Objective
The aim of this study was to investigate which nonclassic cardiovascular biomarkers are associated with persistent endothelial dysfunction after pregnancy in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancies.
Study Design
This was a systematic review and metaanalysis of observational studies. A search was performed in PubMed, Embase, Cochrane, and Cinahl including articles from inception to Feb. 27, 2013. Included were cohort studies and case-control studies. Cases were women with a history of hypertension in pregnancy, control subjects were women with a history of uncomplicated pregnancies. Of the 3136 found, 21 studies on 16 nonclassic cardiovascular biomarkers are described in this review; 12 studies on 5 biomarkers were included in the metaanalysis.
Results
Women with a history of hypertensive pregnancy disorders had a higher homocysteine level compared with women with a history of uncomplicated pregnancies (5 studies; pooled mean difference, 0.77 ng/mL; 95% confidence interval, 0.27–1.26; P < .01). For the other nonclassic cardiovascular biomarkers including markers in areas of inflammation, thrombosis, and angiogenesis, we found no significant differences.
Conclusion
This review and metaanalysis showed that women with a history of hypertensive pregnancy disorders have higher homocysteine levels compared with women with a history of uncomplicated pregnancies. These data suggest persistent endothelial alteration after pregnancies complicated by hypertensive disorders.
Cardiovascular disease is the leading cause of death in women in the Western world. Cardiovascular disease manifests itself differently different between men and women; diagnostic tools in women are less sensitive and specific. Therefore, it is of additional value to identify specific risk factors for cardiovascular disease in women. For women, preeclampsia has been suggested as a specific risk factor for cardiovascular disease later in life.
Hypertensive disorders in pregnancy are hypothesized to act as a stress test for cardiovascular disease later in life ; women who fail this stress test by developing hypertensive pregnancy disorders have an increased cardiovascular risk, which will become apparent during pregnancy. Both disorders, hypertensive pregnancy disorders and cardiovascular disease later in life, share a common pathophysiologic pathway of endothelial dysfunction.
Recently a review on classic cardiovascular biomarkers showed higher levels of glucose, insulin, triglycerides, total cholesterol, low-density lipoprotein cholesterol, and microalbumin and lower levels of high-density lipoprotein cholesterol in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancies. These biochemical cardiovascular risk markers are predictors for cardiovascular disease later in life and might identify high-risk young women early enough to benefit from screening and intervention.
In addition to classic cardiovascular biomarkers, there is a wide variety of nonclassic cardiovascular biomarkers associated with endothelial dysfunction. These latter mentioned biomarkers might be more associated with future risk of cardiovascular disease and therefore might give options for preventive interventions.
In this systematic review and metaanalysis, we focus on 16 nonclassic cardiovascular biomarkers associated with persistent endothelial dysfunction, including inflammation (intercellular adhesion molecule [ICAM], vascular cell adhesion molecule [VCAM], interleukin-6 [IL-6], interleukin-10 [IL-10], and E-selectin), thrombosis (homocysteine, von Willebrand factor [VWF], fibrinogen, fibronectin, endothelin, D-dimer, plasminogen activator inhibitor-1 [PAI-1], tissue plasminogen activator [tPA]), and angiogenesis (vascular endothelial growth factor [VEGF], soluble Fms-like tyrosine kinase-1 [sFLT-1], and tumor necrosis factor alpha ([TNF-α]).
Previous research suggests that these cardiovascular biomarkers could be useful in prediction, identification, and assessment of hypertensive pregnancy disorders, especially preeclampsia. Because cardiovascular disease and hypertensive pregnancy disorders share a (partly) common pathophysiology, we used these biomarkers in our search for this systematic review and metaanalysis.
The aim of this study was to investigate whether nonclassic cardiovascular biomarkers are associated with persistent endothelial dysfunction after pregnancy in women with a history of hypertensive pregnancy disorders compared with women who have had uncomplicated pregnancies.
