Human Reproduction Update 1997, Vol. 3, No. 5 pp.

467503

European Society for Human Reproduction and Embryology

Doppler ultrasound investigation of uterine and ovarian blood flow in infertility and early pregnancy
Richard P.Dickey1
Fertility Institute of New Orleans, New Orleans, LA and Division of Reproductive Endocrinology, Department Obstetrics and Gynecology, Louisiana State University School of Medicine, New Orleans, LA, USA

TABLE OF CONTENTS Introduction Basis of Doppler ultrasound Methods of analysis Velocity waveform analysis Uterine blood flow Ovarian blood flow Uterine blood flow in the first 16 weeks of normal pregnancy Uterine blood flow in abnormal pregnancy New areas of investigation Conclusions References 467 468 470 470 477 484 486 493 494 498 499

infertility and early pregnancy and methods of treatment for the same. Key words: abortion (miscarriage)/Doppler ultrasound/pregnancy first trimester/uterine blood flow Introduction Doppler ultrasound analysis has been used in gynaecology primarily to determine blood flow in ovarian tumours with neoplastic characteristics, and in obstetrics to examine the relationship of blood flow in the uterine and umbilical artery to adverse fetal outcome. These topics have been the subject of numerous reports and are extensively covered in a number of excellent text books. The relationship of uterine blood flow to ovulation and implantation and to normal development in early pregnancy has been less extensively studied. Because these events represent a continuum from preovulation to development of the placental circulation, they are subject to many of the same physiological and hormonal influences. The present paper begins with a review of Doppler ultrasound techniques from the viewpoint of the clinician unfamiliar with principles of Doppler physics, and compares methods of describing blood flow results. The seminal article on use of Doppler ultrasound in infertility was that of Goswamy and Steptoe (1988). Those authors were the first to suggest that abnormal uterine artery blood flow might be associated with infertility, and to develop a classification of uterine artery blood flow waveforms. They related specific abnormal waveforms with repeated failure of implantation in in-vitro fertilization (IVF) patients and successfully improved uterine circulation and implantation by use of therapeutic doses of oestrogen (Goswamy et al., 1988). Subsequently, other authors have confirmed a relationship of uterine and

This review describes the current use of Doppler ultrasound to examine blood flow in the uterus and ovaries in infertile patients and during early pregnancy. The basics of Doppler ultrasound and the different methods of measuring blood flow are discussed from the viewpoint of the clinician who may be unfamiliar with Doppler physics and terminology. Normal values in the menstrual cycle and the relationship of uterine and ovarian blood flow to infertility and to implantation following in-vitro fertilization are presented. Normal values for uterine blood flow in the first 16 weeks of pregnancy and the effect of sex steroids and ovulation induction on their values are described. The possible relationship of defective uterine blood flow to recurrent abortion is examined. New areas of investigation, such as the effect of standing on blood flow, and the effect of drugs are explored. The findings of this review indicate that Doppler blood flow studies may provide significant information about possible causes of some disorders of
1Correspondence

should be sent to: Fertility Institute of New Orleans, 6020 Bullard Avenue, New Orleans, LA 70128, USA. Telephone: (504) 2468971;

fax: (504) 2469778

468

R.P.Dickey

ovarian blood flow to unexplained infertility (Kurjak et al., 1991; Steer et al., 1994) and to successful implantation following IVF (Sterzik et al., 1989; Steer et al., 1992; Favre et al., 1993; Balakier and Stronell, 1994; Coulam et al., 1994; Serafini et al., 1994; Tekay et al., 1995a; Cacciatore et al., 1996; Zaidi et al., 1996a,c). This review continues with an examination of uterine blood flow in normal and abnormal early pregnancy, because by understanding uterine blood flow requirements of early pregnancy, we may eventually be able to learn more about the causes of early pregnancy loss and abnormal placental development associated with complications of late pregnancy. This section begins with a review of the development of the placental circulation. Also considered is the relationship of sex steroids (Jauniaux et al., 1992a, 1994; Dickey and Hower, 1996) and ovulation induction (Dickey and Hower, 1995c) to uterine blood flow in pregnancy. Lastly, two new areas are reviewed: the effect of standing on uterine blood flow and the effect of drugs on uterine blood flow.

Basis of Doppler ultrasound Doppler ultrasound is based on the Doppler principle, named after Christian Andreas Doppler (died 1853), that sound and light waves change in frequency or wavelength when either the source or the receiver is moving. Doppler-shifted echoes are generated when vessel walls or blood are in motion. These waveforms are familiar as sounds of heart motion on fetal ultrasound and of pulsatile flow of blood through vessels. The frequency of sound waves is expressed in Hertz, named after Heinvich Rudolf Hertz (died 1894). One Hertz (Hz) is one sound wave cycle or pulse occurring in 1 s. Pulse repetition frequency is usually given in kilohertz (kHz); 1000 Hz/s are equal to 1 kHz and 1 106 Hz/s are equal to 1 megahertz (MHz). Sound with a frequency 20 000 Hz is called ultrasound, because it is beyond the frequency range of human hearing. Ultrasound transducers operate on the principle of piezoelectricity, by which certain materials produce a voltage when deformed by applied pressure and produce a pressure when voltage is applied. Various formulations of lead zirconate titante (PZT) are commonly used in transducers. Source transducers operating in a continuous mode are driven by an alternating voltage and produce an alternating pressure that propagates as a sound wave. The frequency of sound, called resonance frequency, is equal to the frequency of the driving voltage. For each pulse of ultrasound, a series of echoes are returned as the ultrasound pulse is reflected off objects at a greater or lesser distance. These echoes are received by the transducer and converted

into electrical energy, which is processed electronically and displayed as a series of dots in a single scan line on the display. Additional pulses travelling along the same path result in the same scan line being displayed, but if the starting point for each subsequent pulse is different while the direction of the path is unchanged, a cross-sectional image builds up. The rectangular display that results is called a linear scan. If each pulse originates from the same starting point but the direction of the path is slightly different from the previous point, the result is a pie-shaped sector scan. The terms B scan and grey scale are often used to describe linear and sector scans, because the strength of each returning echo is represented by its brightness on the display (B for brightness and grey because the resulting image is grey). These B or grey scale images are refreshed with current data at a rate of 30 frames/s. Colour flow Doppler instruments provide twodimensional, colour-coded Doppler information superimposed on the real time anatomical display. In Doppler ultrasound, the change in frequency of the returning echoes, with respect to the emitted frequency, results in the so-called Doppler shift. Doppler shifts are usually in the range of 100 Hz to 11 kHz. The Doppler shift is dependent on the speed of blood flow (blood circulates at a speed of 10100 cm/s), the angle between the source of sound (the transducer), the direction of the blood vessel being measured and the operating frequency of the Doppler. Higher operating frequencies, greater flow speeds and smaller Doppler angles produce larger Doppler shifts. At the operating frequency of 5 MHz typically found in vaginal transducers, blood flowing through the uterine artery at a velocity of 50 cm/s produces a Doppler shift of 6.5 kHz at an angle of 0, 5.6 kHz at 30 and 4.6 kHz at 60. Blood flowing through spiral arteries or in ovarian stroma at 10 cm/s produces Doppler shifts of 0.65 kHz at an angle of 0, 0.56 kHz at 30 and 0.32 kHz at 60 (Kremkau, 1990). Continuous wave Doppler utilizes separate voltage generators and receivers to create a picture of the Doppler shifts in the area being scanned similar to the grey scale, except that moving objects, e.g. blood flow, appear as red or blue against a black background. Slow movements, such as bowel peristalsis and vessel walls are scanned out by wall (also called wall thump) filters which reject selected ultrasound frequencies, usually between 50 and 3200 Hz. Continuous wave systems provide motion and flow information without depth information or selection capability. Pulsed-Doppler systems have the ability to select the depth from which Doppler information is received, thus

Doppler ultrasound study of uterine and ovarian blood flow

469

allowing analysis of blood flow within a single vessel. To do this, the vessel to be studied is first located with continuous wave ultrasound. Next, a gate is placed over the vessel which passes only signals that are returned within a defined time. For example, a gate that passes echoes averaging 1315 s after pulse generation is listening at a depth range of 10.011.6 mm. The width of the gate (also called the volume box) is adjusted to the diameter of the vessel. The returning Doppler frequency shift echoes are converted electronically by a mathematical technique called fast Fourier transformation and displayed as the Doppler shift versus time waveform. The Doppler waveform represents changes in the velocity of blood flow during the cardiac cycle. Flow conditions at the site of measurement are indicated by the width of the velocity spectrum, with spectral broadening indicative of disturbed and turbulent flow. Flow condition downstream, particularly distal flow impedance, is indicated by the relationship between peak systolic and end diastolic flow speeds. Continuous forward blood flow is possible because of vessel wall elasticity, also called the windkessel effect. When pulse pressure forces fluid into a compliant vessel, it expands and increases the volume within it. Later, when the pressure is reduced, it contracts, producing extended flow later in the cycle. In the aorta and large vessels with valves, and also in small vessels with little distal impedance, the windkessel effect produces continuous forward blood flow. In the iliac arteries with no valves, reversal of flow occurs in early diastole (triphasic flow) as pulse pressure decreases and distended vessels contract. In the ascending uterine artery, early diastolic flow is usually absent or reversed in non-pregnant patients during menses and the early proliferative phase of the cycle. Absence of flow at any time during diastole on the day of human chorionic gonadotrophin (HCG) administration in IVF cycles is associated with failure of implantation. In pregnancy, absence of early diastolic flow or a sharp decrease in early diastolic flow, also called a protodiastolic notch, is usually due to increased impedance in distal vessels, and is associated with poor pregnancy outcome if it persists. Lack of diastolic flow that occurs only at the end of diastole may be due to low pulse pressure or to a prolonged interval between heart beats, and is considered less important than absence of early diastolic flow. There is an upper limit to the Doppler shift that can be detected by pulse instruments. If the Doppler shift frequency exceeds one half the pulse repetition frequency, a phenomenon called aliasing occurs, in which the peak of the velocity waveform appears below the base line. Aliasing is the most common artefact encountered in Doppler ultrasound. Aliasing can be eliminated by increasing pulse repetition frequency, by increasing the

Doppler angle, which decreases the Doppler shift for a given flow, or by baseline shifting. As frequency is increased, imaging depth decreases and ambiguity occurs because of attenuation, and because the next pulse is emitted before all the echoes from the previous pulse have been returned. Attenuation is the decrease in amplitude and intensity which occurs due to absorption (conversion to heat), reflection and scattering as sound travels through tissue. For soft tissue, attenuation in decibels (dB) is ~0.5 dB of attenuation per cm for each MHz of frequency. Pulse repetition frequencies are usually in the range 520 kHz. At a pulse repetition frequency of 5 kHz, ambiguity occurs at a depth beyond 15 cm, and aliasing occurs with Doppler shift above 2.5 kHz. At a pulse repetition frequency of 20 kHz, ambiguity occurs at a depth beyond 4 cm and aliasing occurs with Doppler shifts above 10.0 kHz. Increasing the Doppler angle to eliminate aliasing increases the chance of error in detecting flow velocity. Flow velocity and flow volume can only be determined accurately if the angle of insonation is accurate. Measurement of the angle of insonation is usually done by the operator orienting a line on the anatomic display, so that it is parallel to the direction of blood flow. The resulting angle between the transducer and vessel may be included in the mathematical calculation of velocity by the ultrasound computer program. A 5 error by the operator in placing the directional line causes error in measurement of velocity of <2.0% when the real angle is <10%. This increases to 5.4% if the real angle is 30, and to 12% when the angle is 60. Accurate determination of velocity is not necessary when ratios of systolic to diastolic velocity are used to interpret blood flow, but these ratios themselves become inaccurate when blood flow is not continuous through the cardiac cycle. It is important, therefore, to keep the angle of insonation as small as possible, consistent with determination of continuous waveforms. Moving the baseline is successful in eliminating aliasing only if there are no legitimate Doppler shifts in the region of aliasing. The width of the Doppler ultrasound volume box, compared to the width of the vessel wall, may also affect velocity estimation. In small vessels, the average velocity may be only one half the velocity at the centre of the stream, because of turbulence caused by friction from the vessel wall. Volume box widths which are too small in relation to vessel size and which are aimed at the centre of the stream may result in overestimation of velocity. The ascending uterine artery diameter width ranges from 0.2 to 0.5 cm in non-pregnant patients and becomes much larger in pregnant patients. Spiral arteries are closer to 12 mm in diameter. Volume box widths that are available usually start at 1 mm and increase to 10 mm in 1 mm increments.

470

R.P.Dickey
Table I. Waveform classification
Original classificationa Type C Revised classificationb Type C Description Diastolic component continuous with previous systolic component and present throughout the cardiac cycle Diastolic component continuous with the previous systolic component but not present at the end of the cardiac cycle Diastolic component present at the end of the cardiac cycle, but not continuous with the systolic component Diastolic component present, but neither continuous with the systolic component nor present at the end of the cardiac cycle No diastolic component present No systolic or diastolic component present Present classification Type C

Type B

Type D-I

Type B

Type A

Type D-II

Type A

None

Type D-III

Type D

Type 0 None
aReprinted bFrom

Type D-IV Type N

Type 0 Type N

with permission from Goswamy and Steptoe (1988). Dickey et al. (1994).

Visual and mathematical analyses of pulse waveforms using velocity ratios and measurement of flow volume, obtained by multiplying average velocity by vessel cross-sectional area, are the tools used to determine blood flow status of vessels. Analysis of waveforms in larger vessels can be used to determine blood flow in smaller distal vessels, even when the distal (downstream) vessels are too small to be visualized individually, as in the case of spiral arterioles in the myometrium or stroma of the ovary. Methods of analysis Doppler blood flow may be analysed in three ways by: (i) waveform, (ii) resistance indices and (iii) flow volume or velocity. There are situations in which each of these methods of analysis is best and should be used, and situations in which each is unreliable. Flow volume analysis is the closest to true blood flow, but the most difficult to perform, because it is dependent on the angle of insonation, accurate measurement of vessel diameter, the tortuosity of the vessel and the analytical power of the instrumentation. Flow volume analysis can only be used for larger vessels, such as the ascending uterine arteries, but may sometimes be used for ovarian arteries and umbilical vessels. Resistance indices measure downstream impedance to blood flow and are independent of the angle of insonation, but are only indirect estimates of flow volume. They are most useful for estimating blood flow in vessels distal to the point of examination. Despite claims to the contrary, resistance indices may be highly inaccurate when blood flow in the vessel being measured is not continuous throughout the cardiac cycle. Waveform analysis provides the most accurate estimate of blood flow in conditions

when blood flow is not continuous throughout the cardiac cycle, as occurs in the uterine and, sometimes spiral, arteries in the proliferative phase of the cycle. Non-continuous flow may persist into the luteal phase and early weeks of pregnancy. Waveform analysis utilizes descriptive terms, but because it is not subject to computer analysis, it is too often omitted in clinical publications of studies of infertility and early pregnancy. Velocity waveform analysis Velocity waveform analysis of the umbilical and uterine arteries is the gold standard for evaluating normal and abnormal blood flow in the second and third trimesters (Fleischer et al., 1986; Harrington et al., 1991, 1996; North et al., 1994). A deflection or notch in late systolic or early diastolic flow is characteristic of normal uterine artery blood flow waveforms until the end of the second trimester. Persistence of an early diastolic notch into the 24th post-menstrual week, especially if it occurs in both uterine arteries, is associated with maternal hypertension and intrauterine growth retardation (IUGR; Harrington et al., 1996). Absence or reversal of early diastolic flow in the umbilical artery is associated with IUGR in 84% of pregnancies in cumulative series and perinatal mortality in 36% of pregnancies (Farin et al., 1995). Goswamy and Steptoe (1988) developed a highly valuable classification of flow velocity waveforms for use in non-pregnant patients (Table I). In their classification, the term type C was used for waveforms in which diastolic flow was continuous with systolic flow and present through the cardiac cycle (Figure 1). Type A was used to designate waveforms in which the diastolic component was present in

Doppler ultrasound study of uterine and ovarian blood flow

471

Figure 1. Waveform type C. Uterine flow velocity waveform showing high systolic flow and diastolic wave extending to the next cardiac cycle. This indicates good uterine perfusion. Reprinted with permission from Goswamy and Steptoe (1988).

Figure 3. Waveform type B. Uterine flow velocity waveform showing diastolic wave continuous with systole, but not extending to the next cardiac cycle. This indicates good uterine perfusion. Reprinted with permission from Goswamy and Steptoe (1988).

Figure 2. Waveform type A. Uterine flow velocity waveform showing high systolic flow and absence of early diastolic flow. This indicates poor uterine perfusion. Reprinted with permission from Goswamy and Steptoe (1988).

mid diastole, but was absent in early diastole, and might or might not be present in later diastole (Figure 2). They considered this to indicate poor perfusion and to be the result of distal impedance to blood flow. Type B waveforms were those in which the diastolic component was continuous with the preceding systolic component, but did not extend to the systolic component in the next cardiac cycle (Figure 3). They considered this type of waveform to indicate low impedance and adequate perfusion. Type 0 waveforms were those in which no diastolic component was present at any time and

indicated high distal impedance and low perfusion (Figure 4). Dickey et al. (1994) revised this classification by substituting the terms D-I, D-II and D-IV (D for discontinuous) for waveforms B, A and 0 respectively, and by adding two additional types: D-III for the combination of type A and type B when flow was present in mid diastole, but absent in both early and late diastole, and type N, when blood flow was undetectable in both systole and diastole. This proved too cumbersome and, as a result, they subsequently used Goswamy and Steptoes original classification with two additions: type D to represent the combination of type A and type B, i.e. absence of both early and end diastolic flow, with some flow in between, and type N for undetectable blood flow in systole, as well as in diastole. In Goswamy and Steptoes original (1988) classification, type D would still have been classified as type A. Goswamy and Steptoe (1988) developed their waveform classification because they believed that resistance indices, such as the pulsatility index of Gosling et al. (1971) and the resistance index of Plainiol et al. (1972), inaccurately estimated uterine artery blood flow when diastolic flow was absent during any part of the cardiac cycle. They subsequently proved the validity of their classification system when they demonstrated that total absence of diastolic flow (type 0) and absence of early diastolic flow (type A and type D) waveforms were associated with repeated failure of implantation in IVF patients, whereas continuous flow (type C) and absence of flow only at the end of diastole (type B) were associated with normal implantation rates (Goswamy et al., 1988). In this authors

472

R.P.Dickey

Figure 5. Common resistance indices: (S) peak systolic velocity, (D) end diastolic velocity, (A) time averaged maximum velocity. RI = resistance index, (S D)/S;, PI = pulsatility index, S D/A; S/D ratio, S/D. TAMV = time averaged maximum velocity. Figure 4. Waveform type O. Uterine flow velocity waveform showing absence of any diastolic flow. This indicates very high impedance. Reprinted with permission from Goswamy and Steptoe (1988).

Fourier pulsatility index (PIF)

experience, undetectable blood flow in both systole and diastole (type N) in spiral arteries is no more serious than type A or D, because it is often due to lack of sensitivity of the instrumentation. Waveforms are unequalled for detection of downstream impedance. They should be used instead of resistance indices, flow volume or velocity estimations whenever blood flow is not continuous throughout the cardiac cycle.
Resistance indices

This index is considered most accurate for detecting downstream resistance, but is too complicated to be calculated automatically by present day software packages. It is expressed by: PIF = N = 1[Vn2/Vo2] where Vo is the mean forward velocity of the waveform, N is the harmonic number from 1 to infinity, and Vn is the harmonic of the velocity waveform (Powis and Schwartz, 1991).

