This project responds to timely methodology needs and opportunities in biomarker evaluations for cancer risk prediction, diagnosis, early detection, and prognosis In the Early Detection Research Network (EDRN). Nevertheless, our work will have implications more broadly for studies of medical tests.
The first aim proposes the development of ROC analysis that allows combination of datasets from multiple studies, populations or platforms including determining if ROC performance is common across datasets;providing efficient estimation of the common ROC performance;and quantifying factors either within or across datasets that affect ROC performance when performance is not common. The proposed methodology provides a unified framework for risk prediction modeling and ROC regression methdologies that have been disjoint and potentially inconsistent.
A second aim i s concerned with group sequential design methods development with the goal of preserving high quality biospecimens in biomarker validation studies, including developing methods for estimating the performance of single markers or marker combinations that accommodate the potential for early study termination;and investigating optimization of design parameters such as timing of interim analyses and criteria for early termination.
A final aim responds to methodology needs in study designs for prognostic biomarker evaluations, including efficient ways of selecting for ascertainment of data and specimens;estimating a variety of biomarker performance metrics from these designs;and comparing them with existing designs and estimation approaches. The general approach involves the specification of pertinent statistical methods and the use of both theoretical probablitistic methods, and computer simulations that are informed by application to data from substantive research contexts in which the four participating biostatistical investigators are engaged.

Public Health Relevance

This project proposes to develop statistical methods to allow combining data sources to identify and confirm new medical tests;to allow early termination of nonperforming biomarkers yet rigorous and efficient estimation of new medical test performance;and to enable rigorous and efficient study designs for disease prognosis studies when ascertainment of clinical and biomarker data from all patients in existing biorepositories is not feasible.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Research Program Projects (P01)
Project #
Application #
Study Section
Special Emphasis Panel (ZCA1-GRB-S)
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Fred Hutchinson Cancer Research Center
United States
Zip Code
Prentice, Ross L; Zhao, Shanshan (2016) Nonparametric estimation of the multivariate survivor function: the multivariate Kaplan-Meier estimator. Lifetime Data Anal :
Dai, James Y; Zhang, Xinyi Cindy; Wang, Ching-Yun et al. (2016) Augmented case-only designs for randomized clinical trials with failure time endpoints. Biometrics 72:30-8
Prentice, R L (2016) Higher Dimensional Clayton-Oakes Models for Multivariate Failure Time Data. Biometrika 103:231-236
Wang, Zhu; Ma, Shuangge; Zappitelli, Michael et al. (2016) Penalized count data regression with application to hospital stay after pediatric cardiac surgery. Stat Methods Med Res 25:2685-2703
Koopmeiners, Joseph S; Feng, Ziding (2016) Group sequential testing of the predictive accuracy of a continuous biomarker with unknown prevalence. Stat Med 35:1267-80
Petralia, Francesca; Song, Won-Min; Tu, Zhidong et al. (2016) New Method for Joint Network Analysis Reveals Common and Different Coexpression Patterns among Genes and Proteins in Breast Cancer. J Proteome Res 15:743-54
Bryan, Matthew; Heagerty, Patrick J (2016) Multivariate analysis of longitudinal rates of change. Stat Med 35:5117-5134
Cheng, Yichen; Dai, James Y; Kooperberg, Charles (2016) Group association test using a hidden Markov model. Biostatistics 17:221-34
Dai, James Y; Tapsoba, Jean de Dieu; Buas, Matthew F et al. (2016) Constrained Score Statistics Identify Genetic Variants Interacting with Multiple Risk Factors in Barrett's Esophagus. Am J Hum Genet 99:352-65
Maziarz, Marlena; Heagerty, Patrick; Cai, Tianxi et al. (2016) On longitudinal prediction with time-to-event outcome: Comparison of modeling options. Biometrics :

Showing the most recent 10 out of 279 publications