Not Logged In

PapersDB

Prediction of overall survival for patients with metastatic castration-resistant prostate cancer: development of a prognostic model through a crowd-sourced challenge with open clinical trial data.

• J Guinney
• T Wang
• Prostate Cancer Challenge DREAM Community
• Russ Greiner, Dept of Computing Science; PI of AICML
• Roberto Vega
• Luke Kumar, University of Alberta (Computing Science)
• Jaykumar Patel

Abstract: Background Improved prognostic models in metastatic castration-resistant prostate cancer (mCRPC) have the potential to improve clinical trial design and to guide treatment strategies. In partnership with Project Data Sphere, LLC (PDS), a not-for-profit initiative allowing cancer clinical trial data to be shared broadly with researchers, we designed an open-data, crowdsourced DREAM (Dialogue for Reverse Engineering Assessments and Methods) Challenge to both identify a better prognostic model and engage a community of international data scientists to study mCRPC. Methods Comparator arm data from four phase III clinical trials in first-line mCRPC were obtained from PDS, including 476 patients treated with docetaxel and prednisone from the ASCENT2 trail, 598 patients treated with docetaxel, prednisone/prednisolone, and placebo in the VENICE trail, 526 patients treated with docetaxel, prednisone, and placebo in the MAINSAIL trial, and 528 patients treated with docetaxel and placebo in the ENTHUSE 33 trial. Datasets comprising over 150 clinical variables were centrally curated, including demographics, lab values, medical history, lesion measures, and prior therapies. Data from ASCENT2, VENICE, and MAINSAIL were released publicly to be used as training data to predict the outcome of interest, overall survival (OS). ENTHUSE 33 was used for independent model validation. The outcome variables in ENTHUSE 33, OS and event status, were hidden from the model builders, thus representing an unbiased and rigorous evaluation. Methods were evaluated using the integrated time-dependent area under the curve (iAUC). The Reference model, based on eight clinical variables and a penalized Cox's proportional hazard model, was used to compare method performance. Further validation was done using a fifth trial, ENTHUSE M1, where 266 patients were treated with placebo alone. Findings A total of 50 independent models were developed to predict OS and evaluated through the Challenge. The top-performer, based on an ensemble of penalized Cox regression models (ePCR) uniquely identified predictive interaction effects with immune biomarkers and markers of hepatic and renal function. Overall, it significantly outperformed all other methods (iAUC=0·791) and surpassed the Reference model (iAUC=0·743). The ePCR model significantly stratified the ENTHUSE 33 patients (p = 2·4e-14; HR=3·32 (95% CI 2·39-4·62)) and outperformed the Reference model (p = 1·4e-9; HR=2·56 (95% CI 1·85-3·53)). The new model was validated further on the ENTHUSE M1 cohort with robust performance (iAUC=0·77). Metaanalysis across all models confirmed previously identified predictive clinical variables and revealed aspartate aminotransferase as an important albeit previously under-reported prognostic biomarker. Interpretation Novel prognostic factors were delineated and the assessment of 50 methods developed by independent international teams establishes a benchmark for the development of methods in the future. The results of this effort demonstrate that data sharing, when combined with a crowdsourced challenge, is a robust and powerful framework to develop new prognostic models in advanced prostate cancer patients.

Citation

BibTeX

@article{Guinney+al:17,
  author = {J Guinney and T Wang and Prostate Cancer Challenge DREAM Community
    and Russ Greiner and Roberto Vega and Luke Kumar and Jaykumar Patel},
  title = {Prediction of overall survival for patients with metastatic
    castration-resistant prostate cancer: development of a prognostic model
    through a crowd-sourced challenge with open clinical trial data.},
  Volume = "18",
  Number = "1",
  Pages = {132-142},
  journal = {Lancet Oncology},
  year = 2017,
}

Last Updated: February 10, 2020
Submitted by Sabina P

• Home
• Search
  • Advanced Search
• Publications
  • View Publication Entries
• Authors
  • View Authors
• User
  • Login or Register