Peak to peak pulsatility index (PPI)

Resistance indices are preferred for measurement of small and tortuous vessels, because they are based on the ratio of peak systolic flow to some measure of diastolic flow and thus, are independent of the angle of insonance (Figure 5). However, the angle of insonance is of little significance if blood flow is not continuous, because the true ratio cannot be known. Resistance indices are necessary when waveforms are continuous to define the magnitude of downstream resistance. Resistance indices were developed for use in cardiovascular disease and were specifically intended to detect stenosis of distal (downstream) vessels. As such, they may not be entirely applicable to evaluation of infertility and early pregnancy. The principle reason for using indices is that they cancel out errors caused by the Doppler angle, machine gain, and hypo- or hypertension. They are, however, affected by the heart rate (Maulik, 1993).

This index is used in cardiovascular disease for measurement of high resistance vascular beds where flow is reversed in early diastole (triphasic flow), and is rarely calculated in obstetric or infertility studies. The PPI measures the total distance from the top to the bottom of the systolic peak and divides this by the mean velocity over the cardiac cycle. It is expressed by: PPI = S/velocitym where S is the peak systolic velocity and velocitym is the time averaged maximum velocity over the cardiac cycle. Results are closely similar to the Fourier pulsatility index. The PPI normally increases as increasingly distal vessels are examined (Baker et al., 1986). The presence of long-term occlusive disease proximal to the site of measurement decreases the index as a result of distal vasculative dilation to compensate for the proximal stenosis.

Doppler ultrasound study of uterine and ovarian blood flow

473

Resistance index (RI)

Other ratios

This index, also known as Pourcelots ratio (Pourcelot, 1974), examines the difference between the peak systolic and end diastolic velocity and is expressed by: RI = (S D)/S where S is the peak systolic velocity and D is the minimum or end diastolic velocity. The RI is suitable for low resistance vascular beds with continuous flow throughout diastole. If the end diastole value (D) goes to zero, the ratio converges to 1.
Pulsatility index (PI)

Goswamy and Steptoe (1988) modified the S/D ratio and RI by substituting the peak diastolic frequency for the end diastolic frequency, but these modified ratios have not been used by others. The (A B)/A ratio tests for the presence of the late systolic inflection in the carotid artery waveform. In this ratio, A is the peak systolic velocity and B is the lowest velocity at the inflection in late systole. Designed specifically for use in the common carotid artery, it has not, as yet, been used for the uterine, spiral, or umbilical arteries, but could conceivably be used for vessels, such as the uterine artery prior to the 20th week of gestation, where an early systolic notch is present.
Blood flow volume and velocity

This index, also known as the mean pulsatility index to distinguish it from the peak to peak pulsatility index, is expressed by: PI = (S D)/velocitym where S is the peak systolic velocity, D is the end diastolic velocity and velocitym is the time averaged maximum velocity over the cardiac cycle (Gosling et al., 1971). Velocitym is calculated from the average of three or four cardiac cycles. No correction for heart rate is required (Gosling et al., 1991). Because it does not go to 1.0 if early or end diastolic flow goes to 0, the PI is often used for vessels where flow is absent during all or part of diastole. In a comment made 20 years after first describing the PI, Gosling stated, When blood flow sonograms are studied just proximal to a vascular bed of low impedance, it is likely that small changes in something itself already small will make little difference to the waveform shape, which will mostly be affected by changes in pressure drop across the vascular bed and not the impedance per se (Gosling et al., 1991). The PI is not as accurate as RI because of the variability inherent in measurement of mean velocity with present day software programs. Trudinger (1991) states, In determining the PI, it is most unlikely that the true mean velocity can be calculated entirely accurately.
Systolic/diastolic ratio (S/D)

By far the most direct way to describe blood flow is to report blood flow volume or average velocity. Either flow volume or average velocity plus a resistance index or waveform (when diastolic flow is not continuous) are necessary to analyse uterine blood flow satisfactorily. Flow volume and mean velocity describe the actual volume or velocity of blood utilized by an organ, while resistance indices and waveforms indicate the state of smaller vessels downstream from the vessel being analysed.
Flow volume

The systolic/diastolic ratio is the simplest of all indices and is expressed by S/D, where S is the peak systolic frequency and D is the end diastolic frequency. It is less frequently used, now that the PI and RI can be readily calculated by built-in software programs. Errors in the S/D ratio increase as diastolic velocity becomes small. Spencer et al. (1991), using an in-vitro flow model, found that when flow was reduced in a stepwise fashion, PI and RI increased in linear fashion, whereas the S/D ratio increased exponentially.

Flow volume is the only true measurement of blood flow. It is expressed by: Vol = Vel A where Vol = flow volume (ml/min), Vel = time averaged mean velocity throughout the cardiac cycle and A = cross-sectional area of the vessel, derived from the diameter (A = r2). Note: Time averaged mean velocity as used here is not the same as time averaged maximum velocity sometimes reported independently, and used in calculating PI and RI. Blood flow volume through vessels of 2 mm can be measured by Doppler ultrasound. Velocity should be calculated from the average of three or four cardiac cycle waveforms. Diameter should be calculated from the mean of at least four measurements of vessel diameter made from frozen two-dimensional grey-scale images. Flow volume measurements are dependent on the operators accuracy in aligning the angle of insonation and in measuring diameter, plus the instruments ability to calculate the average velocity. It is difficult to measure accurately in tortuous vessels.
Time averaged maximum velocity (TAMV)

Time averaged maximum velocity (TAMV) is frequently used when analysing blood flow through small vessels with no collateral circulation, such as the umbilical and ovarian

474

R.P.Dickey

arteries, when diameter is not or cannot be measured. It is determined by measuring the average maximum velocity over a minimum of three cardiac cycles. In many cases, TAMV mirrors flow volume; however, velocity is dependent on the angle of isonation and, at points of vessel narrowing, increases to compensate for decrease in vessel diameter. Therefore, TAMV may give a false impression of blood flow in vessels narrowed by disease or stretching.
Other measurements of velocity

Maximum peak systolic velocity (MPSV), or simply peak systolic velocity (PSV) and minimum diastolic velocity (MDV), are sometimes reported. These expressions of velocity are highly unreliable, since they are dependent on the angle of insonation, display a high degree of variability on repeat examination (Tekay and Jouppila, 1996), and cannot describe blood flow over the entire cardiac cycle. Common abbreviations used for blood flow measurements are shown in Table II.
Table II. Common abbreviations used when reporting blood flow measurement
PI RI S/D ratio Vmax MDV PPI PSV MPSV TAMV TAMX TAPV Pulsatility index Resistance index, Pourcelots index Systolic/diastolic ratio Velocity maximum (same as PSV and MPSV) Minimum diastolic velocity Peak to peak pulsatility index Peak systolic velocity (same as MPSV and Vmax) Maximum peak systolic velocity (same as PSV and Vmax) Time averaged maximum velocity (same as TAMX and TAPV) Time averaged maximum velocity (same as TAMV and TAPV) Time averaged peak velocity (same as TAMV and TAMX)

Sources of error in measurement

All blood flow measurements, whether they are of the uterus and ovary or the aorta and carotid artery, are subject to many potential sources of error. Certainly, the most important source in the uterus and ovary is the operators judgement in selecting the vessel to be examined and the particular part of the vessel on which to focus the Doppler. Selection of the particular waveform or waveforms to analyse, out of the many available, is an equally important operator decision. It must be decided whether to select ones with the highest peak systolic or diastolic velocity, least velocity, or an average example. Doppler blood flow studies of the uterus and umbilical circulation have been published which use each of these choices. Doppler ultrasound operators may cause errors in PI and RI if they mistake brief gaps in diastolic flow as absence in diastolic flow. Analysis of flow volume or resistance indices in a

single uterine artery, instead of both arteries, can lead to a false interpretation of uterine blood flow because of the considerable cross circulation in the uterus. Analysis of the relationship between uterine artery flow volume or RI and spiral artery waveforms or RI show that spiral artery blood flow is not affected by a marked decrease of flow in a single uterine artery, but is affected by lesser decreases in flow in both uterine arteries (Dickey et al., 1994). The reader must determine from the Methods section of a paper which methods have been used to select vessels, in order to interpret the results and the relationship of the authors findings to other studies. The angle of insonation is of greatest concern when measuring velocity and volume. The importance of angle of insonation is obviated for resistance indices which do not incorporate velocity in their calculations (e.g. RI, but not PI). For angles <20, the maximum error in calculating velocity and volume caused by a 5% error in estimating the angle of insonation is 1.5% and is entirely obviated when angle adjustment is incorporated in the analytical software. Another potential source of error in determining flow volume is measurement of vessel diameter, both because the diameter changes during cardiac pulsation and because any errors in measurement are magnified when squared. However, the normal distensibility of arteries in adults is <5% for a 40 mm Hg pulse pressure (Gosling et al., 1991). Errors due to measurement become less important as vessel diameter increases. For example, a 1 mm error in measuring a 20 mm artery diameter results in 10% error in flow volume, but if the diameter is 50 mm, a 1 mm error causes only a 4% error. At least four separate measurements of diameter taken from sequential cardiac cycles must be made. The actual difference found in vessel diameter when repetitive measurements of diameter were made by the same observer was 3%, and when measurements were made by different observers was 5.6% (Dickey et al., 1994). The relative importance of different sources of variability on blood flow volume measurement is illustrated in Table III, where interoperator variability was determined in 10 subjects using the paired t-test, as described by Bewley et al. (1993). After determining that waveforms were continuous in all vessels, the right and left ascending uterine arteries were examined in 10 non-pregnant volunteers during the mid-luteal phase by one of two experienced ultrasonographers and immediately afterward by the second operator, with the order reversed in half of the patients. Mean velocity was measured with the transducer held in the position which resulted in the highest PSV, and the average of three cardiac cycles was determined by 64 point algorithm. The angle of insonation was corrected by software. The least correlation between operators was found for flow volume (r = 0.88),

Doppler ultrasound study of uterine and ovarian blood flow

475

followed by mean velocity (r = 0.90), with much less variability found for vessel diameter (r = 0.94). The best correlation between operators was for RI (r = 0.97). Other comparisons of intra-observer variability have yielded similar results (Pearce et al., 1988; Bewley et al., 1993). Tekay and Jouppila (1996) assessed intra-observer variability for three sources of variation (beat to beat, between frame, and temporal variability) in measurements of PI and MPSV in uterine arteries and PI, RI and MPSV of intra-ovarian arteries in 10 normally cycling women with transvaginal Doppler ultrasound using a 6 MHz transducer. The intraclass correlation coefficients of temporal variability for uterine artery PI and MPSV were 0.99 and 0.88 respectively. In contrast, correlation coefficients for intra-ovarian PI, RI and MPSV were 0.780.86, 0.760.89 and 0.630.68. They concluded that intra-ovarian velocity measurements were too inconsistent to be reliable. An important source of error for measurements of flow volume, TAMV and PI is the instruments ability to determine average velocity. It is impossible to measure actual velocity of all Doppler signals over even a single cardiac cycle with present technology. The number of discrete velocity points or columns which each instrument actually analyses varies from approximately 60 to 540, depending on pulse frequency and software. A 60 point analysis may overestimate or underestimate velocity, compared to a 120 or 240 point analysis, depending on where the measurement points fall over the cardiac cycle. Errors due to infrequent point analyses decrease when downstream resistance is low and the difference between systole and diastole is small. The number of points analysed over a cardiac cycle varies with scroll speed and heart rate. A faster scroll speed or slower heart rate will result in more points being analysed per cardiac cycle. For example, when using the ATL Ultramark 9-HDI Doppler Ultrasound, if the

scroll speed is increased from slow (20 ms/Doppler column) to fast (5 ms/Doppler column) and the heart rate remains constant (100 bpm), then the number of columns or points analysed in one cardiac cycle would increase from 30 to 120. At a medium scroll speed (10 ms/Doppler column), 60 Doppler columns are analysed per cardiac cycle when the heart rate is 100 bpm. A decrease in heart rate from 100 to 60 bpm at a medium scroll speed would increase the number of Doppler columns or points analysed in one cardiac cycle from 60 to 100. As a consequence, three or more cardiac cycles should be analysed whenever blood flow volume, mean velocity, or PI is determined.
Relationship of flow volume and resistance indices to waveforms

The relationship of uterine artery waveforms to a modification of RI and S/D ratio was initially reported by Goswamy and Steptoe (1988). More recently, the relationship of ascending uterine artery and spiral artery waveforms to uterine artery flow volume, PI and RI was investigated by Dickey et al. (1994) (Table IV). Uterine artery flow volume was significantly lower and PI was significantly higher when waveforms were not continuous. Recumbent flow volume in single (right or left) uterine arteries averaged 40.8 ml/min when waveforms were continuous, compared to 27.0 ml/min when only end diastole flow was absent (type B, D-I) and to 19.722.0 ml/min when flow was absent in early diastolic flow (A and D, D-II and D-III). Uterine artery PI was significantly higher for all non-continuous waveforms, compared to continuous waveforms, and did not distinguish waveforms of less clinical significance with absent end diastolic flow only (type B, D-I) from waveforms of greater clinical significance (A and D, D-II and D-III). Tekay et al. (1996b) also reported on the relationship between waveform type and PI.

Table III. Interobserver variability, showing the mean difference between observers 1 and 2, the correlation coefficient (r), standard error (SE) and standard deviation (SD). Reprinted with permission from Dickey et al. (1994)
Measurement Uterine artery diameter (cm) time averaged maximum velocity flow volume (ml/min) pulsatility index resistance index Spiral artery pulsatility index resistance index 10 10 0.208 0.052 0.87 0.85 0.061 0.009 0.193 0.028 20 20 20 20 20 0.056 1.070 4.500 0.067 0.004 0.94 0.90 0.88 0.95 0.97 0.067 2.180 2.398 0.066 0.006 0.300 0.470 10.721 0.296 0.027 n Mean difference between 1 & 2 r SE of mean difference SD of mean difference

476

R.P.Dickey

Table IV. Relationship of ascending uterine artery blood flow volume per minute (UA-VOL, ml/min), pulsatility index (PI) and resistance index (RI) to uterine artery waveforms and spiral artery waveforms while subjects were recumbent and while they were standing. Values in each column for UA-VOL, UA-PI and UA-RI are mean and SE respectively. Reprinted with permission from Dickey et al. (1994)
Waveform classification* C D-I D-II D-III D-IV C D-I D-II D-III D-IV D-V (B) (A) (D) (0) (N) (B) (A) (D) (0) Recumbent Standing UA-VOL 40.8, 2.3 27.0, 4.8a 19.7, 3.3b 22.0, 1.6b 74.4, 5.8 72.8, 11.6 46.0, UA-PI 2.47, 0.05 3.47, 0.17b 3.59, 0.13b 3.36, 0.09b 2.63, 0.08 2.90, 0.11 2.79, UA-RI 0.86, 0.01 0.88, 0.01 0.92, 0.02 0.88,

n
115 9 7 17 0 59 14 0 1 0 0

n
98 0 31 17 2 59 5 1 3 0 6

UA-VOL 37.9, 2.6 19.4, 1.9b 15.2, 2.8b 3.0, 0.90b

UA-PI 2.47, 0.05 3.59, 0.13b 4.45, 0.41b 9.95, 0.05b

UA-RI 0.87, 0.01 0.87, 0.01 0.93, 0.02 0.92, 1.00, 0.015 0.94, 0.04

Uterine artery waveforms (n = 148)

Spiral artery waveforms (n = 74) 65.4, 6.1 80.4, 13.9 45.0, 22.3, 3.0b 40.0, 15.8 2.67, 0.15 2.77, 0.18 3.31, 4.30, 64.0 4.24, 1.18c

*See Table I for waveform classification. a,b,cSignificance, compared to C (continuous flow) (t-test): aP < 0.05, bP < 0.001, cP < 0.01. Table V. Relationship between individual ascending uterine artery blood flow volume per minute and uterine artery pulsatility index and resistance index, and between total uterine artery blood flow volume per minute, average uterine artery pulsatility or resistance index, and spiral artery pulsatility index and resistance index while subjects were recumbent and while they were standing for 914 min. Reprinted with permission from Dickey et al. (1994)
Recumbent Standing (914 min) Significance

r
Uterine artery blood flow volume per minute Uterine artery pulsatility index Uterine artery resistance index Spiral artery pulsatility index Spiral artery resistance index Uterine artery pulsatility index Spiral artery pulsatility index Spiral artery resistance index Uterine artery resistance index Spiral artery pulsatility index Spiral artery resistance index 0.24 0.33 0.24 0.26 0.49 0.46 0.08 0.09

r
0.46 0.66 0.28 0.21 0.31 0.30 0.54 0.54

Significance

P < 0.001 P < 0.001 NS

NS

P < 0.001 P < 0.001 P = 0.010 P = 0.042 P = 0.005 P = 0.006 P < 0.001 P < 0.001

P = 0.019 P = 0.013 P = 0.017 P = 0.002

r = Pearsons correlation coefficient; NS = not significant.

Absence of end diastolic flow in spiral artery waveforms (type V, D-I) was not associated with changes in total (right plus left) uterine artery flow volume or change in uterine artery PI or RI, compared to continuous flow (C). However, absence of early diastolic flow (A and D, D-II and III) and absence of any visible flow (N) in spiral arteries were associated with notably lower total uterine artery flow volume and higher uterine artery mean PI. Non-continuous spiral artery waveforms were not seen in any subjects, unless flow volume was notably decreased in both uterine arteries.

Relationship of flow volume to resistance indices

Relationships of uterine and spiral artery RI to uterine artery flow volume are shown in Table V (reprinted from Dickey et al., 1994). Only cases in which blood flow was continuous were analysed. Slightly less than 50% of uterine artery PI and RI was related to flow volume. This should not be surprising, since PI and RI are significantly influenced by downstream resistance. Spiral artery PI and RI were unrelated to the uterine artery blood flow volume

Doppler ultrasound study of uterine and ovarian blood flow

477

in the recumbent position and only marginally related to flow volume when standing. Spiral artery PI and RI were weakly related to uterine artery PI and RI while recumbent, but were strongly related to uterine artery RI (r = 0.54; P < 0.001) while standing. In in-vitro Doppler ultrasound studies in which water was driven through rubber tubing by a Harvard pulsatile pump, measured flow volume was closely related to Doppler flow volume (r = 0.991), lowest systolic velocity (r = 0.987), peak systolic velocity (r = 0.972), PI (r = 0.940) and RI (r = 0.940). However, this study differed significantly from clinical assessment of blood flow, because the angle of insonation was fixed and could be entered into the calculation (Miles et al., 1987).
Safety

Figure 6. Uterine artery waveform type, S/D (systolic/diastolic) ratio and resistance index (RI) in a multiparous subject during the menstrual cycle. Reprinted with permission from Goswamy and Steptoe (1988). For abbreviations, see Table I.

