Impact of maternal bariatric surgery on offspring perinatal cardiac function: A prospective study

To assess perinatal cardiac function in offspring of women with previous bariatric surgery and examine its association with maternal glucose control.


| I N TRODUC TION
Worldwide, the rates of obesity have tripled since 1975.In 2016, almost 40% of women were overweight and 15% were obese. 1 In the recent Mother and Babies Reducing Risk through Audits and Confidential Enquiries across the UK (MBRRACE-UK) report, obesity is noted to be an important contributing factor to adverse pregnancy/maternal outcomes, so an effective way for obese women to achieve normal weight before conception is needed. 2nfortunately, lifestyle interventions and pharmacological therapy have not been successful in achieving durable weight loss. 3,46][7][8] As a result, the number of women presenting pregnant, having had a previous bariatric surgery, is increasing. 9It is now well established that pregnancy following bariatric surgery is associated with a reduction in the rates of gestational diabetes mellitus (GDM), hypertensive disorders and macrosomic babies, but an increase in the rates of small-for-gestationalage neonates. 10,11Studies on the effects of maternal bariatric surgery on the short-term and long-term cardiometabolic health of their offspring have given conflicting results.3][24] Several animal studies have shown impaired fetal cardiac function among obese mothers. 18Some human studies have also demonstrated that fetuses of obese women have signs of cardiac dysfunction from as early as 14 weeks of gestation, 19,20 and altered autonomic nervous system activity. 21Although the exact mechanisms are unknown, obesity-mediated maternal hyperglycaemia, leading to fetal hyperinsulinaemia 25 and upregulation of the inflammatory mediators of the myocardium have been implicated. 26he aim of the current study was to examine the perinatal cardiac function, in both the last trimester of pregnancy and early infancy, of offspring of women with previous bariatric surgery compared with offspring of women without surgery, and to investigate its association with maternal glucose control.As pre-pregnancy maternal bariatric surgery is associated with an improvement in maternal insulin resistance, 27 and considering the link between maternal hyperglycaemia and fetal cardiac dysfunction, 26 we hypothesised that offspring of post-bariatric women will demonstrate better perinatal cardiac function.

| Study population
This study was part of a large prospective longitudinal study aiming to investigate the effect of maternal obesity, and bariatric surgery in particular, on pregnancy outcomes.The study design has been previously described. 27In brief, women with previous bariatric surgery were matched for early pregnancy body mass index (BMI) with pregnant women with no such surgery.Women were seen; their weight (kg), height (cm), BMI (kg/m 2 ) and blood pressure were measured; blood and urine samples were collected and fetal growth ultrasound scans were performed longitudinally during pregnancy at 11-14, 20-24, 30-32 and 35-37 weeks of gestation.From 20 to 24 weeks onwards, the estimated fetal weight was calculated, based on measurements of fetal head, abdominal circumference and femur length. 28For women with previous bariatric surgery, the date of surgery and their weight before surgery were recorded.Based on this information, the time from surgery to conception, total weight loss (%) [((pre-surgery weight-early pregnancy weight)/pre-surgery weight) × 100] and excess weight loss (%) [((pre-surgery excess weight-early pregnancy excess weight)/pre-surgery excess weight) × 100] were calculated.At 27-30 weeks all women underwent screening for GDM.All no-bariatric and the majority of post-bariatric women (n = 20) underwent a 75-g, 2-hour full oral glucose tolerance test as a diagnostic test for GDM, which was defined according to National Institute for Health and Care Excellence guidelines. 29A small number of postbariatric women (n = 9), recruited after December 2017, had fasting glucose levels measured and capillary blood glucose monitoring for 2 weeks, as studies have suggested that the oral glucose tolerance test may not be a reliable method to diagnose GDM in this group of women. 30At the same time, 27-30 weeks of gestation, maternal levels of glycated haemoglobin (HbA1c), which can reflect the overall glucose control over a period of weeks/months, and haemoglobin were measured in all participants (G8 Glycohemoglobin Analyser, Tosoh Corporation; intra-assay variability 0.6%, inter-assay variability: 0.56%).
Information on pregnancy outcome was obtained from the hospital's perinatal database.Birthweight centiles were calculated based on birthweight and gestational age at delivery. 31Offspring were seen again at 5-7 weeks of age, had their weight measured and infant weight centile calculated. 32ll women gave written informed consent, and the study was approved by the Local REC (No: 14/LO/0592).