Materials and Methods
Definitions
Hypertensive pregnancy disorders include gestational hypertension, preeclampsia, superimposed preeclampsia, and eclampsia. Gestational hypertension was defined according to the International Society for the Study of Hypertension in Pregnancy criteria as diastolic blood pressure of 90 mm Hg or greater measured on 2 occasions at least 6 hours apart in a woman who was normotensive at the start of pregnancy until week 20 of gestational age.
Preeclampsia was defined according to the International Society for the Study of Hypertension in Pregnancy criteria: onset of a blood pressure exceeding 140/90 mm Hg with proteinuria greater than 0.3 g per 24 hours after 20 weeks’ gestation. Severe preeclampsia was defined as a blood pressure exceeding 160/110 mm Hg or proteinuria greater than 5 g per 24 hours, or both. Eclampsia was defined as the presence of seizures. Superimposed preeclampsia was defined as preexisting hypertension with new onset proteinuria greater than 0.3 g per 24 hours after 20 weeks’ gestation.
Sources
We searched PubMed, Embase.com, Cochrane Library (via Wiley), and Cinahl (via EBSCO) (J.C.F.K., R.H.J.O., and S.V.) from inception to Feb. 27, 2013. The following search terms with synonyms were used: gestational hypertension, preeclampsia, and eclampsia.
The search on nonclassic cardiovascular biomarkers in the area of inflammation included the following terms with synonyms: ICAM, VCAM, IL-6, IL-10,and E-selectin. The search on nonclassic cardiovascular biomarkers in the area of thrombosis included the following terms with synonyms: homocysteine, VWF, fibrinogen, fibronectin, endothelin, D-dimer, PAI-1, and tPA. The search on nonclassic cardiovascular biomarkers in the area of angiogenesis included the following terms with synonyms: VEGF, sFLT-1, and TNF-α. The search for study type included the following terms with synonyms: systematic reviews, metaanalyses, cohort studies, and case-control studies. The choice of biochemical markers was based on the knowledge of their role in hypertensive disorders in pregnancy and suspected endothelial activation.
Study selection
We included case-control studies and cohort studies. Further inclusion criteria hold cases described in the articles had to be women with a history of hypertensive pregnancy disorder by previous mentioned definition, control subjects had to be women with a history of uncomplicated pregnancies, the articles should describe measurements in blood samples of one of the nonclassic cardiovascular biomarkers as described above, and blood sample should be drawn more than 6 weeks after delivery. There were no limitations made on publication date or patients sample size.
We excluded studies without the definitions of gestational hypertension or (pre)eclampsia or without baseline criteria of the study population mentioned above. Studies that included superimposed preeclampsia were excluded from this review.
Two reviewers (S.V. and W.H.) screened the title, abstract and key words ( Figure ). All titles were screened by the 2 reviewers (S.V. and W.H.). If the title was not specific enough for decision on inclusion or exclusion, we reviewed the abstract according to the following criteria: case-control or cohort study and blood samples drawn after pregnancy complicated by hypertensive disorders. No exclusions were made on publication date or language of article. If a reference was potentially eligible, a full-text article was scored using a scoring list conducted by the reviewers (S.V., W.H., and C.J.M.d.G.) to assess the article’s quality.
Explicit definitions of gestational hypertension and (pre)eclampsia, parity, mean gestational age, preterm or term pregnancy, number of cases and controls included, follow-up time, and adequate information about blood collection and blood analysis was given (ie, if blood samples were taken sober) were required for inclusion in this systematic review. During the comparison of these scoring lists by the reviewers (S.V. and W.H.), any inconsistencies were resolved by discussion with a third reviewer (C.J.M.d.G.).