The effects of Doppler ultrasound on embryonic and fetal development are not known with certainty. Bioeffects of Doppler ultrasound are related to the spatial peak time average intensity (ISPTA) and duration of exposure. The American Institute of Ultrasound in Medicine (1993) has recommended that Doppler ultrasound during early pregnancy be limited to 500 s (8.3 min) at an ISPTA <94 mW/cm2. Uterine blood flow
Uterine blood flow in the normal cycle

Studies of uterine blood flow during the normal cycle are listed in Table VI. Ultrasound measurements of uterine artery and ovarian artery blood flow were initially described by Taylor et al. (1985) using Duplex ultrasound with an abdominal transducer. Goswamy and Steptoe (1988) improved on previous methods by using an offset abdominal Doppler transducer to plot uterine artery waveforms and their own modification of RI and S/D ratios in the mid-proliferative to pre-menstrual phase of the menstrual cycle. They found waveforms were continuous in most multiparous patients throughout the cycle, except for a few days immediately after ovulation and during menstruation (Figure 6), but were only continuous in nulliparous patients immediately before ovulation (Figure 7). In multiparas, RI fell with the rise in oestrogen concentrations during the early follicular phase, then rose along with a post-ovulatory fall in oestrogen concentrations. A second decrease in RI occurred along with rising oestrogen and progesterone concentrations during the mid-luteal phase of the cycle. In nulliparas, RI was markedly different, especially on days 14 and 21. This difference may have been due to the fact that waveforms were not continuous in nulliparas on most days, or due to

Figure 7. Uterine artery waveform type, S/D (systolic/diastolic) ratio and resistance index (RI) in a nulliparous subject during the menstrual cycle. Reprinted with permission from Goswamy and Steptoe (1988). For abbreviations, see Table I.

real differences in vascular development and blood flow following completion of a full-term pregnancy. Two important principles may be derived from this seminal study: (i) normal uterine blood flow values must be obtained from patients with known fertility and ability to carry to term, and (ii) normal values can only be obtained from studies in which flow is continuous throughout the cardiac cycle. Although Goswamy and Steptoe (1988) established the pattern for blood flow in normal cycles and the relationship of blood flow to hormone concentrations, their modified RI and S/D values cannot be easily translated to standard values. Also, results with transvaginal ultrasound could be different from those obtained with abdominal ultrasound.

478

R.P.Dickey

Steer et al. (1995a) found that uterine artery PI and S/D ratio values differed significantly between abdominal ultrasound and vaginal ultrasound during the normal cycle. Mean PI values during the mid-luteal phase in non-pregnant patients for a 3 MHz abdominal transducer, a 5 MHz abdominal transducer with full bladder and a 5 MHz vaginal transducer were respectively 4.08, 3.24 and 1.97. Moreover, intra-observer variability in measurement of PI decreased from a coefficient of variation of 11.3 and 7.4% respectively with a 3 MHz and 5 MHz abdominal transducer to 4.1% with a 5 MHz vaginal transducer. Zaidi et al. (1995c) studied circadian rhythm in uterine artery blood flow. They found a nadir in PI and a peak in TAMV at 0600 h, following sleep, which was unrelated to changes in oestradiol, progesterone, luteinizing hormone (LH), or follicle stimulating hormone (FSH). The increased blood flow which they found following sleep may have been the result of overnight bed rest, since a similar improvement is seen in pregnant patients who have been at bed rest continuously for several days (R.P.Dickey, unpublished data). Authors other than Goswamy and Steptoe (1988), who used abdominal ultrasound to measure uterine blood flow, found no differences in blood flow at different phases of the cycle (Thompson et al., 1988; Long et al., 1989). Scholtes
Table VI. Studies of uterine blood flow in the normal cycle
Investigation Taylor et al. (1985) (A,O) Goswamy and Steptoe (1988) (A,E) Thompson et al. (1988) (A) Long et al. (1989) (A) Scholtes et al. (1989) (A) Steer et al. (1990) (E) Kupesic & Kurjak (1993) (O,S) Sladkevicius et al. (1993) (S) Weiner et al. (1993) (E,O) Dickey et al. (1994) (S) Deichert & Buurman (1995) (IA) Steer et al. (1995a) (A/V) Strigini et al. (1995) (E,O) Tekay et al. (1996b) (E) Zaidi et al. (1995c) (C) Tan et al. (1996) (E) x x x x x x x x Wave form x x x Flow volume TAMV, TAMX

et al. (1989), using a 4.25 MHz vaginal transducer, found little variation in PI in the phases of the cycle and additionally found no difference in PI on the side of the dominant follicle or corpus luteum. However, Steer et al. (1990), using vaginal ultrasound with a 6 MHz transducer, found marked variation in uterine artery PI during the cycle in 23 patients, 20 of whom were nulliparous, which was similar to that reported by Goswamy and Steptoe (1988) for multiparous patients. Values for PI in cycle days 16, when diastolic flow was frequently absent, ranged from 2.0 to 5.6, with an average of 3.8. Values were lowest 6 days before the LH peak (average 2.6) and 9 days after the LH peak (average 1.9). The highest values occurred at the mid-cycle plasma oestradiol peak (3.7) and 3 days after the LH peak (3.9). These authors found no difference in PI on the side of the dominant follicle. Tan et al. (1996) found the PI significantly lower (3.05 versus 2.34) and TAMV higher (16.3 versus 13.2 cm) in the uterine artery on the side of the dominant follicle during the mid-luteal phase, but no significant changes in either measurement during the follicular phase of the cycle. However, the lack of significance may have been due to studying only seven patients, all with unproven fertility.

Vmax, PSV, MPSV

MDV

PI x

RI

S/D ratio

x x x

x x

x x x x x x x x x x x x x x x x x

A = abdominal ultrasound; A/V = abdominal and vaginal ultrasound; C = circadian rhythm also studied; E = relationship to oestrogen, progesterone and ovulation induction studied; IA = common, external and internal iliac arteries also measured; O = ovarian blood flow also measured; S = spiral arteries also measured; STD = standing blood flow also measured. See Table II for other abbreviations.

Doppler ultrasound study of uterine and ovarian blood flow

Table VII. Studies of uterine blood flow in in-vitro fertilization and infertility
Investigation and when measured Goswamy and Steptoe (1988) (A) IV Sterzik et al. (1989) (O) II Strohmer et al. (1991) I Kurjak et al. (1991) (INF) Steer et al. (1992) III Favre et al. (1993) III Fujino et al. (1993) (INF) Steer et al. (1994) (INF) Bassil et al. (1995) I,II,III Levi-Setti et al. (1995) I,II Tekay et al. (1995a) III Zaidi et al. (1995b) (INF) Cacciatore et al. (1996) III Tekay et al. (1996b) III Zaidi et al. (1996c) I x x x x x x x x x x x x x x x x x x x x x x Wave form x Flow volume TAMV, TAMX Vmax, PSV, MPSV MDV PI RI S/D ratio

479

A = abdominal ultrasound; O = ovarian blood flow also measured; INF = infertility. I = measured on or before day of human chorionic gonadotrophin administration; II = measured on day of oocyte retrieval; III = measured on day of embryo transfer; IV = measured in mid-luteal phase. See Table II for other abbreviations.

Deichert and Buurman (1995) examined blood flow in the common, external and internal iliac arteries every 3 days during the menstrual cycle with a 2.2 MHz Doppler abdominal transducer in nine ovulatory volunteers. The mean right and left PI of the internal iliac artery rose slightly the day before ovulation, and fell significantly the day after ovulation. No significant cyclic changes in PI and RI were found in the external and common iliac arteries. The mean right and left internal iliac artery PI was lower than those of the external iliac and common iliac arteries. The mean PI of the external iliac artery was higher than that of the common iliac artery. Continuous diastolic waveforms were found only in the internal iliac arteries, and only following ovulation. Spiral artery, as well as the uterine artery, PI and TAMV were measured throughout the cycle with vaginal ultrasound using a 5 MHz transducer by Sladkevicius et al. (1993). Continuous diastolic flow in the uterine artery was present in all examinations. Spiral artery PI was much lower than uterine artery PI (0.81.0 versus 2.43.4), but mirrored the changes in average values for the uterine arteries throughout the cycle. Values for PI were lowest for the spiral and uterine artery on days 7 and 12 post-ovulation and highest 2 days after the LH surge. Velocity was highest for the spiral artery on days 1, +7 and +12 from the LH peak, and highest in uterine arteries on day 2. These authors found marked differences between the dominant and non-dominant ovary side in mean uterine artery velocity, and to a lesser extent in PI, in the peri-ovulatory and early luteal phase. Spiral artery

and uterine artery PI and RI were measured during the peri-ovulatory period in 27 patients with spontaneous cycles and 51 patients with stimulated cycles by Kupesic and Kurjak (1993). All had male factor infertility, but parity was not stated. Results from right and left uterine arteries were averaged when they found no consistent differences between the side with the dominant follicle and the contralateral side. Diastolic flow was absent in both uterine arteries in 15% of spontaneous cycles. At their nadir the day before ovulation, the average uterine artery PI and spiral artery PI were 2.22 and 1.13 respectively. No periodical variations in PI were noted in patients stimulated with clomiphene citrate or human menopausal gonadotrophin (HMG).
Relationship to oestrogen, progesterone and ovulation induction

Several researchers have measured serum oestradiol and progesterone, as well as blood flow during the normal cycle (Goswamy and Steptoe, 1988; Steer et al., 1990; Tan et al., 1996). Goswamy and Steptoe (1988) noted that uterine blood flow increased with rising concentrations of oestradiol and progesterone and decreased when oestradiol fell immediately following ovulation. In marked contrast, Steer et al. (1990) found that maximum uterine artery PI values, indicating highest impedance to blood flow, occurred on the day of serum oestradiol peak, then fell by 25% to a mid-cycle nadir 2 days following the urine LH peak, rose again the next day for 1 day only before falling

480

R.P.Dickey

again to an even lower mid-luteal phase nadir 10 days after the LH peak, while oestradiol concentrations were still high and progesterone concentrations were high but declining. Tan et al. (1996) found that the TAMV of the dominant uterine artery was correlated with the rise in progesterone concentrations in the luteal phase (r = 0.52) and with oestradiol concentrations throughout the cycle (r = 0.27), but found no relationship between PI and oestradiol or progesterone at any time in the cycle. The relationship of uterine and ovarian artery PI to number of follicles >15 mm diameter and to serum oestradiol and progesterone concentrations was studied in 11 patients receiving HMG by Weiner et al. (1993). They found a consistent decrease in PI from the early follicular to early luteal phase of the cycle which was negatively correlated with serum oestradiol concentrations (r = 0.51), but not with progesterone concentrations (r = 0.25) or follicle number. However, Kupesic and Kurjak (1993) found no cyclic change in uterine artery PI in patients who had received clomiphene citrate or HMG. Tekay et al. (1995b) found no difference in mean uterine artery PI and MPSV between patients with ovarian hyperstimulation syndrome (OHSS) and asymptomatic IVF patients when first measured 813 days after oocyte retrieval, but found significantly lower PI at gestational week 9 in OHSS patients (1.78 0.22), compared to pregnant controls (2.43 0.46). The difference may have been due to an 80% multiparity rate in the OHSS group. Strigini et al. (1995) measured uterine artery and intra-ovarian PI in spontaneous and FSH stimulated cycles and also following administration of 100 mg progesterone. Uterine artery PI was significantly lower in FSH cycles than in spontaneous cycles on the day of oestradiol peak and 5 days later. Progesterone significantly decreased uterine artery PI further in spontaneous cycles and FSH cycles, but did not affect intra-ovarian PI. Thus, the relationship between serum hormone concentrations and uterine blood flow appears to differ, depending on which authors are consulted and which blood vessels were sampled. Uterine artery flow volume was not measured in any of the studies thus far reported. Spiral arteries were rarely sampled. It may be that factors produced in the ovary, other than oestradiol and progesterone, are responsible for apparent cyclic changes in uterine vessels. This is discussed below, with regard to pregnancy.
Relationship to IVF outcome

Studies of uterine blood flow in IVF and infertility are listed in Table VII. The relationship between Doppler ultrasound and IVF outcome has been reviewed by Tekay et al. (1996a). A relationship between failure of

implantation and inadequate uterine blood flow was established by Goswamy et al. (1988). A total of 153 patients with three IVF failures, despite transfer of good quality embryos, was examined with a 3 MHz offset Doppler abdominal ultrasound in the mid-luteal phase of spontaneous cycles. All patients had received oestradiol valerate, 2 mg daily, continuously from the fifth cycle day during their last IVF cycle. Abnormal uterine artery waveforms, in which diastolic flow was absent or not continuous with preceding systolic flow (type 0, A, or D, Table I), were present in 43% of the patients with repeated implantation failure. Patients with abnormal waveforms were treated with oral oestradiol valerate, 2 mg, and norgestrel, 0.5 mg, for 21 days from the fifth cycle day for three cycles; 82% of these patients with abnormal waveforms responded to this treatment with improvement in waveforms to continuous diastolic flow or absent end diastolic flow (type C and B). Patients with three previous transfer failures who had normal waveforms proceeded directly to another IVF attempt. All patients with initially normal waveforms who chose to attempt another IVF cycle received 2 mg oestradiol valerate from cycle day 5. The pregnancy rate for patients with initially normal waveforms who repeated IVF was 31%. The pregnancy rate for patients with initially abnormal waveforms which improved after therapy was 43%. Patients with initially abnormal waveforms who did not improve after therapy had embryos cryopreserved and later received oestradiol valerate in incremental doses, up to 6 mg daily, before embryo thaw and transfer plus i.m. progesterone, 50 mg daily before and 100 mg daily after transfer, with a resultant 50% pregnancy rate. This study established a role for Doppler ultrasound in assessment of patients undergoing IVF and for two methods of treatment: hormonal therapy before IVF, and cryopreservation with subsequent hormone therapy and transfer. It did not determine the parameters of blood flow compatible with fertility in normal cycles, since blood flow may have been increased due to the high oestradiol and progesterone concentrations which occur during ovulation induction (Robertson et al., 1971; Ghosh et al., 1994; Dickey and Hower 1995c). Furthermore, findings with abdominal ultrasound needed to be repeated with transvaginal ultrasound utilizing at least a 5 MHz transducer, which could notably improve measurements of waveforms, velocity, and RI (Steer et al., 1995a). Strohmer et al. (1991) investigated uterine artery PI in 323 IVF patients on the day of HCG administration using vaginal Doppler ultrasound with a variable transducer (2.254.45 MHz), but examined only one uterine artery and were unable to measure continuous flow in many

Doppler ultrasound study of uterine and ovarian blood flow

481

patients. They nevertheless found significant differences between PI in 46 pregnant (2.55 0.78 SD) and 277 non-pregnant (2.84 1.29 SD) cycles. They declined to suggest a cut-off value above which HCG should not be administered, nor IVF attempted. From their SD, that value would be >3.33. Zaidi et al. (1996c) measured uterine artery PI the day of HCG injection, using vaginal ultrasound with a 5 MHz transducer, in 139 IVF cycles. They found no difference in pregnancy rate or implantation rate for PI values of 1.02.0, compared to 2.02.9 or 3.0. They also found no significant difference in PI for pregnant cycles (2.52 0.50), compared to non-pregnant cycles (2.64 0.80). Furthermore, they found no significant correlation between serum oestradiol concentrations and PI. Sterzik et al. (1989) measured uterine artery and ovarian artery RI by vaginal ultrasound with a 4.5 MHz transducer on the day of follicle aspiration in 45 IVF patients. They used only the uterine and ovarian arteries with the lowest RI in their analysis. These authors found a difference between uterine artery RI in 12 patients who conceived (0.78 0.06 SE) and 43 who did not conceive (0.84 0.05 SE); however, there was considerable overlap. Early diastolic flow (type A, D, or 0) was absent in seven patients, none of whom conceived. Steer et al. (1992) measured mean uterine artery PI (right and left) on the day of embryo transfer, using vaginal ultrasound with a 4 MHz transducer, in 82 IVF patients. No pregnancies occurred when the average PI was 3.0. There were no differences in pregnancy rate, implantation rate, or multiple pregnancy rate between patients with PI <2.0 and 2.03.0. Favre et al. (1993) measured mean right and left uterine artery PI in 198 IVF patients on the day of embryo transfer using vaginal ultrasound with a 4.5 MHz transducer. In patients treated with gonadotrophin-releasing hormone (GnRH), the PI was significantly lower (2.87) than in non-GnRH-treated patients. The PI was not significantly different in conception and non-conception cycles when GnRH and non-GnRH patients were analysed separately. No ongoing pregnancies occurred when the PI was >3.55 on the day of embryo transfer. Cacciatore et al. (1996) measured uterine artery PI and RI on the day of embryo transfer in 200 patients using a vaginal ultrasound with a 5.06.5 MHz transducer. Mean right and left PI and RI were significantly lower in pregnant than in non-pregnant patients, but there was considerable overlap (2.45 0.54 SD, 0.85 0.04 versus 2.66 0.39, 0.87 0.04 respectively). There was no difference between PI and RI values in pregnancies which went to term (2.42 0.95, 0.85 0.05 respectively) and those which aborted (2.45 0.40, 0.85 0.05 respectively). The pregnancy rate dropped from 41% when the PI was <3.0 to

15% when the PI was 3.0 and to 10% when the PI was 3.5. The pregnancy rate was 39% when the RI was <0.91, compared to 13% when the RI was 0.92. The best cut-off values were 3.3 for PI and 0.95 for RI. Absent diastolic velocity was noted in only six patients, one of whom conceived but aborted. Tekay et al. (1996b) measured flow velocity waveform type, PI, MPSV and MDV with a 6 MHz vaginal transducer on the day of embryo transfer in 25 patients during a GnRHHMG cycle, and in 32 patients having frozen embryo transfer in spontaneous cycles. There were no conceptions when both uterine arteries had flow velocity waveforms with absent end diastolic flow (type B and D or D-II and D-III). The mean PI was significantly lower, and the mean MDV was significantly higher, but mean MPSV was not different in GnRH-HMG cycles, compared to spontaneous cycles. There were no differences in mean PI, MPSV, or MDV between conception and non-conception cycles among patients with continuous diastolic flow. However, individual PI values were always <4.0 in conception cycles. Bassil et al. (1995) measured mean (right and left) uterine artery RI with a 6.5 MHz vaginal transducer from the beginning of HMG administration to the day of embryo transfer in 152 patients (196 cycles) after GnRH. Only 102 cycles in which three good quality embryos were transferred and in which endometrial thickness was 9 mm were analysed. A significant increase in RI was seen during 7 days of stimulation, which stabilized for 3 days before decreasing markedly from day 9 until day 13, when HCG was given. After HCG administration, the RI increased significantly until the day of oocyte retrieval, then decreased slightly until embryo transfer. In cycles which resulted in pregnancy, RI values were initially lower on cycle day 2 before starting HMG (0.775 0.003 versus 0.790 0.003), started to decline earlier (cycle day 8 versus cycle day 10), and were significantly lower on the day of HCG (0.773 0.003 versus 0.790 0.003), compared to non-pregnant cycles, but were no different on the day of oocyte retrieval or embryo transfer. Bassil et al. (1995) concluded that an RI value >0.79 before starting HMG was associated with a poor uterine response, and suggested that an increased dose of HMG might improve blood flow and uterine receptability. Levi-Setti et al. (1995) also measured mean (right and left) uterine artery PI with a 6.5 MHz vaginal transducer, from 5 days before HCG administration until oocyte retrieval, in 87 IVF patients, 17 of whom had multiple cycles of treatment. Three or more embryos were transferred in all cycles. The number of patients with continuous waveforms was not stated. During the first IVF cycle, the PI was significantly lower from 4 days before HCG through retrieval in patients

