Testing and estimating gene-environment interactions in family-based association studies

Summary .   We propose robust and efficient tests and estimators for gene–environment/gene–drug interactions in family-based association studies in which haplotypes, dichotomous/quantitative phenotypes, and complex exposure/treatment variables are analyzed. Using causal inference methodology, we show that the tests and estimators are robust against unmeasured confounding due to population admixture and stratification, provided that Mendel's law of segregation holds and that the considered exposure/treatment variable is not affected by the candidate gene under study. We illustrate the practical relevance of our approach by an application to a chronic obstructive pulmonary disease study. The data analysis suggests a gene–environment interaction between a single nucleotide polymorphism in the Serpine2 gene and smoking status/pack-years of smoking. Simulation studies show that the proposed methodology is sufficiently powered for realistic sample sizes and that it provides valid tests and effect size estimators in the presence of admixture and stratification.     

Estimation and testing of gene-environment interactions in family-based association studies

Gene-environment interactions are of interest in genetic association studies for several reasons. First, the power to detect genetic effects may be substantially decreased if those effects differ according to environmental exposure and if no account is taken of this interaction with environmental exposure in the analysis. Second, such interactions may indicate a phenomenon of genuine biological interest (whereby a particular genetic effect operates only in the presence of an environmental trigger, or vice versa), understanding of which can lead us to a greater understanding of possible mechanisms and pathways in disease progression. Here I discuss the testing and estimation of gene-environment interactions via the case/pseudocontrol and related approaches. As originally proposed, the case/pseudocontrol approach applies to case/parents trios with no missing genotype data. I discuss some recent extensions that allow larger pedigree structures with some missing genotype data and present computer simulations to compare the performance of several competing approaches.     

A combinatorial approach to detecting gene-gene and gene-environment interactions in family studies

Widespread multifactor interactions present a significant challenge in determining risk factors of complex diseases. Several combinatorial approaches, such as the multifactor dimensionality reduction (MDR) method, have emerged as a promising tool for better detecting gene-gene (G x G) and gene-environment (G x E) interactions. We recently developed a general combinatorial approach, namely the generalized multifactor dimensionality reduction (GMDR) method, which can entertain both qualitative and quantitative phenotypes and allows for both discrete and continuous covariates to detect G x G and G x E interactions in a sample of unrelated individuals. In this article, we report the development of an algorithm that can be used to study G x G and G x E interactions for family-based designs, called pedigree-based GMDR (PGMDR). Compared to the available method, our proposed method has several major improvements, including allowing for covariate adjustments and being applicable to arbitrary phenotypes, arbitrary pedigree structures, and arbitrary patterns of missing marker genotypes. Our Monte Carlo simulations provide evidence that the PGMDR method is superior in performance to identify epistatic loci compared to the MDR-pedigree disequilibrium test (PDT). Finally, we applied our proposed approach to a genetic data set on tobacco dependence and found a significant interaction between two taste receptor genes (i.e., TAS2R16 and TAS2R38) in affecting nicotine dependence.     

Modeling gene-environment interactions in malignant melanoma

Many cancers are the pathological consequence of environmentally initiated disruptions to cellular genetic control mechanisms. For most cancers the relevant environmental carcinogens have not been identified, but one major exception is cutaneous malignant melanoma, for which the primary environmental agent is solar ultraviolet (UV) radiation. Hence, melanomagenesis represents a potential model of detrimental gene-environment interaction. Although the underlying genetic basis of melanoma is currently being elucidated, fundamental questions concerning UV and the mechanisms by which it operates remain unanswered. Significant progress has recently been made in creating UV-responsive, genetically tractable mouse models of melanoma that accurately recapitulate human disease. These models are providing novel insights into how the genome and environment interact in vivo.     

