DNA methylation-based predictors of health: applications and statistical considerations.

DNA methylation data have become a valuable source of information for biomarker development, because, unlike static genetic risk estimates, DNA methylation varies dynamically in relation to diverse exogenous and endogenous factors, including environmental risk factors and complex disease pathology. Reliable methods for genome-wide measurement at scale have led to the proliferation of epigenome-wide association studies and subsequently to the development of DNA methylation-based predictors across a wide range of health-related applications, from the identification of risk factors or exposures, such as age and smoking, to early detection of disease or progression in cancer, cardiovascular and neurological disease. This Review evaluates the progress of existing DNA methylation-based predictors, including the contribution of machine learning techniques, and assesses the uptake of key statistical best practices needed to ensure their reliable performance, such as data-driven feature selection, elimination of data leakage in performance estimates and use of generalizable, adequately powered training samples.

An epigenome-wide view of osteoarthritis in primary tissues.

Osteoarthritis is a complex degenerative joint disease. Here, we investigate matched genotype and methylation profiles of primary chondrocytes from macroscopically intact (low-grade) and degraded (high-grade) osteoarthritis cartilage and from synoviocytes collected from 98 osteoarthritis-affected individuals undergoing knee replacement surgery. We perform an epigenome-wide association study of knee cartilage degeneration and report robustly replicating methylation markers, which reveal an etiologic mechanism linked to the migration of epithelial cells. Using machine learning, we derive methylation models of cartilage degeneration, which we validate with 82% accuracy in independent data. We report a genome-wide methylation quantitative trait locus (mQTL) map of articular cartilage and synovium and identify 18 disease-grade-specific mQTLs in osteoarthritis cartilage. We resolve osteoarthritis GWAS loci through causal inference and colocalization analyses and decipher the epigenetic mechanisms that mediate the effect of genotype on disease risk. Together, our findings provide enhanced insights into epigenetic mechanisms underlying osteoarthritis in primary tissues.

Epigenetic Regulation of F2RL3 Associates With Myocardial Infarction and Platelet Function.

BACKGROUND: DNA hypomethylation at the F2RL3 (F2R like thrombin or trypsin receptor 3) locus has been associated with both smoking and atherosclerotic cardiovascular disease; whether these smoking-related associations form a pathway to disease is unknown. F2RL3 encodes protease-activated receptor 4, a potent thrombin receptor expressed on platelets. Given the role of thrombin in platelet activation and the role of thrombus formation in myocardial infarction, alterations to this biological pathway could be important for ischemic cardiovascular disease. METHODS: We conducted multiple independent experiments to assess whether DNA hypomethylation at F2RL3 in response to smoking is associated with risk of myocardial infarction via changes to platelet reactivity. Using cohort data (N=3205), we explored the relationship between smoking, DNA hypomethylation at F2RL3, and myocardial infarction. We compared platelet reactivity in individuals with low versus high DNA methylation at F2RL3 (N=41). We used an in vitro model to explore the biological response of F2RL3 to cigarette smoke extract. Finally, a series of reporter constructs were used to investigate how differential methylation could impact F2RL3 gene expression. RESULTS: Observationally, DNA methylation at F2RL3 mediated an estimated 34% of the smoking effect on increased risk of myocardial infarction. An association between methylation group (low/high) and platelet reactivity was observed in response to PAR4 (protease-activated receptor 4) stimulation. In cells, cigarette smoke extract exposure was associated with a 4.9% to 9.3% reduction in DNA methylation at F2RL3 and a corresponding 1.7-(95% CI, 1.2-2.4, P=0.04) fold increase in F2RL3 mRNA. Results from reporter assays suggest the exon 2 region of F2RL3 may help control gene expression. CONCLUSIONS: Smoking-induced epigenetic DNA hypomethylation at F2RL3 appears to increase PAR4 expression with potential downstream consequences for platelet reactivity. Combined evidence here not only identifies F2RL3 DNA methylation as a possible contributory pathway from smoking to cardiovascular disease risk but from any feature potentially influencing F2RL3 regulation in a similar manner.