Outcomes of interest
Outcomes were differences in nonclassic cardiovascular biomarkers, measured after pregnancy, between women with a history of hypertensive pregnancy disorders and women with a history of uncomplicated pregnancies. Nonclassic cardiovascular biomarkers found in case-control and cohort studies were numerous ( Table ).
| Study, country | Study design | Exposure | Parity index pregnancy | Mean gestational age, wks, index pregnancy | Preterm and/or term pregnancy | Cases, n | Controls, n | Follow-up period, y | Nonclassic cardiovascular marker described |
|---|---|---|---|---|---|---|---|---|---|
| 1. Deng et al 1994, Sweden a | Prospective, cohort | PE | NP and MP | Not reported | T | 63 | 29 | 1 (5-15 mo) | VWF, fibronectin (means) |
| 2. Bremme and Blombäck, 1996, Sweden a | Retrospective, case-control | PE | NP and MP | Severe PE, 33 Mild PE, 35 | PT and T | 42 | 26 | 1 (6-15 mo) | Fibronectin, fibrinogen, PAI-1, D-dimer (means) |
| 3. Laivuori et al, 1998, Finnland | Retrospective, case-control | PE | Not reported | Not reported | Not reported | 22 | 22 | Cases, 16.9, Controls, 17.0 | Endothelin (means) |
| 4. Barden et al, 1999, Australia | Retrospective, case-control | PE | NP and MP | Not reported | Not reported | 62 | 84 | 0.5 (6 mo) | Endothelin (means) |
| 5. He et al, 1999, Sweden | Retrospective, case-control | PE | NP and MP | Not reported | Not reported | 24 | 23 | Range, 2–5 | Fibrinogen, VWF, PAI-1, tPA (medians) |
| 6. Chambers et al, 2001, England a | Prospective, case-control | PE | NP and MP | Not reported | Not reported | 78 | 48 | Mean, 3 | Homocysteine, E-selectin, ICAM (means) |
| 7. Sattar et al, 2003, Scotland | Retrospective, cohort | PE | NP | Cases, 36 Controls, 40 | PT and T | 40 | 40 | Range, 18–28 | ICAM, VCAM, E-selectin (medians) |
| 8. Vickers et al, 2003, Scotland a | Prospective, cohort | PE and PIH | NP | Not reported | Not reported | 392 and 297 | 163 | Range, 33–52 | Fibrinogen, VWF (means) |
| 9. Freeman et al, 2004, Scotland a | Retrospective, case-control | PE | NP | Not reported | Not reported | 40 | 40 | Mean, 19.8 (cases), 19.9 (controls) | ICAM, VCAM, IL-6, IL-10, TNF-α (means) |
| 10. Raijmakers et al, 2004, The Netherlands a | Retrospective, case-control | PE | NP and MP | Not reported | Not reported | 113 | 94 | 2 (28 mo) | Homocysteine (means) |
| 11. Wolf et al, 2004, United States | Retrospective, case-control | PE | NP | Not reported | Not reported | 29 | 32 | 2 (18 ± 7.9 mo) | sFlt-1 (means), VEGF (medians) |
| 12. Paradisi et al, 2006, Italy a | Prospective, cohort | PIH | NP and MP | Cases, 37.9 Controls, 39.2 | T | 15 | 15 | >1 (20 mo) | Homocysteine (means) |
| 13. Girouard et al, 2007, Canada a | Prospective, cohort | PE and PIH | NP and MP | Cases, 38.8 Controls, 39.4 | T | 168, 63, and 105 | 168 | Mean, 7.8 | Homocysteine (means), IL-6, PAI-1, TNF-α (medians) |
| 14. Hamad et al, 2007, Sweden a | Retrospective, case-control | PE | NP | Not reported | Not reported | 18 | 17 | 1 (15 mo) | Fibrinogen, VWF, tPA ICAM, VCAM, E-selectin (means), PAI-1 (medians) |
| 15. van Rijn et al, 2008, The Netherlands a | Retrospective, case-control | PE | NP | Cases, 29.9 Controls, 40.0 | PT and T | 340 | 113 | 0.5 (>6 mo) | Fibrinogen, IL-6, ICAM, VWF (means) |
| 16. Lampinen et al, 2008, Finnland | Retrospective, case-control | PE | NP and MP | Cases, 33 Controls, 40 | PT and T | 28 | 20 | Range, 5–6 | IL-6, VWF (medians) |
