Maternal-placental syndrome and future risk of accelerated cardiovascular events in Parous Swedish women with systemic lupus erythematosus — a population-based retrospective cohort study with time-to-event analysis

Abstract

Objectives. Women with SLE are at increased risk of cardiovascular events (CVEs), but a relationship with traditional cardiovascular and SLE-specific risk factors has not been established. In unselected populations, adverse pregnancy outcomes linked to maternal-placental syndrome (MPS) are associated with an increased risk of CVEs. However, the effect of MPS on CVEs is unknown in women with SLE. The aim of this study was to determine if MPS increased the risk and accelerated the development of CVEs in women with SLE.

Methods. Utilizing Swedish population registries, parous women with SLE were identified. Exposures were the following: MPS defined as hypertensive disorders of pregnancy; small-for-gestational-age; placental abruption and stillbirth; and preterm delivery <34 weeks. Outcomes were CVE encompassing cardiovascular morbidity and mortality. Risk of an event was modelled using Cox proportional hazards adjusted for year of delivery, age at CVE, severity of SLE and cardiovascular risk factors. Time-to-CVE was estimated using Kaplan–Meier methods.

Results. Over the 38-year study period, there were 3977 women with 7410 pregnancies, of whom 413 (10.2%) suffered a CVE. Hazard of CVE was higher in those with MPS, particularly when MPS (adjusted HR = 1.64; 95% CI: 1.31, 2.05) was combined with preterm delivery < 34 weeks’ gestation (adjusted HR 1.99; 95% CI 1.39, 2.84). There was accelerated development of CVEs in women with MPS of 10.5% ( vs 7.3% in uncomplicated pregnancies) over the 38-year interval (P < 0.05).

Conclusion. Pregnancy complicated by MPS and preterm delivery exerts an independent effect to increase the risk and accelerate the development of CVEs in parous women with SLE.

Rheumatology key messages

• Women with SLE who had adverse pregnancy outcomes developed cardiovascular events earlier.

• Delivery <34 weeks and maternal-placental syndrome increased the woman’s women’s risk of cardiovascular events 2-fold.

  Adverse pregnancy outcomes could be included when screening for cardiovascular disease in women with SLE.

Introduction

Prevalence of cardiovascular events (CVEs) in North American and international SLE cohorts followed longitudinally is ∼10% [ 1 , 2 ]. There is a 5-fold increase in the risk of premature cardiovascular disease, which is the major cause of morbidity and mortality in patients with SLE [ 1 , 3 , 4 ]. This phenomenon is especially pronounced in younger women [ 4 ]. Using the Nationwide Inpatient Sample—the largest all-payer inpatient care database in the USA—and matching for gender and ethnicity, researchers found the mean age of CVEs requiring hospitalization in women with SLE was 10.5 years younger compared with women without SLE [ 5 ]. The finding that young women with SLE appear to be particularly vulnerable to accelerated vascular events raises the possibility that pregnancy may play a role in the accelerated development of cardiovascular disease in women with SLE.

The relationship of pre-eclampsia to future cardiovascular disease has been well established [ 6 ]. Women who develop pre-eclampsia during pregnancy are four times more likely to develop hypertension later in life, and are twice as likely to develop heart disease, stroke and thrombosis in the future [ 6 , 7 ]. The development of pre-eclampsia is one of the earliest clinically identifiable markers of a woman’s heightened risk of cardiovascular morbidity and mortality [ 8 ]. Cardiovascular risk is also increased significantly in women with a previous history of preterm delivery, stillbirth or fetal growth restriction, which are also related to placental disease. The Cardiovascular Health After Maternal Placental Syndrome study reported that the highest risk of future cardiovascular disease occurred in women with an intrauterine death [ 9 ].

Women with SLE have increased risk of adverse pregnancy outcomes [ 10 ]. They have a much higher risk of developing pre-eclampsia (14–23%) and eclampsia, and preterm delivery occurs in up to 20–31% of women [ 11–13 ]. Fetal growth restriction rates range from 5 to 23% in women with SLE [ 13 , 14 ]. These conditions are collectively referred to as maternal-placental syndrome (MPS). Flares of SLE are associated with poorer obstetric outcomes [ 15 ]. As women with SLE often have pregnancies complicated by MPS, it is possible that the high prevalence of premature cardiovascular morbidity and mortality seen in this cohort without traditional cardiovascular risk factors could be associated with MPS in pregnancy.

