Placental vascular abnormalities in the mouse alter umbilical artery wave reflections

Abstract

Current methods to detect placental vascular pathologies that monitor Doppler ultrasound changes in umbilical artery (UA) pulsatility have only moderate diagnostic utility, particularly in late gestation. In fetal mice, we recently demonstrated that reflected pressure waves propagate counter to the direction of flow in the UA and proposed the measurement of these reflections as a means to detect abnormalities in the placental circulation. In the present study, we used this approach in combination with microcomputed tomography to investigate the relationship between altered placental vascular architecture and changes in UA wave reflection metrics. Fetuses were assessed at embryonic day (E)15.5 and E17.5 in control C57BL6/J mice and dams treated with combination antiretroviral therapy (cART), a known model of fetal growth restriction. Whereas the reflection coefficient was not different between groups at E15.5, it was 27% higher at E17.5 in cART-treated mice compared with control mice. This increase in reflection coefficient corresponded to a 36% increase in the total number of vessel segments, a measure of overall architectural complexity. Interestingly, there was no difference in UA pulsatility index between groups, suggesting that the wave reflections convey information about vascular architecture that is not captured by conventional ultrasound metrics. The wave reflection parameters were found to be associated with the morphology of the fetoplacental arterial tree, with the area ratio between the UA and first branch points correlating with the reflection coefficient. This study highlights the potential for wave reflection to aid in the noninvasive clinical assessment of placental vascular pathology.

NEW & NOTEWORTHY We used a novel ultrasound methodology based on detecting pulse pressure waves that propagate along the umbilical artery to investigate the relationship between changes in wave reflection metrics and altered placental vascular architecture visualized by microcomputed tomography. Using pregnant mice treated with combination antiretroviral therapy, a model of fetal growth restriction, we demonstrated that reflections in the umbilical artery are sensitive to placental vascular abnormalities and associated with the geometry of the fetoplacental tree.

INTRODUCTION

Placental vascular pathology is a causal factor in a large proportion of pregnancy complications and is often associated with fetal growth restriction (10). Current methods to detect placental abnormalities rely on Doppler ultrasound to monitor changes in the pulsatility of the umbilical artery (UA) blood velocity waveforms, an indirect measure of downstream placental vascular resistance (3, 13, 33). However, these methods have poor sensitivity for detecting growth restriction and fetal distress, especially during late gestation (18). Our group has recently developed a novel approach using ultrasound to noninvasively detect changes in the fetoplacental vasculature of the mouse (25). The technique is based on detecting reflected pulse pressure waves in the UA, uniquely decomposing the observed Doppler waveforms into its forward and reflected components. Using two strains of experimental mice, we have previously shown that the reflected waves were significantly different in late gestation, mirroring reported changes in the fractal geometry of the fetoplacental arterial trees. We have also shown that reflected waves are present in the human UA and that these waves explain the variation in Doppler waveforms along the umbilical cord (J. G. Sled, G. Stortz, L. S. Cahill, N. Milligan, V. Ayyathurai, L. Serghides, E. Morgen, V. Seravallli, C. Delp, C. McShane, A. A. Baschat, J. Kingdom, and C. K. Macgowan, unpublished observations). Having identified that reflections exist in the UA of both the mouse and human, it remains unknown from where within the placenta the reflections originate. Additionally, it is unknown if reflection metrics are sensitive to placental vascular pathology.

In the present study, we investigated the changes in UA wave reflection metrics at two time points in late gestation in control C57BL6/J mice and dams treated with combination antiretroviral therapy (cART), a known model of fetal growth restriction (17, 22). One proposed contribution to fetal growth restriction in human immunodeficiency virus (HIV)-positive pregnancies is lower progesterone levels induced by cART exposure. Progesterone is known to play a key role in regulating placenta angiogenesis (12), and decreased progesterone levels correlate with both birth weight percentile in cART-exposed HIV-positive women and fetal weight in cART-treated mice (22). HIV antiretroviral drugs have wide-ranging effects, and some have been shown to influence vascular formation in the placenta. Hypervascularization with cART exposure has been reported in the terminal villi capillaries of placentas from HIV-positive women (17, 21) and in the arterioles in placentas of mice (17). Using high-frequency ultrasound and microcomputed tomography (micro-CT) in the same specimen, we investigated whether pulse wave velocity (PWV) and wave reflections in the UA are sensitive to this pathological pregnancy model and established whether the observed wave reflection metrics are associated with the geometry of the fetoplacental arterial tree.

