Complex traits exhibit polygenic inheritance patterns, where many genes and environmental factors together influencing the traits. The primary goals of genetics research on complex traits include quantifying the relative importance of genes and environment in the total trait variation, dissecting genetic components, and elucidating underlying genetic architecture. Recent development in genotyping and sequencing technologies has largely facilitated gene mapping for complex traits using approaches such as genome-wide association studies (GWAS). GWAS allow the majority of genetic variants with population allele frequency over 1-5% to be tested for association with the trait. This has been proved as a powerful approach for gene mapping, with more than 7,000 common genetic loci being identified in the last five years. The three complex traits studied in this thesis, namely central corneal thickness (CCT), ovarian cancer risk and treatment response in ovarian cancer patients, are three representatives from a wide range of quantitative traits, disease traits and pharmacogenetic traits that GWASs have illuminated.
This thesis starts with a general introduction on polygenic inheritance, and gene mapping strategies with an emphasis on GWAS. In the latter part of the Introduction, I present overview of genetics research on the traits of interest and specific study aims of the main chapters.
I conducted two gene mapping studies on human corneal thickness, as described in Chapter 2 and 3. Chapter 2 was a GWAS on 5,000 individuals which led to the identification of two genetic loci at ZNF469 and FOXO1. The rare sequence variants in ZNF469 are known to cause the Mendelian disease, Brittle Cornea Syndrome (BCS), which has clinical feature of extremely thin cornea. The results here showed that while the rare variants in ZNF469 result in dramatic change in corneal thickness in patients with BCS, common variants near the gene account for variation of corneal thickness in the general population. I followed up this GWAS with a consortium-based meta-analysis in Chapter 3, in order to enhance study power to identify more CCT-associated variants. With a total of 20,000 individuals, this study identified 27, including 16 new, genetic loci. I performed meta-analyses in both European and Asian populations, and showed that the effect directions of these 27 loci were remarkably consistent across these two populations. I then explored whether these individual loci cluster into biological pathways. Results from pathway analysis convincingly showed that collagen and extracellular matrix pathways regulate corneal thickness in both populations. Thin cornea is a risk factor for complex eye diseases such as keratoconus and glaucoma. So in Chapter 3, I tested the hypothesis whether some of the genes associated with thin cornea would confer susceptibility to these diseases. I found that six CCT-loci carry relatively large risk in keratoconus and one also associated with glaucoma risk. Therefore, this study clearly showed that the endophenotype approach yielded disease relevant loci.
In Chapter 4, I conducted a time- and cost-effective GWAS on ovarian cancer risk using pooled DNA, to test the hypothesis that common variants, which are not well tagged by the lower-density arrays used in the previous GWAS, also account for some of the residual ovarian cancer risk. This study lacked power to identify common variants with small effect. However, I concluded in this study that there are unlikely to be any moderate or large effects on ovarian cancer risk untagged by the less dense arrays.
The pharmacological trait of treatment response in epithelial ovarian cancer (EOC) patients was studied in Chapter 5. The successful identification of loci associated with response to treatment could have profound clinical implications for individualizing anti-cancer treatment. This study represented the first genome-wide scan for genes associated with chemotherapy responses in EOC patients. Using progression-free survival as the measurement for treatment response, this study identified two polymorphisms with low allele frequency in TTC39B that are associated with PFS in serous EOC patients following carboplatin/paclitaxel treatment.
As in other GWAS, the hitherto identified genetic loci here explain a small fraction of trait heritability, giving rise to the case of “missing heritability”. In the last chapter, I provided additional results on estimating the heritability which is attributable to all common variants that are tagged on the genotyping arrays, and then continued discussion on general topics such as missing heritability and genetic architecture for the three traits studied in this thesis.