PREcisE’s goal is to use longitudinal data from the pre-conceptional period up until adulthood to explore the life-course and molecular pathways pertaining the mother to offspring transmissibility of impaired glycaemic health. The project harnesses very large data from cohorts/biobanks from Spain, Germany, Finland, France and the Netherlands. It also joins hands with related projects (e.g., ALPHABET & NutriPROGRAM) and other consortia (e.g, PACE: Pregnancy And Childhood Epigenetics, EGG: Early Growth Genetics, MethQTL: Global Methylation Collaboration) to study the mother-to-offspring transmission of glycaemic health through epigenetic factors.

PREcisE’s main objectives are:

1.To identify robust epigenetic markers in the offspring demarking an exposure to adverse maternal glucose metabolism during pregnancy.
2.To characterize the underpinning molecular pathways and the tissue specificity of such epigenetic markers through gene expression analysis in human tissues.
3.To explore the role of pre- and postnatal nutrition in modifying impaired maternal glucose programming.
4.To exemplify the lasting impact of adopting healthy dietary behaviours from pregnancy onwards on lifelong health.
5.To use state of the art statistical methodology (Bayesian Structural Equation Models) to examine the interplays between maternal glycaemic health indicators, epigenetics, and offspring impaired glycaemic health.
6.Causal analyses of maternal glucose response related epigenetic marker - outcome association using Mendelian Randomisation (MR).

Key results of the project are:

1.Association of maternal glucose area under the curve with child DNA methylation (DNAm) at two adjacent CpGs in the well-known Thioredoxin Interacting (TXNIP) gene.
2. Maternal T1D associated with childhood DNAm at TXNIP. DNAm at TXNIP was also associated with multiple metabolic phenotypes in childhood, adulthood and with some early growth factors in infancy (BMI at Adiposity(A)Peak, Age at ARebound).
3.The two methylation markers were associated with TXNIP expression in the liver.
4. Harmonisation of the dietary indices; glycaemic index (GI) and glycaemic load (GL) as well as the dietary inflammatory index. In addition, GI has been revised to improve its application as pre- and postnatal exposure variable, in ongoing meta-analysis on DNAm during childhood and adolescence.
5. Maternal GI and GL (mostly in mothers with overweight/obesity), adherence to the Mediterranean diet (excluding alcohol) during pregnancy were associated with cord blood DNAm.
6. Maternal early-pregnancy glucose concentrations, but not insulin concentrations, were associated with DNAm in all weight groups, and maternal plasma fatty acid pattern characterized by higher concentrations of n-3 polyunsaturated fatty acids may be associated with accelerated epigenetic gestational ageing.
7. Dietary GL in children and adolescents was positively associated with the methylation of cg20274553 (WDR27). Stratification in children by weight identified several DNAm sites related to GI or GL. Among all identified DNAm sites (N = 537), 76 in blood and 89 in adipose tissue were related to the expression of genes that modulate metabolic processes.
8. Higher sugar-containing beverage intake in infancy was associated with NAFLD in school-aged children, independent of sugar-containing beverage intake and BMI at school age.
9. DNAm at three CpGs (cg05937453, cg25212453, and cg10040131), each in a different age range, was associated with BMI at Bonferroni significance, P < 1.06 × 10−7. DNA methylation at the 187 CpGs previously identified to be associated with adult BMI, increased with advancing age across childhood and adolescence in our analyses. In addition, correlation coefficients between effect estimates for those CpGs in adults and in children and adolescents also increased. Among the top findings for each age range, we observed increasing enrichment for the CpGs that were previously identified in adults (birth Penrichment = 1; childhood Penrichment = 2.00 × 10−4; adolescence Penrichment = 2.10 × 10−7).
10. Development of a life-course model using data from 7 life stages: pre-natal, birth, infancy, childhood, adolescence, early adulthood at 31 years and late adulthood at 46 years, to explore the dynamic determinants of metabolic heath.
11. Testing and application of the above life course model in NFBC1966 data showed that early life up to age of 11y and early adulthood are critical periods to tackle later BMI development.
12. Proxies for the two identified methylation markers were not yet  available for MR analyses but MR analyses were carried out using another methylation marker at TXNIP (from published literature) that was associated with T2D in EWAS meta-analyses as a proof of principle. This, however, showed a null causal association between TXNIP and T2D.
We advanced our knowledge on mechanisms of how environment may impact on people’s health and identified elements for preventative measures.

