In 2009, two large genome wide association studies (GWAS) found associations between common single nucleotide polymorphisms (SNPs) at three loci (CLU, PICALM and CR1) and Alzheimer’s disease (AD) risk. The causal variants underlying these associations and how these impact on AD susceptibility remain unclear. Target enrichment and next generation sequencing (NGS) were used to completely resequence the three associated loci in 96 AD patients in an attempt to uncover potentially causative and rare variants that may explain the observed association signals. A pipeline was developed for the handling of pooled NGS data following a comparison of several different combinations of programs. 33 exonic SNPs were found within the three genes, along with over 1000 non-coding variants. To identify the variants most likely to be affecting AD risk, a two pronged approach was adopted. The variants were imputed in a large case-control cohort (2067 cases, 7376 controls) to test for association with AD, and the likely functional consequences of the variants were assessed using in silico resources. Several of the analysed variants showed suggestive or significant association with AD in the imputed data, and/or were predicted to have consequences on the function or regulation of the genes, suggesting avenues for future research in AD genetics. The whole method of pooled, targeted NGS and prioritisation using imputed data for association testing and in silico resources for functional analysis represents a new strategy for tracking down the illusive causation of GWAS signals.

This thesis explores doctors’ and medical students’ attitudes toward older patients in UK hospital settings. There have been regular and strong assertions in the grey literature and the news media that negative attitudes toward older patients may contribute to the inequality of healthcare service provision and treatment for older patients, compared to younger patients (those aged under 65 years), in UK hospital settings. However, much of the evidence does not investigate or explore these attitudes using a theoretical framework of attitudes outlined in the scientific research literature. This thesis comprises three studies. Firstly, a systematic search and review (Study 1) was undertaken in order to determine how attitudes toward older patients had been explored to date in the English-language, scientific research literature. Results demonstrated that previous studies had focused on attitude measurement rather than exploring the content of attitudes toward older patients. In fact, there was little evidence that previous research had ever explored these attitudes, despite the number of studies attempting to measure them. Furthermore, the review indicated the lack of research emanating from UK settings. In Study 2, attitudes toward older patients and their care were explored in twenty-five in-depth interviews with medical students and doctors in a UK NHS Hospital trust. Data were thematically analysed and findings indicated that attitudes toward older patients and their care could be conceptualised as: (1) attitudes toward older patients and their healthcare needs, and (2) attitudes toward providing care for older patients (e.g. the social and organisational barriers and facilitators). Within these two domains, the themes, subthemes and nodes, which represent attitude content with increasing levels of specificity, are presented. The findings from Study 2 mark one of the first attempts in this research area to explore and describe the content of attitudes in line with a theoretical framework of attitudes. The final study, Study 3, explored the devaluation and unpopularity of the specialty of geriatric medicine as a future career choice in a sample of junior doctors. Having identified, in Study 2, that geriatric medicine was not highly regarded in a range of doctors and medical students, Study 3 aimed to ascertain whether this was due to the organisational and working environment or due to older patient-related factors in a recently-qualified sample of doctors. The findings indicated that organisational and work-related factors serve to discourage junior doctors from pursuing geriatric medicine, rather than factors related to the older patients treated on geriatric wards. This thesis contributes to the research literature in two main ways. Firstly, this thesis outlines the research gaps in the worldwide English-language scientific research. Secondly, this thesis presents a conceptualisation of doctors’ and medical students’ attitudes toward older patients in a UK hospital setting. Importantly, this conceptualisation provides research that is relevant to UK settings and is in line with a theoretical framework of attitudes that has been identified from the scientific research literature. The strengths and limitations of this work are discussed.

Alzheimer’s disease (AD) is the commonest form of dementia and is characterised with neuropathological hallmarks such as aggregated amyloid plaques and hyper-phosphorylated tau protein. One type of AD is autosomal dominant AD (ADAD) which is caused by highly penetrant variants in one of three genes (APP, PSEN1 and PSEN2), other cases of AD are described as sporadic and can have a late onset of disease symptoms (> 65 years of age) or early onset (≤ 65 years or age). Late-onset Alzheimer’s disease (LOAD) is estimated to be 70% heritable and is common. Conversely sporadic early-onset Alzheimer’s disease (sEOAD) is estimated to 90% heritable but is relatively rare. The difference in prevalence between the two types of AD has resulted in genome wide association studies focusing on LOAD with sEOAD receiving little attention. Here we use an Illumina human exome genotyping chip customised with neurodegenerative markers (NeuroX) to genotype the coding region of sEOAD samples in a hope to elucidate the genetic aetiology of sEOAD. Sanger sequencing exons 16 and 17 of APP was conducted in a sEOAD cohort (n=451) to screen for variants known to cause ADAD; 9% (n=4) of the cohort were heterozygous for known causative variants and where subsequently removed from the sEOAD NeuroX genotyping data before analyses. Screening also highlighted an intronic 6bp deletion downstream of exon 17 in APP with a non-significant increased minor allele frequency (MAF) in sEOAD, however it did not appear to influence splicing of exon 17. Screening the sEOAD cohort for other variants known to cause neurodegenerative disease was conducted using the NeuroX genotyping data (n=408) which identified two samples with variants in PARK2, these variants are thought to contribute susceptibility to Parkinson’s disease (PD) suggesting these variants might elicit risk for multiple diseases. A further study with increased power would ascertain if the 6bp deletion and PARK2 variants are associated with sEOAD. Statistical analyses of the sEOAD NeuroX genotypes highlighted many variants, genes and pathways that could be contributing to susceptibility to disease; however no tests reached significance after adjusting for multiple testing. The genes most associated (PDZK1, DCLK3, SLC33A1 and BLOC1S2) appear to be biologically relevant and would be ideal candidates for further study. Additionally, just under half of the variants that are significant associated with LOAD were genotyped on the NeuroX and two of these were significantly associated with sEOAD after correcting for multiple testing (rs3851179 and rs3764650). The genotypes of all the variants highlighted would need to be verified before their functionalities were investigated further.