METHODS

Animals.

Healthy adult C57BL6/J virgin female mice (age: 7–10 wk) from Jackson Laboratories were used and mated in house. Embryonic day (E)0.5 was designated as the morning that a vaginal plug was detected. Mice were housed in a standard cage under a 12:12-h light-dark cycle with ad libitum access to food and water. Pregnant mice were randomized to either a treatment group that received a combination of antiretroviral drugs or a control group. Antiretrovirals were purchased as prescription medication under the name Kaletra (lopinavir-ritonavir) and Kivexa (zidovudine-lamivudine), pulverized, incorporated into the regular chow (1.25 g/kg zidovudine-lamivudine and 1.05 g/kg lopinavir-ritonavir in Teklad 18% protein rodent diet), and administered starting on E0.5. The drug doses were chosen to yield clinically relevant plasma drug levels as determined experimentally (22). This was based on the assumption that a 25-g mouse eats 5 g/day (resulting in 100/50 mg·kg⁻¹·day⁻¹ zidovudine-lamivudine and 33/8.3 mg·kg⁻¹·day⁻¹ lopinavir-ritonavir). Dams were imaged using ultrasound biomicroscopy at either E15.5 (cART: 11 fetuses from 5 litters and control: 24 fetuses from 8 litters) or E17.5 (cART: 15 fetuses from 5 litters and control: 24 fetuses from 8 litters) and then immediately euthanized for fixed specimen micro-CT imaging. All animal experiments were approved by the Animal Care Committee of The Centre for Phenogenomics, conducted in accordance with guidelines established by the Canadian Council on Animal Care, and compliant with Animal Research: Reporting of In Vivo Experiment guidelines.

Ultrasound biomicroscopy.

A high-frequency ultrasound system with a 40-MHz linear array transducer (Vevo 2100, VisualSonics, Toronto, ON, Canada) was used to image the fetal end of the UA, as previously described in detail (25, 40). Briefly, dams were anesthetized with isoflurane (4% for induction and 2.5% for maintenance in 21% O₂) with maternal heart rate and respiration monitored throughout the imaging session. M-mode and pulsed wave Doppler recordings were made at approximately the same location in the UA. For M-mode recording of the dynamic change of the UA dimension throughout the fetal cardiac cycle, the ultrasound beam direction was positioned perpendicular to the UA. For recording UA blood flow velocity waveforms, the Doppler sample volume was adjusted to cover the entire vessel lumen and the angle of insonation was kept as small as possible (<60°). All UAs that were in a favorable spatial orientation for both M-mode and Doppler were imaged (2–4 UAs/dam). The fetal and placental orientation/locations inside the maternal abdomen were carefully recorded and used to identify the fetoplacental unit at dissection.

Micro-CT imaging.

Immediately after ultrasound imaging was completed, dams were euthanized by cervical dislocation and prepared for micro-CT imaging as previously described (28, 37). The embryo and placenta were placed in warm PBS to resume blood circulation, and a cannula was inserted into the UA. Heparinized saline followed by the contrast agent (MV-122 Microfil, Flow Technology, Carver, MA) were manually infused. After perfusion, placental and embryonic weights were recorded and the umbilical cord length was measured. Placentas were immersed in 10% formalin for 24–48 h and then mounted in agar for imaging. Three-dimensional data sets were acquired using a Bruker Skyscan 1272 micro-CT scanner (Bruker Skyscan, Antwerp, Belgium). With the X-ray source at 50 kV and 201 μA, the specimen was rotated 360° in 0.2° increments, generating 1,800 views that were reconstructed into data blocks with a 7.1-μm voxel size.

Ultrasound image analysis.