| Fetal and neonatal echocardiography
For the purposes of this study, cardiac function of the offspring was assessed by two-dimensional (2D) echocardiography at 35-37 weeks of gestation and at 5-7 weeks of age.All echocardiograms were performed by experienced fetal and paediatric cardiologists. 33etal and postnatal studies included B-mode, Mmode, spectral pulsed-wave (PW) Doppler, spectral tissue Doppler imaging (PW-TDI) and speckle tracking imaging echocardiography (STE).Three investigators (JSC, MB, OP) performed all fetal and neonatal ultrasound examinations using an ARIETTA V70 (Hitachi Ltd, Tokyo, Japan).Fetal M-mode, B-mode and PW Doppler measurements were made with the convex array transducer C251 (5-1 MHz), while the Sector Single Crystal probes S31 (9-2 MHz) and S42 were used for the neonatal and paediatric heart examination.Cross-sectional images for STE analysis was recorded as raw data using the same ultrasound transducer (adult linear transducer S12 5-1 MHz) in both fetal and neonatal groups.
M-mode ultrasound was used for assessment of cardiac systolic function (shortening fraction estimation) and longitudinal annular plane systolic excursion of tricuspid and mitral valves and interventricular septum (TAPSE, MAPSE and SAPSE, respectively).B-mode imaging was performed for obtaining measurements of valve dimensions, ventricular chambers, and both left ventricular (LV) and right ventricular (RV) sphericity indices calculated by dividing ventricular end-diastolic dimension by end-diastolic length for each ventricle. 34,35PW Doppler technique was used to obtain Doppler signals from the inflow and outflow tracts for evaluation of diastolic and systolic function, respectively.LV and RV stroke volume (SV), cardiac output (CO) and cardiac index (CI) were calculated as follows: SV = outflow valve area × Velocity Time Integral, CO = SV × Heart Rate, and CI = CO/body weight.PW-TDI technique was applied to measure systolic and diastolic myocardial velocities and time intervals with calculation of LV and RV myocardial performance index.STE was used to derive myocardial deformation parameters such as longitudinal strain and systolic strain rate with frame rate greater than 100 frames per second.5][36] Several digital clips were obtained and then transferred as raw data to the Tissue Tracking software dedicated for 2D (DAS-RS1; Hitachi Ltd, Japan) for further analysis, which was performed by one investigator (OP).The time-interval values were adjusted by cardiac cycle length considering differences in heart rate.][38][39] Finally, we investigated any possible associations between maternal glucose control, as assessed by HbA1c, and fetal/ infant cardiovascular parameters.

| Statistical analysis
The Kolmogoroff-Smirnoff test was used to assess normality of the data distribution.Normally and not normally distributed continuous variables were expressed as mean ± standard deviation or as median (interquartile range), respectively.Logarithmic transformation was performed, when necessary.Categorical data are summarised using count and percentages.In order to examine the differences between the study groups, unpaired t test/Mann-Whitney U test and chi-square tests were used for numerical and categorical data, respectively.Pearson's correlation and multiple linear regression analyses were used to examine the relation between maternal variables and prenatal/postnatal cardiac parameters in the groups, where appropriate.Correlation coefficient (r) and 95% confidence intervals are given, when appropriate.
There is no previous study on the cardiac function of offspring of women with previous bariatric surgery, so we based our sample calculation on previous studies in fetuses of diabetic pregnancies. 34Power analysis indicated that 20 pregnant women in each group (no-bariatric and post-bariatric) would have a power over 80% with α = 0.05 (two tails) to detect a mean difference of 0.46 in the relative interventricular septum thickness; a difference reported between fetuses of normal and diabetic mothers. 34The statistical software package SPSS Statistics 23.0 (SPSS Inc., Chicago, IL, USA) was used for data analyses.Differences were considered as significant only at p < 0.01 (Bonferroni correction for type 1 error or false-positive results of multiple measurements).

