Mendelian randomization: Using genes as instruments for making causal inferences in epidemiology

Causal effects of body mass index on cardiometabolic traits and events: a Mendelian randomization analysis

Elevated body mass index (BMI) associates with cardiometabolic traits on observational analysis, yet the underlying causal relationships remain unclear. We conducted Mendelian randomization analyses by using a genetic score (GS) comprising 14 BMI-associated SNPs from a recent discovery analysis to investigate the causal role of BMI in cardiometabolic traits and events. We used eight population-based cohorts, including 34,538 European-descent individuals (4,407 type 2 diabetes (T2D), 6,073 coronary heart disease (CHD), and 3,813 stroke cases). A 1 kg/m(2) genetically elevated BMI increased fasting glucose (0.18 mmol/l; 95% confidence interval (CI) = 0.12-0.24), fasting insulin (8.5%; 95% CI = 5.9-11.1), interleukin-6 (7.0%; 95% CI = 4.0-10.1), and systolic blood pressure (0.70 mmHg; 95% CI = 0.24-1.16) and reduced high-density lipoprotein cholesterol (-0.02 mmol/l; 95% CI = -0.03 to -0.01) and low-density lipoprotein cholesterol (LDL-C; -0.04 mmol/l; 95% CI = -0.07 to -0.01). Observational and causal estimates were directionally concordant, except for LDL-C. A 1 kg/m(2) genetically elevated BMI increased the odds of T2D (odds ratio [OR] = 1.27; 95% CI = 1.18-1.36) but did not alter risk of CHD (OR 1.01; 95% CI = 0.94-1.08) or stroke (OR = 1.03; 95% CI = 0.95-1.12). A meta-analysis incorporating published studies reporting 27,465 CHD events in 219,423 individuals yielded a pooled OR of 1.04 (95% CI = 0.97-1.12) per 1 kg/m(2) increase in BMI. In conclusion, we identified causal effects of BMI on several cardiometabolic traits; however, whether BMI causally impacts CHD risk requires further evidence.

Genome-wide association study of metabolic traits reveals novel gene-metabolite-disease links

Metabolic traits are molecular phenotypes that can drive clinical phenotypes and may predict disease progression. Here, we report results from a metabolome- and genome-wide association study on (1)H-NMR urine metabolic profiles. The study was conducted within an untargeted approach, employing a novel method for compound identification. From our discovery cohort of 835 Caucasian individuals who participated in the CoLaus study, we identified 139 suggestively significant (P<5×10(-8)) and independent associations between single nucleotide polymorphisms (SNP) and metabolome features. Fifty-six of these associations replicated in the TasteSensomics cohort, comprising 601 individuals from São Paulo of vastly diverse ethnic background. They correspond to eleven gene-metabolite associations, six of which had been previously identified in the urine metabolome and three in the serum metabolome. Our key novel findings are the associations of two SNPs with NMR spectral signatures pointing to fucose (rs492602, P = 6.9×10(-44)) and lysine (rs8101881, P = 1.2×10(-33)), respectively. Fine-mapping of the first locus pinpointed the FUT2 gene, which encodes a fucosyltransferase enzyme and has previously been associated with Crohn's disease. This implicates fucose as a potential prognostic disease marker, for which there is already published evidence from a mouse model. The second SNP lies within the SLC7A9 gene, rare mutations of which have been linked to severe kidney damage. The replication of previous associations and our new discoveries demonstrate the potential of untargeted metabolomics GWAS to robustly identify molecular disease markers.

Should African Americans with hypertension be treated differently than non-African Americans?

African Americans have a higher burden of hypertension, more severe blood pressure (BP) elevations, more concurrent risk-enhancing co-morbidities (e.g., diabetes), sub-clinical vascular injury at lower non-hypertensive BP levels, lower BP control rates, and significantly greater risk for adverse pressure-related clinical complications (e.g., stroke, heart failure) than whites. Randomized prospective data from hypertension endpoint trials show a virtually identical percentage reduction in CVD risk for a given magnitude of BP lowering, irrespective of the presence or absence of pre-treatment CVD across a broad range of BP down to pre-treatment BP levels of 110/70 mm Hg. These data, mostly emanating from white populations, do not necessarily inform practitioners as to the level below which BP should be lowered in those with established, long-standing hypertension; however, these data do provide support for initiating hypertension treatment at lower than conventional BP thresholds. A Mendelian randomized study examining the impact of life-long lower SBP levels showed that lifelong exposure to 10 mm Hg lower SBP was associated with an 82 % lesser rate of SBP rise per decade and a 58 % lower CHD risk that was much greater than the 22 % reduction in CHD reported for the same magnitude of SBP reduction in clinical trials. Arguably, it is the hypertension treatment paradigm that merits reexamination. Earlier hypertension treatment in all populations prior to the onset of significant pressure-related target organ injury might conceivably prevent, or at least significantly attenuate, the well documented age-related rise in BP seen in most Western societies. In addition, this treatment paradigm might also reduce the significant residual CVD risk observed under the current recommended approach to hypertension treatment. This new approach to therapy would likely have substantial clinical and public health benefits in the high-risk, under-treated African American population that suffers outsized devastating consequences from inadequate control of BP.

A note on the control function approach with an instrumental variable and a binary outcome

Unobserved confounding is a well known threat to causal inference in non-experimental studies. The instrumental variable design can under certain conditions be used to recover an unbiased estimator of a treatment effect even if unobserved confounding cannot be ruled out with certainty. For continuous outcomes, two stage least squares is the most common instrumental variable estimator used in epidemiologic applications. For a rare binary outcome, an analogous linear-logistic two-stage procedure can be used. Alternatively, a control function approach is sometimes used which entails entering the residual from the first stage linear model as a covariate in a second stage logistic regression of the outcome on the treatment. Both strategies for binary response have previously formally been justified only for continuous exposure, which has impeded widespread use of the approach outside of this setting. In this note, we consider the important setting of binary exposure in the context of a binary outcome. We provide an alternative motivation for the control function approach which is appropriate for binary exposure, thus establishing simple conditions under which the approach may be used for instrumental variable estimation when the outcome is rare. In the proposed approach, the first stage regression involves a logistic model of the exposure conditional on the instrumental variable, and the second stage regression is a logistic regression of the outcome on the exposure adjusting for the first stage residual. In the event of a non-rare outcome, we recommend replacing the second stage logistic model with a risk ratio regression.

Long-term health outcomes in offspring born to women with diabetes in pregnancy

In this review, we critically assess recent evidence from human studies regarding the potential implications of exposure to maternal diabetes in-utero for long-term adiposity, cardiometabolic outcomes, and cognitive ability of the offspring. Evidence supports a direct causal role for exposure to maternal diabetes in utero in determining offspring long-term greater adiposity and adverse cardiometabolic health. Although a majority of observational studies report associations of exposure to maternal pregnancy diabetes with lower cognitive ability, there is also evidence supporting an opposite 'protective' intrauterine effect of exposure to maternal pregnancy diabetes on offspring cognitive ability. Epigenetic modification has been suggested as a mediator on the pathways from maternal pregnancy diabetes to long-term offspring outcomes and several recent studies that are reviewed here lend some support to this notion, but research in this area is still too novel to be conclusive.

The EPIC-InterAct Study: A Study of the Interplay between Genetic and Lifestyle Behavioral Factors on the Risk of Type 2 Diabetes in European Populations

The rising prevalence of type 2 diabetes around the world and the global pattern of variation in risk between countries have been widely attributed to an interplay between rising rates of obesity and poor lifestyles, and genetic or developmental susceptibility to disease. Although this general hypothesis has been in existence for more than 50 years, the precise mechanisms that may explain it have remained uncertain. Advances in technology and the application of new methods in large scale population studies have made it possible to study these mechanisms. The InterAct project, funded by the European Commission, is a large case-cohort study which has verified 12,403 incident cases of type 2 diabetes, facilitating the study of genetic and lifestyle factors on the risk of type 2 diabetes among European populations.

The association between circulating lipoprotein(a) and type 2 diabetes: is it causal?

