Maternal endothelial function, circulating endothelial cells, and endothelial progenitor cells in pregnancies conceived with or without in vitro fertilization

Abstract

In women who conceived with or without assisted reproduction, we evaluated endothelial function by EndoPAT [reactive hyperemia index (RHI)], circulating numbers of endothelial cells (CEC) and endothelial progenitor cells (EPC), and their function before during and after pregnancy. In vitro fertilization (IVF) pregnancies were stratified by method of conception and corpus luteum (CL) number—controlled ovarian stimulation (>1 CL) or programmed (0 CL) cycles and spontaneous singleton pregnancies (1 CL). We observed 1) comparable gestational decline of RHI in the three participant groups secondary to gestational rise of baseline preocclusion pulse-wave amplitude (PWA) incorporated into the RHI calculation by EndoPAT software; 2) progressive rise in “normalized” RHI throughout pregnancy (calculated by substituting prepregnancy baseline preocclusion PWA into the RHI equation), greater in spontaneous conception vs. IVF cohorts; 3) similar gestational increase of maximum PWA and time to maximum PWA after the ischemia stimulus among the three participant groups; 4) modest gestational increase of ischemia response (reactive hyperemia) in the spontaneous conception group and no change or significant decline, respectively, in women who conceived using programmed or controlled ovarian stimulation cycles; 5) enhanced basal nitric oxide production by early (primitive) outgrowth EPC during pregnancy in women who conceived spontaneously, but not through IVF; and 6) gestational increase in CEC in all three participant cohorts, more pronounced in women who conceived by IVF using programmed cycles. On balance, the evidence supported enhanced endothelial function during pregnancy in spontaneous conceptions but less so in IVF pregnancies using either controlled ovarian stimulation or programmed cycles.

INTRODUCTION

Human pregnancy is accompanied by hemodynamic changes that include a profound decrease in systemic vascular resistance and rise in cardiac output, particularly during the first trimester (4, 6, 26). Generally speaking, vascular tone is regulated by numerous vasoactive substances, including endothelium-derived relaxing factor or nitric oxide (NO), which may contribute to the remarkable vasodilation of pregnancy (31). In contrast, endothelial dysfunction has been implicated in the pathogenesis of preeclampsia, mediated at least in part by reduced NO bioavailability, which hinders vasodilation, thereby contributing to hypertension, organ hypoperfusion, and other disease manifestations (25). The endothelial dysfunction of preeclampsia may also have remote adverse health consequences, insofar as women who suffered the disease are at increased risk for developing hypertension, atherosclerosis, and other cardiovascular diseases later in life (1). Moreover, they show persistent endothelial dysfunction as reflected by aberrant vascular response to shear stress and reduced arterial compliance (22).

Women conceiving by in vitro fertilization (IVF) are at increased risk for adverse pregnancy outcomes, in particular hypertensive disorders of pregnancy and preeclampsia. Previous investigations specifically found programmed (artificial) IVF cycles to be associated with the enhanced risk (reviewed in Ref. 7). More recently, the absence of the corpus luteum (and circulating CL products, e.g., relaxin) during pregnancy, which fails to develop during programmed cycles due to hypothalamic-pituitary suppression, was specifically implicated (34). That is, women conceiving by autologous frozen embryo transfers in programmed IVF cycles (0 CL) were at increased risk for developing preeclampsia and preeclampsia with severe features compared with women conceiving by autologous frozen embryo transfers in a natural cycle or spontaneously conceived pregnancies (1 CL), and with women who underwent fresh autologous embryo transfer in controlled ovarian stimulation cycles (>1 CL; see Refs. 18 and 34). Furthermore, women conceiving using programmed cycles demonstrated significant impairment of the cardiovascular changes that normally transpire during the first trimester compared with the ovarian stimulation cohort, which was similar to spontaneous conceptions (11, 33, 34). One possible explanation for this deficit in women who conceived using programmed cycles with absent CL is the lack of circulating relaxin and its vasodilatory influence on vasculature that is in part mediated through NO (8, 9).

Endothelial health can be noninvasively assessed by the technique of flow-mediated dilation (FMD), which reflects NO bioavailability (3, 35). When examining healthy pregnancies, several studies demonstrated that FMD following reactive hyperemia was significantly increased in pregnant compared with nonpregnant women (13, 14, 27, 32). FMD rose during early pregnancy and reached a peak in the third trimester, thereafter reaching a plateau or decreasing toward nonpregnant or postpartum levels (13, 28, 29). However, some investigators did not observe changes in FMD during normal pregnancy (21, 36), or they reported decreases (2, 19, 23).

Whereas most studies of endothelial function during pregnancy have used ultrasound to measure the change in brachial artery diameter after transient ischemia (FMD), the relatively newer technology of EndoPAT has been suggested to be a more convenient and reproducible method of determining endothelial function from arterial pulsatile volume changes in the finger after transient ischemia, thereby negating the necessity of ultrasonography and associated training (reviewed in Refs. 2 and 15). Endothelial health is assessed using the reactive hyperemia index (RHI), quantified as the ratio of the postocclusion to preocclusion peripheral arterial tonometry (PAT) signal amplitude of the tested arm normalized to the control arm. Measurement of the control arm ensures normalization of confounding factors such as autonomic nervous system influence (2, 15).

However, EndoPAT was developed in the nonpregnant population, and its use in pregnancy has been less thoroughly investigated. In one study, endothelial function measured by EndoPAT was assessed in the same pregnant women at gestational weeks 16 and 28. There was a statistically significant increase in baseline PAT signal amplitude (i.e., before occlusion) from weeks 16 to 28 reflecting peripheral vasodilation (2). However, the associated RHI values generated by EndoPAT after transient ischemia in this study demonstrated a significant decrease during the same time period. These findings were corroborated in a parallel study using FMD, in which baseline brachial artery diameter increased from 16 to 28 gestational weeks, while FMD declined. The authors concluded that EndoPAT and FMD may be unreliable for evaluating changes in endothelial function during pregnancy, because the marked rise in baseline vasodilation before the ischemia stimulus, presumably itself due to enhanced endothelial function, may limit or preclude further increases after ischemia (“ceiling effect”; see Ref. 2). In another study, Fujita and colleagues (15) reported that RHI progressively deteriorated throughout normal pregnancy. Although one interpretation is that endothelial function deteriorated with advancing pregnancy as concluded by these investigators, another is that increasing baseline vasodilation undermined the RHI as a meaningful reflection of endothelial function in pregnancy [as advanced by Carty and colleagues (2), vide supra].

