Developing Estimand Frameworks in Registrational Clinical Trials for Gene Therapies and Rare Diseases

Yaohua Zhang (Vertex), Bingming Yi (Vertex), Lanju Zhang (Vertex)

With the introduction of the International Council for Harmonization (ICH) E9(R1) guideline in 2019, estimand frameworks (EFWs) have provided a structured approach to defining precise trial objectives. The U.S. Food and Drug Administration (FDA) has actively promoted EFW adoption through working groups, training initiatives, and integration into investigational new drug (IND) review templates.

While progress has been notable—mentions of “estimand” in submissions have grown exponentially from 9,000 between 2014 and 2019 to 50,000 between 2019 and 2024. Several challenges persist, especially in trials for rare diseases. Furthermore, to authors’ best knowledge, discussion on implementation of EFWs in cell and gene therapies (CGTs) and rare diseases is particularly limited. With the help of three speakers (Yunfan Deng from FDA, Pengyu Liu from Vertex Pharmaceuticals, Xian Sun from Regeneron), and one discussant (Lihan Yan from FDA), this session offered an in-depth examination of how EFWs are shaping clinical trial design and regulatory evaluations for cell and gene therapies (CGTs) and rare diseases.

Key Challenges in CGTs and Rare Disease Trials

Rare diseases, defined as conditions affecting fewer than 200,000 individuals in the U.S., present unique challenges in clinical trial design. Ultra-rare diseases, with patient populations in the hundreds or thousands, add further complexity. The natural history of these conditions is often poorly understood, and clinical presentations are highly heterogeneous. This variability complicates the selection of efficacy endpoints and handling of intercurrent events (ICEs) such as rescue treatments, adverse events, or deaths.

Small patient populations necessitate innovative trial designs, such as single-arm studies, which often rely on external controls or historical benchmarks. However, these approaches require rigorous justification, particularly when aligning trial objectives with statistical methods. ICEs, for instance, can introduce bias if handled conservatively, as is common with non-responder imputations.

CGTs, which often provide one-time treatment for genetic or cellular defects, amplify these challenges. Trials frequently involve binary endpoints (e.g., responder/non-responder) and single-arm designs. This makes defining estimands particularly complex, as it demands clarity in identifying the target population, analyzing treatment effects, and handling ICEs.

The session highlighted key strategies for addressing ICEs in CGTs, such as composite strategies (considering subjects with ICEs as non-responders) or hypothetical strategies (estimating treatment effects assuming ICEs did not occur). Simulation studies demonstrated the value of multiple imputation (MI) methods for handling missing data, showing superior performance in terms of bias and power compared to traditional methods.

Progress, Collaboration and Implications for Practice

Despite challenges, the adoption of EFWs is growing steadily. Early engagement with sponsors during study design stages fosters alignment on trial objectives, particularly in defining estimands and managing ICEs. Collaborative efforts, including international working groups and disease-specific guidance, are helping to standardize practices. Regulatory agencies and industry stakeholders are increasingly incorporating EFWs into protocols for diseases like oncology, CNS disorders, and chronic conditions.

The session emphasized the importance of clinician involvement. Clinicians play a critical role in identifying ICEs and defining appropriate handling strategies. Targeted training for clinicians and statisticians has improved collaboration and understanding of EFWs across therapeutic areas.

The session also underscored the need for proactive planning and clear communication in trial design. For CGTs and rare disease trials, early discussions on estimand definitions and ICE handling strategies can significantly improve regulatory submissions and outcomes. Leveraging advanced statistical methods, such as MI and simulations, can address challenges like missing data and small sample sizes.

Collaboration remains key. Ongoing dialogue between clinicians, statisticians, and regulatory bodies will help refine estimand frameworks and enhance their application across diverse therapeutic areas. Furthermore, expanding disease-specific guidelines and leveraging machine learning tools for data analysis could further streamline implementation.

Conclusion and Future Directions

The development of estimand frameworks represents a paradigm shift in clinical trial methodology. While challenges persist, particularly for rare diseases and CGTs, the progress made through regulatory and industry collaboration is promising. Future efforts should focus on enhancing education and training, standardizing practices globally, and fostering innovation in trial design.

This session highlighted the critical role of estimand frameworks in advancing treatments for complex conditions, offering valuable insights into both the potential and limitations of current approaches. As research in rare diseases and CGTs continues to grow, EFWs will remain an essential tool in bridging clinical objectives and statistical rigor.