Epidemiological evidence supports a direct and causal association between lipoprotein(a) [Lp(a)] levels and coronary risk, but the nature of the association between Lp(a) levels and risk of type 2 diabetes (T2D) is unclear. In this study, we assessed the association of Lp(a) levels with risk of incident T2D and tested whether Lp(a) levels are causally linked to T2D. We analyzed data on 18,490 participants from the European Prospective Investigation of Cancer (EPIC)-Norfolk cohort that included adults aged 40-79 years at baseline 1993-1997. During an average 10 years of follow-up, 593 participants developed incident T2D. Cox regression models were used to estimate the association between Lp(a) levels and T2D. In Mendelian randomization analyses, based on EPIC-Norfolk combined with DIAbetes Genetics Replication And Meta-analysis data involving a total of 10,088 diabetes case participants and 68,346 control participants, we used a genetic variant (rs10455872) as an instrument to test whether the association between Lp(a) levels and T2D is causal. In adjusted analyses, there was an inverse association between Lp(a) levels and T2D: hazard ratio was 0.63 (95% CI 0.49-0.81; P trend = 0.003) comparing the top versus bottom quintile of Lp(a). In EPIC-Norfolk, a 1-SD increase in logLp(a) was associated with a lower risk of T2D (odds ratio [OR] 0.88 [95% CI: 0.80-0.95]). However, in Mendelian randomization analyses, a 1-SD increase in logLp(a) due to rs10455872, which explained 26.8% of the variability in Lp(a) levels, was not associated with risk of T2D (OR 1.03 [0.96-1.10]; P = 0.41). These prospective findings demonstrate a strong inverse association of Lp(a) levels with risk of T2D. However, a genetic variant that elevated Lp(a) levels was not associated with risk of T2D, suggesting that elevated Lp(a) levels are not causally associated with a lower risk of T2D.

Vitamin D and high blood pressure: causal association or epiphenomenon?

High plasma levels of vitamin D are associated with a reduced risk of high blood pressure, but whether this association is causal remains to be ascertained. We performed a meta-analysis of randomized clinical trials, to examine the effect of vitamin D supplementation on both systolic blood pressure (SBP) and diastolic blood pressure (DBP) and supplemented these results with a Mendelian randomization analysis to investigate the causal relationship between vitamin D status (25-hydroxyvitamin D [25(OH)D]) and BP. Pooled random effects meta-analysis of weighted mean differences across 16 trials of vitamin D supplementation showed a non-significant reduction in SBP (-0.94, 95% CI -2.98, 1.10 mmHg) and DBP (-0.52, 95% CI -1.18, 0.14 mmHg), with evidence of heterogeneity (I(2) = 67.9%, P < 0.001) and publication bias (P = 0.02) among trials of SBP. There was a significant reduction in DBP (-1.31, 95% CI -2.28, -0.34 mmHg, P = 0.01) in participants with pre-existing cardiometabolic disease. Variants at three published loci (GC, DHCR7, CYP2R1, and CYP24A1) for 25(OH)D, were not significantly associated with BP, but rs6013897 in CYP24A1 gene region had nominally significant associations with both SBP and DBP (P < 0.05). Evidence from the associations of the genetic variants with the risk of vitamin D deficiency (defined as a 25(OH)D level < 50 nmol/L) and BP showed that the causal effects of a doubling of genetically-elevated risk of vitamin D deficiency were 0.14 mmHg (95% CI -0.19, 0.47, P = 0.42), and 0.12 mmHg (95% CI -0.09, 0.33, P = 0.25) on SBP and DBP respectively. Additional evidence from genetic data are directionally consistent with clinical trial data, though underpowered to reliably demonstrate a strong causal effect of vitamin D status on BP. Further investigation may be warranted.

Body composition and arsenic metabolism: a cross-sectional analysis in the Strong Heart Study

OBJECTIVE

The objective of this study was to evaluate the association between measures of body composition and patterns of urine arsenic metabolites in the 1989-1991 baseline visit of the Strong Heart Study, a cardiovascular disease cohort of adults recruited from rural communities in Arizona, Oklahoma, North Dakota and South Dakota.

METHODS

We evaluated 3,663 Strong Heart Study participants with urine arsenic species above the limit of detection and no missing data on body mass index, % body fat and fat free mass measured by bioelectrical impedance, waist circumference and other variables. We summarized urine arsenic species patterns as the relative contribution of inorganic (iAs), methylarsonate (MMA) and dimethylarsinate (DMA) species to their sum. We modeled the associations of % arsenic species biomarkers with body mass index, % body fat, fat free mass, and waist circumference categories in unadjusted regression models and in models including all measures of body composition. We also considered adjustment for arsenic exposure and demographics.

RESULTS

Increasing body mass index was associated with higher mean % DMA and lower mean % MMA before and after adjustment for sociodemographic variables, arsenic exposure, and for other measures of body composition. In unadjusted linear regression models, % DMA was 2.4 (2.1, 2.6) % higher per increase in body mass index category (< 25, ≥25 & <30, ≥30 & <35, ≥35 kg/m2), and % MMA was 1.6 (1.4, 1.7) % lower. Similar patterns were observed for % body fat, fat free mass, and waist circumference measures in unadjusted models and in models adjusted for potential confounders, but the associations were largely attenuated or disappeared when adjusted for body mass index.

CONCLUSION

Measures of body size, especially body mass index, are associated with arsenic metabolism biomarkers. The association may be related to adiposity, fat free mass or body size. Future epidemiologic studies of arsenic should consider body mass index as a potential modifier for arsenic-related health effects.

Elevated body mass index as a causal risk factor for symptomatic gallstone disease: a Mendelian randomization study

Elevated body mass index (BMI) is associated with an increased risk of gallstone disease. Whether this reflects a causal association is unknown. Using a Mendelian randomization approach, we studied 77,679 individuals from the general population. Of these, 4,106 developed symptomatic gallstone disease during up to 34 years of follow-up. Subjects were genotyped for three common variants known to associate with BMI: FTO(rs9939609); MC4R(rs17782313); and TMEM18(rs6548238). The number of BMI-increasing alleles was calculated for each participant. In observational analyses, mean baseline BMI was 55% (11.6 kg/m(2) ) increased in individuals in the fifth quintile versus the first quintile, similar in women and men. The corresponding multifactorially adjusted hazard ratio (HR) for symptomatic gallstone disease was 2.84 (95% confidence interval [CI]: 2.32-3.46) overall, 3.36 (95% CI: 2.62-4.31) in women, and 1.51 (95% CI: 1.09-2.11) in men (P trend: 0.001 to <0.001; P interaction: BMI*sex on risk = 0.01). In genetic analyses, carrying 6 versus 0-1 BMI-increasing alleles was associated with a 5.2% (1.3 kg/m(2) ) increase in BMI overall and with increases of 4.3% in women and 6.1% in men (all P trend: <0.001). Corresponding HRs for symptomatic gallstone disease were 1.43 (95% CI: 0.99-2.05) overall, 1.54 (95% CI: 1.00-2.35) in women, and 1.19 (95% CI: 0.60-2.38) in men (P trend = 0.007, 0.02, and 0.26, respectively; P interaction allele score*sex on risk = 0.49). The estimated causal odds ratio (OR) for symptomatic gallstone disease, by instrumental variable analysis for a 1 kg/m(2) increase in genetically determined BMI, was 1.17 (95% CI: 0.99-1.37) overall and 1.20 (95% CI: 1.00-1.44) and 1.02 (95% CI: 0.90-1.16) in women and men, respectively. Corresponding observational HRs were 1.07 (95% CI: 1.06-1.08), 1.08 (95% CI: 1.07-1.10), and 1.04 (95% CI: 1.02-1.07), respectively.

CONCLUSION

These results are compatible with a causal association between elevated BMI and increased risk of symptomatic gallstone disease, which is most pronounced in women.

Plasma C-reactive protein, genetic risk score, and risk of common cancers in the Atherosclerosis Risk in Communities study

PURPOSE

Many studies, including the Atherosclerosis Risk in Communities (ARIC) cohort, reported a positive association between plasma C-reactive protein (CRP)-a biomarker of low-grade chronic inflammation-and colorectal cancer risk, although it is unclear whether the association is causal. Our aims were to assess the associations of a CRP genetic risk score (CRP-GRS) created from single-nucleotide polymorphisms (SNPs) with colorectal cancer risk, as well as examine plasma CRP and CRP-GRS in relation to common cancers in the ARIC cohort.

