An ambispective cohort study was conducted in adults born between 1975 and 1995 from a birth registry at a tertiary university hospital in Barcelona, Spain. SGA was defined as birth weight <10th centile for gestational age. Unexposed individuals with appropriate fetal growth for gestational age (AGA) were defined as birth weight ≥10th centile. Two different birthweight standards were used: local standards from 1975 to 1995 (22) (contemporary but limited as not customized by sex and with small sample sizes at extreme gestational ages), and current local standards customized by gestational age and sex (23). Participants with discordant birthweight centile results on the two standards, congenital malformations, structural or chromosomal abnormalities, twin pregnancies, professional athletes, major mental disorder, and those with birthweight >95th centile were excluded. To prevent the inclusion of constitutionally small cases, those born from mothers with a height <1.46 m were also excluded. Biological sex was obtained from the birth register. Self-reported gender coincided with birth recorded biological sex in all participants. The study was approved by the Ethical Committees of Hospital San Joan de Déu and Hospital Clinic in Barcelona, Spain.

The birth register contained birthdate, demographic, pregnancy, and perinatal information. Deliveries were recorded following birthdate order and each birth was identified by a consecutive number throughout the years. First, a random selection was performed using this identification number. Then, the birth weight percentile was determined, and subjects were classified as SGA or AGA. As this selection generated more AGA than SGA, additional SGA cases were randomly selected from among those that were born within the same year as those born AGA. Next, each child's surname was obtained from mother's clinical record. Based on birthdate and surname, the participants were located in the local register of the relevant health affiliation and contact data obtained. Thereafter, a letter of invitation was sent to the participants and they were subsequently contacted via phone calls. If the individual was interested in participating, inclusion and exclusion criteria were verified. Then, a detailed explanation of the study protocol was provided and informed consent signed.

Sample size was calculated assuming an unknown but equal variance [as suggested by previous studies (12, 24)], 80% power, 5% alpha-error, 1:1 allocation index. A standardized mean difference of 0.25 was used to maximize the sample size. Then, a sample size of 250 SGA and 250 AGA individuals was estimated for echocardiography. Assuming an overall response and recruitment rate of ∼25%, 2,000 patients were initially selected. The stratified analysis was unplanned, therefore, the width and precision of the 95% confidence interval (95% CI) of measures of associations were considered for the interpretation of the data (25, 26).

The study protocol included echocardiography, electrocardiogram (ECG), measurement of blood pressure, vascular ultrasound, medical examination including a questionnaire, anthropometry, and collection of peripheral blood samples. These measurements were obtained by a trained physicians and cardiologists blinded to the individual's SGA categories. In addition, covariables were obtained via participant interviews and reviewing birth medical records.

Perinatal data was obtained from the delivery register. Hypertensive disorder in pregnancy included preeclampsia/eclampsia or pregnancy hypertension according to the birth register. Current characteristics including sociodemographic and personal antecedents were obtained by a structured questionnaire. Family history of cardiovascular disease was self-reported and was considered as positive whether any of the participant's parents suffered from myocardial infarction, hypercholesterolemia, diabetes, or hypertension. Physical activity was assessed by the international physical activity questionnaire (short version) (29). Physical activity was classified as: high (vigorous activity on at least 3 days achieving a minimum total physical activity of at least 1,500 metabolic equivalent of task (MET)-minutes/week, or five or more days of any combination of walking or moderate or vigorous activities that achieve at least 3,000 MET-minutes/week), moderate (at least 20 min of vigorous activity per day for three or more days per week, or at least 30 min of moderate intensity activity per day for five or more days per week, or five or more days of any combination of walking, moderate-intensity, or vigorous intensity activities achieving at least 600 MET minutes/week). When the participants did not meet the above criteria, they were classified as sedentary.

Stata 14.0 (StataCorp. LP. College Station. TX) was used for statistical analysis. We first determined the difference between AGA and SGA in the overall population and a stratified analysis by sex was performed similarly. For quantitative variables, normality assumption was checked using the Shapiro–Wilk test. Then, study groups were described using mean (standard deviation) or median (interquartile range). Student's t-test or the Mann–Whitney test were used to compare the mean/median, respectively. The X² test and the Fisher's exact test were used as appropriate for comparison of categorical variables. A P value <0.05 was accepted as statistically significant. Kernel-weighted local polynomial regression was used to explore the relationship between heart rate and blood pressure by categories of sex and SGA background.

