The Biostatistics Shared Resource program contributes to a range of research conducted at UC Irvine's Chao Family Comprehensive Cancer Clinic (CFCCC). Some collaborations require the application of standard statistical methods when addressing the scientific goals of interest.
Others require the development and application of novel statistical methods for a proper data analysis and, ultimately, a valid scientific inference.
Two research highlights are given below. Our first illustrates the fruits of a close scientific collaboration between Dr. Dan Gillen and former CFCCC member Dr. Susan Neuhausen in the development of novel methods of survival analysis in the presence of unknown genetic haplotypes. The second describes methods and software as used by BSR faculty for the efficient design and analysis of sequential clinical trials.
Dr. Gillen recently collaborated with Dr. Susan Neuhausen on the analysis of genetic modifiers of the risk of breast cancer among BRCA1 and BRCA2 carriers. This research involved the analysis of censored survival data stemming from a complex sampling design as well as the imputation of unknown genetic haplotypes.
As no existing methodology or software for these data existed at the time, Dr. Gillen received a NCI R03 grant to develop new methodology as well as software for applying the proposed methods.
Figure 1 depicts a subset of results in a forthcoming publication by Drs. Neuhausen and Gillen (among others) illustrating the relationship between SNPs on the MMP7 gene and the relative risk of breast cancer diagnosis among BRCA1/2 carriers. In addition, Dr. Gillen presented this new methodology during a talk at the 2010 International Biometrics Conference held in Flioronopolis, Brazil.
Group sequential methods provide more efficient and hence more ethical design options for later stage clinical trials by allowing for interim testing of accruing trial data. Such methods allow for early stopping when confidence in a decision is reached while controlling statistical operating characteristics such as type I and II error rates. Dr. Gillen is a collaborator in the development of RCTDesign, a software package for the R statistical programming language. RCTDesign is a comprehensive package for the design, monitoring, and reporting of group sequential clinical trials.
Figure 2 displays example plots that are produced by this research endeavor. Specifically, these are the 95% confidence intervals that would be observed at each stopping time for a symmetric 5%-level O’Brien-Fleming group sequential boundary. This software is freely available and used by BSR faculty.
Selected BSR Publications
Examples of other notable research highlights from BSR faculty include:
Seminar and Lecture Notes