482

R.P.Dickey

who conceived. First cycle PI values were also lower in patients with multiple cycles who conceived after the first cycle. The greatest difference between pregnant and non-pregnant cycles was observed on the day after HCG administration. A PI 3.0 on the day of HCG administration was associated with a 42% pregnancy rate, compared to 24% if the PI was >3.0. Thus, a consensus of published work indicates that implantation in IVF cycles is decreased when uterine artery PI is 3.33.5 and when the RI is 0.95, or when waveforms show absence of early or end diastolic flow at the time of HCG administration, oocyte retrieval, or embryo transfer. Studies of uterine blood flow relationship to IVF outcome performed thus far strongly suggest that the uterine blood flow can help to define which patients will become pregnant. More longitudinal studies like those of Bassil et al. (1995) and Levi-Setti et al. (1995) are needed. An important element missing from all previous studies is that they were performed only while patients were recumbent. Standing upright may markedly change blood flow in some patients (see the section on postural changes). Also, the effect of blood flow on the ovary and on oocyte development must be considered (see the section on the ovary).
Relationship to infertility

Following the pioneer work of Goswamy et al. (1988), Kurjak et al. (1991), Fujino et al. (1993) and Steer et al. (1994) investigated uterine blood flow in women with a history of infertility who were not undergoing IVF cycle stimulation. Kurjak et al. (1991) measured waveforms and RI by vaginal ultrasound with a 6 MHz transducer in 100 infertile women. They noted infertility in 11 of 12 women who lacked end diastolic flow during both the proliferative and luteal phases of the cycle, but no difference in mean right and left uterine artery RI between infertile and fertile women. Fujino et al. (1993) measured uterine artery PI by vaginal ultrasound with a 5.0 MHz transducer throughout the cycle in 60 infertile women. The mean right and left PI was low in pregnant, compared to non-pregnant cycles, during the follicular phase (1.67 0.22 SD versus 2.30 0.78 SD) and mid-luteal phase (2.33 0.69 versus 2.50 0.97), but not during the ovulatory phase (1.89 0.41 versus 2.5 0.51). Fujino et al. (1993) did not suggest a cut-off point. They also did not mention whether waveforms were continuous during diastole. Steer et al. (1994) measured mid-luteal phase mean right and left uterine artery PI by vaginal ultrasound with a 4 MHz transducer in 161 patients with various infertility diagnoses

and in 23 presumably fertile women with azoospermic husbands. Supposedly normal patients had an average PI of 1.9 (range 0.82.7). Among patients with various causes of infertility, the values and ranges were: unexplained (n = 35) 2.45 (range 1.07.0), tubal damage (n = 92) 2.65 (1.38.0), endometriosis (n = 8) 2.32 (2.15.1) and anovulation (n = 22) 3.03 (1.67.0). The PI was higher than the normal range in 50% of patients with endometriosis and anovulation, in 35% with tubal damage and in 23% with unexplained infertility. Significant numbers of patients with infertility had absence or reverse of flow during diastole. Right and left uterine artery PI were similar, except in cases where a Fallopian tube had been removed. There was no correlation between PI and age, number of years of infertility, serum oestradiol, or serum progesterone. The authors concluded that decreased uterine blood flow may be an important factor in upwards of 23% of infertility patients. Right and left uterine artery PSV and RI, as well as perifollicular PSI and PI, were measured throughout the cycle, with multiple measurements on the day of expected ovulation in one patient with luteinized unruptured follicle syndrome (Zaidi et al., 1995b). No significant difference was found between uterine arteries at any time during the cycle. Many questions remain to be answered about the role of uterine blood flow in infertility, not the least of which is how much blood flow is enough. None of the studies of infertility and very few of IVF cycles have measured velocity, and none have measured flow volume, even though these might be as important as impedance to flow as represented by PI and RI. Moreover, according to Goswamy and Steptoe (1988), the only parameter measured in these three studies, PI, is unreliable when flow is absent at any part of diastole, and waveforms must be reported instead. Finally, the important effect of standing on blood flow in infertility remains to be addressed (Dickey et al., 1994).

Relationship to endometrial thickness and pattern

Studies of the relationship of uterine blood flow to endometrial thickness are listed in Table VIII. The relative importance of uterine blood flow and endometrial thickness or pattern as predictors of implantation was investigated by Coulam et al. (1994) and Serafini et al. (1994) on the day of HCG injection, by Cacciatore et al. (1996) and Tekay et al. (1996b) on the day of embryo transfer in IVF cycles, and by Steer et al. (1995b) in frozen embryo transfer cycles. Steer et al. (1994) investigated the relationship of uterine blood flow to endometrial thickness in non-IVF infertility patients.

Doppler ultrasound study of uterine and ovarian blood flow

483

Table VIII. Studies of the relationship of uterine blood flow to endometrial thickness and pattern. See Table II for abbreviations
Wave form Coulam et al. (1994) Serafini et al. (1994) Steer et al. (1994) Steer et al. (1995b) Cacciatore et al. (1996) Tekay et al. (1996b) x x x x x x x x x Flow volume TAMV, TAMX Vmax, PSV, MPSV MDV PI x x RI S/D ratio

Table IX. Studies of ovarian blood flow in normal cycles

Investigation Taylor et al. (1985) (A) Scholtes et al. (1989) Bourne et al. (1991) Collins et al. (1991) Kurjak et al. (1991) Jurkovic et al. (1991) Campbell et al. (1993) Kupesic and Kurjak (1993) (E) Sladkevicius et al. (1993) Strigini et al. (1995) (E) Tekay et al. (1995) (E,H) Bourne et al. (1996) Tan et al. (1996) Zaidi et al. (1996b) (C) x x x x x x x x x x x x x x x x x x x Wave form x Flow volume TAMV, TAMX Vmax, PSV, MPSV MDV PI x x x x x RI S/D ratio

A = abdominal ultrasound; C = circadian rhythm also studied; E = relationship to oestrogen, progesterone, or ovulation induction also studied; H = hyperstimulation syndrome also studied. See Table II for other abbreviations.

In the first study (Coulam et al., 1994), mean right and left uterine artery PI and endometrial thickness and pattern were measured by transvaginal ultrasound with a 5 MHz transducer on the day of HCG administration in 41 IVF cycles, on the day of progesterone administration in 20 donor oocyte cycles, and the day of ovulation in 24 natural cycles with transfer of cryopreserved embryos. The PI was significantly lower in conception cycles (2.79) than in non-conception cycles (3.50), for natural cycles with transfer of cryopreserved embryos, but no significant differences were found in IVF or oocyte donor cycles. The 95th centile of PI in conception cycles was >3.3, similar to that reported by Strohmer et al. (1991). There was no difference in average endometrial thickness in conception and non-conception cycles; however, no pregnancies occurred when the thickness was <6 mm. Lack of a triple-layered endometrial pattern was significantly associated with failure of conception in donor oocyte cycles, but not in other cycles. Unfortunately, the correlation of PI with endometrial thickness and pattern was not analysed.

In the second study, both discontinuous flow during diastolic and triple pattern endometrium were significantly associated with failure to achieve a term pregnancy, but uterine artery RI and endometrial thickness were not (Serafini et al., 1994). Non-continuous diastolic flow (types A, B, D, and 0) was present on the day of HCG injection in 12% of patients who delivered term pregnancies, in 27% who had clinical abortions and in 34% who did not become pregnant. Endometrial pattern was triple in 63% with term pregnancy, 27% with clinical abortion and 24% who failed to become pregnant. However, Serafini et al. (1994) measured blood flow in only one uterine artery and included absent end diastolic flow (type B) among the abnormal waveforms (types A, D, 0). Steer et al. (1994) found a significant correlation between mean right and left uterine artery PI measured in the mid-luteal phase and endometrial thickness in normal patients (r = 0.85), those with unexplained infertility (r = 0.84), endometriosis (r = 0.90) and anovulation (r = 0.77) and also a lesser correlation with tubal infertility (r = 0.37).

484

R.P.Dickey

Steer et al. (1995b) subsequently reported that uterine artery PI was significantly lower on the day of starting progesterone before frozen embryo transfer in 19 patients who became pregnant, compared to 57 who did not conceive; however, there was significant overlap: 2.85 (range 1.43.6) versus 4.15 (range 2.16.8). There was a significant correlation between PI and a 24-kDa protein (r = 0.58), oestradiol receptor (r = 0.71) and endometrial histological dating (r = 0.43), but not endometrial thickness, on the same day in a trial cycle consisting of GnRH and oestrogen preceding the cycle of transfer. Cacciatore et al. (1996) found a significant correlation between mean right and left uterine artery PI and both oestradiol (r = 0.37) and progesterone (r = 0.32), but not endometrial thickness on the day of embryo transfer. None of these authors speculated about methods of treatment to increase uterine blood flow.

Ovarian blood flow Studies of ovarian blood flow in normal cycles are listed in Table IX. Ovarian blood flow has been studied extensively in recent years, in the hope that knowledge of normal and abnormal blood flow would provide an explanation for some cases of infertility and for failure of ovarian response to gonadotrophic stimulation. The majority of studies have come from a single laboratory: Kings College, London (Bourne et al., 1991, 1996; Collins et al., 1991; Jurkovic et al., 1991; Campbell et al., 1993; Zaidi et al., 1995a,b, 1996a,b; Tan et al., 1996). Taylor et al. (1985), in a classic study, measured ovarian artery blood flow by Doppler ultrasound, using an abdominal transducer, and by an invasive technique employing a 10 MHz directional continuous wave flow meter applied directly to the ovarian, iliac, and uterine arteries. They demonstrated that ovarian and uterine arteries, waveforms and PI were identical for both techniques, and that the PI was lower on the side with the dominant follicle. They also performed longitudinal studies during the menstrual cycle which demonstrated that increased blood flow on the side of the dominant follicle occurred by day 5, before the dominant follicle could be distinguished by ultrasound, and continued into early pregnancy. Continuous diastolic flow was present on the side of the dominant follicle and corpus luteum, whereas no diastolic flow could be found on the side of the quiescent ovary. They proposed that Doppler ultrasound could be used to study luteal phase defect. Subsequently, Scholtes et al. (1989), using vaginal ultrasound with a 4.25 MHz transducer, were able to detect continuous diastolic flow in 74% of ovarian arteries. They confirmed a significantly lower PI in the ovarian artery on

the side of the ovary bearing the dominant follicle, from cycle day 13 preceding ovulation through day 21, but found no difference on cycle day 7. Sladkevicius et al. (1993) observed similar changes in the dominant follicles peak velocity and PI, with less changes in the ovarian hilum and stroma during the preovulatory period. Increased blood flow on the side of the dominant follicle was also demonstrated by decreased RI (Kurjak et al., 1991), decreased PI (Weiner et al., 1993) and increased PSV (Tan et al., 1996). Bourne et al. (1996) noted that PSV of corpus luteum vessels rose following ovulation and was correlated with serum progesterone (r = 0.77). In studies of corpus luteum blood flow during the first trimester of pregnancy, the peak velocity and TAMV did not change, and the PI was decreased in some reports (Valentin et al., 1996) but not in others (Jurkovic et al., 1992). Bourne et al. (1991) first measured intra-ovarian blood flow changes during the hours preceding and immediately following ovulation in a single patient. Collins et al. (1991) and Campbell et al. (1993) subsequently measured intrafollicular flow in the peri-ovulatory period in larger numbers of normal women. They observed a marked increase in PSV in the presence of a relatively constant PI, beginning 12 h after the LH surge or 29 h before follicular rupture, with a peak velocity immediately following ovulation. Unlike uterine artery blood flow (Zaidi et al., 1995c), no significant circadian rhythm was found in ovarian follicular or stromal PI or PSV (Zaidi et al., 1996b). Tekay and Jouppila (1996) found that, while intra-observer reproducibility of transvaginal Doppler measurement of intra-ovarian artery PI and RI was satisfactory, repeated measurements of PSV showed too much variability to be reliable.
Relationship to oestrogen, progesterone and ovulation induction

In patients stimulated with HMG, the PI of the ovarian vessels supplying the dominant follicle and corpus luteum was positively correlated with the number of follicles 15 mm and with serum oestradiol concentrations, but not with serum progesterone in one study whose authors suggested that oestradiol increased stromal blood flow, but not corpus luteum blood flow (Weiner et al., 1993). However, in another study, no difference was seen in ovarian PI or RI in HMG stimulated cycles, compared to spontaneous cycles, but a relationship was seen between follicular response and stromal PSV in HMG-stimulated patients (Kupesic and Kurjak, 1993). Strigini et al. (1995) measured intraovarian and uterine artery PI on the day of oestradiol peak and 5 and 10 days later in spontaneous cycles and in

Doppler ultrasound study of uterine and ovarian blood flow

485

FSH-stimulated cycles. Intra-ovarian PI was significantly lower in FSH-stimulated cycles than in spontaneous cycles on the day of oestradiol peak, but did not show any significant change after ovulation. Moreover, during the luteal phase, the PI of intra-ovarian arteries was not significantly different in FSH-treated cycles than in spontaneous cycles, whereas uterine artery PI was lower in FSH cycles. Administration of 100 mg progesterone i.m. during the luteal phase significantly suppressed uterine artery PI further, but did not affect intra-ovarian artery PI. A strong negative correlation (r = 0.80) was noted between serum progesterone and intra-ovarian vessel RI during the mid-luteal phase of the cycle by Glock and Brumsted (1995). Tekay et al. (1995b) could find no difference in intra-ovarian arterial PI or MPSV between OHSS patients and asymptomatic IVF patients 813 days after oocyte retrieval.

Relationship to IVF outcome

Studies of ovarian blood flow in IVF and infertility are listed in Table X. Barber et al. (1988), using transvaginal ultrasound to measure blood flow to the corpus luteum after embryo transfer, found that no pregnancies occurred when the RI was >0.5 on day 3 after embryo transfer. Weiner et al. (1993) reported that failure of the stromal PI to fall during HMG stimulation was associated with less than three follicles at retrieval. In the same study, the ovarian artery PI remained low in the late luteal phase in patients who became pregnant. The relationship of ovarian stromal blood flow on the second or third cycle day to follicular development in response to HMG was studied in 105 IVF patients by Zaidi et al. (1996a). They defined poor response as fewer than six follicles aspirated at retrieval. PSV was, but PI was not, related to subsequent follicular response, suggesting to Zaidi et al. (1996a) that peak velocity could be used to screen for poor responders prior to initiating HMG. The mean peak velocity prior to HMG was 10.2 cm/s in the good response groups and 5.2 cm/s in poor responders. Only 10 patients met the criteria for poor response, and two of these were cancelled because of detection of only a single follicle. The pregnancy rate in the eight who went to retrieval, despite poor response, was no different than in the 95 normal responders. Oyesanya et al. (1996) noted that recovery of oocytes at IVF retrieval was related to the presence of a detectable velocity waveform, but not to the size of follicles. However, they did not indicate the size of the follicles that they measured. Their recovery rate was low; only 17% of patients had oocytes recovered from 75% of follicles, and the correlation

between the number of follicles with pulsatility and number of oocytes recovered was modest (r = 0.47). Nargund et al. (1996) attempted to establish a relationship between Doppler ultrasound indices of follicular quality and oocyte or embryo quality. They measured PI and PSV in 126 follicles from 27 patients immediately before aspiration. The absence of follicular blood flow was highly correlated with failure to retrieve an oocyte. PSV was significantly higher in follicles associated with a good-quality preimplantation embryo, compared with follicles which yielded oocytes that did not fertilize or produced inferior embryos (19.7 10.8 versus 9.9 5.3 cm/s). When PSV was 10 cm/s, there was a 70% chance of producing a good-quality embryo, compared to a 14% chance if PSV was 10 cm/s. There was no corresponding relationship to PI. There were only six clinical pregnancies, so that they were unable to associate blood flow with whether or not pregnancy ended in abortion. Nargund et al. suggested that use of Doppler ultrasound might help to determine which embryos to transfer. Thus, investigations of ovarian and follicular blood flow have so far failed to realize the hope that they would yield information that could be used to predict or alter the outcome of IVF. Because oxygen availability may affect meiotic division during the peri-ovulatory period, blood flow studies may yet reveal significant information about the quality or even the genetic makeup of human embryos.
Ovarian blood flow in polycystic ovarian syndrome and luteal insufficiency

PSV, TAMV and PI of ovarian stroma were measured on cycle day 2 or 3 in 25 patients with polycystic ovarian syndrome (PCOS) (Zaidi et al., 1995a). Mean PSV and TAMV were significantly higher in patients with PCO without elevated LH or anovulation (16.9 and 10.5 respectively) and PCO with elevated LH or anovulation (16.9 and 10.9), than in patients with normal ovaries (8.7 and 5.4). There was no difference in PI. Zaidi et al. (1995a) speculated that the increased stromal blood flow may explain the excessive response seen in PCO patients during HMG stimulation. Aleem and Predanic (1996) studied 40 PCO patients throughout the cycle. Blood flow in the stroma was visualized in 88% of PCO versus 50% of normal ovaries from cycle day 3 to 8 and in all ovaries after ovulation when a corpus luteum was present. Significantly lower PI and RI were seen in PCO patients on days 38, compared to normal ovaries. Values for PI and RI in normal ovaries on days 1522 equalled those seen in PCO patients on days 38. PSV was not significantly different between PCO and normal patients on days 38, but was markedly higher in normal patients on days 1522. Vascular impedance was not influenced by ovarian volume or the number of follicles. No

486

R.P.Dickey

correlation was found between serum oestradiol, progesterone, or testosterone and RI, but a positive correlation was present between LH and PSV. Additionally, there was a trend towards lower PI and RI with higher LH. Aleem and Predanic (1996) speculated that LH acts to increase blood flow by stimulating prostaglandin synthesis. Intra-ovarian RI, PSV and TAMV were measured by vaginal ultrasound with a 5 MHz transducer five times during the cycle in patients with presumptive luteal phase defect by Glock and Brumsted (1995). Women with luteal phase defect by histological dating were found to have higher RI and lower TAMV during the late proliferative to mid-luteal phase, but no difference in PSV, compared to women with in-phase endometrium. Perifollicular blood flow, as well as uterine artery blood flow, was measured throughout the cycle, with multiple measurements near the time of expected ovulation, in one patient with a luteinized unruptured follicle (Zaidi et al., 1995b). PSV was low before and after the LH surge, compared to the dominant follicle in normal cycles, suggesting to Zaidi et al. (1995b) that the primary defect was deficient follicular development during the proliferative phase. Uterine blood flow in the first 16 weeks of normal pregnancy
Development of placental circulation

Doppler ultrasound has greatly enhanced understanding of uteroplacental circulation. The ascending uterine arteries give rise to approximately eight arcuate arteries that run in parallel to the surface of the uterus and form anastomoses

with the contralateral ascending uterine arteries. The arcuate arteries, in turn, give rise to centripedal radial arteries, which penetrate the middle third of the myometrium and, in turn, give rise to ~200 spiral arteries. After implantation, trophoblastic cells spread out from the ends of anchoring villi penetrating the lumina of the intradecidual portion of the spiral arteries, replacing the musculoelastic architecture, thereby converting the spiral arteries into low resistance uteroplacental arteries. Recent anatomical and ultrasonographic studies suggest that infiltration of the decidua down to the superficial myometrium by columns of extravillous trophoblastic cells and the transformation of spiral arteries is not completed until mid-gestation in normal pregnancies. Before the 10th week of gestation, the trophoblastic shell is continuous and forms plugs inside the lumens of spiral arteries, allowing only filtered plasma to penetrate the placenta. During the 10th week, these plugs begin to break up, so that limited circulation of maternal blood around villi begins. This process is not completed, and full maternal circulation within the whole intervillous chamber is not established until the 12th week in normal pregnancy (Hustin and Schaaps, 1987; Hustin et al., 1988, 1995; Jauniaux et al., 1991a,b, 1992b; Jaffe and Woods, 1993). Recently, Jauniaux (1996) reported that limited areas of intervillous blood flow can sometimes be detected before the 10th week in normal pregnancy. Merce et al. (1996), however, using Doppler ultrasound, found intervillous blood flow from 6 weeks onward, in accord with classical embryological studies which reported the establishment of intervillous blood flow around day 14 or 15 of embryonic life (Wilken, 1958).

