Maternal metabolic health and neurodevelopmental conditions in offspring

Abstract

Background: Observational studies published in the last decade have indicated relationships between maternal “overnutrition” states and offspring neurodevelopmental conditions (NDCs), such as autism, attention deficit/hyperactivity disorder (ADHD), and intellectual disability (ID). “Maternal overnutrition” states have been characterized by a series of metabolic conditions before pregnancy (i.e., overweight/obesity, Type I [T1DM] and II [T2DM] diabetes) and during pregnancy (i.e., gestational diabetes mellitus [GDM] and excessive gestational weight gain [GWG]). NDCs often co-occur and have multifactorial etiologies, shaped by both genetic and environmental factors. However, previous studies have not thoroughly considered these complex etiologies when examining associations. For instance, they did not explore whether the relationships between maternal diabetes and offspring NDCs could differ based on the cooccurrence of NDCs or be influenced by genetic predispositions. Moreover, as the fetal brain evolves dramatically and sequentially during pregnancy, it accentuates the need for epidemiological studies to account for the timing and intensity of perturbations during this period. Prior research hasn’t determined whether relationships between maternal conditions such as excessive GWG or hyperglycemia and offspring NDCs could differ based on the GWG rate and glucose concentrations at different pregnancy phases. Maternal overweight/obesity and diabetes might be associated with offspring NDCs due to complications encountered during pregnancy, childbirth, and the neonatal period. Research has suggested that these complications lie at the intersection of, and relate to, maternal metabolic conditions and offspring NDCs. Grasping the mediation of these complications can offer deeper insights into preventative measures during these stages; however, no studies have yet quantified these complications’ mediating impact on the associations. Lastly, the causal influence of BMI, including maternal BMI, on offspring autism and ADHD has seldom been thoroughly explored. In the absence of compelling evidence, the question remains as to whether better weight management among obese women before conception could help reduce the potential risks of offspring NDCs.

Methods: We used two databases, “Psychiatry Sweden (PS), 1987-2016” and “Developmental Origins of Health And Disease (DOHAD), 1997-2021”, which are register linkages across Swedish nationwide registers using the unique identification number assigned to each Swedish resident. Offspring were linked to their biological mothers, fathers, and maternal grandparents using the Total Population Register (Study I, IV, V). We also used a series of maternal weight and capillary glucose records across pregnancy from the Stockholm Obstetrix system, an electronic medical journal of antenatal care, which was nested within the “Stockholm Youth Cohort (SYC)”. The SYC is a part of PS that also includes regional health and administrative registers (Study II, III). In Studies I, IV, and V, we used the National Patient Register to identify offspring NDCs (i.e., autism, ADHD, and ID), which was supplemented by regional register information in Studies II and III as well as the National Prescribed Drug Register (for ADHD). Finally, we utilized genetic data and information from mothers and children in the “Avon Longitudinal Study of Parents and Children (ALSPAC)” cohort (Study V).

In Study I, we utilized a generalized estimating equation (GEE)/population average model with a logit link. This model was clustered based on pseudonymized maternal identification numbers and employed robust standard errors for the computation of odds ratios (ORs) and 95% confidence intervals (CIs) regarding neurodevelopmental conditions (NDCs) in offspring. In Study II, we used Cox regression models, again clustered on maternal numbers and with robust standard errors, to determine hazard ratios (HRs) and 95% CIs for offspring NDCs. In Study III, we employed group-based trajectory modeling (GBTM) to ascertain the varying patterns of glucose alteration throughout pregnancy. GEE models were utilized to evaluate the associations with both obstetric and neonatal outcomes and offspring NDCs. In Study IV, we used a parametric regression approach within a counterfactual framework to conduct both single and multiple mediation analyses. This study aimed to quantify the total effect (TE), natural indirect effects (NIE), and natural direct effects (NDE) in the associations of maternal diabetes (both pregestational diabetes mellitus [PGDM] and GDM) and overweight/obesity with NDCs through individual components of mediators. We employed a paternal negative control comparison analysis in Study I to examine if the associations of maternal T1DM and T2DM with offspring NDCs could be confounded by genetic predispositions to diabetes and NDCs. In Study V, we applied a “triangulation” approach. Analyses were performed using maternal cousin and full sibling comparisons to address unobserved, shared genetic and environmental factors in the associations between maternal BMI and offspring autism and ADHD. In addition, we explored the genetic correlation through Linkage Disequilibrium Score Regression (LDSC). Moreover, we examined the association between the genetic predisposition to both maternal and children’s BMI and various traits of children’s autism and ADHD using Polygenic Risk Score (PRS) analysis. Lastly, we employed a two-sample Mendelian randomization analysis (MR) in Study V to evaluate the causal impacts of BMI on NDCs, including autism and ADHD.

