Department of Obstetrics and Gynecology, Research School GROW, Maastricht University Medical Center, Maastricht, The Netherlands
Correspondence to: Ms N. M. Breetveld, Department of Obstetrics & Gynecology, Department of Obstetrics and Gynecology, P. Debyelaan 25, 6229 HX Maastricht, The Netherlands (e-mail: email@example.com)
After pre-eclampsia (PE), the prevalence of structural heart disease without symptoms, i.e. heart failure Stage B (HF-B), may be as high as one in four women in the first year postpartum. We hypothesize that a significant number of formerly pre-eclamptic women with HF-B postpartum are still in their resolving period and will not have HF-B during follow-up.
In this prospective longitudinal cohort study, we included 69 formerly pre-eclamptic women who underwent serial echocardiographic measurements at 1 and 4 years postpartum. HF-B was diagnosed as left ventricular hypertrophy (left ventricular mass index (LVMi) > 95 g/m2), concentric remodeling (relative wall thickness > 0.42 and LVMi ≤ 95 g/m2), mild systolic dysfunction (left ventricular ejection fraction > 40% and < 55%) or asymptomatic valvular disease. Women were subdivided and analyzed according to HF-B outcome: no HF-B at either visit; HF-B at first visit only; HF-B at second visit only; HF-B at both visits.
The prevalence of HF-B in formerly pre-eclamptic women was 23% in the first year postpartum and 23% after 4 years. At the second visit, HF-B had resolved in 62.5% of affected women but was newly developed in 19% of initially unaffected women. At the first visit, 56% of women diagnosed with HF-B had reduced systolic function whereas at the second visit 69% of women with HF-B had concentric remodeling with mostly normal ejection fraction, consistent with diastolic dysfunction.
Cardiovascular disease (CVD) is the leading cause of death worldwide[1-3]. Although more men are affected by CVD than women, more women die of the disease. Aside from the classical risk factors that contribute to the development of CVD, women also have gender-specific risk factors such as pre-eclampsia (PE), a hypertensive pregnancy syndrome, which increases the risk for CVD developing within 15 years after pregnancy by two- to seven-fold. In contrast to healthy pregnancies in which left ventricular (LV) eccentric hypertrophy occurs, during a pre-eclamptic pregnancy the left ventricle undergoes remodeling and concentric hypertrophy occurs, resembling that seen in heart failure (HF)[7, 8]. The additional PE-induced increase in left ventricular mass (LVM) does not always resolve completely postpartum. In fact, the incidence of preclinical HF Stage B (HF-B) is 24% at 1 year postpartum in formerly pre-eclamptic patients.
HF is becoming a major health problem as life expectancy and prevalence of risk factors rise globally, leading to high healthcare costs, exceeding even those of cancer. The progression of preclinical HF-B to the clinical stage C is associated with a five-fold increase in cardiovascular-related morbidity, mortality and decrease in quality of life[11, 12]. This highlights the importance of identifying patients early in Stage B, which is potentially reversible when treated appropriately.
More women are affected by HF than are men, and therefore sex differences in this condition should be evaluated. Currently, it is not known whether HF-B persists in formerly pre-eclamptic women. On the one hand, if HF-B persists, secondary prevention programs could be implemented. On the other hand, if HF-B resolves spontaneously over time, expectant management seems a reasonable option. In our study, we hypothesize that a significant number of women with HF-B at 1 year postpartum are still in their resolving period and will no longer have HF-B 4 years later. To this end, we performed serial cardiac ultrasound examinations in women with a history of PE at 1 and 4 years postpartum.
This longitudinal cohort study was approved by the medical ethics committee of the Radboud University Medical Center (NL32718.091.10) before patient enrollment and all subjects provided written informed consent before participation. The procedures we followed conformed with institutional guidelines and adhered to the principles of the Declaration of Helsinki and Title 45, U.S. Code of Federal Regulation, Part 46, Protection of Human Subjects, Revised 13 November 2001, effective 13 December 2001. We included women who had a 1-year postpartum clinical cardiovascular risk assessment following a pregnancy complicated by PE and invited them for a second postpartum follow-up screening between 2009 and 2011.
