Comparison of Illumina 450K and EPIC arrays in placental DNA methylation

Consortium Profile: The Methylation, Imaging and NeuroDevelopment (MIND) Consortium

Epigenetic processes, such as DNA methylation, show potential as biological markers and mechanisms underlying gene-environment interplay in the prediction of mental health and other brain-based phenotypes. However, little is known about how peripheral epigenetic patterns relate to individual differences in the brain itself. An increasingly popular approach to address this is by combining epigenetic and neuroimaging data; yet, research in this area is almost entirely comprised of cross-sectional studies in adults. To bridge this gap, we established the Methylation, Imaging and NeuroDevelopment (MIND) Consortium, which aims to bring a developmental focus to the emerging field of Neuroimaging Epigenetics by (i) promoting collaborative, adequately powered developmental research via multi-cohort analyses; (ii) increasing scientific rigor through the establishment of shared pipelines and open science practices; and (iii) advancing our understanding of DNA methylation-brain dynamics at different developmental periods (from birth to emerging adulthood), by leveraging data from prospective, longitudinal pediatric studies. MIND currently integrates 15 cohorts worldwide, comprising (repeated) measures of DNA methylation in peripheral tissues (blood, buccal cells, and saliva) and neuroimaging by magnetic resonance imaging across up to five time points over a period of up to 21 years (Npooled DNAm = 11,299; Npooled neuroimaging = 10,133; Npooled combined = 4,914). By triangulating associations across multiple developmental time points and study types, we hope to generate new insights into the dynamic relationships between peripheral DNA methylation and the brain, and how these ultimately relate to neurodevelopmental and psychiatric phenotypes.

Development and Validation of a Novel Placental DNA Methylation Biomarker of Maternal Smoking during Pregnancy in the ECHO Program

BACKGROUND

Maternal cigarette smoking during pregnancy (MSDP) is associated with numerous adverse health outcomes in infants and children with potential lifelong consequences. Negative effects of MSDP on placental DNA methylation (DNAm), placental structure, and function are well established.

OBJECTIVE

Our aim was to develop biomarkers of MSDP using DNAm measured in placentas (N = 96 ), collected as part of the Vitamin C to Decrease the Effects of Smoking in Pregnancy on Infant Lung Function double-blind, placebo-controlled randomized clinical trial conducted between 2012 and 2016. We also aimed to develop a digital polymerase chain reaction (PCR) assay for the top ranking cytosine-guanine dinucleotide (CpG) so that large numbers of samples can be screened for exposure at low cost.

METHODS

We compared the ability of four machine learning methods [logistic least absolute shrinkage and selection operator (LASSO) regression, logistic elastic net regression, random forest, and gradient boosting machine] to classify MSDP based on placental DNAm signatures. We developed separate models using the complete EPIC array dataset and on the subset of probes also found on the 450K array so that models exist for both platforms. For comparison, we developed a model using CpGs previously associated with MSDP in placenta. For each final model, we used model coefficients and normalized beta values to calculate placental smoking index (PSI) scores for each sample. Final models were validated in two external datasets: the Extremely Low Gestational Age Newborn observational study, N = 426 ; and the Rhode Island Children's Health Study, N = 237 .

RESULTS

Logistic LASSO regression demonstrated the highest performance in cross-validation testing with the lowest number of input CpGs. Accuracy was greatest in external datasets when using models developed for the same platform. PSI scores in smokers only (n = 72 ) were moderately correlated with maternal plasma cotinine levels. One CpG (cg27402634), with the largest coefficient in two models, was measured accurately by digital PCR compared with measurement by EPIC array (R 2 = 0.98 ).

DISCUSSION

To our knowledge, we have developed the first placental DNAm-based biomarkers of MSDP with broad utility to studies of prenatal disease origins. https://doi.org/10.1289/EHP13838.

