Uterine haemodynamics in young and aged pregnant mares measured using Doppler ultrasonography



Reason for performing study: Aged mares with endometrosis suffer higher rates of pregnancy loss than young mares, due to poor placental development. Reduced uterine blood supply may be one contributory factor.

Objectives: To measure uterine artery (UA) blood flow and other Doppler indices throughout pregnancy and compare placental and foal development in young mares and aged mares.

Methods: Thoroughbred mares were grouped according to age and endometrial biopsy score: 1) 6 young mares (mean age 7.3 years, Category I); 2) 6 aged mares (mean age 18.3 years, Category II). Vascular pathology was nil or mild except in one aged mare with moderate perivasculitis. Both UA were scanned fortnightly throughout pregnancy. Total blood flow volume (BFV, ml/min/kg bwt), peak systolic velocity (PSV, cm/s) and resistance index (RI) were determined by pulsed wave, Doppler ultrasound and UA diameter using B-mode. Mixed-effects regression analyses were used to relate vascular parameters with different predictive variables, whilst accounting for the multiple repeated measurements taken from individual horses through the duration of their pregnancies.

Results: PSV, RI and total BFV were best predicted by stage of pregnancy (P<0.001; r2>78%). The UA diameter was also associated with stage of pregnancy (P<0.001; r2= 87%) and was significantly greater in the gravid horn (P<0.001). There was a tendency for lower total BFV in older mares (P<0.05) and they delivered lighter foals than young mares (P<0.05). Gross placental morphometry was similar, but microscopic surface density of the microvilli was lower (P<0.02), in aged than young mares.

Conclusions and potential relevance: Increased uterine blood flow and decreased vascular resistance reflect fetal growth and development of the placental microcirculation. Older mares have poorer placental microvillus development and lighter foals with reduced UA blood flow. Poor uterine blood flow may be an important contributory factor for pregnancy loss in aged mares.


Mares are capable of producing viable foals throughout their lives (>20 years). However, there is an inevitable decline in their fertility and ability to maintain pregnancies with age (Morris and Allen 2002). A survey of Thoroughbred (TB) mares in the UK demonstrated that total pregnancy losses increased significantly, from 10.6 to 27.9%, in mares aged 3–8 and >18 years respectively, and the percentage of live foals born decreased from 81.8 to 62.8% (Allen et al 2007). A number of factors contribute to the decline in reproductive efficiency in older mares including chronic degenerative changes within the endometrial glands and stroma (Ricketts and Alonso 1991), sclerosis of the uterine vasculature (Grüninger et al. 1998; Schoon et al. 1999), impaired oviductal environment (Allen et al. 2006), and chromosomal or morphological abnormalities within the developing embryo (Ball 1988; Carnevale et al. 1993). In aged mares with endometrosis, placental development is compromised, particularly during the first half of gestation, by poor interdigitation and attachment between the developing allantochorion and the lumenal surface of the endometrium (Bracher et al. 1996). Consequently fetal growth is retarded and newborn foals from older mares weigh less than foals born from younger mares (Wilsher and Allen 2003). It is hypothesised that the poor placental development is associated with impaired uteroplacental (UP) blood flow, which limits fetal growth.

Measurement of UP and fetal blood flow forms an essential part of health monitoring in human obstetrics. Doppler ultrasonography measures blood flow parameters based on the Doppler shift effect (Dickey 1997). Colour Doppler imaging provides information about the direction of blood flow whereas power Doppler reflects the strength of the Doppler signal and is independent of direction. Doppler velocity signals displayed against time create a spectral waveform with systolic (S) and diastolic (D) components from which several indices can be measured. One, the resistance index (RI) of Pourcelot (RI = S – D/S), provides information about the vascular perfusion of the organ downstream from the measurement site; for example, uterine artery (UA) and umbilical artery (UmbA) RI represent impedance to blood flow within the maternal and fetal compartments of the placenta, respectively (Abramowicz and Sheiner 2008). Blood flow volume (BFV) may be calculated when the angle of insonation and the vessel diameter are also known.