Our previous work has demonstrated that women with SLE and a history of MPS have a 2-fold greater risk of death from primary cardiovascular causes; this risk was over 3-fold when combined with preterm delivery <34 weeks’ gestation (odds ratio 3.6; 95% CI: 1.6, 8.1) [ 16 ]. What remained unknown was whether MPS and preterm delivery <34 weeks’ gestation was associated with the accelerated development of all CVEs in parous women with SLE.

Methods

This was a population-based retrospective cohort study of women with SLE in Sweden between 1973 and 2011. The Regional Board of Ethics in Stockholm approved this study and granted access to the population databases. Women with SLE were identified from the National Patient Register (NPR) and the Hospital-based Outpatient Care Register. These databases were then linked using the personal identity number to the Swedish Medical Birth Register (MBR) and Cause of Death Register to obtain information on pregnancies and cause of death during the period of study [ 17 ]. The duration of SLE was calculated from the date of the first documented inpatient admission for SLE (or with SLE as one of the comorbid conditions listed).

Since the 1960s the NPR has been collecting information on all inpatients at public hospitals in Sweden. Compulsory reporting of all inpatient admissions was established from 1987, with >99% of all hospital discharges now included in the NPR [ 18 ]. Diagnoses made during each admission were coded using the Swedish international classification of diseases (ICD) system based on the WHO ICD classification [ 18 , 19 ]. All procedures were coded using the Swedish version of the NOMESCO classification of surgical procedures [ 18 ]. The ICD codes used during the study period ranged between ICD 8 and 10. The recording of diagnostic codes is well validated, and specific examples include a >90% positive predictive value for coronary artery disease and >85% positive predictive value for RA [ 18 ]. In 2001, the Hospital-based Outpatient Care Register was established, but its coverage was only 80% of all outpatient attendees [ 18 ]. The MBR, founded in 1973, includes data on 97.0–99.5% of all births in Sweden (from 1973 to 2007, births >28 weeks, and after 2007, all births from 22 weeks) [ 20 ]. The Cause of Death Register contains information on all deaths of Swedish nationals irrespective of whether the death occurred in Sweden or elsewhere. This register does not include data on visitors, asylum seekers or individuals who have yet to obtain residence permits, or those who are no longer registered in Sweden ( http://www.socialstyrelsen.se/statistics ).

Study cohort

The study cohort included women with SLE identified from the NPR between 1 January 1973 and 31 December 2010 with a previous pregnancy registered in the MBR. All co-morbidities listed in this cohort were based on information captured during inpatient admissions utilizing ICD codes.

The outcomes of interest were CVEs defined as presence of any cardiovascular disease (i.e. coronary artery disease, stroke and peripheral vascular disease) or death from cardiovascular causes. Data on cardiovascular disease were captured in NPR using ICD codes ( supplementary Table S1 , available at Rheumatology Online). Death from cardiovascular causes was recorded in the Cause of Death Register.

The exposure of interest was the history of any features of MPS in any pregnancy. In this study MPS was defined as any hypertensive disorders in pregnancy, intrauterine death, stillbirth, placental abruption or the delivery of a small-for-gestational-age (SGA) infant. SGA was coded independently in the MBR and is defined as a birthweight of >2 s.d. below the mean weight for gestational age, according to the Swedish sex-specific fetal growth curve [ 21 ]. The other features were identified through ICD codes ( supplementary Table S1 , available at Rheumatology Online). In addition, we also explored the effect of preterm delivery <34 weeks in isolation and in combination with features of MPS on the risk of CVEs.

Statistical analyses

Continuous variables were summarized using medians and interquartile ranges and the Mann–Whitney U -test for comparisons between groups. The chi-square test was used for univariate comparisons of dichotomous data. Multivariable analysis of survival was performed using Cox proportional-hazards models adjusting for the year of first delivery to account for cohort effect, year of maternal birth, number of inpatient admissions, cancer and infection as a measure of the severity of SLE, and cardiovascular risk factors (inclusive of renal disease, diabetes and hypertension). The effects of preterm delivery <34 weeks on future CVEs were explored independently and in combination with MPS. An interaction (Wald) test was performed to determine the effect of maternal SLE at the time of pregnancy on the association between MPS and future cardiovascular morbidity and mortality. The cumulative probabilities of survival in parous women with and without MPS and those with a delivery <34 weeks’ gestation were assessed using Kaplan–Meier survival curves and compared by the log-rank test. Women who developed a CVE prior to pregnancy were excluded from the analysis.