Green space exposure and blood DNA methylation at birth and in childhood - A multi-cohort study

Green space exposure has been associated with improved mental, physical and general health. However, the underlying biological mechanisms remain largely unknown. The aim of this study was to investigate the association between green space exposure and cord and child blood DNA methylation. Data from eight European birth cohorts with a total of 2,988 newborns and 1,849 children were used. Two indicators of residential green space exposure were assessed: (i) surrounding greenness (satellite-based Normalized Difference Vegetation Index (NDVI) in buffers of 100 m and 300 m) and (ii) proximity to green space (having a green space ≥ 5,000 m2 within a distance of 300 m). For these indicators we assessed two exposure windows: (i) pregnancy, and (ii) the period from pregnancy to child blood DNA methylation assessment, named as cumulative exposure. DNA methylation was measured with the Illumina 450K or EPIC arrays. To identify differentially methylated positions (DMPs) we fitted robust linear regression models between pregnancy green space exposure and cord blood DNA methylation and between cumulative green space exposure and child blood DNA methylation. Two sensitivity analyses were conducted: (i) without adjusting for cellular composition, and (ii) adjusting for air pollution. Cohort results were combined through fixed-effect inverse variance weighted meta-analyses. Differentially methylated regions (DMRs) were identified from meta-analysed results using the Enmix-combp and DMRcate methods. There was no statistical evidence of pregnancy or cumulative exposures associating with any DMP (False Discovery Rate, FDR, p-value < 0.05). However, surrounding greenness exposure was inversely associated with four DMRs (three in cord blood and one in child blood) annotated to ADAMTS2, KCNQ1DN, SLC6A12 and SDK1 genes. Results did not change substantially in the sensitivity analyses. Overall, we found little evidence of the association between green space exposure and blood DNA methylation. Although we identified associations between surrounding greenness exposure with four DMRs, these findings require replication.

Head and neck cancer of unknown primary: unveiling primary tumor sites through machine learning on DNA methylation profiles

BACKGROUND

The unknown tissue of origin in head and neck cancer of unknown primary (hnCUP) leads to invasive diagnostic procedures and unspecific and potentially inefficient treatment options for patients. The most common histologic subtype, squamous cell carcinoma, can stem from various tumor primary sites, including the oral cavity, oropharynx, larynx, head and neck skin, lungs, and esophagus. DNA methylation profiles are highly tissue-specific and have been successfully used to classify tissue origin. We therefore developed a support vector machine (SVM) classifier trained with publicly available DNA methylation profiles of commonly cervically metastasizing squamous cell carcinomas (n = 1103) in order to identify the primary tissue of origin of our own cohort of squamous cell hnCUP patient's samples (n = 28). Methylation analysis was performed with Infinium MethylationEPIC v1.0 BeadChip by Illumina.

RESULTS

The SVM algorithm achieved the highest overall accuracy of tested classifiers, with 87%. Squamous cell hnCUP samples on DNA methylation level resembled squamous cell carcinomas commonly metastasizing into cervical lymph nodes. The most frequently predicted cancer localization was the oral cavity in 11 cases (39%), followed by the oropharynx and larynx (both 7, 25%), skin (2, 7%), and esophagus (1, 4%). These frequencies concord with the expected distribution of lymph node metastases in epidemiological studies.

CONCLUSIONS

On DNA methylation level, hnCUP is comparable to primary tumor tissue cancer types that commonly metastasize to cervical lymph nodes. Our SVM-based classifier can accurately predict these cancers' tissues of origin and could significantly reduce the invasiveness of hnCUP diagnostics and enable a more precise therapy after clinical validation.

Comprehensive evaluation of the implementation of episignatures for diagnosis of neurodevelopmental disorders (NDDs)

