The importance of modelling heterogeneity in complex disease: application to NIMH Schizophrenia Genetics Initiative data

As for other complex diseases, linkage analyses of schizophrenia (SZ) have produced evidence for numerous chromosomal regions, with inconsistent results reported across studies. The presence of locus heterogeneity appears likely and may reduce the power of linkage analyses if homogeneity is assumed. In addition, when multiple heterogeneous datasets are pooled, inter-sample variation in the proportion of linked families (alpha) may diminish the power of the pooled sample to detect susceptibility loci, in spite of the larger sample size obtained. We compare the significance of linkage findings obtained using allele-sharing LOD scores (LOD(exp))-which assume homogeneity-and heterogeneity LOD scores (HLOD) in European American and African American NIMH SZ families. We also pool these two samples and evaluate the relative power of the LOD(exp) and two different heterogeneity statistics. One of these (HLOD-P) estimates the heterogeneity parameter alpha only in aggregate data, while the second (HLOD-S) determines alpha separately for each sample. In separate and combined data, we show consistently improved performance of HLOD scores over LOD(exp). Notably, genome-wide significant evidence for linkage is obtained at chromosome 10p in the European American sample using a recessive HLOD score. When the two samples are combined, linkage at the 10p locus also achieves genome-wide significance under HLOD-S, but not HLOD-P. Using HLOD-S, improved evidence for linkage was also obtained for a previously reported region on chromosome 15q. In linkage analyses of complex disease, power may be maximised by routinely modelling locus heterogeneity within individual datasets, even when multiple datasets are combined to form larger samples.

Rare genomic structural variants in complex disease: lessons from the replication of associations with obesity.

The limited ability of common variants to account for the genetic contribution to complex disease has prompted searches for rare variants of large effect, to partly explain the 'missing heritability'. Analyses of genome-wide genotyping data have identified genomic structural variants (GSVs) as a source of such rare causal variants. Recent studies have reported multiple GSV loci associated with risk of obesity. We attempted to replicate these associations by similar analysis of two familial-obesity case-control cohorts and a population cohort, and detected GSVs at 11 out of 18 loci, at frequencies similar to those previously reported. Based on their reported frequencies and effect sizes (OR≥25), we had sufficient statistical power to detect the large majority (80%) of genuine associations at these loci. However, only one obesity association was replicated. Deletion of a 220 kb region on chromosome 16p11.2 has a carrier population frequency of 2×10(-4) (95% confidence interval [9.6×10(-5)-3.1×10(-4)]); accounts overall for 0.5% [0.19%-0.82%] of severe childhood obesity cases (P = 3.8×10(-10); odds ratio = 25.0 [9.9-60.6]); and results in a mean body mass index (BMI) increase of 5.8 kg.m(-2) [1.8-10.3] in adults from the general population. We also attempted replication using BMI as a quantitative trait in our population cohort; associations with BMI at or near nominal significance were detected at two further loci near KIF2B and within FOXP2, but these did not survive correction for multiple testing. These findings emphasise several issues of importance when conducting rare GSV association, including the need for careful cohort selection and replication strategy, accurate GSV identification, and appropriate correction for multiple testing and/or control of false discovery rate. Moreover, they highlight the potential difficulty in replicating rare CNV associations across different populations. Nevertheless, we show that such studies are potentially valuable for the identification of variants making an appreciable contribution to complex disease.

Bullous pemphigoid: simple measures for a complex disease

Simple descriptive population data are potentially helpful in understanding how bullous pemphigoid (BP) originates and evolves over time. Before embarking with etiological correlations, artifacts and biases should be ruled out. Ideally, epidemiological data should be complemented by immunological and genetic analyses aimed at providing better insight into the causation and prognosis of BP.

Vaccination of cattle with the N terminus of LppQ of Mycoplasma mycoides subsp. mycoides results in type III immune complex disease upon experimental infection.