The data obtained from the Swedish registries were comprehensive, but a few variables that were not available throughout the study interval—for example maternal height, weight and nicotine use were recorded from 1982—were excluded from the survival analysis. All P-values were two-sided and the significance was set at P < 0.05. Analyses were performed with Stata IC version 13.1 (StataCorp LP, College Station, TX, USA).

Results

Between 1973 and 2011 there were 12 702 women in Sweden with SLE identified from the NPR. In this cohort there were 3977 women with 7410 pregnancies. All descriptive and analytical data relate to women with history of pregnancy (n = 3977). Of these, there were 413 (10.3%) women with subsequent CVEs. At the conclusion of the period of observation, the median age of the women with any CVE was 50 years (IQR 41–57). A total of 105 women (25.4%) had died from cardiovascular causes with the median age of death at 54 years of age. Coronary artery disease was the most common cause of death (69.6%). In those who did not die of cardiovascular causes, 7.7% (n = 308) had some form of cardiovascular morbidity at a median age of 50 years. Again coronary artery disease was the most common cardiovascular morbidity (51.0%) ( Table 1 ).

T able 1 .

Cardiovascular morbidity and mortality in parous women with SLE in Sweden (n = 413)

All CVE	n (%)	Median age (IQR), years
Cardiovascular disease (in those without cardiovascular deaths)
Composite cardiovascular morbidity	308 (74.58)	50 (41–57)
Coronary artery disease	157 (50.97)	52 (46–59)
Stroke	126 (40.91)	47 (39–54)
Peripheral vascular disease	60 (19.48)	44 (37–52)
Death from cardiovascular causes
Composite cardiovascular mortality	105 (25.42)	54 (47–60)
Coronary artery disease	71 (69.62)	56 (48–62)
Stroke	31 (29.52)	51 (42–56)
Peripheral vascular disease	8 (07.62)	55 (45–61)

CVE: cardiovascular event.

Those with CVEs had a longer duration of SLE diagnosis, had more inpatient admissions and medical comorbidities (i.e. renal disease, diabetes, hypertension, cancer and infections) and were also more likely to die (both from cardiovascular and non-cardiovascular causes). In pregnancy, these women were more likely to smoke, have higher stillbirth rates, have a higher prevalence of MPS (28.5% vs 21.4%, P = 0.001) and have an SGA infant and stillbirth ( Table 2 ).

T able 2 .