Episignatures are popular tools for the diagnosis of rare neurodevelopmental disorders. They are commonly based on a set of differentially methylated CpGs used in combination with a support vector machine model. DNA methylation (DNAm) data often include missing values due to changes in data generation technology and batch effects. While many normalization methods exist for DNAm data, their impact on episignature performance have never been assessed. In addition, technologies to quantify DNAm evolve quickly and this may lead to poor transposition of existing episignatures generated on deprecated array versions to new ones. Indeed, probe removal between array versions, technologies or during preprocessing leads to missing values. Thus, the effect of missing data on episignature performance must also be carefully evaluated and addressed through imputation or an innovative approach to episignatures design. In this paper, we used data from patients suffering from Kabuki and Sotos syndrome to evaluate the influence of normalization methods, classification models and missing data on the prediction performances of two existing episignatures. We compare how six popular normalization methods for methylarray data affect episignature classification performances in Kabuki and Sotos syndromes and provide best practice suggestions when building new episignatures. In this setting, we show that Illumina, Noob or Funnorm normalization methods achieved higher classification performances on the testing sets compared to Quantile, Raw and Swan normalization methods. We further show that penalized logistic regression and support vector machines perform best in the classification of Kabuki and Sotos syndrome patients. Then, we describe a new paradigm to build episignatures based on the detection of differentially methylated regions (DMRs) and evaluate their performance compared to classical differentially methylated cytosines (DMCs)-based episignatures in the presence of missing data. We show that the performance of classical DMC-based episignatures suffers from the presence of missing data more than the DMR-based approach. We present a comprehensive evaluation of how the normalization of DNA methylation data affects episignature performance, using three popular classification models. We further evaluate how missing data affect those models' predictions. Finally, we propose a novel methodology to develop episignatures based on differentially methylated regions identification and show how this method slightly outperforms classical episignatures in the presence of missing data.

Sex-based differences in placental DNA methylation profiles related to gestational age: an NIH ECHO meta-analysis

The placenta undergoes many changes throughout gestation to support the evolving needs of the foetus. There is also a growing appreciation that male and female foetuses develop differently in utero, with unique epigenetic changes in placental tissue. Here, we report meta-analysed sex-specific associations between gestational age and placental DNA methylation from four cohorts in the National Institutes of Health (NIH) Environmental influences on Child Health Outcomes (ECHO) Programme (355 females/419 males, gestational ages 23-42 weeks). We identified 407 cytosine-guanine dinucleotides (CpGs) in females and 794 in males where placental methylation levels were associated with gestational age. After cell-type adjustment, 55 CpGs in females and 826 in males were significant. These were enriched for biological processes critical to the immune system in females and transmembrane transport in males. Our findings are distinct between the sexes: in females, associations with gestational age are largely explained by differences in placental cellular composition, whereas in males, gestational age is directly associated with numerous alterations in methylation levels.

Identification of gene-level methylation for disease prediction

DNA methylation is an epigenetic alteration that plays a fundamental part in governing gene regulatory processes. The DNA methylation mechanism affixes methyl groups to distinct cytosine residues, influencing chromatin architectures. Multiple studies have demonstrated that DNA methylation's regulatory effect on genes is linked to the beginning and progression of several disorders. Researchers have recently uncovered thousands of phenotype-related methylation sites through the epigenome-wide association study (EWAS). However, combining the methylation levels of several sites within a gene and determining the gene-level DNA methylation remains challenging. In this study, we proposed the supervised UMAP Assisted Gene-level Methylation method (sUAGM) for disease prediction based on supervised UMAP (Uniform Manifold Approximation and Projection), a manifold learning-based method for reducing dimensionality. The methylation values at the gene level generated using the proposed method are evaluated by employing various feature selection and classification algorithms on three distinct DNA methylation datasets derived from blood samples. The performance has been assessed employing classification accuracy, F-1 score, Mathews Correlation Coefficient (MCC), Kappa, Classification Success Index (CSI) and Jaccard Index. The Support Vector Machine with the linear kernel (SVML) classifier with Recursive Feature Elimination (RFE) performs best across all three datasets. From comparative analysis, our method outperformed existing gene-level and site-level approaches by achieving 100% accuracy and F1-score with fewer genes. The functional analysis of the top 28 genes selected from the Parkinson's disease dataset revealed a significant association with the disease.

