The aims of this proposal are to develop statistical methods to make appropriate and efficient use of the data collected in epidemiologic and clinical studies in which a fraction of the subjects will not develop the endpoint. This is a frequent occurrence in cancer studies when some patients can be cured. Statistical models, called cure models, have been proposed to analyze such data. Methods for incorporating covariates into cure models will be developed. Both time fixed and time dependent covariates will be considered. The methodology will be applied to datasets from prostate cancer, head and neck cancer, breast cancer and osteosarcoma. An accelerated failure time regression cure model is proposed for datasets with time fixed covariates. An EM algorithm estimation method incorporating a non-parametric baseline latency distribution will be investigated. A method is proposed for estimating non-linear covariate effects in cure models, using penalised likelihood. Proper statistical modeling with time dependent covariates is much more difficult. In the second project, the applicants will investigate a method of joint modeling of failure times and time dependent (internal) covariates in cure models. The proposed model combines aspects of a mixture model for the cured and uncured groups, a random effects model for longitudinal data and a proportional hazards model for failure time data. They will develop a Gibbs sampling approach for this model. Application os this model to auxiliary variables in clinical trials will be given. The methodology proposed will allow time dependent auxiliary variables to provide additional information for censored cases, and thus give more accurate and efficient conclusions from the trial.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
5R01CA072495-04
Application #
2895739
Study Section
Special Emphasis Panel (ZRG7-STA (01))
Program Officer
Erickson, Burdette (BUD) W
Project Start
1997-09-30
Project End
2001-08-31
Budget Start
1999-09-01
Budget End
2001-08-31
Support Year
4
Fiscal Year
1999
Total Cost
Indirect Cost
Name
University of Michigan Ann Arbor
Department
Biostatistics & Other Math Sci
Type
Schools of Public Health
DUNS #
791277940
City
Ann Arbor
State
MI
Country
United States
Zip Code
48109
Li, Chin-Shang; Taylor, Jeremy M G (2002) A semi-parametric accelerated failure time cure model. Stat Med 21:3235-47
Zhuang, D; Schenker, N; Taylor, J M et al. (2000) Analysing the effects of anaemia on local recurrence of head and neck cancer when covariate values are missing. Stat Med 19:1237-49
Sy, J P; Taylor, J M (2000) Estimation in a Cox proportional hazards cure model. Biometrics 56:227-36