All ultrasound image analysis was performed as previously described (25). Briefly, the M-mode time series showing the two walls of the UA were automatically traced and the waveform shapes were smoothed using a low-pass, second-order Butterworth filter. Smoothed waveforms were separated into individual fetal cardiac cycles based on the onset of systole. The diameter estimates as a function of time were converted to cross-sectional area, and the individual UA area waveforms were temporally aligned and averaged. Similarly, from Doppler flow spectra, the mean velocity waveforms were manually outlined, smoothed, separated into individual fetal cardiac cycles, temporally aligned, and averaged together. Ultrasound data quality was controlled using the acceptance criteria outlined in Rahman et al. (25). Before wave reflection analysis, data sets were excluded if high levels of maternal gasping or fetal movements were present. After wave reflection analysis, further postanalysis criteria were applied for the M-mode traces. Data sets were excluded if 1) >35% of the aligned waveforms had at least some points that were outside bands representing ±20% of the average area change or 2) there were insufficient individual area traces to average (<4 area traces).

The average area and velocity waveforms were multiplied to calculate the average flow waveform. On a plot of flow versus area, a line was fit to that portion of the cycle between 20% and 80% of the maximum systolic flow, and the slope of this line was taken as PWV. With the use of PWV, the observed flow waveforms [Q_m(t)] were decomposed into their forward [Q_f(t)] and reflected [Q_r(t)] components. Reflection waveforms were summarized in terms of reflection coefficient, time delay, and dispersion. The reflection coefficient is defined as the ratio of the peak to peak amplitude between the backward and forward waves. The time delay is the time difference between the peak of the backward and forward waves. Dispersion is the difference in the full-width at half-maximum of the backward and forward wave (Fig. 1). Based on the original Doppler velocity waveforms, the pulsatility index (PI) was calculated as the difference between peak systolic and end-diastolic velocities divided by the mean velocity over the fetal cardiac cycle.

Fig. 1.

Representative umbilical artery wave decomposition, showing the flow (Q_m), forward (Q_f), and reflected (Q_r) waveforms. The parameters used to calculate the reflection coefficient, time delay, and dispersion are shown.

Vascular segmentation.

The placental vascular structure was automatically segmented using a previously described algorithm that identifies vessel segments and bifurcations (26). Detection of vessels with a diameter smaller than 35 μm was unreliable; therefore, the terminal vessel segments were pruned to 35 μm to improve data consistency. Fetoplacental arterial span and depth as well as the diameter of the UA and first-order branches were measured directly from surface renderings of the micro-CT data using digital calipers in the Amira software package (Visage Imaging, San Diego, CA). The area ratio was calculated as the ratio of the combined cross-sectional area of daughter vessels to that of the parent vessel (34). The experimenter who performed the manual measurements was blinded to the specimen details (group and gestational age).

Statistical analysis.

All statistical tests were conducted using the R statistical software package (www.r-project.org). Fetal and placental parameters (fetal weight, placental weight, and umbilical cord diameter and length), physiological parameters (fetal heart rate, mean UA velocity, peak UA velocity, UA blood flow, and PWV), wave reflection metrics (reflection coefficient, time delay, and dispersion), and UA PI and vascular segmentation metrics (depth, span, vascular volume, and total number of vessel segments) were analyzed using a linear mixed effects model with group (cART treatment, control) and gestational age (E15.5, E17.5) as fixed effects and litter as a random effect. If the ANOVA was significant, post hoc t-tests were performed. Three-way ANOVAs were used to assess the variance in the reflection coefficient and UA PI due to group, gestational age, and either fetal heart rate or UA blood flow. A linear model was used to determine if the reflection coefficient depended on the total number of vessel segments or vessel type (subdivided by vessel size). A linear model using a second-order polynomial function was used to determine if the area ratio at the first branch of the UA depended on the reflection coefficient. All data are reported as means with 95% parametric confidence interval (CI) values. P values of <0.05 were considered significant.

