|
Witte Lab
Last updated: September 11, 2003 |
ResearchThe Witte Lab's work encompasses a synthesis of applied and methodologic cancer, genetic, and epidemiologic research. At the present time our applied research is primarily focused on deciphering risk factors for the initiation and progression of prostate cancer. In particular, we are leading a thriving multi-institutional prostate cancer research program (CaP Genes) comprised of genetic epidemiologists, molecular biologists, pathologists, statisticians, and urologists. We have had numerous successes toward sorting out the complex nature of this disease, including isolating chromosomal regions that may harbor prostate cancer genes (Am J Hum Genet 2000;67:92-99; Cancer Res 2000;60:3645-3649; Am J Hum Genet 2001;68:1197-1206; Neoplasia 2002;4:424-431; Genes Chrom Cancer 2002) and detecting an association between a polymorphism in the gene CYP3A4 and more advanced disease among African-American men (Cancer Epidemiol Biomarkers Prev 1999;8:901-906). We have also shown that a relatively common variant in the putative HPC1 gene RNASEL may increase prostate cancer risk 1.5- to 2-fold (Nat Genet 2002;32:581-583). Meanwhile, our other work includes looking at the impact of diet—and diet ´ gene interactions—on the risk of prostate cancer. This applied work helps motivate our methodological research, which primarily involves issues surrounding the design and analysis of genetic and epidemiologic studies. For example, a key aspect of our research is the further development of hierarchical modeling—a potentially valuable statistical analysis approach. We have shown how this approach can be used to incorporate genotype- and haplotype-level information in linkage disequilibrium mapping (Am J Hum Genet 2003;72:351-363). We previously provided an extensive application of hierarchical modeling in analyzing case-control data on diet and breast cancer (Epidemiology 1994;5:612-621). We have also undertaken a simulation study showing that hierarchical modeling generally gives more accurate effect estimates than standard analytic techniques (Stat Med 1996;15:1161-1170). Other work has shown how this approach can be used in genetic epidemiologic research (Genet Epidemiol 1997;14:1137-1142), and provided to the scientific community software for undertaking hierarchical modeling (Epidemiology 1998;9:563-566; Epidemiology 2000;11:684-688). In related work we have investigated the analytic approaches for detecting hot-spots in mutational spectra (Stat Med 2002;21:1867-1885). Another key area of our research is focused on the use of case-control (“association”) studies in genetic epidemiology. For example, we have shown that using as controls some types of family members, such as siblings, can reduce power for detecting main genetic effects, but can provide improved power for detecting gene-environment interactions (Am J Epidemiol 1999;149:693-705; JNCI Mono 1999;26:31-37). Other related work has shown that standard panels of single nucleotide polymorphisms (SNPs) may not suffice for genome-wide association studies across different ethnic groups (Am J Hum Genet 2000;66:216-234), and we are currently evaluating the use of haplotype tagging SNPs in association studies. Furthermore, since a large number of SNPs will likely be required for such studies, we have considered the ensuing problems of multiple comparisons (Nat Genet 1996;12:355-356), showing that the required increase in sample size for testing large numbers of SNPs is surprisingly small (Stat Med 2000;19:369-372). Finally, we are actively investigating the impact of incorporating genetic information into the design and analysis of clinical trials. This research shows how one can drastically reduce clinical trial size and duration by pre-genotyping potential study subjects (Control Clin Trials 2000;21:7-20; Pharmacogenetics 2000;10:503-10; Pharmacogenetics 2001;11:459-60). |