Association between Cardiorespiratory Fitness and Hypertensive Disorders of Pregnancy: A Systematic Review and Meta-Analysis

Hypertensive disorders of pregnancy (HDP) are associated with future cardiovascular disease (CVD), which may be mediated by diminished cardiorespiratory fitness (CRF). In this systematic review and meta-analysis, we summarize evidence linking CRF with HDP before, during, and after pregnancy. We searched relevant databases to identify observational or randomized studies that measured CRF (VO2 max or peak, VO2 at anaerobic threshold, or work rate at peak VO2) in women with and without HDP. We pooled results using random effects models. Fourteen studies (n = 2406 women) reporting on CRF before, during, and after pregnancy were included. Before pregnancy, women who developed HDP had lower CRF (e.g., VO2max < 37 vs. ≥37 mL O2/min) than those without HDP (two studies, 811 women). VO2max at 14–18 weeks of pregnancy was marginally lower among women who developed preeclampsia vs. normotensive women (three studies, 275 women; mean difference 0.43 mL/kg/min [95% CI 0.97, 0.10]). Postpartum, there was a trend towards lower VO2peak in women with previous preeclampsia (three studies, 208 women; 0.26 mL/kg/min [−0.54, 0.02]). While exploratory, our findings raise the possibility that CRF can identify women at risk for HDP, and furthermore, that HDP confers a hit to a woman’s cardiorespiratory reserve.


Introduction
Hypertensive disorders of pregnancy (HDP), including gestational hypertension (GH) and preeclampsia (PE), are leading causes of maternal and neonatal morbidity and mortality [1,2]. Beyond the immediate implications for mother and child, numerous studies have demonstrated an association between HDP and future cardiovascular disease (CVD) [3][4][5][6]. However, whether HDP contributes causally to CVD, merely highlights pre-existing CVD risk, or both [7] and what specific mechanisms interact to lead to HDP and future CVD [5] are still uncertain.
Cardiorespiratory fitness (CRF) reflects the combined functional capacity of the cardiovascular and respiratory systems to provide and maintain an adequate oxygen supply to working skeletal muscles over prolonged periods of physical activity [8,9]. Pregnancy alters both the cardiovascular and respiratory systems during rest and exercise, acting as a natural cardiorespiratory stress test [10]. A recent meta-analysis in normotensive women showed that prenatal exercise interventions help improve maternal CRF, indicating that it is an actionable and modifiable metric that has the potential to improve cardiovascular health during and after pregnancy [11].
Focusing on peripartum CRF offers a dynamic and mechanistic approach to understanding the interplay between potential risk factors for HDP and their roles in the trajectory towards CVD [12,13]. However, it is not known whether women who develop HDP have reduced CRF, or if HDP accelerates CRF decline after pregnancy. We therefore conducted this systematic review to summarize the literature on CRF before, during, and after HDP, compared with pregnant women without HDP, with CRF assessed using maximal or peak oxygen consumption (VO 2 ), VO 2 at anaerobic threshold (VO 2 AT), and work rate or test distance or duration.

Study Eligibility and Inclusion Criteria
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were used to conduct this systematic review and meta-analysis based on an a priori protocol [14,15]. The protocol was registered and can be accessed at the International Prospective Register of Systematic Reviews (PROSPERO; Registration No. CRD42019126663).
We included observational (case-control, prospective and retrospective cohort, and cross-sectional) and interventional (RCT or non-randomized intervention) studies, with no date or language restrictions. Eligible studies were performed in women of any age and parity. Conference abstracts were included only if a corresponding article published in a peer-reviewed journal was not found. Animal studies, case reports and series, comments and editorial letters, and reviews were excluded, as were study protocols. The exposure of interest was HDP of any severity (including PE and gestational hypertension), which needed to be clearly defined either via direct clinical assessment, medical chart review, or by self-report. Studies were only included if there was a control group of unexposed women without HDP, thereby enabling between-group comparisons of outcomes. The outcome in all eligible studies was CRF, measured using any type of cardiopulmonary exercise test or aerobic fitness field test. CRF measures were reported as VO 2 max (mL/kg/min) or VO 2 peak (mL/kg/min), predicted or estimated VO 2 max (mL/kg/min), VO 2 AT (mL/kg/min), and work rate (watts) or aerobic test distance (metres) or duration (seconds). We considered all studies with a measure of HDP and of CRF, irrespective of the timing of these measurements or the presence of other gestational disorders. We reported and analyzed antepartum and postpartum measurements separately.