RESULTS

After treatment with a cART regimen of zidovudine-lamivudine-lopinavir-ritonavir, fetal and placental weights were significantly smaller compared with controls at both E15.5 and E17.5 (Fig. 2, A and B). The difference in fetal weights between groups was more pronounced at later gestation, with a 35% decrease in fetal weight at E17.5 in cART-treated mice compared with control mice [cART: 0.62 g (CI: 0.57–0.67) vs. control: 0.95 g (CI: 0.89–1.01), P < 0.0001]. Although the diameter of the UA measured on micro-CT did not show an effect of group or gestational age (Fig. 2C), the length of the umbilical cord was smaller in cART-treated placentas compared with control placentas at both gestational ages [E15.5: 8.9 mm (CI: 8.3–9.5) in the cART-treated group vs. 10.4 mm (CI: 9.7–11.1) in the control group, P = 0.009; E17.5: 9.8 mm (CI: 9.3–10.3) in the cART-treated group vs. 10.9 mm (CI: 10.4–11.4) in the control group, P = 0.003; Fig. 2D]. Additionally, the thickness of the UA vessel wall, determined from M-mode ultrasound, was significantly increased in the cART-treated group compared with the control group at both gestational ages [E15.5: 65 μm (CI: 56–74) in the cART-treated group vs. 53 μm (CI: 47–59) in the control group, P = 0.02; E17.5: 69 μm (CI: 59–79) in the cART-treated group vs. 56 μm (CI: 50–62) in the control group, P = 0.02; Fig. 2E]. cART-treated dams had smaller litter sizes [cART: 4 fetuses/litter (CI: 3–5) vs. control: 7 fetuses/litter (CI: 6–8), P = 0.006], and resorptions were more frequent compared with control mice. The fetal-to-placental weight ratio (23) increased with gestational age for both groups; however, there was no difference between groups.

Fig. 2.

Effects of combination antiretroviral therapy (cART) treatment on fetal and placental parameters compared with controls. A: fetal weights (in g). B: placental weights (in g). C: umbilical artery (UA) diameter (in mm). D: umbilical cord length (in mm). E: UA wall thickness (in μm). *P < 0.0001, significant group and gestational age interaction, as determined by a two-way ANOVA. Main effects of group or gestational age are noted as P_group and P_age. Data are shown as means ± 95% confidence intervals; n refers to the number of fetuses or placentas. A different symbol is used for each litter to aid in data visualization.

The reflection coefficient is sensitive to placental vascular abnormalities.

After application of the preprocessing and postprocessing steps to the M-mode images, 51% of the UA scans met the quality control criteria and were included in the wave decomposition analysis (E15.5: 7 cART-treated and 12 control scanes and E17.5: 8 cART-treated and 11 control scans). This is consistent with our previous work, where 46% of the UA scans met the acceptance criteria (25). Of these data sets, 79% had a successful micro-CT fetoplacental perfusion (E15.5: 5 cART-treated and 8 control scans and E17.5: 7 cART-treated and 10 control scans). The physiological and hemodynamic parameters for the data sets that met the acceptance criteria are shown in Table 1.

Table 1.

Physiological and hemodynamic parameters

Group	Gestational Age	Fetal Heart Rate, beats/min	Mean UA Velocity, mm/s	Peak UA Velocity, mm/s	UA Blood Flow, ml/min
Control	E15.5	184 (CI: 170–198)*†	25 (CI: 20–30)	56 (CI: 47–65)	0.09 (CI: 0.07–0.11)†
cART	E15.5	215 (CI: 194–236)	24 (CI: 17–31)	52 (CI: 35–69)	0.07 (CI: 0.05–0.09)‡
Control	E17.5	211 (CI: 195–227)	26 (CI: 23–29)	57 (CI: 50–64)	0.12 (CI: 0.10–0.14)
cART	E17.5	232 (CI: 206–258)	38 (CI: 21–55)	78 (CI: 46–110)	0.15 (CI: 0.09–0.21)

Data are presented as means and 95% confidence interval (CI) values. cART, combination antiretroviral therapy; E15.5, embryonic day 15.5; E17.5, embryonic day 17.5; UA, umbilical artery.

^*P < 0.01 compared with cART-treated fetuses at E15.5;

^†P < 0.01 compared with control fetuses at E17.5;

^‡P < 0.01 compared with cART-treated fetuses at E17.5 (as determined by two-way ANOVA followed by a post hoc t-test).