Formerly pre-eclamptic women were recruited by an obstetric clinician at the 6-week-postpartum visit to participate in a 1-year postpartum cardiovascular screening assessment. At 4 years postpartum, women were invited by mail to participate in the cardiovascular follow-up study. The catchment area of this study population was in the east side of The Netherlands in which the socioeconomic status is reported to be average, with a low prevalence of immigrants. For the second visit, we included 69 women without any pre-existing comorbidity (diabetes mellitus, autoimmune disease, or pre-existing hypertension) and with a complete workup at both visits that allowed us to diagnose HF-B at two separate time points following a pregnancy complicated by PE.
PE in the index pregnancy was diagnosed according to the criteria set by the International Society of Hypertension in Pregnancy: new-onset hypertension, systolic blood pressure (SBP) ≥ 140 mmHg and/or diastolic blood pressure (DBP) ≥ 90 mmHg, after 20 weeks' gestation and proteinuria exceeding 0.3 g/day. Early-onset PE was diagnosed as PE developing before 34 weeks' gestation. Preterm PE was defined as PE requiring delivery before 37 weeks' gestation. Four women in the study population gave birth to twins. All birth weights were included in our analysis. A birth weight ≤ 10th percentile was defined as small-for-gestational age. Patients underwent cardiovascular screening and echocardiographic measurements according to a standardized protocol at both postpartum intervals. Most women were of continental European ancestry; two women were of Turkish descent and one was of Moroccan descent. At the time of the first set of measurements, no woman was pregnant, all had stopped breastfeeding and none was using oral contraceptives. Women who had become pregnant again after their index pregnancy had to be at least 6 months postpartum for their second measurement.
Each cardiovascular screening examination started at 8:00 am in a temperature-controlled room (22 °C), after an overnight fast and was performed following a standardized study protocol by an experienced physician. The cardiovascular screening consisted of determination of body mass index (BMI) by measuring body weight (Seca 888 scale, Hamburg, Germany) and height. After 15 min of rest, SBP, DBP and mean arterial pressure (MAP) were measured for 30 min (at 3-min intervals) with the patient in an upright sitting position, using a semi-automatic oscillometric device (Dinamap Vital Signs Monitor 1846; Critikon, Tampa, FL, USA) with a cuff size appropriate for arm circumference. We used the median of each patient's measurements for statistical analysis. During the measurements, the participants were not allowed to talk and external disturbances were kept to a minimum. Hypertension was defined as SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg and/or if taking antihypertensive medication. Prehypertension was defined as SBP of 120–139 mmHg and/or DBP of 80–89 mmHg. Blood samples were taken by venepuncture at the level of the antecubital vein and analyzed for fasting glucose (mmol/L), fasting insulin (mU/L) and lipids (mmol/L; low-density lipoprotein (LDL), high density lipoprotein (HDL), total cholesterol and triglycerides). The homeostasis model assessment index for insulin resistance (HOMAIR) was calculated by insulin (mU/L) × glucose (mmol/L)/22.5. All participants collected their urine 24 h preceding the measurements. The 24-h urine sample was assayed for albumin and creatinine to determine the (micro)albuminuria corrected for creatinine output (g/mol creatinine) (Aeroset, Abbot Laboratories, Chicago, IL, USA). Participants completed a questionnaire consisting of general history, current medication intake, intoxications (smoking was defined as ≥ one cigarette per day), lifestyle factors and family history of CVD (in first-degree relatives < 60 years old), occurrence of PE in first-degree relatives and gestational age at which participants themselves were born and their birth weight. Preterm birth was defined as a birth before 37 weeks' gestation.
Echocardiographic measurements were performed with a phased-array echocardiographic Doppler system. We performed two-dimensional, M-mode and Doppler echocardiography according to the guidelines of the American Society of Echocardiography (ASE). We measured LV end-diastolic (LVEDd) and LV end-systolic (LVESd) diameters as well as end-diastolic interventricular septum thickness (IVST) and posterior wall thickness (PWT) using M-mode in the parasternal long-axis view. LVM was calculated using the formula 0.8 × (1.04 ((LVEDd + PWT + IVST) − LVEDd)) + 0.6 and indexed for body surface area (BSA; Dubois formula), as recommended by the ASE. The relative wall thickness (RWT) was calculated using the formula (2 × PWT)/LVEDd. LV end-diastolic volume (LVEDV) and LV end-systolic volume (LVESV) were estimated using the Teichholz formula. Left ventricular ejection fraction (LVEF) was calculated by [(LVEDV – LVESV)/LVEDV] × 100. In all cardiac assessments, heart rate (HR) was obtained by taking the reciprocal of the mean of five consecutive R–R intervals on the electrocardiogram multiplied by 60. Stroke volume and cardiac output were indexed for BSA.
