Winners have been announced for the Prostate Cancer DREAM Challenge, the first research challenge in prostate cancer to marry crowdsourcing with data sharing, paving a new way to tackle key research questions about metastatic castration-resistant prostate cancer (mCRPC), an advanced form of the disease with poor outcomes.
The challenge called upon the cancer research and computational biology community to find solutions to key open clinical research questions about mCRPC and explore innovative research and modeling approaches. The three specific questions posed were to:
Predict overall survival for prostate cancer patients using clinical data.
Predict the exact time to event for prostate cancer patients using clinical data.
Predict treatment discontinuation for prostate cancer patients treated with docetaxel.
The challenge produced unprecedented levels of participation, with more than 550 registrants, comprising more than 60 teams. Three rounds of leader-board scoring allowed teams to submit up to five predictions per round; in total, nearly 1,200 model predictions and almost 160 final submissions were received across two sub-challenges.
Best performers include Team FIMM-UTU from the University of Turku, Finland, and the University of Helsinki, Finland, for the first question regarding prediction of overall survival; and Yuanfang Guan from the University of Michigan for the second question regarding prediction of treatment discontinuation.
“The DREAM Challenge was rewarding in that it enabled us to participate in a collaborative project in which researchers from multiple institutes worked together and compared findings with other teams,” said Tero Aittokallio, a member of Team FIMM-UTU. “The most rewarding part, however, will be to see how our findings will eventually help patients with metastatic castrate resistant prostate cancer.”
Challenge winners and results can be found on the Prostate Cancer DREAM Challenge homepage. The winners will be invited to co-author a challenge overview paper that will be submitted for peer review to Nature Biotechnology. Winners also will each receive a share of an educational award from Project Data Sphere, sponsored by AstraZeneca. Solvers will be able to submit write-ups on their models to F1000Research, an open science publishing platform, where the models can be reviewed.
“More than 550 researchers from around the world accessed the data over the course of the challenge,” said Dr. James Costello, University of Colorado Anschutz Medical Campus. “Without the efforts from Project Data Sphere, Sage Bionetworks with the support and experience of the DREAM Challenges Initiative, and our clinical trial collaborators, these data would have been difficult to access and even harder to use. Project Data Sphere performed a great deal of data curation to make the clinical variates readily accessible.”
Data for the challenge was standardized and integrated from four different, de-identified clinical trials with more than 2,000 mCRPC patients treated with first-line docetaxel. The data sets were provided to Project Data Sphere by AstraZeneca, Celgene, Memorial Sloan-Kettering Cancer Center and Sanofi U.S.
