The vascular response to pregnancy has been frequently studied in mesenteric artery models by investigating endothelial cell (EC)- and smooth muscle cell (SMC)-dependent responses to mechanical (flow-mediated vasodilation, myogenic reactivity, and vascular compliance) and pharmacological stimuli (G protein-coupled receptor responses: GqEC, GsSMC, GqSMC). It is unclear to what extent these pathways contribute to normal pregnancy-induced vasodilation across species, strains, and/or gestational age and at which receptor level pregnancy affects the pathways. We performed a meta-analysis on responses to mechanical and pharmacological stimuli associated with pregnancy-induced vasodilation of mesenteric arteries and included 55 (188 responses) out of 398 studies. Most included studies (84%) were performed in Wistar and Sprague-Dawley rats (SDRs) and compared late gestation versus nonpregnant controls (80%). Pregnancy promotes flow-mediated vasodilation in all investigated species. Only in SDRs, pregnancy additionally stimulates both vasodilator GqEC sensitivity (EC50 reduced by −0.76 [−0.92, −0.60] log[M]) and GsSMC sensitivity (EC50 reduced by −0.51 [−0.82, −0.20] log[M]), depresses vasopressor GqSMC sensitivity (EC50 increase in SDRs by 0.23 [0.16, 0.31] log[M]), and enhances arterial compliance. We conclude that 1) pregnancy facilitates flow-mediated vasodilation at term among all investigated species, and the contribution of additional vascular responses is species and strain specific, and 2) late pregnancy mediates vasodilation through changes at the receptor level for the substances tested. The initial steps of vasodilation in early pregnancy remain to be elucidated.
- systematic review
this article is part of a collection on Cardiovascular Regulation in Pregnancy. Other articles appearing in this collection, as well as a full archive of all collections, can be found online at http://ajpheart.physiology.org/.
Vascular adaptation to pregnancy is characterized by marked vasodilation. Vasodilation is reflected by a reduction in systemic vascular resistance. Systemic vascular resistance drops early in gestation, is at its lowest at midgestation, and gradually increases toward term. This is true for both humans (3, 41) and animals (69).
The local vascular mechanisms underlying adaptation to pregnancy have been studied extensively, mainly in animal experiments. They involve both endothelial cell (EC)- and smooth muscle cell (SMC)-dependent changes. The underlying mechanisms are complex and probably gestational age-specific and species- or strain-specific. As a result of such confounding factors, conclusions from various studies may seem inconsistent. Systematic review is a tool to determine qualitative and quantitative responses objectively while taking into account methodological quality, study design, and various confounders. It has been used to evaluate human research on a variety of topics, as collected in the Cochrane database (48). The same technique is applicable to animal experimental work (41).
Adaptation to pregnancy has been studied in several vascular beds and by a variety of stimuli. The mesenteric circulation has been commonly used because this vascular bed is quantitatively important, since it receives one-third of cardiac output in pregnancy (15, 41) and because it is easily accessible for study. The local vascular responses can be arranged by the stimuli imposed or the pathways involved. Mechanical stimuli include flow-mediated vasodilation, myogenic reactivity, and arterial compliance. Electric stimuli concern experiments on electric field stimulation (EFS). Pharmacological stimuli consist of responses to agents such as acetylcholine, norepinephrine, and phenylephrine. As most of these pharmacological stimuli involve G protein-coupled receptors, the response to the agents can be arranged according to their receptor pathways. These include the EC (GqEC) pathway and two SMC (GqSMC and GsSMC) pathways (8, 25, 28, 29, 56, 59, 62), as shown in Fig. 1.
Many authors have generalized the conclusions of their vascular studies from species and/or strain and/or gestational age period to pregnancy in mammals in general. The aim of our study was to evaluate to what extent vascular adaptation to pregnancy is indeed similar across species, strains, and the course of pregnancy. To study this effect, we applied the methods of meta-analysis to experimental data, limiting ourselves for practical reasons to one important vascular bed, that of the mesenteric circulation. The questions to be answered were the following: 1) To what extent do mechanical stimuli and/or pharmacological receptor pathways contribute to normal vasodilation of the mesenteric circulation in pregnancy across species, strains, and/or gestational age? and 2) At which receptor level does pregnancy affect the different pathways in this circulation across species, strains, and/or gestational age?