Characteristics of the parous women with SLE who have had cardiovascular events

	Cardiovascular events (n = 413; 10.38%)		No cardiovascular events (n = 3564; 89.62%)	P-value
Characteristics of these parous women at the end of the study (31 December 2010)
Duration of SLE, median (IQR), years	13 (7–23)		9 (5–15)	<0.001
Number of inpatient admissions, median (IQR)	11 (4–32)		2 (0–5)	<0.001
Current age of those alive, median (IQR), years	57 (49–63)		50 (41–58)	<0.001
Dead, n (%)	135 (32.69)		190 (5.33)	<0.001
Age of death, median (IQR), years	54 (47–60)		50 (43–58)	0.036
Medical comorbidities accumulated by the end of the study interval
Renal disease, n (%)	131 (31.72)		454 (12.74)	<0.001
Diabetes, n (%)	50 (12.11)		125 (3.51)	<0.001
Hypertension, n (%)	155 (37.53)		335 (9.40)	<0.001
Cancer, n (%)	86 (20.82)		271 (7.60)	<0.001
Infection, n (%)	225 (54.48)		718 (20.15)	<0.001
Pregnancy history—characteristics of the women during their pregnancy
Number of pregnancies, n		679	6731
Year of first birth, n (%)				<0.001
1973–86		340 (82.32)	1901 (53.34)
1987–99		60 (14.53)	1060 (29.74)
2000–11		13 (3.15)	603 (16.92)
Parity, median (IQR), n		2 (1–3)	2 (1–2)	0.009
Maternal age at first birth, median (IQR), years		27 (23–31)	27 (23–30)	0.30
History of maternal SLE at the time pregnancy, n (%)		108 (26.15)	1161 (32.58)	0.008
Maternal obesity a , n (%)				<0.001
Not obese		193 (58.48)	3242 (64.48)	0.002
Obese		8 (2.42)	281 (5.59)	0.002
Missing		129 (39.09)	1505 (29.93)	0.039
Maternal smoking in pregnancy a , n (%)				<0.001
Non-smokers		161 (59.41)	3417 (77.66)	<0.001
Smokers		110 (40.59)	983 (22.34)	<0.001
Missing		59 (17.88)	628 (12.49)	0.30
Maternal diabetes in pregnancy, n (%)		17 (4.12)	115 (3.23)	0.34
Livebirth, n (%)		661 (97.35)	6650 (98.80)	0.002
Gestation for live births, median (IQR), weeks		39 (38–40)	39 (38–40)	0.019
Birthweight b , median (IQR), g		3230 (2720–3640)	3355 (2960–3730)	0.014
Preterm deliveries <34 weeks b , n (%)		49 (7.41)	284 (4.27)	0.019
Neonatal survival at 1 month b , n (%)		657 (99.39)	6612 (99.43)	0.96
Features of MPS in any of the pregnancies
History of any MPS, n (%)	117 (28.33)		762 (21.38)	0.001
History of hypertensive disorders in pregnancy, n (%)	51 (12.35)		350 (9.82)	0.11
History of small for gestational age b , n (%)	64 (16.49)		429 (12.4)	0.022
History of stillbirth, n (%)	18 (4.36)		76 (2.13)	0.005
History of placental abruption, n (%)	5 (1.21)		36 (1.01)	0.61

^a Data only available from 1982.

^b Only for livebirths. n: number.

The study duration was 38 years with total person-time for follow-up of 147 246.9 years with a CVE incidence rate in those with a history of MPS complicating their pregnancies of 3.5 per thousand (95% CI: 2.9, 4.3) compared with 2.5 per thousand (95% CI: 2.3, 2.8) (P < 0.0001) for those without MPS. In the univariate analysis, history of MPS was associated with 1.4-fold hazard of CVE, and this finding persisted after adjustment (Adj HR = 1.6; 95% CI: 1.3, 2.1) ( Table 3 ). Preterm delivery <34 weeks’ gestation was independently associated with a 1.8-fold increased risk of CVE (adjusted HR = 1.80; 95% CI: 1.33, 2.43). When MPS was combined with preterm delivery <34 weeks’ gestation, the hazard was 2-fold (Adj HR = 2.0; 95% CI: 1.4, 2.8). These associations were independent of the presence of maternal SLE at the time of pregnancy (P > 0.05).

T able 3 .

Adjusted and unadjusted hazard ratios for cardiovascular events in parous women with SLE

	HR; 95% CIs for CVE
History of MPS or its features in any pregnancy	Unadjusted HR	Adjusted HR for year delivered	Adjusted HR for year delivered + age + SLE-related comorbidities + cardiovascular risk factors
Any MPS in any pregnancy	1.43 (1.15, 1.78)	1.65 (1.33, 2.06)	1.64 (1.31, 2.05)
Any hypertensive disorders in pregnancy	1.31 (0.98, 1.76)	1.82 (1.35, 2.44)	1.68 (1.24, 2.26)
SGA	1.50 (1.17, 1.94)	1.63 (1.27, 2.11)	1.68 (1.30, 2.17)
Stillbirth	2.12 (1.32, 3.40)	2.24 (1.40, 3.60)	2.02 (1.26, 3.25)
Placental abruption	1.26 (0.52, 3.06)	1.17 (0.49, 2.84)	1.16 (0.498, 2.81)
Preterm delivery <34 weeks	1.72 (1.28, 2.31)	2.18 (1.63, 2.94)	1.80 (1.33, 2.43)
MPS and preterm delivery <34 weeks	1.88 (1.33, 2.66)	2.69 (1.89, 3.81)	1.99 (1.39, 2.84)

CVE: cardiovascular event; HR: hazard ratio; SGA: small for gestational age.