Genome-Wide DNA Methylation Profiles in Whole-Blood and Buccal Samples-Cross-Sectional, Longitudinal, and across Platforms

The field of DNA methylation research is rapidly evolving, focusing on disease and phenotype changes over time using methylation measurements from diverse tissue sources and multiple array platforms. Consequently, identifying the extent of longitudinal, inter-tissue, and inter-platform variation in DNA methylation is crucial for future advancement. DNA methylation was measured in 375 individuals, with 197 of those having 2 blood sample measurements ~10 years apart. Whole-blood samples were measured on Illumina Infinium 450K and EPIC methylation arrays, and buccal samples from a subset of 58 participants were measured on EPIC array. The data were analyzed with the aims to examine the correlation between methylation levels in longitudinal blood samples in 197 individuals, examine the correlation between methylation levels in the blood and buccal samples in 58 individuals, and examine the correlation between blood methylation profiles assessed on the EPIC and 450K arrays in 83 individuals. We identified 136,833, 7674, and 96,891 CpGs significantly and strongly correlated (>0.50) longitudinally, across blood and buccal samples as well as array platforms, respectively. A total of 3674 of these CpGs were shared across all three sets. Analysis of these shared CpGs identified previously found associations with aging, ancestry, and 7016 mQTLs as well.

A meta-analysis of pre-pregnancy maternal body mass index and placental DNA methylation identifies 27 CpG sites with implications for mother-child health

Higher maternal pre-pregnancy body mass index (ppBMI) is associated with increased neonatal morbidity, as well as with pregnancy complications and metabolic outcomes in offspring later in life. The placenta is a key organ in fetal development and has been proposed to act as a mediator between the mother and different health outcomes in children. The overall aim of the present work is to investigate the association of ppBMI with epigenome-wide placental DNA methylation (DNAm) in 10 studies from the PACE consortium, amounting to 2631 mother-child pairs. We identify 27 CpG sites at which we observe placental DNAm variations of up to 2.0% per 10 ppBMI-unit. The CpGs that are differentially methylated in placenta do not overlap with CpGs identified in previous studies in cord blood DNAm related to ppBMI. Many of the identified CpGs are located in open sea regions, are often close to obesity-related genes such as GPX1 and LGR4 and altogether, are enriched in cancer and oxidative stress pathways. Our findings suggest that placental DNAm could be one of the mechanisms by which maternal obesity is associated with metabolic health outcomes in newborns and children, although further studies will be needed in order to corroborate these findings.

Low reliability of DNA methylation across Illumina Infinium platforms in cord blood: implications for replication studies and meta-analyses of prenatal exposures

Background

There is an increasing interest in the role of epigenetics in epidemiology, but the emerging research field faces several critical biological and technical challenges. In particular, recent studies have shown poor correlation of measured DNA methylation (DNAm) levels within and across Illumina Infinium platforms in various tissues. In this study, we have investigated concordance between 450 k and EPIC Infinium platforms in cord blood. We could not replicate our previous findings on the association of prenatal paracetamol exposure with cord blood DNAm, which prompted an investigation of cross-platform DNAm differences.

Results

This study is based on two DNAm data sets from cord blood samples selected from the Norwegian Mother, Father and Child Cohort Study (MoBa). DNAm of one data set was measured using the 450 k platform and the other data set was measured using the EPIC platform. Initial analyses of the EPIC data could not replicate any of our previous significant findings in the 450 k data on associations between prenatal paracetamol exposure and cord blood DNAm. A subset of the samples (n = 17) was included in both data sets, which enabled analyses of technical sources potentially contributing to the negative replication. Analyses of these 17 samples with repeated measurements revealed high per-sample correlations (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\stackrel{\mathrm{-}}{\text{R}}\hspace{0.17em}\approx$$\end{document}R-≈ 0.99), but low per-CpG correlations (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\stackrel{\mathrm{-}}{\text{R}}$$\end{document}R- ≈ 0.24) between the platforms. 1.7% of the CpGs exhibited a mean DNAm difference across platforms > 0.1. Furthermore, only 26.7% of the CpGs exhibited a moderate or better cross-platform reliability (intra-class correlation coefficient ≥ 0.5).