| 17. Portelinha et al, 2009, Portugal | Retrospective, case-control | PE | Not reported | Cases, 35 Controls, 39 | PT and T | 65 | 54 | Mean, 6 | PAI-1 (means) D-dimer, fibrinogen, tPA (medians) |
| 18. Porthelina et al, 2008, Portugal a | Retrospective, case-control | PE | NP | Cases, 34 Controls, 39 | PT and T | 58 | 49 | Mean, 6 | VCAM, ICAM (means) |
| 19. Coffeng et al, 2011, The Netherlands a | Retrospective, case-control | PE | NP and MP | Not reported | Not reported | 16 | 17 | Median, 4 | Homocysteine (means) |
| 20. Stepan et al, 2011, Germany | Prospective, cohort | PE | Not reported | Not reported | Not reported | 37 | 37 | 0.5 (6 mo) | TNF-α (medians) |
| 21. Gaugler-Senden et al, 2012, The Netherlands | Retrospective, case-control | PE | NP and MP | Cases, 22.8 Controls, 40.2 | PT and T | 16 | 18 | Median, 9.4 (cases), 9.7 (controls) | sFlt, VEGF, E-selectin (medians) |
In this review, we describe a specific selection of nonclassic cardiovascular biomarkers (n = 16), ICAM, VCAM, IL-6, IL-10, E-selectin, homocysteine, VWF, fibrinogen, fibronectin, endothelin, D-dimer, PAI-1, tPA, VEGF, sFLT-1, and TNF-α. Only if 3 or more studies reported on a nonclassic biomarker and reported outcome values as means ± SEM or SD, we preformed a metaanalysis including the nonclassic biomarker. If studies reported values as medians (range/interquartile range) or if less than 3 studies reported on the nonclassic biomarker, we described the nonclassic biomarker separately.
Data synthesis
We used Review Manager 5.0 (Cochrane Collaboration, Oxford, UK) for statistical analyses. Outcomes were reported as continuous data and analyzed using a mean difference. Raw numbers were used from each study, as adjustments for confounding effects varied among the different studies. Weighting of studies in the metaanalyses was calculated on the basis of the inverse variance of the study. The random effects model was chosen because a clinical and statistical heterogeneity was expected among the studies. We used forest plots to visualize data and assessed heterogeneity using the I² test. For all effect estimates, a value of P < .05 indicated statistical significance.
Results
Literature identification and study quality
The initial search produced 4995 articles: PubMed 1775 articles, Embase 2757 articles, Cinahl 260 articles, and Cochrane 119 articles. After removing duplicates of references that were selected from more than 1 database, 3136 papers remained. After screening the titles and abstracts, we retrieved 73 articles for detailed evaluation of the full text. As described in the Figure , we included 21 articles of these 73 in this review. For the metaanalyses we included 12 articles that reported on 5 different nonclassic cardiovascular biomarkers. The other nonclassic cardiovascular biomarkers will be described separately.
Study characteristics
The studies included in this review were published from 1994 to 2012. Study characteristics are shown in the Table . Data collection of studies in this review was prospective design in 15 studies (71%). Six studies (29%) were designed as cohort studies and 15 (71%) were case control studies.
In 1 study (5%), only women with a history of pregnancy induced hypertension were included ; 2 studies (9%) included women with a history of pregnancy induced hypertension and preeclampsia, and 18 studies (86%) included women with a history of preeclampsia.