Progress in the epidemiological understanding of gene-environment interactions in major diseases: cancer

Cancer epidemiology has undergone marked development since the 1950s. One of the most spectacular and specific contributions was the demonstration of the massive effect of smoking on the occurrence of lung, larynx, and bladder cancer. Major chemical, physical, and biological carcinogenic agents have been identified in the working environment and in the overall environment. The chain of events from environmental exposures to cancer requires hundreds of polymorphic genes coding for proteins involved in the transport and metabolism of xenobiotics, or in repair, or in an immune or inflammatory response. The multifactorial and multistage characteristics of cancer create the theoretical conditions for statistical interactions that have been exceptionally detected. Over the last two decades, a considerable mass of data has been generated, mostly addressing the interactions between smoking and xenobiotic-metabolizing enzymes in smoking-related cancers. They were sometimes considered disappointing, but they actually brought a lot of information and raised many methodological issues. In parallel, the number of polymorphisms that can be considered candidate per function increased so much that multiple testing has become a major issue, and genome wide-screening approaches have more and more gained in interest. Facing the resulting complexity, some instruments are being set up: our studies are now equipped with carefully sampled biological collections, high-throughput genotyping systems are becoming available, work on statistical methodologies is ongoing, bioinformatics databases are growing larger and access to them is becoming simpler; international consortiums are being organized. The roles of environmental and genetic factors are being jointly elucidated. The basic rules of epidemiology, which are demanding with respect to sampling, with respect to the histological and molecular criteria for cancer classification, with respect to the evaluation of environmental exposures, their timeframes, quantification and covariables, with respect to study size and with respect to the rigor of multivariate analyses, are more pertinent than ever before.     

Bioremediation of radioactive waste: radionuclide-microbe interactions in laboratory and field-scale studies

Given the scale of the contamination associated with 60 years of global nuclear activity, and the inherent high financial and environmental costs associated with invasive physical and chemical clean-up strategies, there is an unparalleled interest in new passive in situ bioremediation processes for sites contaminated with nuclear waste. Many of these processes rely on successfully harnessing newly discovered natural biogeochemical cycles for key radionuclides and fission products. Recent advances have been made in understanding the microbial colonization of radioactive environments and the biological basis of microbial transformations of radioactive waste in these settings.     

Subtle gene-environment interactions driving paranoia in daily life

It has been suggested that genes impact on the degree to which minor daily stressors cause variation in the intensity of subtle paranoid experiences. The objective of the present study was to test the hypothesis that catechol- O -methyltransferase (COMT) Val¹⁵⁸Met and brain-derived neurotrophic factor (BDNF) Val⁶⁶Met in part mediate genetic effects on paranoid reactivity to minor stressors. In a general population sample of 579 young adult female twins, on the one hand, appraisals of (1) event-related stress and (2) social stress and, on the other hand, feelings of paranoia in the flow of daily life were assessed using momentary assessment technology for five consecutive days. Multilevel regression analyses were used to examine moderation of daily life stress-induced paranoia by COMT Val¹⁵⁸Met and BDNF Val⁶⁶Met genotypes. Catechol- O -methyltransferase Val carriers displayed more feelings of paranoia in response to event stress compared with Met carriers. Brain-derived neurotrophic factor Met carriers showed more social-stress-induced paranoia than individuals with the Val/Val genotype. Thus, paranoia in the flow of daily life may be the result of gene–environment interactions that can be traced to different types of stress being moderated by different types of genetic variation.     

The value of studying gene-environment interactions in culturally diverse populations

Cardiovascular disease (CVD) is the leading cause of death and disability in the world. It is anticipated that CVD will reach pandemic proportions by the year 2020. Although the major causes of CVD are well documented and explain the majority of cardiovascular deaths, the prevalence of conventional cardiovascular risk factors vary substantially across diverse cultural groups. These differences are attributed to cultural or genetic differences or to interactions between genes and environmental factors. Substantial efforts have been invested in determining the genetic influences on CVD development, and it is unlikely that a single gene is responsible for the development of atherosclerotic CVD or its classical risk factors such as blood pressure or plasma lipids. It is more plausible that multiple genes, acting either alone or in concert with one another, which display effect modification in the presence of certain environmental factors, are modestly associated with CVD or its main risk factors. Following this hypothesis, studying populations with diversity in environmental factors may increase the discovery potential of gene-environmental interactions. In this brief review, the advantage of studying gene-environment interactions across heterogeneous groups with diverse lifestyles is discussed.     