Statistical and integrative system-level analysis of DNA methylation data.

Epigenetics plays a key role in cellular development and function. Alterations to the epigenome are thought to capture and mediate the effects of genetic and environmental risk factors on complex disease. Currently, DNA methylation is the only epigenetic mark that can be measured reliably and genome-wide in large numbers of samples. This Review discusses some of the key statistical challenges and algorithms associated with drawing inferences from DNA methylation data, including cell-type heterogeneity, feature selection, reverse causation and system-level analyses that require integration with other data types such as gene expression, genotype, transcription factor binding and other epigenetic information.

Ultra-processed foods, adiposity and risk of head and neck cancer and oesophageal adenocarcinoma in the European Prospective Investigation into Cancer and Nutrition study: a mediation analysis.

PURPOSE: To investigate the role of adiposity in the associations between ultra-processed food (UPF) consumption and head and neck cancer (HNC) and oesophageal adenocarcinoma (OAC) in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. METHODS: Our study included 450,111 EPIC participants. We used Cox regressions to investigate the associations between the consumption of UPFs and HNC and OAC risk. A mediation analysis was performed to assess the role of body mass index (BMI) and waist-to-hip ratio (WHR) in these associations. In sensitivity analyses, we investigated accidental death as a negative control outcome. RESULTS: During a mean follow-up of 14.13 ± 3.98 years, 910 and 215 participants developed HNC and OAC, respectively. A 10% g/d higher consumption of UPFs was associated with an increased risk of HNC (hazard ratio [HR] = 1.23, 95% confidence interval [CI] 1.14-1.34) and OAC (HR = 1.24, 95% CI 1.05-1.47). WHR mediated 5% (95% CI 3-10%) of the association between the consumption of UPFs and HNC risk, while BMI and WHR, respectively, mediated 13% (95% CI 6-53%) and 15% (95% CI 8-72%) of the association between the consumption of UPFs and OAC risk. UPF consumption was positively associated with accidental death in the negative control analysis. CONCLUSIONS: We reaffirmed that higher UPF consumption is associated with greater risk of HNC and OAC in EPIC. The proportion mediated via adiposity was small. Further research is required to investigate other mechanisms that may be at play (if there is indeed any causal effect of UPF consumption on these cancers).

Investigating causality in the association between DNA methylation and type 2 diabetes using bidirectional two-sample Mendelian randomisation.

AIMS/HYPOTHESIS: Several studies have identified associations between type 2 diabetes and DNA methylation (DNAm). However, the causal role of these associations remains unclear. This study aimed to provide evidence for a causal relationship between DNAm and type 2 diabetes. METHODS: We used bidirectional two-sample Mendelian randomisation (2SMR) to evaluate causality at 58 CpG sites previously detected in a meta-analysis of epigenome-wide association studies (meta-EWAS) of prevalent type 2 diabetes in European populations. We retrieved genetic proxies for type 2 diabetes and DNAm from the largest genome-wide association study (GWAS) available. We also used data from the Avon Longitudinal Study of Parents and Children (ALSPAC, UK) when associations of interest were not available in the larger datasets. We identified 62 independent SNPs as proxies for type 2 diabetes, and 39 methylation quantitative trait loci as proxies for 30 of the 58 type 2 diabetes-related CpGs. We applied the Bonferroni correction for multiple testing and inferred causality based on p<0.001 for the type 2 diabetes to DNAm direction and p<0.002 for the opposing DNAm to type 2 diabetes direction in the 2SMR analysis. RESULTS: We found strong evidence of a causal effect of DNAm at cg25536676 (DHCR24) on type 2 diabetes. An increase in transformed residuals of DNAm at this site was associated with a 43% (OR 1.43, 95% CI 1.15, 1.78, p=0.001) higher risk of type 2 diabetes. We inferred a likely causal direction for the remaining CpG sites assessed. In silico analyses showed that the CpGs analysed were enriched for expression quantitative trait methylation sites (eQTMs) and for specific traits, dependent on the direction of causality predicted by the 2SMR analysis. CONCLUSIONS/INTERPRETATION: We identified one CpG mapping to a gene related to the metabolism of lipids (DHCR24) as a novel causal biomarker for risk of type 2 diabetes. CpGs within the same gene region have previously been associated with type 2 diabetes-related traits in observational studies (BMI, waist circumference, HDL-cholesterol, insulin) and in Mendelian randomisation analyses (LDL-cholesterol). Thus, we hypothesise that our candidate CpG in DHCR24 may be a causal mediator of the association between known modifiable risk factors and type 2 diabetes. Formal causal mediation analysis should be implemented to further validate this assumption.

