Bayesian methodology for genetics of complex diseases

Genetic research of complex diseases is a challenging, but exciting, area of research. The early development of the research was limited, however, until the completion of the Human Genome and HapMap projects, along with the reduction in the cost of genotyping, which paves the way for understanding the genetic composition of complex diseases. In this thesis, we focus on the statistical methods for two aspects of genetic research: phenotype definition for diseases with complex etiology and methods for identifying potentially associated Single Nucleotide Polymorphisms (SNPs) and SNP-SNP interactions. With regard to phenotype definition for diseases with complex etiology, we firstly investigated the effects of different statistical phenotyping approaches on the subsequent analysis. In light of the findings, and the difficulties in validating the estimated phenotype, we proposed two different methods for reconciling phenotypes of different models using Bayesian model averaging as a coherent mechanism for accounting for model uncertainty. In the second part of the thesis, the focus is turned to the methods for identifying associated SNPs and SNP interactions. We review the use of Bayesian logistic regression with variable selection for SNP identification and extended the model for detecting the interaction effects for population based case-control studies. In this part of study, we also develop a machine learning algorithm to cope with the large scale data analysis, namely modified Logic Regression with Genetic Program (MLR-GEP), which is then compared with the Bayesian model, Random Forests and other variants of logic regression.

FGFR2 mutations are rare across histologic subtypes of ovarian cancer

OBJECTIVE: Ovarian cancer is the leading cause of death from gynecologic malignancies in the Western world. Fibroblast growth factor receptor (FGFR) signaling has been implicated to play a role in ovarian tumorigenesis. Mutational activation of one member of this receptor family, FGFR2, is a frequent event in endometrioid endometrial cancer. Given the similarities in the histologic and molecular genetics of ovarian and endometrial cancers, we hypothesized that activating FGFR2 mutations may occur in a subset of endometrioid ovarian tumors, and possibly other histotypes. METHODS: Six FGFR2 exons were sequenced in 120 primary ovarian tumors representing the major histologic subtypes. RESULTS: FGFR2 mutation was detected at low frequency in endometrioid (1/46, 2.2%) and serous (1/41, 2.4%) ovarian cancer. No mutations were detected in clear cell, mucinous, or mixed histology tumors or in the ovarian cancer cell lines tested. Functional characterization of the FGFR2 mutations confirmed that the mutations detected in ovarian cancer result in receptor activation. CONCLUSIONS: Despite the low incidence of FGFR2 mutations in ovarian cancer, the two FGFR2 mutations identified in ovarian tumors (S252W, Y376C) overlap with the oncogenic mutations previously identified in endometrial tumors, suggesting activated FGFR2 may contribute to ovarian cancer pathogenesis in a small subset of ovarian tumors.

Evidence for u.v. induction of CDKN2 mutations in melanoma cell lines

The CDKN2 gene, encoding the cyclin dependent kinase inhibitor p16, is a tumour suppressor gene involved in melanoma and maps to chromosome band 9p22. Mutations or interstitial deletions of this gene have been found both in the germline of familial melanoma cases and somatically in melanoma cell lines. Previous mutation analyses of melanoma cell lines have indicated a high frequency of C:G to T:A transitions, with all of these mutations occurring at dipyrimidine sites. Including three melanoma cell lines carrying tandem CC to TT mutations, the spectrum of mutations so far reported indicates a possible role for u.v. radiation in the mutagenesis of this gene in some tumours. To further examine this hypothesis we have characterised mutations of the CDKN2 gene in 30 melanoma cell lines. Nineteen lines carried complete or partial homozygous deletions of the gene. Of the remaining cell lines, eight were shown by direct sequencing of PCR products from exon 1 and exon 2 to carry a total of nine different mutations of CDKN2. Two cell lines carried tandem CC to TT mutations and a high rate of C:G to T:A transitions was observed. This study provides further evidence for the role of u.v. light in the genesis of melanoma, with one target being the CDKN2 tumour suppressor gene.