| R E SU LTS
The study period was May 2016 to August 2019.The study included 29 fetuses of post-bariatric pregnant women (4 with gastric band, 8 with sleeve gastrectomy and 17 with gastric bypass) and 25 fetuses of women without surgery but similar maternal booking BMI.One woman from the post-bariatric group and one from the no-bariatric group did not attend for the postnatal echocardiogram, leaving 28 post-bariatric and 24 no-bariatric infants for assessment.Descriptive maternal, pregnancy and offspring characteristics of the study population are given in Table 1.

| Maternal and pregnancy characteristics
Women with previous bariatric surgery were older compared with the no-surgery group but there were no other differences in the maternal demographics.The maternal BMI of women at booking and at 36 weeks of gestation, when fetal echocardiography was performed, was similar in the groups.In the post-bariatric group, the mean surgery to conception interval was 60.9 ± 36.7 months and women lost, on average, 11.8 ± 8.0 BMI units, had a total weight loss of 27.35 ± 19.80% and excess body weight loss of 54.85 ± 38.81%, from presurgery to early pregnancy.None of the participants had preexisting diabetes and only one woman in the post-bariatric group developed GDM requiring metformin in the late third trimester.

| Fetal and infant parameters
Compared with the no-bariatric group, fetuses of postbariatric women tended to be smaller at the 36-week ultrasound assessment, birth and 5-7 weeks of age (Table 1).There were no major structural cardiac abnormalities in any of the fetuses.In five fetuses (four from the no-bariatric group and one from the post-bariatric group), a small muscular ventricular septal defect was suspected, which was confirmed in two of the no-bariatric infants at the time of the postnatal scan.
The prenatal and postnatal cardiac parameters of the two groups are given (unadjusted values) in Table 2. On both fetal and infant echocardiograms, valve diameters and    PW Doppler velocities were within the normal range.There were no significant differences between the no-bariatric and post-bariatric fetuses or infants.In particular, there were no differences in the STE parameters between the groups, and using multiple regression, we found that bariatric surgery was not a significant determinant of any of the STE parameters, after adjusting for possible confounders including maternal age, ethnicity, smoking, method of conception, BMI, gestational age and estimated fetal weight at the time of the fetal echocardiography, or age and infant weight, at the time of the postnatal assessment.
Offspring from both groups demonstrated similar cardiovascular transitional changes, from fetal to neonatal circulation, with improvement in indices of LV function, as assessed by myocardial function parameters (data not provided but available on request).

| Correlation between maternal indices and fetal/infant cardiovascular parameters
We investigated the correlation between maternal glucose control, as assessed by measurement of HbA1c, and fetal/infant cardiovascular parameters.In the fetal cohort, overall, we found a positive correlation between maternal HbA1c and fetal LV longitudinal strain (r = 0.33), LV longitudinal strain normalised by LV length (r = 0.34), LV longitudinal strain rate (r = 0.29) and LV longitudinal strain rate normalised by LV length (r = 0.30), indicating an inverse correlation between maternal glucaemic levels and fetal LV systolic function, which, nevertheless, remained within the normal range.When we examined the groups separately, we found positive correlations between maternal HbA1c and fetal LV longitudinal strain (r = 0.56), LV longitudinal strain normalised by LV length (r = 0.56), LV longitudinal strain rate (r = 0.50) (Figure 1) and LV longitudinal strain rate normalised by LV length (r = 0.52) in fetuses of women with no previous surgery but no correlations in fetuses of post-bariatric women (r values of 0.06, 0.08, 0.16 and 0.18, respectively).
In the infants, there was no correlation between maternal glucose control, as assessed by HbA1c, and any of the STE indices in the whole group or any of the subgroups.