Use of Correct and Incorrect Methods of Accounting for Age in Studies of Epigenetic Accelerated Aging: Implications and Recommendations for Best Practices.

Motivated by our conduct of a literature review on social exposures and accelerated aging as measured by a growing number of epigenetic "clocks" (which estimate age via DNA methylation (DNAm) patterns), we report on 3 different approaches in the epidemiologic literature-1 incorrect and 2 correct-on the treatment of age in these and other studies using other common exposures (i.e., body mass index and alcohol consumption). Among the 50 empirical articles reviewed, the majority (n = 29; 58%) used the incorrect method of analyzing accelerated aging detrended for age as the outcome and did not control for age as a covariate. By contrast, only 42% used correct methods, which are either to analyze accelerated aging detrended for age as the outcome and control for age as a covariate (n = 16; 32%) or to analyze raw DNAm age as the outcome and control for age as a covariate (n = 5; 10%). In accord with prior demonstrations of bias introduced by use of the incorrect approach, we provide simulation analyses and additional empirical analyses to illustrate how the incorrect method can lead to bias towards the null, and we discuss implications for extant research and recommendations for best practices.

Multi-omics analyses of cognitive traits and psychiatric disorders highlights brain-dependent mechanisms.

Integrating findings from genome-wide association studies with molecular datasets can develop insight into the underlying functional mechanisms responsible for trait-associated genetic variants. We have applied the principles of Mendelian randomization (MR) to investigate whether brain-derived gene expression (n = 1194) may be responsible for mediating the effect of genetic variants on eight cognitive and psychological outcomes (attention deficit hyperactivity disorder (ADHD), Alzheimer's disease, bipolar disorder, depression, intelligence, insomnia, neuroticism and schizophrenia). Transcriptome-wide analyses identified 83 genes associated with at least one outcome (PBonferroni < 6.72 × 10-6), with multiple-trait colocalization also implicating changes to brain-derived DNA methylation at nine of these loci. Comparing effects between outcomes identified evidence of enrichment which may reflect putative causal relationships, such as an inverse relationship between genetic liability towards schizophrenia risk and cognitive ability in later life. Repeating these analyses in whole blood (n = 31 684), we replicated 58.2% of brain-derived effects (based on P < 0.05). Finally, we undertook phenome-wide evaluations at associated loci to investigate pleiotropic effects with 700 complex traits. This highlighted pleiotropic loci such as FURIN (initially implicated in schizophrenia risk (P = 1.05 × 10-7)) which had evidence of an effect on 28 other outcomes, as well as genes which may have a more specific role in disease pathogenesis (e.g. SLC12A5 which only provided evidence of an effect on depression (P = 7.13 × 10-10)). Our results support the utility of whole blood as a valuable proxy for informing initial target identification but also suggest that gene discovery in a tissue-specific manner may be more informative. Finally, non-pleiotropic loci highlighted by our study may be of use for therapeutic translational endeavours.

Epigenome-wide change and variation in DNA methylation in childhood: trajectories from birth to late adolescence.