CVE-free survival was longest amongst the parous women with SLE who had uncomplicated pregnancies ( Fig. 1 ). Log-rank tests for equality of survival function confirmed that those with MPS or MPS and preterm delivery <34 weeks had more CVEs compared with those who had uncomplicated pregnancies (P < 0.05).

F ig . 1 .

Kaplan–Meier survival estimates for CVE in parous women with SLE

CVE: cardiovascular event.

Discussion

This study demonstrates that there was an increased risk and accelerated development of CVEs in parous women with SLE with a history of MPS, particularly when combined with preterm delivery <34 weeks, compared with those with uncomplicated pregnancies. The risk of future CVEs was increased by 2-fold when MPS was combined with <34-week delivery. This association occurred independent of any cohort effect, duration and severity of SLE and cardiovascular risk factors (diabetes, hypertension and renal disease).

The contribution of traditional cardiovascular risk factors (i.e. age, diabetes, hypertension, smoking, etc.) to the development of CVEs in women with SLE has not been consistently demonstrated [ 22 ]. Large multi-national cohorts of patients with SLE have failed to identify specific risk factors—either cardiovascular or disease-specific—that lead to the accelerated development of CVEs, particularly in women [ 23 ]. Unlike male SLE patients, CVD in women with SLE appears to occur independent of traditional cardiovascular risk factors and disease-specific factors [ 2 , 24 ]. More recently, researchers have begun to explore the relationship between pregnancy and CVEs in the SLE population [ 25 ]. Wu et al. performed a population-based retrospective study of 1 132 089 parturients from 1999 to 2003 to determine whether successful delivery was associated with CVEs in young women with or without SLE. The researchers found that the women with SLE had the highest risk of major CVEs including myocardial infarction. However, unlike our study, which had 38 years of follow-up, the duration of follow-up by Wu et al. was only 4.4 years, with only 12 major adverse CVEs in the lupus cohort. They adjusted for maternal education, age, marital status, parity, multiple pregnancy, diabetes and hypertension, but did not adjust for any lupus-specific confounders.

One of the strengths of our study includes the population-based retrospective design over 38 years. The data collected were complete (>99% of all inpatients and deliveries) and well validated [ 18 ]. A long duration of follow-up is necessary for the development of CVEs in this relatively young cohort of women. Nevertheless, our study clearly demonstrates that accelerated development of CVEs occurs prematurely in the women who have had pregnancies complicated by MPS. As the cohort is predominantly identified from the NPR, which required an inpatient diagnosis of SLE, there may be a bias towards patients with more severe manifestations of SLE. From the data available, we were able to stratify the women with more severe SLE by using duration of SLE and number of inpatient admissions as a surrogate marker for disease severity. Even after adjusting for these disease-specific factors and for traditional cardiovascular risk factors, the association between MPS and an increased risk for future CVEs persisted. Limitations of our study included the lack of information on pharmacological intervention—particularly cumulative steroid and HCQ use—presence of APAs and history of smoking beyond pregnancy. The NPR has limited information on cardiovascular risk factors and past papers have highlighted the need for it to be combined with outpatient data [ 26 ].

The pathophysiological process by which MPS leads to future cardiovascular disease is unclear. Researchers have postulated that there are shared pathological pathways between pre-eclampsia and future CVEs that are unmasked by the physiological stress of pregnancy [ 6 ]. A population-based study has found pre-existing cardiovascular risk factors, for example, hypertension and dyslipidaemia, are more prevalent in women who then develop pre-eclampsia in pregnancy [ 27 ]. Pre-eclampsia is a disease of the maternal endothelium; the placenta is one of the most vascular structures [ 28 ]. Hence, the earliest manifestation of any predilection for vascular disease may present in the form of placental insufficiency. This is manifested clinically as pre-eclampsia or hypertensive disease in pregnancy, poor fetal growth (defined as SGA when the baby is delivered), stillbirth and placental abruption—or features of MPS as defined in this study. Preterm delivery <34 weeks is a useful surrogate marker for underlying maternal disease and likely MPS, but can also occur spontaneously, although the number of spontaneous preterm deliveries <34 weeks is likely to be quite small (<3%) in this study cohort.

