MRC Integrative Epidemiology Unit

Background: Mendelian randomization (MR) is being increasingly used to strengthen causal inference in observational studies. Availability of summary data of genetic associations for a variety of phenotypes from large genome-wide association studies (GWAS) allows straightforward application of MR using summary data methods, typically in a two-sample design. In addition to the conventional inverse variance weighting (IVW) method, recently developed summary data MR methods, such as the MR-Egger and weighted median approaches, allow a relaxation of the instrumental variable assumptions. Methods: Here, a new method - the mode-based estimate (MBE) - is proposed to obtain a single causal effect estimate from multiple genetic instruments. The MBE is consistent when the largest number of similar (identical in infinite samples) individual-instrument causal effect estimates comes from valid instruments, even if the majority of instruments are invalid. We evaluate the performance of the method in simulations designed to mimic the two-sample summary data setting, and demonstrate its use by investigating the causal effect of plasma lipid fractions and urate levels on coronary heart disease risk. Results: The MBE presented less bias and lower type-I error rates than other methods under the null in many situations. Its power to detect a causal effect was smaller compared with the IVW and weighted median methods, but was larger than that of MR-Egger regression, with sample size requirements typically smaller than those available from GWAS consortia. Conclusions: The MBE relaxes the instrumental variable assumptions, and should be used in combination with other approaches in sensitivity analyses.

Large-scale cross-sectional and cohort studies have transformed our understanding of the genetic and environmental determinants of health outcomes. However, the representativeness of these samples may be limited-either through selection into studies, or by attrition from studies over time. Here we explore the potential impact of this selection bias on results obtained from these studies, from the perspective that this amounts to conditioning on a collider (i.e. a form of collider bias). Whereas it is acknowledged that selection bias will have a strong effect on representativeness and prevalence estimates, it is often assumed that it should not have a strong impact on estimates of associations. We argue that because selection can induce collider bias (which occurs when two variables independently influence a third variable, and that third variable is conditioned upon), selection can lead to substantially biased estimates of associations. In particular, selection related to phenotypes can bias associations with genetic variants associated with those phenotypes. In simulations, we show that even modest influences on selection into, or attrition from, a study can generate biased and potentially misleading estimates of both phenotypic and genotypic associations. Our results highlight the value of knowing which population your study sample is representative of. If the factors influencing selection and attrition are known, they can be adjusted for. For example, having DNA available on most participants in a birth cohort study offers the possibility of investigating the extent to which polygenic scores predict subsequent participation, which in turn would enable sensitivity analyses of the extent to which bias might distort estimates.

INTRODUCTION The cerebral cortex underlies our complex cognitive capabilities. Variations in human cortical surface area and thickness are associated with neurological, psychological, and behavioral traits and can be measured in vivo by magnetic resonance imaging (MRI). Studies in model organisms have identified genes that influence cortical structure, but little is known about common genetic variants that affect human cortical structure. RATIONALE To identify genetic variants associated with human cortical structure at both global and regional levels, we conducted a genome-wide association meta-analysis of brain MRI data from 51,665 individuals across 60 cohorts. We analyzed the surface area and average thickness of the whole cortex and 34 cortical regions with known functional specializations. RESULTS We identified 369 nominally genome-wide significant loci ( P < 5 × 10 −8 ) associated with cortical structure in a discovery sample of 33,992 participants of European ancestry. Of the 360 loci for which replication data were available, 241 loci influencing surface area and 66 influencing thickness remained significant after replication, with 237 loci passing multiple testing correction ( P < 8.3 × 10 −10 ; 187 influencing surface area and 50 influencing thickness). Common genetic variants explained 34% (SE = 3%) of the variation in total surface area and 26% (SE = 2%) in average thickness; surface area and thickness showed a negative genetic correlation ( r G = −0.32, SE = 0.05, P = 6.5 × 10 −12 ), which suggests that genetic influences have opposing effects on surface area and thickness. Bioinformatic analyses showed that total surface area is influenced by genetic variants that alter gene regulatory activity in neural progenitor cells during fetal development. By contrast, average thickness is influenced by active regulatory elements in adult brain samples, which may reflect processes that occur after mid-fetal development, such as myelination, branching, or pruning. When considered together, these results support the radial unit hypothesis that different developmental mechanisms promote surface area expansion and increases in thickness. To identify specific genetic influences on individual cortical regions, we controlled for global measures (total surface area or average thickness) in the regional analyses. After multiple testing correction, we identified 175 loci that influence regional surface area and 46 that influence regional thickness. Loci that affect regional surface area cluster near genes involved in the Wnt signaling pathway, which is known to influence areal identity. We observed significant positive genetic correlations and evidence of bidirectional causation of total surface area with both general cognitive functioning and educational attainment. We found additional positive genetic correlations between total surface area and Parkinson’s disease but did not find evidence of causation. Negative genetic correlations were evident between total surface area and insomnia, attention deficit hyperactivity disorder, depressive symptoms, major depressive disorder, and neuroticism. CONCLUSION This large-scale collaborative work enhances our understanding of the genetic architecture of the human cerebral cortex and its regional patterning. The highly polygenic architecture of the cortex suggests that distinct genes are involved in the development of specific cortical areas. Moreover, we find evidence that brain structure is a key phenotype along the causal pathway that leads from genetic variation to differences in general cognitive function. Identifying genetic influences on human cortical structure. ( A ) Measurement of cortical surface area and thickness from MRI. ( B ) Genomic locations of common genetic variants that influence global and regional cortical structure. ( C ) Our results support the radial unit hypothesis that the expansion of cortical surface area is driven by proliferating neural progenitor cells. ( D ) Cortical surface area shows genetic correlation with psychiatric and cognitive traits. Error bars indicate SE. IMAGE CREDITS: (A) K. COURTNEY; (C) M. R. GLASS

The Avon Longitudinal Study of Parents and Children (ALSPAC) is a prospective population-based study. Initial recruitment of pregnant women took place in 1990-1992 and the health and development of the index children from these pregnancies and their family members have been followed ever since. The eligible sampling frame was constructed retrospectively using linked recruitment and health service records. Additional offspring that were eligible to enrol in the study have been welcomed through major recruitment drives at the ages of 7 and 18 years; and through opportunistic contacts since the age of 7. This data note provides a status update on the recruitment of the index children since the age of 7 years with a focus on enrolment since the age of 18, which has not been previously described. A total of 913 additional G1 (the cohort of index children) participants have been enrolled in the study since the age of 7 years with 195 of these joining since the age of 18. This additional enrolment provides a baseline sample of 14,901 G1 participants who were alive at 1 year of age.