DNA methylation (DNAm) is known to play a pivotal role in childhood health and development, but a comprehensive characterization of genome-wide DNAm trajectories across this age period is currently lacking. We have therefore performed a series of epigenome-wide association studies in 5019 blood samples collected at multiple time-points from birth to late adolescence from 2348 participants of two large independent cohorts. DNAm profiles of autosomal CpG sites (CpGs) were generated using the Illumina Infinium HumanMethylation450 BeadChip. Change over time was widespread, observed at over one-half (53%) of CpGs. In most cases, DNAm was decreasing (36% of CpGs). Inter-individual variation in linear trajectories was similarly widespread (27% of CpGs). Evidence for non-linear change and inter-individual variation in non-linear trajectories was somewhat less common (11 and 8% of CpGs, respectively). Very little inter-individual variation in change was explained by sex differences (0.4% of CpGs) even though sex-specific DNAm was observed at 5% of CpGs. DNAm trajectories were distributed non-randomly across the genome. For example, CpGs with decreasing DNAm were enriched in gene bodies and enhancers and were annotated to genes enriched in immune-developmental functions. In contrast, CpGs with increasing DNAm were enriched in promoter regions and annotated to genes enriched in neurodevelopmental functions. These findings depict a methylome undergoing widespread and often non-linear change throughout childhood. They support a developmental role for DNA methylation that extends beyond birth into late adolescence and has implications for understanding life-long health and disease. DNAm trajectories can be visualized at http://epidelta.mrcieu.ac.uk.

Characterizing the Causal Pathway for Genetic Variants Associated with Neurological Phenotypes Using Human Brain-Derived Proteome Data.

Leveraging high-dimensional molecular datasets can help us develop mechanistic insight into associations between genetic variants and complex traits. In this study, we integrated human proteome data derived from brain tissue to evaluate whether targeted proteins putatively mediate the effects of genetic variants on seven neurological phenotypes (Alzheimer disease, amyotrophic lateral sclerosis, depression, insomnia, intelligence, neuroticism, and schizophrenia). Applying the principles of Mendelian randomization (MR) systematically across the genome highlighted 43 effects between genetically predicted proteins derived from the dorsolateral prefrontal cortex and these outcomes. Furthermore, genetic colocalization provided evidence that the same causal variant at 12 of these loci was responsible for variation in both protein and neurological phenotype. This included genes such as DCC, which encodes the netrin-1 receptor and has an important role in the development of the nervous system (p = 4.29 × 10-11 with neuroticism), as well as SARM1, which has been previously implicated in axonal degeneration (p = 1.76 × 10-08 with amyotrophic lateral sclerosis). We additionally conducted a phenome-wide MR study for each of these 12 genes to assess potential pleiotropic effects on 700 complex traits and diseases. Our findings suggest that genes such as SNX32, which was initially associated with increased risk of Alzheimer disease, may potentially influence other complex traits in the opposite direction. In contrast, genes such as CTSH (which was also associated with Alzheimer disease) and SARM1 may make worthwhile therapeutic targets because they did not have genetically predicted effects on any of the other phenotypes after correcting for multiple testing.

Smoking, DNA Methylation, and Lung Function: a Mendelian Randomization Analysis to Investigate Causal Pathways.

Whether smoking-associated DNA methylation has a causal effect on lung function has not been thoroughly evaluated. We first investigated the causal effects of 474 smoking-associated CpGs on forced expiratory volume in 1 s (FEV1) in UK Biobank (n = 321,047) by using two-sample Mendelian randomization (MR) and then replicated this investigation in the SpiroMeta Consortium (n = 79,055). Second, we used two-step MR to investigate whether DNA methylation mediates the effect of smoking on FEV1. Lastly, we evaluated the presence of horizontal pleiotropy and assessed whether there is any evidence for shared causal genetic variants between lung function, DNA methylation, and gene expression by using a multiple-trait colocalization ("moloc") framework. We found evidence of a possible causal effect for DNA methylation on FEV1 at 18 CpGs (p < 1.2 × 10-4). Replication analysis supported a causal effect at three CpGs (cg21201401 [LIME1 and ZGPAT], cg19758448 [PGAP3], and cg12616487 [EML3 and AHNAK] [p < 0.0028]). DNA methylation did not clearly mediate the effect of smoking on FEV1, although DNA methylation at some sites might influence lung function via effects on smoking. By using "moloc", we found evidence of shared causal variants between lung function, gene expression, and DNA methylation. These findings highlight potential therapeutic targets for improving lung function and possibly smoking cessation, although larger, tissue-specific datasets are required to confirm these results.