Conclusion

The observations of low cross-platform probe correlation and reliability corroborate previous reports in other tissues. Our study cannot determine the origin of the differences between platforms. Nevertheless, it emulates the setting in studies using data from multiple Infinium platforms, often analysed several years apart. Therefore, the findings may have important implications for future epigenome-wide association studies (EWASs), in replication, meta-analyses and longitudinal studies. Cognisance and transparency of the challenges related to cross-platform studies may enhance the interpretation, replicability and validity of EWAS results both in cord blood and other tissues, ultimately improving the clinical relevance of epigenetic epidemiology.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-022-01299-3.

An effective processing pipeline for harmonizing DNA methylation data from Illumina's 450K and EPIC platforms for epidemiological studies

OBJECTIVE

Illumina BeadChip arrays are commonly used to generate DNA methylation data for large epidemiological studies. Updates in technology over time create challenges for data harmonization within and between studies, many of which obtained data from the older 450K and newer EPIC platforms. The pre-processing pipeline for DNA methylation is not trivial, and influences the downstream analyses. Incorporating different platforms adds a new level of technical variability that has not yet been taken into account by recommended pipelines. Our study evaluated the performance of various tools on different versions of platform data harmonization at each step of pre-processing pipeline, including quality control (QC), normalization, batch effect adjustment, and genomic inflation. We illustrate our novel approach using 450K and EPIC data from the Diabetes Autoimmunity Study in the Young (DAISY) prospective cohort.

RESULTS

We found normalization and probe filtering had the biggest effect on data harmonization. Employing a meta-analysis was an effective and easily executable method for accounting for platform variability. Correcting for genomic inflation also helped with harmonization. We present guidelines for studies seeking to harmonize data from the 450K and EPIC platforms, which includes the use of technical replicates for evaluating numerous pre-processing steps, and employing a meta-analysis.

AgeGuess, a Methylomic Prediction Model for Human Ages

Aging was a biological process under regulations from both inherited genetic factors and various molecular modifications within cells during the lifespan. Multiple studies demonstrated that the chronological age may be accurately predicted using the methylomic data. This study proposed a three-step feature selection algorithm AgeGuess for the age regression problem. AgeGuess selected 107 methylomic features as the gender-independent age biomarkers and the Support Vector Regressor (SVR) model using these biomarkers achieved 2.0267 in the mean absolute deviation (MAD) compared with the real chronological ages. Another regression algorithm Ridge achieved a slightly better MAD 1.9859 using the same biomarkers. The gender-independent age prediction models may be further improved by establishing two gender-specific models. And it's interesting to observe that there were only two methylation biomarkers shared by the two gender-specific biomarker sets and these two biomarkers were within the two known age-associated biomarker genes CALB1 and KLF14.

Correlation of Infinium HumanMethylation450K and MethylationEPIC BeadChip arrays in cartilage

DNA methylation of CpG sites is commonly measured using Illumina Infinium BeadChip platforms. The Infinium MethylationEPIC array has replaced the Infinium Methylation450K array. The two arrays use the same technology, with the EPIC array assaying almost double the number of sites than the 450K array. In this study, we compare DNA methylation values of shared CpGs of the same human cartilage samples assayed using both platforms. DNA methylation was measured in 21 human cartilage samples using the both 450K and EPIC arrays. Additional matched 450K and EPIC data in whole tumour and whole blood were downloaded from GEO GSE92580 and GSE86833, respectively. Data were processed using the Bioconductor package Minfi. DNA methylation of six CpG sites was validated for the same 21 cartilage samples by pyrosequencing. In cartilage samples, overall sample correlations of methylation values between arrays were high (Pearson’s r > 0.96). However, 50.5% of CpG sites showed poor correlation (r < 0.2) between arrays. Sites with limited variance and with either very high or very low methylation levels in cartilage exhibited lower correlation values, corroborating prior studies in whole blood. Bisulphite pyrosequencing did not highlight one array as generating more accurate methylation values. For a specific CpG site, the array methylation correlation coefficient differed between cartilage, tumour, and whole blood, reflecting the difference in methylation variance between cell types. Researchers should be cautious when analysing methylation of CpG sites that show low methylation variance within the cell type of interest, regardless of the method used to assay methylation.