| Main findings
In the current study, we have shown that, reassuringly, the prenatal and postnatal cardiac function of offspring of women with previous bariatric surgery is similar to that in offspring of women with no history of weight loss surgery, suggesting that maternal bariatric surgery is unlikely to inadvertently affect the cardiovascular performance of their children, at least in the late third trimester of pregnancy and early infancy.Despite having undergone bariatric surgery, the majority of women in the post-bariatric group remained overweight/obese and the values of fetal cardiovascular parameters reported here are similar to the values previously published in obese pregnant women, 18 adding external validity to our results.

| Strengths and limitations
This was a longitudinal study from fetal to postnatal life and therefore we were able to assess perinatal changes.All cardiac assessments were performed by expert fetal cardiologists using advanced echocardiographic methods.We examined the fetal cardiac function in late pregnancy, at the time when the fetal heart is under more stress compared  with earlier gestational ages.Additionally, we assessed the infant's cardiac function at 6 weeks of life, when the heart has already adapted to the postnatal environment and circulation, so avoiding transient perinatal changes.Our sample size was calculated to detect a difference in the relative interventricular septum thickness, similar to the one seen in diabetic compared with uncomplicated pregnancies. 34owever, because of the small number of cases, our study may not have had enough power to detect differences in all cardiac indices and correlations assessed or the effect of the different types of maternal weight loss surgery on the offspring heart performance.Despite having undergone a weight loss surgery, most women in this group continued to be overweight/obese and our findings may not be applicable to women who achieve a normal BMI, as a result of the surgery.As we assessed the cardiac function at 6 weeks of age, we cannot comment on the offspring cardiovascular health later in life and, therefore, long-term follow-up studies will be required.Although HbA1c levels reflect the average glucose levels over the preceding 3 months, more detailed assessment of the maternal glucose homeostasis throughout pregnancy may be required to confirm our findings.

| Interpretation
9][20][21][22][23][24] Although, reassuringly, we did not find a difference between the cardiac performance of offspring of women with previous bariatric surgery and those without, we have shown an inverse correlation between maternal glucose control, as assessed by HbA1c, and fetal cardiac systolic function; although this was maintained within the normal range.In animal models, and in the context of overt diabetes, maternal hyperglycaemia has been associated with increased cardiac growth, as a result of elevated collagen synthesis, suppressed fibronectin synthesis, profibrosis, apoptosis, together with altered expression of cardiac genes involved in contractile, electrical, endocrine and metabolic function of the heart. 40,41Human studies have also shown that fetuses of women with established pre-gestational diabetes demonstrate altered cardiac geometry, myocardial deformation and ventricular function, 34 and these changes appear to persist in infancy, 42 when abnormal cardiovascular structure and function, as assessed by carotid intima-media thickness, carotid-femoral pulse wave velocity and aortic augmentation index 43 are also seen.
1][42][43] This is comparable to the associations, described in other studies, between increasing (but still normal) maternal glucose levels, across a continuum, and increasing birthweight, cord C-peptide and cord insulin levels. 44Of note, in our study, the inverse correlation between maternal glucose control and fetal cardiac systolic function was seen only in the no-bariatric group.It appears that in the post-bariatric group there may be a disruption of the maternal-fetal metabolic signalling, similar to the lack of correlation between maternal and neonatal insulin resistance described in this population. 27The reasons for this dissociation are unknown but could be related to hormonal and metabolic changes induced by the bariatric surgery and include Bile flow alteration, Reduction in stomach size, Anatomical gut rearrangement and altered flow of nutrients, Vagal stimulation and Enteric gut hormone modulation (the BRAVE effect), 45 which could lead to changes in the maternal metabolome, gut microbiome and metabolic signature of the neonate. 46Currently, we are uncertain whether the F I G U R E 1 Scatterplots illustrating the correlation (regression line) between maternal glycated haemoglobin (HbA1c) and fetal left ventricular (LV) strain in the no-bariatric (A) (r = 0.56; 95% CI 0.20-0.78)and post-bariatric (B) (r = 0.06; 95% CI: −0.32 to 0.43) groups.'independence' of the fetal unit from the maternal unit, seen in the post-bariatric pregnancies, could play a role in fetal development and adaptation and, as such, influence the future metabolic health of these children.We did not find an association between maternal glucose control and the infant's cardiac function, suggesting that either maternal glucose homeostasis during pregnancy does not play a role or that other environmental factors, such as type of feeding, may be involved in determining the future cardiovascular performance of the offspring. 47At least a third of our neonates were on mixed or formula feed at the time of hospital discharge (Table 1), and probably more at the time of the postnatal echocardiography assessment.We and others have previously shown that offspring of women with previous bariatric surgery are smaller and thinner during pregnancy 48,49 and at delivery, as assessed by skin-fold measurements and dual-energy-X-ray absorptiometry scanning. 50Although in the current study, the rate of small babies in our groups was not significantly different, probably due to the small sample size, we have found post-bariatric offspring to have lower weight at birth and in early infancy (5-7 weeks of age), compared with those of women without surgery.3][14][15] More studies are needed to investigate whether maternal bariatric surgery itself, low birthweight, postnatal weight gain or diet has greater impact on the offspring's weight later in life.

