clinically meaningful benefit. A recommended strategy from FDA is to consider early development work on biomarkers as surrogate endpoints that may support approval (either for traditional or accelerated approval). Hence, initial evaluation of the literature to identify such biomarkers, early work performed on translational animal models, and leveraging data from natural history cohorts before initiation of clinical development is essential. Clinical data in patients with the disease, and clinical pharmacodynamic (PD) data from early clinical investigations with the drug, can contribute to substantiate the use of the proposed biomarker as a surrogate. A well-developed strategy and initial information on a proposed surrogate endpoint are required prior to initial discussions with FDA (e.g., pre-IND)
Biomarkers can serve as indicators of disease progression or therapeutic response but must be specific to the condition being studied. Accelerated approval pathways may allow for endpoints that are “reasonably likely” to predict clinical benefit, or surrogate endpoints that may correlate with an endpoint meant to predict clinical outcomes. In these cases, rigorous validation remains essential, and decisions should be justified and supported by evidence [5].
Patient focused drug development initiatives aim to incorporate measurements that are important to patients. These include patient and observer / caregiver-reported outcomes, such as pain levels or quality of life, which are gaining importance, especially with increased involvement from patient advocacy groups (PAGs) [4]. These endpoints should reflect how a patient feels, functions, or survives but may require a more objective anchor such as mortality. This provides a mechanism that is useful in clinical trials for rare diseases, particularly indications with limited or no objective measurements, high-heterogeneity, or very slow progressing diseases requiring very long assessment periods.
Composite endpoints can also be useful in increasing event counts across multiple outcomes [3]. An example is Major Adverse Cardiovascular Events (MACE) which combines CV death, MI, and stroke. For rare diseases, composites may combine respiratory failure with functional decline (e.g., ppFVC drop). Ensure components are clinically meaningful and move in the same direction; report individual component results to avoid misleading interpretation. Yet, statistical considerations such as multiplicity and heterogeneity across components can introduce complexity and potential penalties in hypothesis testing and regulatory review when working with composite endpoints. Careful planning and justification of component selection are therefore essential to ensure meaningful interpretation and regulatory acceptability.
Statistical approach for rare disease trials
Individual vs. population-level analyses: In rare and highly heterogeneous conditions, patient-level trajectories can be more informative than aggregate means. Consider mixed models or Bayesian approaches that borrow strength across patients while preserving individual patterns.
Absolute vs. percentage change: Define upfront whether efficacy will be assessed by absolute change (e.g., +5 points on a scale) or percentage change from baseline, as this impacts interpretability and comparability across patients.
Establishing a robust baseline is critical. Strategies include:
Lead-in periods to prospectively collect multiple pre-dose measurements.
Leveraging medical records and natural history data to define a patient’s trajectory before dosing.
For slowly progressive diseases, analyzing the slope of decline or improvement (rather than a single timepoint change) can detect meaningful differences. This is particularly valuable in neurodevelopmental and rare conditions.
When realized benefit may emerge over years rather than within a trial window, consider:
Incorporating longitudinal modeling and extrapolation methods.
Using external controls or natural history cohorts to contextualize observed changes.
Planning for post-marketing follow-up to confirm durability of effect.
Because no single endpoint can capture the full spectrum of disease impact, sponsors should consider including secondary clinical endpoints and ensure robust safety monitoring throughout the trial. Exploratory endpoints, particularly innovative / novel biomarkers or digital tools, can be useful to incorporate early in designs to progress science and have ability to be utilized in a more robust way supporting data package confirmatory evidence.
Furthermore, FDA recommends including Auxiliary cohorts to augment the safety and efficacy database if the data is rigorously collected and analyzed [6]. That is for a clinical investigation protocol a safety cohort running parallel to the efficacy cohorts. This cohort would include patients with the disease who investigators think might benefit from the investigational drug but who do not meet all the registration clinical investigation eligibility criteria. Such patients can be enrolled in clinical investigation, avoiding the need for a separate clinical investigation and protocol. However, these patients are not randomized and are excluded from the efficacy analysis.