Cloning and characterization of a novel human gene related to vascular endothelial growth factor

This paper describes the cloning and characterization of a new member of the vascular endothelial growth factor (VEGF) gene family, which we have designated VRF for VEGF-related-factor. Sequencing of cDNAs from a human fetal brain library and RT-PCR products from normal and tumor tissue cDNA pools indicate two alternatively spliced messages with open reading frames of 621 and 564 bp, respectively. The predicted proteins differ at their carboxyl ends resulting from a shift in the open reading frame. Both isoforms show strong homology to VEGF at their amino termini, but only the shorter isoform maintains homology to VEGF at its carboxyl terminus and conserves all 16 cysteine residues of VEGF165. Similarity comparisons of this isoform revealed overall protein identity of 48% and conservative substitution of 69% with VEGF189. VRF is predicted to contain a signal peptide, suggesting that it may be a secreted factor. The VRF gene maps to the D11S750 locus at chromosome band 11q13, and the protein coding region, spanning approximately 5 kb, is comprised of 8 exons that range in size from 36 to 431 bp. Exons 6 and 7 are contiguous and the two isoforms of VRF arise through alternate splicing of exon 6. VRF appears to be ubiquitously expressed as two transcripts of 2.0 and 5.5 kb; the level of expression is similar among normal and malignant tissues.

CDKN2A mutation in a non-FAMMM kindred with cancers at multiple sites results in a functionally abnormal protein

The CDKN2A gene encodes p16 (CDKN2A), a cell-cycle inhibitor protein which prevents inappropriate cell cycling and, hence, proliferation. Germ-line mutations in CDKN2A predispose to the familial atypical multiple-mole melanoma (FAMMM) syndrome but also have been seen in rare families in which only 1 or 2 individuals are affected by cutaneous malignant melanoma (CMM). We therefore sequenced exons 1alpha and 2 of CDKN2A using lymphocyte DNA isolated from index cases from 67 families with cancers at multiple sites, where the patterns of cancer did not resemble those attributable to known genes such as hMLH1, hMLH2, BRCA1, BRCA2, TP53 or other cancer susceptibility genes. We found one mutation, a mis-sense mutation resulting in a methionine to isoleucine change at codon 53 (M531) of exon 2. The individual tested had developed 2 CMMs but had no dysplastic nevi and lacked a family history of dysplastic nevi or CMM. Other family members had been diagnosed with oral cancer (2 persons), bladder cancer (1 person) and possibly gall-bladder cancer. While this mutation has been reported in Australian and North American melanoma kindreds, we did not observe it in 618 chromosomes from Scottish and Canadian controls. Functional studies revealed that the CDKN2A variant carrying the M531 change was unable to bind effectively to CDK4, showing that this mutation is of pathological significance. Our results have confirmed that CDKN2A mutations are not limited to FAMMM kindreds but also demonstrate that multi-site cancer families without melanoma are very unlikely to contain CDKN2A mutations.

The genetics of cutaneous melanoma

Although germline mutations in CDKN2A are present in approximately 25% of large multicase melanoma families, germline mutations are much rarer in the smaller melanoma families that make up most individuals reporting a family history of this disease. In addition, only three families worldwide have been reported with germline mutations in a gene other than CDKN2A (i.e., CDK4). Accordingly, current genomewide scans underway at the National Human Genome Research Institute hope to reveal linkage to one or more chromosomal regions, and ultimately lead to the identification of novel genes involved in melanoma predisposition. Both CDKN2A and PTEN have been identified as genes involved in sporadic melanoma development; however, mutations are more common in cell lines than uncultured tumors. A combination of cytogenetic, molecular, and functional studies suggests that additional genes involved in melanoma development are located to chromosomal regions 1p, 6q, 7p, 11q, and possibly also 9p and 10q. With the near completion of the human genome sequencing effort, combined with the advent of high throughput mutation analyses and new techniques including cDNA and tissue microarrays, the identification and characterization of additional genes involved in melanoma pathogenesis seem likely in the near future.