BACKGROUND: Reducing saturated fat reduces serum cholesterol, but effects on other intermediate outcomes may be less clear. Additionally, it is unclear whether the energy from saturated fats eliminated from the diet are more helpfully replaced by polyunsaturated fats, monounsaturated fats, carbohydrate or protein. OBJECTIVES: To assess the effect of reducing saturated fat intake and replacing it with carbohydrate (CHO), polyunsaturated (PUFA), monounsaturated fat (MUFA) and/or protein on mortality and cardiovascular morbidity, using all available randomised clinical trials. SEARCH METHODS: We updated our searches of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (Ovid) and Embase (Ovid) on 15 October 2019, and searched Clinicaltrials.gov and WHO International Clinical Trials Registry Platform (ICTRP) on 17 October 2019. SELECTION CRITERIA: Included trials fulfilled the following criteria: 1) randomised; 2) intention to reduce saturated fat intake OR intention to alter dietary fats and achieving a reduction in saturated fat; 3) compared with higher saturated fat intake or usual diet; 4) not multifactorial; 5) in adult humans with or without cardiovascular disease (but not acutely ill, pregnant or breastfeeding); 6) intervention duration at least 24 months; 7) mortality or cardiovascular morbidity data available. DATA COLLECTION AND ANALYSIS: Two review authors independently assessed inclusion, extracted study data and assessed risk of bias. We performed random-effects meta-analyses, meta-regression, subgrouping, sensitivity analyses, funnel plots and GRADE assessment. MAIN RESULTS: We included 15 randomised controlled trials (RCTs) (16 comparisons, 56,675 participants), that used a variety of interventions from providing all food to advice on reducing saturated fat. The included long-term trials suggested that reducing dietary saturated fat reduced the risk of combined cardiovascular events by 17% (risk ratio (RR) 0.83; 95% confidence interval (CI) 0.70 to 0.98, 12 trials, 53,758 participants of whom 8% had a cardiovascular event, I² = 67%, GRADE moderate-quality evidence). Meta-regression suggested that greater reductions in saturated fat (reflected in greater reductions in serum cholesterol) resulted in greater reductions in risk of CVD events, explaining most heterogeneity between trials. The number needed to treat for an additional beneficial outcome (NNTB) was 56 in primary prevention trials, so 56 people need to reduce their saturated fat intake for ~four years for one person to avoid experiencing a CVD event. In secondary prevention trials, the NNTB was 53. Subgrouping did not suggest significant differences between replacement of saturated fat calories with polyunsaturated fat or carbohydrate, and data on replacement with monounsaturated fat and protein was very limited. We found little or no effect of reducing saturated fat on all-cause mortality (RR 0.96; 95% CI 0.90 to 1.03; 11 trials, 55,858 participants) or cardiovascular mortality (RR 0.95; 95% CI 0.80 to 1.12, 10 trials, 53,421 participants), both with GRADE moderate-quality evidence. There was little or no effect of reducing saturated fats on non-fatal myocardial infarction (RR 0.97, 95% CI 0.87 to 1.07) or CHD mortality (RR 0.97, 95% CI 0.82 to 1.16, both low-quality evidence), but effects on total (fatal or non-fatal) myocardial infarction, stroke and CHD events (fatal or non-fatal) were all unclear as the evidence was of very low quality. There was little or no effect on cancer mortality, cancer diagnoses, diabetes diagnosis, HDL cholesterol, serum triglycerides or blood pressure, and small reductions in weight, serum total cholesterol, LDL cholesterol and BMI. There was no evidence of harmful effects of reducing saturated fat intakes. AUTHORS' CONCLUSIONS: The findings of this updated review suggest that reducing saturated fat intake for at least two years causes a potentially important reduction in combined cardiovascular events. Replacing the energy from saturated fat with polyunsaturated fat or carbohydrate appear to be useful strategies, while effects of replacement with monounsaturated fat are unclear. The reduction in combined cardiovascular events resulting from reducing saturated fat did not alter by study duration, sex or baseline level of cardiovascular risk, but greater reduction in saturated fat caused greater reductions in cardiovascular events.

Across time and place, right hand preference has been the norm, but what is the precise prevalence of left- and right-handedness? Frequency of left-handedness has shaped and underpinned different fields of research, from cognitive neuroscience to human evolution, but reliable distributional estimates are still lacking. While hundreds of empirical studies have assessed handedness, a large-scale, comprehensive review of the prevalence of handedness and the factors that moderate it, is currently missing. Here, we report 5 meta-analyses on hand preference for different manual tasks and show that left-handedness prevalence lies between 9.3% (using the most stringent criterion of left-handedness) to 18.1% (using the most lenient criterion of nonright-handedness), with the best overall estimate being 10.6% (10.4% when excluding studies assessing elite athletes' handedness). Handedness variability depends on (a) study characteristics, namely year of publication and ways to measure and classify handedness, and (b) participant characteristics, namely sex and ancestry. Our analysis identifies the role of moderators that require taking into account in future studies on handedness and hemispheric asymmetries. We argue that the same evolutionary mechanisms should apply across geographical regions to maintain the roughly 1:10 ratio, while cultural factors, such as pressure against left-hand use, moderate the magnitude of the prevalence of left-handedness. Although handedness appears as a straightforward trait, there is no universal agreement on how to assess it. Therefore, we urge researchers to fully report study and participant characteristics as well as the detailed procedure by which handedness was assessed and make raw data publicly available. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Human longevity is heritable, but genome-wide association (GWA) studies have had limited success. Here, we perform two meta-analyses of GWA studies of a rigorous longevity phenotype definition including 11,262/3484 cases surviving at or beyond the age corresponding to the 90th/99th survival percentile, respectively, and 25,483 controls whose age at death or at last contact was at or below the age corresponding to the 60th survival percentile. Consistent with previous reports, rs429358 (apolipoprotein E (ApoE) ε4) is associated with lower odds of surviving to the 90th and 99th percentile age, while rs7412 (ApoE ε2) shows the opposite. Moreover, rs7676745, located near GPR78, associates with lower odds of surviving to the 90th percentile age. Gene-level association analysis reveals a role for tissue-specific expression of multiple genes in longevity. Finally, genetic correlation of the longevity GWA results with that of several disease-related phenotypes points to a shared genetic architecture between health and longevity.