Placental dysfunction in man impairs the development of the placental villous tree and its capillaries (Kingdom et al. 2000; Kaufmann et al. 2004). Consequently, placental vascular resistance increases in the UA and UmbA, and UP blood flow is reduced, thereby limiting gas and nutrient exchange and eventually fetal growth (Viero et al. 2004; Abramowicz and Sheiner 2008). When vascular impedance is high, diastolic flow may be absent or reversed and, if this persists after the first trimester, there is a greater risk of retarded growth, premature delivery and infant mortality (Harrington et al. 1997; Baschat and Hecher 2004; Urban et al. 2007). Meta-analysis demonstrates that UmbA and UA Doppler monitoring accurately predicts neonatal outcome in high-risk pregnant women (Papageorghiou et al. 2004). Therefore Doppler indices provide an important diagnostic tool for assessment of placental and fetal health during pregnancy.

Doppler ultrasound is now being applied in equine reproduction. Nonpregnant, aged mares with endometrosis have high UA vascular resistance compared with young, reproductively healthy mares (Bollwein et al. 1998; Blaich et al. 1999). Age related degeneration of the equine endometrium may contribute to poor uterine perfusion but this has not been measured during pregnancy. Therefore the study aims were to compare: 1) UA Doppler indices during pregnancy in aged mares with endometrosis vs. young, reproductively healthy mares; and 2) the effects of ageing and uterine blood flow on placental and foal development.

Materials and methods


Twelve multiparous TB mares were selected on the basis of age: 6 young (mean ± s.e. age 7.3 ± 0.2 years) and 6 aged mares (age 18.3 ± 0.7 years). An endometrial biopsy was collected from each mare during dioestrus and graded according to Kenney and Doig (1986). Microscopic vascular changes, including perivasculitis and vessel fibrosis, were also graded as none (0), mild (+), moderate (++) or severe (+++) according to Grüninger et al. (1998). The numbers of foals produced previously by the young and aged mares were 1–2 and ≥5, respectively.

The mares were maintained at pasture, stabled individually at night and fed concentratesa with hay ad libitum. They were mated to the same TB stallion and pregnancy and location (side) of the embryo determined by transrectal linear ultrasound. Doppler scanning was performed fortnightly and the pregnant mares weighed on an electronic weighbridge (Salter Universal Weighing Machine)b. At delivery, the placental membranes were collected, weighed and the gross surface area measured; 10 random biopsies of the allantochorion were recovered and fixed in 10% buffered formalin for histological processing and stereological assessment (Wilsher and Allen 2003). Newborn foals were weighed and crown–rump lengths (CRL) and abdominal circumferences (girth) were measured.

Doppler ultrasonography

The mares' right and left UA were examined per rectum. The measurement site was the main UA branch within 2–5 cm of the external iliac artery (Bollwein et al. 1998). Measurements were obtained using colour flow, pulsed wave Doppler with a 5–8 mHz microconvex probe (Sonosite Titan)c. Once the UA was visualised, the on-screen Doppler gate was positioned over the artery and the probe rotated until an angle of insonation <60° (mean ± s.e. 28 ± 3.5°) and a consistent spectral display, were obtained. Doppler indices, measured from 2 consecutive spectral waveforms using the online software, included peak systolic velocity (PSV, cm/s), RI and the time averaged mean velocities (TAM, cm/s) over one cardiac cycle. Three UA diameter and area measurements were determined at the same location using a linear array ultrasound transducer. Blood flow volume (BFV, ml/min) for the right and left UA was calculated (TAM x UA area x 60). Total uterine blood flow was determined by summing the results for both UA and expressed /kg mare bodyweight (total BFV, ml/min/kg bwt).


