BACKGROUND AND AIMS
Epigenetics is defined as the study of mechanisms that control gene expression in a mitotically-heritable manner and are influenced by genetic, environmental, and developmental factors.1 Growing evidence has suggested that the adverse health outcomes reported in IVF-born offspring might have underlying epigenetic mechanisms. Both the features of an infertile couple, as well as the IVF procedure itself, have been shown to alter the epigenetic signature in the offspring and placental tissue.2 As most studies have investigated, DNA methylation changes in cord blood/placental tissue3-5 and a recent study reported that these changes are mitigated by adulthood.6 It is essential to further investigate the potential effects of IVF on the DNA methylation profiles in adolescents using whole blood and to compare the findings to a large representative control group. In this study, the authors had a unique opportunity to investigate the differences in the epigenetic markers between the IVF-conceived adolescents from the Growing Up Healthy Study (GUHS) cohort and their naturally conceived, age-matched counterparts from the Raine study by comparing their DNA methylation signature (epigenomes) using epigenome-wide association studies (EWAS).
MATERIALS AND METHODS
Genomic DNA from whole blood was used to generate the epigenome profiles of the adolescents from the GUHS cohort using the Infinium Methylation Epic Bead Chip (Illumina Inc., San Diego, California, USA) which measures and quantifies approximately 850,000 DNA methylation probes. The epigenomes of the Raine cohort participants were generated from whole blood using the Illumina 450K array (Illumina Inc., San Diego, California, USA). The authors employed three analytical methods: EWAS, gene set enrichment analysis (GSEA), and four measures of epigenetic age. Differential DNA methylation differences between participants in the Raine and GUHS cohorts at age 17 were investigated followed by GSEA. Furthermore, differences in the methylation signatures between IVF and intracytoplasmic sperm injection (ICSI) offspring and frozen versus fresh embryo transfers within the GUHS cohort were investigated. GSEA and comparisons versus chronological age were also explored within the IVF cohort. The authors tested for an association between the cohorts applying Firth’s bias reduced logistic regression against the outcome of IVF versus naturally conceived between the Raine study and GUHS. The effect of IVF on DNA methylation levels of 238 IVF-born adolescents, mean age 16.06±1.67 years (52.94% male), was compared to 1,188 naturally conceived, age-matched controls, mean age 17.25±0.58 years (50.93% male), from the Raine study. Results across all EWAS analyses were investigated to identify enriched biological pathways amongst the most significantly altered probes. Additionally, within the GUHS cohort, the authors investigated 792,104 DNA methylation probes for difference in methylation status applied GSEA to identify enriched pathways and compared four estimates (Horvath, Hannum, Levine, and Skin Horvath) of epigenetic age and their correlation with chronological age.
Between the two cohorts, a total of 401,022 DNA methylation probes overlapped. After adjustment for batch effects, DNA methylation probes as well as technical variation caused by different methylation platforms used between studies, none of the compared probes reached a Bonferroni correction of 1.24E-0.7 (0.05/402,022) required for statistical significance of a positive correlation. Of the analysed DNA methylation probes, 3,850 (0.96%), a small minority, showed nominal significance with a p value <0.05, most likely to be false positives after controlling for cross-study comparisons. Between the cohorts, 1,810 differentially methylated regions were identified; however, none reached statistical significance after correcting for multiple testing. In the comparison between Raine study and GUHS participants no significant enriched gene pathways were identified.
Within the GUHS cohort, when comparing the IVF versus ICSI offspring, and after adjusting for age, sex, maternal smoking, multiple births, batch effect, and cell type, the authors identified 5
CpG probes (cg 15016734; cg 26744878; cg 0331628; cg 20235051; cg20233073) that reached a Bonferroni correction of 6.31E-8. A further 20 probes were identified at a false discovery rate of 5%. Within the IVF cohort, the functional GSEA identified the neuroactive ligand–receptor interaction pathway, which remained significant (p=0.00048) after adjusting for age and sex. In the analysis of epigenetic ageing, the authors found that all four measures were highly correlated with chronological age and did not demonstrate evidence of significant accelerated ageing within the GUHS cohort. The Levine method provided the weakest correlation for accelerated ageing (r2=0.23) and Skin Horvath had the best fit (r2=0.61), followed by Horvath (r2=0.35) and Hannum (r2=0.28).
The authors observed no significant differences in the DNA methylation profiles of adolescents born from IVF when compared to their naturally conceived, age-matched counterparts. A statistically significant difference in the methylation profiles was identified within the IVF cohort when comparing the IVF and ICSI-conceived offspring. Furthermore, an enriched gene pathway among the altered methylation profiles was identified. The IVF cohort showed no evidence of accelerated epigenetic ageing within their whole blood.