Dental caries and periodontitis account for a vast burden of morbidity and healthcare spending, yet their genetic basis remains largely uncharacterized. Here, we identify self-reported dental disease proxies which have similar underlying genetic contributions to clinical disease measures and then combine these in a genome-wide association study meta-analysis, identifying 47 novel and conditionally-independent risk loci for dental caries. We show that the heritability of dental caries is enriched for conserved genomic regions and partially overlapping with a range of complex traits including smoking, education, personality traits and metabolic measures. Using cardio-metabolic traits as an example in Mendelian randomization analysis, we estimate causal relationships and provide evidence suggesting that the processes contributing to dental caries may have undesirable downstream effects on health.

BACKGROUND: Bacterial infections are a major cause of morbidity and mortality in patients who are neutropenic following chemotherapy for malignancy. Trials have shown the efficacy of antibiotic prophylaxis in reducing the incidence of bacterial infections but not in reducing mortality rates. Our systematic review from 2006 also showed a reduction in mortality. OBJECTIVES: This updated review aimed to evaluate whether there is still a benefit of reduction in mortality when compared to placebo or no intervention. SEARCH METHODS: We searched the Cochrane Cancer Network Register of Trials (2011), Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library Issue 2, 2011), MEDLINE (1966 to March 2011), EMBASE (1980 to March 2011), abstracts of conference proceedings and the references of identified studies. SELECTION CRITERIA: Randomised controlled trials (RCTs) or quasi-RCTs comparing different types of antibiotic prophylaxis with placebo or no intervention, or another antibiotic, to prevent bacterial infections in afebrile neutropenic patients. DATA COLLECTION AND ANALYSIS: Two authors independently appraised the quality of each trial and extracted data from the included trials. Analyses were performed using RevMan 5.1 software. MAIN RESULTS: One-hundred and nine trials (involving 13,579 patients) that were conducted between the years 1973 to 2010 met the inclusion criteria. When compared with placebo or no intervention, antibiotic prophylaxis significantly reduced the risk of death from all causes (46 trials, 5635 participants; risk ratio (RR) 0.66, 95% CI 0.55 to 0.79) and the risk of infection-related death (43 trials, 5777 participants; RR 0.61, 95% CI 0.48 to 0.77). The estimated number needed to treat (NNT) to prevent one death was 34 (all-cause mortality) and 48 (infection-related mortality).Prophylaxis also significantly reduced the occurrence of fever (54 trials, 6658 participants; RR 0.80, 95% CI 0.74 to 0.87), clinically documented infection (48 trials, 5758 participants; RR 0.65, 95% CI 0.56 to 0.76), microbiologically documented infection (53 trials, 6383 participants; RR 0.51, 95% CI 0.42 to 0.62) and other indicators of infection.There were no significant differences between quinolone prophylaxis and TMP-SMZ prophylaxis with regard to death from all causes or infection, however, quinolone prophylaxis was associated with fewer side effects leading to discontinuation (seven trials, 850 participants; RR 0.37, 95% CI 0.16 to 0.87) and less resistance to the drugs thereafter (six trials, 366 participants; RR 0.45, 95% CI 0.27 to 0.74). AUTHORS' CONCLUSIONS: Antibiotic prophylaxis in afebrile neutropenic patients significantly reduced all-cause mortality. In our review, the most significant reduction in mortality was observed in trials assessing prophylaxis with quinolones. The benefits of antibiotic prophylaxis outweighed the harm such as adverse effects and the development of resistance since all-cause mortality was reduced. As most trials in our review were of patients with haematologic cancer, we strongly recommend antibiotic prophylaxis for these patients, preferably with a quinolone. Prophylaxis may also be considered for patients with solid tumours or lymphoma.

BACKGROUND: Reduction and modification of dietary fats have differing effects on cardiovascular risk factors (such as serum cholesterol), but their effects on important health outcomes are less clear. OBJECTIVES: To assess the effect of reduction and/or modification of dietary fats on mortality, cardiovascular mortality, cardiovascular morbidity and individual outcomes including myocardial infarction, stroke and cancer diagnoses in randomised clinical trials of at least 6 months duration. SEARCH METHODS: For this review update, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and EMBASE, were searched through to June 2010. References of Included studies and reviews were also checked. SELECTION CRITERIA: Trials fulfilled the following criteria: 1) randomised with appropriate control group, 2) intention to reduce or modify fat or cholesterol intake (excluding exclusively omega-3 fat interventions), 3) not multi factorial, 4) adult humans with or without cardiovascular disease, 5) intervention at least six months, 6) mortality or cardiovascular morbidity data available. DATA COLLECTION AND ANALYSIS: Participant numbers experiencing health outcomes in each arm were extracted independently in duplicate and random effects meta-analyses, meta-regression, sub-grouping, sensitivity analyses and funnel plots were performed. MAIN RESULTS: This updated review suggested that reducing saturated fat by reducing and/or modifying dietary fat reduced the risk of cardiovascular events by 14% (RR 0.86, 95% CI 0.77 to 0.96, 24 comparisons, 65,508 participants of whom 7% had a cardiovascular event, I(2) 50%). Subgrouping suggested that this reduction in cardiovascular events was seen in studies of fat modification (not reduction - which related directly to the degree of effect on serum total and LDL cholesterol and triglycerides), of at least two years duration and in studies of men (not of women). There were no clear effects of dietary fat changes on total mortality (RR 0.98, 95% CI 0.93 to 1.04, 71,790 participants) or cardiovascular mortality (RR 0.94, 95% CI 0.85 to 1.04, 65,978 participants). This did not alter with sub-grouping or sensitivity analysis.Few studies compared reduced with modified fat diets, so direct comparison was not possible. AUTHORS' CONCLUSIONS: The findings are suggestive of a small but potentially important reduction in cardiovascular risk on modification of dietary fat, but not reduction of total fat, in longer trials. Lifestyle advice to all those at risk of cardiovascular disease and to lower risk population groups, should continue to include permanent reduction of dietary saturated fat and partial replacement by unsaturates. The ideal type of unsaturated fat is unclear.