For PWV, gestational age had a significant effect (Fig. 3A). Post hoc analysis using t-tests showed that while the control group had a similar PWV across gestation, there was a 171% increase in PWV from E15.5 to E17.5 in the cART-treated group [E15.5: 1.7 m/s (CI: 0.7–2.7) vs. E17.5: 4.6 m/s (CI: 2.6–6.6), P = 0.01]. Both group and gestational age had a significant effect on the average reflection coefficient (Fig. 3B). Post hoc analysis showed that, although the reflection coefficient was not different between the two groups at E15.5 [cART: 0.41 (CI: 0.32–0.50) vs. control: 0.34 (CI: 0.30–0.38), P = 0.08], it was 27% higher at E17.5 in the cART-treated group compared with the control group [cART: 0.53 (CI: 0.43–0.63) vs. control: 0.42 (CI: 0.37–0.47), P = 0.02]. Both cART-treated and control mice had a significant increase in the reflection coefficient across gestation (P = 0.05 and P = 0.02, respectively). For both groups and gestational ages, there was no correlation between the reflection coefficient and fetal heart rate (P = 0.4) or between the reflection coefficient and UA blood flow (P = 0.3).

Fig. 3.

Wave reflection parameters and umbilical artery (UA) pulsatility index (PI) for combination antiretroviral therapy (cART)-treated mice compared with control mice. A: pulse-wave velocity (in m/s). B: reflection coefficient. C: time delay (in s). D: UA PI. E: dispersion (in s). Main effects of group or gestational age as determined by a two-way ANOVA are noted as P_group and P_age. Data are shown as means ± 95% confidence intervals; n refers to the number of fetuses. A different symbol is used for each litter to aid in data visualization.

In terms of time delay, both group and gestational age had a significant effect (Fig. 3C). Post hoc analysis using t-tests showed there was no significant difference in time delay between the groups at either E15.5 or E17.5; however, there was a decrease of 24% from E15.5 to E17.5 in the control group [E15.5: 0.07 s (CI: 0.06–0.08) vs. E17.5: 0.05 s (CI: 0.04–0.06), P = 0.01]. There was a strong negative correlation between PWV and time delay (P = 0.008). Group had a significant effect on the UA PI (Fig. 3D). The UA PI was 26% lower in cART-treated mice compared with control mice at E15.5 [cART: 2.9 (CI: 2.0–3.8) vs. control: 3.9 (CI: 3.5–4.3), P = 0.01], whereas there was no difference between groups at E17.5. cART-treated mice had a similar UA PI across gestation. The UA PI had no correlation with UA blood flow (P = 0.1); however, unlike the reflection coefficient, there was a strong negative correlation with fetal heart rate (P = 0.008). Dispersion did not show an effect of group or gestational age (Fig. 3E).

The reflection coefficient is associated with a greater area mismatch at the first branching point.

Of the 74 placental samples that were prepared for micro-CT imaging, 44 placental samples were suitable for analysis (E15.5: 7 cART-treated and 11 control placental samples and E17.5: 12 cART-treated and 14 control placental samples). Samples were excluded because of incomplete Microfil perfusion or vessel ruptures causing leakage. The morphology of the fetoplacental arterial vasculature was visualized using surface renderings of the micro-CT data (Fig. 4, A and B) and is summarized in Table 2.

Fig. 4.

Distribution of vessel segments and association with wave reflection parameters. A and B: representative microcomputed tomography images of control embryonic day (E)17.5 (A) and combination antiretroviral therapy (cART) E17.5 (B) fetoplacental arterial vascular trees color coded by vessel diameter. C: cumulative distributions of vessel diameters in cART-treated and control placentas. D and E: the number of arterioles showed a negative dependence on fetoplacental weight (P = 0.01; D), and there was a trend toward a correlation between the reflection coefficient and number of intraplacental arteries (P = 0.09; E). The shaded gray area represents 95% confidence intervals. F: average reflection coefficient as a function of the area ratio at the first branch point of the umbilical artery for cART-treated and control placentas. The black curve represents a quadratic model with 95% confidence intervals (adjusted R² = 0.35).

Table 2.