Placental abruption is rare and is a difficult clinical entity to define. This may account for the small numbers reported in this study and hence the lower utility as marker for future CVEs. There also appears to be a close interaction between poor fetal growth and future maternal cardiovascular disease. Severe SGA (defined as an expected birthweight for gestational age below the 0.75 centile, or ∼2 s.d. below the mean) infants at term have been linked to 1.4-fold increase in cardiovascular disease in the mother (1.38, 95% CI: 1.15, 1.65) with increasing risk as length of gestation decreases [ 29 ].

A recently published population-based study covering a 25-year period has demonstrated that stillbirth is a risk factor for CVEs and cardiovascular admissions [ 30 ]. The Cardiovascular Health After Maternal Placental Syndrome study has linked stillbirth with a 4-fold increased risk for cardiovascular disease—the highest risk amongst all other features of MPS [ 9 ]. Stillbirth is a robust outcome that is well captured in the Swedish MBR. When used in combination with preterm deliveries <34 weeks, we believe this provides a particularly robust marker for clinicians to identify women at risk for future CVEs. The caveat is that not all stillbirths are related to MPS, as they can occur for a variety of other reasons, for example, fetal anomalies and poorly controlled diabetes.

Preterm delivery on the other hand is more common and well captured in the MBR. We selected <34 weeks to define our preterm deliveries as the incidence of iatrogenic preterm deliveries is increasing, but the decision to deliver <34 weeks is usually driven by clinical concern regarding either poor maternal health (reasons for which may not be adequately captured in the MBR) or fetal wellbeing—which is usually a result of placental insufficiency. Hence, deliveries <34 weeks are likely a surrogate marker for MPS or more severe maternal disease. Preterm deliveries per se have also been linked to future maternal cardiovascular disease, particularly if recurrent [ 31–33 ]. A recent systematic review estimates the risk of ischaemic heart disease to be 1.3- to 2.1-fold, stroke 1.7-fold increased in women who have had preterm deliveries [ 34 ]. The subclinical risk for vascular disease is measurable within 12 years of a preterm delivery with more women with preterm births having elevated blood pressure, atherogenic lipid profiles and evidence of carotid intimal media thickening [ 35 ].

Most of the features of MPS are linked to inflammatory processes resulting in maternal vascular endothelial dysfunction, particularly in the placenta [ 36 , 37 ]. Inflammatory molecules within the maternal circulation during a pregnancy complicated by MPS can persist for many months after delivery with potential for permanent vascular damage thereby accelerating atherosclerosis. In particular, soluble fms-like tyrosine kinase-1—a response to uterine hypoxia—is present 5 weeks prior to any clinical manifestation of pre-eclampsia and can persist in the maternal circulation for 18 months afterwards [ 36 , 38–40 ]. It is also present in women with SLE who develop pre-eclampsia [ 41–43 ].

In the present study, only 32% of women had SLE documented at the time of their pregnancy. Maternal SLE at the time of pregnancy did not affect the association between MPS and future CVEs. This could be, in part, because diagnosis was not captured via the NPR as they had predominantly outpatient treatment, or diagnosis was not recorded on the MBR when they were pregnant. On the other hand, a diagnostic lag-time for SLE is well documented; many women may exhibit lupus-related pregnancy complications prior to a formal diagnosis [ 44 ]. APS has been shown to contribute to CVEs in some studies, and also plays a significant role in MPS-related adverse pregnancy outcomes seen in women with SLE [ 45 , 46 ]. Hence, it is very likely that our data have included women with underlying APS. Specific data on socioeconomic status were only available from after 2001. However, Sweden is a relatively homogeneous society with equitable access to healthcare.

While we cannot infer causality from these data, there appears to be a close association between MPS, particularly when combined with preterm delivery <34 weeks’ gestation, and accelerated development of CVEs in parous women with SLE. Preterm delivery <34 weeks’ gestation is also a useful and independent marker of CVEs in this group. It is particularly useful as a screening tool as most women will be able to recollect the gestation their infant was delivered, but not necessarily details of all the complications encountered in the pregnancy. Hence, clinicians caring for these young women with SLE may wish to consider screening for cardiovascular disease earlier in women who have had a history of MPS and/or a preterm delivery <34 weeks’ gestation