MATERIALS AND METHODS
We searched the PubMed and Embase databases for (all) original studies on adaptive responses of mesenteric arteries to a first pregnancy, from 1948 (PubMed) and 1980 (Embase) until April 2012. No restriction was used for language or species (studies on humans were included). The search strategies focused on pregnancy, mesenteric arteries, and vasoconstrictor and/or vasodilator responses, as detailed in Tables 1 and 2.
Selection of studies.
From the original studies identified by the literature search, we selected studies qualified for inclusion in the meta-analysis through the three-phase selection process shown in Fig. 2.
In phase 1, all potential studies were screened for inclusion criteria on the basis of title and abstract. The screening was performed independently by two investigators (first and second author). In case of any doubt, the full publication was evaluated. Differences between investigators were resolved by mutual agreement or, in case of remaining disagreement, by a third investigator (M. Leenaars, Central Animal Laboratory, SYRCLE). Inclusion criteria were 1) original data, 2) healthy subjects, 3) first pregnancy, 4) nulliparous controls, 5) age-matched subjects, 6) mesenteric artery vasoconstrictor and/or vasodilator responses, and 7) standard medium used in the experiments. Since some studies did not literally describe information concerning inclusion criteria 3, 4, and 5, studies were included if they were very assumable that they complied.
In phase 2, articles that passed the initial screening were analyzed in full by the same two investigators for the same inclusion criteria.
In phase 3, full articles that passed the inclusion criteria were evaluated for the assessability of the responses, based on the presence of greater of equal to five measurements for each response. From these, we included all responses in the absence of any blockade in the presence of denuded endothelium or blockade of well-known, commonly studied, EC vasodilator mechanisms [blockade of nitric oxide (NO) and/or prostacyclin (PGI2)].
From each article with assessable responses, data were extracted. We recorded primary data on species, strain, gestation age (early, mid-, and late gestation, defined by trimester), estrous/cycle stage, and type of myograph. From each vascular response we recorded the number of subjects, effect size, and standard deviation or standard error of the mean. In case of missing, incomplete, or indeterminate data, an effort was made to complete the data by approaching the authors by e-mail. The response rate to these requests was 46%. In case of duplicate reporting, we used only the results of the most recent publication.
From the pharmacological and electrical responses, we extracted the sensitivity (EC50 = effective concentration, needed to obtain 50% of maximum effect) and maximum effect (Emax = not responding to ≥2 further increases in concentration or intensity). From the mechanical responses, we determined only the direction of the effect (favoring vasodilation, vasoconstriction, or no effect), because quantitative comparison was not feasible. This was caused by the highly variable outcome measures reported in these studies and the use of statistics (often ANOVA) that did not allow calculation of overall effect sizes.
The quality of each study was assessed by two independent investigators (first and second author), based on randomization, blinding of outcome assessment, age or weight of subjects, and number of animals used at study entry and analysis. Subsequently, each response was ranked on three points: 1) clear number of subjects used for analyses, 2) presence of a response graph (based on at least 5 measurements), and 3) achievement of Emax. The overall quality of the responses was expressed as the percentage of responses that fulfilled the criteria compared with the total number of responses in each study.
Values of EC50 and Emax for pharmacological and electrical stimuli were analyzed and displayed in forest plots (Review Manager 5, The Cochrane Collaboration, 2008).
Responses to pharmacological stimuli were grouped together by the type of stimulus and type of G protein-coupled pathway (GqEC, GqSMC, and GsSMC) involved. If they could not be attributed to such pathway, they were arranged by type of stimulus (NO donors, EFS, MgSO4, and potassium). Subgroup analysis was performed to assess species and strain differences and to explore possible causes for heterogeneity. The responses were calculated as weighed mean differences (pregnant minus nonpregnant) for EC50 or as standardized mean differences (pregnant minus nonpregnant) for Emax, depending on whether the variable was uniformly expressed among studies or not. These data were presented as means ± 95% confidence interval. P < 0.05 was considered statistically significant. Responses to mechanical stimuli were analyzed qualitatively, and the direction of combined responses was determined relative to the number of animals.