Measurements of the fetoplacental arterial tree

Group	Gestational Age	Vascular Depth, mm	Vascular Span, mm	Vascular Volume, mm3	No. Vessel Segments
Control	E15.5	1.5 (CI: 1.4–1.6)a	6.8 (CI: 6.6–7.0)	3.4 (CI: 2.8–4.0)a,e	3,500 (CI: 2,900–4,100)
cART	E15.5	1.4 (CI: 1.2–1.6)b	6.7 (CI: 6.5–6.9)	2.6 (CI: 2.1–3.1)b	4,200 (CI: 3,400–5,000)
Control	E17.5	1.8 (CI: 1.7–1.9)c	7.0 (CI: 6.8–7.2)d	5.0 (CI: 4.3–5.7)f	3,300 (CI: 2,800–3,800)d
cART	E17.5	1.6 (CI: 1.5–1.7)	6.6 (CI: 6.5–6.7)	3.6 (CI: 2.9–4.3)	4,500 (CI: 4,000–5,000)

Data are presented as means and 95% confidence interval (CI) values. cART, combination antiretroviral therapy; E15.5, embryonic day 15.5; E17.5, embryonic day 17.5.

^aP < 0.001 compared with control fetuses at E17.5;

^bP < 0.05 compared with cART-treated fetuses at E17.5;

^cP < 0.05 compared with cART-treated fetuses at E17.5;

^dP < 0.001 compared with cART-treated fetuses at E17.5;

^eP < 0.05 compared with cART-treated fetuses at E15.5;

^fP < 0.005 compared with cART-treated fetuses at E17.5 (as determined by a two-way ANOVA followed by a post hoc t-test).

Compared with the control group, the arterial tree in the cART-treated group was significantly smaller in vascular volume at both E15.5 and E17.5 and in overall dimensions at later gestation. Vascular segmentation revealed that there was a significant effect of group on the total number of vessel segments in the fetoplacental arterial tree, with no difference at E15.5 and a 36% increase in vessel segments at E17.5 compared with the control group (P = 0.0008). To determine if this difference was confined to a specific part of the placenta, the number of vessel segments was determined within diameter ranges corresponding to approximate anatomic locations within the vascular tree: arterioles [35–75 μm], intraplacental arteries [75–150 μm], and chorionic plate arteries [>200 μm] (26). The increase in vessel segment number at E17.5 in the cART-treated group was confined primarily to the arterioles [cART: 3,500 (CI: 3,100–3,900) vs. control: 2,400 (CI: 2000–2800), P = 0.0004], whereas the number of intraplacental and chorionic plate arteries was similar between groups. This difference between groups is apparent when the cumulative number of vessel segments was plotted as a function of diameter (Fig. 4C). Additionally, with gestational age, there was a significant rightward shift in the mid diameter range in both control and cART-treated placentas. Because the total number of vessel segments remains constant, this shift corresponds to vessels in this size range increasing in diameter. Consistent with our previous work (17), the number of arterioles showed a negative correlation with fetoplacental weight (the weight of the fetus and placenta combined, P = 0.01; Fig. 4D).

By combining noninvasive in vivo ultrasound measurements and ex vivo micro-CT imaging in the same fetoplacental unit, we were able to determine if wave reflection in the UA was explained by the geometry of the fetoplacental arterial tree. Our primary hypothesis was that group and gestational age would have an effect on the reflection coefficient. Although there was a trend toward a positive correlation between the reflection coefficient and number of intraplacental arteries (P = 0.09; Fig. 4E), after accounting for the contributions from the group and gestational age, there was no additional effect of the number of vessel segments on the reflection coefficient.

In our previous work (25), we did not have access to the corresponding micro-CT data and instead used reported diameter scaling coefficients (26) to compute the average area ratio (the ratio of the combined cross-sectional area of the daughter vessels to that of the parent vessel) in the fetoplacental tree. Here, we measured the area ratio at the first branching point of the UA from the micro-CT images. The majority of placentas had three branches at the first branch point. Figure 4F shows the experimental curve relating the reflection coefficient to the area ratio at the first branch point (quadratic model, adjusted R² = 0.35). The reflection coefficient approached a minimum at an area ratio close to 1.0 and increased as the area mismatch increased.