METHODS

Cox proportional hazards models were used to prospectively estimate hazard ratios (HRs) and 95 % confidence interval (95 % CI) of total, colorectal, lung, prostate, and breast cancers in relation to: (1) CRP-GRS among 8,657 Whites followed in 1987-2006 and (2) log-transformed plasma CRP among 7,603 Whites followed in 1996-2006. A weighted CRP-GRS was comprised of 20 CRP-related SNPs located in/near CRP, APOC1, HNF1A, LEPR, and 16 other genes that were identified in genome-wide association studies.

RESULTS

After multivariable adjustment, one standard deviation increment of the CRP-GRS was associated with colorectal cancer risk (HR 1.19; 95 % CI 1.03-1.37), but not with any other cancer. One unit of log-transformed plasma CRP was associated with the risk of total, colorectal, lung, and breast cancers: HRs (95 % CIs) were 1.08 (1.01-1.15), 1.24 (1.01-1.51), 1.29 (1.08-1.54), and 1.27 (1.07-1.51), respectively. HRs remained elevated, although lost statistical significance for all but breast cancer, after excluding subjects with <2 years of follow-up.

CONCLUSIONS

The study corroborates a causative role of chronic low-grade inflammation in colorectal carcinogenesis.

Androgen activity, ischaemic heart disease and risk factors among men in NHANES III

AIM

Observationally, low serum testosterone among men is associated with cardiovascular diseases and its risk factors, but it is unclear whether raising endogenous androgens would be protective. To clarify the role of androgens, the association of two different androgen biomarkers (serum testosterone and androstanediol glucuronide) with cardiovascular disease risk factors and mortality was examined in a nationally representative sample of US men.

RESEARCH DESIGN AND METHODS

Multivariable linear and proportion hazards regression were used to examine the adjusted associations of serum testosterone and androstanediol glucuronide with cardiovascular disease risk factors and death from major cardiovascular diseases in 1460 men from NHANES III phase 1 (1988-1991) followed-up through 2006.

RESULTS

Serum testosterone and androstanediol glucuronide were weakly correlated (0·13). Serum testosterone was associated with healthier values of most cardiovascular disease risk factors but not with death from ischaemic heart disease or stroke, adjusted for age, education, race/ethnicity, smoking and alcohol use. Similarly adjusted, androstanediol glucuronide was associated with unhealthier values of some cardiovascular risk factors and death from ischaemic heart disease (hazard ratio 1·16, 95% confidence interval 1·003-1·33 per standard deviation).

CONCLUSIONS

Androgen biomarkers had inconsistent associations with cardiovascular disease risk factors and ischaemic heart disease. Androstanediol glucuronide, rather than serum testosterone, had associations with cardiovascular disease risk factors more similar to those seen in randomized controlled trials of testosterone therapy, with corresponding implications for raising androgens.

Identifying the odds ratio estimated by a two-stage instrumental variable analysis with a logistic regression model

An adjustment for an uncorrelated covariate in a logistic regression changes the true value of an odds ratio for a unit increase in a risk factor. Even when there is no variation due to covariates, the odds ratio for a unit increase in a risk factor also depends on the distribution of the risk factor. We can use an instrumental variable to consistently estimate a causal effect in the presence of arbitrary confounding. With a logistic outcome model, we show that the simple ratio or two-stage instrumental variable estimate is consistent for the odds ratio of an increase in the population distribution of the risk factor equal to the change due to a unit increase in the instrument divided by the average change in the risk factor due to the increase in the instrument. This odds ratio is conditional within the strata of the instrumental variable, but marginal across all other covariates, and is averaged across the population distribution of the risk factor. Where the proportion of variance in the risk factor explained by the instrument is small, this is similar to the odds ratio from a RCT without adjustment for any covariates, where the intervention corresponds to the effect of a change in the population distribution of the risk factor. This implies that the ratio or two-stage instrumental variable method is not biased, as has been suggested, but estimates a different quantity to the conditional odds ratio from an adjusted multiple regression, a quantity that has arguably more relevance to an epidemiologist or a policy maker, especially in the context of Mendelian randomization.

Secretory phospholipase A(2)-IIA and cardiovascular disease: a mendelian randomization study

OBJECTIVES

This study sought to investigate the role of secretory phospholipase A2 (sPLA2)-IIA in cardiovascular disease.

BACKGROUND

Higher circulating levels of sPLA2-IIA mass or sPLA2 enzyme activity have been associated with increased risk of cardiovascular events. However, it is not clear if this association is causal. A recent phase III clinical trial of an sPLA2 inhibitor (varespladib) was stopped prematurely for lack of efficacy.

METHODS

We conducted a Mendelian randomization meta-analysis of 19 general population studies (8,021 incident, 7,513 prevalent major vascular events [MVE] in 74,683 individuals) and 10 acute coronary syndrome (ACS) cohorts (2,520 recurrent MVE in 18,355 individuals) using rs11573156, a variant in PLA2G2A encoding the sPLA2-IIA isoenzyme, as an instrumental variable.

RESULTS

PLA2G2A rs11573156 C allele associated with lower circulating sPLA2-IIA mass (38% to 44%) and sPLA2 enzyme activity (3% to 23%) per C allele. The odds ratio (OR) for MVE per rs11573156 C allele was 1.02 (95% confidence interval [CI]: 0.98 to 1.06) in general populations and 0.96 (95% CI: 0.90 to 1.03) in ACS cohorts. In the general population studies, the OR derived from the genetic instrumental variable analysis for MVE for a 1-log unit lower sPLA2-IIA mass was 1.04 (95% CI: 0.96 to 1.13), and differed from the non-genetic observational estimate (OR: 0.69; 95% CI: 0.61 to 0.79). In the ACS cohorts, both the genetic instrumental variable and observational ORs showed a null association with MVE. Instrumental variable analysis failed to show associations between sPLA2 enzyme activity and MVE.

CONCLUSIONS

Reducing sPLA2-IIA mass is unlikely to be a useful therapeutic goal for preventing cardiovascular events.

Lipoprotein(a) concentrations, isoform size, and risk of type 2 diabetes: a Mendelian randomisation study

BACKGROUND

Low concentrations of lipoprotein(a) in plasma are associated with increased risk of type 2 diabetes, but whether this association is causal is unclear. Variations in the LPA gene affect lipoprotein(a) isoform size and concentrations in plasma. We therefore did a Mendelian randomisation study to investigate whether large isoform size, low concentrations in plasma, or both, are causally associated with type 2 diabetes.

METHODS

We assessed data for adults from the Danish general population enrolled in the Copenhagen City Heart Study and the Copenhagen General Population Study, with and without type 2 diabetes. Eligible participants had data for lipoprotein(a) concentrations in plasma, LPA kringle IV type 2 (KIV-2) sums of repeats (affecting both isoform size and plasma concentrations), and carrier status for the LPA single-nucleotide polymorphism rs10455872 (mainly affecting concentrations in plasma).

FINDINGS

77,901 individuals had lipoprotein(a) data, of whom 28,567 (36·7%) had all three measurements. Low concentrations of lipoprotein(a) in plasma were associated with risk of type 2 diabetes, with adjusted odds ratios of 1·26 (1·09-1·45), 1·17 (1·01-1·36), 1·04 (0·90-1·21), and 1·05 (95% CI 0·90-1·22), respectively, for quintiles 1-4, compared with quintile 5 concentrations. High KIV-2 sums of repeats were associated with risk of type 2 diabetes (adjusted odds ratio 1·16, 95% CI 1·05-1·28) for KIV-2 quintile 5 versus quintiles 1-4 combined. Being a carrier of rs10455872 did not affect risk of type 2 diabetes. For a halving of lipoprotein(a) concentrations, the instrumental variable estimate of the causal odds ratio for type 2 diabetes was 1·15 (95% CI 1·05-1·27) for KIV-2 sum of repeats and 0·99 (0·95-1·03) for rs10455872 genotype.

INTERPRETATION

Low lipoprotein(a) concentrations alone seem not to be causally associated with type 2 diabetes, but a causal association for large lipoprotein(a) isoform size cannot be excluded.

FUNDING

Danish Heart Foundation, Danish Council for Independent Research-Medical Sciences, IMK Almene Fund, and Johan and Lise Boserup's Fund.