Table X. Studies of ovarian blood flow in in-vitro fertilization and infertility

Investigation Barber et al. (1988) Kurjak et al. (1991) Weiner et al. (1993) (E) Glock & Brumsted (1995) (INF,LI,N,E) Zaidi et al. (1995a) (INF,PCO) Zaidi et al. (1995b) (INF,LI) Aleem & Predanic (1996) (E,N,PCO) Nargund et al. (1996) Oyesanya et al. (1996) Zaidi et al. (1996a) x x x x x x x x x x x x x x x x x x x x x Wave form Flow volume TAMV, TAMX Vmax, PSV, MPSV MDV PI RI x x S/D ratio

E = relationship to oestrogen, progesterone and ovulation induction also studied; INF = infertility LI = luteal insufficiency; N = normal cycle also studied; PCO = polycystic ovaries. See Table II for other abbreviations.

Doppler ultrasound study of uterine and ovarian blood flow

Table XI. Studies of uterine blood flow in normal early pregnancy
Investigation Deutinger et al. (1988) Stabile et al. (1990) (S) Thaler et al. (1990) Jaffe and Warsof (1991) (S) Jauniaux et al. (1991b) Jurkovic et al. (1991) (S) Jauniaux et al. (1992a) (E,O,S) Jauniaux et al. (1992b) Jaffe and Woods (1993) (S) Kurjak et al. (1993 (S) Jauniaux et al. (1994b) (E) Van Zalen-Sprock et al. (1994) (S) Dickey and Hower (1995a) (S) Dickey and Hower (1995b) (C,E,S) Dickey and Hower (1995c) (E,S) Tekay et al. (1996b) Coppens et al. (1996) (S) Dickey and Hower (1996) Guanes et al. (1996) Merce et al. (1996) (S) Valentin et al. (1996) (O,S) x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x Wave form Flow volume TAMV, TAMX Vmax, PSV, MPSV MDV PI x x RI S/D ratio x

487

C = relationship to crownrump length and chorionic sac diameter also studied; E = relationship to oestrogen, progesterone and ovulation induction also studied; O = ovarian blood flow also measured; S = spiral artery also measured. See Table II for other abbreviations.

In complicated early pregnancies, trophoblastic invasion of both the decidua and the lumen in spiral arteries is reduced, the trophoblastic shell is thinner, and circulation within the placenta may be seen earlier than in normal pregnancy (Khong et al., 1987; Hustin et al., 1988). The presence of numerous echo-poor areas with flow in the first trimester may be an indication of subsequent pregnancy complications, according to Jaffe et al. (1995) and Jauniaux (1996). Independent evidence in support of the onset of intervillous circulation at between 10 and 12 weeks is the notable decrease in uterine artery PI and RI at this time (Jauniaux et al., 1992b; Jaffe and Woods, 1993; Kurjak et al., 1993; Dickey and Hower, 1995a; Guanes et al., 1996), and the corresponding increase in uterine artery velocity and volume at the same time (Dickey and Hower, 1995a). Additionally, the diastolic notch in spiral arteries disappears between at between 10 and 13 weeks (Coppens et al., 1996). The development of the fetal placental circulation is discordant with changes from the maternal side. At the end of the sixth menstrual week, the villous vasculature is connected with the embryonic heart and umbilical circulation begins (Benirschke and Kauffmann, 1990). The umbilical circulation may be detected by Doppler from 7 weeks of gestation as systolic flow (Den Ouden et al., 1990;

Jauniaux et al., 1991a,b). End diastolic flow is usually absent from umbilical arteries until the 12th week of gestation and does not become detectable in all pregnancies until 15 weeks (Den Ouden et al., 1990; Arduini and Rizzo, 1991). The appearance of end diastolic flow is accompanied by a sharp decrease in umbilical artery PI.
Uterine blood flow during early normal pregnancy

Studies of uterine blood flow in normal early pregnancy are listed in Table XI. Normal embryonic and fetal growth and development depend on adequate perfusion of the intervillous space through the maternal spiral arterioles (Clapp et al., 1982; Campbell et al., 1983). Although the first 16 gestational (post-menstrual) weeks are the most critical period in human development, uterine artery blood flow volume before week 16 has been measured in only two studies (Thaler et al., 1990; Dickey and Hower, 1995a). Thaler et al. (1990) reported that blood flow volume and vessel diameter increased at a constant rate between the mid-luteal phase of the cycle and 3639 weeks. However, they measured only the left uterine artery, performed measurements only twice during the first 16 weeks, at 68 weeks and at 1416 weeks, and used midluteal phase measurements from non-pregnant patients as a

488

R.P.Dickey

Figure 8. Uterine artery; total (right plus left) flow volume (A), mean cross-sectional area (B), mean time averaged maximum velocity (C), heart rate (D), mean resistance index (E), and spiral artery resistance index (F). Mean (SE) filled symbols, smoothed curve open symbols. Reprinted with permission from Dickey and Hower (1995a).

starting point. A linear increase in uterine artery MPSV between weeks 4 and 5 and weeks 13 through 16 was reported by Jauniaux et al. (1992a) and Tekay et al. (1995a). However, Jauniaux et al. (1992b) reported elsewhere that peak uterine artery velocity doubled between weeks 13 and 14. Valentin et al. (1996) found that TAMV and PSV, measured from the fifth to the 11th

gestational week, increased exponentially in both the dominant and non-dominant uterine artery, and in the subchorionic (spiral) arteries, in contrast to PI, which decreased linearly in both vessels. Dickey and Hower (1995a) measured uterine blood flow with transvaginal Doppler ultrasound at 13 week intervals in 44 patients with primary or secondary infertility during

Doppler ultrasound study of uterine and ovarian blood flow

489

the first 16 weeks of pregnancy. They found that total (right plus left) uterine artery flow volume increased very gradually, an average of 16 ml/min per week from 77 ml/min at post-menstrual week 5 to 159 ml/min at post-menstrual week 10, and then more rapidly to reach 665 ml/min at week 16 (Figure 8A). Uterine artery cross-sectional area increased linearly through the first 16 weeks (Figure 8B), while TAMV remained level or increased slowly until week 8, then rapidly until week 16, with a tendency to plateau from week 12 to 15 (Figure 8C). Studies of indices of resistance begun before the eighth week are rare because of the infrequency with which patients are referred for scanning in early pregnancy. Linear decreases in uterine artery PI and RI in early pregnancy between weeks 46 and weeks 1632 have been reported by several authors (Deutinger et al., 1988; Jauniaux et al., 1991a,b; Jurkovic et al., 1991; Van Zalen-Sprock et al., 1994; Coppens et al., 1996; Valentin et al., 1996), but others (Jauniaux et al., 1992b; Jaffe and Woods, 1993; Kurjak et al., 1993; Guanes et al., 1996) have reported no change in PI and RI until week 812 (the onset of intervillous circulation), with a rapid decrease thereafter. Dickey and Hower (1995a) found that uterine artery RI (Figure 8E) and PI (not shown) were the reciprocal of velocity, and decreased slowly from week 5 to week 10, then rapidly from week 10 to week 16. A linear decrease in spiral artery indices of resistance between post-menstrual weeks 46 and weeks 1516 has been consistently reported by all authors [Stabile et al., 1990; Jurkovic et al., 1991; Jaffe and Warsof, 1991; Jauniaux et al., 1992b; Jaffe and Woods, 1993; Kurjak et al., 1993; VanZalen-Sprock et al., 1994; Dickey and Hower, 1995a (Figure 8F); Coppens et al., 1996; Merce et al., 1996; Valentin et al., 1996]. Analysis of covariance performed before and after week 10, the putative onset of intravillous flow, by Dickey and Hower (1995a), revealed that cross-sectional area of the uterine artery was the most important factor in determining uterine artery flow volume before week 10, while TAMV was the most important factor after week 10 (Table XII). Uterine artery PI and RI were related more closely to velocity than to flow volume for all weeks. Spiral artery PI and RI were positively related to flow volume, that is spiral artery impedance decreased and flow increased when uterine artery flow volume decreased before week 10, but not from week 10 to 16. This may be due to the ability of myometrial vessels, but not placental vessels, to respond to decreased uterine artery volume by vasodilatation (Kauppila et al., 1980). The total flow volume from both uterine arteries, measured during the fifth to sixth gestational weeks by

Dickey and Hower (1995a) (Figure 8) was equal to that reported by Thaler et al. (1990) for a single artery of non-pregnant patients during the mid-luteal phase. By the 16th week, however, total flow volume in Figure 8 was twice that reported by Thaler et al. (1990) for a single artery during the 31st to the 33rd weeks, but was similar to the 500 ml/min reported by Maini et al. (1985) for the same weeks of gestation using an invasive radioisotope method. The most obvious explanation for the marked decrease in uterine artery RI and PI and consequent increase in velocity, which began at 8 weeks, is the onset of intervillous circulation. The linear decrease in spiral artery RI and PI from week 4 or 5 is consistent with the finding of Jaffe and Woods (1993) that trophoblastic invasion of the spiral artery musculo-elastic architecture begins soon after implantation. According to Duvekot et al. (1993), a decrease in systemic vascular tone, lowering both afterload and preload, begins early in pregnancy between post-menstrual weeks 5 and 8 (Duvekot et al., 1993; Duvekot and Peeters, 1994). Cardiac output increases in response to an increase in heart rate in early pregnancy. Later, increases in the stroke volume are more important as ventricular filling status normalizes. According to Duvekot and Peeters (1994), 60% of all of the systemic and cardiac changes have occurred by mid-pregnancy. In the study by Dickey and Hower (1995a), however, uterine artery flow volume showed no relationship to maternal heart rate and, after the 10th week, 80% of the difference in uterine artery flow volume between patients could be explained by velocity and vessel size alone. In a review of recent developments in understanding uteroplacental circulation in early pregnancy, Jauniaux (1996) stated, Longitudinal evaluation of placental blood flow is the only possible approach for future study of in vivo early pregnancy. He could have added that studies should be started in the fourth week of gestation, should be carried out weekly, should include either flow volume or TAMV in addition to indices of resistance, and should include only pregnancies delivered of normal babies at term. A longitudinal study of uterine artery and spiral artery blood flow in patients with normal outcome is underway in the our clinic, with results reported as centiles. Preliminary results indicate that normal flow volumes for the 10th and 50th centiles are approximately 30 ml/min and 100 ml/min respectively from the mid-luteal phase of the cycle through the fifth post-menstrual week, rise slowly until the 10th or 12th post-menstrual week, and then increase rapidly thereafter until the end of study at week 16.

490

R.P.Dickey
Table XII. Analysis by covariance of the relative effects of uterine blood flow variables to uterine artery blood flow volume, with gestational age, pulse rate, other uterine blood flow variables, and subject as covariants. Data are expressed as F-values. Reprinted with permission from Dickey et al. (1995a)
Flow volume Week 59 Week 1016 Significance <0.001 <0.001 NS 0.038 NS NS

F
Velocity Cross-sectional area Uterine artery RI Spiral artery RI Gestational age Pulse rate 9.54 11.89 1.35 2.00 0.80 0.90

F
9.77 5.58 1.32 1.27 2.77 1.16

Significance <0.001 <0.001 NS NS NS NS

RI = resistance index; NS = not significant.

Table XIII. Analysis of covariance of the relative effects of serum oestradiol or progesterone on uterine blood flow variables in early pregnancy with gestational age, subject, and oestrogen or progesterone as covariants. Data are expressed as F-values. Signs are derived from multiple linear regression. Reprinted with permission from Dickey and Hower (1996)
Uterine artery Flow volume Weeks 516 (n = 118) Oestradiol Progesterone Weeks 59 (n = 48) Oestradiol Progesterone Weeks 1016 (n = 72) Oestradiol Progesterone
a,bSignificance: aP

Spiral artery Average velocity 0.87 2.05a 1.71 1.30 0.60 1.41 Average diameter 1.94a 1.24 1.03 1.10 1.89a 0.99 Pulsatility index 1.88a 0.96 1.53 0.55 1.23 1.23 Resistance index 2.49b 1.22 1.07 0.44 1.72 0.96 Pulsatility index 1.24 0.77 0.40 0.73 1.24 0.49 Resistance index 1.12 0.89 0.30 0.92 1.37 0.65

1.70a 1.89a 0.36 2.23a 1.40 1.05

< 0.05, bP < 0.01.

Effect of oestrogen and progesterone

Oestradiol administered to postmenopausal women causes decreased resistance in the systemic circulation (Magness and Rosenfeld, 1989; Magness et al., 1993; Penotti et al., 1993), and causes dilation of the uterine arteries in sheep (Anderson et al., 1977) and women (de Ziegler et al., 1991; Hillard et al., 1992). Progesterone opposes the effect of oestradiol in some studies (Hillard et al., 1992; Marsh et al., 1994), but not in others (de Ziegler et al., 1991). These observations suggest that rising oestrogen and progesterone concentrations might be the cause of increased uterine blood flow in early pregnancy. However, such findings may not be transferable to pregnancy, where the uterine vasculature may already be responding maximally to oestrogen and progesterone as a prerequisite for implantation (Sterzik et al., 1989; Steer et al., 1992). Indeed, Anderson et al. (1977), using chronically

implanted electromagnetic flow probes around the left middle uterine artery, determined that oestradiol secretion was not responsible for the increase in uterine blood flow in ovine pregnancies. Oestrogen-induced vasodilatation of uterine and systemic blood vessels is now believed to be mediated through nitric oxide (Kharitonov et al., 1994; Ramsey et al., 1995; Cicinelli et al., 1996) and is antagonized by nitric oxide synthesis inhibitors (Van Buren et al., 1992). Two Doppler ultrasound studies appear to support the premise that oestradiol and progesterone are responsible for increased uterine blood flow during the first 16 weeks of pregnancy, because both hormones were found to be negatively related to uterine artery RI (Jauniaux et al., 1992a, 1994a). However, in the same studies, no association was found between oestradiol or progesterone and uterine artery PI or peak velocity, although both are

Doppler ultrasound study of uterine and ovarian blood flow

491

ordinarily closely associated with RI. Also, uterine artery blood flow volume, a more direct indication of uterine perfusion, was not measured. Multiple linear regression analysis, which was performed for the entire period of weeks 5 through 16 in these studies, may have missed important changes in uterine blood flow that occur after the beginning of intervillous circulation during the 10th week of gestation. Dickey and Hower (1996) analysed the relationship between uterine blood flow and serum oestradiol and progesterone, first for weeks 5 through 16 inclusive, and then separately before and after 10 weeks. They used analysis of covariance to adjust for gestational age, effect of sex steroids not in the primary analysis, and repeat analysis of the same subjects (Bland and Altman, 1995). When studies with non-continuous uterine artery waveforms were excluded from analysis, both PI and RI were negatively related to oestradiol for weeks 516 (Table XIII). Uterine artery flow volume and diameter were also negatively related to oestradiol for weeks 516, despite the decrease in downstream resistance represented by the negative relationships of oestradiol to PI and RI. Progesterone concentrations were positively related to both total flow volume and average velocity for weeks 516. When ultrasound examinations with non-continuous waveforms were included in the analysis, uterine artery RI was negatively related to oestradiol, but uterine artery PI was unrelated to oestradiol for weeks 516 (data not shown), as previously reported by Jauniaux et al. (1994b). When relationships were reanalysed separately before and after week 10, oestradiol concentrations were negatively related to diameter, but not to flow volume, PI, or RI for weeks 1016. Progesterone was positively related to flow volume during weeks 59 and was unrelated to volume or velocity during weeks 1016. Thus, during the first 10 weeks of pregnancy, the net effect of progesterone was to increase uterine blood flow, while the effect of oestradiol was nil, a decrease in uterine artery diameter being cancelled by a decrease in distal vessel impedance. After the 10th week, the effect of progesterone was negligible, while oestradiol had an adverse effect on uterine artery diameter and flow volume. The paradoxical finding that oestradiol is associated with both decreased uterine artery blood flow volume and decreased downstream resistance can be explained if oestradiol affects uterine blood flow by two different mechanisms, one acting on the uterine arteries and the other acting on smaller vessels. Vasodilatation mediated through nitric oxide is now a well-known effect of oestrogen (Kharitonov et al., 1994; Ramsey et al., 1995; Cicinelli et al., 1996). The negative association between oestradiol and uterine artery diameter found in spontaneous cycles could be

explained by the finding that oestrogens can potentiate -adrenergic vasoconstriction in response to prostaglandins (Colucci et al., 1982; Miller and Vanhoutte, 1990). On the other hand, apparent relationships between oestradiol or progesterone and increased uterine blood flow during early pregnancy may be due to nothing more than colinearity. Duvekot et al. (1993), in a study of cardiovascular, renal and endocrine changes throughout pregnancy, concluded that cardiovascular changes in pregnancy were not the result of increased oestradiol, progesterone, or plasma renin concentrations. Anderson et al. (1977) also concluded that oestrogen was not responsible for the increase in uterine blood flow volume in ovine pregnancy. The premise that oestradiol and progesterone have a direct effect on uterine blood flow during pregnancy is supported by the findings that receptors for these hormones are present in the muscle cells of non-pregnant uterine arteries (Perrot-Applanat et al., 1988), that oestradiol may have a direct, smooth muscle relaxing effect, mediated by a calcium-channelled blockage mechanism (Jiang et al., 1991), and that oestradiol may potentiate vasodilatation, via endothelial-derived relaxin factor and endothelialindependent vasodilators (Gilligan et al., 1994; Kharitonov et al., 1994; Ramsey et al., 1995; Cicinelli et al., 1996). Irrespective of the relationship to uterine blood flow, oestrogen apparently has no role in the maintenance of early pregnancy, since oestradiol replacement without progesterone replacement is unable to prevent abortion in patients lutectomized in early pregnancy (Csapo et al., 1973) or in ovariectomized primates (Ghosh et al., 1994). The relationship between uterine blood flow, hormone concentrations and embryo development during the first 16 weeks of pregnancy is difficult to assess and to analyse, due to the rapid changes which occur, problems of colinearity, and the onset of intervillous circulation. Studies on uterine artery blood flow and steroid measurements summarized in Table VI were performed in patients with a history of infertility or previous pregnancy loss and, therefore, results of similar studies in an entirely normal population may be different.
Effect of ovulation induction drugs

Ovulation induction with HMG and clomiphene citrate results in marked increases in oestradiol and progesterone which continue into early pregnancy (Robertson et al., 1971; Dickey et al., 1991, 1992a, 1993c; Ghosh et al., 1994). Uterine blood flow was previously measured during ovulation induction with HMG (Weiner et al., 1993) and at the time of embryo transfer (Sterzik et al., 1989; Steer et al., 1992), but such studies were not continued into pregnancy.