Cord blood epigenome-wide meta-analysis in six European-based child cohorts identifies signatures linked to rapid weight growth.

BACKGROUND: Rapid postnatal growth may result from exposure in utero or early life to adverse conditions and has been associated with diseases later in life and, in particular, with childhood obesity. DNA methylation, interfacing early-life exposures and subsequent diseases, is a possible mechanism underlying early-life programming. METHODS: Here, a meta-analysis of Illumina HumanMethylation 450K/EPIC-array associations of cord blood DNA methylation at single CpG sites and CpG genomic regions with rapid weight growth at 1 year of age (defined with reference to WHO growth charts) was conducted in six European-based child cohorts (ALSPAC, ENVIRONAGE, Generation XXI, INMA, Piccolipiù, and RHEA, N = 2003). The association of gestational age acceleration (calculated using the Bohlin epigenetic clock) with rapid weight growth was also explored via meta-analysis. Follow-up analyses of identified DNA methylation signals included prediction of rapid weight growth, mediation of the effect of conventional risk factors on rapid weight growth, integration with transcriptomics and metabolomics, association with overweight in childhood (between 4 and 8 years), and comparison with previous findings. RESULTS: Forty-seven CpGs were associated with rapid weight growth at suggestive p-value <1e-05 and, among them, three CpGs (cg14459032, cg25953130 annotated to ARID5B, and cg00049440 annotated to KLF9) passed the genome-wide significance level (p-value <1.25e-07). Sixteen differentially methylated regions (DMRs) were identified as associated with rapid weight growth at false discovery rate (FDR)-adjusted/Siddak p-values < 0.01. Gestational age acceleration was associated with decreasing risk of rapid weight growth (p-value = 9.75e-04). Identified DNA methylation signals slightly increased the prediction of rapid weight growth in addition to conventional risk factors. Among the identified signals, three CpGs partially mediated the effect of gestational age on rapid weight growth. Both CpGs (N=3) and DMRs (N=3) were associated with differential expression of transcripts (N=10 and 7, respectively), including long non-coding RNAs. An AURKC DMR was associated with childhood overweight. We observed enrichment of CpGs previously reported associated with birthweight. CONCLUSIONS: Our findings provide evidence of the association between cord blood DNA methylation and rapid weight growth and suggest links with prenatal exposures and association with childhood obesity providing opportunities for early prevention.

Meta-analysis of epigenome-wide associations between DNA methylation at birth and childhood cognitive skills.

Cognitive skills are a strong predictor of a wide range of later life outcomes. Genetic and epigenetic associations across the genome explain some of the variation in general cognitive abilities in the general population and it is plausible that epigenetic associations might arise from prenatal environmental exposures and/or genetic variation early in life. We investigated the association between cord blood DNA methylation at birth and cognitive skills assessed in children from eight pregnancy cohorts within the Pregnancy And Childhood Epigenetics (PACE) Consortium across overall (total N = 2196), verbal (total N = 2206) and non-verbal cognitive scores (total N = 3300). The associations at single CpG sites were weak for all of the cognitive domains investigated. One region near DUSP22 on chromosome 6 was associated with non-verbal cognition in a model adjusted for maternal IQ. We conclude that there is little evidence to support the idea that variation in cord blood DNA methylation at single CpG sites is associated with cognitive skills and further studies are needed to confirm the association at DUSP22.

Chromosome 18 Loss of Heterozygosity in Small Intestinal Neuroendocrine Tumours: Multi-Omic and Tumour Composition Analyses.