Definition of heart failure Stage B
HF was diagnosed according to the guidelines of the American Heart Association. HF-B was defined as the presence of previous myocardial infarction, LV hypertrophy (left ventricular mass index (LVMi) > 95 g/m2), concentric remodeling (RWT > 0.42 and LVMi ≤ 95 g/m2), mildly impaired LVEF (> 40% and < 55%) or asymptomatic valvular disease. We defined asymptomatic valvular disease as mild aortic valve insufficiency or central aortic valve insufficiency. HF with preserved ejection fraction (HFpEF) in this subclinical stage was defined as LVEF ≥ 55% but with the occurrence of one of the other criteria for HF-B.
We performed all statistical analyses using SPSS Statistics version 21.0 (IBM Corp., Armonk, NY, USA). We subdivided the PE group into four subgroups based on the HF-B outcome: (1) formerly pre-eclamptic women with no HF-B at either visit were categorized as HF-B−/−; (2) formerly pre-eclamptic women with HF-B at both visits were categorized as HF-B+/+; (3) formerly pre-eclamptic women with HF-B at the first visit but no HF-B at the second visit were categorized as HF-B+/−; (4) formerly pre-eclamptic women who developed de-novo HF-B between the two visits were categorized as HF-B−/+. We analyzed our data non-parametrically and reported continuous data as median (interquartile range) due to the modest sample size after subdividing the study population. We analyzed the continuous data with Wilcoxon's signed-rank test for intragroup differences and the Mann–Whitney U-test for intergroup differences. Dichotomous data were analyzed using McNemar's test for intragroup differences and Fisher's exact test for intergroup differences and were reported as number (percentage). A P-value of < 0.05 was considered statistically significant. We determined the sample size needed for our study based on previous findings by Melchiorre et al. on the prevalence of altered geometry at 1 year postpartum in formerly preterm and term pre-eclamptic women (30%) compared with controls (6%). The required sample size was calculated with a desired power of 0.95 and a two-sided α of 0.05. A minimum of 64 patients at 1 year and at 4 years postpartum was necessary for determining statistical significance.
Of the 69 formerly pre-eclamptic women included, 16 (23%) had HF-B at the first visit, 1 year postpartum, and 16 (23%) had HF-B at the second visit, 4 years postpartum (Figure 1). Six (37.5%) of the 16 women with HF-B at the first visit sustained this condition, which was observed at the second visit (HF-B+/+), and in 10 (62.5%) of the 16 women HF-B resolved before the second visit (HF-B+/−). Of the 53 patients with no HF-B at the first visit, 43 (81%) preserved normal cardiac function and had a normal echocardiography scan at the second visit (HF-B−/−); however, 10 (19%) developed de-novo HF-B within the subsequent 4 years (HF-B−/+).
Characteristics of the formerly pre-eclamptic women are shown in Table 1. Body weight was comparable between the two visits. The index pregnancy was complicated by early-onset PE in 68% of cases and by preterm PE in 75%. Of the 69 women included, 44 (64%) had a subsequent pregnancy at the time of the second visit of which nine (20%) had recurrent PE. Of the study population, 29% had a first-degree relative who had a pregnancy complicated by PE. Characteristics of the cohort after excluding women with twin pregnancy are shown in Table S1.
Table 1. Characteristics of 69 formerly pre-eclamptic women at 1 and 4 years postpartum
1 year (n = 69)
4 years (n = 69)
Data are given as median (interquartile range), median [range] or n/N (%).
*Time of onset of pre-eclampsia (PE) unknown in one case.
†Pregnancy between index pregnancy and second visit.
‡Pregnancy of which woman in this study was offspring.
