Largest genetic study of gestational diabetes discovers nine novel genetic regions linked to severe and common pregnancy complication

By Vijay Kumar MalesuReviewed by Danielle Ellis, B.Sc.Jan 9 2024

In a recent study published in Nature Genetics, a group of researchers investigated the genetic underpinnings of Gestational Diabetes Mellitus (GDM) and its relationship with Type 2 Diabetes (T2D) through a genome-wide association study ((GWAS), identifying distinct and shared genetic factors.

Background 

GDM, increasingly prevalent in diverse populations over the past 15 years, poses significant risks to mothers and children, yet its genetic basis, particularly in relation to T2D, remains largely unexplored. Previous GWAS on GDM identified five significant loci, mostly overlapping with T2D, suggesting a shared genetic etiology. Further research is essential to fully understand the unique genetic factors and mechanisms underlying GDM, distinct from those of T2D.

About the study

In the present study, participants gave informed consent for biobank research in accordance with the Finnish Biobank Act. For cohorts collected prior to the Act and FinnGen's initiation, study-specific consents were used, and later, these cohorts were transferred to Finnish biobanks with approval from Fimea, Finland's national supervisory authority for welfare and health. Specific Biobank Access Decisions were also in place for FinnGen samples and data used in this research.

The FinnGen study, a public-private partnership, integrates data from Finnish biobanks with national registry electronic health records, including hospital and outpatient visits, causes of death, primary care, and medication records. The study used data from FinnGen release R8, encompassing 342,499 individuals. Phenotyping was carefully carried out, with clinical endpoints and corresponding dates constructed for gestational diabetes and related diagnoses. A 'pregnancy window' was defined, and pregnancies were categorized based on gestational diabetes criteria and exclusion factors.

Genotyping and GWAS methodologies were detailed in online documentation. FinnGen individuals were genotyped using Illumina and Affymetrix chip arrays, followed by quality control and imputation using a population-specific reference panel. Unrelated individuals of Finnish ancestry were identified for the GWAS, conducted using Ridge Regression-based Efficient GENome-wide association study Imputation method for Exploratory large-scale data analysis (REGENIE) software with various covariates.

Fine-mapping was performed using the Sum of Single Effects (SuSiE) algorithm, focusing on a 1.5-Mb locus around GWAS lead Single Nucleotide Polymorphisms (SNPs). Independent signals were identified based on primary and secondary signals, each with genome-wide significance. Replication studies included samples from FinnGen, the Estonian Biobank, and meta-analyses of these cohorts, as well as the GenDIP consortium meta-analysis.

Annotation of variants involved using Ensembl Variant Effect Predictor and comparison with non-Finnish European populations. Colocalization was performed using the probabilistic model expression Coloc Association Visualization, Annotation, and Integration Resource (eCAVIAR), integrating GWAS and Expression Quantitative Trait Loci (eQTL) data. Gene enrichment analysis was conducted using Multi-marker Analysis of GenoMic Annotation (MAGMA) results, identifying tissue and pathway enrichments.

Genetic correlations between GDM and related diseases or traits were estimated using Linkage Disequilibrium Score Regression (LDSC) software. The study also developed SCOUTJOY, an algorithm to compare the heterogeneity of GDM-associated loci's genetic effects across disorders. Shared variants analysis was applied to GWAS summary statistics from T2D and GDM GWAS, classifying variants based on their bivariate effect sizes.

Cell-type specificity analyses were conducted using high-quality single-cell mouse datasets. Tissue-level associations were identified using Tabula Muris data, and specific cell types were analyzed to gain better resolution on their involvement in GDM and T2D. This analysis included comparing pancreatic results to human Single-Cell RNA Sequencing (scRNA-seq) to assess differences in pancreatic cellular function and physiology.

Study results

The researchers conducted a GWAS on GDM involving 12,332 cases and 131,109 controls of Finnish ancestry from the FinnGen study. They identified cases using Finnish health and population registries, focusing on diagnoses within a specific pregnancy window and excluding pre-existing diabetes. This study significantly advanced GDM knowledge, identifying 13 chromosomal regions associated with GDM, thereby tripling the known loci.

The research included replication studies with new samples from FinnGen and the Estonian Biobank. Fine-mapping of these loci revealed 14 independent signals, including nine new GDM associations, and integrated data from over 3,800 GWAS and other genomic resources.

The research also included a detailed analysis of the shared genetic etiology with T2D. Using the Significant Cross-trait Outliers and Trends in Joint York regression algorithm, the study found that the GDM-associated loci exhibited significant heterogeneity in their relationship to T2D. Notably, five of these loci were not significantly associated with T2D, highlighting distinct genetic factors in GDM. The study also revealed significant genetic correlations between GDM and 12 diseases or traits, as well as eight blood laboratory values related to the disorder.

Further, a Bayesian classification algorithm was applied to compare the effects of GDM and T2D. This analysis revealed two distinct classes of significant variants: one predominantly affecting GDM and the other T2D. This suggested that the genetic risk of GDM falls into two categories: one shared with T2D risk and another predominantly influencing gestational mechanisms.

Additionally, the study explored cell-type specific expression patterns, highlighting significant associations with certain cell populations within maternal tissues, like pancreatic β cells and hypothalamic neurons. This analysis was crucial given the major adaptive changes induced by pregnancy.

The study's findings underscore the partial distinctness of GDM's genetic etiology from T2D. Although GDM shares a polygenic predisposition with T2D, there is a separate category of genetic risk factors predominantly gestational in nature. This is evident in the substantial effect of the melatonin receptor 1B (MTNR1B) locus, which, along with other loci, indicates a larger impact on GDM than on T2D.

Finally, the research acknowledged the need for further studies to understand the molecular underpinnings of GDM fully. This includes characterizing the precise molecular effects predominant in GDM, exploring the roles of gestational hormones and their impact on glycemic pathways, and considering genetic effects in specific tissues during pregnancy.

The findings from this study, conducted in a Finnish population, highlight the need for additional studies in diverse populations to gain a comprehensive understanding of GDM's genetic basis. This work underscores the importance of focusing on pregnancy disorders to discover new physiological mechanisms of glycemic or homeostatic control.