INTRODUCTION: Small intestinal neuroendocrine tumours (siNETs) are rare neoplasms which present with low mutational burden and can be subtyped based on copy number variation (CNV). Currently, siNETs can be molecularly classified as having chromosome 18 loss of heterozygosity (18LOH), multiple CNVs (MultiCNV), or no CNVs. 18LOH tumours have better progression-free survival when compared to MultiCNV and NoCNV tumours, however, the mechanism underlying this is unknown, and clinical practice does not currently consider CNV status. METHODS: Here, we use genome-wide tumour DNA methylation (n = 54) and gene expression (n = 20 matched to DNA methylation) to better understand how gene regulation varies by 18LOH status. We then use multiple cell deconvolution methods to analyse how cell composition varies between 18LOH status and determine potential associations with progression-free survival. RESULTS: We identified 27,464 differentially methylated CpG sites and 12 differentially expressed genes between 18LOH and non-18LOH (MultiCNV + NoCNV) siNETs. Although few differentially expressed genes were identified, these genes were highly enriched with the differentially methylated CpG sites compared to the rest of the genome. We identified differences in tumour microenvironment between 18LOH and non-18LOH tumours, including CD14+ infiltration in a subset of non-18LOH tumours which had the poorest clinical outcomes. CONCLUSIONS: We identify a small number of genes which appear to be linked to the 18LOH status of siNETs, and find evidence of potential epigenetic dysregulation of these genes. We also find a potential prognostic marker for worse progression-free outcomes in the form of higher CD14 infiltration in non-18LOH siNETs.

Harnessing Whole Genome Polygenic Risk Scores to Stratify Individuals Based on Cardiometabolic Risk Factors and Biomarkers at Age 10 in the Lifecourse-Brief Report.

BACKGROUND: In this study, we investigated the capability of polygenic risk scores to stratify a cohort of young individuals into risk deciles based on 10 different cardiovascular traits and circulating biomarkers. METHODS: We first conducted large-scale genome-wide association studies using data on adults (mean age 56.5 years) enrolled in the UK Biobank study (n=393 193 to n=461 460). Traits and biomarkers analyzed were body mass index, systolic blood pressure, diastolic blood pressure, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, apolipoprotein B, apolipoprotein A-I, C-reactive protein and vitamin D. Findings were then leveraged to build whole genome polygenic risk scores in participants from the Avon Longitudinal Study of Parents and Children (mean age, 9.9 years) which were used to stratify this cohort into deciles in turn and analyzed against their respective traits. RESULTS: For each of the 10 different traits assessed, we found strong evidence of an incremental trend across deciles (all P<0.0001). Large differences were identified when comparing top and bottom deciles; for example, using the apolipoprotein B polygenic risk scores there was a mean difference of 13.2 mg/dL for this established risk factor of coronary heart disease in later life. CONCLUSIONS: Although the use of polygenic prediction in a clinical setting may currently be premature, our findings suggest they are becoming increasingly powerful as a means of predicting complex trait variation at an early stage in the lifecourse.

Bayesian network analysis incorporating genetic anchors complements conventional Mendelian randomization approaches for exploratory analysis of causal relationships in complex data.

Mendelian randomization (MR) implemented through instrumental variables analysis is an increasingly popular causal inference tool used in genetic epidemiology. But it can have limitations for evaluating simultaneous causal relationships in complex data sets that include, for example, multiple genetic predictors and multiple potential risk factors associated with the same genetic variant. Here we use real and simulated data to investigate Bayesian network analysis (BN) with the incorporation of directed arcs, representing genetic anchors, as an alternative approach. A Bayesian network describes the conditional dependencies/independencies of variables using a graphical model (a directed acyclic graph) with an accompanying joint probability. In real data, we found BN could be used to infer simultaneous causal relationships that confirmed the individual causal relationships suggested by bi-directional MR, while allowing for the existence of potential horizontal pleiotropy (that would violate MR assumptions). In simulated data, BN with two directional anchors (mimicking genetic instruments) had greater power for a fixed type 1 error than bi-directional MR, while BN with a single directional anchor performed better than or as well as bi-directional MR. Both BN and MR could be adversely affected by violations of their underlying assumptions (such as genetic confounding due to unmeasured horizontal pleiotropy). BN with no directional anchor generated inference that was no better than by chance, emphasizing the importance of directional anchors in BN (as in MR). Under highly pleiotropic simulated scenarios, BN outperformed both MR (and its recent extensions) and two recently-proposed alternative approaches: a multi-SNP mediation intersection-union test (SMUT) and a latent causal variable (LCV) test. We conclude that BN incorporating genetic anchors is a useful complementary method to conventional MR for exploring causal relationships in complex data sets such as those generated from modern "omics" technologies.

