Project Data Sphere® (PDS), an independent, non-profit initiative of the CEO Roundtable on Cancer, was founded in 2014 with the goal of accelerating cancer research by sharing de-identified patient-level data from completed cancer clinical trials. PDS operates through two main mechanisms: an open-access platform and closed-access precompetitive research initiatives.
The open-access platform provides a secure, online environment for researchers from academia, industry, and government agencies to access, analyze, and share historical clinical trial data. This platform promotes collaboration and allows for the aggregation and utilization of large datasets that would otherwise be difficult to access, facilitating the discovery of new insights and advancing the understanding of cancer biology, treatment response, and patient outcomes.
Concurrently, PDS conducts closed-access precompetitive research initiatives that bring together experts from various disciplines to address specific challenges in cancer research and treatment. These initiatives leverage advanced methods based on artificial intelligence (AI), machine learning (ML), and sophisticated analytical tools, to explore novel approaches to drug discovery, biomarker identification, and treatment optimization. By employing these technologies, PDS aims to identify new therapeutic targets, develop predictive models for treatment response, and ultimately improve patient care.
A key aspect of PDS's closed-access initiatives is the development of large AI models capable of integrating and analyzing vast amounts of data from multiple sources, including clinical trials, electronic health records, and genomic databases. These models have the potential to uncover fundamental insights into cancer biology and identify novel patterns and associations that may inform the development of new diagnostic and therapeutic strategies. Project Data Sphere's initiatives showcase the organization's commitment to advancing the use of cutting-edge technologies and fostering collaboration between academia, industry, and regulatory bodies.
Images and Algorithms (I&A): Machine Learning in Medical Imaging
Project Data Sphere (PDS), in collaboration with the US Food and Drug Administration (FDA), is harnessing the power of AI and ML to advance medical imaging. By leveraging vast imaging datasets, the I&A program aims to streamline workflows, improve efficiency, reduce costs, and enhance precision in across various medical imaging modalities.
Radiology
Project Data Sphere is developing advanced imaging biomarkers that combine the latest advances in radiomics and AI to enhance the utility of radiographic examination of tumor dynamics in cancer clinical trials and patient care. The project involves creating automated tools for detecting, segmenting, and measuring lesions using radiomics, an emerging field that involves the extraction of quantitative features from medical images, such as CT scans, MRI scans, and PET scans. These radiomic features, which include intensity-based, shape-based, and texture-based features, can be used to characterize the properties of tissues or lesions within the images.
AI algorithms, such as machine learning and deep learning, are being employed to analyze these radiomic features and develop predictive models for diagnosis, prognosis, and treatment response assessment. The combination of radiomics and AI has the potential to significantly improve the accuracy and efficiency of image analysis in clinical trials and patient management.
The automated tools developed by PDS will support participating member companies in performing standard assessments such as those based on RECIST 1.1 while enabling the development of next-generation imaging biomarkers for total tumor burden and tumor growth kinetics. These tools leverage DICOM images, the standard format for storing, transmitting, and viewing medical images, which ensures the consistency and reproducibility of image data across different studies and institutions.
Histopathology
Project Data Sphere is developing innovative digital imaging biomarkers to optimize the application of pathology in clinical trials and patient care. Digital histopathology involves the digitization of traditional glass slides containing tissue samples using specialized scanners, creating high-resolution digital images that can be viewed, analyzed, and shared electronically.
In partnership with academic institutions and member companies, PDS is working to create a comprehensive solution for the discovery, development, validation, and deployment of novel digital histopathologic biomarkers. These biomarkers are quantitative features extracted from digital histopathology images using computational algorithms, providing objective and reproducible measurements of various tissue characteristics, such as cell morphology, tissue architecture, and the presence of specific proteins or molecular markers. The digital biomarkers will be utilized in clinical trials for patient selection and treatment monitoring, as well as in routine patient care.
The use of digital histopathology and digital biomarkers has the potential to improve the accuracy, efficiency, and reproducibility of pathological assessments, ultimately leading to better patient care and precision therapeutics. Project Data Sphere and the US FDA are organizing a digital histopathology symposium scheduled for spring 2025 to further advance these efforts and discuss the latest developments in this rapidly evolving field.
Immunotherapies
The Immune-related Adverse Events (irAEs) Initiative was inaugurated in 2019 at a symposium co-sponsored by Project Data Sphere and the US FDA. The primary objective of this initiative is to improve the clarity and classification of adverse events associated with immunotherapies. The impetus for this initiative stemmed from observations of rare, life-threatening events, such as myocarditis, in patients undergoing checkpoint inhibitor therapy. As more case reports emerged, it became evident that checkpoint inhibition had the potential to activate the immune system in any organ or system. Notably, patients who experience these immune-mediated adverse reactions often exhibit superior anti-tumor responses, which complicates clinical management decisions.
PDS's efforts in this area have identified gaps in data recording and have led to significant contributions, including the introduction of new Medical Dictionary for Regulatory Activities (MedDRA) terms for clinical development reporting and new International Classification of Diseases, 10th Revision (ICD-10) codes. PDS is currently collaborating with a network of physicians, patients, and patient advocates to further advance this work and refine the understanding and management of irAEs.
External Control Arms
Project Data Sphere has successfully developed external control arms for clinical trials in two rare cancer types: small cell lung cancer (SCLC) and glioblastoma (GBM). By leveraging historical data to supplement or replace control arm data, PDS has demonstrated the value of this approach when traditional randomized designs are impractical or unfeasible.
PDS's work in SCLC was showcased at the US FDA in 2019, highlighting the potential of external control arms. Building upon this success, PDS developed an external control arm for newly diagnosed glioblastoma, which has enhanced interim decision-making in the GBM-INSIGhT study, a Bayesian adaptive platform trial.
The external control arms developed by PDS for these two cancer types exemplify the organization's commitment to advancing innovative trial designs and utilizing historical data to support the development of new therapies, especially in rare diseases. This approach serves as a model for future clinical trials, emphasizing the importance of alternative trial designs and the utilization of historical data in advancing the development of safe and effective treatments for patients in need.
Global Oncology Big Data Alliance (GOBDA)
A federation of industry stakeholders collaborating to support 21st-century biomedical research. Participating companies and organizations share knowledge, data, and advanced analytical tools to accelerate the translation of research innovations into safe and effective cancer treatments.