Pro tip:
When possible, continuous or quantitative measures and longitudinal data should be prioritized to enhance statistical sensitivity.
Leveraging Historical Controls and Real-World Evidence
FDA recognizes that for diseases that are very rare or have very slow and variable progression over years, the use of clinical endpoints may be challenging. In these situations, several strategies may be considered, such as using data from natural history or registry-based studies, to identify clinically relevant changes that are most prominent and most rapidly progressive that could serve as the basis for a clinical endpoint [6].
Hence, given the inherent limitations in rare disease trial populations, external data sources can play a vital role in contextualizing results. Historical controls, natural history studies, and real-world evidence (RWE) can supplement single-arm trials and help establish baseline trajectories. Robust natural history data can also help distinguish drug-related adverse events from underlying disease manifestations. Robust observational data can inform endpoint selection and provide meaningful comparators when placebo use is not feasible. When using historical data, it’s essential to match trial and control cohorts on key characteristics such as genotypes, disease stage, outcome assessment windows, and other potential heterogeneities to ensure the comparability of the populations [7]. Using historical data can be complex, it is advised to consult a statistician when considering its incorporation into your trial design.
Existing data and registries (academic, PAG supported, or other sources) may already maintain long-term natural history datasets that can be leveraged for this purpose. Collaborations with organizations that support disease registries or patient groups (such as National Organization for Rare Disorders (NORD)) can provide access to valuable data and patient networks. As always, early discussion with the FDA is recommended to align on the use and acceptance of external controls and RWE within the design of the clinical trial and statistical analysis plan (SAP). Additionally, alternative models such as a crossover design can help increase statistical significance when working in small patient populations.
Adaptive and Innovative Trial Designs
Identifying early biomarkers of disease or of intervention effects and biomarkers that could be used in adaptive and enrichment designs can enhance efficiency. For example, values of a laboratory measurement expected to be sensitive to a drug’s effect could be used to screen potential responders for inclusion in efficacy clinical investigations.
To address patient and family concerns on the study design, adaptive designs with interim analysis can be considered by using modified clinical investigation designs to demonstrate effectiveness and identify important safety signals early in clinical development.
Additionally, when planning to use innovative clinical investigation designs, it is important to discuss in advance with the review division, ideally at the pre-IND application meeting. Some examples of innovative or non-traditional approaches in rare diseases include Bayesian methods, n-of-1 clinical investigations, randomized delayed-start designs, and crossover designs. For example, Bayesian methods to maximize the use of information gleaned from early-phase studies or natural history studies. Bayesian methods may also inform pediatric clinical investigations through incorporation of adult clinical data.
In January 2026, the FDA issued draft guidance promoting Bayesian methods. The guidance provides recommendations on the appropriate use of Bayesian methods, with an emphasis on the use of these methods to support primary inference. Bayesian methods may be especially valuable for sponsors targeting rare or pediatric indications, where patient populations are smaller. This approach is particularly important for rare diseases were incorporating prior information (e.g., natural history studies) augments the trials’ size and hence improve clinical trial efficiency, allowing for smaller, more flexible, and faster studies to be performed; hence accelerates drug development and reducing patient burden and cost [8].
If an adaptive clinical investigation design is under consideration, a detailed SAP, including the key features of the clinical investigation design and preplanned analyses (including interim analyses), may need to be discussed with the review division before clinical investigation initiates.
In cases where recruitment is particularly challenging, adaptive trial designs may offer a flexible alternative. This provides an efficient way to expand the eligible population and generate broader insights while maximizing the data captured within a rare patient population and maintaining scientific rigor.
Conclusion
Designing and planning for rare disease trials demands creativity, precision, and proactive regulatory engagement. Sponsors should prioritize early and frequent dialogue with the FDA to align trial design, endpoint selection, and control strategies. By embracing tailored methodologies and leveraging external data sources, developers can overcome the inherent challenges of rare disease research and accelerate the path to a meaningful clinical impact. Ultimately, robust statistical justification, framed within the feasibility constraints of small populations and complex endpoints, often forms the foundation for regulatory acceptance of gene therapy for rare diseases.