492

R.P.Dickey

Figure 9. Mean (SE) serum oestradiol-17 (A) and progesterone (B) after treatment with human menopausal gonadotrophin (HMG, ) or clomiphene citrate (CC, s), or spontaneous ovulation (). Reprinted with permission from Dickey and Hower (1995c).

Dickey and Hower (1995c) studied the effects of HMG or clomiphene citrate on uterine artery and spiral artery blood flow and their relationship to oestradiol and progesterone concentrations in patients who conceived singleton pregnancies after treatment for primary or secondary infertility. During the first 8 post-menstrual weeks, serum oestradiol and progesterone concentrations were increased by 300% in patients who had received HMG and 60100% in patients who had received clomiphene citrate, compared to spontaneous ovulation (Figure 9A and B). Despite these considerable changes, uterine artery blood flow volume increased by only 33% following HMG and by 20% following clomiphene citrate, compared to spontaneous ovulation (Figure 10A). Uterine artery diameter was increased by 15% for both HMG and clomiphene citrate groups (Figure 10B). Time averaged maximum velocity increased by 37% in patients who received HMG (Figure 10C). Uterine artery RI decreased by 35% only in patients who received HMG (Figure 10D). When analysed by multiple linear regression analysis and analysis of covariance, oestradiol concentrations were negatively related to uterine artery diameter, but only for spontaneously pregnant patients, and were unrelated to flow volume and TAMV or RI for any group. Progesterone concentrations were positively related to TAMV in spontaneous pregnancy, but not in clomiphene citrate- or HMG-induced pregnancy. These findings strongly suggest that there is no direct relationship between either oestradiol or progesterone and uterine artery blood flow during early pregnancy, and that the relationships observed by ourselves (Dickey and Hower, 1996) and others (Jauniaux et al., 1992a, 1994a) were the result of colinearity or due to other factors.

Relationship to chorionic sac and crownrump length

Dickey and Hower (1995b) examined the relationship between recumbent uterine blood flow and average chorionic sac diameter (CSD) and crownrump length (CRL) during the first 16 gestational weeks in normal singleton pregnancies to determine whether differences in blood flow could explain some of the variability that occurs in normal human embryos in early pregnancy (Nishimura et al., 1968; Rossavik et al., 1988; Dickey and Gasser, 1993; Guirgus et al., 1993). During gestational weeks 5 through 12, CSD increased 12-fold and CRL increased 13-fold, while mean total (right plus left) uterine artery blood flow increased only ~3-fold. After adjustment for the effect of gestational age and oestradiol, by multiple regression analysis, CSD was negatively related to spiral artery PI and RI, but not to uterine artery blood flow volume PI or RI. CRL was positively related to uterine artery blood flow volume and negatively related to uterine artery PI and RI. The findings of a relationship between uterine artery resistance indices and CRL suggest that there may also be an association between uterine blood flow and embryonic growth and well-being during the first trimester. In late pregnancy, increased uterine artery resistance and impaired uterine blood flow are associated with intrauterine growth retardation (Fleischer et al., 1986; Jacobson et al., 1990; Kay et al., 1991; North et al., 1994). Thus, a lag in CSD or CRL found using grey scale ultrasound in the first trimester may be an indication that additional studies of uterine blood flow need to be performed and, if abnormal, that efforts should be directed towards improving blood flow.

Doppler ultrasound study of uterine and ovarian blood flow

493

Figure 10. Mean (SE) uterine artery total (right plus left) flow volume (A), mean diameter (B), mean time averaged maximum velocity (C) and mean resistance index (D), following treatment with human menopausal gonadotrophin (HMG, ) or clomiphene citrate (CC, s), or spontaneous ovulation (). Reprinted with permission from Dickey and Hower (1995c).

Uterine blood flow in abnormal pregnancy Increased impedance of uterine and umbilical artery blood flow, presenting as elevated RI and PI ratios and persistence of the early diastolic notch into the third trimester is associated with intrauterine growth retardation (IUGR), premature delivery, maternal hypertension and in severe cases, fetal death (Gerretsen et al., 1981; Trudinger et al., 1985; Fleischer et al., 1986; Jacobson et al., 1990; Harrington et al., 1991; Meekins et al., 1994; Murakoshi et al., 1996). The first 12 weeks of gestation are the period of most rapid embryonic development (ORahilly and Muller, 1987). Impairment of subplacental blood flow at this time by vasoactive drugs has been shown to result in birth defects (Danielsson et al., 1989) and may be the mechanism by which diabetes results in birth defects associated with a lag in embryo growth (Pedersen and Moldsted-Pedersen, 1981). Jaffe et al. (1995) noted that 50% of complicated pregnancies demonstrated abnormal first trimester Doppler characteristics. They later followed up 32 patients who had spiral artery RI above 2 SD and blood flow in the intervillous

space before 12 weeks. They found that when the intraplacental RI was greater than the umbilical RI between 22 and 25 weeks, 80% of pregnancies had an abnormal outcome (Jaffe and Woods, 1996). Chianchianco et al. (1991) found an association with abnormal uteroplacental blood flow, pregnancy related hypertension and a history of recurrent abortion. However, most uterine blood flow studies have failed to show significant differences in resistance indices between pregnancies with normal outcomes and pregnancies which subsequently ended in abortion (Alfirevic and Kurjak, 1990; Arduini et al., 1991; Jaffe and Warsof, 1992). Jauniaux et al. (1994b) found that mean uterine artery PI was elevated in missed abortions, but that mean uterine artery RI and spiral artery PI and RI were within normal limits. Pathological examination of the missed abortion specimens in their study showed a fragmented or absent trophoblastic shell in 53%, reduced or absent trophoblastic infiltration in 43%, and absent or reduced physiological changes in spiral arteries in 63% of cases. Other histological studies of spontaneous abortion

494

R.P.Dickey

have also demonstrated defective transformation of the spiral arteries and a reduced trophoblastic infiltration of the decidua (Khong et al., 1987; Hustin et al., 1990). None of the studies of the relationship of uterine blood flow to abortion have reported the karyotype of the aborti. As a result, given the high degree of variability in individual values in these studies, it is impossible to know whether abnormal blood flow was responsible for abortion of a karyotypically normal embryo, was the result of inadequate trophoblastic invasion of a karyotypically abnormal embryo, or was the end result of embryo demise irrespective of whether the karyotype was normal or not. In an ongoing study in our clinic, in which recumbent and standing uterine artery flow volume and uterine and spiral artery PI and RI are recorded at the time of the first positive quantitative pregnancy test and at each subsequent visit, patients are encouraged to undergo a dilatation and curettage procedure to obtain tissue for karyotype diagnosis whenever a diagnosis of inevitable abortion is made by ultrasound. Recumbent blood flow volume prior to loss of viability was below the 50th centile in 8/10 subsequent pregnancy losses with normal karyotype (46XX = 5, 46XY = 5), in 5/5 biochemical pregnancies, and in 1/3 with an abnormal karyotype. A 50% decrease in blood flow volume during standing occurred in 5/10 losses with normal karyotype (46XY = 4, 46XX = 1) and 1/3 with an abnormal karyotype. Uterine RI was 0.80 in 6/10 pregnancy losses with normal karyotypes, 4/5 biochemical pregnancies, and 1/3 with abnormal karyotypes. Uterine artery PI was 2.50 in 3/10 with normal karyotypes (all 46XY), in 3/5 biochemical pregnancies, and in 0/3 with abnormal karyotypes. Spiral artery RI was 0.80 in 7/10 with normal karyotypes, 5/5 biochemical pregnancies, and 2/3 with abnormal karyotypes. In preliminary studies of uterine blood flow in the small sac syndrome, where the difference between mean gestational sac diameter and embryo crownrump length is <5 mm before gestational day 65 (Bromely et al., 1988; Nazari et al., 1991) and most embryos are karyotypically normal (Dickey et al., 1992c), small gestational sac size was associated with higher than normal spiral artery PI and RI (R.P.Dickey, unpublished observations). Administration of oral micronized progesterone or i.m. progestin was followed by significant decreases in spiral artery PI and RI and by significant increases in gestational sac diameter (R.P.Dickey, unpublished observations). New areas of investigation
Postural studies

A consistent limitation of Doppler ultrasound studies of uterine blood flow is that measurements have been obtained

only in the recumbent position. Blood flow measurements while standing may be equally or more important physiologically, and may differ significantly from recumbent measurements. Standing causes a change in position and/or downward movement of the uterus, which may result in acute angulation or stretching of the uterine arteries, especially in patients with poor uterine support. As a consequence, arterial diameter may be decreased, and if uncompensated for by increased pulse rate or decreased downstream resistance, can result in a decrease in blood flow volume. Dickey et al. (1994) measured blood flow by Doppler ultrasound during the mid-luteal phase in 74 non-pregnant patients, first while recumbent and then twice while standing: once immediately, and again after 914 min of standing, without removal of the vaginal probe. Blood pressure and pulse rate were recorded during each measurement period (Table XIV). Total right plus left uterine artery blood flow volume fell progressively while standing in 70% of patients by an average 11% after 38 min and by an average 34% after 914 min (Figure 11). Decreases in blood flow volume while standing were accompanied by significant decreases in mean uterine artery diameter and significant increases in mean uterine artery PI, with no change in uterine artery RI nor in spiral artery PI and RI. In subjects with unchanged or increased blood flow volume while standing, there was no change in mean diameter, but mean uterine artery PI and RI were decreased. Pulse rate and mean blood pressure were increased equally in patients both with decreased and with increased flow volume. Decreases in blood flow volume at 38 min and at 914 min of standing could be entirely accounted for by decreases in cross-sectional area, which averaged 22% at 38 min and 34% at 914 min. Increased flow volume while standing was due to both an increase in pulse rate, which resulted in increased end diastolic flow, and to a decrease in downstream vessel impedance. Waveforms became worse in 16% of individual uterine arteries (right or left) while standing (Table IV). Uterine artery waveforms with absent end diastolic flow only while recumbent (B or D-I) became continuous in seven of nine cases while standing, due to increased pulse rate, and degraded to absence of early diastolic flow (A or D, D-II or D-III) in two cases (22%). Twelve percent of patients demonstrated either absent early diastolic flow (A or D, D-II or D-III) or no visible flow (N) for the first time while standing. Poor uterine support was associated with increased uterine artery PI (r = 0.33) while standing, but not while recumbent. A history of previous spontaneous abortion was associated with decreased uterine blood flow and increased

Doppler ultrasound study of uterine and ovarian blood flow

495

Figure 11. Effect of standing (percentage change), mean (SE) on (A) ascending uterine artery [flow volume (UA-VOL, unbroken line), pulsatility index (dashed line), resistance index (dotted line)] and (B) spiral artery [pulsatility index (dashed line), resistance index (dotted line)] in all subjects (a), subjects with increased UA-VOL, (b), and subjects with decreased UA-VOL (c). Reprinted with permission from Dickey et al. (1994).

uterine artery PI or RI while recumbent, but not while standing. Spiral artery PI initially rose in all subjects upon standing and then decreased at 914 min in subjects with increased uterine artery flow volume, but remained elevated in subjects with decreased flow volume (Figure 11). This suggested that these two groups respond differently to the challenge of changing from a recumbent to a standing position. Spiral artery flow which had indicated absence of end diastolic flow (B or D-I) while recumbent converted to continuous flow in 13 of 14 cases and degraded to A or D-II in one case while standing (Table IV). Spiral artery PI and RI were related to uterine artery blood flow volume while standing, but not while recumbent, and were more strongly related to

uterine artery PI and RI while standing than while recumbent (Table V). Possible explanations for the initial rise in spiral artery PI in all subjects, followed by a fall only in those with unchanged or increased uterine artery flow volume, include individual differences in endothelium-dependent relaxing factor (Meyer et al., 1993), prostaglandins (Greiss, 1972), or adrenergic innervation (Ford et al., 1984). Alternatively, a fall in uterine artery blood flow may be the normal response to standing, and the increase in blood flow may be aberrant. This possibility is suggested by the finding that average recumbent uterine blood flow volume was initially lower in subjects who experienced a rise in flow volume while standing (Table XIV).

496

R.P.Dickey

Table XIV. Effect of standing on ascending uterine artery (UA) and spiral artery (SA) blood flow volume per minute (VOL), diameter (DIA), pulsatility index (PI) and resistance index (RI). Values are mean SE. Modified and reprinted with permission from Dickey et al. (1994)
Recumbent Value UA-VOL (ml) UA-DIA (mm) UA-PI UA-RI SA-PI SA-RI Pulse rate Mean BP 61.6 6.6 Standing (38 min) Value 80. 10.9 % Change 126.2 6.6b 97.6 2.0 89.8 3.4b 96.3 1.6a 102.5 6.7 96.2 2.8 124.2 4.6c 104.2 2.4 78.5 4.3c 94.1 1.5c 103.5 2.9 101.5 1.1 108.2 4.8 99.6 2.4 122.3 2.7c 102.7 1.4 Standing (914 min) Value 90.0 12.8 % Change 142.8 9.1c 100.0 2.5 86.6 2.5c 94.8 1.4c 96.4 5.8 97.4 4.3 126.4 6.0c 105.2 1.6+ 65.8 2.8c 92.8 1.5c 115.2 6.7a 102.0 1.2 109.0 5.2 98.4 2.4c 125.0 3.6c 105.3 1.6b

Subjects with increased or unchange UA-VOL (n = 22) 2.78 0.06 2.76 0.14 0.90 0.02 2.10 0.10 0.84 0.02 72.4 89.4 2.4 1.9 6.6 2.70 0.05 2.50 0.11 0.86 0.02 2.07 0.11 0.80 0.02 89.4 92.8 4.2 2.1 4.5 2.76 0.05 2.40 0.13 0.85 0.02 0.95 0.08 0.81 0.02 91.1 94.3 5.0 1.8 4.4

Subjects with decreased UA-VOL (n = 52) UA-VOL (ml) UA-DIA (mm) UA-PI UA-RI SA-PI SA-RI Pulse rate Mean BP 78.9 57.6 50.5 2.88 0.05 2.65 0.08 0.89 0.01 1.95 0.08 0.83 0.02 75.4 91.4 1.5 1.0 2.69 0.05 2.73 0.11 0.90 0.01 2.05 0.09 0.82 0.02 91.1 93.5 2.2 1.3 2.65 0.04 3.08 0.22 0.90 0.06 2.06 0.09 0.81 94.0 96.1 0.02 2.6 1.6

BP = blood pressure. a,b,cPaired t-test, standing versus recumbent, 38 min versus 914 min: aP < 0.05, bP < 0.01, cP < 0.001.

Studies of the effect of standing in pregnant patients are now underway in our clinic. Pregnant patients are allowed to stand for no more than 56 min while repeating blood flow measurements. Early results indicate that 70% of pregnant patients experience a decrease in blood flow volume while standing during the fourth and fifth gestational weeks, the same proportion as for non-pregnant patients. Approximately 4% of pregnant patients experience a fall in uterine blood flow of 50% while standing during the first 10 weeks of gestation. Both the percentage of patients who experience a decrease in blood flow while standing and the extent of the decrease appear to decline as pregnancy progresses. This may be an important adaptation to the need for increased blood flow and oxygenation. The reasons for this adaptation may be entirely mechanical, due to enlargement of the uterus and its vessels and to changes in its position within the pelvis, or may be due to maturation or adaptation of humeral or neurological factors or a combination of these. Vasodilation has been shown to occur in response to reduced uterine perfusion during the second and third trimesters of normal pregnancy, presumably due to release of endogenous nitrous oxide in the vascular epithelium (Cameron et al., 1993; Ramsey et al., 1994). Improvement in the response to decreased uterine artery

blood flow volume caused by postural changes may be a necessary physiological adaptation, because the placental vessels are unable to increase perfusion by vasodilation (Kauppila et al., 1980). Although previous ultrasound studies of systemic (Duvekot et al., 1993) and uterine (Thaler et al., 1990; Jurkovic et al., 1991; Jauniaux et al., 1992b; Jaffe and Woods, 1993) blood flow in early pregnancy have failed to consider the effect of posture, assessment of the effect of posture on cardiac and uterine blood flow is not new. Lindhard (1913) reported a decrease in cardiac output in men upon changing from the recumbent to the standing position. In 1956, Morris et al., using the 24NA clearance method, found increased uterine blood flow during bed rest, compared with exercise during pregnancy. Suonio et al. (1976) noted a 23% decrease in placental blood flow in late pregnancy when moving from the left lateral recumbent to a standing position. Dickey et al. (1966) reported an 80% decrease in urinary oestriol excretion when changing from a recumbent to a standing position during the third trimester, which they attributed to a change in placental perfusion. Schiff et al. (1991) studied the use of Doppler ultrasound of uterine arteries to increase the sensitivity of the rollover test of Gant et al. (1974) to predict patients at increased risk of

Doppler ultrasound study of uterine and ovarian blood flow

497

pregnancy-induced hypertension. Indeed, this new evidence of the effect of standing on uterine blood flow may explain why bed rest has been traditionally considered to be beneficial in preventing recurrent or threatened abortion (Hertig and Livingstone, 1944). Standing studies are not without difficulty. Transvaginal scanning of both ascending uterine arteries and of myometrial spiral arteries while the subjects are standing is not possible in very obese subjects, in subjects with very large ovaries due to hyperstimulation, and when the uterus is greatly enlarged by fibroids. Some subjects became faint while standing, especially in early pregnancy. Measurement of flow volume requires more time than measurement of uterine and spiral artery indices, so that acute changes may be missed. More precise assessment of time-related uterine blood flow changes after standing, with instrumentation which allows instantaneous and measurement of minute blood flow volume, PI and RI, is needed. However, the findings of standing studies to date indicate that, when possible, measurements of uterine blood flow should be performed both with patients standing and recumbent.
Non-steroidal drugs