BACKGROUND: The gut microbiome is a complex and metabolically active community that directly influences host phenotypes. In this study, we profile gut microbiota using 16S rRNA gene sequencing in 531 well-phenotyped Finnish men from the Metabolic Syndrome In Men (METSIM) study. RESULTS: We investigate gut microbiota relationships with a variety of factors that have an impact on the development of metabolic and cardiovascular traits. We identify novel associations between gut microbiota and fasting serum levels of a number of metabolites, including fatty acids, amino acids, lipids, and glucose. In particular, we detect associations with fasting plasma trimethylamine N-oxide (TMAO) levels, a gut microbiota-dependent metabolite associated with coronary artery disease and stroke. We further investigate the gut microbiota composition and microbiota-metabolite relationships in subjects with different body mass index and individuals with normal or altered oral glucose tolerance. Finally, we perform microbiota co-occurrence network analysis, which shows that certain metabolites strongly correlate with microbial community structure and that some of these correlations are specific for the pre-diabetic state. CONCLUSIONS: Our study identifies novel relationships between the composition of the gut microbiota and circulating metabolites and provides a resource for future studies to understand host-gut microbiota relationships.

ARIES is a population-based resource of DNA methylation data accessible by the research community, established with the aim of improving our understanding of the determinants and consequences of epigenetic variation, in particular the role of DNA methylation in health and development. To this end, ARIES has profiled genome-wide DNA methylation on samples selected from approximately 1000 mother-child pairs at three time points in children and two time points in their mothers. ARIES has also profiled a wide range of fetal and adult tissues for reference purposes, with which to compare cord blood and peripheral blood DNA methylation patterns. A web portal has been constructed to allow openly accessible browsing of aggregate ARIES DNA methylation data in the context of other publicly available data on many aspects of gene function, using a custom-built data integration system (ARIES-Explorer). ARIES is a sub-study of the Avon Longitudinal Study of Parents and Children (ALSPAC).1,2 A sub-set of 1018 mother-offspring pairs were included in ARIES and were selected based on availability of DNA samples at two time- for the mother (at an antenatal clinic and at a follow-up clinic when their offspring were mean age 15.5 years) and three time points for the offspring [at birth, childhood, (mean age 7.5 years) and adolescence at mean age 15.5 years]. Figure 1 provides a schematic of the ALSPAC participants and time points profiled as part of ARIES. Schematic of the ALSPAC participants included in ARIES. A total of 1018 mother-child pairs have been profiled (v.1 data release 2014) at each of the five time points indicated, using the Illumina Infinium HumanMethylation450K BeadChip array. Ten of the same mother-child pairs have been profiled using whole-genome bisulphite sequencing at each of the five time points shown. GA, gestational age. ARIES children provided a DNA sample extracted from cord blood drawn from the umbilical cord upon delivery. A second DNA sample was extracted from peripheral blood drawn at the 7-year clinic visit, and a third at a clinic held either at age 15 or at age 17 years. Maternal samples were provided during the antenatal period [mean 25.7 (SD 9.5) weeks of gestation] and approximately 15–17 years later when mothers accompanied their child to a clinic visit or when they attended a clinic themselves. DNA methylation patterns differ between tissues. To generate a reference resource to facilitate comparison of peripheral blood DNA methylation with that of other tissues, we undertook profiling of DNA methylation from a range of other sources, where possible with paired peripheral blood DNA. Tissues were provided from three sources (as described below) for genome-wide DNA methylation profiling in ARIES using the Illumina Infinium HumanMethylation450 BeadChip (450 K) array. The tissues represented are summarized in Table 1 . Tissues profiled as part of ARIES in addition to cord blood and peripheral blood DNA samples from ALSPAC DNAm, DNA methylation data available; RNA, gene expression data available. Tissues profiled as part of ARIES in addition to cord blood and peripheral blood DNA samples from ALSPAC DNAm, DNA methylation data available; RNA, gene expression data available. Adult autopsy tissues. Twelve tissue types were harvested from 16 autopsy cases (10 male, 6 female) and provided for DNA methylation profiling and genotyping (through collaboration with Leiden University Medical Center). Tissue included peripheral blood (collected from the thoracic cavity) to allow cross-reference between blood and other tissue profiles from the same individuals. The samples were collected within 12 h of the death of the subjects with a mean age of 62.8 years. Further details can be found elsewhere.3 All samples are anonymized and procedures were performed according to the ethical guidelines in the Code for Proper Secondary Use of Human Tissue in The Netherlands (Dutch Federation of Medical Scientific Societies). Post-mortem brain tissues. Post-mortem brain tissue was provided by the Newcastle Brain Tissue Resource. A total of 20 samples underwent DNA methylation profiling from each of 10 male and 10 female donors. Multiple regions were sampled from post-mortem brain tissue; frontal cortex, temporal cortex, hippocampus, substantia nigra, dorsal raphe nucleus, putamen, hypothalamus, amygdala, cerebellum and peripheral blood DNA was also provided from the same individuals. Fetal tissues. Fetal tissues were provided by the MRC Human Developmental Biology Resource at Newcastle University from terminations of pregnancy. Twenty different tissues (listed in Table 1) were harvested from two male and two female fetuses of gestational age 12–14 weeks. No associated information is available for the fetal tissues (e.g. maternal age or reason for termination). Samples from 10 mother-child pairs selected from the ALSPAC cohort, that underwent Illumina 450 K array analysis at each of the available time points, were also whole genome-bisulphite sequenced (BS-Seq) at the same time points (Figure 1). Samples were selected based on the completeness of the Illumina 450 K data and availability of additional DNA at the same time points. All tissues undergoing DNA methylation profiling, with the exception of peripheral blood (as no RNA was available in this instance), were considered for gene expression analysis. Samples were limited by the availability of good quality RNA. Those tissue types where successful gene expression data were generated are summarized in Table 1. Approximately 1000 ALSPAC lymphoblastoid cell lines derived from blood samples drawn from participants at age 9 years had previously undergone gene expression profiling. There is an overlap of 319 ALSPAC participants for whom both DNA methylation and gene expression data are available. ARIES-Explorer [www.ariesepigenomics.org.uk] provides a web interface to browse methylation variation between groups of individuals and across time. Supporting this web interface, a highly powerful platform for graph-based data mining and data integration, termed Entanglement, was constructed to integrate ARIES DNA methylation data and other information to allow ‘on-the-fly’ querying of relationships with publicly available epigenomics data sets. This system is highly scalable, allowing large data sources to be integrated through use of disposable virtual machines and no-SQL databases [http://intbio.ncl.ac.uk/?projects=entanglement]. The extensive body of genetic and life-course data available on the ALSPAC study are detailed in the cohort profiles1,2 and can be browsed via the study website [http://www.bristol.ac.uk/alspac/] which contains details of all the data that are available through a fully searchable data dictionary [http://www.bris.ac.uk/alspac/researchers/data-access/data-dictionary/].The potential exists to link ARIES DNA methylation data to any of this information, which covers a plethora of measures spanning social, behavioural, psychiatric, environmental, genetic, metabolomic and many other domains. Ethical approvals are in place for all sources of biological samples and data in ARIES and include appropriate permissions from the Human Development Biology Resource, Newcastle Brain Tissue Resource and Leiden University Medical Center. Ethical approval for the ALSPAC study was obtained from the ALSPAC Ethics and Law Committee and the local research ethics committees. ALSPAC is a long-term health research project that started in the early 1990 s. It recruited more than 14 000 pregnant women in the former county of Avon, who were due to give birth between April 1991 and December 1992. The study currently holds longitudinal data and biosamples for these women, their offspring and their partners. Table 2 compares the characteristics of mothers in ARIES with those in the rest of ALSPAC. Mothers included in ARIES were slightly older, more likely to have a non-manual occupation and less likely to have smoked throughout pregnancy. The ARIES sub-sample is considered to be reasonably representative of the main study population. Comparison of a selection of maternal characteristics in ALSPAC mothers included and not included in ARIES * For mothers or offspring as appropriate. Comparison of a selection of maternal characteristics in ALSPAC mothers included and not included in ARIES * For mothers or offspring as appropriate. The first public release of ARIES data became available in June 2014. Subsequent annual updates will incorporate all new data generated and processed by that date and will be annotated with version numbers, with all previous versions stored and available on request. Notification of version updates will be posted on the study website. Additional data generation is under way, with further work planned. This includes Illumina 450 K array analysis of more ALSPAC participants including fathers (n = 312 ARIES fathers who have been profiled to date), samples from ARIES mothers attending further follow-up clinics, additional time points for ARIES children with further DNA samples available during infancy, children of ALSPAC participants, various case series and specific exposure sub-groups. The ARIES-Explorer web portal has the facility to import the browser’s own Illumina 450 K array data as a track for visualization and comparison alongside ARIES data in the genome browser. The functionality of ARIES-Explorer is also under development and additional features will become available in due course. Plans are under way to add data from several ethnic groups (South Asian, South African, Brazilian). DNA methylation was quantified using the Illumina Infinium® HumanMethylation450K BeadChip assay (Illumina 450 K array) (Illumina Inc., CA). Cord and peripheral blood samples (whole blood or buffy coat) were collected according to standard procedures. Following extraction, DNA was bisulfite-converted using the Zymo EZ DNA MethylationTM kit (Zymo, Irvine, CA) then genome-wide methylation status of over 485 000 CpG sites was measured using the Illumina 450 K array according to the standard protocol. The arrays were scanned using an Illumina iScan and initial quality review was assessed using