The objectives of the present investigation were threefold: first, to revisit the use of EndoPAT in uncomplicated pregnancies to ascertain whether RHI indeed deteriorates as reported by others (vide supra) and, if so, to further explore potential explanations for this apparent paradox; second, to apply EndoPAT to pregnancies conceived by IVF in which we previously documented widespread dysregulation of maternal cardiovascular function during the first trimester and elevated preeclampsia risk in those IVF protocols involving programmed cycles that precluded formation of a corpus luteum (11, 34). More specifically, the aim here was to determine whether the previously documented dysregulation of cardiovascular function and increased preeclampsia risk in the women conceiving using programmed IVF cycles without a CL would be accompanied by endothelial dysfunction as assessed by EndoPAT; finally, we measured circulating endothelial and endothelial progenitor cells as correlates of endothelial cell damage and repair, respectively (5, 17). Both colony formation and flow cytometry were employed as two established methodologies for assessing endothelial progenitor cells.

METHODS

Study Participants

After we obtained written informed consent in accordance with the Declaration of Helsinki, study participants were enrolled in this study approved by the University of Florida Institutional Review Board. Women planning to conceive by IVF were identified by Reproductive Endocrinology and Infertility physicians at the University of Florida. The women planning to become pregnant without IVF were recruited by advertisement. The three participant groups were women conceiving 1) without IVF (singleton pregnancies/single CL); 2) by transfer of frozen embryo(s) generated using donor or autologous eggs or fresh donor embryo transfer in programmed (artificial) cycles (absent CL); or 3) by IVF and fresh embryo transfer after controlled ovarian stimulation (multiple CL).

In brief, baseline measurements were performed in the absence of circulating CL factors before pregnancy (follicular phase or last 4 days of leuprolide suppression). Next, participants were evaluated six times during pregnancy (5.8 ± 0.1, 8.3 ± 0.1, 11.6 ± 0.1, 15.2 ± 0.1, 24.0 ± 0.1, and 33.6 ± 0.1 wk of gestation) and finally 45.9 ± 1.8 wk postdelivery (mean ± SE). The number of women investigated at each study visit is presented in Supplemental Tables S4 and S5 (Supplemental data for this article can be found at https://doi.org/10.6084/m9.figshare.11709984) containing the mean ± SE of all measured variables. Women included in the analysis of endothelial function by Digital Plethysmography (EndoPAT) had at least three pregnancy visits, in addition to the prepregnancy baseline, while those women with at least one pregnancy visit, in addition to the prepregnancy baseline, were included for the analysis of Circulating Endothelial Cells and Endothelial Progenitor Cells. After overnight fast and abstinence from alcohol, caffeinated products, and pain medications during the preceding 24 h, the women reported to the Clinical Research Center at ~8:00 AM where they participated in a battery of physiological assessments over an ~5-h period, including EndoPAT, and underwent phlebotomy for fasting blood samples.

Routine controlled ovarian stimulation protocols were used for fresh IVF cycles. Identical programmed IVF cycle protocols were implemented for donor oocyte-derived fresh or frozen embryo transfer and autologous oocyte-derived frozen embryo transfer. In the fresh IVF cycles involving controlled ovarian stimulation, the number of CL was estimated by the number of retrieved eggs. Absent CL in the programmed cycles was verified by ultrasound, and by undetectable concentrations of circulating relaxin at all time points during pregnancy (10). For further details about the IVF protocols, see Supplemental Table S1.

EndoPAT

We evaluated 19–24 participants in each cohort who had at least four study visits, including the prepregnancy baseline. Peripheral arterial plethysmography was performed using the Endo-PAT2000 device (Itamar Corporation) after Yinon et al. (37). All tests were performed in a quiet low-light environment at ~22°C. Briefly, a pneumatic finger probe was placed on the index finger of both hands, and a manual blood pressure cuff was placed on the nondominant arm in preparation for occlusion. After a period of at least 15 min of supine rest, PAT was recorded for 5 min at baseline (preocclusion), during brachial occlusion of one arm with an inflated cuff (5 min), and finally after occlusion, continuously for ≥5 min (postocclusion). Simultaneous measurements were made in both the occluded and control arms. The RHI or Endoscore derived from software of the Endo-PAT2000 device represented the ratio between post- and preocclusion PAT signal amplitude adjusted for the ratio recorded in the nonoccluded arm.

The formula used by the EndoPAT software to convert the PAT signal amplitude to the Reactive Hyperemic Index (RHI; Supplemental Fig. S1) is the following:

$$\text{RHI}=\text{indexed\hspace{0.17em}PAT\hspace{0.17em}ratio}\times \text{baseline\hspace{0.17em}correction\hspace{0.17em}factor}$$ (1)

$$\text{Indexed\hspace{0.17em}PAT\hspace{0.17em}ratio}=\left(\text{RH\hspace{0.17em}occluded\hspace{0.17em}arm\hspace{0.17em}}90-120\text{\hspace{0.17em}s}+\text{RH\hspace{0.17em}occluded\hspace{0.17em}arm\hspace{0.17em}}120-150\text{\hspace{0.17em}s}\right)/\left(\text{RH\hspace{0.17em}control\hspace{0.17em}arm\hspace{0.17em}}90-120\text{\hspace{0.17em}s}+\text{RH\hspace{0.17em}control\hspace{0.17em}arm\hspace{0.17em}}120-150\text{\hspace{0.17em}s}\right)$$ (2)

$$\text{RH}=\text{mean\hspace{0.17em}pulse-wave\hspace{0.17em}amplitude\hspace{0.17em}postocclusion}/\text{mean\hspace{0.17em}pulse-wave\hspace{0.17em}amplitude\hspace{0.17em}preocclusion}\left(\text{baseline}\right)$$ (3)

$$\text{Baseline\hspace{0.17em}correction\hspace{0.17em}factor}=\left[\left(0.25\times {\text{log}}_{\text{3}}\text{\hspace{0.17em}mean\hspace{0.17em}occluded\hspace{0.17em}arm\hspace{0.17em}preocclusion\hspace{0.17em}amplitude}\right)-0.2\right]$$ (4)