Quality of life and genetics in men with prostate cancer

As family history has been established as a risk factor for prostate cancer, attempts have been made to isolate predisposing genetic variants that are related to hereditary prostate cancer. With many genetic variants still to be identified and investigated, it is not yet possible to fully understand the impact of genetic variants on prostate cancer development. The high survival rates among men with prostate cancer have meant that other issues, such as quality of life (QoL), have also become important. Through their effect on a person’s health, a range of inherited genetic variants may potentially influence QoL in men with prostate cancer, even prior to treatment. Until now, limited research has been conducted on the relationship between genetics and QoL. Thus, this study contributes to an emerging field by aiming to identify certain genetic variants related to the QoL found in men with prostate cancer. It is hoped that this study may lead to future research that will identify men who have an increased risk of a poor QoL following prostate cancer treatment, which will aid in developing treatments that are individually tailored to support them. Previous studies have established that genetic variants of Vascular Endothelial Growth Factor (VEGF) and Insulin-like Growth Factor 1 (IGF-1) may play a role in prostate cancer development. VEGF and IGF-1 have also been reported to be associated with QoL in people with ovarian cancer and colorectal cancer, respectively. This study completed a series of secondary analyses using two major data-sets (from 850 men newly diagnosed with prostate cancer, and approximately 550 men from the general Queensland population), in which genetic variants of VEGF and IGF-1 were investigated for associations with prostate cancer susceptibility and QoL. The first aim of this research was to investigate genetic variants in the VEGF and IGF-I gene for an association with the risk of prostate cancer. It was found that one IGF-1 genetic variant (rs35765) had a statistically significant association with prostate cancer (p = 0.04), and one VEGF genetic variant (rs2146323) had a statistically significant association with advanced prostate cancer (p = 0.02). The estimates suggest that carriers of the CA and AA genotype for rs35765 may have a reduced risk of developing prostate cancer (Odds Ratio (OR) = 0.72, 95% Confidence Interval (CI) = 0.55, 0.95, OR = 0.60, 95% CI = 0.26, 1.39, respectively). Meanwhile, carriers of the CA and AA genotype for rs2146323 may be at increased risk of advanced prostate cancer, which was determined by a Gleason score of above 7 (OR = 1.72, 95% CI = 1.12, 2.63, OR = 1.90, 95% CI = 1.08, 3.34, respectively). Utilising the widely used short-form health survey, the SF-36v2, the second aim of this study was to investigate the relationship between prostate cancer and QoL prior to treatment. Assessing QoL at this time-point was important as little research has been conducted to evaluate if prostate cancer affects QoL regardless of treatment. The analyses found that mean SF-36v2 scale scores related to physical health were higher by at least 0.3 Standard Deviations (SD) among men with prostate cancer than the general population comparison group. This difference was considered clinically significant (defined by group differences in mean SF-36v2 scores by at least 0.3 SD). These differences were also statistically significant (p<0.05). Mean QoL scale scores related to mental health were similar between men with prostate cancer and those from the general population comparison group. The third aim of this study was to investigate genetic variants in the VEGF and IGF-1 gene for an association with QoL in prostate cancer patients prior to their treatment. It was essential to evaluate these relationships prior to treatment, before the involvement of these genes was potentially interrupted by treatment. The analyses found that some genetic variants had a small clinically significant association (0.3 SD) to some QoL domains experienced by these men. However, most relationships were not statistically significant (p>0.05). Most of the associations found identified that a small sub-group of men with prostate cancer (approximately 2%) reported, on average, a slightly better QoL than the majority of the prostate cancer patients. The fourth aim of this research was to investigate whether associations between genetic variants in VEGF and IGF-1 and QoL were specific to men with prostate cancer, or were also applicable to the general male population. It was found that twenty out of one-hundred relationships between the genetic variants of VEGF and IGF-1 and QoL health-measures and scales examined differed between these groups. In the majority of the relationships involving VEGF SNPs that differed, a clinically significant difference (0.3 or more SD) between mean scores among the genotype groups in prostate cancer patients was found, while mean scores among men from the general-population comparison group were similar. For example, prostate cancer participants who carried at least one T allele (CT or TT genotype) for rs3024994 had a clinically significant higher (0.3 SD) mean QoL score in terms of the role-physical scale, than participants who carried the CC genotype. This was not seen among men from the general population sample, as the mean score was similar between genotype groups. The opposite was seen in regards to the IGF-1 SNPs examined. Overall, these relationships were not considered to directly impact on the clinical options for men with prostate cancer. As this study utilised secondary data from two separate studies, there are a number of important limitations that should be acknowledged including issues of multiple comparisons, power, and missing or unavailable data. It is recommended that this study be replicated as a better-designed study that takes greater consideration of the many factors involved in prostate cancer and QoL. Investigation into other genetic variants of VEGF or IGF-1 is also warranted, as is consideration of other genes and their relationship with QoL. Through identifying certain genetic variants that have a modest association to prostate cancer, this project adds to the knowledge surrounding VEGF and IGF-1 and their role in prostate cancer susceptibility. Importantly, this project has also introduced the potential role genetics plays in QoL, through investigating the relationships between genetic variants of VEGF and IGF-1 and QoL.