DISCUSSION

In the present study, we found that wave reflection measurements in the murine UA are altered in the presence of placental vascular abnormalities during late gestation. At E17.5, cART treatment resulted in decreased dimensions and increased vascularization of the fetoplacental arterial tree compared with the control group, specifically an increase in the number of small-diameter vessels. This increase in the number of arterioles correlated with fetoplacental weight, leading us to speculate that it was detrimental to fetal growth. An increase in the number of small-diameter vessels has been previously associated with increased calculated vascular resistance in endothelial nitric oxide synthase-deficient (eNOS^−/−) mice (27) and in a mouse model of experimental malaria in pregnancy (7). These differences in placental vascular morphology corresponded to a significant increase in the reflection coefficient. There was no difference in the UA PI at E17.5, suggesting an increased sensitivity in the reflection coefficient metric compared with the PI. This is consistent with our previous receiver-operating characteristic analysis (25), which showed that the reflection coefficient was better able to distinguish fetoplacental vascular differences between healthy CD1 and C57BL6/J mice.

The work of Adamson and Langille (1, 2) has shown that the UA PI is affected by multiple factors in addition to placental vascular resistance. For example, previous studies have reported that the UA PI is negatively correlated with fetal heart rate in both sheep (19) and humans (39). This is consistent with our finding of a strong negative correlation between the UA PI and fetal heart rate. In contrast, the reflection coefficient was not correlated with fetal heart rate or UA blood flow, suggesting that the reflections are dependent on the geometry of the fetoplacental arterial tree.

Using high-resolution micro-CT imaging, we found that the area ratio at the first branch of the UA was related to the observed average reflection coefficient. As the area mismatch increased (the total area after the branch point is either larger or smaller than the area of the parent vessel), so did the reflection coefficient. This is consistent with our previous theoretical prediction (25), based on the square law for scaling vessels (30), that deviations away from an area ratio of 1.0 would result in greater reflections. The theoretical curve predicted that wave reflections would be completely eliminated when the area ratio is 1.0 and impedances are matched. Our experimental curve (Fig. 4F) showed a minimum reflection coefficient close to 1.0; however, the reflection coefficient was nonzero. This is likely explained by the fact that the average reflection coefficient is an aggregate of reflections from all bifurcations in the tree, with subsequent bifurcations having an area ratio other than 1.0. This is consistent with the nonzero dispersion metric, indicating that multiple reflection sites contribute to the observed waveform and the first reflection site only explained 35% of the variance of the data. The trend toward a positive correlation between the reflection coefficient and number of intraplacental arteries suggests that these vessels also serve as a significant source of reflection. Interestingly, both cART-treated and control placentas fell on both sides of the area ratio curve in Fig. 4F. This shows that even in healthy placentas, the morphological characteristics of the first branches vary between specimens in hemodynamically significant ways.

From E15.5 to E17.5, the reflection coefficient increased in both cART-treated and control placentas. Whereas UA blood flow increased during late gestation, the number of fetoplacental arteries did not change. To lower fetoplacental arterial resistance, there is an increase in the diameter of the arteries upstream of the capillary bed from E15.5 to E17.5, indicated by the rightward shift of the cumulative curves between E15.5 and E17.5 for both cART-treated and control placentas (Fig. 4C). Assuming that capillary diameter remains the same, this increase in arterial vessel diameter would result in a greater area ratio mismatch throughout the fetoplacental arterial tree, consistent with the increased pulse wave reflections found in this study. These morphological changes are in agreement with our previous observations in control C57BL6/J mice (26). Moreover, it should be noted that this study is a good opportunity to evaluate replication of the wave reflection methodology. Here, the reflection coefficients for C57BL6/J control data are a good match with our previous work [E15.5: Rahman et al. (25), 0.35 (CI: 0.30–0.40), vs. this study, 0.34 (CI: 0.30–0.38); E17.5: Rahman et al. (25), 0.44 (CI: 0.38–0.50), vs. this study, 0.42 (CI: 0.37–0.47)].