At the present time, not enough is known about the effect of drugs, other than oestrogen and progesterone, on human uterine and ovarian blood flow in infertility and early pregnancy. This situation should change because of the ease with which blood flow may be measured with vaginal Doppler ultrasound under physiological conditions. Investigation of drug effects on uterine blood flow in infertility and early pregnancy can be ranked in order of significance as (i) studies performed during late pregnancy, (ii) studies of non-pregnant reproductive age women, (iii) inferences from animal studies during pregnancy, (iv) inferences from human investigations of the effects of drugs on the systemic circulation and (v) anecdotal reports. The largest number of reports concern late pregnancy, and are concerned with three types of drugs: nitrous oxide, aspirin and vasodilators. It is now known that arterial circulation is continuously dilated by nitric oxide released from arterial and arteriole endothelium (Vallance and Collier, 1994). The role of nitric oxide in late pregnancy, its relationship to pre-eclampsia, and its role as a widespread biological mediator have been reviewed recently (Vallance and Collier, 1994; Morris et al., 1996a). Endothelial-derived relaxing factor, discovered by Furchgott and Zawadski (1980), was proven to be nitric oxide by two groups acting independently (Ignarro et al., 1987; Palmer et al., 1987). Nitric oxide has been shown to mediate oestrogen-induced

vasodilation (Van Buren et al., 1992). Endothelial-derived nitric oxide may be a key regulator of placental blood flow during pregnancy (Moncada et al., 1991; Chang et al., 1992; Myatt et al., 1992; Hull et al., 1994). Circulating concentrations of nitrite, an oxidative metabolite of nitric oxide, are reduced in patients with pre-eclampsia (Seligman et al., 1994). In non-pregnant women of reproductive age, circulating concentrations of nitric oxide metabolites are higher in the proliferative phase of the menstrual cycle than in the luteal phase and they peak at mid cycle (Cicinelli et al., 1996). In patients undergoing IVF, nitrite and nitrate (NO3/NO2) concentrations in follicles 4 mm diameter are related to ovarian artery PI (Anteby et al., 1996). In animals, nitric oxide synthesis activity in maternal tissue begins to increase early in pregnancy (Weiner et al., 1994). Prolonged blockage of nitric oxide synthesis produces a pre-eclampsia-like syndrome in rats (Molnar et al., 1994). Inhibition of nitric oxide synthesis results in growth retardation and limb defects in rats (Diket et al., 1994). Glycerol trinitrate (GTN) is a nitric oxide donor which preferentially dilates veins and other vessels that have low basal output of nitric oxide. Ramsey et al. (1994) have shown that administration of intravenous GTN to pregnant patients with abnormally elevated uterine artery RI (> 0.6) and bilateral notching resulted in lowering of RI and PI with no change in TAMV and no associated changes in heart rate, systemic blood pressure, or resistance indices in the umbilical or maternal carotid artery. During GTN infusion, 50% of uterine artery waveforms with discontinuous early diastole flow (A, D or D-II, D-III) were converted to continuous flow. Changes due to GTN were greater than those caused by prostacyclin in the same study. The use of organic nitrates as active vasodilators in clinical medicine dates back to 1847. Therapeutically administered organic nitrates became the drugs of choice when nitrates were indicated. Tolerance is one of the main problems of nitrate use. After 1214 h of continuous administration, almost complete tolerance occurs. The drug must be discontinued for 12 h before the vasodilator effect is again observed within the coronary arteries. Tolerance to the vasodilator effect in the uterine arteries is not as yet known. Low-dose aspirin (ASA) began to be investigated as a means of preventing pre-eclampsia in the mid 1980s because of its antiplatelet activity and because it decreased the synthesis of thromboxane A2 (Beaufils et al., 1985; Benigni et al., 1989; Uzan et al., 1994). The effect of ASA on uterine blood flow in patients with pre-eclampsia was investigated by McParland et al. (1990) and Bower et al. (1996). A large multicentre collaborative study of ASA in

498

R.P.Dickey

pregnancy (CLASP, 1994) reached the conclusion that there was no significant reduction in the incidence of pre-eclampsia, intrauterine growth retardation, stillbirth, or neonatal death. However, some centres with expertise in Doppler ultrasound diagnosis of pre-eclampsia reached the opposite conclusion. Bower et al. (1996), who participated in the CLASP study, found that ASA reduced the incidence of pre-eclampsia and severe eclampsia by ~50%. They suggested that ASA should be started earlier than 24 weeks to achieve maximum results. In a study of nulliparous patients who had abnormal waveforms, there was no overall difference in total adverse outcomes between 84 patients who were untreated and 102 who were started on 100 mg ASA daily beginning at 18 weeks. Although the incidence of pre-eclampsia was 4% in the treatment group versus 11% in the placebo group, this was offset by a higher incidence of birth weight in the <10th centile (27 versus 22%) and antepartum haemorrhage (4 versus 2%) in the treatment group (Morris et al., 1996b). In a study of hormone replacement for frozen embryo replacement, patients with impaired uterine perfusion were given ASA, 150 or 300 mg, once daily starting from day 13. The cancellation rate was lower (33 versus 48%) and the pregnancy rate was higher (25 versus 15%) in patients receiving 300 mg, compared to 150 mg per day. In a second phase, patients cancelled initially received ASA from the first day of hormonal replacement, with the result that 83% underwent replacement and 47% of these became pregnant (Wada et al., 1994). None of the studies examined the effect of ASA on the spiral arteries. In a preliminary study from our clinic, Doppler ultrasound screening was performed during the mid-luteal phase prior to IVF cycles. Patients with abnormal waveforms and elevated RI and PI were given 300 mg ASA orally. Waveforms became continuous and RI and PI were notably reduced in 80% of patients 12 h after ASA administration. Animal studies conducted with electromagnetic flow probes in the 1970s demonstrated that ovine myometrial vasculature responded to very small doses of the -adrenergic drugs epinephrine and norepinephrine by vasoconstriction, was resistant to -adrenergic blockade, and responded to -adrenergic stimulation with vasodilation (Greiss, 1972). The same studies found that ovine placental vascularization was less sensitive to -adrenergic stimulation and more sensitive to blockade than uterine vasculature. They also established that the response of myometrial vasculature did not change, even in late pregnancy, indicating that myometrial vasculative response to sympathetic neurohormone stimulation was

unaltered by the high concentrations of sex steroids that occur in pregnancy. The -adrenergic agents fenoterol and isoxsuprine are commonly used in pregnancy as uterine relaxants. Their effect on myometrial and intervillous blood flow was studied with intravenous 133Xe before the availability of Doppler ultrasound (Kauppila et al., 1978). Fenoterol, but not isoxsuprine, significantly increased myometrial blood flow. Neither drug increased intervillous blood flow. Doppler ultrasound studies of the vasodilators nifedipine (Lindow et al., 1988) and atenolol (Montane et al., 1987) have shown that these drugs do not increase uterine blood flow significantly. Vasoconstrictive drugs have been shown to cause birth defects in animals by causing vasoconstriction of uterine vessels, but vasodilating drugs may also cause birth defects (Danielsson et al., 1989). The latter are believed to act by a decrease in uteroplacental blood flow caused by excessive hypotension. In studies of other drugs, no significant changes were found in uterine or umbilical artery blood flow following maternal betamethasone administration (Cohlen et al., 1996) or following methyldopa (Montan et al., 1993). During the oxytocin challenge tests, fetal heart rate decelerations were associated with rapid and exaggerated increases in uterine and umbilical artery PI during uterine contractions, but returned to resting levels during uterine relaxation (Olofsson et al., 1996). Growth hormone and insulin-like growth factor were shown to increase ovarian vascularization by Ryan and Mackis (1987). Among anecdotal reports is that of Bonilla-Musoles et al. (1995), who reported that 100 mg of Paracetainol, an anti-inflammatory agent, caused immediate disappearance of ovarian vascular flow, and that 600 mg per day Elorgan, a xanthine-type vasodilator, caused increased uterine artery blood flow in 60% of patients.

Conclusions Doppler ultrasound measurement of uterine blood flow holds the promise of being as helpful for investigation of infertility and early pregnancy loss as Doppler analysis of uterine and umbilical blood flow has been in the second and third trimesters of pregnancy. This promise has not yet been fulfilled, in part due to the difficulty of interpreting blood flow measurements performed during the menstrual cycle and early pregnancy when diastolic blood flow is often not continuous in uterine and spiral arteries. Lack of fulfilment is also due, in part, to the fact that most studies have measured only downstream impedance, which is critical in the second and third trimesters, but may not be as

Doppler ultrasound study of uterine and ovarian blood flow

499

important in infertility and early pregnancy, and to the fact that studies have been performed only in the recumbent position. Despite these limitations, the majority of studies agree that implantation cannot occur when the mean (right and left) uterine artery PI is 3.04.0 or diastolic flow is not continuous in at least one artery on the days of HCG administration, oocyte retrieval, or embryo transfer in IVF and donor embryo cycles. Investigation of ovarian blood flow has also proven disappointing, not allowing the predetermination of which follicles will produce oocytes capable of fertilization and implantation, but it may soon help our understanding of luteal insufficiency and the cause of poor oocyte quality in polycystic ovaries. Even more intriguing is the possibility that ovarian blood flow studies may link intrafollicular hypoxia to errors of nondysjunction during meiosis. Early pregnancy, insofar as uterine blood flow is concerned, can be divided into two periods, i.e. before and after the onset of maternal intervillous circulation, which begins, according to the most recent authorities, between the 10th and 12th gestational weeks. Notable changes occur during the 10th to 12th weeks in uterine artery RI, PI, mean velocity and flow volume. The 10th week coincides with the end of embryological development, according to classical embryology studies. Uterine blood flow during the earlier period is most like that in the mid to late follicular phase of non-pregnant fertile women, while uterine blood flow in the later period is more like that of the second or third trimester of pregnancy. It is interesting to speculate that deficient uterine blood flow in the earlier period may lead to infertility and abortion, just as deficient blood flow in the later period can lead to intrauterine growth retardation in the baby and to hypertension in the mother. Additional Doppler investigations into physiological and pharmacological methods to improve uterine blood flow in infertility and early pregnancy are clearly needed. Present pharmacological methods are limited to ASA, which has its own inherent side-effects, and to nitric oxide donors, whose length of use is severely limited by rapid development of tolerance. Doppler ultrasound may establish a scientific basis for use of physiological methods, such as bed rest and progesterone, in preventing abortion. Doppler ultrasound measurement of uterine and ovarian blood flow is a relatively new technique. Its role in clinical medicine has still to be established for infertility and early pregnancy. As more investigations are performed, as clinicians become familiar with its use, and as instrumentation is improved, it may yet prove to be an

indispensable tool for clinical investigation of disorders of oocyte development, implantation and early pregnancy.

References
Aleem, F.A. and Predanic, M. (1996) Transvaginal color Doppler determination of the ovarian and uterine blood flow characteristics in polycystic ovary disease. Fertil. Steril., 65, 510516. Alfirevic, Z. and Kurjak, A. (1990) Transvaginal color Doppler ultrasound in normal and abnormal early pregnancy. J. Perinatal Med., 18, 173180. American Institute of Ultrasound in Medicine (1993) Bioeffects and Safety of Diagnostic Ultrasound. AIUM, Laurel, MD, p. 96. Anderson, S., Hacksaw, B., Still, G. and Greiss, F. (1977) Uterine blood flow and its distribution after chronic estrogen and progesterone administration. Am. J. Obstet. Gynecol., 127, 138142. Anteby, E.Y., Hurwitz, A., Korach, O. et al. (1996) Human follicular nitric oxide pathway: relationship to follicular size, estradiol concentrations, and ovarian blood flow. Hum. Reprod., 11, 19471951. Arduini, D. and Rizzo, G. (1991) Umbilical artery velocity waveforms in early pregnancy: a transvaginal color Doppler study. J. Clin. Ultrasound, 12, 335339. Arduini, D., Rizzo, G. and Romanini, C. (1991) Doppler ultrasonography in early pregnancy does not predict adverse outcome. Ultrasound Obstet. Gynecol., 1, 180185. Baker, A.R., Evans, D.H., Prytherch, D.R. et al. (1986) Some failings of the pulsatility index and damping factor. Ultrasound Med. Biol., 12, 878881. Balakier, H. and Stronell, R.D. (1994) Color Doppler assessment of folliculogenesis in in vitro fertilization patients. Fertil. Steril., 62, 12111216. Barber, R.J., McSweeney, M.B., Gill, R.W. et al. (1988) Transvaginal pulsed Doppler ultrasound assessment of blood flow to the corpus luteum in IVF patients following embryo transfer. Br. J. Obstet. Gynaecol., 95, 12261230. Bassil, S., Magritte, J.P., Roth, J. et al. (1995) Uterine vascularity during stimulation and its correlation with implantation in in vitro fertilization. Hum. Reprod., 10, 14971501. Beaufils, M., Uzan, S., Donsimoni, R. and Colau, J.C. (1985) Prevention of preeclampsia by early antiplatelet therapy. Lancet, i, 840842. Benigni, A., Gregorini, G., Frusca, T. et al. (1989) Effect of low dose aspirin on fetal and maternal generation of thromboxane by platelets in women at risk for pregnancy-induced hypertension. N. Engl. J. Med., 321, 357362. Benirschke, K. and Kauffmann, P. (1990) Pathology of the Human Placenta, 3rd edn. Springer Verlag, New York. Bewley, S., Chang, T.C. and Campbell, S. (1993) Uteroplacental resistance index: analysis of intra- and interobserver variability. Ultrasound Obstet. Gynecol., 3, 417421. Bland, J.M. and Altman, D.G. (1995) Calculating correlation coefficients with repeated observations: Part I: Correlations with subjects. Br. Med. J., 310, 446. Bonilla-Musoles, F., Ballester, M.J., Raga, R. et al. (1995) Transvaginal color Doppler sonography from folliculogensis to implantation: normal and abnormal evolution. Assist. Reprod. Rev., 5, 158169. Bourne, T.H., Jurkovic, D., Waterstone, J. et al. (1991) Intrafollicular blood flow during human ovulation. Ultrasound Obstet. Gynecol., 1, 5359. Bourne, T.H., Hagstrom, H.G., Hahlin, M. et al. (1996) Ultrasound studies of vascular and morphological changes in the human corpus luteum during the menstrual cycle. Fertil. Steril., 65, 753758. Bower, S.J., Harrington, K.F., Schuchter, K. et al. (1996) Prediction of preeclampsia by abnormal uterine Doppler ultrasound and modification by aspirin. Br. J. Obstet. Gynaecol., 103, 625629. Bromely, B., Harlow, B.L., Laboda, L.A. and Benecerraf, B.R. (1988) Small sac size in the first trimester: a predictor of poor fetal outcome. Radiology, 71, 8183. Cacciatore, B., Simberg, N., Fusaro, P. and Titinen, A. (1996) Transvaginal Doppler study of uterine artery blood flow in in vitro fertilizationembryo transfer cycles. Fertil. Steril., 66, 130134.

500

R.P.Dickey
Dickey, R.P. and Hower, J.F. (1995b) Relationship of uterine blood flow to chorionic sac and embryo growth rates. Hum. Reprod., 10, 26762679. Dickey, R.P. and Hower, J.F. (1995c) Effect of ovulation induction on uterine blood flow and oestradiol and progesterone concentrations in early pregnancy. Hum. Reprod., 10, 28752879. Dickey, R.P. and Hower, J.F. (1996) Relationship of estradiol and progesterone levels to uterine blood flow during early pregnancy. Early Preg. Biol. Med., 2, 113120. Dickey, R.P., Carter, W.T., Besch, P.K. and Vorys, N. (1966) Relationship of posture to estrogen excretion. Am. J. Obstet. Gynecol., 96, 127130. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1991) Relationship of follicle number, serum estradiol, and other factors to birth rate and multiparity in human menopausal gonadotropin-induced intrauterine insemination cycles. Fertil. Steril., 56, 8992. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1992a) Relationship of follicle number and other factors to fecundability and multiple pregnancy in clomiphene citrate induced intrauterine insemination cycles. Fertil. Steril., 57, 613619. Dickey, R.P., Olar, T.T., Curole, D.N. et al. (1992b) Endometrial pattern and thickness associated with pregnancy outcome after assisted reproduction technologies. Hum. Reprod., 7, 418421. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1992c) Relationship of small gestational sac-crown rump difference to abortion and abortus karyotypes. Obstet. Gynecol., 79, 554557. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1993a) Relationship of biochemical pregnancy to preovulatory endometrial thickness and pattern in patients undergoing ovulation induction. Hum. Reprod., 8, 327330. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1993b) Relationship of endometrial thickness and pattern to fecundity in ovulation induction cycles: effect of clomiphene citrate alone and with human menopausal gonadotropin. Fertil. Steril., 59, 756760. Dickey, R.P., Olar, T.T., Taylor, S.N. et al. (1993c) Sequential clomiphene citrate and human menopausal gonadotropin for ovulation induction: comparison to clomiphene citrate alone and human menopausal gonadotropin alone. Hum. Reprod., 8, 5659. Dickey, R.P., Hower, J.F., Matulich, E.M. and Brown, G.T. (1994) Effect of standing on non-pregnant uterine blood flow. Ultrasound Obstet. Gynecol., 4, 480487. Diket, A.L., Pierce, M.R., Munshi, U.K. et al. (1994) Nitric oxide inhibition causes intrauterine growth retardation and hind-limb disruptions in rats. Am. J. Obstet. Gynecol., 171, 12431250. Duvekot, J.J. and Peeters, L.L.H. (1994) Maternal cardiovascular hemodynamic adaptation to pregnancy. Obstet. Gynecol. Surv. (Suppl.), 49, 114. Duvekot, J.J., Cheriex, E.C., Piters, F.A.A. et al. (1993) Early pregnancy changes in hemodynamics and volume homeostasis are consecutive adjustments triggered by a primary fall in systemic vascular tone. Am. J. Obstet. Gynecol., 169, 13821392. Farin, D., Kelly, N., Ryan, G. et al. (1995) In Copel, J.F. and Reed, K.L. (eds), Doppler Ultrasound in Obstetrics and Gynecology, Chapter 20. Raven Press, New York, pp. 187197. Favre, R., Bettahar, K., Grange, G. et al. (1993) Predictive value of transvaginal uterine Doppler assessment in an in vitro fertilization program. Ultrasound Obstet. Gynecol., 3, 350353. Fleischer, A., Schulman, H., Farmakides, G. et al. (1986) Uterine artery Doppler velocimetry in pregnant women with hypertension. Am. J. Obstet. Gynecol., 154, 806813. Ford, S.P., Reynolds, L.P. and Farley, D.B. (1984) Interaction of ovarian steroids and periarterial alpha-1-adrenergic receptors in altering uterine blood flow during the estrous cycle of gilts. Am. J. Obstet. Gynecol., 150, 480484. Fujino, Y., Ito, F., Matuoka, I. et al. (1993) Pulsatility index of uterine artery in pregnancy and non-pregnant women. Hum. Reprod., 8, 11261128. Furchgott, R.F. and Zawadski, J.V. (1980) The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature, 288, 373376. Gant, N.F., Chand, S. and Worley, R.T. (1974) A clinical test useful for predicting the development of acute hypertension in pregnancy. Am. J. Obstet. Gynecol., 120, 17.