Variation of the combined responses was expressed as heterogeneity, calculated as I2, under the assumption of a random effects model (Review Manager 5). A value of I2 < 60% represents moderate heterogeneity, and I2 > 60% is substantial to considerable heterogeneity (7).
Publication bias was assessed by the method of Egger et al. (for 10 or more studies) (26) or by the amount of funnel plot asymmetry (for >10 studies).
The robustness of results from meta-analysis was assessed by sensitivity analysis (7). To this effect we repeated all analyses in the presence and absence of the excluded studies to determine the extent to which decisions made during in- and exclusion based on parity or age-matching had any major effect on our results.
Our search identified 398 studies on vascular adaptation to pregnancy, involving the mesenteric vascular bed. From these, 345 studies were not included for the reasons shown in Fig. 2, leaving 55 studies that met the criteria for inclusion in the meta-analysis. Human studies did not meet the criteria for inclusion. In one case double reporting was suspected (37, 38).
The characteristics of the included studies are shown in Table 3. Most studies concerned rodents [rats = 46 Spraque-Dawley rats (SDRs) and Wistar rats (WRs), mice = 5 (C57BL/6J), and guinea pigs = 2 (Hartley)], one study used sheep, and one used pigs. Most of the studies had methodological shortcomings (Table 4). Only 25% (14/55) reported the number of animals used in the study, 53% (29/55) parity; only 4% (2/55) performed randomization, and 0% blinded outcome.
From the 55 included studies with 266 vascular responses, 188 responses fully met the criteria for inclusion in the meta-analysis, as shown in Fig. 2. Stimuli were pharmacological (82%, 154/188), electrical (3%, 6/188), and mechanical (15%, 28/188) and were performed in early (2%, 3/188), mid-(12%, 23/188), and late (86%, 161/188) pregnancy (one undefined pregnancy stage). For pharmacological and electrical responses, EC50 was detectable in 52% (83/160) of cases and Emax in 39% (63/160) and a graph was present in 81% (130/160). For mechanical responses the effect size was quantified in 7% (2/28) and a graph was present in 96% (27/28). Responses were obtained by wire myography (49%, 93/188), pressure myography (18%, 33/188), pressure-perfusion myography (9%, 16/188), and whole organ perfusion (24%, 46/188).
Vascular adaptation to pregnancy, assessed in mesenteric arteries through pharmacological stimuli related to the GqEC-coupled pathway, is shown in Fig. 3. Responses were available for mid- and late gestation, mainly in rats. In midgestation, two studies in SDRs showed no significant effect of acetylcholine or bradykinin on EC50, Emax, or the overall (acetylcholine plus bradykinin) GqEC response. In late gestation, responses differed for SDRs and WRs. In WRs, data from four studies showed no significant effect of pregnancy on EC50 or Emax for any stimulus. In SDRs, the responses from three studies showed considerable heterogeneity for EC50 and Emax. This was largely caused by one whole organ perfusion myography response [Parent et al. (61)], as exclusion of this response markedly reduced heterogeneity. The two other wire myograph studies in SDRs showed a reduction in EC50 (by −0.76 [−0.92, −0.60] log[M]) in late gestation. In guinea pigs, pregnancy slightly reduced EC50 (by −0.34 [−0.47, −0.20] log[M]). This suggests that in SDRs and to a lesser amount in guinea pigs, but not WRs, that GqEC-mediated vasodilation is activated in late pregnancy.
