Linkage analysis of anorexia and bulimia nervosa cohorts using selected behavioral phenotypes as quantitative traits or covariates.

To increase the likelihood of finding genetic variation conferring liability to eating disorders, we measured over 100 attributes thought to be related to liability to eating disorders on affected individuals from multiplex families and two cohorts: one recruited through a proband with anorexia nervosa (AN; AN cohort); the other recruited through a proband with bulimia nervosa (BN; BN cohort). By a multilayer decision process based on expert evaluation and statistical analysis, six traits were selected for linkage analysis (1): obsessionality (OBS), age at menarche (MENAR), and anxiety (ANX) for quantitative trait locus (QTL) linkage analysis; and lifetime minimum body mass index (BMI), concern over mistakes (CM), and food-related obsessions (OBF) for covariate-based linkage analysis. The BN cohort produced the largest linkage signals: for QTL linkage analysis, four suggestive signals: (for MENAR, at 10p13; for ANX, at 1q31.1, 4q35.2, and 8q13.1); for covariate-based linkage analyses, both significant and suggestive linkages (for BMI, one significant [4q21.1] and three suggestive [3p23, 10p13, 5p15.3]; for CM, two significant [16p13.3, 14q21.1] and three suggestive [4p15.33, 8q11.23, 10p11.21]; and for OBF, one significant [14q21.1] and five suggestive [4p16.1, 10p13.1, 8q11.23, 16p13.3, 18p11.31]). Results from the AN cohort were far less compelling: for QTL linkage analysis, two suggestive signals (for OBS at 6q21 and for ANX at 9p21.3); for covariate-based linkage analysis, five suggestive signals (for BMI at 4q13.1, for CM at 11p11.2 and 17q25.1, and for OBF at 17q25.1 and 15q26.2). Overlap between the two cohorts was minimal for substantial linkage signals.     

Meta-Analysis of Associations Between Hypothalamic-Pituitary-Adrenal Axis Genes and Risk of Posttraumatic Stress Disorder

The hypothalamic-pituitary-adrenal (HPA) axis has been of interest in attempts to identify genetic vulnerability for posttraumatic stress disorder (PTSD). Although numerous HPA-axis genes have been implicated in candidate gene studies, the findings are mixed and interpretation is limited by study design and methodological inconsistencies. To address these inconsistencies in the PTSD candidate gene literature, we conducted meta-analyses of HPA-related genes from both a traditional single nucleotide polymorphism (SNP)–level analysis and a gene-level analysis, using novel methods aggregating markers in the same gene. Database searches (PubMed and PsycINFO) identified 24 unique articles examining six HPA-axis genes in PTSD; analyses were conducted on four genes (ADCYAP1R1, CRHR1, FKBP5, NR3C1) that met study eligibility criteria (original research, human subjects, main effect association study of selected genes, PTSD as an outcome, trauma-exposed control group) and had sufficient data and number of studies for use in meta-analysis, within 20 unique articles. Findings from SNP-level analyses indicated that two variants (rs9296158 in FKBP5 and rs258747 in NR3C1) were nominally associated with PTSD, ps = .001 and .001, respectively, following multiple testing correction. At the gene level, significant relations between PTSD and both NR3C1 and FKBP5 were detected and robust to sensitivity analyses. Although study limitations exist (e.g., varied outcomes, inability to test moderators), taken together, these results provide support for FKBP5 and NR3C1 in risk for PTSD. Overall, this work highlights the utility of meta-analyses in resolving discrepancies in the literature and the value of adopting gene-level approaches to investigate the etiology of PTSD.     

Linkage analysis of Alzheimer disease with psychosis

Although a portion of risk for late-onset AD (LOAD) is attributable to APOE, the search for other loci is ongoing. The authors hypothesize that psychotic symptoms with LOAD (LOAD+P) identify a potentially more etiologically homogeneous form of AD. Linkage analysis of families with LOAD+P identified one significant and several suggestive novel linkage signals, which bolsters the conjecture of greater etiologic homogeneity.     

On optimal gene-based analysis of genome scans

Univariate analysis of markers has modest power when there are multiple causal variants within a gene. Under this scenario, combining the effects of all variants from a gene in a gene-wide statistic is thought to increase power. However, it is not really clear (1) what is the performance of most commonly used gene-wide methods for whole genome scans and (2) how scalable these methods are for more computationally intensive analyses, e.g. analysis of genome-wide sequence data. We attempt to answer these questions by using realistic simulations to assess the performance of a range of gene-based methods: (1) commonly used, e.g. VEGAS and GATES; (2) less commonly used, e.g. Simes, adaptive sum (aSUM), and kernel methods; and (3) a combination of univariate and multivariate tests we proposed for the analysis of markers in linkage disequilibrium. Simes is the fastest method and has good power for single causal variant models. aSUM method has good power for multiple causal variant models, especially at lower gene lengths. Our proposed statistic yields good power for all causal models. Given the extreme data volumes coming from sequencing studies, we recommend a two step analysis of genome scans. The initial step uses the very fast Simes procedure to flag possibly interesting genes. The second step refines interesting signals by using more computationally intensive methods, e.g. (1) aSUM for shorter and (2) VEGAS for larger gene lengths. Alternatively, genome scans can be analyzed using only our proposed method while sacrificing only a modest amount of power.     

