Real-World Data and AI Transform Rare Disease Clinical Trial Design Through Synthetic Control Arms
核心洞察
Industry experts at the Clinical Trials in Rare Diseases Europe 2025 conference highlighted how synthetic control arms using real-world data can overcome ethical and feasibility challenges in rare disease trials.
The FDA has shown acceptance of synthetic control arm approaches, exemplified by the approval of Bavencio (avelumab) for metastatic Merkel cell carcinoma (搜索) based on single-arm trials with real-world comparators.
Artificial intelligence integration could enhance synthetic control arm matching on prognostic factors, accelerate enrollment timelines, and reduce trial costs while allowing more patients to receive active treatment.
Pharmaceutical companies are increasingly turning to innovative trial designs that leverage real-world data to address the unique challenges of conducting clinical research in rare diseases, where traditional randomized controlled trials may be unfeasible or unethical due to limited patient populations.
Speaking at the Clinical Trials in Rare Diseases Europe 2025 conference in Munich, Germany, on November 26-27, Channa Debruyne, global clinical development lead for late-stage and LCM oncology at Servier, outlined how real-world data from similar patient populations can serve as external comparators in synthetic control arms. "The data should preferably be at the patient level and populations should be matched in terms of key demographics and prognostic baseline variables," Debruyne explained.
Regulatory Acceptance and Precedent
The approach has gained regulatory traction, with the US Food and Drug Administration indicating acceptance of synthetic control arm data for rare diseases. A notable example is Bavencio (avelumab), co-developed by Merck KGaA and Pfizer, which received approval for treating metastatic Merkel cell carcinoma (搜索) based on single-arm trials using real-world data as an external comparator.
Several drugs have recently been approved using this methodology, though Debruyne emphasized this represents "just the beginning" of broader adoption across the industry.
Data Quality and Harmonization Challenges
The success of synthetic control arms depends heavily on data quality and standardization. Debruyne stressed the importance of "standardised and robust data collection and storage, as well as careful management of potential biases" in developing these external comparators.
A significant challenge lies in harmonizing data from multiple registries with different parameters, which requires close collaboration among stakeholders. This represents "currently one of the biggest challenges in this space," according to Debruyne.
Clara Cali Mella, data strategy lead at Bayer, noted that while "the data is there," the main obstacles are access and regulatory barriers. She highlighted the complexity of navigating data protection regulation laws while encouraging greater willingness to share anonymized data among data owners.
Finland was cited as a positive example of a region openly sharing anonymized patient data, with Mella calling for more countries to adopt similar approaches, describing this as potentially "a game changer for rare diseases."
AI Integration and Future Optimization
Artificial intelligence is emerging as a transformative tool for enhancing synthetic control arm development. Eslam Katab, global clinical development manager at Sandoz, described how AI can improve the process of leveraging historical real-world data by ensuring appropriate matching on key prognostic factors and applying propensity weighting.
The integration of AI technology could deliver multiple benefits, including accelerated enrollment timelines, reduced trial costs, and enabling more patients to receive active treatment rather than placebo. These improvements address critical concerns in rare disease research where patient populations are inherently limited.
Strategic Recommendations for Industry
Looking ahead, experts emphasized the need for expanded high-quality patient registries and strengthened partnerships across the healthcare ecosystem. Debruyne called for enhanced "interactions between the clinicians, researchers, non-profit organisations, regulatory bodies, patient associations, and industry."
The convergence of real-world data utilization, regulatory acceptance, and AI-enhanced methodologies represents a significant evolution in rare disease clinical trial design, potentially accelerating drug development timelines while maintaining scientific rigor and patient safety standards.