In addition to an increase in the reflection coefficient in cART-treated placentas across gestation, there was a significant increase in UA PWV, a known measure of arterial stiffness. This corresponded to an increase in UA wall thickness. To investigate this further, we calculated Young’s modulus of elasticity using the Moens-Korteweg equation (16), assuming the density of blood was the same between groups. Young’s modulus was significantly elevated at E17.5 compared with E15.5 [E17.5: 133 Pa (CI: 36–230) vs. E15.5: 21 Pa (CI: −6–48), P = 0.03], indicating that in addition to the UA wall becoming thicker, the wall stiffens with gestational age in cART-treated placentas. As observed in a model of fetal growth restriction in pregnant sheep, the increased UA stiffness is most likely associated with an increased collagen-to-elastin ratio in the UA (9). Stiffer UAs in growth-restricted infants have also been reported (5) and have been implicated in fetal programming of cardiovascular diseases (20). Patients that are HIV positive and exposed to cART have an increased risk of cardiovascular disease (11), and studies have reported increased aortic stiffness in patients that are HIV positive (6, 14, 15, 31, 32). To our knowledge, the effect of cART on arterial stiffness has not been studied during pregnancy.

Previously, exposure to cART has been reported to result in abnormal angiogenesis in vitro using human umbilical vein endothelial cells (8) and in murine and human placentas (17, 21). In the mouse, we previously found dysregulation of angiogenic factors and an increase in the number of arterioles in cART-treated placentas at E14.5. Compared with the present study where we only observed a significant increase in the number of arterioles later in gestation in cART-treated placentas, the difference may be explained by the fact that the cART regimen was administered via the chow instead of an oral gavage. This approach decreases the maternal stress that results from daily handling; however, the amount of chow consumed was not recorded and is a limitation of this study.

The present study has several additional limitations. Currently, our ex vivo micro-CT imaging does not reliably provide resolution of vessels <35 μm; therefore, we are unable to determine if there is capillary remodeling in cART-treated placentas. However, we do not expect changes in the capillary bed to make a significant contribution to the observed reflections, as arterial pulsations have been shown to be highly attenuated by the time they reach the capillaries (1). Another limitation is that the sex of the fetus was not determined. Several recent animal and human studies have reported sexual dimorphism in both placental structure and function (4, 29, 36), including UA Doppler indexes (24, 38); therefore, it is possible that some of the variability in our measurements could be the result of sex differences. A final limitation is that a large proportion of the wave reflection data sets had to be excluded (49%), an important consideration when developing an experimental design using this wave reflection methodology in pregnant mice.

The principal finding of this work is that wave reflections in the UA are sensitive to placental vascular abnormalities in the mouse. Treatment of dams with cART resulted in fetal growth restriction and an increase in the number of vessel segments in the placenta that corresponded to a significant increase in the reflection coefficient. With an increased sensitivity compared with the UA PI, this technique shows great potential for detecting placental vascular pathologies in human pregnancy. The observed reflections are associated with the geometry of the fetoplacental arterial tree, specifically the area ratio at the first branching point. Future studies using biomechanical modeling of each vessel segment will provide further information about additional sources of reflection. This information is essential for interpretation of the hemodynamic wave reflection measures, a necessary step toward translating this novel methodology to a clinical population.

GRANTS

Funding for this work was provided by Eunice Kennedy Shriver National Institute of Child Health and Human Development of Health Grant U01-HD-087177-01.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.S.C., J.C.K., C.K.M., L.S., and J.G.S. conceived and designed research; L.S.C., Y.-Q.Z., J.H., and L.X.Y. performed experiments; L.S.C. and J.H. analyzed data; L.S.C., A.R., C.K.M., L.S., and J.G.S. interpreted results of experiments; L.S.C. prepared figures; L.S.C. and J.G.S. drafted manuscript; L.S.C., Y.-Q.Z., J.H., L.X.Y., A.R., G.S., C.L.W., A.B., J.C.K., C.K.M., L.S., and J.G.S. edited and revised manuscript; L.S.C., Y.-Q.Z., J.H., L.X.Y., A.R., G.S., C.L.W., A.B., J.C.K., C.K.M., L.S., and J.G.S. approved final version of manuscript.