GenomeStudio (version 2011.1). Samples from all participant ages in ARIES were distributed across slides using a semi-random approach (sampling criteria were in place to ensure that all time points were represented on each array) to minimize the possibility of confounding by batch effects. In addition, during the data generation process a wide range of batch variables were recorded in a purpose-built laboratory information management system (LIMS). The LIMS also reported quality control (QC) metrics from the standard control probes on the Illumina 450 K array for each sample. Samples failing QC (average probe P-value ≥0.01) were excluded from further analysis and scheduled for repeat assay. As an additional QC step, genotype probes were compared with SNP-chip data from the same individual to identify and remove any sample mismatches. For individuals with no genome-wide SNP data, samples were flagged if there was a sex-mismatch based on X-chromosome methylation. Tissue samples from autopsy specimens, post-mortem brain and fetal samples were processed in exactly the same way as blood samples described above. The Illumina 450 K BeadChip assay detects the proportion of molecules methylated at each CpG site on the array. For each sample, the estimated methylation level at each CpG site was expressed as a beta value (β), which is the ratio of the methylated probe intensity and the overall intensity and ranges from 0 (no cytosine methylation) to 1 (complete cytosine methylation). Methylation data were pre-processed using R (version 3.0.1), with background correction and subset quantile normalization performed within each time point using the pipeline described by Touleimat and Tost.4 Data are available both pre- and post-normalization. It is widely recognized that white blood cell heterogeneity can confound DNA methylation measurement.5 The overwhelming majority of ALSPAC biosamples did not have whole blood cell counts taken prior to DNA extraction and therefore the only viable option is to apply a post hoc correction for cellular heterogeneity to the DNA methylation data generated. In order to ensure that the results of any downstream analysis are not influenced by variation in cell type fraction between samples, we estimated the fraction of CD8 T cells, CD4 T cells, NK cells, B cells, monocytes and granulocytes in the samples using the estimateCellCounts function in the minfi Bioconductor package implemented in R.6 Both cell-type corrected and uncorrected data are available to facilitate use of either data set in a main analysis or the corrected data set as a sensitivity analysis. For BS-seq library preparation, DNA samples were fragmented by sonication (Covaris) and adaptor ligated (using Illumina-supplied methylated adaptors and NEBnext library preparation kit). DNA was then bisulphite-treated using the Sigma Imprint kit, according to the manufacturer’s instructions (one-step protocol). Final library amplification (16 cycles) was done using Pfu Turbo Cx (Agilent), after which the libraries were gel-purified using QIAGEN Minelute kit. Libraries were sequenced on either an Illumina GAIIx or an Illumina HiSeq using the default RTA analysis software.7 Pre-existing LCL gene expression data were generated using the Illumina HT-12 V3 BeadChip (Illumina Inc., San Diego, CA) which included 48 804 probes.8 Additional gene expression data on selected tissue samples were generated by ServiceXS (Leiden, The Netherlands) using the same array. Participants of ALSPAC, including those in ARIES, were previously genotyped. Children were genotyped using the Illumina HumanHap550 quad genome-wide SNP genotyping platform (Illumina Inc., San Diego, CA) by the Wellcome Trust Sanger Institute (WTSI, Cambridge, UK) and the Laboratory Corporation of America (LCA, Burlington, NC, USA) with support from 23andMe. Mothers were genotyped using the Illumina Human660W-quad genome-wide SNP genotyping platform (Illumina Inc., San Diego, CA) at the Centre National de Génotypage (CNG, Paris, France). Imputation has been performed to increase SNP density for all mothers and children. The final imputed data set contains >8 million SNPs with imputation quality >0.8 and minor allele frequency (MAF) > 1%. The ARIES sample also overlaps with participants included in a large-scale whole-genome sequencing initiative to identify low frequency and rare variants, the UK10K project [http://www.uk10k.org/]. Genotyping of autopsy cases was undertaken using the Illumina HumanCoreExome-24 BeadChip SNP array (Illumina Inc., San Diego, CA), at the University of Bristol, following the manufacturer’s protocol. Longitudinal studies provide a powerful opportunity to investigate the role of epigenetic variation in both normal development and the evolution of adverse traits and diseases.9 To date the use of ARIES data has been motivated by six main areas of activity, illustrative examples of which are provided below. There are currently in excess of 70 projects approved to access and use DNA methylation data in ALSPAC, so it is anticipated that wider usage of ARIES will prompt many further applications of the data in due course. Epigenome-wide association studies (EWAS) have been undertaken to uncover associations between DNA methylation and a range of exposures or outcomes. These have included analysis of the influence of: maternal prenatal smoking on offspring DNA methylation10 (Figure 2A) and the mediation of lower infant birthweight in smokers;11 maternal pre-pregnancy BMI on offspring DNA methylation and subsequent childhood adiposity;12 and maternal vitamin B12 status on offspring DNA methylation and subsequent childhood IQ. In an effort to provide a rigorous evidence base for future epigenetic studies, a Pregnancy and Childhood Epigenetics Consortium has been established (PACE), to facilitate meta-analyses across several exposures and outcomes pertinent to this sector of the population. EWAS of many other exposures and phenotypes have been conducted (e.g. alcohol intake, trace elements, heavy metals, stress, autism spectrum disorders, children born following assisted reproductive technologies, etc.). Replication studies. Frequent requests are made by other investigators to access ARIES data relating to or a that have been as associated with a specific or exposure in an is an in a study and ARIES has for with several other cohort studies. to access and use ARIES data for can be made at any time with no to ALSPAC association studies have been undertaken to identify genetic variation associated with DNA methylation A large-scale has been a genetic influence of genotype on DNA methylation across the course. are currently for and for all associations in excess of 000 associations at each time in the of a methylation will be available via the study website A of DNA Methylation Consortium has been established to facilitate of studies. Development and of for DNA methylation data, including longitudinal of DNA methylation over time (Figure and the of to in epigenetic (Figure have been a of ARIES The of epigenetic studies to confounding and have been previously by as has the of using a of is the of which is to be in studies ARIES data (Figure The availability of and to be for the of on DNA as as of within A further of development using ARIES data is the of in DNA methylation patterns over time. to date have evidence of of in DNA methylation at birth, with methylation for than (Figure of of in genetic association studies has been undertaken using ARIES Following the of a it is to the SNP of in the ARIES to the SNP of is likely to be influence via of DNA is the SNP also a an approach was using ARIES to the of associated with of DNA methylation patterns in fetal tissues, brain autopsy tissues and peripheral blood are in including a analysis of methylated sites and Use of ARIES association study that maternal prenatal smoking is associated with DNA methylation in DNA of Longitudinal of methylated over time to the of DNA methylation from prenatal exposure to The of to epigenetic is the analysis of the of maternal vitamin DNA methylation and at age years. as a for maternal vitamin B12 and as a for DNA methylation. ARIES is a longitudinal population-based resource which includes and and which is within the ALSPAC, and so provides an opportunity to link DNA methylation data to a array of exposure and variables as as in these over time. It is that DNA methylation data are in of the of genetic and this is possible the genome-wide SNP data also available in ARIES. The majority of ARIES children are also part of the UK10K sequencing initiative which has generated whole-genome The study has provided the first which will in of of as as for the of genetic for DNA methylation in studies. ALSPAC and ARIES place on data To this end, ARIES data are available and a web portal has been to data browsing and integration with other data data are also available through standard ALSPAC data access which are to all of ARIES include on DNA extracted from peripheral This upon the that can be made DNA methylation patterns. The of a reference data series of tissue types this to is an include on the Illumina Infinium HumanMethylation450K BeadChip which has limited of all CpG sites in the and recognized can be have for and of DNA as or can be made to access and use biological samples from the ALSPAC to additional epigenetic data can be browsed via the ARIES-Explorer web portal Additional data are available by following the ALSPAC project approval data are to browse the of approved projects on the ALSPAC website using to of work and potential in their ALSPAC a of and therefore of effort can and is the of the to the time of approximately of ALSPAC data requests include DNA methylation. The ALSPAC access and data dictionary can be via the following Illumina 450 K array data are provided as for each time point probe methylated probe and P-value data are available as proportion methylated and for each reported in the Illumina 450 K array data can be R and with a wide range of for analysis of 450 K data, including and for and for association analysis many other have established a to and from the analysis of epigenetic data, including Illumina 450 K array data, at All studies to use ARIES data to other ALSPAC data are to give due to both studies. ARIES in a ARIES was set to support population-based epigenetic studies. It is the only study of to have large DNA methylation data on the same individuals and on mother-child in ARIES includes mother-child pairs from the Avon Longitudinal Study of Parents and Children recruited during Participants have undergone genome-wide DNA methylation analysis at birth and childhood and adolescence and In addition a range of reference tissues from samples, post-mortem brain tissue and autopsy samples have been Data can be to a wide range of environmental, social, genetic, and other data collected as part of ALSPAC. data are openly available to browse via the study web portal and individual level data are available upon via ALSPAC ARIES was by the and support for ALSPAC is provided by the Medical and the Wellcome Trust and the University of to generate DNA methylation data which is will included in ARIES has been obtained from the and other ARIES is under the of the MRC at the University of and of the autopsy series was by the under are to all the who part in ALSPAC, the for their in and the whole ALSPAC which includes and laboratory research and those who have made a to ARIES, including and the Newcastle Brain Tissue Resource, the Human Developmental Biology Resource and the ARIES Scientific and of