We observed a decrease in RHI throughout pregnancy (results) as previously reported by other investigators (introduction) and determined that this paradoxical finding was mostly likely because of the large increase in baseline preocclusion PWA (as presented in results), which is in the denominator of the equation used to calculate RH (Eq. 3) that, in turn, is a variable used in the calculation of the indexed PAT ratio (Eq. 2) and RHI (Eq. 1). To underscore this point, we substituted the prepregnant baseline preocclusion PWA (or prepregnant pre- and postocclusion PWA in the case of the control arm) in the RH calculation (Eq. 3) and baseline correction factor (Eq. 4) at each stage of pregnancy, ultimately yielding “normalized” RHI (Supplemental Fig. S1). It is important to point out, however, that the normalized RHI was no longer a measure of the response to ischemia alone but also reflected the progressive rise in baseline preocclusion PWA during pregnancy. To more precisely quantify the reactive hyperemia or ischemia response, we recorded the maximum PWA in the occluded arm, which reflected the summation of baseline preocclusion amplitude and the ischemia response. Therefore,

$$\text{Ischemia\hspace{0.17em}response}=\text{maximum\hspace{0.17em}postocclusion\hspace{0.17em}PWA\hspace{0.17em}in\hspace{0.17em}occluded\hspace{0.17em}arm}–\text{maximum\hspace{0.17em}postocclusion\hspace{0.17em}PWA\hspace{0.17em}in\hspace{0.17em}control\hspace{0.17em}arm}$$ (5)

where postocclusion PWA in the control arm takes into account baseline preocclusion PWA and potential confounding influences, e.g., autonomic activity consequent to the 5-min arm occlusion.

Enumeration of Circulating Endothelial Cells from Peripheral Blood

Because evaluation of late (outgrowth) endothelial progenitor cells (EPC) and measurement of baseline NO obtained from early (primitive) EPCs as described below were not initiated until later in our study, we liberalized the inclusion criteria for all cell analyses [i.e., circulating endothelial cells (CEC), EPC colonies, and flow cytometry] to include participants who had at least two visits, including the baseline prepregnancy visit, to increase the number of participants in the analyses.

CEC were isolated from peripheral blood samples using immunomagnetic Dynabeads (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. Briefly, peripheral blood mononuclear cells (PBMCs) were prepared from 1.0 ml of blood after eliminating red blood cells (RBC) with RBC Lysing Solution (BD Biosciences). The PBMCs were then diluted with phosphate-buffered saline (PBS) supplemented with 2 mM EDTA and centrifuged for 10 min at 300 g. After a wash, the cell pellet was resuspended and then incubated with magnetic beads coated with rabbit monoclonal anti-CD146 antibody (BD Biosciences) for 60 min at 4°C with rotation. The beads were then washed three times with PBS, and the cell-bead conjugates were next deposited on glass slides using a Cytospin centrifuge (250 g for 5 min). Last, the cells were fixed with 4% paraformaldehyde for 15 min and stained with propidium iodide and DAPI. The number of CEC was counted using a Fluorescent microscope.

Isolation and Culture of Human EPC

Early (primitive) EPC.

Early EPC were cultured from PBMCs isolated from 8 ml of peripheral blood using cell preparation tubes with heparin (BD Biosciences). After centrifugation at room temperature in a swinging bucket rotor for 20 min at 1,500 g, the mononuclear cells were diluted with PBS supplemented with 2% FBS. The cells were centrifuged for 15 min at 300 g and then plated on fibronectin-coated six-well dishes and incubated at 37°C in humidified 5% CO₂. After 2 days, the nonadherent cells were replated in 12-well dishes at a concentration of 10⁶ cells/well. At day 5, the appearance of the colonies was observed, and the cells were supplemented with fresh EGM-2 medium (Lonza Switzerland). At day 7, the colonies were counted under a light microscope. The endothelial phenotype of the cells was confirmed by staining with fluorescently labeled acetylated low-density lipoprotein (Biomedical Technologies) and FITC-conjugated Ulex europaeus agglutinin-1, as well as by detecting expression of mRNA for endothelial nitric oxide synthase and live imaging of NO production.

Late (outgrowth) EPC.

Late-outgrowth EPC were cultured from PBMCs isolated from 32 ml of peripheral blood. After 4 days, nonadherent cells were removed by aspiration and washing. The adherent cells were then supplemented with fresh EGM-2 medium. The medium was changed every 3 days during culture. Colonies generally appeared after 18–21 days in culture. At day 21, they were counted, and the endothelial phenotype of cells was confirmed as described above.

Baseline NO production by early (primitive) EPC.

Early (primitive) EPC were cultured on 35-mm dishes with glass-bottom inserts (MatTek). Bioavailable NO was determined as previously described (20). Briefly, EPCs were incubated with 5 μM 4-amino-5-methylamino-2,7-difluzorofluorescein (DAF-FM) diacetate (Invitrogen) for 30 min at 37°C in the dark. Excess extracellular probe was removed by being washed in Hank’s balanced salt solution followed by incubation for 10 min at room temperature to allow for probe deesterification. DAF-FM fluorescence increases by ~160-fold when it reacts with NO. Green fluorescence was measured using an Axiovert 200 (Carl Zeiss) inverted microscope equipped with a charge-coupled device camera and AxioVision (version 4.5) image acquisition/analysis software. Images were acquired every 1 min for at least 20 min, and fluorescence intensity was measured in 20–30 cells/field in at least 6 fields/experiment. Alternatively, imaging was performed using a confocal microscope at excitation and emission maxima of 495 and 515 nm, respectively. Intensity of fluorescence was quantified using LSM 510 (version 3.0 SP3) software for the Carl Zeiss Laser Scanning Microscope.