| CONCLUSIONS
In summary, the study has shown that the cardiac performance of the offspring of women with previous bariatric surgery, assessed in late pregnancy and early infancy, is not altered.In no-bariatric pregnancies, we also found an inverse correlation between maternal glucose control and fetal cardiac systolic function but this association was not seen in the post-bariatric cohort.The extent to which these differences impact the future metabolic health of the offspring remains to be determined.

AU T HOR C ON T R I BU T ION S
The study was conceived and planned by MS, MB and JSC.The study was carried out by OP, MB, TM, DP and JSC.Analysis of the data was performed by OP, MB, MS and JSC.The manuscript was written by MS and JSC and all authors approved the paper.

AC K NO W L E D GE M E N T S
The authors would like to thank Miriam Bourke and Allessandra Rocco for their assistance in the research clinic and all the women taking part in the study.The authors would also like to thank the statistician, Dr Sebastian Nastuta, for his statistical advice and input.

F U N DI NG I N FOR M AT ION
The study was supported by CW+ (CWPLUS Registered Charity No 1169897).The funding sources had no role in the study design, interpretation of data, writing of the report and the decision to submit the article for publication.

E T H IC S A PPROVA L
The study was approved by the Local REC (No: 14/LO/0592date of approval: 3 June 2014) and all women gave written, informed consent.

T
A B L E 1 Maternal, pregnancy and offspring characteristics of the study participants.Maternal and neonatal characteristics No-bariatric cohort, n = 25 Post-bariatric cohort, n = 29 p value Maternal age (years)

T A B L E 2
Fetal and infant parameters (unadjusted values) of cardiovascular function in the study groups.
None declared.DATA AVA I L A BI L I T Y S TAT E M E N TData are available from the authors upon request.
Data are given as mean ± SD, median (interquartile range) or n (%), as indicated.Significant p values are denoted in bold. Note: Data are given as mean ± SD and median (interquartile range).Values are normalised by ventricular length or cardiac cycle length as appropriate.The wall thickness values are normalised by end-diastolic ventricular dimensions.Abbreviations: A, atrial contraction diastolic peak velocity derived by pulsed wave Doppler; A′, atrial contraction diastolic peak tissue velocity derived by TDI; CCO/CCI, combined CO/CI; CI, cardiac index; CO, cardiac output; E, early diastolic peak velocity derived by pulsed wave Doppler; E′, early diastolic peak tissue velocity derived by TDI; IVS, interventricular septum; MAPSE/TAPSE/SAPSE, mitral/tricuspid/septal annular plane systolic excursion; MPI′, myocardial performance index; RV/LV, right/left ventricular.
Note: a Comparisons were made between no-bariatric and post-bariatric fetuses and infants.T A B L E 2 (Continued)