Pre-pregnancy maternal obesity is associated with adverse offspring outcomes at birth and later in life. Individual studies have shown that epigenetic modifications such as DNA methylation could contribute. Within the Pregnancy and Childhood Epigenetics (PACE) Consortium, we meta-analysed the association between pre-pregnancy maternal BMI and methylation at over 450,000 sites in newborn blood DNA, across 19 cohorts (9,340 mother-newborn pairs). We attempted to infer causality by comparing the effects of maternal versus paternal BMI and incorporating genetic variation. In four additional cohorts (1,817 mother-child pairs), we meta-analysed the association between maternal BMI at the start of pregnancy and blood methylation in adolescents. In newborns, maternal BMI was associated with small (<0.2% per BMI unit (1 kg/m2), P < 1.06 × 10-7) methylation variation at 9,044 sites throughout the genome. Adjustment for estimated cell proportions greatly attenuated the number of significant CpGs to 104, including 86 sites common to the unadjusted model. At 72/86 sites, the direction of the association was the same in newborns and adolescents, suggesting persistence of signals. However, we found evidence for acausal intrauterine effect of maternal BMI on newborn methylation at just 8/86 sites. In conclusion, this well-powered analysis identified robust associations between maternal adiposity and variations in newborn blood DNA methylation, but these small effects may be better explained by genetic or lifestyle factors than a causal intrauterine mechanism. This highlights the need for large-scale collaborative approaches and the application of causal inference techniques in epigenetic epidemiology.