Identification of genetic variants that affect histone modifications in human cells

Histone modifications are important markers of function and chromatin state, yet the DNA sequence elements that direct them to specific genomic locations are poorly understood. Here, we identify hundreds of quantitative trait loci, genome-wide, that affect histone modification or RNA polymerase II (Pol II) occupancy in Yoruba lymphoblastoid cell lines (LCLs). In many cases, the same variant is associated with quantitative changes in multiple histone marks and Pol II, as well as in deoxyribonuclease I sensitivity and nucleosome positioning. Transcription factor binding site polymorphisms are correlated overall with differences in local histone modification, and we identify specific transcription factors whose binding leads to histone modification in LCLs. Furthermore, variants that affect chromatin at distal regulatory sites frequently also direct changes in chromatin and gene expression at associated promoters.

A meta-analysis of genome-wide association studies for adiponectin levels in East Asians identifies a novel locus near WDR11-FGFR2

Blood levels of adiponectin, an adipocyte-secreted protein correlated with metabolic and cardiovascular risks, are highly heritable. Genome-wide association (GWA) studies for adiponectin levels have identified 14 loci harboring variants associated with blood levels of adiponectin. To identify novel adiponectin-associated loci, particularly those of importance in East Asians, we conducted a meta-analysis of GWA studies for adiponectin in 7827 individuals, followed by two stages of replications in 4298 and 5954 additional individuals. We identified a novel adiponectin-associated locus on chromosome 10 near WDR11-FGFR2 (P = 3.0 × 10(-14)) and provided suggestive evidence for a locus on chromosome 12 near OR8S1-LALBA (P = 1.2 × 10(-7)). Of the adiponectin-associated loci previously described, we confirmed the association at CDH13 (P = 6.8 × 10(-165)), ADIPOQ (P = 1.8 × 10(-22)), PEPD (P = 3.6 × 10(-12)), CMIP (P = 2.1 × 10(-10)), ZNF664 (P = 2.3 × 10(-7)) and GPR109A (P = 7.4 × 10(-6)). Conditional analysis at ADIPOQ revealed a second signal with suggestive evidence of association only after conditioning on the lead SNP (Pinitial = 0.020; Pconditional = 7.0 × 10(-7)). We further confirmed the independence of two pairs of closely located loci (<2 Mb) on chromosome 16 at CMIP and CDH13, and on chromosome 12 at GPR109A and ZNF664. In addition, the newly identified signal near WDR11-FGFR2 exhibited evidence of association with triglycerides (P = 3.3 × 10(-4)), high density lipoprotein cholesterol (HDL-C, P = 4.9 × 10(-4)) and body mass index (BMI)-adjusted waist-hip ratio (P = 9.8 × 10(-3)). These findings improve our knowledge of the genetic basis of adiponectin variation, demonstrate the shared allelic architecture for adiponectin with lipids and central obesity and motivate further studies of underlying mechanisms.

Approaches for strengthening causal inference regarding prenatal risk factors for childhood behavioural and psychiatric disorders

BACKGROUND

The risk of childhood behavioural and psychiatric diseases could be substantially reduced if modifiable risk factors for these disorders were identified. The critical period for many of these exposures is likely to be in utero as this is the time when brain development is most rapid. However, due to confounding and other limitations of traditional epidemiological studies, identification of causal risk factors has proved challenging and on the whole research in this area has not been fruitful.

SCOPE

In this review, we highlight several alternative approaches including; comparisons across settings, the use of negative controls and natural experiments, which includes migration studies, studies of individuals conceived using in vitro fertilisation and not least Mendelian randomisation. We have illustrated these approaches using examples of behavioural and psychiatric disorders.

CONCLUSION

By having these approaches outlined together in one review, researchers can consider which of these methods would be most suitable for their study question. We have particularly focussed on Mendelian randomisation, as this is a relatively novel concept, in doing so, we have illustrated the concept and discussed the implementation and the limitations of this approach.

Efficient design for Mendelian randomization studies: subsample and 2-sample instrumental variable estimators

Mendelian randomization (MR) is a method for estimating the causal relationship between an exposure and an outcome using a genetic factor as an instrumental variable (IV) for the exposure. In the traditional MR setting, data on the IV, exposure, and outcome are available for all participants. However, obtaining complete exposure data may be difficult in some settings, due to high measurement costs or lack of appropriate biospecimens. We used simulated data sets to assess statistical power and bias for MR when exposure data are available for a subset (or an independent set) of participants. We show that obtaining exposure data for a subset of participants is a cost-efficient strategy, often having negligible effects on power in comparison with a traditional complete-data analysis. The size of the subset needed to achieve maximum power depends on IV strength, and maximum power is approximately equal to the power of traditional IV estimators. Weak IVs are shown to lead to bias towards the null when the subsample is small and towards the confounded association when the subset is relatively large. Various approaches for confidence interval calculation are considered. These results have important implications for reducing the costs and increasing the feasibility of MR studies.

Mendelian randomization studies do not support a causal role for reduced circulating adiponectin levels in insulin resistance and type 2 diabetes

Adiponectin is strongly inversely associated with insulin resistance and type 2 diabetes, but its causal role remains controversial. We used a Mendelian randomization approach to test the hypothesis that adiponectin causally influences insulin resistance and type 2 diabetes. We used genetic variants at the ADIPOQ gene as instruments to calculate a regression slope between adiponectin levels and metabolic traits (up to 31,000 individuals) and a combination of instrumental variables and summary statistics-based genetic risk scores to test the associations with gold-standard measures of insulin sensitivity (2,969 individuals) and type 2 diabetes (15,960 case subjects and 64,731 control subjects). In conventional regression analyses, a 1-SD decrease in adiponectin levels was correlated with a 0.31-SD (95% CI 0.26-0.35) increase in fasting insulin, a 0.34-SD (0.30-0.38) decrease in insulin sensitivity, and a type 2 diabetes odds ratio (OR) of 1.75 (1.47-2.13). The instrumental variable analysis revealed no evidence of a causal association between genetically lower circulating adiponectin and higher fasting insulin (0.02 SD; 95% CI -0.07 to 0.11; N = 29,771), nominal evidence of a causal relationship with lower insulin sensitivity (-0.20 SD; 95% CI -0.38 to -0.02; N = 1,860), and no evidence of a relationship with type 2 diabetes (OR 0.94; 95% CI 0.75-1.19; N = 2,777 case subjects and 13,011 control subjects). Using the ADIPOQ summary statistics genetic risk scores, we found no evidence of an association between adiponectin-lowering alleles and insulin sensitivity (effect per weighted adiponectin-lowering allele: -0.03 SD; 95% CI -0.07 to 0.01; N = 2,969) or type 2 diabetes (OR per weighted adiponectin-lowering allele: 0.99; 95% CI 0.95-1.04; 15,960 case subjects vs. 64,731 control subjects). These results do not provide any consistent evidence that interventions aimed at increasing adiponectin levels will improve insulin sensitivity or risk of type 2 diabetes.

Mendelian randomization analysis associates increased serum urate, due to genetic variation in uric acid transporters, with improved renal function

Increased serum urate predicts chronic kidney disease independent of other risk factors. The use of xanthine oxidase inhibitors coincides with improved renal function. Whether this is due to reduced serum urate or reduced production of oxidants by xanthine oxidase or another physiological mechanism remains unresolved. Here we applied Mendelian randomization, a statistical genetics approach allowing disentangling of cause and effect in the presence of potential confounding, to determine whether lowering of serum urate by genetic modulation of renal excretion benefits renal function using data from 7979 patients of the Atherosclerosis Risk in Communities and Framingham Heart studies. Mendelian randomization by the two-stage least squares method was done with serum urate as the exposure, a uric acid transporter genetic risk score as instrumental variable, and estimated glomerular filtration rate and serum creatinine as the outcomes. Increased genetic risk score was associated with significantly improved renal function in men but not in women. Analysis of individual genetic variants showed the effect size associated with serum urate did not correlate with that associated with renal function in the Mendelian randomization model. This is consistent with the possibility that the physiological action of these genetic variants in raising serum urate correlates directly with improved renal function. Further studies are required to understand the mechanism of the potential renal function protection mediated by xanthine oxidase inhibitors.

Elevated Remnant Cholesterol Causes Both Low-Grade Inflammation and Ischemic Heart Disease, Whereas Elevated Low-Density Lipoprotein Cholesterol Causes Ischemic Heart Disease Without Inflammation

Background—

Elevated nonfasting remnant cholesterol and low-density lipoprotein (LDL) cholesterol are causally associated with ischemic heart disease (IHD), but whether elevated nonfasting remnant cholesterol and LDL cholesterol both cause low-grade inflammation is currently unknown.