Simulating gene-gene and gene-environment interactions in complex diseases: Gene-Environment iNteraction Simulator 2

ABSTRACT: BACKGROUND: The analysis of complex diseases is an important problem in human genetics. Because multifactoriality isexpected to play a pivotal role, many studies are currently focused on collecting information on the geneticand environmental factors that potentially influence these diseases. However, there is still a lack of efficientand thoroughly tested statistical models that can be used to identify implicated features and theirinteractions. Simulations using large biologically realistic data sets with known gene-gene andgene-environment interactions that influence the risk of a complex disease are a convenient and useful wayto assess the performance of statistical methods. RESULTS: The Gene-Environment iNteraction Simulator 2 (GENS2) simulates interactions among two genetic and oneenvironmental factor and also allows for epistatic interactions. GENS2 is based on data with realisticpatterns of linkage disequilibrium, and imposes no limitations either on the number of individuals to besimulated or on number of non-predisposing genetic/environmental factors to be considered. The GENS2tool is able to simulate gene-environment and gene-gene interactions. To make the Simulator more intuitive,the input parameters are expressed as standard epidemiological quantities. GENS2 is written in Pythonlanguage and takes advantage of operators and modules provided by the simuPOP simulation environment.It can be used through a graphical or a command-line interface and is freely available fromhttp://sourceforge.net/projects/gensim. The software is released under the GNU General Public Licenseversion 3.0. CONCLUSIONS: Data produced by GENS2 can be used as a benchmark for evaluating statistical tools designed for theidentification of gene-gene and gene-environment interactions.     

Analyses of longitudinal effects of gene-environment interactions on plasma C-reactive protein levels: the Hallym Aging Study

High C-reactive protein (CRP) level (above 3 mg/L), an inflammatory biomarker, is a well-known risk factor for cardiovascular disease. We investigated the longitudinal effects of interaction between genetic variants of seven candidate loci (i.e., IL6R, CRP, GCKR, IL6, CYP17A1, HNF1A, and APOE) and eight non-genetic factors (i.e., body mass index (BMI), total cholesterol to high density lipoprotein cholesterol ratio (T/HDL), systolic blood pressure, diastolic blood pressure (DBP), current smoking, alcohol consumption, regular exercise, and sleeping hours) on plasma hsCRP levels in a community-based cohort composed of 1,051 elderly Korean participants with a 6-year follow up. We applied a recently developed nonparametric approach, Survival Dimensionality Reduction (SDR) to evaluate gene-environment interactions using follow up data, and compared the results to those of conventional statistical approaches, Cox proportional hazard (CPH) regression model and log-rank test. Four gene variants significantly interacted with three non-genetic factors: SNPs rs2293571 (GCKR), rs1004467 (CYP17A1) and high DBP (HR = 3.22 and 2.95, P _G×E = 0.013 and 0.017, respectively); rs2464196 (HNF1A) and two non-genetic factors, regular exercise (HR = 3.78, P _G×E = 0.043) and high T/HDL (HR = 4.54, P _G×E = 0.042); and rs439401 (APOE) and regular exercise (HR = 3.01, P _G×E = 0.049). The interaction between the rs2464196 and T/HDL was consistently observed in both CPH model and SDR model (Accuracy = 0.86, P = 0.011). Investigating the effects of gene-environment interactions on baseline plasma hsCRP concentrations will provide clues to identify the pathway involved in increasing hsCRP level and the risks of related diseases.     

Detection of gene-environment interactions in joint segregation and linkage analysis.