We conducted a systematic review and meta-analysis to investigate the associations between menarcheal age and all-cause and cardiovascular death. Medline, Embase, Scopus, and Web of Knowledge were searched for articles published prior to March 2013 reporting on the associations between menarcheal age and death from all causes or from cardiovascular disease (total cardiovascular disease, ischemic heart disease (IHD), and stroke) in adult women. Nine articles were eligible for inclusion; these reported 5 estimates each for death from all causes and total cardiovascular death, 6 estimates for IHD, and 7 estimates for death from stroke. Our meta-analysis showed that each 1-year increase in age at menarche was associated with a 3% lower relative risk of death from all causes (pooled hazard ratio = 0.97, 95% confidence interval: 0.96, 0.98) with low heterogeneity (I(2) = 32.2%). Meta-analysis of 2 cohorts showed a higher risk of death from all causes for women who experienced early menarche (at <12 years of age) versus "not early" menarche (at ≥ 12 years of age) (pooled hazard ratio = 1.23, 95% confidence interval: 1.10, 1.38; I(2) = 0%). An inverse association between age at menarche and death from IHD was observed only among nonsmoking populations or populations with low prevalence of smoking. We found no evidence of association between age at menarche and death from all cardiovascular diseases or stroke. Early menarche was consistently associated with higher risk of death from all causes. Further studies are needed to clarify the role of menarcheal age on cardiovascular outcomes and to investigate the potential modifying role of smoking.

DNA methylation-based biomarkers of aging are highly correlated with actual age. Departures of methylation-estimated age from actual age can be used to define epigenetic measures of child development or age acceleration (AA) in adults. Very little is known about genetic or environmental determinants of these epigenetic measures of aging. We obtained DNA methylation profiles using Infinium HumanMethylation450 BeadChips across five time-points in 1018 mother-child pairs from the Avon Longitudinal Study of Parents and Children. Using the Horvath age estimation method, we calculated epigenetic age for these samples. AA was defined as the residuals from regressing epigenetic age on actual age. AA was tested for associations with cross-sectional clinical variables in children. We identified associations between AA and sex, birth weight, birth by caesarean section and several maternal characteristics in pregnancy, namely smoking, weight, BMI, selenium and cholesterol level. Offspring of non-drinkers had higher AA on average but this difference appeared to resolve during childhood. The associations between sex, birth weight and AA found in ARIES were replicated in an independent cohort (GOYA). In children, epigenetic AA measures are associated with several clinically relevant variables, and early life exposures appear to be associated with changes in AA during adolescence. Further research into epigenetic aging, including the use of causal inference methods, is required to better our understanding of aging.

IMPORTANCE: Schizophrenia is a highly heritable, polygenic condition characterized by a relatively diverse phenotype and frequent comorbid conditions, such as anxiety and depression. At present, limited evidence explains how genetic risk for schizophrenia is manifest in the general population. OBJECTIVE: To investigate the extent to which genetic risk for schizophrenia is associated with different phenotypes during adolescence in a population-based birth cohort. DESIGN, SETTING, AND PARTICIPANTS: This cohort study used data from the Avon Longitudinal Study of Parents and Children (ALSPAC). Of 14,062 children in the birth cohort, genetic data were available for 9912 adolescents. Data were collected periodically from September 6, 1990, and collection is ongoing. Data were analyzed from March 4 to August 13, 2015. EXPOSURES: Polygenic risk scores (PRSs) for schizophrenia generated for individuals in the ALSPAC cohort using results of the second Psychiatric Genomics Consortium Schizophrenia genome-wide association study as a training set. MAIN OUTCOMES AND MEASURES: Logistic regression was used to assess associations between the schizophrenia PRS and (1) psychotic experiences (Psychosis-Like Symptom Interview at 12 and 18 years of age), (2) negative symptoms (Community Assessment of Psychic Experiences at 16.5 years of age), (3) depressive disorder (Development and Well-Being Assessment at 15.5 years of age), and (4) anxiety disorder (Development and Well-Being Assessment at 15.5 years of age) in adolescence. RESULTS: Of the 8230 ALSPAC participants whose genetic data passed quality control checks (51.2% male, 48.8% female), 3676 to 5444 participated in assessments from 12 to 18 years of age. The PRSs created using single-nucleotide polymorphisms with a training-set P ≤ .05 threshold were associated with negative symptoms (odds ratio [OR] per SD increase in PRS, 1.21; 95% CI, 1.08-1.36; R(2) = 0.007) and anxiety disorder (OR per SD increase in PRS, 1.17; 95% CI, 1.06- 1.29; R(2) = 0.005). No evidence was found of an association between schizophrenia PRS and psychotic experiences (OR per SD increase in PRS, 1.08; 95% CI, 0.98-1.19; R(2) = 0.001) or depressive disorder (OR per SD increase in PRS, 1.02; 95% CI, 0.91-1.13; R(2) = 0.00005). Results were mostly consistent across different training-set P value thresholds and using different cutoffs and measures of the psychopathological outcomes. CONCLUSIONS AND RELEVANCE: This study demonstrates polygenic overlaps between common genetic polymorphisms associated with schizophrenia and negative symptoms and anxiety disorder but not with psychotic experiences or depression. Because the genetic risk for schizophrenia appears to be manifest as anxiety and negative symptoms during adolescence, a greater focus on these phenotypes rather than on psychotic experiences might be required for prediction of transition in at-risk samples.

