Insights from whole exome sequencing in a Maltese cohort with gestational diabetes

Abstract

BACKGROUND: Gestational diabetes (GDM) can be driven by mutations or rare variants in various genes associated with monogenic or atypical forms of diabetes. The reported frequency of monogenic defects of beta cell function in GDM varies extensively, in part due to differences in ethnicity, patient ascertainment criteria and techniques used for genetic analysis. The objective was to evaluate the frequency and molecular spectrum of mutations in a curated list of genes associated with monogenic/atypical diabetes in non-obese women of Maltese ethnicity with GDM.

METHOD: 30 non-obese Maltese women who met the International Association of the Diabetes and Pregnancy Study Group (IADPSG) criteria for diagnosis of GDM and with a first-degree relative with non-autoimmune diabetes were included in this study. Whole exome capture and high throughput sequencing was carried out. Rare sequence variants were filtered, annotated and prioritized according to the American College for Medical Genetics guidelines. For selected missense variants we explored effects on protein stability and structure through homology predictions or PDB structures using in-silico tools.

RESULTS: In total, we identified three pathogenic mutations and twelve variants of uncertain significance (VUS). The disease-causing mutations comprise a nonsense mutation in GCK, an insertion-frameshift at a mutational hotspot in HNF1A and a missense substitution in ABCC8. Critically, the ABCC8 mutation leads to significant changes in interatomic interactions and to expansion of protein cavity volume, with resulting destabilising effects. Damaging VUS in PDX1, KLF11, DYRK1B, TRMT10A, AKT2, BLK, GLIS3 and NKX6-1 were detected, having either conflicting pathogenicity interpretationsor insufficient evidence for pathogenicity from in-vitro studies. Novel NEUROG3 and CEL VUS were also detected. Stereochemical analysis reveals that the missense variants described in NEUROG3, DYRK1B, TRMT10AandAKT2 have destabilising effects. Genotype-phenotype correlations for all detected variants are described, including associations with anthropometric traits, OGTT, HOMA-IR, treatment and post-pregnancy follow-up data where available. We show that GDM cases who were carriers of either pathogenic mutations or damaging VUS had a younger age of GDM diagnosis than females where no variant of interest was identified. (29 vs 32 years, P = 0.039).

CONCLUSION: This study provides the first insight into an underlying monogenic aetiology in non-obese women with GDM from a high-prevalence island population. It suggests that monogenic variants constitute an underestimated cause of diabetes detected in pregnancy, and that careful evaluation of GDM probands to identify monogenic disease subtypes is indicated.