Contagious bovine pleuropneumonia (CBPP) is a serious respiratory disease of cattle caused by Mycoplasma mycoides subsp. mycoides. Current vaccines against CBPP induce short-lived immunity and can cause severe postvaccine reactions. Previous studies have identified the N terminus of the transmembrane lipoprotein Q (LppQ-N') of M. mycoides subsp. mycoides as the major antigen and a possible virulence factor. We therefore immunized cattle with purified recombinant LppQ-N' formulated in Freund's adjuvant and challenged them with M. mycoides subsp. mycoides. Vaccinated animals showed a strong seroconversion to LppQ, but they exhibited significantly enhanced postchallenge glomerulonephritis compared to the placebo group (P = 0.021). Glomerulonephritis was characterized by features that suggested the development of antigen-antibody immune complexes. Clinical signs and gross pathological scores did not significantly differ between vaccinated and placebo groups. These findings reveal for the first time the pathogenesis of enhanced disease as a result of antibodies against LppQ during challenge and also argue against inclusion of LppQ-N' in a future subunit vaccine for CBPP.

Nonlinear tests for genetic studies of complex disease

Linkage and association studies are major analytical tools to search for susceptibility genes for complex diseases. With the availability of large collection of single nucleotide polymorphisms (SNPs) and the rapid progresses for high throughput genotyping technologies, together with the ambitious goals of the International HapMap Project, genetic markers covering the whole genome will be available for genome-wide linkage and association studies. In order not to inflate the type I error rate in performing genome-wide linkage and association studies, multiple adjustment for the significant level for each independent linkage and/or association test is required, and this has led to the suggestion of genome-wide significant cut-off as low as 5 × 10 −7. Almost no linkage and/or association study can meet such a stringent threshold by the standard statistical methods. Developing new statistics with high power is urgently needed to tackle this problem. This dissertation proposes and explores a class of novel test statistics that can be used in both population-based and family-based genetic data by employing a completely new strategy, which uses nonlinear transformation of the sample means to construct test statistics for linkage and association studies. Extensive simulation studies are used to illustrate the properties of the nonlinear test statistics. Power calculations are performed using both analytical and empirical methods. Finally, real data sets are analyzed with the nonlinear test statistics. Results show that the nonlinear test statistics have correct type I error rates, and most of the studied nonlinear test statistics have higher power than the standard chi-square test. This dissertation introduces a new idea to design novel test statistics with high power and might open new ways to mapping susceptibility genes for complex diseases. ^

The importance of modelling heterogeneity in complex disease: application to NIMH Schizophrenia Genetics Initiative data

As for other complex diseases, linkage analyses of schizophrenia (SZ) have produced evidence for numerous chromosomal regions, with inconsistent results reported across studies. The presence of locus heterogeneity appears likely and may reduce the power of linkage analyses if homogeneity is assumed. In addition, when multiple heterogeneous datasets are pooled, intersample variation in the proportion of linked families ( a) may diminish the power of the pooled sample to detect susceptibility loci, in spite of the larger sample size obtained. We compare the significance of linkage. findings obtained using allele- sharing LOD scores ( LODexp) - which assume homogeneity - and heterogeneity LOD scores ( HLOD) in European American and African American NIMH SZ families. We also pool these two samples and evaluate the relative power of the LODexp and two different heterogeneity statistics. One of these ( HLOD- P) estimates the heterogeneity parameter a only in aggregate data, while the second ( HLOD- S) determines a separately for each sample. In separate and combined data, we show consistently improved performance of HLOD scores over LODexp. Notably, genome-wide significant evidence for linkage is obtained at chromosome 10p in the European American sample using a recessive HLOD score. When the two samples are combined, linkage at the 10p locus also achieves genome-wide significance under HLOD- S, but not HLOD- P. Using HLOD- S, improved evidence for linkage was also obtained for a previously reported region on chromosome 15q. In linkage analyses of complex disease, power may be maximised by routinely modelling locus heterogeneity within individual datasets, even when multiple datasets are combined to form larger samples.

Bayesian mixtures for modelling complex medical data : a case study in Parkinson’s disease

