Not Logged In

A DREAM Challenge to build prediction models for short-term discontinuation of docetaxel in metastatic castration-resistant prostate cancer

Full Text: cci.17.00018.pdf PDF

PURPOSE Docetaxel has a demonstrated survival benefit for patients with metastatic castration-resistant prostate cancer (mCRPC). However, 10-20% of patients discontinue docetaxel prematurely because of toxicity-induced adverse events, and the management of risk factors for toxicity remains an ongoing challenge.

PATIENTS AND METHODS The comparator arms of four phase III clinical trials in first-line mCRPC were collected, annotated and compiled (2070 patients). Early discontinuation was defined as treatment stoppage within three months due to adverse treatment effects; 10% of patients discontinued. We designed an open-data, crowdsourced DREAM Challenge for developing models to predict early discontinuation of docetaxel. Clinical features for all four trials and outcomes for three of the trials were made publicly available with the fourth trial held back for unbiased model evaluation. Challenge participants from around the world trained models and submitted their predictions; area under precision-recall curve (AUPRC) was the main metric used for performance assessment.

RESULTS In total, 34 separate teams submitted predictions. Seven models with statistically similar AUPRCs (Bayes factor ≤ 3) outperformed all other models. A post-challenge analysis of risk prediction using these seven models revealed three patient subgroups: high-risk, low-risk, or discordant-risk. Early discontinuation events were two-times higher in the high- versus low-risk subgroup. Simulation studies demonstrated that utilization of patient discontinuation prediction models can reduce patient enrollment in clinical trials without loss of statistical power.

CONCLUSION This work represents a successful collaboration between 34 international teams leveraging open clinical trial data. Our results are the first to demonstrate that routinely collected clinical features can be used to identify mCRPC patients who are likely to discontinue treatment due to adverse events, and establishes a robust benchmark with implications in clinical trial design.

Citation

F. Seyednasrollah, D. Koestler, T. Wang, S. Piccolo, R. Vega, R. Greiner, C. Fuchs, E. Gofer, L. Kumar. "A DREAM Challenge to build prediction models for short-term discontinuation of docetaxel in metastatic castration-resistant prostate cancer". Journal of Clinical Oncology: Clinical Cancer Informatics, (1), pp 1-15, July 2017.

Keywords: machine learning, prostate cancer, DREAM Challenge, medical informatics
Category: In Journal
Web Links: Journal site

BibTeX

@article{Seyednasrollah+al:JCO:CCI17,
  author = {Fatemeh Seyednasrollah and Devin C Koestler and Tao Wang and
    Stephen R Piccolo and Roberto Vega and Russ Greiner and Christiane Fuchs
    and Eyal Gofer and Luke Kumar},
  title = {A DREAM Challenge to build prediction models for short-term
    discontinuation of docetaxel in metastatic castration-resistant prostate
    cancer},
  Number = "1",
  Pages = {1-15},
  journal = {Journal of Clinical Oncology: Clinical Cancer Informatics},
  year = 2017,
}

Last Updated: February 07, 2020
Submitted by Sabina P

University of Alberta Logo AICML Logo