Cloning of a novel insulin-regulated ghrelin transcript in prostate cancer

Ghrelin is a multifunctional hormone, with roles in stimulating appetite and regulating energy balance, insulin secretion and glucose homeostasis. The ghrelin gene locus (GHRL) is highly complex and gives rise to a range of novel transcripts derived from alternative first exons and internally spliced exons. The wild-type transcript encodes a 117 amino acid preprohormone that is processed to yield the 28 amino acid peptide ghrelin. Here, we identified insulin-responsive transcription corresponding to cryptic exons in intron 2 of the human ghrelin gene. A transcript, termed in2c-ghrelin (intron 2-cryptic), was cloned from the testis and the LNCaP prostate cancer cell line. This transcript may encode an 83 AA preproghrelin isoform that codes for the ghrelin, but not obestatin. It is expressed in a limited number of normal tissues and in tumours of the prostate, testis, breast and ovary. Finally, we confirmed that in2c-ghrelin transcript expression, as well as the recently described in1-ghrelin transcript, is significantly upregulated by insulin in cultured prostate cancer cells. Metabolic syndrome and hyperinsulinaemia has been associated with prostate cancer risk and progression. This may be particularly significant after androgen deprivation therapy for prostate cancer, which induces hyperinsulinaemia, and this could contribute to castrate resistant prostate cancer growth. We have previously demonstrated that ghrelin stimulates prostate cancer cell line proliferation in vitro. This study is the first description of insulin regulation of a ghrelin transcript in cancer, and should provide further impetus for studies into the expression, regulation and function of ghrelin gene products.

Evolutionary genetics and human assisted movement of a globally invasive pest (Russian wheat aphid : Diuraphis noxia)

This PhD study has examined the population genetics of the Russian wheat aphid (RWA, Diuraphis noxia), one of the world’s most invasive agricultural pests, throughout its native and introduced global range. Firstly, this study investigated the geographic distribution of genetic diversity within and among RWA populations in western China. Analysis of mitochondrial data from 18 sites provided evidence for the long-term existence and expansion of RWAs in western China. The results refute the hypothesis that RWA is an exotic species only present in China since 1975. The estimated date of RWA expansion throughout western China coincides with the debut of wheat domestication and cultivation practices in western Asia in the Holocene. It is concluded that western China represents the limit of the far eastern native range of this species. Analysis of microsatellite data indicated high contemporary gene flow among northern populations in western China, while clear geographic isolation between northern and southern populations was identified across the Tianshan mountain range and extensive desert regions. Secondly, this study analyzed the worldwide pathway of invasion using both microsatellite and endosymbiont genetic data. Individual RWAs were obtained from native populations in Central Asia and the Middle East and invasive populations in Africa and the Americas. Results indicated two pathways of RWA invasion from 1) Syria in the Middle East to North Africa and 2) Turkey to South Africa, Mexico and then North and South America. Very little clone diversity was identified among invasive populations suggesting that a limited founder event occurred together with predominantly asexual reproduction and rapid population expansion. The most likely explanation for the rapid spread (within two years) from South Africa to the New World is by human movement, probably as a result of the transfer of wheat breeding material. Furthermore, the mitochondrial data revealed the presence of a universal haplotype and it is proposed that this haplotype is representative of a wheat associated super-clone that has gained dominance worldwide as a result of the widespread planting of domesticated wheat. Finally, this study examined salivary gland gene diversity to determine whether a functional basis for RWA invasiveness could be identified. Peroxidase DNA sequence data were obtained for a selection of worldwide RWA samples. Results demonstrated that most native populations were polymorphic while invasive populations were monomorphic, supporting previous conclusions relating to demographic founder effects in invasive populations. Purifying selection most likely explains the existence of a universal allele present in Middle Eastern populations, while balancing selection was evident in East Asian populations. Selection acting on the peroxidase gene may provide an allele-dependent advantage linked to the successful establishment of RWAs on wheat, and ultimately their invasion potential. In conclusion, this study is the most comprehensive molecular genetic investigation of RWA population genetics undertaken to date and provides significant insights into the source and pathway of global invasion and the potential existence of a wheat-adapted genotype that has colonised major wheat growing countries worldwide except for Australia. This research has major biosecurity implications for Australia’s grain industry.

Molecular genetics of migraine

Migraine is a common neurological disorder with a significantly heritable component. It is a complex disease and despite numerous molecular genetic studies, the exact pathogenesis causing the neurological disturbance remains poorly understood. Although several known molecular mechanisms have been associated with an increased risk for developing migraine, there remains significant scope for future studies. The majority of studies have investigated the most plausible candidate genes involved in common migraine pathogenesis utilising criteria that takes into account a combination of physiological functionality in conjunction with regions of genomic association. Thus, far genes involved in neurological, vascular or hormonal pathways have been identified and investigated on this basis. Genome-wide association studies (GWAS) studies have helped to identify novel regions that may be associated with migraine and have aided in providing the basis for further molecular investigations. However, further studies utilising sequencing technologies are required to characterise the genetic basis for migraine.

