Project Data Sphere has amassed data on nearly 30,000 patients over the past 2 years across many types of cancer. An independent not-for-profit initiative, the project is an online resource of comparator-arm clinical trial data provided by pharmaceutical companies and research institutions to spur drug discovery and development.

Project Data Sphere (PDS) has amassed data from nearly 30,000 patients since 2014, providing a valuable resource for cancer drug discovery and development ( The project, an online trove of clinical trial data supplied by pharmaceutical companies and research institutions, most recently added data from the PROCLAIM trial on etoposide, cisplatin, and radiotherapy in patients with non–small cell lung cancer (J Clin Oncol 2016;34:953–62).

An independent not-for-profit initiative of the CEO Roundtable on Cancer's Life Sciences Consortium, the PDS platform was designed to help cancer researchers access and share comparator-arm data from phase III clinical trials, says PDS chief executive officer Martin Murphy, DMedSc, PhD. The data and accompanying analytic tools are freely available to the program's 1,200 registered users.

“PDS is like a digital laboratory,” he says. “It's not just lumps of data but curated, coherent information that can be aggregated and analyzed.”

PDS's focus on clinical trial data differentiates it from similar efforts, such as the American Association for Cancer Research's Project GENIE and the American Society of Clinical Oncology's CancerLinQ, says Stephen Friend, MD, PhD, president and co-founder of Seattle, WA–based Sage Bionetworks, and an advisor to PDS.

Project GENIE currently includes seven institutions that have agreed to collect and share patients' deidentified, clinical-grade sequencing data that have been used to make treatment decisions, says Friend, who helped launch the initiative in 2015. In contrast, CancerLinQ aids clinicians with decision-making, he says. It amasses data from patients' electronic health records to reveal patterns and identify the best evidence-based therapies.

PDS data could help researchers better understand and predict disease progression and treatment toxicity, says Murphy. For example, the data were used recently in the Prostate Cancer DREAM Challenge, which aimed to improve the prediction of survival and potential toxicity of docetaxel chemotherapy in patients with metastatic castrate-resistant prostate cancer (mCRPC). Results from the challenge, which attracted more than 50 teams of almost 600 cancer researchers and computational biologists, are pending publication.

According to Murphy, the following are potential uses for PDS data:

  • Prognostic modeling. Pooled data from the control arms of multiple studies can model the effect of pretreatment factors, such as overall survival (OS), on endpoints. For example, researchers created a prognostic model of OS for patients with mCRPC who had received docetaxel, using patient data available through PDS (J Natl Cancer Inst 2013;105:1729–37).

  • Comparative effectiveness research. Combining data from many trials allows for more in-depth analyses of patient subgroups. For example, one study compared mitoxantrone plus prednisone with prednisone alone in advanced mCRPC, concluding that mitoxantrone confers no significant survival benefit for patients who have received docetaxel (Oncologist 2015;20:516–22). The large data pool also allows researchers to determine how the dose, frequency, and duration of therapy may affect outcomes, such as survival and side effects.

PDS's industry collaborators include Amgen, AstraZeneca, Bayer, Celgene, EMD Serono, Janssen, Lilly, Millennium Pharmaceuticals, Pfizer, Sanofi, and Synta Pharmaceuticals. Institutional participants include Memorial Sloan Kettering Cancer Center and the NCI's National Clinical Trials Network cooperative group, the Alliance for Clinical Trials in Oncology.

In addition to prostate and lung cancers, the platform contains data on blood and other cancers, such as breast, kidney, and pancreatic, to name a few.

Organizers want to continue to expand and diversify the types of data they collect, says Friend. “There might be other tumor types that make sense to share.” –Janet Colwell

For more news on cancer research, visit Cancer Discovery online at