Resampling to Address the Winner's Curse in Genetic Association Analysis of Time to Event

Gene-environment interactions in the associations of PFAS exposure with insulin sensitivity and beta-cell function in a Faroese cohort followed from birth to adulthood

BACKGROUND

Exposure to perfluoroalkyl substances (PFAS) has been associated with changes in insulin sensitivity and pancreatic beta-cell function in humans. Genetic predisposition to diabetes may modify these associations; however, this hypothesis has not been yet studied.

OBJECTIVES

To evaluate genetic heterogeneity as a modifier in the PFAS association with insulin sensitivity and pancreatic beta-cell function, using a targeted gene-environment (GxE) approach.

METHODS

We studied 85 single-nucleotide polymorphisms (SNPs) associated with type 2 diabetes, in 665 Faroese adults born in 1986-1987. Perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) were measured in cord whole blood at birth and in participants' serum from age 28 years. We calculated the Matsuda-insulin sensitivity index (ISI) and the insulinogenic index (IGI) based on a 2 h-oral glucose tolerance test performed at age 28. Effect modification was evaluated in linear regression models adjusted for cross-product terms (PFAS*SNP) and important covariates.

RESULTS

Prenatal and adult PFOS exposures were significantly associated with decreased insulin sensitivity and increased beta-cell function. PFOA associations were in the same direction but attenuated compared to PFOS. A total of 58 SNPs were associated with at least one PFAS exposure variable and/or Matsuda-ISI or IGI in the Faroese population and were subsequently tested as modifiers in the PFAS-clinical outcome associations. Eighteen SNPs showed interaction p-values (PGxE) < 0.05 in at least one PFAS-clinical outcome association, five of which passed False Discovery Rate (FDR) correction (PGxE-FDR<0.20). SNPs for which we found stronger evidence for GxE interactions included ABCA1 rs3890182, FTO rs9939609, FTO rs3751812, PPARG rs170036314 and SLC12A3 rs2289116 and were more clearly shown to modify the PFAS associations with insulin sensitivity, rather than with beta-cell function.

DISCUSSION

Findings from this study suggest that PFAS-associated changes in insulin sensitivity could vary between individuals as a result of genetic predisposition and warrant replication in independent larger populations.

Estimation of genetic variance contributed by a quantitative trait locus: correcting the bias associated with significance tests

The Beavis effect in quantitative trait locus (QTL) mapping describes a phenomenon that the estimated effect size of a statistically significant QTL (measured by the QTL variance) is greater than the true effect size of the QTL if the sample size is not sufficiently large. This is a typical example of the Winners' curse applied to molecular quantitative genetics. Theoretical evaluation and correction for the Winners' curse have been studied for interval mapping. However, similar technologies have not been available for current models of QTL mapping and genome-wide association studies where a polygene is often included in the linear mixed models to control the genetic background effect. In this study, we developed the theory of the Beavis effect in a linear mixed model using a truncated noncentral Chi-square distribution. We equated the observed Wald test statistic of a significant QTL to the expectation of a truncated noncentral Chi-square distribution to obtain a bias-corrected estimate of the QTL variance. The results are validated from replicated Monte Carlo simulation experiments. We applied the new method to the grain width (GW) trait of a rice population consisting of 524 homozygous varieties with over 300 k single nucleotide polymorphism markers. Two loci were identified and the estimated QTL heritability were corrected for the Beavis effect. Bias correction for the larger QTL on chromosome 5 (GW5) with an estimated heritability of 12% did not change the QTL heritability due to the extremely large test score and estimated QTL effect. The smaller QTL on chromosome 9 (GW9) had an estimated QTL heritability of 9% reduced to 6% after the bias-correction.

The genetics of drug-related movement disorders, an umbrella review of meta-analyses

This umbrella review investigates which genetic factors are associated with drug-related movement disorders (DRMD), in an attempt to provide a synthesis of published evidence of candidate-gene studies. To identify all relevant meta-analyses, a literature search was performed. Titles and abstracts were screened by two authors and the methodological quality of included meta-analyses was assessed using 'the assessment of multiple systematic reviews' (AMSTAR) critical appraisal checklist. The search yielded 15 meta-analytic studies reporting on genetic variations in 10 genes. DRD3, DRD2, CYP2D6, HTR2A, COMT, HSPG2 and SOD2 genes have variants that may increase the odds of TD. However, these findings do not concur with early genome-wide association studies. Low-power samples are susceptible to 'winner's curse', which was supported by diminishing meta-analytic effects of several genetic variants over time. Furthermore, analyses pertaining to the same genetic variant were difficult to compare due to differences in patient populations, methods used and the choice of studies included in meta-analyses. In conclusion, DRMD is a complex phenotype with multiple genes that impact the probability of onset. More studies with larger samples using other methods than by candidate genes, are essential to developing methods that may predict the probability of DRMD. To achieve this, multiple research groups need to collaborate and a DRMD genetic database needs to be established in order to overcome winner's curse and publication bias, and to allow for stratification by patient characteristics. These endeavours may help the development of a test with clinical value in the prevention and treatment of DRMD.