Statistical analyses were conducted using StataTM 9.2 softwared with statistical significance as P≤0.05. The continuous outcome (or dependent) variables were PSV, RI and total BFV and UA diameter. The predictor (or independent) variables were stage of pregnancy (weeks), mare age group (young vs. old), mare age (years), pregnant side (gravid or nongravid). Uterine biopsy score (Category I or II) was equivalent to mare age group because all young mares were Category I and all old mares were Category II.

In order to control for the potentially significant and biologically plausible effects of having multiple observations coming from the same animals (rather than truly independent observations from separate animals), mixed-effects, linear and polynomial regression models with random-effects were used; the random-effect components were applied to the horses from which the observations were taken. Preliminary univariable regression analyses were conducted to examine the relationship between Doppler blood flow indices and UA diameter and each predictor variable separately. In addition, the relationships between stage of pregnancy and these measurements were examined using fractional polynomial regression analysis to investigate possible nonlinear associations.

Multivariable linear and polynomial regression analyses were then used to investigate the relationship between these indices and simultaneous multiple predictor variables. Model building was by forward stepwise selection of variables with the final model retaining variables that had categories that were significantly associated with the outcome and/or that significantly improved the overall fit of the model. When an outcome measure was not significantly associated with the gravid or nongravid horn, a single summary value of the outcome measure was used in the statistical modelling. Average values for PSV and RI and a summed value for total BFV expressed /kg bodyweight of the mare were used. The means ± s.e. were calculated for the 2 mare age groups and plotted at fortnightly intervals through the pregnancies. Foal and placenta data for each mare group were analysed by unpaired t test using SigmaStat 2.0 statistical softwaree.

Ethical approval

All studies were performed under the Animals (Scientific Procedures) Act (1986) following approval by the Ethical Review Committees at both institutions.


Endometrial biopsies

The endometrial biopsies from the young mares were classified as healthy (Category I, Kenney and Doig 1986) with either no or mild fibrosis and perivasculitis of the vasculature (Table 1). The aged mares showed more widespread inflammatory changes (Categories IIA and IIB), although vascular fibrosis and perivasculitis were mild in all but one aged mare with moderate degeneration.

Table 1. Mare age and endometrial biopsy results during dioestrus in young (n = 6) and aged (n = 6), mares
Mare IDAge (years)Age groupBiopsy score1Vessel fibrosis2Perivasculitis2
  1. 1 Kenney and Doig (1986). 2Grüninger et al. (1998).


Doppler indices and uterine artery diameter

Table 2 summarises the results of univariable and multivariable mixed-effect linear and polynomial regression analyses of the Doppler indices and UA diameter with different predictor variables with horse as a random-effect variable. The Doppler indices of PSV, RI and BFV were similar for the UA ipsilateral or contralateral to the gravid horn both within and between mare age groups.

Table 2. Summary of univariable and multivariable mixed-effect linear and polynomial regression analyses of (uterine artery) UA Doppler indices (peak systolic velocity [PSV], resistance index [RI] and total blood flow volume [BFV]) and UA diameter with different predictor variables and horse included as a random-effect variable. (Best fitting/final models for each outcome variable are shaded)Thumbnail image of

Peak systolic velocity, RI and total BFV were best predicted only by the stage of pregnancy variable (P<0.001; r2>78%). Other significant predictors (mare age for PSV and total BFV and age group for total BFV) in univariable associations were no longer statistically significant when included with stage of pregnancy in multivariable models. Regression modelling demonstrated a positive, linear increase in PSV throughout pregnancy (Fig 1, P<0.001) which decreased with mare age (P<0.05). The RI demonstrated a nonlinear, predominantly negative association. The RI was high (>0.79) until 42 days of pregnancy, then declined rapidly to 0.54 by approximately 150 days (Fig 2, P<0.001); thereafter RI declined marginally to 0.51 at term. Uterine artery spectral waveforms with absent early diastolic flow were occasionally observed during early pregnancy (<56 days); reversed diastolic flow was never observed. Total BFV increased nonlinearly (P<0.001) as pregnancy progressed from 0.28 and 0.24 ml/min/kg bwt at 14 days after ovulation to 27.7 and 23.8 ml/min/kg bwt at full term, in young and aged mares, respectively (Fig 3). There was a trend (P<0.05) for lower total BFV in the aged than young mares, particularly during the last trimester of pregnancy.