BMI, body mass index; CVD, cardiovascular disease; GA, gestational age; IUFD, intrauterine fetal death; LVEF, left ventricular ejection fraction; NA, not applicable; SGA, small-for-gestational age.
Inter- and intragroup differences in HF-B vs no-HF-B
Table 2 shows the cardiac and metabolic characteristics in women with and without HF-B at each follow-up visit. Data excluding women with a twin pregnancy are shown in Table S2. The phenotype of the group of women with HF-B at the first visit differed in some aspects from the phenotype of the group of women with HF-B at the second visit. Concentric remodeling was present in 31% of women at visit 1 compared with 69% at visit 2 (P < 0.05). On the other hand, systolic dysfunction, as indicated by mildly impaired LVEF, was present in 56% at visit 1 compared with 25% at visit 2 (P = 0.15). The prevalence of LV hypertrophy and asymptomatic valvular disease did not differ when comparing data from both visits.
Table 2. Cardiac and metabolic characteristics of heart failure Stage B (HF-B) in 69 formerly pre-eclamptic women categorized according to presence or absence of HF-B at 1 and 4 years postpartum
(n = 16)
(n = 53)
(n = 16)
(n = 53)
Data are given as n (%) or median (interquartile range).
*P < 0.05 compared with HF-B at 1 year postpartum.
BMI, body mass index; DBP, diastolic blood pressure; HDL, high density lipoprotein; HOMAIR, homeostatic model assessment for insulin resistance; LDL, low density lipoprotein; LV, left ventricular; LVEF, left ventricular ejection fraction; MAP, mean arterial pressure; NA, not applicable; SBP, systolic blood pressure.
At visit 1, the HF-B group had comparable fasting glucose, triglycerides, total cholesterol, LDL, HDL and albumin/creatinine ratio levels compared with the no-HF-B group (Table 2). Fasting insulin and HOMA levels were higher in the HF-B group compared with the no-HF-B group, along with higher SBP, DBP, MAP and BMI. At visit 2, no significant differences were seen between the groups with and without HF-B.
Inter- and intragroup differences in HF-B−/−vs HF-B−/+
We first compared HF-B−/− and HF-B−/+ subgroups (Table 3; data excluding women with twin pregnancy are given in Table S3). Baseline characteristics, obstetric characteristics of index pregnancy complicated by PE and offspring pregnancy were comparable between the two groups except for a higher prevalence of primiparous women in the HF-B−/− group compared with the HF-B−/+ group at visit 1 (93% vs 60%). LVMi did not differ between subgroups at visit 1 or 2. RWT did not differ between subgroups at visit 1, but it was higher in the HF-B−/+ group compared with HF-B−/− at visit 2. LVEF did not differ between subgroups at visit 1 or 2. However, in the HF-B−/+ group, LVEF had decreased significantly at visit 2 when compared with visit 1. Stroke volume (SV) and SV index did not differ between groups at visit 1 or 2. However, in the HF-B−/− group, both SV and SV index had increased at visit 2 compared with visit 1, whereas they did not change over time in the HF-B−/+ group. The ratio of early to late ventricular filling velocity (EA ratio) did not differ between the two groups at visit 1 or 2, whereas cardiac output (CO) and CO index were higher in the HF-B−/+ group compared with the HF-B−/− group at visit 1, but this difference was no longer present at visit 2. Moreover, in the HF-B−/− group, CO and CO index had increased at visit 2 compared with visit 1, whereas they did not change over time in the HF-B−/+ group. HR was significantly higher in the HF-B−/+ group compared with the HF-B−/− group at visit 1, but this intergroup difference was not present at visit 2. LVEDV did not differ between groups at visit 1 or 2. However, in the HF-B−/− group, LVEDV was increased at visit 2 compared with visit 1, while in the HF-B−/+ it remained unchanged.
Table 3. Pregnancy characteristics and cardiac indices in 69 formerly pre-eclamptic women, categorized according to absence of heart failure Stage B (HF-B) at 1 year and 4 years postpartum (HF-B−/−), absence of HF-B at 1 year but presence at 4 years (HF-B−/+), presence of HF-B at both 1 and 4 years (HF-B+/+) and presence of HF-B at 1 year but absence at 4 years (HF-B+/−)
(n = 43)
(n = 10)
(n = 6)
(n = 10)
Data are given as median [range], n/N (%) or median (interquartile range). Comparison with 1 year postpartum:
*P < 0.01;
†P < 0.05.