The effect of pregnancy on the GqEC pathway in mesenteric arteries under blockade is shown in Fig. 4. Responses were available for late gestation only in WRs, SDRs, mice, and guinea pigs. In both WRs, SDRs, and mice, blockade of NO and/or PGI2 abolished any GqEC-mediated vasodilation, if present. In guinea pigs, however, the reduction in EC50 in pregnancy persisted under blockade of NO and PGI2. Apparently, in WRs the GqEC pathway is not activated, in SDRs and mice it is activated and NO and PGI2 dependent, and in guinea pigs the pathway is activated and independent of NO and PGI2.
Vasoconstriction through pharmacological stimuli related to the GqSMC-coupled pathway in mesenteric arteries in pregnancy is shown in Figs. 5 and 6. The responses to the various agents showed considerable heterogeneity. The overall GqSMC response at midgestation showed no effect of pregnancy. Responses for late gestation were available in SDRs, WRs, mice, and guinea pigs. The overall GqSMC response suggested downregulation of vasoconstriction, but the responses differed markedly between species and strains. In WRs and guinea pigs, pregnancy did not affect EC50 or Emax for all agents studied, except for one response (32) in WRs of increased sensitivity to norepinephrine. This response caused considerable heterogeneity. Additionally, the data showed funnel plot asymmetry, which may suggest publication bias. SDRs and mice showed reduced vasoconstrictor sensitivity, as demonstrated by the increase in EC50 and no change in Emax for all agents tested. Apparently, in WRs and guinea pigs the GqSMC-mediated vasoconstrictor sensitivity is not affected, whereas in SDRs and mice the pathway is reduced in late pregnancy.
The effect of late pregnancy on the GqSMC pathway in mesenteric arteries under blockade is shown in Fig. 7. Responses were available for late gestation only in SDRs, mice, and guinea pigs, but not in WRs. In these species, blockade of NO, PGI2, or absence of endothelium did not affect EC50 or Emax, except for the responses in SDRs in which EC50 remained elevated (24). This suggests that the downregulation in late pregnancy of the vasoconstrictor GqSMC-mediated pathway is dependent on endothelium-related changes in mice and guinea pigs, but not convincingly in SDRs.
The effect of pregnancy on the SMC relaxing GsSMC pathway in mesenteric arteries is shown in Fig. 8. Responses were available for mid- and late gestation mainly in SDRs, to a limited extent in WRs, and not in other species. Human chorion gonadotrophin (hCG) responses were qualitatively different from all other GsSMC-coupled responses. In SDRs, hCG increased EC50 in the presence of endothelium and reduced EC50 in the absence of it. It seems likely that hCG activates both EC and SMC receptors. For this reason hCG responses were not included in further analysis of the GsSMC pathway. In WRs, vasodilation was upregulated in midgestation, as CGRP was associated with a marked reduction in EC50 and an increase in Emax. Late in pregnancy, isoproterenol increased EC50 and did not affect Emax, indicating less vasodilation compared with controls. In late-pregnant SDRs, adrenomedullin and CGRP reduced EC50, whereas isoproterenol did not. Overall, in late pregnancy, SDRs were more sensitive to GsSMC-coupled vasodilation than nonpregnant controls, whereas WRs were not.
The effect of pregnancy on non-G protein-coupled responses in mesenteric arteries is presented in Figs. 9 and 10. Responses were available for SDRs in mid- and late gestation and for WRs and mice in late gestation. Independent of gestational age, species, or strain, pregnancy did not affect EC50 or Emax to any type of NO donor or potassium. This suggests that pregnancy does not affect SMC sensitivity to EC-dependent stimuli. Single responses showed that vasoconstrictor response to electrical field stimulation (EFS) in late-pregnant WRs reduced Emax (17, 24); in SDR, MgSO4 reduced EC50. This suggests that late pregnancy may affect vascular tone also through non-G protein-coupled responses.