Flow Cytometry

Whole blood (0.5 ml) was probed for CD34, CD45, CXCR4, VEGFR2, and CD133 using the following antibodies (all at a 1:100 dilution): PerCP-conjugated CD34-percep (BD Biosciences, Franklin Lakes, NJ), FITC-conjugated CD45 (Miltenyi Biotec, Bergisch Gladbach, Germany), phycoerythrin (PE)-conjugated CXCR4 (eBioscience), allophycocyanin-conjugated VEGFR2 (R&D, Minneapolis, MN), and phycoerythrin (PE)-conjugated CD133 (Miltenyi Biotec). Antibodies and whole blood were incubated for 30 min in the dark at room temperature. Next, 2.0 ml of freshly made RBC lysis buffer (BD Biosciences, Franklin Lakes, NJ) were added, and, after incubation at room temperature for 5 min, the whole blood was centrifuged at 500 g for 5 min. The cell pellet was washed in 2.0 ml Dulbecco's PBS containing 5% fetal bovine serum, 5 mmol/L EDTA, and 0.1% sodium azide (Sigma-Aldrich, St. Louis, MO); resuspended in 200 μL of PBS-0.5% BSA-NaN₃-formaldehyde; and maintained at 4°C until FACS analysis. Analysis was performed on a BD FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ) using the manufacturer's recommended isotype control antibodies. A total of 100,000 events/sample were acquired, and the data were analyzed with FCS data analysis software version 4 (De NOVO Software, Pasadena, CA).

Statistics

Numerical values for participant demographic and clinical characteristics, primary infertility diagnoses, adverse pregnancy outcomes, and neonatal characteristics were summarized by means ± SE [range]. All other data were expressed as means ± SE of the absolute values or as percent change: minus baseline [before pregnancy (BP)] divided by baseline. Area under the curve was calculated from BP to weeks 32–35 (or to weeks 14–16 for ischemia response and CEC) with missing data extrapolated linearly. Nonparametric ANOVA (Kruskal-Wallis) test was used to compare numerical variables between two or more groups. Wilcoxon signed-rank test with Bonferroni correction was used to compare numerical variables for each of the three cohorts, separately, between subsequent time points and the prepregnancy baseline (BP). Categorical characteristics were described using frequencies (proportions) and compared between groups using the χ² or Fisher’s exact test (if an expected cell count was <5). Linear mixed model with cubic splines for gestation weeks was used to model the trajectory for the absolute values before and during pregnancy. The full model included the main effects of time and group and group by time interactions. Linear mixed models account for the correlation between repeated measurements made in the same women at multiple time points and include random effects to model the correlation and allow for missing observations to occur during the study. Statistical significance was set at the 0.05 level. SAS (9.4) and R (3.5.1) software were used for the analysis.

RESULTS

Participant Characteristics, Infertility Diagnoses, Obstetrical and Neonatal Outcomes

History of hypertensive disease of pregnancy and maternal smoking, as well as parity, race, and ethnicity were similar among the three cohorts. Maternal age and body mass index (BMI) were 4 yr and 3–4 kg/m² greater, respectively, in the 0 CL compared with the multiple and single CL cohorts (both P < 0.05, Table 1). Diminished ovarian reserve and male factor infertility were more frequent in the 0 CL and >1 CL cohorts, respectively (Supplemental Table S1). Although the study was not powered to detect statistical differences in obstetrical outcomes, more pathologic obstetrical outcomes occurred in women conceiving by IVF, which reached significance for preterm delivery (Supplemental Table S2). The number of twin pregnancies was significantly greater and gestational age at delivery and newborn weight significantly less in IVF compared with spontaneously conceived pregnancies (Supplemental Table S3). (Because the participant characteristics, infertility diagnoses, obstetrical and neonatal outcomes for the inclusion criteria of at least 2 or 4 study visits, including the prepregnancy baseline were similar, only the results for the latter are reported.)

Table 1.

Participant demographic and clinical characteristics

	0 CL	1 CL	Multiple CL	P Value
n	24	22	19
Maternal age at 5–6 wk, yr	36 ± 1 [29–46]	32 ± 1 [23–45]	32 ± 1 [26–41]	0.03
BMI prepregnancy, kg/m2	27 ± 1 [18–36]	24 ± 1 [18–43]	23 ± 1 [18–38]	0.02
Body wt prepregnancy, kg	74 ± 3 [47–101]	66 ± 3 [44–115]	65 ± 3 [48–100]	0.16
Height, cm	165 ± 2 [154–183]	165 ± 1 [152–172]	167 ± 2 [157–182]	0.87
Participant race, n (%)
White	20 (83)	20 (91)	16 (84)	0.54
American Indian or Alaska Native	0	0	0
Asian	2 (8)	1 (5)	0
Black or African American	2 (8)	1 (5)	1 (5)
Other			2 (10)
Nulliparity, n (%)	11 (46)	11 (50)	12 (63)	0.55
Maternal smoking, n (%)	2 (10)	1 (4)	0	0.77
Hypertensive disease in prior pregnancy, n	0	0	0	NA

Values are means ± SE with ranges in square brackets; n, no. of subjects. CL, corpus luteum; BMI, body mass index; NA, not applicable.

EndoPAT

RHI as reported by EndoPAT (Eq. 1 in methods) progressively decreased with advancing gestational age, which was comparable among the three participant cohorts (main effect of time: P < 0.02; Fig. 1A and Supplemental Table S4). The percent change of RHI showed a similar pattern of change (Fig. 1B). The prepregnancy baseline PWA in both control and occluded arms (i.e., before occlusion) was significantly higher for the two IVF cohorts (which themselves were similar) compared with the spontaneous conception group (both P < 0.002; Fig. 2, A and B, left, and Supplemental Table S4). Thereafter, the absolute levels of baseline preocclusion PWA progressively rose during pregnancy in a similar manner among the three cohorts but mostly after the first trimester (P < 0.001; Supplemental Table S4). Because the absolute level before pregnancy of baseline (preocclusion) PWA was higher in the two IVF cohorts (Fig. 2A and B; left panels vide supra), the percent increase during pregnancy was significantly lower at several gestational time points relative to spontaneous conceptions in both control and occluded arms (Fig. 2, A and B, right).