For each variable, an inverse probability weighting (IPW) was used with the aim of achieving a balanced distribution of confounders across exposure groups and simulating a randomized trial. Weights were assigned to individuals based on the inverse of their probability of being SGA, using a propensity score probit model. Variables that are both related to the exposure and to the exposure–outcome relationship were included in the model. This results in a pseudo-population in which patients with a high probability of being exposed have a smaller weight and patients with a low probability of exposure have a larger weight and thus the distribution of measured patient characteristics used to calculate the propensity score becomes independent of the exposure.

The average exposure effect with robust standard errors was then estimated and called the adjusted absolute difference (AAD), i.e., the average effect of moving an entire population from AGA to SGA (30). Standardized mean differences (SMD) and the variance ratio were used to verify the baseline covariates balance between SGA and AGA in the overall and stratified analyses. As a rule of thumb, the SMD should be lower than 0.1 and the variance ratio close to 1. Overall balance was also inspected by the overidentification test (31, 32).

The flow diagram of the study population is shown in Figure 1. In total, 262 AGA and 261 SGA individuals were included in the analyses. The proportions of females were 58.62% and 45.80% among the SGA and AGA groups, respectively (P = 0.003). In all participants, biological sex assigned at birth coincided with self-reported gender at the time of the assessment. By definition, birthweight and birthweight centile were lower in the SGA group. The prevalence of preeclampsia during pregnancy and asthma was higher among those SGA. Higher prevalence of asthma was observed among the SGA group when compared to the AGA group (Table 1 and Supplementary Table SM1), with this difference more pronounced in males. The group balance obtained using IPW was adequate in both overall and stratified analyses (Supplementary Table SM2). The overall differences between the SGA and AGA groups regarding cardiovascular endpoints are shown in the supplemental material (Supplementary Tables SM3–SM6) and the results of stratified analyses are described below.

After indexing for BSA, most LV and RV dimensions were significantly increased in SGA individuals, particularly among females (Table 2). Ventricular sphericity and relative wall thickness were similar among groups. Female born SGA showed a significant reduction in ejection fraction, compensated by increased heart rate and with preserved cardiac output. Cardiac function in male born SGA was mainly preserved except for reduced mitral displacement compared to those born with AGA. A sensitivity analysis excluding the eight female of the study group (seven from the SGA group and one from the AGA group) with reported preeclampsia during pregnancy of their own offspring showed the same direction and significance of the previously reported associations (data not shown).

A positive correlation between blood pressure and heart rate was observed if heart rate values were above 60 bpm (Figure 2). Regarding the ECG results, a reduction in the duration of the P and QRS waves and PR interval were observed in the SGA group, with more pronounced changes in males (Table 3). There was a non-significant trend to higher values of blood pressure in male born SGA compared to those born with AGA (Table 4). We found no differences between males and females regarding the effect of SGA on cIMT.

As expected, SGA individuals had lower height and weight (Table 5). While body mass index was similar among the study populations, both females and males born SGA showed a significant increase in waist-to-hip ratio. SGA males also had higher central-to-peripheral fat ratios and fat mass percentages. Laboratory results on glycemia and lipids profile were similar among the study groups; a slight increment in triglyceride values was observed in SGA males.

Our echocardiographic data is in line with previous reports suggesting subtle cardiac structural and functional changes in adults related to fetal growth (13, 33). Both males and females born SGA showed subtle increases in indexed ventricular dimensions, with changes being more pronounced in female. In addition, we first identified SGA female to be more susceptible to cardiac dysfunction. Ejection fraction was preserved (>40%) in all participants; however, females born SGA showed a reduction in ejection fraction compensated by a heart rate increase. These changes were less pronounced and not statistically significant in males. While systolic dysfunction has been described to be more associated with male sex (34), a previous study in older patients with preserved ejection fraction showed more frequent diastolic dysfunction and readmission for heart failure in female than male (35). These findings are also consistent with female being more prone to heart failure (5).