‡Pregnancy between index pregnancy and second visit.
CO, cardiac output; EA ratio, ratio of early-to-late ventricular filling velocity; GA, gestational age; IUFD, intrauterine fetal death; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVMi, left ventricular mass index; PE, pre-eclampsia; RWT, relative wall thickness; SGA, small-for-gestational age; SV, stroke volume.
Table 4 presents the cardiometabolic variables at visits 1 and 2 in each subgroup of women (Table S4 shows data after excluding women with a twin pregnancy). No metabolic variable differed between the HF-B−/− and HF-B−/+ subgroups at either visit. Moreover, in the HF-B−/− group, fasting glucose had decreased between visits 1 and 2 (4.8 vs 4.7 mmol/L, P < 0.05), DBP had increased (68 vs 72 mmHg, P < 0.05) and BMI had increased (22.5 vs 23.5 kg/m2, P < 0.05). In the HF-B−/+ group, all metabolic variables remained unchanged over time.
Table 4. Metabolic variables in 69 formerly pre-eclamptic women categorized according to absence of heart failure Stage B (HF-B) at 1 year and 4 years postpartum (HF-B−/−), absence of HF-B at 1 year but presence at 4 years (HF-B−/+), presence of HF-B at both 1 and 4 years (HF-B+/+) and presence of HF-B at 1 year but absence at 4 years (HF-B+/−)
(n = 43)
(n = 10)
(n = 6)
(n = 10)
Data are given as median (interquartile range) or n/N (%). Comparison with 1 year postpartum:
*P < 0.01;
†P < 0.05.
BMI, body mass index; DBP, diastolic blood pressure; HDL, high density lipoprotein; HOMA IR, homeostatic model assessment for insulin resistance; LDL, low density lipoprotein; MAP, mean arterial pressure; SBP, systolic blood pressure.
Inter- and intragroup differences in HF-B+/+vs HF-B+/−
We subsequently compared HF-B+/+ and HF-B+/− subgroups (Table 3; data excluding women with twin pregnancy are given in Table S3). Baseline characteristics, obstetric characteristics of index pregnancy and offspring pregnancy were comparable between the two groups. However, among the HF-B+/+ group there was a significantly greater proportion of women with a family history of PE compared with the HF-B+/− group (67% vs 10%). LVMi did not differ between subgroups at either visit. RWT did not differ between subgroups at visit 1 but was higher in the HF-B+/+ group compared with the HF-B+/− group at visit 2. LVEF did not differ between subgroups at visit 1. LVEF was significantly higher in the HF-B+/− group compared with the HF-B+/+ group at visit 2, and was increased in this subgroup compared to the first visit. SV, SV index, EA ratio, CO, CO index, HR and LVEDV did not differ between subgroups at visit 1 or 2.
When analyzing metabolic risk factors (Table 4), no variable differed between the two subgroups at either visit (Table S4 shows data after excluding women with a twin pregnancy). DBP was significantly higher in the HF-B+/+ group at visits 1 and 2 compared with the HF-B+/− group. Moreover, at visit 1, hypertension was significantly more prevalent in the HF-B+/+ group compared with the HF-B+/− group (83% vs 20%). However, at visit 2, this significant difference in hypertension prevalence was no longer present between the two subgroups.
To the best of our knowledge, this is the first presentation of a longitudinal cohort study on the prevalence of asymptomatic HF in formerly pre-eclamptic women. The prevalence of preclinical HF-B is similar at 1 and 4 years postpartum (23%) suggesting a persistent cardiac condition. However, HF-B resolved in two-thirds of originally affected women, and one in five initially unaffected women developed HF-B in the years thereafter.
The cardiovascular implications of PE do not end with delivery. In fact, in contrast to the physiological LV remodeling in normal pregnancy, which resolves in the first weeks postpartum, the PE-induced cardiac adaptation may persist for at least several months postpartum[8, 18]. To date, it is unknown whether (and when) PE-induced cardiac remodeling resolves completely. Traditional cardiovascular risk factors may affect the resolving process after PE and could be responsible for the differences in postpartum recovery between healthy and PE pregnancies.