Mixture models are a flexible tool for unsupervised clustering that have found popularity in a vast array of research areas. In studies of medicine, the use of mixtures holds the potential to greatly enhance our understanding of patient responses through the identification of clinically meaningful clusters that, given the complexity of many data sources, may otherwise by intangible. Furthermore, when developed in the Bayesian framework, mixture models provide a natural means for capturing and propagating uncertainty in different aspects of a clustering solution, arguably resulting in richer analyses of the population under study. This thesis aims to investigate the use of Bayesian mixture models in analysing varied and detailed sources of patient information collected in the study of complex disease. The first aim of this thesis is to showcase the flexibility of mixture models in modelling markedly different types of data. In particular, we examine three common variants on the mixture model, namely, finite mixtures, Dirichlet Process mixtures and hidden Markov models. Beyond the development and application of these models to different sources of data, this thesis also focuses on modelling different aspects relating to uncertainty in clustering. Examples of clustering uncertainty considered are uncertainty in a patient’s true cluster membership and accounting for uncertainty in the true number of clusters present. Finally, this thesis aims to address and propose solutions to the task of comparing clustering solutions, whether this be comparing patients or observations assigned to different subgroups or comparing clustering solutions over multiple datasets. To address these aims, we consider a case study in Parkinson’s disease (PD), a complex and commonly diagnosed neurodegenerative disorder. In particular, two commonly collected sources of patient information are considered. The first source of data are on symptoms associated with PD, recorded using the Unified Parkinson’s Disease Rating Scale (UPDRS) and constitutes the first half of this thesis. The second half of this thesis is dedicated to the analysis of microelectrode recordings collected during Deep Brain Stimulation (DBS), a popular palliative treatment for advanced PD. Analysis of this second source of data centers on the problems of unsupervised detection and sorting of action potentials or "spikes" in recordings of multiple cell activity, providing valuable information on real time neural activity in the brain.

Epigenetics and migraine; complex mitochondrial interactions contributing to disease susceptibility

Migraine is a common neurological disorder classified by the World Health Organisation (WHO) as one of the top twenty most debilitating diseases in the developed world. Current therapies are only effective for a proportion of sufferers and new therapeutic targets are desperately needed to alleviate this burden. Recently the role of epigenetics in the development of many complex diseases including migraine has become an emerging topic. By understanding the importance of acetylation, methylation and other epigenetic modifications, it then follows that this modification process is a potential target to manipulate epigenetic status with the goal of treating disease. Bisulphite sequencing and methylated DNA immunoprecipitation have been used to demonstrate the presence of methylated cytosines in the human D-loop of mitochondrial DNA (mtDNA), proving that the mitochondrial genome is methylated. For the first time, it has been shown that there is a difference in mtDNA epigenetic status between healthy controls and those with disease, especially for neurodegenerative and age related conditions. Given co-morbidities with migraine and the suggestive link between mitochondrial dysfunction and the lowered threshold for triggering a migraine attack, mitochondrial methylation may be a new avenue to pursue. Creative thinking and new approaches are needed to solve complex problems and a systems biology approach, where multiple layers of information are integrated is becoming more important in complex disease modelling.

Using genome-Wwide complex trait analysis to quantify 'missing heritability' in Parkinson's disease

Genome-wide association studies (GWASs) have been successful at identifying single-nucleotide polymorphisms (SNPs) highly associated with common traits; however, a great deal of the heritable variation associated with common traits remains unaccounted for within the genome. Genome-wide complex trait analysis (GCTA) is a statistical method that applies a linear mixed model to estimate phenotypic variance of complex traits explained by genome-wide SNPs, including those not associated with the trait in a GWAS. We applied GCTA to 8 cohorts containing 7096 case and 19 455 control individuals of European ancestry in order to examine the missing heritability present in Parkinson's disease (PD). We meta-analyzed our initial results to produce robust heritability estimates for PD types across cohorts. Our results identify 27% (95% CI 17-38, P = 8.08E - 08) phenotypic variance associated with all types of PD, 15% (95% CI -0.2 to 33, P = 0.09) phenotypic variance associated with early-onset PD and 31% (95% CI 17-44, P = 1.34E - 05) phenotypic variance associated with late-onset PD. This is a substantial increase from the genetic variance identified by top GWAS hits alone (between 3 and 5%) and indicates there are substantially more risk loci to be identified. Our results suggest that although GWASs are a useful tool in identifying the most common variants associated with complex disease, a great deal of common variants of small effect remain to be discovered. © Published by Oxford University Press 2012.

Entropy Based Approach to Finding Interacting Genes Responsible for Complex Human Disease

2000 Mathematics Subject Classification: 62P10, 92D10, 92D30, 94A17, 62L10.

Probabilistic subgroup identification using Bayesian finite mixture modelling : a case study in Parkinson’s disease phenotype identification

This article explores the use of probabilistic classification, namely finite mixture modelling, for identification of complex disease phenotypes, given cross-sectional data. In particular, if focuses on posterior probabilities of subgroup membership, a standard output of finite mixture modelling, and how the quantification of uncertainty in these probabilities can lead to more detailed analyses. Using a Bayesian approach, we describe two practical uses of this uncertainty: (i) as a means of describing a person’s membership to a single or multiple latent subgroups and (ii) as a means of describing identified subgroups by patient-centred covariates not included in model estimation. These proposed uses are demonstrated on a case study in Parkinson’s disease (PD), where latent subgroups are identified using multiple symptoms from the Unified Parkinson’s Disease Rating Scale (UPDRS).