Figure 1.

Mean ± s.e. peak systolic velocity (PSV) in uterine artery during pregnancy in young and aged mares.

Figure 2.

Mean ± s.e. resistance index (RI) in uterine artery during pregnancy in young and aged mares.

Figure 3.

Mean ± s.e. uterine artery total blood flow volume (BFV) during pregnancy in young and aged mares.

Uterine artery diameter was significantly predicted by stage of pregnancy (Fig 4) and gravid or nongravid uterine horn (both P<0.001; r2= 87.5%). The UA ipsilateral to the gravid horn was larger (P<0.001) than the contralateral UA (Table 2). The UA diameters were highest in the oldest mare.

Figure 4.

Mean ± s.e. uterine artery diameters during pregnancy in young and aged mares.

Pregnancy outcome

Eleven of the 12 mares delivered healthy foals at term. The oldest mare (age 21 years) died suddenly from a UA haemorrhage at 246 days of gestation, one week after her last examination. In the remaining mares, gestation length was similar for both age groups (Table 3). The mare and newborn foal bodyweights, and foal dimensions, were not statistically different between the 2 age groups. However, when foal bodyweights were expressed as a proportion of maternal weight, they were significantly greater (P<0.05) for foals from the young mares. The gross weights, surface areas and volumes of the placentas were similar for both young and aged mares. Stereological analyses demonstrated that the surface density of the microvilli were less (P<0.02) in the placentas from the older mares (Table 4); all other stereological measurements were similar.

Table 3. Mean ± s.e. gestation length, mare and foal morphometry
VariableYoungAged1P value
  1. CRL = crown-rump length. 1Data for the mare that died are excluded. 2Nonpregnant weight.

Gestation length (days)343 ± 3.5334 ± 3.20.105
Mare bodyweight2 (kg)547 ± 11.6601 ± 34.00.137
Foal bodyweight (kg)54.6 ± 2.149.9 ± 2.50.182
Foal/mare weight (%)10.0 ± 0.58.4 ± 0.50.042
Foal CRL (cm)95.2 ± 2.393.2 ± 1.990.540
Foal girth (cm)88.3 ± 0.6786.6 ± 1.720.537
Table 4. Macroscopic and microscopic morphometry of the placentas
VariableYoungAged1P value
  • 1

    Data for the mare that died are excluded.

Gross weight (kg)3.7 ± 0.213.4 ± 0.330.462
Gross surface area (m2)12.5 ± 0.5513.2 ± 0.820.486
Gross volume (m3)3.7 ± 0.273.3 ± 0.300.388
Volume of chorion, Vc (cm3)1317 ± 641350 ± 1660.701
Depth of chorion (mm)1.07 ± 0.071.02 ± 0.100.870
Microscopic surface density of microcotyledons, Sv (/µm)0.035 ± 0.00070.033 ± 0.00020.018
Total microscopic area of feto–maternal contact, SA (m2)46.0 ± 2.044.5 ± 5.40.662
Rv (Sv x Vc) (m2)37.2 ± 2.633.7 ± 3.40.421
Placental efficiency ( = foal body weight/SA) (kg/m2)1.20 ± 0.061.17 ± 0.120.858