We compare approaches for analysis of gene-environment (G x E) interaction, using segregation and joint segregation and linkage analyses of a quantitative trait. Analyses of triglyceride levels in a single large pedigree demonstrate the two methods and show evidence for a significant interaction (P=.015 when segregation analysis is used; P=.006 when joint analysis is used) between a codominant major gene and body-mass index. Genotype-specific correlation coefficients, between triglyceride levels and body-mass index, estimated from the joint model are rAA=.72, rAa=.49, and raa=. 20. Several simulation studies indicate that joint segregation and linkage analysis leads to less-biased and more-efficient estimates of a G x E-interaction effect, compared with segregation analysis alone. Depending on the heterozygosity of the marker locus and its proximity to the trait locus, we found joint analysis to be as much as 70% more efficient than segregation analysis, for estimation of a G x E-interaction effect. Over a variety of parameter combinations, joint analysis also led to moderate (5%-10%) increases in power to detect the interaction. On the basis of these results, we suggest the use of combined segregation and linkage analysis for improved estimation of G x E-interaction effects when the underlying trait gene is unmeasured.     

Powerful multilocus tests of genetic association in the presence of gene-gene and gene-environment interactions

In modern genetic epidemiology studies, the association between the disease and a genomic region, such as a candidate gene, is often investigated using multiple SNPs. We propose a multilocus test of genetic association that can account for genetic effects that might be modified by variants in other genes or by environmental factors. We consider use of the venerable and parsimonious Tukey's 1-degree-of-freedom model of interaction, which is natural when individual SNPs within a gene are associated with disease through a common biological mechanism; in contrast, many standard regression models are designed as if each SNP has unique functional significance. On the basis of Tukey's model, we propose a novel but computationally simple generalized test of association that can simultaneously capture both the main effects of the variants within a genomic region and their interactions with the variants in another region or with an environmental exposure. We compared performance of our method with that of two standard tests of association, one ignoring gene-gene/gene-environment interactions and the other based on a saturated model of interactions. We demonstrate major power advantages of our method both in analysis of data from a case-control study of the association between colorectal adenoma and DNA variants in the NAT2 genomic region, which are well known to be related to a common biological phenotype, and under different models of gene-gene interactions with use of simulated data.     

Different data from different labs: lessons from studies of gene-environment interaction

It is sometimes supposed that standardizing tests of mouse behavior will ensure similar results in different laboratories. We evaluated this supposition by conducting behavioral tests with identical apparatus and test protocols in independent laboratories. Eight genetic groups of mice, including equal numbers of males and females, were either bred locally or shipped from the supplier and then tested on six behaviors simultaneously in three laboratories (Albany, NY; Edmonton, AB; Portland, OR). The behaviors included locomotor activity in a small box, the elevated plus maze, accelerating rotarod, visible platform water escape, cocaine activation of locomotor activity, and ethanol preference in a two-bottle test. A preliminary report of this study presented a conventional analysis of conventional measures that revealed strong effects of both genotype and laboratory as well as noteworthy interactions between genotype and laboratory. We now report a more detailed analysis of additional measures and view the data for each test in different ways. Whether mice were shipped from a supplier or bred locally had negligible effects for almost every measure in the six tests, and sex differences were also absent or very small for most behaviors, whereas genetic effects were almost always large. For locomotor activity, cocaine activation, and elevated plus maze, the analysis demonstrated the strong dependence of genetic differences in behavior on the laboratory giving the tests. For ethanol preference and water escape learning, on the other hand, the three labs obtained essentially the same results for key indicators of behavior. Thus, it is clear that the strong dependence of results on the specific laboratory is itself dependent on the task in question. Our results suggest that there may be advantages of test standardization, but laboratory environments probably can never be made sufficiently similar to guarantee identical results on a wide range of tests in a wide range of labs. Interpretations of our results by colleagues in neuroscience as well as the mass media are reviewed. Pessimistic views, prevalent in the media but relatively uncommon among neuroscientists, of mouse behavioral tests as being highly unreliable are contradicted by our data. Despite the presence of noteworthy interactions between genotype and lab environment, most of the larger differences between inbred strains were replicated across the three labs. Strain differences of moderate effects size, on the other hand, often differed markedly among labs, especially those involving three 129-derived strains. Implications for behavioral screening of targeted and induced mutations in mice are discussed.     