Sex steroid hormones and risk of breast cancer: a two-sample Mendelian randomization study.

BACKGROUND: Breast cancer (BC) has the highest cancer incidence and mortality in women worldwide. Observational epidemiological studies suggest a positive association between testosterone, estradiol, dehydroepiandrosterone sulphate (DHEAS) and other sex steroid hormones with postmenopausal BC. We used a two-sample Mendelian randomization analysis to investigate this association. METHODS: Genetic instruments for nine sex steroid hormones and sex hormone-binding globulin (SHBG) were obtained from genome-wide association studies (GWAS) of UK Biobank (total testosterone (TT) N: 230,454, bioavailable testosterone (BT) N: 188,507 and SHBG N: 189,473), The United Kingdom Household Longitudinal Study (DHEAS N: 9722), the LIFE-Adult and LIFE-Heart cohorts (estradiol N: 2607, androstenedione N: 711, aldosterone N: 685, progesterone N: 1259 and 17-hydroxyprogesterone N: 711) and the CORtisol NETwork (CORNET) consortium (cortisol N: 25,314). Outcome GWAS summary statistics were obtained from the Breast Cancer Association Consortium (BCAC) for overall BC risk (N: 122,977 cases and 105,974 controls) and subtype-specific analyses. RESULTS: We found that a standard deviation (SD) increase in TT, BT and estradiol increased the risk of overall BC (OR 1.14, 95% CI 1.09-1.21, OR 1.19, 95% CI 1.07-1.33 and OR 1.03, 95% CI 1.01-1.06, respectively) and ER + BC (OR 1.19, 95% CI 1.12-1.27, OR 1.25, 95% CI 1.11-1.40 and OR 1.06, 95% CI 1.03-1.09, respectively). An SD increase in DHEAS also increased ER + BC risk (OR 1.09, 95% CI 1.03-1.16). Subtype-specific analyses showed similar associations with ER+ expressing subtypes: luminal A-like BC, luminal B-like BC and luminal B/HER2-negative-like BC. CONCLUSIONS: TT, BT, DHEAS and estradiol increase the risk of ER+ type BCs similar to observational studies. Understanding the role of sex steroid hormones in BC risk, particularly subtype-specific risks, highlights the potential importance of attempts to modify and/or monitor hormone levels in order to prevent BC.

Impact of sex hormone-binding globulin on the human phenome.

BACKGROUND: Sex hormone-binding globulin (SHBG) is a circulating glycoprotein and a regulator of sex hormone levels, which has been shown to influence various traits and diseases. The molecular nature of SHBG makes it a feasible target for preventative or therapeutic interventions. A systematic study of its effects across the human phenome may uncover novel associations. METHODS: We used a Mendelian randomization phenome-wide association study (MR-pheWAS) approach to systematically appraise the potential functions of SHBG while reducing potential biases such as confounding and reverse causation common to the literature. We searched for potential causal effects of SHBG in UK Biobank (N = 334 977) and followed-up our top findings using two-sample MR analyses to evaluate whether estimates may be biased due to horizontal pleiotropy. RESULTS: Results of the MR-pheWAS across over 21 000 outcome phenotypes identified 12 phenotypes associated with genetically elevated SHBG after Bonferroni correction for multiple testing. Follow-up analysis using two-sample MR indicated the associations of increased natural log SHBG with higher impedance of the arms and whole body, lower pulse rate, lower bone density, higher odds of hip replacement, lower odds of high cholesterol or cholesterol medication use and higher odds of gallbladder removal. CONCLUSIONS: Our systematic MR-pheWAS of SHBG, which was comprehensive to the range of phenotypes available in UK Biobank, suggested that higher circulating SHBG affects the body impedance, bone density and cholesterol levels, among others. These phenotypes should be prioritized in future studies aiming to investigate the biological effects of SHBG or develop targets for therapeutic intervention.