The observed 23% prevalence of HF-B in the first year postpartum is in line with previous findings. Our data show that although the prevalence of HF-B at both visits is comparable, the underlying components are different. More than half of the women diagnosed with HF-B at visit 1 recovered, while about one in five formerly pre-eclamptic women without prior subclinical HF-B developed HF-B in the years thereafter, mostly as a result of concentric remodeling. Considering diastolic dysfunction is the most prevalent form of cardiac dysfunction in women, it may be hypothesized that the development towards concentric remodeling could be a more persistent condition in these women than the initial HF-B variant with mostly reduced systolic function.
HF can occur with a reduced EF (HFrEF) or with a preserved EF (HFpEF), the latter occurring more often in women than in men with a noteworthy 2 : 1 ratio. Despite significant uncertainty and conflicting results, the pathophysiological problem in patients with HFpEF has been related to impaired diastolic reserve. HFpEF is thought to develop from a ‘systemic pro-inflammatory state’ induced by comorbidities and with the involvement of microvascular endothelial inflammation, all of which are thought to be involved in the etiology of PE. Although this terminology is applied to symptomatic HF, specifying different subclinical phenotypes within the HF spectrum may theoretically allow stratification for potential early diagnosis and care to prevent progression. The prevalence of subclinical HFpEF in our population increased more than two-fold between both visits. Although borderline significant, in combination with the shift in concentric remodeling between both visits it is tempting to speculate that this shift may indicate a mechanistic response to the persistently high blood pressure that results in increased concentric remodeling.
Recent evidence suggests that chronic volume and pressure overload determine the resulting phenotype of hypertrophic remodeling. Untreated chronic high blood pressure is a well-known cause of structural cardiovascular alterations and cardiovascular morbidity. In the development of actual CVD, chronically elevated blood pressure is considered to be an important intermediate risk condition, which is easily identifiable and modifiable, enabling the timely institution of measures to prevent the premature development of a more debilitating CVD in women[2, 27]. Besides the mechanical strain and risks, HF risk may also originate from all or a subset of biochemical stressors consistent with metabolic syndrome. This is reflected in our findings, as some of the metabolic syndrome constituents seem to play a role in the prevalence of HF-B at visit 1. However, no association was found between the metabolic syndrome constituents and HF-B at visit 2, suggesting that different mechanisms contribute to the high prevalence at both time points. As most HF-B women have a phenotypical profile of consistent diastolic dysfunction, these findings may indicate a difference in either pressure or volume load, common in formerly pre-eclamptic women[29, 30].
In women, HF is often detected at a late and clinically overt stage. Formerly pre-eclamptic women should be aware of the increased risk for potentially reversible HF-B before it develops to the much less reversible symptomatic Stages C and D. Therefore, patients and clinicians should be aware of the importance of regular postpartum cardiovascular follow-up after a pre-eclamptic pregnancy. The patient may benefit from secondary prevention if CVD is diagnosed at an early stage. However, before specific treatment can be implemented in daily clinical practice, intervention trials are necessary, in order to prevent CVD in formerly pre-eclamptic women. Follow-up studies with a longer postpartum interval are necessary to provide more information on the prevalence of HF-B. Whether PE induces HF or whether PE and HF are a common result of the same susceptibility is yet to be determined.
This prospective longitudinal cohort study has some limitations. First, most women had ancestry originating from Europe. Therefore, our results may not be fully applicable to women of other ethnic origins. Second, after subdividing the cohort according to HF outcome, the number of women per subgroup became modest thus increasing the risk of Type I error. However, we did detect clinically relevant significant differences between subgroups. Therefore, considering the sample size, prevalence of abnormalities, and potential importance of the observations, additional studies confirming our findings are required.
In conclusion, the prevalence of subclinical HF-B is persistently high in the years after a PE pregnancy. Nonetheless, there is a noteable shift in individuals and phenotypes contributing to HF-B in the years after PE. From a clinical perspective, these findings may imply that cardiovascular risk screening at approximately 1 year postpartum is important, and that an additional assessment a few years later is necessary. Whether lifestyle change and pharmacological therapy are effective at preventing progression or development towards clinical HF in these women needs to be investigated.