Dietetic practice by Australian and Canadian dietitians with people with Parkinson's disease

Although Parkinson’s disease (PD) is a complex disease for which appropriate nutrition management is important, limited evidence is currently available to support dietetic practice. Existing PD-specific guidelines do not span all phases of the Nutrition Care Process (NCP). This study aimed to document PD-specific nutrition management practice by Australian and Canadian dietitians. DAA members and PEN subscribers were invited to participate in an online survey (late 2011). Eighty-four dietitians responded (79.8% Australian). The majority (70.2%) worked in the clinical setting. Existing non-PD guidelines were used by 52.4% while 53.6% relied on self-initiated literature reviews. Weight loss/malnutrition, protein intake, dysphagia and constipation were common issues in all NCP phases. Respondents also requested more information/evidence for these topics. Malnutrition screening (82.1%) and assessment (85.7%) were routinely performed. One-third did not receive referrals for weight loss for overweight/obesity. Protein intake meeting gender/age recommendations (69.0%), and high energy/high protein diets to manage malnutrition (82.1%) were most commonly used. Constipation management was through high fibre diets (86.9%). Recommendations for spacing of meals and PD medications varied with 34.5% not making recommendations. Nutritional diagnosis (70.2%) and stage of disease (61.9%) guided monitoring frequency. Common outcome measures included appropriate weight change (97.6%) and regular bowel movements (88.1%). With limited PD-specific guidance, dietitians applied best available evidence for other groups with similar issues. Dietitians requested evidence-based guidelines specifically for the nutritional management of PD. Guideline development should focus on those areas reported as commonly encountered. This process can identify the gaps in evidence to guide future research.

Studies of the genetic epidemiology of cardiovascular disease: focus on inflammatory candidate genes

Cardiovascular disease (CVD) is a complex disease with multifactorial aetiology. Both genetic and environmental factors contribute to the disease risk. The lifetime risk for CVD differs markedly between men and women, men being at increased risk. Inflammatory reaction contributes to the development of the disease by promoting atherosclerosis in artery walls. In the first part of this thesis, we identified several inflammatory related CVD risk factors associating with the amount of DNA from whole blood samples, indicating a potential source of bias if a genetic study selects the participants based on the available amount of DNA. In the following studies, this observation was taken into account by applying whole genome amplification to samples otherwise subjected to exclusion due to very low DNA yield. We continued by investigating the contribution of inflammatory genes to the risk for CVD separately in men and women, and looked for sex-genotype interaction. In the second part, we explored a new candidate gene and its role in the risk for CVD. Selenoprotein S (SEPS1) is a membrane protein residing in the endoplasmic reticulum where it participates in retro-translocation of unfolded proteins to cytosolic protein degradation. Previous studies have indicated that SEPS1 protects cells from oxidative stress and that variations in the gene are associated with circulating levels of inflammatory cytokines. In our study, we identified two variants in the SEPS1 gene, which associated with coronary heart disease and ischemic stroke in women. This is, to our knowledge, the first study suggesting a role of SEPS1 in the risk for CVD after extensively examining the variation within the gene region. In the third part of this thesis, we focused on a set of seven genes (angiotensin converting enzyme, angiotensin II receptor type I, C-reactive protein (CRP), and fibrinogen alpha-, beta-, and gamma-chains (FGA, FGB, FGG)) related to inflammatory cytokine interleukin 6 (IL6) and their association with the risk for CVD. We identified one variant in the IL6 gene conferring risk for CVD in men and a variant pair from IL6 and FGA genes associated with decreased risk. Moreover, we identified and confirmed an association between a rare variant in the CRP gene and lower CRP levels, and found two variants in the FGA and FGG genes associating with fibrinogen. The results from this third study suggest a role for the interleukin 6 pathway genes in the pathogenesis of CVD and warrant further studies in other populations. In addition to the IL6 -related genes, we describe in this thesis several sex-specific associations in other genes included in this study. The majority of the findings were evident only in women encouraging other studies of cardiovascular disease to include and analyse women separately from men.