Data Sources and Search Strategy
An initial MEDLINE strategy was developed by a research librarian (T.L.) based on a scoping search to identify relevant search terms and was then modified with input from the research team. The final MEDLINE strategy was adapted for other databases, with vocabulary and syntax tailored to enable optimal digital searches up to July 2018 of the following databases: MEDLINE (via Ovid and PubMed), EMBASE, Cochrane Library (CENTRAL and CDSR), and Scopus. Search strategies complied with the Institute of Medicine standards [16] and were not limited by language restrictions. The reference lists of included articles and relevant systematic reviews were checked manually for potentially relevant articles. ClinicalTrials.gov and WHO's International Clinical Trials Registry Platform (ICTRP) Search Portal were used to identify clinical trials recently completed, and associated publications were then retrieved. A final MEDLINE search was performed to identify additional references published through 14 January 2021 (L.H.). Search terms and results are shown in Appendix A.

Study Selection
Two reviewers (F.A. and G.D.S.) independently selected studies using the specific eligibility criteria. The first screening was based on titles and abstracts of identified publications. All studies identified by at least one reviewer as potentially relevant were retrieved for full-text evaluation. Both F.A. and G.D.S. independently evaluated full-text references, and reasons for exclusion were recorded. Disagreements were resolved by a third reviewer (N.D.). If studies were found using the same study population, the most recent or most complete publication was selected [17][18][19], unless both publications contained unique, potentially relevant data, in which case both were included but counted as one study in the flow diagram. Citations of relevant systematic reviews and meta-analyses were searched manually (F.A. and G.D.S.), and eligible full texts were retrieved. Authors were contacted when studies reported using cardiopulmonary exercise testing but did not report CRF. For example, studies reporting other cardiorespiratory health measures such as blood pressure and heart rate in response to exercise were excluded if no other CRF values were recorded. References excluded with reasons can be found in Appendix B.

Data Collection Process
Data were extracted from full-text articles using a data extraction spreadsheet in Microsoft Excel by two independent reviewers (F.A. and G.D.S.), with disagreements resolved through discussion and consultation (N.D. and M.H.D.). For the meta-analysis, data were entered into ReviewManager v5.3 (Cochrane Collaboration, Copenhagen, Denmark) by one reviewer and verified by the other (F.A. and G.D.S., respectively).

Data Retrieved
Study characteristics extracted included study design, year and country of publication, definition and subtype of HDP (i.e., GH or PE), method of CRF assessment, and type of exercise intervention (if applicable). From each included study, we extracted sample characteristics including number of participants in each exposure group (GH/PE vs. no GH/PE), age, ethnicity, parity, length of time between pregnancy and CRF measurement, whether participants were pregnant or postpartum, history of underlying conditions, CRF measurements, and other obstetric outcomes as available (e.g., gestational age at delivery, Caesarean delivery).

Assessment of Risk of Bias
Reviewers (F.A. and G.D.S.) independently assessed the quality of each included study with the resolution of disagreements through discussion or involvement of N.D. Risk of bias was assessed using the Joanna Briggs Institute Critical Appraisal Checklist for Cohort and Analytical Cross Sectional Studies [20], which was used for intervention studies as well, as only pre-intervention data were included. All publications meeting inclusion criteria were included regardless of quality. Given the small number of studies obtained for subgroup analysis and meta-regression, it was not possible to examine the impact of study quality or publication bias on pooled outcomes.