Doppler indices measured in the UA relate directly to vascular perfusion of the UP tissues and correlate with fetal and placental growth (Harrington et al. 1997; Konje et al. 2003; Abramowicz and Sheiner 2008). In the present study, both young and aged multiparous TB mares had significant rises in UA PSV and total BFV, and decline in RI, during pregnancy. The timing and the pattern of these changes correlated closely with placental and fetal development. Between fertilisation and Day 40 of gestation in the mare, the embryonic vesicle enlarges slowly and the allantochorion is only partly developed (Allen and Stewart 2001). During this period, total BFV increased slightly while RI values remained high with some spectral waveforms displaying absent early diastolic flows, indicating high uterine vascular impedance (Fig 2), similar to nonpregnant mares (Bollwein et al. 1998). Between Days 40 and 150, UA RI declined substantially which coincided with intense angiogenesis in, and great expansion of, the allantochorion (Allen and Stewart 2001). The nascent microvilli develop into distinct, highly vascularised microcotyledons containing complex, branched capillaries and the microcotyledons interdigitate with the endometrium over the whole uterine surface (Samuel et al. 1975; Abd-Elnaeim et al. 2006). The decrease in UA RI was accompanied by a corresponding rise in total BFV (from approximately 0.3 to 3 ml/min/kg bwt between Days 40 and 120) as the metabolic demands of the placental and fetal tissues increase. Therefore, during the first half of equine pregnancy, the vasculature of the gravid uterus is transformed from high resistance vessels with low flow, to low resistance vessels with high blood flow, and the timing of this transition correlates closely with the onset of placental angiogenesis. A similar decline in UA RI occurs in human pregnancies at the end of the first trimester, as the spiral arteries invade the UP vascular beds (Dickey 1997; Kingdom et al. 2000; Whitley and Cartwright 2010).

During the second half of pregnancy, RI changed little in both mare age groups. Although the microcotyledons continue to grow as the capillaries elongate and branch, these vessels also become narrower or flattened; therefore, the vascular impedance within the UP tissues remained relatively low (MacDonald et al. 2000; Abd-Elnaeim et al. 2006). In contrast, UA total BFV continued to rise throughout pregnancy. Increased blood delivery to the UP tissues is influenced by factors other than vascular impedance. These include counter current blood flow between the maternal and fetal capillaries, a decrease in their perfusion distance, increases in fetal umbilical blood flow, fetal heart rate and blood pressure, and other fetal circulatory changes (Samuel et al. 1975; Fowden et al. 2000; Giussani et al. 2005; Abd-Elnaeim et al. 2006). Moreover, on the maternal side of the placenta, there is redistribution of blood flow from other tissues towards the gravid uterus and maternal cardiac output increases through increased heart rate and stroke volume; >20% of maternal cardiac output is redirected towards the uterus compared to 0.5% when nonpregnant (Silver et al. 1982). These maternal circulatory adaptations are as important for fetal growth as those within the placental vasculature (Konje et al. 2003). In the present study, the significant increases in UA PSV, vessel diameter and reduced RI, observed during pregnancy facilitated increased blood delivery to, and within, the UP tissues. The trend for lower PSV values in the older mares is not surprising given that cardiovascular function declines with age (Betros et al. 2002).

During late gestation (Day 210 to term), total BFV increased approximately 3 fold in both the young and aged mare groups. This rise correlates closely with the reported 3–4 fold increase in fetal body mass over the same gestation period (Fowden et al. 2000). Thus UA total BFV increases in proportion to fetal growth. Comparable correlations have been described for sheep, man and cattle, in which BFV increases 3, 2.5 and 4.5-fold, respectively, during the second half of pregnancy (Reynolds et al. 2006). There was a tendency for greater total BFV in young compared with aged mares, and their foals weighed proportionally more. This lends support to the previous hypothesis that uterine blood flow and fetal growth is reduced in older mares due to compromised placental development (Bracher et al. 1996; Wilsher and Allen 2003). These authors reported reduced microcotyledonary surface density (Sv) with shorter and fewer microvilli in aged pregnant mares exhibiting endometrosis thereby leading to a substantial reduction in fetal bodyweights. In the present study, although the gross placental parameters were similar, at the microscopic level Sv was significantly lower in the older mares. The Sv provides an estimate of villous vascularisation, with lower values indicating a reduced capacity for nutrient and gas exchange (Kaufmann et al. 2004). Therefore impaired placental development appears to have contributed to the reduced fetal growth observed in the aged mare group.