Comparison of methods and sampling designs to test for association between rare variants and quantitative traits

Genome‐wide association studies succeeded in finding genetic variants associated with various phenotypes, but a large portion of the predicted genetic contribution to many traits remains unknown. One plausible explanation is that some missing variation is due to rare variants. Latest sequencing technology facilitates the investigation of such rare variants, but their statistical analysis remains challenging. For quantitative traits, a commonly used approach is to contrast the frequency of putatively functional rare variants between subjects in the two tails of the trait distribution. The contrast is usually performed by Fisher's exact or similar test. These tests are conservative as they discard trait rank information and are most useful under the unrealistic homogeneity assumption (i.e., variants have similar effects). We propose, and investigate via simulations, various designs for resequencing studies and statistical methods that incorporate information about rank, predicted function and allow for heterogeneity of effects. We propose designs which accommodate heterogeneity by sequencing both tails and the middle of the trait and novel statistical tests for trend, for heterogeneity and for a combination of the two. The conclusions of the simulations are four fold: (1) sequencing both tails and the middle of the trait distributions is desirable when heterogeneity is suspected, (2) trend and heterogeneity statistics should be used alongside other methods, (3) using rank information improves power over Fisher's exact test when the number of rare variants is not very large and (4) due to high misclassification rates, incorporating current predictions of a variant's function does not improve power. Genet. Epidemiol. 35: 226‐235, 2011. © 2011 Wiley‐Liss, Inc.     

Comparison of association methods for dense marker data

While data sets based on dense genome scans are becoming increasingly common, there are many theoretical questions that remain unanswered. How can a large number of markers in high linkage disequilibrium (LD) and rare disease variants be simulated efficiently? How should markers in high LD be analyzed: individually or jointly? Are there fast and simple methods to adjust for correlation of tests? What is the power penalty for conservative Bonferroni adjustments? Assuming that association scans are adequately powered, we attempt to answer these questions. Performance of single‐point and multipoint tests, and their hybrids, is investigated using two simulation designs. The first simulation design uses theoretically derived LD patterns. The second design uses LD patterns based on real data. For the theoretical simulations we used polychoric correlation as a measure of LD to facilitate simulation of markers in LD and rare disease variants. Based on the simulation results of the two studies, we conclude that statistical tests assuming only additive genotype effects (i.e. Armitage and especially multipoint T²) should be used cautiously due to their suboptimal power in certain settings. A false discovery rate (FDR)‐adjusted combination of tests for additive, dominant and recessive effects had close to optimal power. However, the common genotypic χ² test performed adequately and could be used in lieu of the FDR combination. While some hybrid methods yield (sometimes spectacularly) higher power they are computationally intensive. We also propose an “exact” method to adjust for multiple testing, which yields nominally higher power than the Bonferroni correction. Genet. Epidemiol. 2008. © 2008 Wiley‐Liss, Inc.     

The impact on estimations of genetic correlations by the use of super-normal,unscreened, and family-history screened controls in genome wide case–control studies

Traditionally, in normal case–control studies of disorder A, the controls are defined as those not developing the disorder. However, in genome wide association (GWA) studies, controls are sometimes (a) unscreened or (b) screened for both disorder A and disorder B, producing super-normal controls. Using simulations, we examine how the observed genetic correlations between two disorders (A and B) are influenced by the use of unscreened, normal, and super-normal controls. Normal controls produce unbiased estimates of the genetic correlation. However, unscreened and super-normal controls both bias upward the genetic correlations. The strength of the bias increases with increasing population prevalences for the two disorders. With super-normal controls, the absolute magnitude of bias is stronger when the true genetic correlation is low. The opposite is seen with the use of unscreened controls. Adding screening of first-degree relatives of controls substantially increases the bias in genetic correlations with super-normal controls but has minimal impact when controls are screened only for the relevant disease.     

Testing for Measured Gene-Environment Interaction: Problems with the use of Cross-Product Terms and a Regression Model Reparameterization Solution

The study of gene-environment interaction (G × E) has garnered widespread attention. The most common way to assess interaction effects is in a regression model with a G × E interaction term that is a product of the values specified for the genotypic (G) and environmental (E) variables. In this paper we discuss the circumstances under which interaction can be modeled as a product term and cases in which use of a product term is inappropriate and may lead to erroneous conclusions about the presence and nature of interaction effects. In the case of a binary coded genetic variant (as used in dominant and recessive models, or where the minor allele occurs so infrequently that it is not observed in the homozygous state), the regression coefficient corresponding to a significant interaction term reflects a slope difference between the two genotype categories and appropriately characterizes the statistical interaction between the genetic and environmental variables. However, when using a three-category polymorphic genotype, as is commonly done when modeling an additive effect, both false positive and false negative results can occur, and the nature of the interaction can be misrepresented. We present a reparameterized regression equation that accurately captures interaction effects without the constraints imposed by modeling interactions using a single cross-product term. In addition, we provide a series of recommendations for making conclusions about the presence of meaningful G × E interactions, which take into account the nature of the observed interactions and whether they map onto sensible genotypic models.