Fig. 1.

Reactive hyperemia index (RHI). Absolute (A) and %change (B) in RHI before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: significant effect of time during pregnancy for each participant cohort: P < 0.02. BP, before pregnancy; PP, postpartum; n, no. of subjects.

Fig. 2.

A: baseline amplitude [arbitrary units (AU)] in control arm. Baseline preocclusion pulse wave amplitude of the control arm before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization (IVF). Left, absolute baseline pulse wave amplitude (PWA) in AU; **P < 0.001 for 1 CL vs. 0 CL and >1 CL. Right, %change in baseline PWA; †P < 0.05 and *P < 0.005 for 1 CL vs. 0 CL and >1 CL. Area under the curves for 1 CL vs. 0 CL and >1 CL through gestational age 32–35 wk: P < 0.01. BP, before pregnancy; PP, postpartum; n, no. of subjects. B: baseline amplitude (AU) in occluded arm. Baseline preocclusion PWA of the occluded arm before, during, and after pregnancy in women conceiving with (0 or >1 CL) or without (1 CL) IVF. Left, absolute baseline PWA in AU; **P < 0.001 for 1 CL vs. 0 CL and > 1 CL. Right, %change in baseline PWA; †P < 0.05, *P < 0.01, and **P < 0.001 for 1 CL vs. 0 CL and >1 CL. Area under the curves for 1 CL vs. 0 CL and >1 CL through gestational age 32–35 wk: P < 0.001.

Because baseline preocclusion PWA was incorporated in the derivation of RH that, in turn, was used in the calculation of RHI (Eqs. 1–3 in methods), the progressive rise in baseline preocclusion PWA during pregnancy was a major factor contributing to the steady decline in RHI (Figs. 1 and 2). To underscore this point, we adjusted for the pregnancy rise in baseline preocclusion PWA by substituting the prepregnancy baseline preocclusion PWA of each participant into the equation and designated the new dependent variable as normalized RHI (Eq. 4 in methods and Supplemental Fig. S1). Normalized RHI rose progressively throughout gestation in all three participant cohorts as anticipated, reflecting the sum of the baseline preocclusion PWA and the reactive hyperemia or ischemia response at each time point (main effect of time: P < 0.001; Fig. 3A and Supplemental Table S4). The rate of rise, which again mostly transpired after the first trimester, was greatest in the spontaneous conceptions, intermediate in the IVF cohort with 0 CL (programmed cycles), and lowest in the >1 CL cohort (controlled ovarian stimulation cycles; group × time P = 0.0125; Fig. 3A). The percent change of normalized “RHI” followed a similar pattern (Fig. 3B). After omitting pathologic obstetrical outcomes or twins from the IVF cohorts, the patterns of change in RHI and “normalized RHI” were comparable to those observed in the analysis including all participants (data not shown).

Fig. 3.

Normalized reactive hyperemia index. Absolute (A) and %change (B) in the “normalized” reactive hyperemia index before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: main effect of time during pregnancy: P < 0.001; group × time interaction, P < 0.0125; *P < 0.008 vs. before pregnancy (BP). B: †P < 0.05 and **P < 0.005 for 0 CL vs. 1 CL vs. >1 CL. Area under the curves for 1 CL vs. 0 CL vs. >1 CL through gestational age 32–35 wk: P < 0.005. PP, postpartum; n, no. of subjects.

We also identified and recorded the maximum PWA reached in the occluded arm after deflation of the cuff, which represented the sum of the baseline preocclusion PWA and the reactive hyperemia or ischemia response at each visit (Eq. 5 in methods). We observed that the prepregnancy levels of the maximum PWA were significantly higher in the two IVF groups (which themselves were similar) relative to the spontaneous conception cohort (P < 0.001; Fig. 4A and Supplemental Table S4), consistent with the higher baseline preocclusion PWA noted above (Fig. 2, A and B). Thereafter, the rate of rise in absolute maximum PWA was similar among the three cohorts. Because of the higher prepregnancy maximum PWA in the two IVF groups, the percent increase was significantly lower during pregnancy relative to spontaneous conceptions at several time points (Fig. 4B).

Fig. 4.

Maximum amplitude [arbitrary units (AU)] in occluded arm. Absolute (A) and %change (B) in the maximum pulse wave amplitude before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: **P < 0.001 for 1 CL vs. 0 CL and > 1 CL. B: †P < 0.05, *P < 0.01, and **P < 0.001 for 1 CL vs. 0 CL vs. >1 CL. Area under the curves for 1 CL vs. 0 CL vs. >1 CL through gestational age 32–35 wk: P < 0.001. BP, before pregnancy; PP, postpartum; n, no. of subjects.

By subtracting the postocclusion PWA of the control arm from the maximum PWA postocclusion in the occluded arm, we were able to estimate the contribution of the reactive hyperemia or ischemia response to the maximum PWA after release of the occlusion (Eq. 5 in methods). Although the group × time interaction was not significant in the full mixed model, using a reduced mixed model, a significant overall main effect of time was observed (P = 0.0463; Fig. 5A and Supplemental Table S4). Notably, in contrast to the 0 (programmed IVF cycles) and 1 (spontaneously conceived pregnancies) CL cohorts, absolute values of ischemia response fell in the >1 CL cohort during the first trimester (controlled ovarian stimulation; P < 0.013 vs. before pregnancy, Fig. 5A). Moreover, the percent decrease of the ischemia response in the >1 CL cohort was significantly lower than the 1 CL cohort at several pregnancy time points (Fig. 5B). After omitting pathological obstetrical outcomes or twins from the IVF cohorts, the patterns of change in the ischemia responses were similar to those observed in the analysis including all participants (data not shown).

Fig. 5.

Ischemia response. Absolute (A) and %change (B) in the reactive hyperemia or ischemia response before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: main effect of time during pregnancy: P < 0.05. *P < 0.013 vs. before pregnancy (BP). B: †P < 0.05 and *P < 0.01 for 1 CL vs. >1 CL. Area under the curves for 1 CL vs. >1 CL through gestational weeks 14–16: P = 0.011. PP, postpartum. AU, arbitrary units; n, no. of subjects.

