Sharron G. Penn

Comprehensive evaluation of the association between prostate cancer and genotypes/haplotypes in CYP17A1, CYP3A4, and SRD5A2

Genes involved in the testosterone biosynthetic pathway – such as CYP17A1, CYP3A4, and SRD5A2 – represent strong candidates for affecting prostate cancer. Previous work has detected associations between individual variants in these three genes and prostate cancer risk and aggressiveness. To more comprehensively evaluate CYP17A1, CYP3A4, and SRD5A2, we undertook a two-phase study of the relationship between their genotypes/haplotypes and prostate cancer. Phase I of the study first searched for single-nucleotide polymorphisms (SNPs) in these genes by resequencing 24 individuals from the Coriell Polymorphism Discovery Resource, 92–110 men from prostate cancer case–control sibships, and by leveraging public databases. In all, 87 SNPs were discovered and genotyped in 276 men from case–control sibships. Those SNPs exhibiting preliminary case–control allele frequency differences, or distinguishing (ie, ‘tagging’) common haplotypes across the genes, were identified for further study (24 SNPs in total). In Phase II of the study, the 24 SNPs were genotyped in an additional 841 men from case–control sibships. Finally, associations between genotypes/haplotypes in CYP17A1, CYP3A4, and SRD5A2 and prostate cancer were evaluated in the total case–control sample of 1117 brothers from 506 sibships. Family-based analyses detected associations between prostate cancer risk or aggressiveness and a number of CYP3A4 SNPs (P-values between 0.006 and 0.05), a CYP3A4 haplotype (P-values 0.05 and 0.009 in nonstratified and stratified analysis, respectively), and two SRD5A2 SNPs in strong linkage disequilibrium (P=0.02). Undertaking a two-phase study comprising SNP discovery, haplotype tagging, and association analyses allowed us to more fully decipher the relation between CYP17A1, CYP3A4, and SRD5A2 and prostate cancer.

Application of DNA microarrays for predicting toxicity and evaluating cross-species extrapolation

The application of DNA microarray technology to the field toxicology has increased significantly. In most cases, the research has monitored global changes in gene expression in order to provide insight into the cellular mechanisms of toxicity. Although assessing global gene expression changes may prove to be important when characterizing the action of a particular chemical, it is not necessarily predictive of the toxicological behavior within an organism or across species. As a first step for developing predictive toxicological models within a species, we developed a statistical model to classify 24 model treatments that fall into five well-studied toxicological categories based on gene expression. Using all the gene expression measurements resulted in relatively poor predictive accuracy. However, focusing on a diagnostic subset of genes greatly increased the predictive accuracy. For evaluating the toxicological behavior across species, a set of orthologous microarrays were developed that allowed a direct comparison of gene expression changes in both organisms. To construct these microarrays, a genome wide comparison of available human and mouse sequence was performed to identify putative orthologous genes. A subset of the orthologous genes were spotted on tandem microarrays (one human and one mouse) and used to evaluate conservation of expression patterns between organisms. The use of predictive statistical models and cross-genome comparisons in chemically induced gene expression are the next logical advancements in the field of toxicogenomics and their application has the potential to be extremely valuable in regulatory decisions and the risk assessment process.