Behavior of QQ-plots and genomic control in studies of gene-environment interaction

Genome-wide association studies of gene-environment interaction (GxE GWAS) are becoming popular. As with main effects GWAS, quantile-quantile plots (QQ-plots) and Genomic Control are being used to assess and correct for population substructure. However, in G x E work these approaches can be seriously misleading, as we illustrate; QQ-plots may give strong indications of substructure when absolutely none is present. Using simulation and theory, we show how and why spurious QQ-plot inflation occurs in G x E GWAS, and how this differs from main-effects analyses. We also explain how simple adjustments to standard regression-based methods used in G x E GWAS can alleviate this problem.     

From 'omics' to complex disease: a systems biology approach to gene-environment interactions in cancer

Background  

Cancer is a complex disease that involves a sequence of gene-environment interactions in a progressive process that cannot occur without dysfunction in multiple systems, including DNA repair, apoptotic and immune functions. Epigenetic mechanisms, responding to numerous internal and external cues in a dynamic ongoing exchange, play a key role in mediating environmental influences on gene expression and tumor development.     

Parallel genetic changes and nonparallel gene-environment interactions characterize the evolution of drug resistance in yeast

Beneficial mutations are required for adaptation to novel environments, yet the range of mutational pathways that are available to a population has been poorly characterized, particularly in eukaryotes. We assessed the genetic changes of the first mutations acquired during adaptation to a novel environment (exposure to the fungicide, nystatin) in 35 haploid lines of Saccharomyces cerevisiae. Through whole-genome resequencing we found that the genomic scope for adaptation was narrow; all adapted lines acquired a mutation in one of four late-acting genes in the ergosterol biosynthesis pathway, with very few other mutations found. Lines that acquired different ergosterol mutations in the same gene exhibited very similar tolerance to nystatin. All lines were found to have a cost relative to wild type in an unstressful environment; the level of this cost was also strongly correlated with the ergosterol gene bearing the mutation. Interestingly, we uncovered both positive and negative effects on tolerance to other harsh environments for mutations in the different ergosterol genes, indicating that these beneficial mutations have effects that differ in sign among environmental challenges. These results demonstrate that although the genomic target was narrow, different adaptive mutations can lead populations down different evolutionary pathways, with respect to their ability to tolerate (or succumb to) other environmental challenges.     

A latent variable approach to study gene-environment interactions in the presence of multiple correlated exposures

Many existing cohort studies initially designed to investigate disease risk as a function of environmental exposures have collected genomic data in recent years with the objective of testing for gene-environment interaction (G × E) effects. In environmental epidemiology, interest in G × E arises primarily after a significant effect of the environmental exposure has been documented. Cohort studies often collect rich exposure data; as a result, assessing G × E effects in the presence of multiple exposure markers further increases the burden of multiple testing, an issue already present in both genetic and environment health studies. Latent variable (LV) models have been used in environmental epidemiology to reduce dimensionality of the exposure data, gain power by reducing multiplicity issues via condensing exposure data, and avoid collinearity problems due to presence of multiple correlated exposures. We extend the LV framework to characterize gene-environment interaction in presence of multiple correlated exposures and genotype categories. Further, similar to what has been done in case-control G × E studies, we use the assumption of gene-environment (G-E) independence to boost the power of tests for interaction. The consequences of making this assumption, or the issue of how to explicitly model G-E association has not been previously investigated in LV models. We postulate a hierarchy of assumptions about the LV model regarding the different forms of G-E dependence and show that making such assumptions may influence inferential results on the G, E, and G × E parameters. We implement a class of shrinkage estimators to data adaptively trade-off between the most restrictive to most flexible form of G-E dependence assumption and note that such class of compromise estimators can serve as a benchmark of model adequacy in LV models. We demonstrate the methods with an example from the Early Life Exposures in Mexico City to Neuro-Toxicants Study of lead exposure, iron metabolism genes, and birth weight.