Results: Maternal T1DM, T2DM, and GDM were all associated with offspring autism, ADHD, and ID, with greater risks linked to comorbid diagnoses involving ID. Stronger associations with GDM were observed when diagnosed between 27-30 wkGA. Paternal T1DM and T2DM were also associated with offspring NDCs, though the strength of these associations was less than those observed with maternal diabetes (Study I). Lower rates of GWG in the second trimester and higher rates of GWG in the third trimester were associated with increased risks for offspring NDCs (Study II). Among those without PGDM, persistently high glucose levels throughout pregnancy demonstrated the strongest association with adverse obstetric/neonatal complications. Transient hyperglycemic states followed by periods of potential glycemic control were also associated with these complications but to a lesser extent. Notably, subclinical states of hyperglycemia, which were less likely to receive a GDM diagnosis, remained associated with these complications, albeit to a lesser degree. A similar pattern of associations was observed for offspring NDCs. Persistently high glucose levels showed stronger associations with offspring NDCs (i.e., ADHD only), while weaker associations were identified with transient hyperglycemic states followed by improved glucose control. Notably, we found that hyperglycemia in early pregnancy, but not in mid-pregnancy, was associated with offspring NDCs when followed by improved glucose control. However, none of these associations regarding NDC outcomes survived the false discovery rate correction using the Benjamini- Hochberg approach (Study III). The joint mediating effects of all obstetric and neonatal complications were more pronounced in the associations between PGDM and offspring NDCs (accounting for 30-50% of the association) than in those concerning maternal GDM and overweight/obesity. Although the mediating effects of obstetric and neonatal complications were generally insignificant for GDM and minor for maternal overweight/obesity, we observed direct associations between GDM (10-30% increased risks compared to non-diabetes) and maternal overweight/obesity (30-60% increased risks compared to normal weight) with the risks of offspring NDCs. However, these associations might still contain residual confounding due to unobserved factors. The combined mediating effects of these complications, especially those emerging during the neonatal period, were particularly strong in the relationship between maternal PGDM and offspring NDCs. For individual mediators, the effects were generally minimal, except for complications such as pregnancy hypertensive diseases, preterm birth, neonatal asphyxia, and hematological comorbidities in the association between PGDM and offspring NDCs (with proportions mediated exceeding 10%) (Study IV). Maternal obesity was linked to increased risks of autism and ADHD in both the full cohort analysis and family designs (i.e., maternal cousin comparisons and full sibling analyses). It is worth noting that when accounting for shared familial factors in family designs, the associations were attenuated but modest associations remained. For instance, among full siblings, children exposed to maternal obesity had a 0.87% higher risk of autism and a 2.13% higher risk of ADHD at age 16, compared to those exposed to mothers of normal weight. The LDSC analysis showed a genetic correlation between BMI and ADHD, but not with autism. The PRS analysis provided less evidence suggesting a relationship between maternal and children’s genetic liability to BMI and various autism and ADHD traits. Specifically, a one-unit increase in BMI was associated with a 12% higher risk for autism and a 77% increased risk for ADHD (Study V).

Conclusions: In conclusion, my research has reaffirmed known associations between maternal metabolic conditions and offspring NDCs while providing new insights into the underlying mechanisms and causal relationships. Greater associations between maternal diabetes and NDCs involving ID suggested distinct pathophysiological mechanisms. The associations involving PGDM and offspring NDCs might be partially confounded by a genetic predisposition to both the exposure and outcomes; however, its intrauterine effects cannot be completely discounted. Further investigation into the causal link is still warranted in the future. To reduce the risk of offspring NDCs associated with maternal PGDM, it can be beneficial to manage specific obstetric and neonatal complications, especially those arising during the neonatal period. For maternal GDM and overweight/obesity, although we found direct associations with NDCs without evident mediating effects from obstetric and neonatal complications, these direct associations might still contain residual confounding due to unobserved factors. GDM during weeks 27-30 of gestation showed a more pronounced association with offspring NDCs. However, women with hyperglycemia in mid-pregnancy who subsequently achieved effective glucose control did not have a notable increased risk of NDCs among their offspring. While this suggests that effective glucose management during mid-pregnancy might benefit offspring neurodevelopment, further studies are needed to confirm a causal link. My research provides evidence of a modest causal relationship between maternal BMI and offspring autism and ADHD. Further, a lower rate of GWG in the second trimester and a higher rate in the third trimester were more strongly associated with offspring NDCs. This suggests that continuous monitoring and potential interventions related to weight and weight gain from conception onward could have a positive, albeit modest, impact on reducing the risks of offspring NDCs, such as autism and ADHD.