Use of phenotypic covariates in association analysis by sequential addition of cases

Genome-wide Association for Major Depression Through Age at Onset Stratification: Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium

BACKGROUND

Major depressive disorder (MDD) is a disabling mood disorder, and despite a known heritable component, a large meta-analysis of genome-wide association studies revealed no replicable genetic risk variants. Given prior evidence of heterogeneity by age at onset in MDD, we tested whether genome-wide significant risk variants for MDD could be identified in cases subdivided by age at onset.

METHODS

Discovery case-control genome-wide association studies were performed where cases were stratified using increasing/decreasing age-at-onset cutoffs; significant single nucleotide polymorphisms were tested in nine independent replication samples, giving a total sample of 22,158 cases and 133,749 control subjects for subsetting. Polygenic score analysis was used to examine whether differences in shared genetic risk exists between earlier and adult-onset MDD with commonly comorbid disorders of schizophrenia, bipolar disorder, Alzheimer's disease, and coronary artery disease.

RESULTS

We identified one replicated genome-wide significant locus associated with adult-onset (>27 years) MDD (rs7647854, odds ratio: 1.16, 95% confidence interval: 1.11-1.21, p = 5.2 × 10^-11). Using polygenic score analyses, we show that earlier-onset MDD is genetically more similar to schizophrenia and bipolar disorder than adult-onset MDD.

CONCLUSIONS

We demonstrate that using additional phenotype data previously collected by genetic studies to tackle phenotypic heterogeneity in MDD can successfully lead to the discovery of genetic risk factor despite reduced sample size. Furthermore, our results suggest that the genetic susceptibility to MDD differs between adult- and earlier-onset MDD, with earlier-onset cases having a greater genetic overlap with schizophrenia and bipolar disorder.

Assessing the validity of asthma associations for eight candidate genes and age at diagnosis effects

BACKGROUND

Before the advent of genome-wide association studies (GWAS), ADAM33, ADRB2, CD14, MS4A2 (alias FCER1B), IL13, IL4, IL4R, and TNF constituted the most replicated non-HLA candidate genes with asthma and related traits. However, except for the IL13-IL4 region, none of these genes have been found in close proximity of genome-wide significant hits among GWAS for asthma or related traits. Here we aimed to assess the reproducibility of these asthma associations and to test if associations were more evident considering the effect of age at diagnosis.

METHODOLOGY/PRINCIPAL FINDINGS

We systematically evaluated 286 common single nucleotide polymorphisms (SNPs) of these 8 genes in a sample of 1,865 unrelated Spanish individuals (606 asthmatics and 1,259 controls). We found that variants at MS4A2, IL4R and ADAM33 genes demonstrated varying association effects with the age at diagnosis of asthma, with 10 SNPs showing study-wise significance after the multiple comparison adjustment. In addition, in silico replication with GWAS data supported the association of IL4R.

CONCLUSIONS/SIGNIFICANCE

Our results support the important role of MS4A2, IL4R and ADAM33 genes in asthma and/or atopy susceptibility. However, additional studies in larger samples sets are needed to firmly implicate these genes in asthma susceptibility, and also to identify the causal variation underlying the associations found.

Dissecting the genetic heterogeneity of depression through age at onset

Genome-wide studies in major depression have identified few replicated associations, potentially due to heterogeneity within the disorder. Several studies have suggested that age at onset (AAO) can distinguish sub-types of depression with specific heritable components. This paper investigates the role of AAO in the genetic susceptibility for depression using genome-wide association data on 2,746 cases and 1,594 screened controls from the RADIANT studies, with replication performed in 1,471 cases and 1,403 controls from two Munich studies. Three methods were used to analyze AAO: First a time-to-event analysis with controls censored, secondly comparing controls to case-subsets defined using AAO cut-offs, and lastly analyzing AAO as a quantitative trait. In the time-to-event analysis three SNPs reached suggestive significance (P < 5E-06), overlapping with the original case-control analysis of this study. In a case-control analysis using AAO thresholds, SNPs in 10 genomic regions showed suggestive association though again none reached genome-wide significance. Lastly, case-only analysis of AAO as a quantitative trait resulted in 5 SNPs reaching suggestive significance. Sex specific analysis was performed as a secondary analysis, yielding one SNP reaching genome-wide significance in early-onset males. No SNPs achieved significance in the replication study after correction for multiple testing. Analysis of AAO as a quantitative trait did suggest that, across all SNPs, common genetic variants explained a large proportion of the variance (51%, P = 0.04). This study provides the first focussed analysis of the genetic contribution to AAO in depression, and establishes a statistical framework that can be applied to a quantitative trait underlying any disorder.