Late onset Alzheimer’s disease (LOAD) is the commonest form of dementia, affecting approximately 850,000 patients in the UK alone, predicted to exceed one million by 2025. The cause of LOAD is complex, but several large Genome Wide Association Studies have highlighted 21 genetic loci associated with this devastating disease and the ATP-Binding Cassette Protein, family A, member 7 (ABCA7) is one of these genetic loci. However, the exact reasons behind this association are still unknown, focusing work on identifying functional, pathogenic mutations within this locus. A total of 240 exonic variations within ABCA7 were therefore annotated in order to identify ones potentially altering the functionality of ABCA7. A total of five variants were predicted to be damaging by in silico annotation tools: rs3752233; rs59851484; rs3752237; rs114782266 and a novel mutation at genomic position 19:1056958. These were genotyped in the ARUK DNA Bank resource and three (rs59851484, rs3752239 and 19:1056958) showed tentative association with LOAD. However, lack of power in this study prevented any definitive associations from being formed. A further two variants were examined within functional cell assays. rs881768 had been predicted to affect the splicing of the ABCA7 protein and appeared to do so within minigene cellular assays. However, this did not appear to be the case when RNA from brain tissue harbouring this variation was examined. rs2020000 was examined through the dual luciferase assays, with the minor allele seeming to down regulate the reporter protein by approximately 30% (p < 0.02) in these in vitro assays. Functional variations within the ABCA7 locus do play a role in LOAD risk and improvements within functional databases and annotation programmes will assist in identifying these causative mutations, in order to put a halt to LOAD, as well as other destructive complex disorders.

Little is known about the molecular biology of late onset Alzheimer’s disease (LOAD), the most common dementia in the elderly. Genetic loci associated with LOAD have been identified through genome-wide association studies (GWAS). However, the functional variants responsible for the observed GWAS association at each of the loci remain unknown. The aim of this project was to identify and assess potential functional rare variants at three associated loci, CD2AP, EPHA1 and CD33. Target enriched, pooled sequencing of 96 post-mortem confirmed LOAD patient samples was used to identify 1273 variants within the three GWAS loci. Variants were prioritised using a combination of in silico functional annotation and putative disease association. Disease association was assessed through comparison to an independent, imputed LOAD GWAS dataset (2067 cases, 7376 controls). 18 coding and untranslated region variants and 9 noncoding variants were prioritised for further investigation. Potential splicing variants in CD2AP (6:47544253A > G) and EPHA1 (rs6967117) were assessed using minigene assays, although neither were found to influence splicing products in vivo. Five untranslated variants from the three genes and a frameshift variant in CD33 (rs201074739) were assessed for potential cis-regulatory consequences using allelic expression imbalance in brain tissues and B-lymphoblast cell lines. Only the frameshift variant displayed significant allelic expression imbalance and was found to be targeted for nonsense-mediated decay. None of the prioritised variants investigated were both functional and significantly associated with LOAD. However, pooled next generation sequencing using target enrichment successfully identified potential functional rare variants in CD2AP, EPHA1 and CD33. Rare variants do have a role to play in late onset Alzheimer’s disease. With the development of additional functional databases and improvements imputing rare variants from GWAS datasets, the combined prioritisation strategy used in this thesis will be useful for similar studies investigating causal GWAS variants.

Alzheimer's disease is the most common form (~60-80%) of dementia, currently affecting approximately half a million people in the UK and ~30 million people worldwide. The autosomal dominant form of AD represents a small proportion (~1-2%) of AD cases and is genetically well characterised. The vast majority of AD cases that show symptoms later in life (>65 years of age) are genetically complex. This type of AD, also known as late onset Alzheimer's disease (LOAD) disease, is still highly heritable with an estimated heritability of up to 76% (Gatz et al., 2006). Unfortunately, there is no cure for this devastating disease. Investigating genetic factors influencing the risk of LOAD is imperative for development of effective therapeutic treatments and more accurate diagnosis. A cross-platform comparison of four Genome-wide association studies (GWAS) was performed in an effort to identify novel genetic associations with LOAD (Chapter 3). A TRIM15 SNP rs929156 demonstrated significant evidence of association with LOAD with a p-value approaching genome-wide significance (p = 8.77 x 10-8) and an odds ratio that showed consistent effect on risk (OR = 1.1, p = 0.03). Within this chapter, a bio-informatic program to automate the process of GWAS meta-analysis taking into account linkage disequilibrium (LD) is also presented. Subsequently two fragments of the TRIM15 gene (including both 5’ and 3’ end flanking regions) were sequenced using the ABI SOLiDTM next generation sequencing technology. This was a pilot study using a DNA pooling strategy to determine whether this region harbours multiple rare variants which are associated with the disease (Chapter 4). Lastly, a candidate gene study combined with whole genome analysis was performed in an effort to search for genetic variants influencing human ageing using LOAD GWAS data (Chapter 5).