Methods and Results—

We studied 60 608 individuals from the Copenhagen General Population Study, the Copenhagen City Heart Study, and the Copenhagen Ischemic Heart Disease study, of whom 10 668 had IHD diagnosed between 1977 and 2011. We genotyped for variants affecting levels of nonfasting remnant cholesterol, LDL cholesterol, C-reactive protein by CRP alleles, and C-reactive protein by IL6R alleles. Using a multidirectional mendelian randomization design, we investigated possible causal associations between the lipoproteins and C-reactive protein and between the lipoproteins and IHD. A 1-mmol/L(39 mg/dL) higher level of nonfasting remnant cholesterol was associated observationally with a 37% (95% confidence interval, 35–39) higher C-reactive protein level and causally with a 28% (95% confidence interval, 10–48) higher level. For LDL cholesterol, a 1-mmol/L (39-mg/dL) higher level was associated observationally with a 7% (95% confidence interval, 6–7) higher C-reactive protein level, but we found no causal association. Likewise, higher levels of C-reactive protein did not associate causally with elevated nonfasting remnant cholesterol or LDL cholesterol. Finally, the causal risk ratio for IHD for a 1-mmol/L (39-mg/dL) higher level was 3.3 (95% confidence interval, 2.1–5.2) for nonfasting remnant cholesterol and 1.8 (95% confidence interval, 1.5–2.2) for LDL cholesterol. The causal associations for remnant cholesterol were present even in those without diabetes mellitus and obesity.

Conclusions—

Elevated nonfasting remnant cholesterol is causally associated with low-grade inflammation and with IHD, whereas elevated LDL cholesterol is associated causally with IHD without inflammation.

A causal role for circulating miR-34b in osteosarcoma

PURPOSE

To investigate the associations between plasma miR-34b/c expression levels and osteosarcoma (OS).

SUBJECTS AND METHODS

A case-control study was conducted in 133 patients with OS and 133 controls. MiR-34b/c levels were detected by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) assays. Genotyping of SNP rs4938723 was done using the TaqMan assay. The causal association was examined by mendelian randomization analysis.

RESULTS

Plasma miR-34b level was significantly lower in OS patients than in controls (P = 0.001). Expression levels of miR-34b in OS tissues decreased (P = 3.22 × 10(-4)) and was significantly related with its expression in plasma (r = 0.21, P = 0.004). Compared with wild-type TT genotype, the variant genotypes of rs4938723 TC/CC were significantly associated with increased OS risk (TC vs. TT: OR, 1.97 [95% CI: 1.40-2.55], P = 0.021; CC vs. TT: OR, 2.76 [95% CI: 2.00-3.53], P = 0.009; TC + CC vs. TT: OR, 2.16 [95% CI: 1.61-2.70], P = 0.006), consistent with its decreased effect on plasma miR-34b (TC vs. TT: -0.32 (-0.43, -0.21), P < 0.001; CC vs. TT: -0.70 (-0.84, -0.56), P < 0.001; TC + CC vs. TT: -0.42 (-0.53, -0.32), P < 0.001). Adjustment for miR-34b completely abolished the association between SNP rs4938723 and OS risk (P > 0.05). In addition, plasma expression levels of miR-34b were significantly decreased in the metastatic patients compared with that in the non-metastatic ones (P = 0.004).

CONCLUSION

Plasma miR-34b was causally associated with OS risk and related with its metastatic status, suggesting that plasma miR-34b might be a novel biomarker and a potential treatment target for OS.

Integrating biomarker information within trials to evaluate treatment mechanisms and efficacy for personalised medicine

BACKGROUND

The development of personalised (stratified) medicine is intrinsically dependent on an understanding of treatment-effect mechanisms (effects on therapeutic targets that mediate the effect of the treatment on clinical outcomes). There is a need for clinical trial data for the joint evaluation of treatment efficacy, the utility of predictive markers as indicators of treatment efficacy, and the mediational mechanisms proposed as the explanation of these effects.

PURPOSE

(1) To review the problem of confounding (common causes) for the drawing of valid inferences concerning treatment-effect mechanisms, even when the data have been generated using a randomised controlled trial, and (2) to suggest and illustrate solutions to this problem of confounding.

RESULTS

We illustrate the potential of the predictive biomarker stratified design, together with baseline measurement of all known prognostic markers, to enable us to evaluate both the utility of the predictive biomarker in such a stratification and, perhaps more importantly, to estimate how much of the treatment's effect is actually explained by changes in the putative mediator. The analysis strategy involves the use of instrumental variable (IV) regression, using the treatment by predictive biomarker interaction as an IV - a refined, much more powerful, and (in the present context) subtle use of Mendelian randomisation.

CONCLUSION

Personalised (stratified) medicine and treatment-effect mechanisms evaluation are inextricably linked. Stratification without corresponding mechanisms evaluation lacks credibility. In the presence of mediator-outcome confounding, mechanisms evaluation is dependent on stratification for its validity. Both stratification and treatment-effect mediation can be evaluated using a biomarker stratified trial design together with detailed baseline measurement of all known prognostic biomarkers and other prognostic covariates. Direct and indirect (mediated) effects should be estimated through the use of IV methods (the IV being the predictive marker by treatment interaction) together with adjustments for all known prognostic markers (confounders) - the latter adjustments contributing to increased precision (as in a conventional analysis of treatment effects) rather than bias reduction.

Linking vitamin d deficiency to inflammatory bowel disease

Inflammatory bowel disease is associated with industrialization, and its incidence has increased markedly over time. The prospect of reversing these trends motivates the search for the agent(s) involved. Modernity entails several physical and behavioral modifications that compromise both the photosynthesis of cholecalciferol in the skin and of its bioavailability. Although deficiency in this "vitamin" has therefore emerged as a leading candidate, and despite the publication of a randomized control trial that showed a trend toward statistically significant benefit in Crohn's disease, its causal agency has yet to be demonstrated by an adequately powered study. We discuss the strengths and weaknesses of the case being made by epidemiologists, geneticists, clinicians, and basic researchers, and consolidate their findings into a model that provides mechanistic plausibility to the claim. Specifically, converging data sets suggest that local activation of vitamin D coordinates the activity of the innate and adaptive arms of immunity, and of the intestinal epithelium, in a manner that promotes barrier integrity, facilitates the clearance of translocated flora, and diverts CD4 T cell development away from inflammatory phenotypes. Because smoking is an important risk-altering exposure, we also discuss its newly established melanizing effect and other emerging evidence linking tobacco use to immune function through vitamin D pathways.

An analysis of the association between the vitamin D pathway and serum 25-hydroxyvitamin D levels in a healthy Chinese population

Vitamin D deficiency has been recognized as a major public health issue worldwide. Recent studies have indicated that genetic factors might play an important role in determining serum 25-hydroxyvitamin D [25(OH)D] levels in Caucasians and African Americans. However, the genes that contribute to the variation in serum 25(OH)D levels in Chinese are unknown. In this study, we screened 15 key genes within the vitamin D metabolic pathway using 96 single-nucleotide polymorphism (SNP) markers in a group of 2897 unrelated healthy Chinese subjects. Significant confounding factors that may influence the variability in serum 25(OH)D levels were used as covariates for association analyses. An association test for quantitative traits was performed to evaluate the association between candidate genes and serum 25(OH)D levels. In the present study, variants and/or haplotypes in GC, CYP2R1, and DHCR7/NADSYN1 were identified as being associated with 25(OH)D levels. Participants with three or four risk alleles of the two variants (GC-rs4588 and CYP2R1-rs10766197) had an increased chance of presenting with a 25(OH)D concentration lower than 20 ng/mL (odds ratio 2.121, 95% confidence interval 1.586-2.836, p = 6.1 × 10(-8) ) compared with those lacking the risk alleles. Each additional copy of a risk allele was significantly associated with a 0.12-fold decrease in the log-25(OH)D concentration (p = 3.7 × 10(-12) ). Haplotype TGA of GC rs705117-rs2282679-rs1491710, haplotype GAGTAC of GC rs842999-rs705120-rs222040-rs4588-rs7041-rs10488854, haplotype CA of GC rs1155563-rs222029, and haplotype AAGA of CYP2R1 rs7936142-rs12794714-rs2060793-rs16930609 were genetic risk factors toward a lower 25(OH)D concentration. In contrast, haplotype TGGGCCC of DHCR7/NADSYN1 rs1790349-rs7122671-rs1790329-rs11606033-rs2276360-rs1629220-rs2282618 were genetic protective factors. The results suggest that the GC, CYP2R1, and DHCR7/NADSYN1 genes might contribute to variability in the serum 25(OH)D levels in a healthy Chinese population in Shanghai. These markers could be used as tools in Mendelian randomization analyses of vitamin D, and they could potentially be drug targets in the Chinese population in Shanghai.