A subset-based approach improves power and interpretation for the combined analysis of genetic association studies of heterogeneous traits

Pooling genome-wide association studies (GWASs) increases power but also poses methodological challenges because studies are often heterogeneous. For example, combining GWASs of related but distinct traits can provide promising directions for the discovery of loci with small but common pleiotropic effects. Classical approaches for meta-analysis or pooled analysis, however, might not be suitable for such analysis because individual variants are likely to be associated with only a subset of the traits or might demonstrate effects in different directions. We propose a method that exhaustively explores subsets of studies for the presence of true association signals that are in either the same direction or possibly opposite directions. An efficient approximation is used for rapid evaluation of p values. We present two illustrative applications, one for a meta-analysis of separate case-control studies of six distinct cancers and another for pooled analysis of a case-control study of glioma, a class of brain tumors that contains heterogeneous subtypes. Both the applications and additional simulation studies demonstrate that the proposed methods offer improved power and more interpretable results when compared to traditional methods for the analysis of heterogeneous traits. The proposed framework has applications beyond genetic association studies.

TDT-HET: a new transmission disequilibrium test that incorporates locus heterogeneity into the analysis of family-based association data

BACKGROUND

Locus heterogeneity is one of the most documented phenomena in genetics. To date, relatively little work had been done on the development of methods to address locus heterogeneity in genetic association analysis. Motivated by Zhou and Pan's work, we present a mixture model of linked and unlinked trios and develop a statistical method to estimate the probability that a heterozygous parent transmits the disease allele at a di-allelic locus, and the probability that any trio is in the linked group. The purpose here is the development of a test that extends the classic transmission disequilibrium test (TDT) to one that accounts for locus heterogeneity.

RESULTS

Our simulations suggest that, for sufficiently large sample size (1000 trios) our method has good power to detect association even the proportion of unlinked trios is high (75%). While the median difference (TDT-HET empirical power - TDT empirical power) is approximately 0 for all MOI, there are parameter settings for which the power difference can be substantial. Our multi-locus simulations suggest that our method has good power to detect association as long as the markers are reasonably well-correlated and the genotype relative risk are larger. Results of both single-locus and multi-locus simulations suggest our method maintains the correct type I error rate.Finally, the TDT-HET statistic shows highly significant p-values for most of the idiopathic scoliosis candidate loci, and for some loci, the estimated proportion of unlinked trios approaches or exceeds 50%, suggesting the presence of locus heterogeneity.

CONCLUSIONS

We have developed an extension of the TDT statistic (TDT-HET) that allows for locus heterogeneity among coded trios. Benefits of our method include: estimates of parameters in the presence of heterogeneity, and reasonable power even when the proportion of linked trios is small. Also, we have extended multi-locus methods to TDT-HET and have demonstrated that the empirical power may be high to detect linkage. Last, given that we obtain PPBs, we conjecture that the TDT-HET may be a useful method for correctly identifying linked trios. We anticipate that researchers will find this property increasingly useful as they apply next-generation sequencing data in family based studies.

Ordered subset analysis for case-control studies

Genetic heterogeneity, which may manifest on a population level as different frequencies of a specific disease susceptibility allele in different subsets of patients, is a common problem for candidate gene and genome-wide association studies of complex human diseases. The ordered subset analysis (OSA) was originally developed as a method to reduce genetic heterogeneity in the context of family-based linkage studies. Here, we have extended a previously proposed method (OSACC) for applying the OSA methodology to case-control datasets. We have evaluated the type I error and power of different OSACC permutation tests with an extensive simulation study. Case-control datasets were generated under two different models by which continuous clinical or environmental covariates may influence the relationship between susceptibility genotypes and disease risk. Our results demonstrate that OSACC is more powerful under some disease models than the commonly used trend test and a previously proposed joint test of main genetic and gene-environment interaction effects. An additional unique benefit of OSACC is its ability to identify a more informative subset of cases that may be subjected to more detailed molecular analysis, such as DNA sequencing of selected genomic regions to detect functional variants in linkage disequilibrium with the associated polymorphism. The OSACC-identified covariate threshold may also improve the power of an additional dataset to replicate previously reported associations that may only be detectable in a fraction of the original and replication datasets. In summary, we have demonstrated that OSACC is a useful method for improving SNP association signals in genetically heterogeneous datasets.