There is controversy regarding the sex differences in the pattern of cardiac remodeling. For example, preclinical echocardiographic changes are mainly characterized by a dilated left atrium in female and by left ventricular hypertrophy in male (36); however, it has also been suggested that hypertrophy is more common and less modifiable in female (37). This discordance could be explained by differences with the study population, as in most published reports female are aged >40 years. Likewise, an animal model of hypoxia demonstrated that males had smaller body weights and signs of left ventricular hypertrophy, but both males and females showed signs of left ventricular diastolic dysfunction by 12 months of age (21). Animal models of fetal programming (38–40) have suggested a protective role of estrogens in females exposed to fetal undernutrition, which was not the case for some cardiac functional parameters in our study.

In addition to previous knowledge, SGA female presented the highest heart rate at rest, which is an independent risk factor for CVD (41). A similar response was reported for occupational stress, where female individuals have a larger percent change in heart rate compared with males (42). These results are in line with previous findings suggesting lower baseline heart rates in male from birth onward (43), even after stratification by race (44). Our sensitivity analysis suggested that pregnancy-related disorders (45) did not account for the observed sex differences in systolic function, nor were the differences explained by the severity of the restriction, as the differences in birthweight and gestational age between SGA and AGA were similar in male and female.

We previously described the presence of electrical remodeling in SGA and preterm preadolescents (46) overall. The present study in adulthood further identifies male born SGA being more susceptible to short PR and QRS intervals. The absolute difference of −6.61 in male indicates that, on average, being SGA leads to a decrease in PR interval of about 7 ms. Our results are in line with previous literature reporting more prevalent ventricular pre-excitation in male than in female (47). However, the pattern seems to be different, as QRS interval was decreased in SGA population (48). It has also been reported that female have smaller calcium currents and lower excitation-contraction coupling gain than male (49).

Despite prolonged PR intervals being a risk factor for atrial fibrillation incidence (50), a U-shaped relationship has been found with an increased risk for individuals with short PR intervals (51), which was the case of our study. In addition, a non-significant increment in QTc, a potential risk factor for ventricular tachycardia and sudden cardiac death (52), was observed among SGA male. Overall, these findings suggest a greater impact of SGA on electrical remodeling among SGA male compared with SGA females.

Even though males have higher blood pressure than age-matched females during early adulthood (53), our study showed those that who were born SGA presented the highest systolic values. While statistically non-significant, the average systolic blood pressure in male SGA was 3 mmHg higher than AGA, with a mean absolute difference of 2.71. This difference seems to be clinically relevant considering that every 2 mmHg increases the risk of fatal stroke and fatal coronary heart disease by 7% and 5%, respectively (54). Human data on hypertension from prenatal origins are controversial. Similar differences in the effect of a birth condition have been reported among adult preterm males (55) and females (56). An increased risk of systolic hypertension without any effect on diastolic blood pressure has been also reported among preterm individuals (57), with an increase in both systolic and diastolic blood pressure in the general adolescent population (58).

Overall, our results reinforce previous findings suggesting that smaller size at birth is associated with higher arterial pressure in males and an increased cardiac sympathetic activation among females (59). Heart rate and blood pressure are both important hemodynamic parameters which are positive correlated with each other in children and adults (60). Our findings suggest that the effect of sex on this correlation seems to be higher than the effect of SGA. However, the direction and strength of the correlation between both parameters might vary whether the heart rate frequency is above or below 60 bpm. Future studies are needed to validate this observation.

These results are consistent with previous papers also suggesting no differences in intima-media thickness in adults born SGA (61). These findings conflict prenatal and pediatric data consistently demonstrating an increased vascular IMT in fetuses, neonates, and children born SGA. This inconsistency may be explained by differences in the study populations (less severe phenotypes in adult cohorts) and the relatively stronger influence of adult lifestyle on IMT.