A genome-wide association study implicates NR2F2 in lymphangioleiomyomatosis pathogenesis

INTRODUCTION

Lymphangioleiomyomatosis (LAM) occurs either associated with tuberous sclerosis complex (TSC) or as sporadic disease (S-LAM). Risk factors for development of S-LAM are unknown. We hypothesised that DNA sequence variants outside of TSC2/TSC1 might be associated with susceptibility for S-LAM and performed a genome-wide association study (GWAS).

METHODS

Genotyped and imputed data on 5 426 936 single nucleotide polymorphisms (SNPs) in 426 S-LAM subjects were compared, using conditional logistic regression, with similar data from 852 females from COPDGene in a matched case-control design. For replication studies, genotypes for 196 non-Hispanic White female S-LAM subjects were compared with three different sets of controls. RNA sequencing and immunohistochemistry analyses were also performed.

RESULTS

Two noncoding genotyped SNPs met genome-wide significance: rs4544201 and rs2006950 (p=4.2×10^-8 and 6.1×10^-9, respectively), which are in the same 35 kb linkage disequilibrium block on chromosome 15q26.2. This association was replicated in an independent cohort. NR2F2 (nuclear receptor subfamily 2 group F member 2), a nuclear receptor and transcription factor, was the only nearby protein-coding gene. NR2F2 expression was higher by RNA sequencing in one abdominal LAM tumour and four kidney angiomyolipomas, a LAM-related tumour, compared with all cancers from The Cancer Genome Atlas. Immunohistochemistry showed strong nuclear expression in both LAM and angiomyolipoma tumours.

CONCLUSIONS

SNPs on chromosome 15q26.2 are associated with S-LAM, and chromatin and expression data suggest that this association may occur through effects on NR2F2 expression, which potentially plays an important role in S-LAM development.

Enrichment of B cell receptor signaling and epidermal growth factor receptor pathways in monoclonal gammopathy of undetermined significance: a genome-wide genetic interaction study

BACKGROUND

Recent identification of 10 germline variants predisposing to monoclonal gammopathy of undetermined significance (MGUS) explicates genetic dependency of this asymptomatic precursor condition with multiple myeloma (MM). Yet much of genetic burden as well as functional links remain unexplained. We propose a workflow to expand the search for susceptibility loci with genome-wide interaction and for subsequent identification of genetic clusters and pathways.

METHODS

Polygenic interaction analysis on 243 cases/1285 controls identified 14 paired risk loci belonging to unique chromosomal bands which were then replicated in two independent sets (case only study, 82 individuals; case/control study 236 cases/ 2484 controls). Further investigation on gene-set enrichment, regulatory pathway and genetic network was carried out with stand-alone in silico tools separately for both interaction and genome-wide association study-detected risk loci.

RESULTS

Intronic-PREX1 (20q13.13), a reported locus predisposing to MM was confirmed to have contribution to excess MGUS risk in interaction with SETBP1, a well-established candidate predisposing to myeloid malignancies. Pathway enrichment showed B cell receptor signaling pathway (P < 5.3 × 10- 3) downstream to allograft rejection pathway (P < 5.6 × 10- 4) and autoimmune thyroid disease pathway (P < 9.3 × 10- 4) as well as epidermal growth factor receptor regulation pathway (P < 2.4 × 10- 2) to be differentially regulated. Oncogene ALK and CDH2 were also identified to be moderately interacting with rs10251201 and rs16966921, two previously reported risk loci for MGUS.

CONCLUSIONS

We described novel pathways and variants potentially causal for MGUS. The methodology thus proposed to facilitate our search streamlines risk locus-based interaction, genetic network and pathway enrichment analyses.

Polygenic influences on dyslipidemias

PURPOSE OF REVIEW

Rare large-effect genetic variants underlie monogenic dyslipidemias, whereas common small-effect genetic variants - single nucleotide polymorphisms (SNPs) - have modest influences on lipid traits. Over the past decade, these small-effect SNPs have been shown to cumulatively exert consistent effects on lipid phenotypes under a polygenic framework, which is the focus of this review.