Plasma levels of the arterial wall protein fibulin-1 are associated with carotid-femoral pulse wave velocity: a cross-sectional study

BACKGROUND

The arterial system in diabetic patients is characterized by generalized non-atherosclerotic alterations in the vascular extracellular matrix causing increased arterial stiffness compared with subjects without diabetes. The underlying pathophysiology remains elusive. The elastin-associated extracellular matrix protein, fibulin-1, was recently found in higher concentrations in the arterial wall and in plasma in patients with long duration type 2 diabetes. Furthermore, plasma fibulin-1 independently predicted total mortality and was associated with pulse pressure, an indirect measure of arterial stiffness. Whether plasma fibulin-1 is associated with arterial stiffness at earlier phases of type 2 diabetes has not been determined.

METHODS

In this cross-sectional study, we examined 90 patients with recently diagnosed type 2 diabetes (< 5 years) and 90 gender- and age-matched controls. Plasma fibulin-1 was measured immunochemically. Arterial stiffness was assessed by carotid-femoral Pulse Wave Velocity (PWV). Differences in means were assessed by t-tests. Associations were assessed by multivariate regression analyses.

RESULTS

Plasma fibulin-1 levels were lower in the diabetic group compared with the control group, 93 ± 28 vs 106 ± 30 μg/mL, p = 0.005. In unadjusted analysis of the total study sample, plasma fibulin-1 was not associated with PWV, p = 0.46. However, with adjustment for the confounders age, gender, mean blood pressure, heart rate, body mass index, diabetes and glomerular filtration rate, a 10 μg/mL increase in plasma fibulin was associated with 0.09 ± 0.04 m/s increase in PWV, p < 0.05. In subgroup analysis, plasma fibulin-1 was associated with PWV in the diabetes group, (0.16 ± 0.07 m/s increase in PWV per 10 μg/mL increase in plasma fibulin-1, p<0.05), but not controls, β = 0.021 ± 0.057 m/s per 10 μg/mL, p = 0.70. The association remained significant in the diabetes group after adjustment for covariates, p < 0.05.

CONCLUSIONS

Plasma fibulin-1 is independently associated with PWV. Yet, as the plasma level of fibulin-1 was lower in patients with recently diagnosed type 2 diabetes than in healthy controls, plasma fibulin-1 levels are not a simple marker of the degree of arterial stiffening. Further studies are needed to determine the exact role of fibulin-1 in arterial stiffness and cardiovascular risk in patients with type 2 diabetes.

Moderate alcohol use and cardiovascular disease from Mendelian randomization

Background

Observational studies show moderate alcohol use negatively associated with ischemic heart disease (IHD) and cardiovascular disease (CVD). However, healthier attributes among moderate users compared to never users may confound the apparent association. A potentially less biased way to examine the association is Mendelian randomization, using alcohol metabolizing genes which influence alcohol use.

Methods

We used instrumental variable analysis with aldehyde dehydrogenase 2 (ALDH2) genotypes (AA/GA/GG) as instrumental variables for alcohol use to examine the association of alcohol use (10 g ethanol/day) with CVD risk factors (blood pressure, lipids and glucose) and morbidity (self-reported IHD and CVD) among men in the Guangzhou Biobank Cohort Study.

Results

ALDH2 genotypes were a credible instrument for alcohol use (F-statistic 74.6). Alcohol was positively associated with HDL-cholesterol (0.05 mmol/L per alcohol unit, 95% confidence interval (CI) 0.02 to 0.08) and diastolic blood pressure (1.15 mmHg, 95% CI 0.23 to 2.07) but not with systolic blood pressure (1.00 mmHg, 95% CI -0.74 to 2.74), LDL-cholesterol (0.03 mmol/L, 95% CI -0.03 to 0.08), log transformed triglycerides (0.03 mmol/L, 95% CI -0.01 to 0.08) or log transformed fasting glucose (0.01 mmol/L, 95% CI -0.006 to 0.03), self-reported CVD (odds ratio (OR) 0.98, 95% CI 0.76 to 1.27) or self-reported IHD (OR 1.10, 95% CI 0.83 to 1.45).

Conclusion

Low to moderate alcohol use among men had the expected effects on most CVD risk factors but not fasting glucose. Larger studies are needed to confirm the null associations with IHD, CVD and fasting glucose.

Prenatal alcohol exposure and childhood balance ability: findings from a UK birth cohort study

OBJECTIVE

To investigate the association of prenatal alcohol exposure with balance in10-year-old children.

DESIGN

Population-based prospective longitudinal study.

SETTING

Former Avon region of UK (Southwest England).

PARTICIPANTS

6915 children from the Avon Longitudinal Study of Parents and Children who had a balance assessment at age 10 and had data on maternal alcohol consumption.

OUTCOME MEASURES

3 composite balance scores: dynamic balance (beam-walking), static balance eyes open, static balance eyes closed (heel-to-toe balance on a beam and standing on one leg, eyes open or closed).

RESULTS

Most mothers (95.5%) consumed no-to-moderate amounts (3-7 glasses/week) of alcohol during pregnancy. Higher total-alcohol consumption was associated with maternal-social advantage, whereas binge drinking (≥4 units/day) and abstinence were associated with maternal social disadvantage. No evidence was found of an adverse effect of maternal-alcohol consumption on childhood balance. Higher maternal-alcohol use during pregnancy was generally associated with better offspring outcomes, with some specific effects appearing strong (static balance eyes open and moderate total alcohol exposure at 18 weeks, adjusted OR 1.23 (95% CI 1.01 to 1.49); static balance eyes closed and moderate total alcohol exposure at 18 weeks, adjusted OR 1.25 (95% CI 1.06 to 1.48). Similar results were found for both paternal and postnatal maternal alcohol exposure. A Mendelian-randomization approach was used to estimate the association between maternal genotype and offspring balance using the non-synonymous variant rs1229984*A (ADH1B) to proxy for lower maternal alcohol consumption; no strong associations were found between this genotype/proxy and offspring balance.

CONCLUSIONS

No evidence was found to indicate that moderate maternal alcohol consumption in this population sample had an adverse effect on offspring balance at age 10. An apparent beneficial effect of higher total maternal alcohol consumption on offspring balance appeared likely to reflect residual confounding.

Genotype misclassification in genetic association studies of the rs1042522 TP53 (Arg72Pro) polymorphism: a systematic review of studies of breast, lung, colorectal, ovarian, and endometrial cancer

Preferential loss of heterozygosity at the rs1042522 locus of the tumor protein 53 gene (TP53) (Arg72Pro) is observed in several tumors. Genetic association studies in oncology often use tumor tissue rather than unaffected tissue for genotyping; in such cases, loss of heterozygosity at the TP53 locus could lead to differential misclassification and could bias estimates of association. We searched multiple databases (through March 8, 2011) for studies investigating the association of Arg72Pro with breast, lung, colorectal, ovarian, or endometrial cancer. Meta-analysis was performed with multilevel Bayesian models. Informative priors for the bias effect were derived from a meta-analysis of the same polymorphism in cervical cancer. Of 160 studies (68 breast, 42 lung, 26 colorectal, 16 ovarian, and 8 endometrial cancer), 22 used tumor tissue as the source of genotyping material for cases. Use of tumor tissue versus other sources of genotyping material was associated with an apparent protective effect of the proline allele (relative odds ratio = 0.78, 95% credible interval: 0.70, 0.88). The probability that use of tumor tissue induced bias was estimated to be higher than 99%. Use of tumor tissue as the source of genotyping material for cases is associated with significant bias in the estimate of the genetic effect in cancer genetic association studies.