Newborn foals from the older mares weighed proportionally less than foals from the younger mares indicating limited nutrient supply. Adequate UP blood flow is critical for fetal nutrient delivery although placental transport capacity, nutrient uptake and placental surface area and vascularity also play a role (Fowden et al. 2000; Reynolds et al. 2006). Weight specific glucose utilisation by the equine fetus declines between mid and late gestation while the UP tissues themselves use a large (>60%) proportion of the nutrients supplied via the UA (Fowden et al. 2000). Hence the fetus is particularly sensitive to small changes in nutrient supply. Measurement of umbilical blood flow will determine whether nutrient (blood) delivery is reduced (in the fetuses of the aged mares); originally this was an objective using a 3.5 MHz Doppler probe and transabdominal approach. However, due to the location, mobility and coiling of the umbilical cord, the angle of insonation (necessary for blood flow calculations) could not be determined precisely. Further work is needed to confirm whether such measurements can be performed accurately in late pregnant mares.

Age-related differences in RI were not observed in this study. In contrast, Blaich et al. (1999) demonstrated that nonpregnant, older mares with Category IIB or III endometrial biopsy scores had higher UA RI than young, healthy mares. Grüninger et al. (1998) suggested that pathological changes within the endometrial vasculature caused high vascular impedance, and is consistent with results from nonpregnant, infertile women (Dickey 1997; Ferreira et al. 2007). The absence of higher RI values in aged mares in the present study may reflect the relatively minor pathological changes observed in their endometrial vasculature, in contrast to more marked pathology within their glandular tissue. Indeed, significantly elevated vascular resistance may only be present when there is moderate or severe vascular pathology. In women, absent or reversed end diastolic flows in UmbA are only observed when 50–70% of the placental vessels are abnormal (Baschat and Hecher 2004).

No differences in Doppler indices were detected between the UA ipsilateral and contralateral to the gravid uterine horn, as reported previously (Bollwein et al. 2004). However, larger diameter UA vessels were observed ipsilateral to the pregnancy in both mare groups. Although no age effect was observed, the ipsilateral UA diameters were highest in the oldest mare that died from spontaneous UA rupture. This is a common cause of death in older, multiparous broodmares because of progressive vascular damage caused by repeated cycles of pregnancy-associated vessel remodelling and involution. Mares with UA diameters above the normal range for their age group may be at greater risk of UA rupture during pregnancy (Grüninger et al. 1998; Schoon et al. 1999).

In conclusion, young and aged mares showed UA haemodynamic changes consistent with a transition from a high resistance, low flow to low resistance, high flow system as pregnancy progressed, and the timing of such changes correlated closely with placental angiogenesis and exponential fetal growth during early and late gestation, respectively. The similarity in Doppler indices between the mare age groups is supported by the absence of severe pathological changes in the endometrial vasculature and glandular tissue of the aged mares. However, the tendency for reduced uterine blood flow together with reduced placental microvillus surface densities and lighter foals at term in the aged mares indicates that they are susceptible to placental and fetal compromise as a consequence of reduced uterine perfusion.

Conflicts of interest

No conflicts of interest have been declared.

Source of funding

This project was funded by the Horserace Betting Levy Board. The Doppler ultrasound scanner was kindly funded by the European Breeders Fund and local Newmarket studfarms.


The authors wish to thank Dr Sandra Wilsher for the stereological analysis, Professor Margaret Ramsey and Dr Eileen Bradley for technical advice and Drs Aenne Honnens, Juliana Körte and Ina Woschee for practical help with the Doppler ultrasonography.

Manufacturers' addresses

a Dodson & Horrell, Islip, Northamptonshire, UK.

b Salter Ltd, Malton, North Yorkshire, UK.

c BCF Technology, Livingston, Scotland.

d StataCorp, College Station, Texas, USA.

e SPSS Inc, Chicago, Illinois, USA.


Author contributions

Contribution by the authors (%)