RECENT FINDINGS

Several groups have reported polygenic risk scores assembled from lipid-associated SNPs, and have applied them to their respective phenotypes. For lipid traits in the normal population distribution, polygenic effects quantified by a score that integrates several common polymorphisms account for about 20-30% of genetic variation. Among individuals at the extremes of the distribution, that is, those with clinical dyslipidemia, the polygenic component includes both rare variants with large effects and common polymorphisms: depending on the trait, 20-50% of susceptibility can be accounted for by this assortment of genetic variants.

SUMMARY

Accounting for polygenic effects increases the numbers of dyslipidemic individuals who can be explained genetically, but a substantial proportion of susceptibility remains unexplained. Whether documenting the polygenic basis of dyslipidemia will affect outcomes in clinical trials or prospective observational studies remains to be determined.

Effects of the Ser326Cys Polymorphism in the DNA Repair OGG1 Gene on Cancer, Cardiovascular, and All-Cause Mortality in the PREDIMED Study: Modulation by Diet

BACKGROUND

Oxidatively induced DNA damage, an important factor in cancer etiology, is repaired by oxyguanine glycosylase 1 (OGG1). The lower repair capacity genotype (homozygote Cys326Cys) in the OGG1-rs1052133 (Ser326Cys) polymorphism has been associated with cancer risk. However, no information is available in relation to cancer mortality, other causes of death, and modulation by diet.

OBJECTIVE

Our aim was to evaluate the association of the OGG1-rs1052133 with total, cancer, and cardiovascular disease (CVD) mortality and to analyze its modulation by the Mediterranean diet, focusing especially on total vegetable intake as one of the main characteristics of this diet.

DESIGN

Secondary analysis in the PREDIMED (Prevención con Dieta Mediterránea) trial is a randomized, controlled trial conducted in Spain from 2003 to 2010.

PARTICIPANTS/SETTING

Study participants (n=7,170) were at high risk for CVD and were aged 55 to 80 years.

INTERVENTION

Participants were randomly allocated to two groups with a Mediterranean diet intervention or a control diet. Vegetable intake was measured at baseline.

MAIN OUTCOME MEASURES

Main outcomes were all-cause, cancer, and CVD mortality after a median follow-up of 4.8 years.

STATISTICAL ANALYSES

Multivariable-adjusted Cox regression models were fitted.

RESULTS

Three hundred eighteen deaths were detected (cancer, n=127; CVD, n=81; and other, n=110). Cys326Cys individuals (prevalence 4.2%) presented higher total mortality rates than Ser326-carriers (P=0.009). The multivariable-adjusted hazard ratio for Cys326Cys vs Ser326-carriers was 1.69 (95% CI 1.09 to 2.62; P=0.018). This association was greater for CVD mortality (P=0.001). No relationship was detected for cancer mortality in the whole population (hazard ratio 1.07; 95% CI 0.47 to 2.45; P=0.867), but a significant age interaction (P=0.048) was observed, as Cys326Cys was associated with cancer mortality in participants <66.5 years (P=0.029). Recessive effects limited our ability to investigate Cys326Cys×diet interactions for cancer mortality. No statistically significant interactions for total or CVD mortality were found for the Mediterranean diet intervention. However, significant protective interactions for CVD mortality were found for vegetable intake (hazard ratio interaction per standard deviation 0.42; 95% CI 0.18 to 0.98; P=0.046).

CONCLUSIONS

In this population, the Cys326Cys-OGG1 genotype was associated with all-cause mortality, mainly CVD instead of cancer mortality. Additional studies are needed to provide further evidence on its dietary modulation.

Illustrating, Quantifying, and Correcting for Bias in Post-hoc Analysis of Gene-Based Rare Variant Tests of Association

To date, gene-based rare variant testing approaches have focused on aggregating information across sets of variants to maximize statistical power in identifying genes showing significant association with diseases. Beyond identifying genes that are associated with diseases, the identification of causal variant(s) in those genes and estimation of their effect is crucial for planning replication studies and characterizing the genetic architecture of the locus. However, we illustrate that straightforward single-marker association statistics can suffer from substantial bias introduced by conditioning on gene-based test significance, due to the phenomenon often referred to as "winner's curse." We illustrate the ramifications of this bias on variant effect size estimation and variant prioritization/ranking approaches, outline parameters of genetic architecture that affect this bias, and propose a bootstrap resampling method to correct for this bias. We find that our correction method significantly reduces the bias due to winner's curse (average two-fold decrease in bias, p < 2.2 × 10^-6) and, consequently, substantially improves mean squared error and variant prioritization/ranking. The method is particularly helpful in adjustment for winner's curse effects when the initial gene-based test has low power and for relatively more common, non-causal variants. Adjustment for winner's curse is recommended for all post-hoc estimation and ranking of variants after a gene-based test. Further work is necessary to continue seeking ways to reduce bias and improve inference in post-hoc analysis of gene-based tests under a wide variety of genetic architectures.