Data Analysis
Statistical analyses were conducted using ReviewManager. Effect measures were reported as mean differences and standard deviations (SD) in CRF between groups. For studies using different exercise or field test modes to estimate CRF, standardised mean differences (SMD) were calculated instead. If studies reported standard error (SE), SD values were calculated from the SE by multiplying by the square root of the sample size [21]. If results were reported as absolute CRF, authors were contacted to obtain CRF relative to weight.
Meta-analyses were conducted using inverse variance random-effects models. If there were at least 2 studies using the same type of variable for CRF, results were pooled, with a p-value < 0.05 considered statistically significant. The I-squared statistic was used to evaluate heterogeneity, with results pooled only for studies with I 2 < 80%. A priori subgroup analyses were carried out for type of CRF measurement (VO 2 peak, estimated VO 2 max, VO 2 AT, work rate, test distance, or test time) and timing of CRF measurement (before, during, or after pregnancy). If data were not suitable for meta-analysis or relevant data were missing or unclear, authors were contacted to obtain additional information. Data were qualitatively synthesized if authors were unable to provide additional numerical data.

Literature Search
The initial literature search yielded 1949 records for title and abstract screening after removal of duplicate results, of which 126 records were retrieved and assessed for eligibility. A total of 116 reports were excluded for reasons including no measure of any HDP or CRF, no normotensive comparison group, CRF not reported in relation to HDP, duplicate study population, or inappropriate study design ( Figure 1). Ten studies were included in the review [17][18][19][22][23][24][25][26][27][28] and five in the meta-analysis [17,19,22,23,28]. Manual citation searching of included articles and relevant systematic reviews yielded 65 additional references, of which 3 were included in our systematic review [29][30][31] (2 in the meta-analysis [29,31]). Finally, the second Medline search conducted on 14 January 2021 yielded 127 additional records for title and abstract screening, among which 13 additional eligible reports were assessed, and 1 [32] was included in the meta-analysis. Authors were contacted to request study data for nine studies for which data relevant to our review were potentially available but not reported [18,19,23,25,[27][28][29][30]32]; authors of seven studies responded and provided relevant data [19,23,25,[28][29][30]32]. A total of 14 studies met final inclusion criteria including 2406 women: 560 with HDP GH/PE and 1846 without GH/PE. Meta-analyses were conducted using inverse variance random-effects models. there were at least 2 studies using the same type of variable for CRF, results were poole with a p-value < 0.05 considered statistically significant. The I-squared statistic was us to evaluate heterogeneity, with results pooled only for studies with I 2 < 80%. A priori su group analyses were carried out for type of CRF measurement (VO2peak, estimat VO2max, VO2AT, work rate, test distance, or test time) and timing of CRF measureme (before, during, or after pregnancy). If data were not suitable for meta-analysis or releva data were missing or unclear, authors were contacted to obtain additional informatio Data were qualitatively synthesized if authors were unable to provide additional nume ical data.

Literature Search
The initial literature search yielded 1949 records for title and abstract screening aft removal of duplicate results, of which 126 records were retrieved and assessed for elig bility. A total of 116 reports were excluded for reasons including no measure of any HD or CRF, no normotensive comparison group, CRF not reported in relation to HDP, dup cate study population, or inappropriate study design ( Figure 1). Ten studies were i cluded in the review [17][18][19][22][23][24][25][26][27][28] and five in the meta-analysis [17,19,22,23,28]. Manu citation searching of included articles and relevant systematic reviews yielded 65 add tional references, of which 3 were included in our systematic review [29][30][31] (2 in the met analysis [29,31]). Finally, the second Medline search conducted on 14 January 2021 yield 127 additional records for title and abstract screening, among which 13 additional eligib reports were assessed, and 1 [32] was included in the meta-analysis. Authors were co tacted to request study data for nine studies for which data relevant to our review we potentially available but not reported [18,19,23,25,[27][28][29][30]32]; authors of seven studies r sponded and provided relevant data [19,23,25,[28][29][30]32]. A total of 14 studies met fin inclusion criteria including 2406 women: 560 with HDP GH/PE and 1846 without GH/P