The genetics of endurance : frequency of the ACTN3 R577X variant in Ironman World Championship athletes

Objectives To investigate the frequency of the ACTN3 R577X polymorphism in elite endurance triathletes, and whether ACTN3 R577X is significantly associated with performance time. Design Cross-sectional study. Methods Saliva samples, questionnaires, and performance times were collected for 196 elite endurance athletes who participated in the 2008 Kona Ironman championship triathlon. Athletes were of predominantly North American, European, and Australian origin. A one-way analysis of variance was conducted to compare performance times between genotype groups. Multiple linear regression analysis was performed to model the effect of questionnaire variables and genotype on performance time. Genotype and allele frequencies were compared to results from different populations using the chi-square test. Results Performance time did not significantly differ between genotype groups, and age, sex, and continent of origin were significant predictors of finishing time (age and sex: p < 5 × 10−6; continent: p = 0.003) though genotype was not. Genotype and allele frequencies obtained (RR 26.5%, RX 50.0%, XX 23.5%, R 51.5%, X 48.5%) were found to be not significantly different from Australian, Spanish, and Italian endurance athletes (p > 0.05), but were significantly different from Kenyan, Ethiopian, and Finnish endurance athletes (p < 0.01). Conclusions Genotype and allele frequencies agreed with those reported for endurance athletes of similar ethnic origin, supporting previous findings for an association between 577X allele and endurance. However, analysis of performance time suggests that ACTN3 does not alone influence endurance performance, or may have a complex effect on endurance performance due to a speed/endurance trade-off.

Genetics of menstrual migraine : the molecular evidence

Migraine is considered to be a multifactorial disorder in which genetic, environmental, and, in the case of menstrual and menstrually related migraine, hormonal events influence the phenotype. Certainly, the role of female sex hormones in migraine has been well established, yet the mechanism behind this well-known relationship remains unclear. This review focuses on the potential role of hormonally related genes in migraine, summarizes results of candidate gene studies to date, and discusses challenges and issues involved in interpreting hormone-related gene results. In light of the molecular evidence presented, we discuss future approaches for analysis with the view to elucidate the complex genetic architecture that underlies the disorder.

Investigation of Gamma-aminobutyric acid (GABA) A receptors genes and migraine susceptibility

Background Migraine is a neurological disorder characterized by recurrent attacks of severe headache, affecting around 12% of Caucasian populations. It is well known that migraine has a strong genetic component, although the number and type of genes involved is still unclear. Prior linkage studies have reported mapping of a migraine gene to chromosome Xq 24–28, a region containing a cluster of genes for GABA A receptors (GABRE, GABRA3, GABRQ), which are potential candidate genes for migraine. The GABA neurotransmitter has been implicated in migraine pathophysiology previously; however its exact role has not yet been established, although GABA receptors agonists have been the target of therapeutic developments. The aim of the present research is to investigate the role of the potential candidate genes reported on chromosome Xq 24–28 region in migraine susceptibility. In this study, we have focused on the subunit GABA A receptors type ε (GABRE) and type θ (GABRQ) genes and their involvement in migraine. Methods We have performed an association analysis in a large population of case-controls (275 unrelated Caucasian migraineurs versus 275 controls) examining a set of 3 single nucleotide polymorphisms (SNPs) in the coding region (exons 3, 5 and 9) of the GABRE gene and also the I478F coding variant of the GABRQ gene. Results Our study did not show any association between the examined SNPs in our test population (P > 0.05). Conclusion Although these particular GABA receptor genes did not show positive association, further studies are necessary to consider the role of other GABA receptor genes in migraine susceptibility.

Migraine genetics and prospects for pharmacotherapy

Migraine is a common complex neurological disorder with a well-known but poorly characterized genetic liability. The search for migraine susceptibility genes has been the focus of intense research. It is now believed that common migraine is not a single gene disorder, but attributable to several potentially interacting genetic variants. These variants may differ in each sufferer and interact with environmental factors to set the individual migraine threshold. This genetic liability may play an important role in the clinical heterogeneity seen in migraine and also in the variability of treatment response. This review will look at genetic loci implicated in migraine to date and consider their current or prospective role in migraine therapy. To elucidate the complex nature of migraine genetic liability, approaches that consider detailed endophenotypic profiles that encompass treatment response may provide much more relevant information than simple end diagnosis.