Maternal diet in pregnancy and child's respiratory outcomes: an individual participant data meta-analysis of 18 000 children.

RATIONALE: Severe fetal malnutrition has been related to an increased risk of respiratory diseases later in life, but evidence for the association of a suboptimal diet during pregnancy with respiratory outcomes in childhood is conflicting. We aimed to examine whether a pro-inflammatory or low-quality maternal diet during pregnancy was associated with child's respiratory health. METHODS: We performed an individual participant meta-analysis among 18 326 mother-child pairs from seven European birth cohorts. Maternal pro-inflammatory and low-quality diets were estimated by energy-adjusted Dietary Inflammatory Index (E-DII) and Dietary Approaches to Stop Hypertension (DASH) scores. Preschool wheezing and school-age asthma were measured using questionnaires and lung function by spirometry. RESULTS: After adjustment for lifestyle and sociodemographic factors, we observed that a higher maternal E-DII score (a more pro-inflammatory diet) during pregnancy was associated only with a lower forced vital capacity (FVC) in children (z-score difference -0.05, 95% CI -0.08- -0.02, per interquartile range increase). No linear associations of the maternal E-DII or DASH score with child's wheezing or asthma were observed. In an exploratory examination of the extremes, a very low DASH score (<10th percentile) (a very low dietary quality) was associated with an increased risk of preschool wheezing and a low forced expiratory volume in 1 s/FVC (z-score <-1.64) (OR 1.20, 95% CI 1.06-1.36 and z-score difference 1.40, 95% CI 1.06-1.85, compared to ≥10th percentile), with corresponding population attributable risk fractions of 1.7% and 3.3%, respectively. CONCLUSION: The main results from this individual participant data meta-analysis do not support the hypothesis that maternal pro-inflammatory or low-quality diet in pregnancy are related to respiratory diseases in childhood.

Applying Mendelian randomization to appraise causality in relationships between nutrition and cancer.

Dietary factors are assumed to play an important role in cancer risk, apparent in consensus recommendations for cancer prevention that promote nutritional changes. However, the evidence in this field has been generated predominantly through observational studies, which may result in biased effect estimates because of confounding, exposure misclassification, and reverse causality. With major geographical differences and rapid changes in cancer incidence over time, it is crucial to establish which of the observational associations reflect causality and to identify novel risk factors as these may be modified to prevent the onset of cancer and reduce its progression. Mendelian randomization (MR) uses the special properties of germline genetic variation to strengthen causal inference regarding potentially modifiable exposures and disease risk. MR can be implemented through instrumental variable (IV) analysis and, when robustly performed, is generally less prone to confounding, reverse causation and measurement error than conventional observational methods and has different sources of bias (discussed in detail below). It is increasingly used to facilitate causal inference in epidemiology and provides an opportunity to explore the effects of nutritional exposures on cancer incidence and progression in a cost-effective and timely manner. Here, we introduce the concept of MR and discuss its current application in understanding the impact of nutritional factors (e.g., any measure of diet and nutritional intake, circulating biomarkers, patterns, preference or behaviour) on cancer aetiology and, thus, opportunities for MR to contribute to the development of nutritional recommendations and policies for cancer prevention. We provide applied examples of MR studies examining the role of nutritional factors in cancer to illustrate how this method can be used to help prioritise or deprioritise the evaluation of specific nutritional factors as intervention targets in randomised controlled trials. We describe possible biases when using MR, and methodological developments aimed at investigating and potentially overcoming these biases when present. Lastly, we consider the use of MR in identifying causally relevant nutritional risk factors for various cancers in different regions across the world, given notable geographical differences in some cancers. We also discuss how MR results could be translated into further research and policy. We conclude that findings from MR studies, which corroborate those from other well-conducted studies with different and orthogonal biases, are poised to substantially improve our understanding of nutritional influences on cancer. For such corroboration, there is a requirement for an interdisciplinary and collaborative approach to investigate risk factors for cancer incidence and progression.