Creating external control arms and improving outcomes prediction in multiple myeloma through the use of synthetic data

Multiple myeloma (MM) is a plasma cell neoplasm and the second most prevalent hematologic cancer. It is incurable but can be controlled for years with appropriate disease management. The treatment armamentarium has improved substantially in recent years as have clinical outcomes, but genetic heterogeneity and high-risk features that affect more than 25% of patients at diagnosis represent continued areas of unmet needs. The SYNTHIA MM use case focuses on creating synthetic data to both 1) build external control arms for existing and future clinical trials which integrate genomic, clinical, demographic, and imaging variables, and 2) improve long-term outcome prediction and advance precision medicine. 

The Challenge 

Multiple myeloma (MM) is a complex blood cancer characterized by clonal plasma cell growth in the bone marrow and significant genetic diversity. High-risk features such as cytogenetic abnormalities, circulating plasma cells, and extramedullary disease impact outcomes and complicate treatment decisions.  Control arms in randomized clinical trials can be difficult to establish in later-line settings or among specialized patient subgroups. Additionally, even in cases where control arms are feasible, the rapid proliferation of treatment options and duration of trials can make it unlikely that a single control arm would meet the needs of all decisions makers. Leveraging synthetic data to create external control arms, integrating imaging, molecular, and clinical data, can help to address these challenges and overcome the barriers to precision medicine and timely access to effective treatments. 

Our Research Questions 

1. How effectively can synthetic data integrate genomic, clinical, demographic, and therapeutic variables?

2. How accurately can synthetic data replicate clinical and real-world outcomes observed among patients with newly diagnosed multiple myeloma (NDMM)?  

3. Can a synthetic external control arm reliably support regulatory and HTA decision-making for new MM treatments?   

Our Approach 

• Evaluate the accuracy of synthetic data in replicating clinical features and endpoints observed in clinical trials and real-world data to determine the suitability and robustness of synthetic control arms for decision making.
• Investigate the impact of rare high-risk genetic alterations on predicting clinical endpoints through synthetic genomic data generation.  

The main data modalities:

• Clinical/EHR: demographics, medical/family history, subtype, lab tests (serum/urine monoclonal component, creatinine, calcium, LDH), cytogenetics, flow cytometry, MRD, therapy details, clinical notes. 
• Imaging: PET-CT, bone and extraosseous lesions. 
• Genomics: FISH, DNA sequencing, CNV, ctDNA. 

Envisioned Impact 

The MM use case is expected to demonstrate the value of synthetic data by providing external control arms for clinical trials, particularly in settings where randomized controls are impractical or unethical. This approach could accelerate access to novel therapies while supporting decision-making by regulatory bodies, HTAs, healthcare providers, and patients. The expected direct impacts are to highlight the utility of synthetic data in advancing the development and applicability of predictive models, contributing to medical research, clinical practice, and regulatory evaluation. This collaborative initiative holds the promise of optimizing the design and outcomes of clinical trials and advancing treatments for MM. 

Use Case Leadership