Metabolomic insights into maternal and neonatal complications in pregnancies affected by type 1 diabetes

Aims/hypothesis Type 1 diabetes in pregnancy is associated with suboptimal pregnancy outcomes, attributed to maternal hyperglycaemia and offspring hyperinsulinism (quantifiable by cord blood C-peptide). We assessed metabolomic patterns associated with risk factors (maternal hyperglycaemia, diet, BMI, weight gain) and perinatal complications (pre-eclampsia, large for gestational age [LGA], neonatal hypoglycaemia, hyperinsulinism) in the Continuous Glucose Monitoring in Women with Type 1 Diabetes in Pregnancy Trial (CONCEPTT). Methods A total of 174 CONCEPTT participants gave ≥1 non-fasting serum sample for the biorepository at 12 gestational weeks (147 women), 24 weeks (167 women) and 34 weeks (160 women) with cord blood from 93 infants. Results from untargeted metabolite analysis (ultrahigh performance LC-MS) are presented as adjusted logistic/linear regression of maternal and cord blood metabolites, risk factors and perinatal complications using a modified Bonferroni limit of significance for dependent variables. Results Maternal continuous glucose monitoring time-above-range (but not BMI or excessive gestational weight gain) was associated with increased triacylglycerols in maternal blood and increased carnitines in cord blood. LGA, adiposity, neonatal hypoglycaemia and offspring hyperinsulinism showed distinct metabolite profiles. LGA was associated with increased carnitines, steroid hormones and lipid metabolites, predominantly in the third trimester. However, neonatal hypoglycaemia and offspring hyperinsulinism were both associated with metabolite changes from the first trimester, featuring triacylglycerols or dietary phenols. Pre-eclampsia was associated with increased abundance of phosphatidylethanolamines, a membrane phospholipid, at 24 weeks. Conclusions/interpretation Altered lipid metabolism is a key pathophysiological feature of type 1 diabetes pregnancy. New strategies for optimising maternal diet and insulin dosing from the first trimester are needed to improve pregnancy outcomes in type 1 diabetes. Graphical Abstract Supplementary Information The online version of this article (10.1007/s00125-023-05989-2) contains peer-reviewed but unedited supplementary material.


Metabolomic and lipidomic changes associated with maternal BMI
Baseline characteristics and pregnancy outcomes of women categorised according to enrolment BMI are shown in ESM Table 1.There were no significant differences in LGA rates between lean (38/90; 42.2%) and overweight/obese (30/84; 35.7%) women.

Metabolomic and lipidomic changes associated with gestational weight gain (ESM Figure A2.2, ESM Table 3 and 4)
Data on gestational weight gain from enrolment to 34 weeks was available on 141 women.Baseline characteristics and pregnancy outcomes of women categorised according to the presence or absence of excessive gestational weight gain (using Institute of Medicine thresholds) are shown in ESM Table 3.As expected, the presence of excessive gestational weight gain influenced the maternal metabolome only in late pregnancy.At 34 weeks there was a positive association with PC(18:0/18:3) and a negative association with alpha-tocopherol.The cord blood metabolome showed no associations with maternal excessive gestational weight gain.