The effect of pregnancy on vascular responses to mechanical stimuli was assessed qualitatively, as shown in Fig. 11. Responses on early gestation were available only in SDRs, responses on midgestation in both WRs and SDRs, whereas qualitative different responses on late gestation were available in WRs, SDRs, and mice. In mid- and late-pregnant WRs, only flow-mediated vasodilation was upregulated. In SDRs, compliance was upregulated in early and midgestation without change in myogenic reactivity. In late-pregnant SDRs, flow-mediated vasodilation and compliance were upregulated, and two out of three studies showed a reduction in myogenic reactivity. One study showed similar results in mice. Single studies on NO blockade showed that flow-mediated vasodilation was inconsistently NO dependent in WRs and not NO dependent in SDRs. In mice, the reduced myogenic reactivity was dependent on NO and independent of PGI2. The data suggest that pregnancy upregulates flow-mediated vasodilation in both WRs and SDRs. SDRs, but not WRs, employ reduced myogenic reactivity and increased compliance.
Robustness of estimates.
Sensitivity analysis was used to assess the robustness of our findings. It showed that the results were not affected by extending the inclusion for studies that did not match the criteria “nulliparity” (n = 11) or “age matching” (n = 18) (data not shown).
Overall effect-size estimates, across species and strains, showed considerable heterogeneity (I2 > 60%). For most responses, this could be reduced to an appropriate effect size (I2 < 60%) only by stratification for stimulus and/or species and/or strain. Stratification by strain was applicable only to rats (SDRs vs. WRs).
Pathways involved in pregnancy-induced vasodilation in mesenteric arteries differ between species and even strains (Table 5). Our meta-analysis showed that the considerable heterogeneity for several overall responses could be reduced to acceptable levels by stratification for species and strain. This was most obvious for the differences between SDRs and WRs. Pregnancy increases flow-mediated vasodilation in both SDRs and WRs. However, while WRs do not use any other mechanism for vasodilation, SDRs additionally increase arterial compliance, activate GqEC and GsSMC receptor-coupled vasodilation pathways, and reduce GqSMC-coupled receptor-mediated vasoconstriction. The differences between the two strains may reflect their vascular health, as SDRs, but not WRs, are known to develop severe vascular dysfunction, including chronic progressive nephropathy, peri-/vasculitis, and chronic cardiomyopathy later in life (internal study by K. Weber, The Researcher, no 28, March 2009, Harlan). Teleologically, one might argue that SDRs must employ the additional mechanisms to cope with the vasodilatory demands of pregnancy, whereas flow-mediated vasodilation suffices in WRs. If that is the case, the WR would be the better model to investigate healthy pregnancy vasodilation, whereas the SDR is the more appropriate model to examine vascular maladaptation.
Vasodilation and the involved mechanisms are gestational age dependent. Data on mechanisms are absent for early pregnancy and limited for midgestation. In mesenteric arteries in midgestation WRs, flow-mediated vasodilation is upregulated, whereas data on the other responses are lacking. In SDRs, GqEC- and GqSMC-coupled pathways and myogenic reactivity are virtually unaffected, whereas GsSMC-coupled vasoconstriction is reduced and arterial compliance is increased. In late gestation WR flow-mediated vasodilation remains increased, whereas the other responses are not affected. In SDRs, GqEC- and GsSMC-coupled vasodilation, flow-mediated vasodilation, and vascular compliance are increased, whereas the GqSMC-coupled vasoconstrictor pathway is downregulated. This suggests that, dependent on the strain, additional vascular responses are elicited with advancing gestation.
Pregnancy enhances relaxation by adaptation of the EC (endothelium-dependent pathways) and the SMC itself (endothelium-independent pathways), depending on species/strain. Endothelium-dependent vasodilator responses are studied through flow and GqEC-related stimuli, which share a common downstream pathway (51, 59). Our meta-analysis showed that endothelium-mediated vasodilation in pregnancy depends on increased production of NO and PGI2 by the EC, as proposed by Carbillon et al. (9), whereas SMC sensitivity to these substances remains unaltered. The effect of pregnancy on SMC itself is studied by deactivation of the EC. Under NO, PGI2 blockade, or endothelium denudation, pregnancy reduces the vasoconstrictor GqSMC-coupled pathway and activates the vasodilator GsSMC-coupled pathway, whereas myogenic reactivity is not consistently affected. These data suggest that pregnancy affects the endothelium-dependent pathways and, dependent on the species/strain, also the endothelium-independent pathways.