Study Quality
Study quality ratings are shown in Table 2. Appropriate conduct of analysis and follow-up was documented in nearly all studies. Exposure measurement was considered valid and reliable in 12 of 14 studies. One study was presented as an abstract, and women with previous severe PE were compared to controls, but assessment of PE was not welldescribed [22]. The second study reported mean watts produced but did not report VO 2 [23]. The outcome was measured in a valid and reliable way in 8 [17,19,22,23,[28][29][30]32] of 14 studies. Unreliable outcome measures for the purposes of our review included selfreported HDP [27], assessment of HDP without reporting diagnostic criteria [18], and measurement of work rate [31], distance travelled [24,25], or exercise test duration [26] without presentation of sufficient data to calculate VO 2 . Were the two groups similar and recruited from the same population?
Were the exposures measured similarly to assign people to both exposed and unexposed groups?
Was the exposure measured in a valid and reliable way?
Were confounding factors identified?
Were strategies to deal with confounding factors stated? Were the groups/participants free of the outcome at the start of the study (or at the moment of exposure)?
Were the outcomes measured in a valid and reliable way?
Was the follow-up time reported and sufficient to be long enough for outcomes to occur?
Was follow-up complete, and if not, were the reasons to loss to follow-up described and explored?
Were strategies to address incomplete follow-up utilized?
Was appropriate statistical analysis used? Confounding factors such as maternal ethnicity, smoking, education, and income were identified in 13 of 14 studies, and strategies to deal with confounding factors were stated in 9 of these studies. However, none of the studies included adjusted analyses of the association between HDP and CRF.

CRF before Pregnancy and Subsequent HDP
Two studies [18,27] examined CRF before pregnancy in relation to future GH and/or PE. Morris et al. [18] compared women with VO 2 max < 37 vs. ≥37 mL O 2 /min and found that pre-pregnancy CRF was low in 100% of women with GH or PE (n = 10) compared to 57.6% (19/33) of women without GH or PE (p = 0.01). Lane-Cordova et al. [27] divided study participants into three fitness tertiles based on performance on a treadmill test and found that GH was diagnosed in 21%, 19%, and 12% of women in the lowest, middle, and highest fitness tertiles, respectively (p = 0.03). However, the study did not report rates of PE.

CRF during Pregnancy and Subsequent GH or PE
Seven studies [19,23,24,[28][29][30][31] examined CRF during pregnancy. Bisson et al. [30] found a mean VO 2 AT of 15.0 and 15.7 mL/kg/min among women who developed GH and normotensive women, respectively, at 14 weeks gestation, with a mean VO 2 AT of 14.9 and 15.0 mL/kg/min in the two groups, respectively, at 28 weeks gestation. Da Silva et al. [24] found women who had been diagnosed with PE completed a shorter distance on the 6 min walk test compared to normotensive women (421 m vs. 497 m, p = 0.001, median gestational age at measurement 37 weeks in both groups, interquartile range 34-38 among women with PE, 33-38 for normotensive women).

HDP and Postpartum CRF
CRF was examined after pregnancy in six studies [17,22,23,25,26,32]. Duration since delivery at the time of CRF measurement ranged from 6 weeks to 20 years. There was a trend towards lower VO2peak in women with prior PE compared with controls (three studies [17,22,32], 208 women; −0.26 mL/kg/min [−0.54, 0.02], p = 0.07; Figure 3). Cottrill et al. [26] found that women with previous GH and/or PE had a lower mean duration of exercise on a modified Balke treadmill test compared to women with previous non-HDP pregnancy (115 women; 660 vs. 738 s; SMD −0.42 [−0.80, 0.05]; p = 0.03; measurement of CRF 4-6 years following pregnancy). Harville et al. [25] found that women with GH and/or PE took longer to complete a 4-metre walk test compared to women with no previous GH or PE (1329 women; 4.