There were no significant differences in the time to maximum ischemia response among the three cohorts (Supplemental Fig. S2, A and B and Supplemental Table S4). However, there was a significant main effect of time (P < 0.001; Supplemental Fig. S2A).

Circulating Endothelial Cells

In all three participant cohorts, the number of CEC rose significantly during the first and early second trimesters, reaching a plateau thereafter (main effect of time: P < 0.001; Fig. 6A and Supplemental Table S5). Although the variation in percent change of CEC was considerable, the 0 CL cohort showed a greater rise relative to the other two participant cohorts that was of borderline significance for both gestational weeks 5–6 and 14–16 (P = 0.07; Fig. 6B).

Fig. 6.

Circulating endothelial cells (per ml blood). Absolute (A) and %change (B) in the circulating endothelial cells before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: main effect of time during pregnancy: P < 0.001. *P < 0.008 vs. before pregnancy (BP). B: †P < 0.07 for 0 CL vs. 1 CL and >1 CL. Area under the curves for 1 CL vs. 0 CL and >1 CL through gestational weeks 14–16: P = 0.085. PP, postpartum; n, no. of subjects.

Endothelial Progenitor Cells

Early (primitive) EPC.

Circulating numbers of early (primitive) EPC generally increased during pregnancy when analyzing all three participant cohorts (main effect of time: P < 0.05; Fig. 7A and Supplemental Table S5). However, there were no significant differences in the absolute or percent change among the three cohorts (Fig. 7B). Interestingly, EPC numbers were significantly higher postpartum compared with before pregnancy in the three participant cohorts (all P = 0.04 by Wilcoxon signed-rank test).

Fig. 7.

Endothelial progenitor cells (per ml blood). Absolute (A) and %change (B) in early outgrowth colonies before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: main effect of time during pregnancy: P < 0.05. BP, before pregnancy; PP, postpartum; n, no. of subjects.

Late (outgrowth) EPC.

There were no significant effects of group, time, or group × time in the number of circulating late (outgrowth) EPC (Supplemental Table S5). Nor were there any significant differences in the percent change of late outgrowth colonies (data not shown).

Baseline NO production by early (primitive) EPC.

Absolute levels of basal NO production by early (primitive) EPC tended to increase by a greater extent during early pregnancy in the 1 vs. 0 and >1 CL cohorts (group × time: P = 0.07; Fig. 8A and Supplemental Table S5). In fact, only the 1 CL cohort demonstrated a significant time effect (P = 0.05; Fig. 8A and Supplemental Table S5). These observations were corroborated by the percent change in basal NO production, which rose in the 1 CL cohort but not in the two IVF groups (Fig. 8B).

Fig. 8.

Nitric oxide. Absolute (A) and %change (B) in basal nitric oxide production by early outgrowth colonies before, during, and after pregnancy in women conceiving with [0 or >1 corpus luteum (CL)] or without (1 CL) in vitro fertilization. A: significant effect of time for 1 CL participant cohort: P = 0.05. B: †P < 0.053 for 1 CL vs. 0 CL and >1 CL. BP, before pregnancy; PP, partum; n, no. of subjects.

Flow Cytometry

We focused on circulating progenitor cells coexpressing CD34⁺CD45dim (Supplemental Fig. S3), CD34⁺CD45dimCXCR4⁺ (Supplemental Fig. S4), and CD34⁺CD45dimCD133⁺ (Supplemental Fig. S5). The pattern of change for the three participant cohorts was similar, each showing a progressive increase in the first trimester, either reaching a plateau thereafter or a peak at the end of pregnancy (main effect of time: all P < 0.001).

DISCUSSION

The objectives of this work were to evaluate 1) endothelial function by EndoPAT using RHI; 2) CEC by CD146 antibody selection; and 3) circulating EPC number and function by colony formation, flow cytometry, and NO production before, during, and after pregnancy in women conceiving spontaneously with or without IVF. These objectives were in part motivated by our previous findings of widespread cardiovascular dysregulation, particularly in the first trimester of women conceiving by IVF using programmed (artificial) cycles in the absence of a CL, who were also at greater risk for developing preeclampsia (11, 34). We reasoned that these women would also show compromised endothelial function as revealed by abnormal RHI, increased CEC reflecting endothelial cell damage, and decreased number and function of circulating EPC suggesting reduced endothelial repair.

A major finding was that the RHI progressively (and paradoxically) decreased during pregnancy, and to a similar degree in all participant cohorts (Fig. 1). Although RHI remained in the healthy range (>1.67) before, during, and after pregnancy in the three subject cohorts, we expected an increase in RHI during pregnancy relative to prepregnant and postpartum values indicative of even more robust endothelial function, especially in the women who conceived spontaneously without assisted reproduction. However, after further consideration, we realized that the unexpected gestational decline in RHI was largely attributable to the progressive gestational rise in baseline preocclusion PWA in both control and occluded arms (Fig. 2, A and B, left), a variable that was incorporated into the denominator of the RHI calculation and baseline correction factor by the EndoPAT software. To adjust for this gestational increase of baseline preocclusion PWA, we substituted prepregnant values for the pregnant baseline preocclusion PWA in the RHI calculation at each visit during pregnancy [normalized RHI (Supplemental Fig. S1)], which yielded a progressive rise that was greatest in women who conceived spontaneously, at least in those who conceived by IVF after controlled ovarian stimulation with formation of multiple CL, and intermediate in the women who conceived by IVF using programmed cycles in the absence of a CL (Fig. 3, A and B). Thus, the greatest gestational increase of normalized RHI was observed in the spontaneously conceiving cohort with 1 CL, consistent with the hypothesis that this (control) cohort would demonstrate the most robust endothelial health during pregnancy. As detailed below, however, normalized RHI at each pregnancy visit represented the sum of two factors, both of which were presumably dependent to a large extent on endothelial function, i.e., baseline preocclusion PWA, which progressively rose during gestation (Fig. 2, A and B), and the reactive hyperemia or ischemia response (Fig. 5).