Detection of susceptibility genes as modifiers due to subgroup differences in complex disease

Complex diseases invariably involve multiple genes and often exhibit variable symptom profiles. The extent to which disease symptoms, course, and severity differ between affected individuals may result from underlying genetic heterogeneity. Genes with modifier effects may or may not also influence disease susceptibility. In this study, we have simulated data in which a subset of cases differ by some effect size (ES) on a quantitative trait and are also enriched for a risk allele. Power to detect this 'pseudo-modifier' gene in case-only and case-control designs was explored blind to case substructure. Simulations involved 1000 iterations and calculations for 80% power at P<0.01 while varying the risk allele frequency (RAF), sample size (SS), ES, odds ratio (OR), and proportions of the case subgroups. With realistic values for the RAF (0.20), SS (3000) and ES (1), an OR of 1.7 is necessary to detect a pseudo-modifier gene. Unequal numbers of subjects in the case groups result in little decrement in power until the group enriched for the risk allele is <30% or >70% of the total case population. In practice, greater numbers of subjects and selection of a quantitative trait with a large range will provide researchers with greater power to detect a pseudo-modifier gene. However, even under ideal conditions, studies involving alleles with low frequencies or low ORs are usually underpowered for detection of a modifier or susceptibility gene. This may explain some of the inconsistent association results for many candidate gene studies of complex diseases.

No association of dysbindin with symptom factors of schizophrenia in an Irish case-control sample

Robust associations between the dysbindin gene (DTNBP1) and schizophrenia have been demonstrated in many but not all samples, and evidence that this gene particularly predisposes to negative symptoms in this illness has been presented. The current study sought to replicate the previously reported negative symptom associations in an Irish case-control sample. Association between dysbindin and schizophrenia has been established in this cohort, and a factor analysis of the assessed symptoms yielded three factors, Positive, Negative, and Schneiderian. The sequential addition method was applied using UNPHASED to assess the relationship between these symptom factors and the high-risk haplotype. No associations were detected for any of the symptom factors indicating that the dysbindin risk haplotype does not predispose to a particular group of symptoms in this sample. Several possibilities, such as differing risk haplotypes, may explain this finding.

Association study of SNAP25 and schizophrenia in Irish family and case-control samples

SNAP25 occurs on chromosome 20p12.2, which has been linked to schizophrenia in some samples, and recently linked to latent classes of psychotic illness in our sample. SNAP25 is crucial to synaptic functioning, may be involved in axonal growth and dendritic sprouting, and its expression may be decreased in schizophrenia. We genotyped 18 haplotype-tagging SNPs in SNAP25 in a sample of 270 Irish high-density families. Single marker and haplotype analyses were performed in FBAT and PDT. We adjusted for multiple testing by computing q values. Association was followed up in an independent sample of 657 cases and 411 controls. We tested for allelic effects on the clinical phenotype by using the method of sequential addition and 5 factor-derived scores of the OPCRIT. Nine of 18 SNPs had P values <0.05 in either FBAT or PDT for one or more definitions of illness. Several two-marker haplotypes were also associated. Subjects inheriting the risk alleles of the most significantly associated two-marker haplotype were likely to have higher levels of hallucinations and delusions. The most significantly associated marker, rs6039820, was observed to perturb 12 transcription-factor binding sites in in silico analyses. An attempt to replicate association findings in the case-control sample resulted in no SNPs being significantly associated. We observed robust association in both single marker and haplotype-based analyses between SNAP25 and schizophrenia in an Irish family sample. Although we failed to replicate this in an independent sample, this gene should be further tested in other samples.

The ordered transmission disequilibrium test: detection of modifier genes

We consider the problem of detection of modifier genes that lead to variations in a disease-related continuous variable (DRCV), such as the age of onset or a measure of disease severity, in a strategy of candidate genes. We propose a novel method, the ordered transmission disequilibrium test (OTDT), to test for a relation between the clinical heterogeneity expressed by a DRCV and marker genotypes of a candidate gene. The OTDT applies to trio families with one patients and his parents, all three genotyped at a bi-allelic marker M. The OTDT aims to find a critical value of the DRCV which separates the sample of families in two subsamples in which the transmission rates are significantly different. We investigate the power of the method by simulations under various genetic models and covariate distributions and compare it with a linear regression analysis.

Polarity at illness onset in bipolar I disorder and clinical course of illness

OBJECTIVES

Studies have suggested that episode polarity at illness onset in bipolar disorder may be predictive of some aspects of lifetime clinical characteristics. We here examine this possibility in a large, well-characterized sample of patients with bipolar I disorder.

METHODS

We assessed polarity at onset in patients with bipolar I disorder (N = 553) recruited as part of our ongoing studies of affective disorders. Lifetime clinical characteristics of illness were compared in patients who had a depressive episode at first illness onset (n = 343) and patients who had a manic episode at first illness onset (n = 210).

RESULTS

Several lifetime clinical features differed between patients according to the polarity of their onset episode of illness. A logistic regression analysis showed that the lifetime clinical features significantly associated with a depressive episode at illness onset in our sample were: an earlier age at illness onset; a predominantly depressive polarity during the lifetime; more frequent and more severe depressive episodes; and less prominent lifetime psychotic features.

CONCLUSIONS

Knowledge of pole of onset may help the clinician in providing prognostic information and management advice to an individual with bipolar disorder.