Discussion
In this systematic review and meta-analysis of 14 studies, we identified a pattern of lower preconception CRF in women who were subsequently diagnosed with GH and/or PE. While no significant differences were noted in weight-adjusted CRF during pregnancy, results suggest that women who developed PE may have lower CRF earlier in pregnancy, as well as months to years after delivery.
It is widely known that higher CRF improves cardiovascular health and lowers cardiovascular mortality [33,34]. Previous studies have shown HDP, especially preeclampsia, to be a risk factor for later CVD [3,4], with a two-fold increased risk of ischemic heart disease 10 to 15 years following the pregnancy [5]. Our findings suggest that reduced CRF may be one pathway connecting HDP to future adverse cardiovascular outcomes. Specifically, lower postpartum CRF in HDP-affected women suggests a possible "hit" to cardiorespiratory reserve brought about by HDP-one that is potentially actionable.
Our study adds to the large body of literature suggesting that pregnancy is a "stress test" identifying women at risk for future cardiovascular risk factors and overt CVD [35]. Our analyses suggest reduced cardiorespiratory capacity both in women who went on to develop GH or PE and in those with previous HDP. Prior systematic reviews have demonstrated that engaging in physical activity during pregnancy reduces the odds of developing HDP by ~40% [36]. That this observed benefit seems mediated by improved CRF is Figure 3. Mean differences in VO 2 peak (in mL/kg/min) in postpartum women with and without previous preeclampsia [17,22,32].