Absolute values of baseline preocclusion PWA rose comparably during pregnancy in the three participant cohorts, consistent with the overall reduction in systemic vascular resistance of pregnancy (Fig. 2, A and B, left). As a percent change of before pregnancy, the baseline preocclusion PWA increased most rapidly during pregnancy in the spontaneous conception cohort with 1 CL (Fig. 2, A and B, right). One possible interpretation of the latter finding is that, during gestation in women who conceived spontaneously, there was more robust baseline preocclusion production of endothelial NO and possibly other vasodilators in resistance arteries stimulated by pregnancy hormones, thereby reducing vascular resistance and enhancing blood flow that, in turn, stimulated endothelial NO production in conduit arteries via increasing shear stress. However, a more likely explanation is strictly a mathematical one, that is, when expressed as a percentage of the values observed before pregnancy (Fig. 2, A and B, right), the more rapid gestational increase of baseline preocclusion PWA in the spontaneous conception cohort compared with the two IVF groups was the result of the lower baseline preocclusion PWA before pregnancy (Fig. 2, A and B, left, and Supplemental Table S4). An explanation for the lower baseline preocclusion PWA values before pregnancy in the spontaneously conceiving participant group, or, alternatively, the higher preocclusion PWA before pregnancy in the two IVF subject cohorts, was not readily apparent.

The robust and progressive percent increase in baseline preocclusion PWA during pregnancy in the women who conceived spontaneously was also reflected in a more rapid percent increase of the maximum PWA (Fig. 4B; again, however, as shown in Fig. 4A, a function of the lower maximum PWA before pregnancy). Of note, the time to maximum PWA after the ischemia stimulus progressively increased during pregnancy, but in a comparable fashion among the three cohorts (Supplemental Fig. S2). Maximum PWA, like the normalized RHI discussed above, represented the sum of two factors, both dependent at least in part on endothelial function, i.e., baseline preocclusion PWA and reactive hyperemia or ischemia response. In contrast to the baseline preocclusion PWA, the reactive hyperemia or ischemia response during pregnancy only modestly increased relative to the values observed before pregnancy in the spontaneously conceiving control participant cohort (P < 0.08; Fig. 5, A and B). Numerically speaking, the ischemia response contributed progressively less to the normalized RHI and maximum PWA with advancing gestation (compare Fig. 5 with Figs. 3 and 4). Of note, however, the ischemia response of the women who conceived using programmed IVF cycles without a CL remained more-or-less unchanged throughout pregnancy while those who conceived by controlled ovarian stimulation with multiple CL actually showed a significant gestational decrease in the ischemia response (Fig. 5, A and B). The gestational decrease of the ischemia response in the controlled ovarian stimulation group contributed to the lesser rise in normalized RHI (Fig. 3, A and B) and maximum PWA (at least as a percentage of the values before pregnancy; Fig. 4B) observed in this participant cohort.

In light of our previous work demonstrating generalized cardiovascular dysregulation, especially in the first trimester of women conceiving by IVF in programmed cycles without a CL (11, 34), we predicted that these women would show the most attenuated gestational increases of baseline preocclusion PWA, ischemia response, maximum PWA, and normalized RHI, reflecting the greatest degree of endothelial impairment, but this was not the case. Rather, on balance, the women who conceived by controlled ovarian stimulation with multiple CL demonstrated the most attenuated gestational increases despite their otherwise apparently normal cardiovascular adaptation to pregnancy as previously reported (11, 34). This unexpected result was not so much a consequence of a more attenuated gestational rise in baseline preocclusion PWA in the controlled ovarian stimulation vs. programmed cycles because baseline preocclusion PWA was comparable in the two IVF cohorts (Fig. 2, A and B); rather, there was a significant decline in the ischemia response, most notably during early pregnancy in the controlled ovarian stimulation group (Fig. 5). Given that gestational changes in preocclusion PWA, maximum PWA, and normalized RHI reflected endothelial function, then both IVF groups may have manifested some degree of endothelial dysfunction during pregnancy before the ischemia stimulus. However, only the IVF group that conceived by controlled ovarian stimulation showed reduced ischemia response during pregnancy, perhaps reflecting greater compromise of endothelial function and impaired endothelial reserve. Finally, it should be noted that, despite the greater BMI in women who conceived by programmed cycles relative to the other two participant cohorts who were similar (Table 1), there were no significant group × BMI interactions, i.e., the relationships between BMI and RHI, RHI norm, and ischemia response were comparable among the three participant cohorts at each study visit (data not shown).

Another major finding of this study was that the number of CEC increased significantly during pregnancy in all three participant cohorts, reinforcing the idea of enhanced systemic inflammation and heightened state of endothelial activation during pregnancy (Fig. 6, A and B, and Ref. 24). Interestingly, the number of CEC appeared to have increased the most during early gestation in the women who conceived by IVF using programmed cycles without the formation of a corpus luteum, perhaps reflecting more endothelial activation or damage. This finding is consistent with our earlier observation of widespread cardiovascular dysregulation in this patient cohort, particularly during the first trimester.

Although the number of circulating early (primitive) EPC as assessed by colony formation rose gradually during pregnancy, they did so in a comparable fashion among the three participant cohorts (Fig. 7). Reassuringly, there was also a comparable gestational increase of circulating EPC as defined by flow cytometry among the three subject groups (Supplemental Figs. S3–S5). Of note, the rise in EPC during pregnancy was independent of CL status, suggesting that hormones uniquely secreted by the CL like relaxin are not likely to be involved (30). Several, but not all, investigators reported elevated EPC during pregnancy; however, the precise hormone(s)/factor(s) mediating the increase and the physiological role(s) of these cells during pregnancy remain incompletely understood (reviewed in Refs. 12 and 16), although they have generally been associated with the development of new endothelium (vasculogenesis) and endothelial repair (5, 17). Interestingly, the basal production of NO by early (primitive) EPC significantly increased during gestation in the control participant cohort that conceived spontaneously (1 CL; Fig. 8). In contrast, basal production of NO by early (primitive) EPC did not rise during pregnancy in either IVF cohort, indicating that relaxin did not play a role in basal production of NO by early (primitive) EPC because the women who conceived by controlled ovarian stimulation developed multiple CL frequently associated with markedly elevated circulating relaxin concentrations (30). Collectively, these results suggested that, despite comparable circulating numbers of EPC during pregnancy among the three cohorts, their function with respect to basal NO production was compromised in both IVF groups. Finally, we did not detect significant gestational changes in the number of circulating late (outgrowth) EPC in the three participant cohorts (Supplemental Table S5).