Pleiotropy in complex traits: challenges and strategies

Genome-wide association studies have identified many variants that each affects multiple traits, particularly across autoimmune diseases, cancers and neuropsychiatric disorders, suggesting that pleiotropic effects on human complex traits may be widespread. However, systematic detection of such effects is challenging and requires new methodologies and frameworks for interpreting cross-phenotype results. In this Review, we discuss the evidence for pleiotropy in contemporary genetic mapping studies, new and established analytical approaches to identifying pleiotropic effects, sources of spurious cross-phenotype effects and study design considerations. We also outline the molecular and clinical implications of such findings and discuss future directions of research.

Genetic variants associated with glycine metabolism and their role in insulin sensitivity and type 2 diabetes

Circulating metabolites associated with insulin sensitivity may represent useful biomarkers, but their causal role in insulin sensitivity and diabetes is less certain. We previously identified novel metabolites correlated with insulin sensitivity measured by the hyperinsulinemic-euglycemic clamp. The top-ranking metabolites were in the glutathione and glycine biosynthesis pathways. We aimed to identify common genetic variants associated with metabolites in these pathways and test their role in insulin sensitivity and type 2 diabetes. With 1,004 nondiabetic individuals from the RISC study, we performed a genome-wide association study (GWAS) of 14 insulin sensitivity-related metabolites and one metabolite ratio. We replicated our results in the Botnia study (n = 342). We assessed the association of these variants with diabetes-related traits in GWAS meta-analyses (GENESIS [including RISC, EUGENE2, and Stanford], MAGIC, and DIAGRAM). We identified four associations with three metabolites-glycine (rs715 at CPS1), serine (rs478093 at PHGDH), and betaine (rs499368 at SLC6A12; rs17823642 at BHMT)-and one association signal with glycine-to-serine ratio (rs1107366 at ALDH1L1). There was no robust evidence for association between these variants and insulin resistance or diabetes. Genetic variants associated with genes in the glycine biosynthesis pathways do not provide consistent evidence for a role of glycine in diabetes-related traits.

The Impact of Cigarette Quitting during Pregnancy on Other Prenatal Health Behaviors

Several economic studies have evaluated the effects of cigarette smoking and quitting on other health behaviors such as alcohol use and weight gain. However, there is little research that evaluates the effects of cigarette quitting during pregnancy on other health behaviors such as caloric intake, alcohol consumption, multivitamin use, and caffeine intake. In this paper, we evaluate these effects and employ a genetic variant that predicts cigarette quitting to aid in identification. We find some evidence that cigarette quitting during pregnancy may increase multivitamin use and caloric intake and reduce caffeine consumption.

Genetic variability related to serum uric acid concentration and risk of Parkinson's disease

BACKGROUND

Low serum uric acid (UA) levels have been associated with increased Parkinson's disease (PD) risk and accelerated disease progression. We analyzed the effect of polymorphisms in 9 genes influencing serum UA concentration on the risk of PD.

METHODS

We genotyped SLC2A9 rs734553, ABCG2 rs2231142, SLC17A1 rs1183201, SLC22A11 rs17300741, SLC22A12 rs505802, GCKR rs780094, PDZK1 rs12129861, LRRC16A+SCGN rs742132, and SLC16A9 rs12356193 in 1061 PD patients and 754 controls. For each subject we calculated a cumulative genetic risk score (GRS), defined as the total number of PD-risk alleles (range, 2-15) associated to lower serum UA levels. Serum UA levels were measured in a subgroup of 365 PD cases and 132 controls.

RESULTS

Serum UA levels were significantly lower in men with PD than in controls. Subjects (both men and women) carrying more than 9 risk alleles (third GRS tertile) had a 1.5 higher risk of developing PD than subjects with less than 8 risk alleles (first GRS tertile). An inverse correlation was observed between higher GRS and lower serum UA concentration in both men and women.

CONCLUSIONS

Genetic variability influencing serum UA levels might modify susceptibility to PD.

Model selection approach suggests causal association between 25-hydroxyvitamin D and colorectal cancer

INTRODUCTION

Vitamin D deficiency has been associated with increased risk of colorectal cancer (CRC), but causal relationship has not yet been confirmed. We investigate the direction of causation between vitamin D and CRC by extending the conventional approaches to allow pleiotropic relationships and by explicitly modelling unmeasured confounders.

METHODS

Plasma 25-hydroxyvitamin D (25-OHD), genetic variants associated with 25-OHD and CRC, and other relevant information was available for 2645 individuals (1057 CRC cases and 1588 controls) and included in the model. We investigate whether 25-OHD is likely to be causally associated with CRC, or vice versa, by selecting the best modelling hypothesis according to Bayesian predictive scores. We examine consistency for a range of prior assumptions.

RESULTS

Model comparison showed preference for the causal association between low 25-OHD and CRC over the reverse causal hypothesis. This was confirmed for posterior mean deviances obtained for both models (11.5 natural log units in favour of the causal model), and also for deviance information criteria (DIC) computed for a range of prior distributions. Overall, models ignoring hidden confounding or pleiotropy had significantly poorer DIC scores.

CONCLUSION

Results suggest causal association between 25-OHD and colorectal cancer, and support the need for randomised clinical trials for further confirmations.

Mendelian randomization and estimation of treatment efficacy for chronic diseases

Prevention and treatment of common noncommunicable chronic diseases have been revolutionized by the development of therapies. Recently, several randomized controlled trials (RCTs) designed to assess the efficacy of new therapies targeted at well-established risk factors for noncommunicable chronic diseases have reported lower benefits than expected. Subsequent observational analysis of the same trial data has not clarified these unexpected findings. Mendelian randomization (MR) provides an approach for estimating causal effects from observational or trial data and thus provides information complementary to that from an RCT. An RCT assesses the efficacy of a therapy but does not usually confirm the underlying mechanistic pathway. In contrast, an MR study does not assess the efficacy of a therapy but rather assesses causal effects on an underlying mechanistic pathway. We suggest that incorporating an MR study into an RCT at the design stage would improve etiologic understanding of current therapies and enhance the search for therapies for the significant amount of noncommunicable chronic diseases that resists current treatments.

Causal inference of the effect of adiposity on bone mineral density in adults

OBJECTIVE

The causal effect of adipose tissue on bone mass and the direction of its net influence have not been directly assessed in adult humans. Using the Mendelian randomization analysis, we assessed the causality of adiposity in measurements of bone mass in adult males and females.

DESIGN AND METHODS

Subjects consisted of 2154 adults aged 25-54 years from a cross-sectional cohort of the employees of the Electricity Generating Authority of Thailand. Body composition was determined after at least 3 h of fasting using multifrequency bioelectrical impedance analysis. Bone mineral density (BMD) was assessed by dual energy X-ray absorptiometry. A polymorphism in the fat mass and obesity-associated gene (FTO rs9939609) was used as an instrument in the Mendelian randomization analysis.

RESULTS

The genotype distribution of the FTO rs9939609 polymorphism was 61·1% TT, 33·9% AT and 5·0% AA. The average body mass index (BMI), body fat mass and percentage body fat were 23·9 kg/m(2) (SD = 3·6), 17·9 kg (SD = 6·6) and 26·8% (SD = 7·2), respectively. The FTO rs9939609 polymorphism was significantly correlated with BMI (coefficient = 0·673 kg/m(2) , P < 0·001), body fat mass (coefficient = 0·948 kg, P < 0·001) and percentage body fat (coefficient = 0·759%, P < 0·01). An instrumental variable (IV) regression model, using BMI as the intermediate phenotype, suggested that FTO was a strong IV. Also, the FTO-BMI polymorphism was significantly associated with total hip and femoral neck BMD but was not correlated with total spine BMD, with estimated correlation coefficients of 0·0189 (95% CI: 0·0046, 0·0332), 0·0149 (95% CI: 0·0030, 0·0268) and 0·0025 (95% CI: -0·0131, 0·0136) g/cm(2) , respectively. The variances of BMDs explained by the FTO-BMI were 19·0%, 21·3% and 1·1%, respectively. Similar trends were also observed for the FTO-body fat mass and FTO-percentage body fat correlations.

CONCLUSIONS

Mendelian randomization analysis suggests that adiposity might be causally related to BMD at the femur but not at the spine.

Adhesion molecules, endothelin-1 and lung function in seven population-based cohorts

CONTEXT

Endothelial function is abnormal in chronic obstructive pulmonary disease (COPD); whether endothelial dysfunction causes COPD is unknown.

OBJECTIVE

Test associations of endothelial biomarkers with FEV1 using instrumental variables.