Discussion
In this systematic review and meta-analysis of 14 studies, we identified a pattern of lower preconception CRF in women who were subsequently diagnosed with GH and/or PE. While no significant differences were noted in weight-adjusted CRF during pregnancy, results suggest that women who developed PE may have lower CRF earlier in pregnancy, as well as months to years after delivery.
It is widely known that higher CRF improves cardiovascular health and lowers cardiovascular mortality [33,34]. Previous studies have shown HDP, especially preeclampsia, to be a risk factor for later CVD [3,4], with a two-fold increased risk of ischemic heart disease 10 to 15 years following the pregnancy [5]. Our findings suggest that reduced CRF may be one pathway connecting HDP to future adverse cardiovascular outcomes. Specifically, lower postpartum CRF in HDP-affected women suggests a possible "hit" to cardiorespiratory reserve brought about by HDP-one that is potentially actionable.
Our study adds to the large body of literature suggesting that pregnancy is a "stress test" identifying women at risk for future cardiovascular risk factors and overt CVD [35].
Our analyses suggest reduced cardiorespiratory capacity both in women who went on to develop GH or PE and in those with previous HDP. Prior systematic reviews have demonstrated that engaging in physical activity during pregnancy reduces the odds of developing HDP by~40% [36]. That this observed benefit seems mediated by improved CRF is perhaps not surprising but highlights the importance of offsetting the complex cardiovascular adaptations of both normotensive and hypertensive pregnancy with exercise. It is therefore likely that CRF is a metric that can be used both to identify higher-risk individuals and also to personalize safe and targeted exercise programs during and after pregnancy.
Our study highlights the need for the standardization of exercise tests used in research on pregnant and postpartum women in order to facilitate meaningful comparison of results between studies. Our findings also have implications for clinical practice. The American Heart Association has recommended that CRF be used as a risk marker for morbidity and mortality in the general population [12], and our review suggests that CRF may be particularly valuable as a clinical vital sign in the peripartum period and during reproductive years. CRF may also prove useful in clinical prediction of maternal morbidity, as adverse cardiovascular outcomes make up a substantial proportion of severe maternal morbidity [37,38]. Finally, our findings suggest that dedicated postpartum cardiovascular rehabilitation after HDP may be worthwhile, using a patient-centred approach sensitive to the needs of new mothers.
In light of evidence for the benefits of physical activity during pregnancy [39], the appropriateness of physical activity in women with established HDP is an area under study. Current GH is a relative contraindication to physical activity, whereas PE is an absolute contraindication [40,41]. Our review was not designed to assess whether current guideline recommendations regarding physical activity for women with HDP are appropriate, and further studies to elucidate the safe CRF threshold customization of physical activity regimens for women with HDP under the guidance of obstetrician-gynecologists or other obstetric care providers, such as kinesiologists, are essential [42].
Our review is subject to several limitations. While the included studies showed high quality ratings with regards to selection of participants and measurement of exposures and covariates, many studies used CRF measurements that did not allow calculation of VO 2 . Accordingly, our ability to combine results from different studies in meta-analysis was limited by the variety of CRF measures employed, as well as by the heterogeneity of study designs. Standardization of CRF measures in perinatal research, including adjustment for BMI, will be essential to understanding the role of CRF in normotensive and hypertensive pregnancy, including its lasting effects in the postpartum period. In addition, most of the included studies did not explicitly aim to measure HDP in relation to CRF and examined HDP as a secondary outcome. Furthermore, the majority of studies examining CRF during pregnancy measured CRF before HDP diagnosis, limiting our ability to draw conclusions about the effects of HDP on CRF in a current pregnancy. Finally, while preeclampsia is understood as a heterogeneous disorder with different phenotypes during pregnancy (i.e., early-onset vs. late-onset) [43], the majority of studies included in this review did not report on the timing or severity of HDP. In order to plan tailored postpartum rehabilitation programs, further studies should evaluate CRF according to the timing and severity of HDP.
HDP often occurs in women with existing comorbidities, such as pregestational diabetes and obesity, which may impact cardiovascular fitness. Women with pre-pregnancy obesity are particularly at elevated risk for PE and CVD and also tend to have reduced CRF. While our analyses focused on weight-adjusted CRF, several of the studies included in our review examined VO 2 without adjustment for maternal weight, which tends to be higher in women with HDP. Additional studies using weight-adjusted VO 2 values before, during, and after HDP are needed to analyze CRF changes with respect to HDP and to shed light on the interplay between HDP and CVD risk factors.
Our analyses may also be subject to confounding by age, which is known to be linked with CRF decline. However, the relatively narrow age range of child-bearing women likely minimizes the importance of this bias. Finally, we were not able to address persistence and duration of hypertension after delivery. However, PE generally normalizes within one to two weeks postpartum in 70% of individuals [44], whereas the earliest postpartum CRF measurements in the included studies were at six weeks.
Despite these limitations, our study was focused on CRF as an actionable risk marker rather than an independent causal factor. Furthermore, our aim was to identify knowledge gaps to guide future research in the prospective evaluation of CRF. Eventually, studies evaluating CRF-guided rehabilitation programs at key periods in women's reproductive trajectory will be needed to contribute evidence-based recommendations for optimal cardiovascular health around the time of pregnancy.

Conclusions
In summary, in the first systematic review to date on CRF before, during, and after HDP suggested that CRF is a valuable marker of perinatal cardiovascular risk, and that postpartum measurement of CRF may shed light on the cardiovascular sequelae of hypertensive pregnancy. Our findings raise the possibility that HDP may impair cardiorespiratory reserve and suggest that dedicated postpartum cardiovascular rehabilitation programs may be indicated. Additional studies using standardized measures of VO 2 are needed to quantify the strength and temporality of the association between CRF and HDP.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/jcm11154364/s1. Figure S1: Mean weight-unadjusted differences in VO 2 max during pregnancy and subsequent development of preeclampsia and/or gestational hypertension; Figure S2: Mean difference in work rate (Watts) during pregnancy and development of preeclampsia and/or gestational hypertension. Acknowledgments: The authors thank Tara Landry and Lindsay Hales for assistance with the literature search.

Conflicts of Interest:
The authors declare no conflict of interest.       (animals or animal or canine* or cat or cats or dog or dogs or feline or hamster* or mice or monkey or monkeys or mouse or murine or pig or pigs or piglet* or porcine or primate* or rabbit* or rats or rat or rodent* or sheep or veterinar*).ti,jw. Appendix B Table A2. References excluded with reasons. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont. Table A2. Cont.