Motivation: Epidemiological cohorts typically contain a diverse set of phenotypes such that automation of phenome scans is non-trivial, because they require highly heterogeneous models. For this reason, phenome scans have to date tended to use a smaller homogeneous set of phenotypes that can be analysed in a consistent fashion. We present PHESANT (PHEnome Scan ANalysis Tool), a software package for performing comprehensive phenome scans in UK Biobank. General features: PHESANT tests the association of a specified trait with all continuous, integer and categorical variables in UK Biobank, or a specified subset. PHESANT uses a novel rule-based algorithm to determine how to appropriately test each trait, then performs the analyses and produces plots and summary tables. Implementation: The PHESANT phenome scan is implemented in R. PHESANT includes a novel Javascript D3.js visualization and accompanying Java code that converts the phenome scan results to the required JavaScript Object Notation (JSON) format. Availability: PHESANT is available on GitHub at [https://github.com/MRCIEU/PHESANT]. Git tag v0.5 corresponds to the version presented here.

Few genome-wide association studies (GWAS) account for environmental exposures, like smoking, potentially impacting the overall trait variance when investigating the genetic contribution to obesity-related traits. Here, we use GWAS data from 51,080 current smokers and 190,178 nonsmokers (87% European descent) to identify loci influencing BMI and central adiposity, measured as waist circumference and waist-to-hip ratio both adjusted for BMI. We identify 23 novel genetic loci, and 9 loci with convincing evidence of gene-smoking interaction (GxSMK) on obesity-related traits. We show consistent direction of effect for all identified loci and significance for 18 novel and for 5 interaction loci in an independent study sample. These loci highlight novel biological functions, including response to oxidative stress, addictive behaviour, and regulatory functions emphasizing the importance of accounting for environment in genetic analyses. Our results suggest that tobacco smoking may alter the genetic susceptibility to overall adiposity and body fat distribution.

IMPORTANCE: The association of copy number variations (CNVs), differing numbers of copies of genetic sequence at locations in the genome, with phenotypes such as intellectual disability has been almost exclusively evaluated using clinically ascertained cohorts. The contribution of these genetic variants to cognitive phenotypes in the general population remains unclear. OBJECTIVE: To investigate the clinical features conferred by CNVs associated with known syndromes in adult carriers without clinical preselection and to assess the genome-wide consequences of rare CNVs (frequency ≤0.05%; size ≥250 kilobase pairs [kb]) on carriers' educational attainment and intellectual disability prevalence in the general population. DESIGN, SETTING, AND PARTICIPANTS: The population biobank of Estonia contains 52,000 participants enrolled from 2002 through 2010. General practitioners examined participants and filled out a questionnaire of health- and lifestyle-related questions, as well as reported diagnoses. Copy number variant analysis was conducted on a random sample of 7877 individuals and genotype-phenotype associations with education and disease traits were evaluated. Our results were replicated on a high-functioning group of 993 Estonians and 3 geographically distinct populations in the United Kingdom, the United States, and Italy. MAIN OUTCOMES AND MEASURES: Phenotypes of genomic disorders in the general population, prevalence of autosomal CNVs, and association of these variants with educational attainment (from less than primary school through scientific degree) and prevalence of intellectual disability. RESULTS: Of the 7877 in the Estonian cohort, we identified 56 carriers of CNVs associated with known syndromes. Their phenotypes, including cognitive and psychiatric problems, epilepsy, neuropathies, obesity, and congenital malformations are similar to those described for carriers of identical rearrangements ascertained in clinical cohorts. A genome-wide evaluation of rare autosomal CNVs (frequency, ≤0.05%; ≥250 kb) identified 831 carriers (10.5%) of the screened general population. Eleven of 216 (5.1%) carriers of a deletion of at least 250 kb (odds ratio [OR], 3.16; 95% CI, 1.51-5.98; P = 1.5e-03) and 6 of 102 (5.9%) carriers of a duplication of at least 1 Mb (OR, 3.67; 95% CI, 1.29-8.54; P = .008) had an intellectual disability compared with 114 of 6819 (1.7%) in the Estonian cohort. The mean education attainment was 3.81 (P = 1.06e-04) among 248 (≥250 kb) deletion carriers and 3.69 (P = 5.024e-05) among 115 duplication carriers (≥1 Mb). Of the deletion carriers, 33.5% did not graduate from high school (OR, 1.48; 95% CI, 1.12-1.95; P = .005) and 39.1% of duplication carriers did not graduate high school (OR, 1.89; 95% CI, 1.27-2.8; P = 1.6e-03). Evidence for an association between rare CNVs and lower educational attainment was supported by analyses of cohorts of adults from Italy and the United States and adolescents from the United Kingdom. CONCLUSIONS AND RELEVANCE: Known pathogenic CNVs in unselected, but assumed to be healthy, adult populations may be associated with unrecognized clinical sequelae. Additionally, individually rare but collectively common intermediate-size CNVs may be negatively associated with educational attainment. Replication of these findings in additional population groups is warranted given the potential implications of this observation for genomics research, clinical care, and public health.

OBJECTIVE: The relation between childhood overweight and adult non-alcoholic fatty liver disease (NAFLD) is largely unknown. We investigated if weight and weight gain in childhood increases the risk of being diagnosed with NAFLD in routine clinical settings in adulthood. PARTICIPANTS: We studied 244,464 boys and girls, born between 1930 and 1989, who attended school in Copenhagen, Denmark. Their heights and weights were measured by physicians or nurses at mandatory school health examinations at ages 7-13 years. Body mass index (BMI) z-scores were calculated from an internal age-specific and sex-specific reference. OUTCOME MEASURES: NAFLD reported in the National Patient Register and the National Register of Pathology at 18 years of age or older. HRs with 95% CIs were estimated. RESULTS: During follow-up, 1264 and 1106 NAFLD cases, respectively, occurred in men and women. In both sexes, childhood BMI z-score was not consistently associated with adult NAFLD. Change in BMI z-score between 7 and 13 years of age was positively associated with NAFLD in both sexes. When adjusted for BMI z-score at age 7 years, the HRs of adult NAFLD were 1.15 (95% CI 1.05 to 1.26) and 1.12 (95% CI 1.02 to 1.23) per 1-unit gain in BMI z-score in men and women, respectively. Associations were similar when adjusted for BMI z-score at age 13 years, and were consistent across birth years. CONCLUSIONS: A BMI gain in school-aged children is associated with adult NAFLD. Intriguingly, BMI gain appears to have an effect on adult NAFLD irrespective of either the initial or the attained BMI. Taken together, our results suggest that BMI gain in childhood, rather than the level of BMI per se, is important in the development of adult NAFLD.