In a different population of women, we recently reported that RHI was reduced during pregnancy in women who conceived with programmed (0 CL) and controlled ovarian stimulation cycles (>1 CL) compared with spontaneous conceptions (33). In the current study, however, we did not observe these differences at comparable stages of gestation (gestational weeks 10–12 or 14–16; Fig. 1 and Supplemental Table S4). Although we do not have a ready explanation for this apparent inconsistency, other than the two participant populations were different, we did observe reductions in the normalized RHI in the two IVF groups throughout pregnancy relative to spontaneous conceptions (Fig. 3). Moreover, comparable to the earlier study (33), we also noted a trend for greater baseline preocclusion PWA in the multiple CL cohort (Fig. 2A, left), although, in the current study, greater baseline preocclusion PWA was also observed in the 0 CL group. Finally, in both studies, EPCs were not significantly different among the three pregnancy cohorts with 0, 1, and >1 CL (33). On balance, there was some, but not complete, congruity between the present work and our earlier investigation.

In summary, we investigated RHI in pregnancies conceived with and without IVF. A major strength of our work is that we serially evaluated RHI in the same women before pregnancy in the follicular phase, at six time points during pregnancy, and, on average, 10.5 mo postpartum. Our study conclusively demonstrated an unexpected progressive decline in RHI during gestation relative to both the before and after pregnancy time points, irrespective of the mode of conception or CL number. The gestational decline in RHI was largely a consequence of a rising baseline preocclusion PWA during pregnancy, which was incorporated into the RHI equation by the EndoPAT software. To underscore this point, we adjusted for this gestational increase in baseline preocclusion PWA by substituting into the RHI equation at every gestational time point for each study participant, the baseline preocclusion PWA measured before pregnancy. As predicted, this substitution yielded a normalized RHI that progressively rose during pregnancy, more so in the spontaneous conception than IVF groups. We further observed that the time to maximum PWA after the ischemia stimulus also increased during pregnancy, but in a comparable manner among the three participant cohorts. We next identified the maximum PWA after the ischemia stimulus for each participant at each study time point and observed that the maximum PWA also progressively rose during pregnancy. Maximum PWA after ischemia and the normalized RHI both represented the sum of two major components: baseline preocclusion PWA and the reactive hyperemia or ischemia response. We determined the reactive hyperemia or ischemia response by taking the difference between the maximum PWA after ischemia in the occluded arm and the postocclusion maximum PWA in the control arm (reflecting baseline preocclusion PWA and potential confounding factors like autonomic activity). There was a modest gestational increase of the reactive hyperemia or ischemia response in the spontaneous conception cohort, no change during pregnancy in women who conceived by IVF using programmed cycles, and, most notably, a significant decline in the reactive hyperemia or ischemia response in the women who conceived by IVF using controlled ovarian stimulation. Other notable findings were that NO production by early (primitive) outgrowth EPCs increased during pregnancy in the spontaneously conceiving women but not in the two IVF cohorts, and, although CEC, a marker of endothelial cell damage, increased during gestation in all three participant cohorts, the increase was greater in the women who conceived by IVF using programmed cycles during the first trimester.

Perspectives and Significance

Taken together, we conclude that there was profound increase in basal endothelial function (i.e., before the ischemia stimulus), and consequently, of forearm blood flow during pregnancy that was attenuated in the two IVF cohorts when expressed as a percentage of prepregnancy values because of an inexplicably higher forearm blood flow in the two IVF groups before pregnancy. Reactive hyperemia or ischemia response showed a modest increase in spontaneous conceptions during pregnancy but was unchanged in pregnancies conceived by programmed cycles. Of note, the ischemia response was significantly decreased in pregnancies conceived by controlled ovarian stimulation, suggesting greater compromise of endothelial function and reduced endothelial reserve. EPCs measured by early (primitive) outgrowth colonies or by flow cytometry increased during pregnancy in all three cohorts in a similar manner, as did CEC, although the latter increased more in the programmed cycles during the first trimester, perhaps reflecting greater endothelial activation or damage. Finally, early (primitive) outgrowth EPCs in the spontaneous conceptions showed a robust gestational increase in basal NO production that was absent in the two IVF cohorts. Thus, although the number of circulating early (primitive) outgrowth EPCs was comparable among the three participant cohorts, their function as measured by NO production was impaired in the pregnancies conceived by IVF.

GRANTS

This work was supported by National Institutes of Health (NIH) Grant P01-HD-065647-01A1 (KPC), J. Robert and Mary Cade Professorship of Physiology (KPC), and matching funds from the University of Florida College of Medicine (KPC). Research reported in this publication was also supported by the University of Florida Clinical and Translational Science Institute, which is underwritten in part by the NIH National Center for Advancing Translational Sciences under award number UL1-TR-001427.

DISCLOSURES

K. P. Conrad discloses use patents for relaxin. The other authors report no conflict of interest. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Study data were collected and managed using REDCap electronic data capture tools hosted at the University of Florida.

AUTHOR CONTRIBUTIONS

K.P.C. and M.S.S. conceived and designed research; K.P.C., M. Lingis, L.S., S.L., Y.-Y.C., Y.Q., M. Li, and M.S.S. analyzed data; M. Lingis, L.S., and S.L. performed experiments; K.P.C. and M.S.S. interpreted results of experiments; Y.-Y.C., Y.Q., and M. Li prepared figures; K.P.C., M. Lingis, L.S., S.L., Y.-Y.C., R.S.W., and A.R.-V. drafted manuscript; K.P.C., M. Lingis, Y.-Y.C., Y.Q., M. Li, and M.S.S. edited and revised manuscript; K.P.C., M. Lingis, L.S., S.L., Y.-Y.C., Y.Q., M. Li, R.S.W., A.R.-V., and M.S.S. approved final version of manuscript.