In 11 studies (52%), both nulliparous and multiparous women were included ; in 7 other studies (32%), only nulliparous women were included. We included 3 studies (14%) in which parity was not described. In 3 studies (14%), women were included only after a term pregnancy ; in 7 studies (33%), women who delivered at both term and at preterm were included. In the other 11 studies (52%), it was not reported whether women delivered at term or preterm. The follow-up period of the included studies varied from 5 months to 52 years. An overall mean weighted follow-up was not calculated because the follow-up periods in the different studies were reported as means (±SD) or medians (ranges/interquartile range).
Biomarkers for inflammation
ICAM and VCAM
Five studies with a total of 534 cases and 264 controls were able to be included in metaanalyses on ICAM, which showed no significant difference in mean ICAM levels in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy (mean difference, 11.04 ng/mL; 95% confidence interval [CI], –12.10 to 34.18; P = .35). The test heterogeneity observed was significant ( P = .04, I² = 61%) ( Figure ). One study described a higher median ICAM level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
Three studies with a total of 116 cases and 106 controls were able to be included in metaanalyses on VCAM, which showed no significant difference in mean VCAM levels in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy (mean difference, 68.72 ng/mL; 95% CI, –150.86 to –288.29). The test heterogeneity observed was significant ( P < .00, I² = 97%) ( Figure ). One study described a higher median VCAM level in women with a history of hypertensive pregnancy disorders compared with women with an uncomplicated pregnancy.
Two studies described higher mean IL-6 levels, and 1 study described a higher median IL-6 level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described a lower median IL-6 level in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy. Overall, IL-6 levels were described higher after pregnancy in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy.
One study described a higher mean IL-10 level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
One study described a higher mean E-selectin level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described the same mean E-selectin level in women with a history of hypertensive pregnancy disorders and women with uncomplicated pregnancy. Two studies described higher median E-selectin levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. Overall, E-selectin levels were described higher after pregnancy in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy.
Biomarkers for thrombosis
Homocysteine, VWF, and fibrinogen
Five studies with a total of 390 cases and 342 controls could be included in metaanalyses on homocysteine, which showed higher mean homocysteine levels in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy (mean difference, 0.77 ng/mL; 95% CI, 0.27–1.26; P < .00). The test heterogeneity observed was not significant ( P = .11, I² = 48%) ( Figure ).
Three studies with a total of 473 cases and 209 controls could be included in metaanalyses on VWF, which showed no significant difference in mean VWF levels in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy (mean difference, 0.06 ng/mL; 95% CI, –0.05 to 0.17; P = .28). The test heterogeneity observed was not significant ( P = .47, I² = 0%) ( Figure ). Two studies described higher VWF median levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
Four studies with a total of 778 cases and 319 controls could be included in metaanalyses on fibrinogen, which showed no significant difference in mean fibrinogen levels in women with a history of hypertensive pregnancy disorders compared with women with a history of uncomplicated pregnancy (mean difference, 0.28 g/L; 95% confidence interval, –0.02 to 0.59; P = .07). The test heterogeneity observed was significant ( P = .02, I² = 70%) ( Figure ). Two studies described higher median fibrinogen levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
Two studies described higher mean cellular fibronectin levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
One study described a higher mean endothelin levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described the same mean endothelin level in women with a history of hypertensive pregnancy disorders and women with uncomplicated pregnancy.
One study described a higher mean D-dimer level and 1 study described a higher median D-dimer level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
One study described a higher mean PAI-1 level and three studies described higher median PAI-1 levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described a lower mean PAI-1 level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
One study described a higher mean tPA level and 2 studies described higher median tPA levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
Biomarkers for angiogenesis
VEGF, sFLT-1, and TNF-α
Two studies described higher median VEGF levels in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described a higher mean sFlt level, and 1 study describes higher median sFlt level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
One study described a higher mean TNF-α level, and 1 study described a higher median TNF-α level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy. One study described a lower median TNF-α level in women with a history of hypertensive pregnancy disorders compared with women with uncomplicated pregnancy.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
