Related Experiment Video
Updated: Jan 30, 2026
Generation of Prostate Cancer Cell Models of Resistance to the Anti-mitotic Agent Docetaxel
Published on: September 8, 2017
A DREAM Challenge to Build Prediction Models for Short-Term Discontinuation of Docetaxel in Metastatic
Fatemeh Seyednasrollah1, Devin C Koestler1, Tao Wang1
1Fatemeh Seyednasrollah and Laura L. Elo, Turku Centre for Biotechnology; University of Turku; Åbo Akademi University, Turku, Finland; Devin C. Koestler, University of Kansas Medical Center, Kansas City, KS; Tao Wang, University of Texas Southwestern Medical Center, Dallas, TX; Stephen R. Piccolo, Brigham Young University, Provo; University of Utah, Salt Lake City, Utah, UT; Roberto Vega, Russell Greiner, and Luke Kumar, University of Alberta; Alberta Innovates Centre for Machine Learning, Edmonton, Alberta, Canada; Christiane Fuchs, Helmholtz Zentrum München, Neuherberg; Technische Universität München, Garching, Germany; Eyal Gofer, The Hebrew University, Jerusalem, Israel; Russell D. Wolfinger, SAS Institute, Cary, NC; Kimberly Kanigel Winner and James C. Costello, University of Colorado, Anschutz Medical Campus, Aurora, CO; Chris Bare, Elias Chaibub Neto, Thomas Yu, Thea Norman, and Justin Guinney, Sage Bionetworks, Seattle, WA; Liji Shen and Fang Liz Zhou, Sanofi, Bridgewater, NJ; Kald Abdallah, AstraZeneca, Gaithersburg, MD; Gustavo Stolovitzky, IBM Research, Yorktown Heights; Howard I. Scher, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY; Howard R. Soule, Prostate Cancer Foundation, Santa Monica; Charles J. Ryan, University of California, San Francisco, CA; Christopher J. Sweeney, Dana-Farber Cancer Institute and Brigham and Women's Hospital, Harvard Medical School, Boston, MA; and Oliver Sartor, Tulane University, New Orleans, LA.
Predicting early docetaxel discontinuation in metastatic castration-resistant prostate cancer (mCRPC) is crucial. Machine learning models using clinical data can identify high-risk patients, improving clinical trial design and patient outcomes.
Area of Science:
- Oncology
- Clinical Trial Design
- Data Science
Background:
- Docetaxel offers survival benefits for metastatic castration-resistant prostate cancer (mCRPC) patients.
- A significant percentage of mCRPC patients discontinue docetaxel due to toxicity, posing a management challenge.
Purpose of the Study:
- To develop predictive models for early docetaxel discontinuation in mCRPC.
- To leverage crowd-sourced, open-data approaches for robust model development.
Main Methods:
- Analysis of comparator arms from four Phase III mCRPC clinical trials (2,070 patients).
- A DREAM Challenge was conducted using publicly available clinical features and outcomes.
- Models were trained by 34 international teams to predict treatment discontinuation within 3 months due to adverse events.
Main Results:
- Seven models demonstrated superior predictive performance based on area under the precision-recall curve.
- Identified high-risk and low-risk patient subgroups for early discontinuation.
- High-risk subgroup experienced double the rate of early discontinuation compared to the low-risk subgroup.
Conclusions:
- Routinely collected clinical features can effectively identify mCRPC patients likely to discontinue docetaxel due to adverse events.
- This approach establishes a benchmark for clinical trial design, potentially optimizing patient enrollment and statistical power.
- Successful international collaboration highlights the power of open data in advancing cancer research.
Related Concept Videos
01:30Dreaming
02:56Treatment Resistant Cancers
01:10Lucid Dreaming
Studies have shown...
01:19Limits with Oscillating Discontinuities
01:28Improper Integrals: Discontinuous Integrands
02:27Predicting Molecular Geometry