Evidence is emerging to support that a range of pre- and postnatal life-style exposures (e.g. gestational diabetes mellitus (GDM) of the mother, maternal smoking, nutrition) may have an impact on the epigenome and can induce epigenetic marks (e.g. DNAm marks) that predict risk of metabolic disease later in life.
Here we will summarise the key highlights by work package. During the first year of the project, we completed the work on prenatal smoke exposure within PREcisE (Wiklund et al, 2019) showing a dose-response relationship between maternal smoking and blood DNAm in the offspring, replicated permanency of several of these methylation markers until middle age and associations with some causal evidence with adult disorders. These observations have clear public health importance. Public policy actions and guidelines may be improved to better guide strategies for prevention. Still in some countries prevalence of maternal smoking is high.
Within WP1 we have undertaken and completed multiple EWAS meta-analyses to investigate the associations of maternal glycaemic health and metabolism during pregnancy with offspring DNAm, discovering one of the best studied epigenetic marks to date. In particular, we have investigated DNAm profiles in new-born blood and blood collected in later childhood in relation to maternal fasting glucose levels (plasma or whole blood values), maternal glucose response to oral glucose tolerance test (OGTT) and maternal fasted serum insulin levels during pregnancy. The projects were led by PREcisE researchers and in collaboration with Pregnancy and Childhood Epigenetics, PACE, consortium. The analyses showed the association of maternal glucose (area under the curve from OGTT) with child DNAm and identified two adjacent CpGs in the well-known TXNIP gene. TXNIP methylation has been previously found to be associated with metabolic phenotypes and future T2D risk. Both CpGs in TXNIP in our analyses are novel, not yet been described in the literature. Follow up in PREcisE cohorts and datasets showed that DNAm in liver was associated with TXNIP expression and that maternal T1D was also associated with childhood DNAm at TXNIP. In addition, DNAm at TXNIP was associated with multiple metabolic phenotypes in childhood and adulthood. A scientific paper detailing the meta-EWAS was published.
Within WP2 we have completed the harmonisation of dietary indices between cohorts, to be included as pre- and postnatal exposure variables in ongoing meta-analysis on DNAm during childhood and adolescence. The harmonised dietary indices include the glycaemic index and glycaemic load as well as the dietary inflammatory index which has been revised to improve its application in multi-cohort collaborations. A manuscript on the revised calculation of the dietary inflammatory index has been published (PMID: 34748580). The harmonized dietary variables from WP2.1 were taken forward to EWAS meta-analyses on the impact of prenatal and postnatal diet on the epigenome at birth, adolescence, and later ages. We found associations between dietary glycaemic index / glycaemic load during pregnancy and early childhood with DNAm in new-born blood and during later childhood/adolescence. While only few associations between GI/GL and DNAm were observed when analysing the total cohort, stratification into subjects with overweight/obesity and normal weight, revealed that dietary GI/GL was associated with different methylation of numerous CpGs, mostly in the overweight/obese stratum. As this was consistently observed in both meta-analyses, the maternal GI/GL and childhood GI/GL analysis, we suggest that pathways, that are affected by dietary GI/GL, are different between mothers and children with overweight/obesity and with normal weight. This was further strengthened by the finding that associations between maternal early blood glucose concentrations and DNAm in new-borns differed between mothers with overweight/obesity and normal weight mothers.
Within WP3 development of a life course model using seven life stages: pre-natal, birth, infancy, childhood, adolescence, early adulthood at 31 years and late adulthood at 46 years, to explore the dynamic determinants of glycaemic and metabolic heath and its application in NFBC66 data showed that early life up to age of 11y and early adulthood are critical periods for designing appropriate interventions to tackle later BMI development. We discovered that many prenatal and very early factors' associations on distal outcomes are mediated through adiposity rebound measures and that CpGs discovered in TXNIP gene associate with age at adiposity rebound. Work demonstrated the necessity of further structural equation model (SEM) development, update and testing because the earlier packages available could not easily handle complex, multiple type of data. Substantive amount of work was spent on this developmental work and the software is now available for future analyses (under review). We have also worked together with DataSHIELD (https://www.datashield.org/) that offers a platform for analyses of multiple datasets at the same time. Our software is implemented in DataSHIELD, tested and ready for use. The SEMs allow, with triangulation, some inferences of causal associations.