METHODS

Among 26 907 participants with spirometry, ICAM-1, P-selectin, E-selectin and endothelin-1 were measured in subsets.

RESULTS

ICAM-1 and P-selectin were inversely associated with FEV1 among European-Americans (-29 mL and -34 mL per standard deviation of log-transformed biomarker, p < 0.001), as was endothelin-1 among African-Americans (-22 mL, p = 0.008). Genetically-estimated ICAM-1 and P-selectin were not significantly associated with FEV1. The instrumental variable for endothelin-1 was non-informative.

CONCLUSION

Although ICAM-1, P-selectin and endothelin-1 were inversely associated with FEV1, associations for ICAM-1 and P-selectin do not appear causal.

Evidence of a causal relationship between adiponectin levels and insulin sensitivity: a Mendelian randomization study

The adipocyte-secreted protein adiponectin is associated with insulin sensitivity in observational studies. We aimed to evaluate whether this relationship is causal using a Mendelian randomization approach. In a sample of Swedish men aged 71 years (n = 942) from the Uppsala Longitudinal Study of Adult Men (ULSAM), insulin sensitivity (M/I ratio) was measured by the euglycemic insulin clamp. We used three genetic variants in the ADIPOQ locus as instrumental variables (IVs) to estimate the potential causal effect of adiponectin on insulin sensitivity and compared these with results from conventional linear regression. The three ADIPOQ variants, rs17300539, rs3774261, and rs6444175, were strongly associated with serum adiponectin levels (all P ≤ 5.3 × 10(-9)) and were also significantly associated with M/I ratio in the expected direction (all P ≤ 0.022). IV analysis confirmed that genetically determined adiponectin increased insulin sensitivity (β = 0.47-0.81, all P ≤ 0.014) comparable with observational estimates (β = 0.50, all P(difference) ≥ 0.136). Adjustment for BMI and waist circumference partly explained the association of both genetically determined and observed adiponectin levels with insulin sensitivity. The observed association between higher adiponectin levels and increased insulin sensitivity is likely to represent a causal relationship partly mediated by reduced adiposity.

Using genetic proxies for lifecourse sun exposure to assess the causal relationship of sun exposure with circulating vitamin d and prostate cancer risk

BACKGROUND

Ecological and epidemiological studies have identified an inverse association of intensity and duration of sunlight exposure with prostate cancer, which may be explained by a reduction in vitamin D synthesis. Pigmentation traits influence sun exposure and therefore may affect prostate cancer risk. Because observational studies are vulnerable to confounding and measurement error, we used Mendelian randomization to examine the relationship of sun exposure with both prostate cancer risk and the intermediate phenotype, plasma levels of vitamin D.

METHODS

We created a tanning, a skin color, and a freckling score as combinations of single nucleotide polymorphisms that have been previously associated with these phenotypes. A higher score indicates propensity to burn, have a lighter skin color and freckles. The scores were tested for association with vitamin D levels (25-hydroxyvitamin-D and 1,25-dihydroxyvitamin-D) and prostate-specific antigen detected prostate cancer in 3,123 White British individuals enrolled in the Prostate Testing for cancer and Treatment (ProtecT) study.

RESULTS

The freckling score was inversely associated with 25(OH)D levels [change in 25(OH)D per score unit -0.27; 95% CI, -0.52% to -0.01%], and the tanning score was positively associated with prostate cancer risk (OR = 1.05; 95% CI, 1.02-1.09), after adjustment for population stratification and potential confounders.

CONCLUSIONS

Individuals who tend to burn are more likely to spend less time in the sun and consequently have lower plasma vitamin D levels and higher susceptibility to prostate cancer.

IMPACT

The use of pigmentation-related genetic scores is valuable for the assessment of the potential benefits of sun exposure with respect to prostate cancer risk.

Genetic risk for earlier menarche also influences peripubertal body mass index

It is unclear whether earlier age at menarche is associated with higher body mass index (BMI) because they share a common genetic underpinning. We investigated the impact of single nucleotide polymorphisms (SNPs) influencing menarche timing on peripubertal BMI. For 556 Fels Longitudinal Study children (277 boys/279 girls) born 1928-1992, a genetic risk score (GRS(42)) was computed as the sum of the number of risk alleles in 42 putative menarche SNPs. Serial BMI Z-scores within ±6.99 years from each individual's age at peak height velocity (Age@PHV) were grouped into seven time points (-6 years, -4 years, -2 years, Age@PHV, +2 years, +4 years, and +6 years). Heritability of BMI ranged from 0.53 to 0.85 across the time points. The effect of GRS(42) on BMI Z-scores at each time point was modeled using variance components-based procedures. GRS(42) had a significant (P < 0.05) effect at every time point; an increase of one risk allele was associated with an increase of 0.03-0.08 BMI Z-scores. A separate score (GRS(29)) was computed that excluded 13 of the menarche SNPs previously documented to also influence adiposity; significant main effects were observed at Age@PHV+4 and +6 years. This finding supports a causal effect of advanced sexual development on post-Age@PHV BMI. Significant positive GRS(42) (or GRS(29))-by-birth year interactions indicate that some genetic influences on BMI have amplified over the 20th century. This gene-by-environment interaction also suggests that children with a genetic predisposition to earlier sexual development might avoid elevated BMI through alteration of their nutritional environment.

Genetic predisposition to higher blood pressure increases coronary artery disease risk

Hypertension is a risk factor for coronary artery disease. Recent genome-wide association studies have identified 30 genetic variants associated with higher blood pressure at genome-wide significance (P<5 × 10(-8)). If elevated blood pressure is a causative factor for coronary artery disease, these variants should also increase coronary artery disease risk. Analyzing genome-wide association data from 22 233 coronary artery disease cases and 64 762 controls, we observed in the Coronary ARtery DIsease Genome-Wide Replication And Meta-Analysis (CARDIoGRAM) consortium that 88% of these blood pressure-associated polymorphisms were likewise positively associated with coronary artery disease, that is, they had an odds ratio >1 for coronary artery disease, a proportion much higher than expected by chance (P=4 × 10(-5)). The average relative coronary artery disease risk increase per each of the multiple blood pressure-raising alleles observed in the consortium was 3.0% for systolic blood pressure-associated polymorphisms (95% confidence interval, 1.8%-4.3%) and 2.9% for diastolic blood pressure-associated polymorphisms (95% confidence interval, 1.7%-4.1%). In substudies, individuals carrying most systolic blood pressure- and diastolic blood pressure-related risk alleles (top quintile of a genetic risk score distribution) had 70% (95% confidence interval, 50%-94%) and 59% (95% confidence interval, 40%-81%) higher odds of having coronary artery disease, respectively, as compared with individuals in the bottom quintile. In conclusion, most blood pressure-associated polymorphisms also confer an increased risk for coronary artery disease. These findings are consistent with a causal relationship of increasing blood pressure to coronary artery disease. Genetic variants primarily affecting blood pressure contribute to the genetic basis of coronary artery disease.

The effect of non-differential measurement error on bias, precision and power in Mendelian randomization studies

BACKGROUND

Mendelian randomization (MR) studies assess the causality of associations between exposures and disease outcomes using data on genetic determinants of the exposure. In this work, we explore the effect of exposure and outcome measurement error in MR studies.

METHODS

For continuous traits, we describe measurement error in terms of a theoretical regression of the measured variable on the true variable. We quantify error in terms of the slope (calibration) and the R(2) values (discrimination or classical measurement error). We simulated cohort data sets under realistic parameters and used two-stage least squares regression to assess the effect of measurement error for continuous exposures and outcomes on bias, precision and power. For simulations of binary outcomes, we varied sensitivity and specificity.

RESULTS

Discrimination error in continuous exposures and outcomes did not bias the MR estimate, and only outcome discrimination error substantially reduced power. Calibration error biased the MR estimate when the exposure and the outcome measures were not calibrated in a similar fashion, but power was not affected. For binary outcomes, exposure calibration error introduced substantial bias (with negligible impact on power), but exposure discrimination error did not. Reduced outcome specificity and, to a lesser degree, reduced sensitivity biased MR estimates towards the null.

CONCLUSIONS

Understanding the potential effects of measurement error is an important consideration when interpreting estimates from MR analyses. Based on these results, future MR studies should consider methods for accounting for such error and minimizing its impact on inferences derived from MR analyses.