Last but not least, our adjusted estimates showed a greater reduction in lean mass, an increased fat mass percentage, and altered fat distribution together with higher central adiposity in SGA male. These results are in line with previous reports on preterm males (62, 63). It has been demonstrated that birth size and adult adiposity independently predict events of cardiovascular disease after controlling for potential confounders (64); therefore, both exposures, prematurity and SGA, could play a crucial role in the pathogenesis of CVD in male. Male sex is an independent risk factor for all components of the metabolic syndrome except for low HDL in subjects with a normal body mass index (65) and our study shows a trend toward a greater lipid profile risk among SGA male. Our results are in line with a previous study reporting young female had a higher prevalence of obesity and low High-density lipoprotein cholesterol (HDL-C), while younger male had a higher prevalence of elevated blood pressure and elevated triglycerides (66). Our study additionally supports a sex-specific effect of SGA on cardiovascular health.

The effect of fetal programming in humans has been mainly apprised by traditional risk factors and vascular parameters (14, 15, 39, 67). To our knowledge, this is the first study showing a sex difference on the effect of fetal programming related to SGA and including a wider variety of risk factors in young adulthood (68). We followed current recommendation regarding the need of reporting sex-disaggregated results even when the chance of finding a sex difference is small (69). In our study, females were not under-represented as in most of the heart failure studies (70). In addition, all participants underwent all cardiometabolic tests, overcoming potential sex biases such as patient or clinician awareness.

However, there are some limitations to be acknowledged. Firstly, sex stratification was not considered in the sample size calculation, and therefore, the results should be considered as exploratory. Secondly, measurement bias could be present for some covariates that were self-reported, including physical activity and smoking. SGA can be considered as a “proxy” of fetal growth restriction, as this definition was exclusively percentile-based; at the time the participants were born, the fetoplacental ultrasound-based definition was not available. Thirdly, participants and those who declined to participate were similar regarding birthweight gestational age and birthweight percentile; however, we could not rule out selection bias due to other postnatal risk factors. Fourthly, despite no female reported systemic autoimmune diseases and only one AGA female reported polycystic ovary syndrome, a causal relationship also requires assumptions about the absence of other confounders across the life course like dietary intake, air pollution exposure, psychological stress, childhood obesity, or early menarche, which we could not control for. Currently, more extreme SGA cases might have greater likelihood of survival; this might explain at least in part the greater difference between SGA and AGA of more recent cohorts of birth (12, 24, 71). Finally, the observed changes are within the normal range of the studied parameters.

The underlying mechanisms of sex differences in CVD are still incompletely understood. Potential explanations include the role of risk factors limited to or more frequent in females: pregnancy, preeclampsia, iron deficiency, contraception, hormone replacement therapy, or autoimmune disorders (8). Molecular mechanisms include an activated renin-angiotensin-aldosterone system in response to low estrogen during menopause, the downregulation by estrogen of proteins related to diastolic function, the progesterone activation of the extracellular signal-regulated kinase 2 inducing cardiac hypertrophy during pregnancy, and lower excitation-contraction mediated by calcium and glucose uptake and utilization compared with male (72). An atheroprotective effect of estrogen in young female (73), and the effect of sex steroids on blood pressure-controlling mechanisms have been also postulated (74). Whether these mechanisms also differ by SGA status or whether prenatal epigenetic changes and the reduction in gonadal organ size associated with SGA contribute differently to sex differences in fetal cardiovascular programming need to be elucidated.

There is a need to better study women's hearts. Females are underrepresented across all aspects of cardiovascular research; therefore, they are more likely to be misdiagnosed (37) or to receive less optimized medical treatment (75). Additionally, common risk factors such as tobacco and type-2 diabetes are more harmful to the cardiovascular system of female (76). More recently, female were at increased risk of acute heart failure and in-hospital mortality due to COVID-19 compared with male (77). Taken together, our results reinforce the need for equitable access to cardiovascular prevention services across the life course, and to improve inclusion, accuracy, and reproducibility of women's cardiovascular research. Further studies explore the role of SGA on the higher prevalence of heart failure with preserved ejection fraction among female and sudden cardiac death among male. Last but not least, we observed a higher prevalence of asthma in SGA males, which might require further research.