Ordered-subset analysis (OSA) for family-based association mapping of complex traits

Association analysis provides a powerful tool for complex disease gene mapping. However, in the presence of genetic heterogeneity, the power for association analysis can be low since only a fraction of the collected families may carry a specific disease susceptibility allele. Ordered-subset analysis (OSA) is a linkage test that can be powerful in the presence of genetic heterogeneity. OSA uses trait-related covariates to identify a subset of families that provide the most evidence for linkage. A similar strategy applied to genetic association analysis would likely result in increased power to detect association. Association in the presence of linkage (APL) is a family-based association test (FBAT) for nuclear families with multiple affected siblings that properly infers missing parental genotypes when linkage is present. We propose here APL-OSA, which applies the OSA method to the APL statistic to identify a subset of families that provide the most evidence for association. A permutation procedure is used to approximate the distribution of the APL-OSA statistic under the null hypothesis that there is no relationship between the family-specific covariate and the family-specific evidence for allelic association. We performed a comprehensive simulation study to verify that APL-OSA has the correct type I error rate under the null hypothesis. This simulation study also showed that APL-OSA can increase power relative to other commonly used association tests (APL, FBAT and FBAT with covariate adjustment) in the presence of genetic heterogeneity. Finally, we applied APL-OSA to a family study of age-related macular degeneration, where cigarette smoking was used as a covariate.

Identifying modifier genes of monogenic disease: strategies and difficulties

Substantial clinical variability is observed in many Mendelian diseases, so that patients with the same mutation may develop a very severe form of disease, a mild form or show no symptoms at all. Among the factors that may explain these differences in disease expression are modifier genes. In this paper, we review the different strategies that can be used to identify modifier genes and explain their advantages and limitations. We focus mainly on the statistical aspects but illustrate our points with a variety of examples from the literature.

The GABA transporter 1 (SLC6A1): a novel candidate gene for anxiety disorders

Recent evidence suggests that the GABA transporter 1 (GAT-1; SLC6A1) plays a role in the pathophysiology and treatment of anxiety disorders. In order to understand the impact of genetic variation within SLC6A1 on pathological anxiety, we performed a case–control association study with anxiety disorder patients with and without syndromal panic attacks. Using the method of sequential addition of cases, we found that polymorphisms in the 5′ flanking region of SLC6A1 are highly associated with anxiety disorders when considering the severity of syndromal panic attacks as phenotype covariate. Analysing the effect size of the association, we observed a constant increase in the odds ratio for disease susceptibility with an increase in panic severity (OR ~ 2.5 in severely affected patients). Nominally significant association effects were observed considering the entire patient sample. These data indicate a high load of genetic variance within SLC6A1 on pathological anxiety and highlight GAT-1 as a promising target for treatment of anxiety disorders with panic symptoms.

MEGF10 association with schizophrenia

BACKGROUND

The 5q21-31 region has been implicated as harboring risk genes for schizophrenia in many linkage studies. In our previous report of stepwise fine mapping of this region, the MEGF10 gene was one of the genes showing consistent associations in our screening subsample. In this study, we carried out independent replication and expression studies of the MEGF10 gene.

METHODS

Association studies with 8 SNPs in the MEGF10 gene were performed in the Irish case-control study of schizophrenia (ICCSS) sample (652 case and 617 control subjects). The expression of MEGF10 was also compared between healthy control subjects and schizophrenia patients using postmortem brain cDNA libraries.

RESULTS

In the ICCSS sample, associations with the disease were found in the same risk alleles and haplotypes as that observed in our fine-mapping studies. The major allele (A) of rs27388 was overrepresented in affected individuals (p = .0169), which remained significant after correction for multiple testing. In expression studies, MEGF10 had higher expression levels in the affected than the unaffected (p = .015). Schizophrenia patients with a 1/1 genotype at rs27388 had higher expressions than those patients with 1/2 and 2/2 genotypes (p = .0008).

CONCLUSIONS

Evidence from both association and expression studies suggests that MEGF10 is likely associated with schizophrenia. The major allele and 1/1 genotype at rs27388 impose higher risks for the disease.

An ordered subset approach to including covariates in the transmission disequilibrium test

Clinical heterogeneity of a disease may reflect an underlying genetic heterogeneity, which may hinder the detection of trait loci. Consequently, many statistical methods have been developed that allow for the detection of linkage and/or association signals in the presence of heterogeneity.

This report describes the work of two parallel investigations into similar approaches to ordered subset analysis, based on an observed covariate, in the framework of family-based association analysis using Genetic Analysis Workshop 15 simulated data.

With an appropriate choice of covariate, both approaches allow detection of two loci that are undetectable by the classical transmission-disequilibrium test. For a third locus, detectable by the classical transmission-disequilibrium test, a substantial increase of power of detection is shown.