Abstract:
Pre-pregnancy obesity, defined as a body mass index (BMI) greater than or equal to 30 kg/m2, can have adverse effects on the health of newborns and can also lead to metabolic, cardiovascular and neurological diseases in the offspring as they grow older. In the area of fetal origins and disease in adult life, a large number of studies have reported a critical role for maternal weight and metabolism before or during gestation in shaping the health of their offspring. Maternal obesity is recognised as a major modifiable contributor to obesity and metabolic syndrome in offspring, but the underlying factors remain unclear. The fetal autonomic nervous system (ANS) is subject to programming during developmental periods and is considered one of the processes by which early programming of disease can take place. The main goal of the present work was to use the fetal heart rate (HR) and heart rate variability (HRV) as proxies for the fetal ANS to study the effects of metabolic and anthropometric maternal (MAM) parameters before and during gestation on the fetuses of healthy, normoglycemic mothers. A total of 184 women in their second/third trimesters of uncomplicated pregnancies were included in this study. Pre-pregnancy BMI and maternal weight gain during pregnancy were recorded. In a subsample (n = 104), maternal insulin sensitivity was measured during an oral glucose tolerance test. Fetal HR and HRV were determined by magnetic recording in all subjects. The influence of pre-pregnancy BMI, maternal weight gain and maternal insulin sensitivity on fetal HR and HRV was evaluated. Associations between MAM parameters and maternal HR and HRV were also assessed. ANCOVA, partial correlation and mediation analysis were applied, all of which were adjusted for gestational age, gender and parity. A regression on fetal HR using a machine learning approach was tested to explore which maternal factor is the driving factor programming the fetal ANS. Four models were tested: Linear regression, Regression Tree, Support Vector Machine and Random Forest. The fetal HR was higher in fetuses of mothers with high pre-pregnancy BMI (overweight/obese) than in mothers with normal weight. The fetal HRV was lower in mothers with high weight gain than in mothers with normal weight gain. The fetal HR was negatively correlated with maternal weight gain and maternal insulin sensitivity. Pre-pregnancy BMI was positively correlated with fetal high frequency and negatively correlated with low frequency and the low to high frequency ratio. Maternal weight gain was associated indirectly with birth weight through fetal HR, while maternal insulin sensitivity was associated with fetal HR through fetal HRV. Separately, fetal HRV was associated with birth weight through the fetal HR. The Random Forest ensemble tree-based model outperformed linear regression as the fetal HR regression model. Fetal HR can be predicted using the following nine relevant variables (sorted from the most important to the least important): pre-pregnancy BMI, gender, maternal fasting insulin, maternal insulin sensitivity, gravidity, maternal age, maternal fasting glucose, gestational age and maternal weight gain. Pre-pregnancy BMI appeared to be the major factor predicting fetal HR. In conclusion, the fetal ANS is sensitive to maternal metabolic and anthropometric influences, and particularly maternal weight before pregnancy. These findings support the concept of the “Developmental Origin of Health and Disease” and increase our knowledge about the importance of the intrauterine environment in the programming of the ANS and the possible programming of disease in later life.