It is unknown how pregnancy influences the endothelium-dependent and -independent responses. For several vasoactive substances, the prereceptor level is unaffected (9). Our meta-analysis shows that pregnancy stimulates the GqEC pathway and blunts the GqSMC pathway, whereas they share similar downstream pathways. This suggests that, for the substances tested, adaptation takes place at the receptor level and not at the pre- or postreceptor level. However, this does not exclude the possibility that other substances known to be related to pregnancy-induced vasodilation, for example relaxin or progesterone (18, 34), might be involved at the prereceptor by acting as a facilitator or inhibitor for the reported pathways.
Our meta-analysis aimed to provide insight in the complexity and heterogeneity across species and strains concerning vascular responses involved in pregnancy-related vasodilation. Several methodological aspects deserve consideration. First, many studies (87%), including six in humans (1, 4–6, 77, 78), did not meet the selection criteria shown in Fig. 2. As a result, the meta-analysis is based on responses in rodents [rats (WRs and SDRs), mice (C57BL/6J), and guinea pigs (Hartley)] only. In addition, several of the results showed considerable heterogeneity that resulted mainly from differences in species and strains. Therefore, the findings should not be extrapolated beyond the limitations of these species/strains. Second, we combined responses to an overall G protein-coupled pathway response, as stimuli that share a common receptor complex induce comparable effects. However, combining results across stimuli, species, and strains may distort reality, especially when responses are qualitatively different. Therefore, the overall effects should be interpreted with caution. Third, publication bias may have affected overall results across species/strains. Most notably, SDRs may have been overrepresented in the meta-analysis. It was the most commonly used animal, possibly due to profound activation of multiple pathways, but not necessarily the best representative of normal pregnancy in general. SDR is a strain known to have vascular dysfunction, and SDRs are prone to vascular disease later in life. Our meta-analysis shows that SDRs employ additional mechanisms to realize the same degree of vascular adaptation to pregnancy, compared with WRs. Therefore, conclusions based on SDRs should not be interpreted as representative of healthy pregnancy in general. Fourth, differences in the degree of precontraction between experiments may have affected flow-mediated vasodilation, GqEC-mediated vasodilation, and myogenic reactivity. This effect is likely to be small because the degree of precontraction apparently has little effect on EC50 (76). Fifth, most animal experimental studies included in the analysis did not meet the stringent criteria for methodological quality used in systematic reviews of human trials, as details on randomization, blinding, withdrawals, and dropouts were often lacking. This may have affected our results in an unpredictable manner. We recommend that stringent methodological criteria are implemented in animal experimental studies through standardized procedures and report guidelines as previously advocated (40). Sixth, the overall effects should be interpreted with some caution, as the number of studies per subgroup is limited and overall estimates suffer from heterogeneity, this despite efforts to reduce heterogeneity by using a random effects model, performing subgroup analysis and sensitivity analysis.
In conclusion, our meta-analysis shows that in the mesenteric arterial bed, pregnancy increases flow-mediated vasodilation across the rat strains tested, whereas activation of additional pathways is species, strain, and gestational age specific. Therefore, results on vascular adaptation to pregnancy should not be generalized lightly beyond the limits of the study. Second, late pregnancy mediates vasodilation through changes at the receptor level for the substances tested. It remains a challenge to decipher the initial steps of vasodilation in early pregnancy.
No conflicts of interest, financial or otherwise, are declared by the author(s).
J.v.D. conception and design of research; J.v.D. and C.R.H. performed experiments; J.v.D. analyzed data; J.v.D. and C.R.H. interpreted results of experiments; J.v.D. prepared figures; J.v.D. and C.R.H. drafted manuscript; J.v.D., C.R.H., F.K.L., P.S., and M.E.S. edited and revised manuscript; J.v.D., C.R.H., F.K.L., P.S., and M.E.S. approved final version of manuscript.
We thank Alice H. J. Tillema, from the Medical Library of the Radboud University Nijmegen, for assistance with the search strategy.
- Copyright © 2012 the American Physiological Society