Metabolomic and lipidomic changes associated with habitual diet (ESM Figure A2.3, ESM Table 5-8)
Baseline characteristics and pregnancy outcomes of women categorised according to dietary practices are shown in ESM Tables 5 and 7. Dietary information was available on 56 women at 12 weeks.Participants reported consuming mean 1652 kcal/day (SD 451.5; median 1729 kcal/day) with a macronutrient composition of 43.9% (7.3) energy from carbohydrates, 37.4% (6.4) from fat and 17.5% (4.2) from protein.Maternal total energy intake was positively associated with piperine sulfate (found in black pepper) at 12 weeks (ESM Figure A2.3).Maternal carbohydrate intake (% of total energy intake) was positively associated with CE(16:1).Maternal protein intake (% of total energy intake) was positively associated with two PE plasmalogens, and negatively associated with monoglyceride (18:1).
Maternal fat intake (% of total energy intake) showed negative associations with a cholesteryl ester, a lysophosphatidylcholine (LPC(16:1)), a lyso-phosphatidylethanolamine (LPE(16:1)), and multiple PCs, PIs and triglycerides containing fatty acids (16:0) and (16:1).Women consuming a higher energy diet (energy intake above the median) had a slightly lower HbA1c at 12 weeks but no significant differences in pregnancy outcomes compared to women consuming a lower energy diet (ESM Table 5).A comparison of women who consumed a low-carbohydrate high-fat diet (LCHF; average 35% carbohydrate, 45% fat and 20% protein) demonstrated no differences in glycaemia or outcomes, except for increased neonatal hypoglycaemia on adjusted regression analysis ESM Table 7).Small numbers in the dietary analysis preclude definitive conclusions.
Maternal BMI was not significantly associated with maternal intake of fat, other macronutrients or polyunsaturated fatty acids (ESM Table 1).Maternal intake of retinol or carotene was not associated with maternal BMI on unadjusted analysis or when adjusted for total maternal energy intake.
ESM Figure A2.1 -Metabolites in maternal blood at 12, 24 and 34 weeks' gestation and in cord blood at birth in association with maternal BMI in early pregnancy (measured at first visit, approx.8 -9 weeks' gestation).
ESM Figure A2.2 -Metabolites in maternal blood at 12, 24 and 34 weeks' gestation and in cord blood at birth in association with gestational weight gain.
Regression of effect of maternal hyperglycaemia at 24 weeks upon skinfold sum, including all listed maternal triglycerides in ESM Figure A2.2:Coeff 13.74 (2.66 to 24.84); p=0.0160

Do triglycerides in CORD blood mediate the relationship between maternal hyperglycaemia and offspring adiposity?
Regression of effect of maternal glycaemia at 24 weeks upon skinfold sum: Regression of effect of maternal hyperglycaemia at 24 weeks upon skinfold sum Coeff 16.20 (10.57 to 21.83); p=9.28 x 10 -8 .Regression of effect of maternal hyperglycaemia at 24 weeks upon skinfold sum, including all listed cord blood triglycerides Coeff 22.16 (9.50 to 34.81); p=0.00132
Regression of effect of maternal hyperglycaemia at 24 weeks upon skinfold sum, including all listed cord blood triglycerides Coeff 18.27 (0.91 to 35.64); p=0.0401

Interpretation:
Adjustment for specific maternal or cord blood triglycerides markedly reduces the significance of the association between maternal hyperglycaemia and offspring adiposity at 24 weeks and attenuates associations at 34 weeks.This suggests that triglyceride and related species have a key role in mediating the relationship between maternal hyperglycaemia and offspring body composition.However, adjusting for triglycerides did not consistently attenuate the association, suggesting that glucose itself and/or other mediators may also play a role.
A note on methods and limitations: studying the physiology and pathophysiology of human pregnancy is challenging.We present our mediation analysis in order to extend the body of knowledge and yield new testable hypotheses.However, lipids are highly inter-related, which causes some issues for a mediation analysis.For example, fatty acids can transfer easily from triglycerides to diglycerides and monoglycerides, or exist as free fatty acids in the blood.Free fatty acids can quickly be incorporated into phospholipids and other larger lipid structures.This means that any findings related to triglycerides need to be interpreted as part of a broad overview of lipid homoeostasis at any one time.
Furthermore, the physiology of lipid transfer across the placenta remains imperfectly understood.While it is tempting to assume that triglycerides passing through the placenta are directly mediating these effects, it is likely that fatty acids, but not assembled triglycerides, can pass through the placenta.Fatty acids may reassemble into triglycerides on the fetal side of the placenta.However, fetal triglycerides are also likely to arise from fetal de novo lipogenesis, making new fat, which is likely to be crucial part of early development.Coeff 17.20 (9.66 to 24.75); p=0.0000166 (1.66 x 10 -5 ).
Regression of effect of maternal glycaemia at 24 weeks upon skinfold sum (all carnitines and plasmalogens mentioned above in ESM Figure A3.3 included in the model): Coeff 5.97 (-6.89 to 18.83) p=0.353

Interpretation:
Carnitines and PE plasmalogens appear to be important mediators of the relationship between maternal hyperglycaemia and offspring adiposity.However, they do not appear to be the sole mediators of this relationship as they do not consistently obliterate the significant relationship between hyperglycaemia and skinfold sum.