Cameron, J.T., Van Pependorp, C.L., Palmer, R.M.J. et al. (1993) Relationship between nitric oxide synthesis and increase in systolic blood pressure in women with hypertension in pregnancy. Hypertension Pregnancy, 12, 8592. Campbell, S., Griffin, D.R., Pearce, J.M. et al. (1983) New Doppler technique for assessing uteroplacental blood flow. Lancet, 1, 675677. Campbell, S., Bourne, T.H., Waterstone, J. et al. (1993) Transvaginal color blood flow imaging of the periovulatory follicle. Fertil. Steril., 60, 433438. Chang, J.K., Roman, C. and Heymann, M.A. (1992) Effect of endothelium-derived relaxing factor inhibition on the umbilical placental circulation in fetal limbs in utero. Am. J. Obstet. Gynecol., 166, 727734. Chianchiano, N., Rossi, C., Bertolino, G. et al. (1991) Uteroplacental blood flow in pregnant women with hypertension and recurrent spontaneous abortion. Acta Eur. Fertil., 22, 151152. Cicinelli, E., Ignarro, L.J., Lograno, M. et al. (1996) Circulating levels of nitric oxide in fertility women in relation to the menstrual cycle. Fertil. Steril., 66, 10361038. Clapp, J.F., McLaughlin, M.K., Larrow, R. et al. (1982) The uterine hemodynamic response to repetitive unilateral vascular embolization in the pregnant ewe. Am. J. Obstet. Gynecol., 144, 309318. CLASP (Collaborative Low Dose Aspirin Study in Pregnancy) Collaborative Group (1994) A randomized trial of low-dose aspirin for the prevention and treatment of preeclampsia among 9364 pregnant women. Lancet, 343, 619629. Cohlen, B.J., Stigter, R.H., Derks, J.B. et al. (1996) Absence of significant hemodynamic changes in the fetus following maternal betamethasone administration. Ultrasound Obstet. Gynecol., 7, 252255. Collins, W., Jurkovic, D., Bourne, T. et al. (1991) Ovarian morphology, endocrine function, and intra-follicular blood flow during the periovulatory period. Hum. Reprod., 6, 319324. Colucci, W.S., Gimbrone, M.A., McLaughlin, M.K. et al. (1982) Increased vascular catecholamine sensitivity and a-adrenergic receptor affinity in female and estrogen-treated male rats. Circ. Res., 50, 805811. Coppens, M., Loquet, P., Kollen, M. et al. (1996) Longitudinal evaluation of uteroplacental and umbilical blood flow changes in normal early pregnancy. Ultrasound Obstet. Gynecol., 7, 114121. Coulam, C.B., Bustillo, M., Soenksen, D.M. and Britten, S.T. (1994) Ultrasonographic predictors of implantation after assisted reproduction. Fertil. Steril., 62, 10041010. Csapo, A.I., Pulkkinen, M.O. and Kaihola, H.L. (1973) The effect of estradiol replacement therapy on early pregnant lutectomized patients. Am. J. Obstet. Gynecol., 117, 987990. Danielsson, B.R., Reiland, S., Rundqvist, E. and Danielson, M. (1989) Digital defects induced by vasodilating agents: relationship to reduction in uteroplacental blood flow. Teratology, 40, 351358. Deichert, U. and Buurman, C. (1995) Transabdominal Doppler sonographic measurements of blood flow differences in the main pelvic arteries during physiological spontaneous ovarian cycles. Hum. Reprod., 10, 15311536. Den Ouden, M., Cohen-Overbeck, T.E. and Waldimiroff, J.W. (1990) Uterine and fetal umbilical artery flow velocity waveforms in normal first trimester pregnancies. Br. J. Obstet. Gynaecol., 97, 716719. Deutinger, J., Rudelstorfer, R. and Bernaschek, G. (1988) Vaginosonographic velocimetry of both main uterine arteries by visual vessel recognition and pulsed Doppler method during pregnancy. Am. J. Obstet. Gynecol., 159, 10721076. Deutinger, J., Reinthaller, A. and Bernaschek, G. (1989) Transvaginal pulsed Doppler measures of blood flow velocity in the ovarian arteries during cycle stimulation and after follicle puncture. Fertil. Steril., 51, 466470. de Ziegler, D., Bessis, R. and Frydman, R. (1991) Vascular resistance of uterine arteries: physiological effects of estradiol and progesterone. Fertil. Steril., 55, 775779. Dickey, R.P. and Gasser, R.F. (1993) Ultrasound evidence for variability in the size and development of normal human embryos before the tenth postinsemination week after assisted reproductive technologies. Hum. Reprod., 8, 331337. Dickey, R.P. and Hower, J.F. (1995a) Ultrasonographic features of uterine blood flow during the first sixteen weeks of pregnancy. Hum Reprod, 10, 24482452.

Doppler ultrasound study of uterine and ovarian blood flow

Gerretsen, G., Huisjes, H.J. and Elema, J.D. (1981) Morphological changes of the spiral arteries in the placental bed in relation to preeclampsia and fetal growth retardation. Br. J. Obstet. Gynaecol., 88, 876881. Ghosh, D., De, P. and Sengupta, J. (1994) Luteal phase ovarian oestrogen is not essential for implantation and maintenance of pregnancy from surrogate embryo transfer in the rhesus monkey. Hum. Reprod., 9, 629637. Gilligan, D.M., Badar, D.M., Panza, J.A. et al. (1994) Acute effects of estrogen in postmenopausal women. Circulation, 90, 786791. Glock, J.L. and Brumsted, J.R. (1995) Color flow pulsed Doppler ultrasound in diagnosing luteal phase defect. Fertil. Steril., 64, 500504. Gosling, R.G., Dunbar, G., King, D.H. et al. (1971) The quantitative analysis of occlusive peripheral arterial disease by a non-intrusive ultrasonic technique. Angiology, 22, 5255. Gosling, R.G., Lo, P.T.S. and Taylor, M.G. (1991) Interpretation of pulsatility index in feeder arteries to low impedance vascular beds. Ultrasound Obstet. Gynecol., 1, 175179. Goswamy, R.K. and Steptoe, R.P. (1988) Doppler ultrasound studies of the uterine artery in spontaneous ovarian cycles. Hum. Reprod., 3, 721726. Goswamy, R.K., Williams, G. and Steptoe, P.C. (1988) Decreased uterine perfusiona cause of infertility. Hum. Reprod., 3, 955959. Greiss, F.C. (1972) Differential reactivity of the myoendometrial and placental vasculatures: adrenergic responses. Am. J. Obstet. Gynecol., 112, 2030. Guanes, P.P., Remohi, J., Gallardo, E. et al. (1996) Age does not affect uterine resistance to vascular flow in patients undergoing oocyte donation. Fertil. Steril., 66, 265270. Guirgis, R.R., Alshawaf, T., Dave, R. and Craft, I.L. (1993) Transvaginal crownrump measurements of 224 successful pregnancies which resulted from gamete intrafallopian transfer or in vitro fertilization. Hum. Reprod., 8, 19331937. Harrington, K.F., Campbell, S., Bewley, S. and Brower, S. (1991) Doppler velocimetry studies of the uterine artery in the early prediction of preeclampsia and intrauterine growth retardation. Eur. J. Obstet. Gynaecol., Reprod. Biol., 42, S14S20. Harrington, K., Cooper, D., Lees, C. et al. (1996) Doppler ultrasound of the uterine artery: the importance of bilateral notching in the prediction of preeclampsia, placental abruption, or delivery of a small for gestational age baby. Ultrasound Obstet. Gynecol., 7, 182188. Hertig, A.T. and Livingstone, R.G. (1944) Spontaneous, threatened, and habitual abortion: their pathogenesis and treatment. N. Engl. J. Med., 230, 797806. Hillard, T.C., Bourne, T.H., Whitehead, M.I. et al. (1992) Differential effects of transdermal estradiol and sequential progestins of impedance to flow within the uterine arteries of postmenopausal women. Fertil. Steril., 58, 959963. Hull, A.D., White, C.R. and Pearce, W.J. (1994) Endothelium-derived relaxing factor and cyclic GMP-dependent vasorelaxation in human chorionic plate arteries. Placenta, 15, 365375. Hustin, J. (1995) Vascular physiology and pathophysiology of early pregnancy. In Vourne, T., Jauniaux, E. and Jurkovic, D. (eds), Transvaginal Colour Doppler, Springer-Verlag, Heidelberg, pp. 4756. Hustin, J. and Schaaps, J.P. (1987) Echographic and anatomic studies of the maternotrophoblastic border during the first trimester of pregnancy. Am. J. Obstet. Gynecol., 157, 162168. Hustin, J., Schaaps, J.P. and Lambotte, R. (1988) Anatomical studies of the utero-placental vascularization in the first trimester of pregnancy. Trophoblast. Res., 3, 4960. Hustin, J., Jauniaux, E. and Schaaps, J.P. (1990) Histological study of the materno-embryonic interface in spontaneous abortion. Placenta, 11, 477486. Ignarro, L.J., Buga, G.M., Wood, K.S. et al. (1987) Endothelium-derived relaxing factor produced and released from artery and vein is nitric oxide. Proc. Natl. Acad. Sci. USA, 84, 92659269. Jacobson, S.L., Imhof, R., Manning, N. et al. (1990) The value of Doppler assessment of the uteroplacental circulation in predicting preeclampsia or intrauterine growth retardation. Am. J. Obstet. Gynecol., 162, 110114. Jaffe, R. and Warsof, S.L. (1991) Transvaginal color Doppler imaging in the assessment of uteroplacental blood flow in the normal first trimester pregnancy. Am. J. Obstet. Gynecol., 164, 781785.

501

Jaffe, R. and Warsof, S.L. (1992) Color Doppler imaging in the assessment of the uteroplacental blood flow in abnormal first trimester intrauterine pregnancies: an attempt to define etiologic mechanisms. J. Ultrasound Med., 11, 4144. Jaffe, R. and Woods, J.R. (1993) Color Doppler imaging and in vivo assessment of the anatomy and physiology of the early uteroplacental circulation. Fertil. Steril., 60, 293297. Jaffe, R. and Woods, J.R. (1996) Doppler velocimetry of intraplacental fetal vessels in the second trimester: improving the prediction of pregnancy complications in high-risk patients. Ultrasound Obstet. Gynecol., n, 262266. Jaffe, R., Dorgan, A. and Abramowicz, J.S. (1995) Color Doppler imaging of the uteroplacental circulation in the first trimester: value in predicting pregnancy failure or complication. Am. J. Roentgenol., 164, 12551258. Jauniaux, E. (1996) Intervillous circulation in the first trimester: the phantom of the color Doppler obstetric opera. Ultrasound Obstet. Gynecol., 8, 7176. Jauniaux, E., Burton, G.J., Moscoso, G.J. and Hustin, J. (1991a) Development of the early human placenta: a morphometric study. Placenta, 12, 269276. Jauniaux, E., Jurkovic, D. and Campbell, S. (1991b) In vivo investigations of anatomy and the physiology of early human placental circulation. Ultrasound Obstet. Gynecol., 1, 435445. Jauniaux, E., Jurkovic, D., Delogne-Desnock, J. and Meuris, S. (1992a) Influence of human chorionic gonadotropin, oestradiol and progesterone on uteroplacental and corpus luteum blood flow in normal early pregnancy. Hum. Reprod., 7, 14671473. Jauniaux, E., Jurkovic, D., Campbell, S. and Hustin, J. (1992b) Doppler ultrasonographic features of the developing placental circulation: correlation with anatomic findings. Am. J. Obstet. Gynecol., 166, 585587. Jauniaux, E., Johnson, M.R., Jurkovic, D. et al. (1994a) The role of relaxin in the development of the uteroplacental circulation in early pregnancy. Obstet. Gynecol., 84, 338342. Jauniaux, E., Zaidi, J., Jurkovic, D. et al. (1994b) Comparison of color Doppler features and pathological findings in complicated early pregnancy. Hum. Reprod., 12, 24322437. Jiang, C., Sarrel, P.M., Lindsay, D.C. et al. (1991) Endothelium-independent relaxation of rabbit coronary artery by 17 oestradiol in vitro. Br. J. Pharmacol., 104, 10331037. Jurkovic, D., Jauniaux, E., Kurjak, A. et al. (1991) Transvaginal color Doppler assessment of the uteroplacental circulation in early pregnancy. Obstet. Gynecol., 77, 365369. Kauppila, A., Kuikka, J. and Tuimala, R. (1978) Effect of fenoterol and isoxsuprine on myometrial and intervillous blood flow during late pregnancy. Obstet. Gynecol., 52, 558562. Kauppila, A., Koskinem, M., Puolakka, J. et al. (1980) Decreased intervillous and unchanged myometrial blood flow in supine recumbancy. Obstet. Gynecol., 55, 203205. Kay, H.H., Carroll, B.B., Dahmus, M. and Killam, A.P. (1991) Sonographic measurements with umbilical and uterine artery Doppler analysis in suspected intrauterine growth retardation. J. Reprod. Med., 36, 6568. Kharitonov, S.A., Logan-Sinclair, R.B., Busset, C.M. and Shinebourne, E.A. (1994) Peal expiratory nitric oxide difference in men and women: relation to the menstrual cycle. Br. Heart J., 72, 243245. Khong, T.Y., Liddell, H.S. and Robertson, W.B. (1987) Defective haemochorioal placentation as a cause of miscarriage. A preliminary study. Br. J. Obstet. Gynaecol., 94, 649655. Kremkau, W. (1990) Doppler Ultrasound: Principles and Instruments. W.B. Saunders, Philadelphia, London. Kupesic, S. and Kurjak, A. (1993) Uterine and ovarian perfusion during the periovulatory period assessed by transvaginal color Doppler. Fertil. Steril., 60, 439443. Kurjak, A., Kupesic-Urek, S., Schulman, H. and Zalud, I. (1991) Transvaginal color flow Doppler in the assessment of ovarian and uterine blood flow in infertile women. Fertil. Steril., 56, 870873. Kurjak, A., Crvenkovic, G., Salihagic, A. et al. (1993) The assessment of normal early pregnancy by transvaginal color Doppler ultrasonography. J. Clin. Ultrasound, 21, 38.

502

R.P.Dickey
and/or intrauterine growth retardation. Ultrasound Obstet. Gynecol., 7, 122128. Myatt, L., Brewer, A.S., Langdon, G. and Brockman, D. (1992) Attenuation of the vasoconstrictor effects of the thromboxane and endothelin by nitric oxide in the human fetal placental circulation. Am. J. Obstet. Gynecol., 166, 224230. Nargund, G., Bourne, T., Doyle, P. et al. (1996) Associations between ultrasound indices of follicular blood flow, oocyte recovery, and preimplantation embryo quality. Hum. Reprod., 11, 109113. Nazari, A., Check, J.H., Epstein, R.H. et al. (1991) Relationship of small for dates sac size to crown-rump length and spontaneous abortion in patients with a known date of ovulation. Obstet. Gynecol., 78, 369373. Nishimura, H., Takano, K., Tanimura, T. and Yasuda, M. (1968) Normal and abnormal development of human embryos: first report of the analysis of 1,213 intact embryos. Teratology, 1, 281290. North, R.A., Ferrier, C., Long, D. et al. (1994) Uterine artery Doppler flow velocity waveforms in the second trimester for the prediction of preeclampsia and fetal growth retardation. Obstet. Gynecol., 83, 378386. Olofsson, P., Thuring-Jonsson, A. and Marsal, K. (1996) Uterine and umbilical circulation during the oxytocin challenge test. Ultrasound Obstet. Gynecol., 7, 247251. ORahilly, R. and Muller, F. (eds) (1987) Developmental Stages in Human Embryos. Meriden-Stienhour Press, Merident, CT. Oyesanya, O.A., Parsons, J.H., Collins, W.P. and Campbell, S. (1996) Prediction of oocyte recovery rate by transvaginal ultrasonography and color Doppler imaging before human chorionic gonadotropin administration in in vitro fertilization cycles. Fertil. Steril., 65, 806809. Palmer, R.M.J., Ferrige, A.G. and Moncada, S. (1987) Nitric oxide released accounts for the biological activity of endothelium-derived relaxing factor. Nature, 327, 524526. Pearce, J.M., Campbell, S., Cohen-Overbeek, T. et al. (1988) References ranges and sources of variation for indices of pulsed Doppler flow velocity waveforms from the uteroplacental and fetal circulation. Br. J. Obstet. Gynaecol., 95, 248256. Pedersen, J.F. and Molsted-Pedersen, L. (1981) Early fetal growth delay detected by ultrasound marks increased risk of congenital malformation in diabetic pregnancy. Br. Med. J., 283, 269271. Penotti, M., Memcioni, T., Gabrielli, L. et al. (1993) Blood flow variations in internal carotid and middle cerebral arteries induced by postmenopausal hormone replacement therapy. Am. J. Obstet. Gynecol., 169, 12261232. Perrot-Applanat, M., Groyer-Picard, M.T., Garcia, E. et al. (1988) Immunocytochemical demonstration of estrogen and progesterone receptors in muscle cells of uterine arteries in rabbits and humans. Endocrinology, 123, 15111519. Planiol, T., Pourcelot, L. and Pottier, J.M. (1972) Etude de la circulation carotidienne par les methodes ultrasongigues de la thermographie. Rev. Neurol., 126, 127141. Pourcelot, L. (1974) Applications clinique de pexamen Doppler transcutance. In Pourcelot, L. (ed.), Belocimetric Ultrasonore Doppler. INSERM, Paris, p. 213. Powis, R.L. and Schwartz, R.A. (1991) Practical Doppler Ultrasound for the Clinician, Williams and Wilkins, Baltimore, MD, p. 187. Ramsay, B., De Belder, A., Campbell, S. et al. (1994) A nitric oxide donor improves uterine artery diastolic blood flow in normal early pregnancy and in women at high risk of preeclampsia. Eur. J. Clin. Invest., 24, 7678. Ramsay, B., Johnson, M.R., Leone, A.M. and Steer, P.J. (1995) The effect of exogenous oestrogen on nitric oxide production in women: a placebo controlled crossover study. Br. J. Obstet. Gynaecol., 102, 417319. Robertson, D.M., Mester, J. and Kellie, A.E. (1971) The measurement of oestradiol and progesterone in plasma from normal, infertile, and clomiphene treated women. Acta Endocrinol., 68, 523533. Rossavik, I.K., Torjusen, G.O. and Gibbons, W.E. (1988) Conceptual age and ultrasound measurements of gestational sac and crown-rump length in in vitro fertilization pregnancies. Fertil. Steril., 49, 10121017. Ryan, K.J. and Mackis, A. (1987) Significance of angiogenic and growth factors in ovarian follicle development. Adv. Exp. Med. Biol., 219, 203209.

Levi-Setti, P.E., Rognoni, G., Bozzo, M. et al. (1995) Color Doppler velocimetry of uterine arteries in pregnant and nonpregnant patients during multiovulation induction for IVF. J. Assist. Reprod. Genet., 12, 413417. Lindhard, J. (1913) Arch. Physiol., 30, 395397. Cited in Vorys, N., Ullery, J. and Hanusek, G. (1961) The cardiac output changes in various positions in pregnancy. Am. J. Obstet. Gynecol., 82, 13121321. Lindow, S.W., Davies, N., Davey, D.A. and Smith, J.A. (1988) The effect of sublingual nifedipine on uteroplacental blood flow in hypertensive pregnancy. Br. J. Obstet. Gynaecol., 95, 12761281. Long, M.G., Boultbee, J.E., Hanson, M.E. and Begent, R.H.J. (1989) Doppler time velocity waveform studies of the uterine artery and uterus. Br. J. Obstet. Gynaecol., 96, 588593. Magness, R.R. and Rosenfeld, C.R. (1989) Local and systemic estradiol-17: effects on uterine and systemic vasodilation. Am. J. Physiol., 256, 536542. Magness, R.R., Parker, C.R. and Rosenfeld, C.R. (1993) Systemic and uterine responses to chronic infusion of estradiol-17. Am. J. Physiol., 265, E690E698. Maini, C.L., Galli, G., Bellati, U. et al. (1985) Non-invasive radioisotopic evaluation of placental blood flow. Obstet. Gynecol. Invest., 19, 196206. Marsh, M.S., Bourne, T.H., Whitehead, M.I. et al. (1994) The temporal effect of progesterone on uterine artery pulsatility index in postmenopausal women receiving sequential hormone replacement therapy. Fertil. Steril., 62, 771774. Maulik, D. (1993) Hemodynamic interpretation of the arterial Doppler waveform. Ultrasound Obstet. Gynecol., 3, 219227. McParland, P., Pearce, J.M. and Chamberlain, G.V.P. (1990) Doppler ultrasound and aspirin in recognition and prevention of pregnancy induced hypertension. Lancet, 335, 15521555. Meekins, J.W., Pijnenborg, R., Hanssens, M. et al. (1994) A study of placental bed spiral arteries and trophoblast invasion in normal and severe preeclamptic pregnancies. Br. J. Obstet. Gynecol., 101, 669674. Merce, L., Barco, M.J. and Bau, S. (1996) Color Doppler sonographic assessment of placental circulation in the first trimester of normal pregnancy. J. Ultrasound Med., 15, 135142. Meyer, M.C., Brayden, J.E. and McLaughlin, M. (1993) Characteristics of vascular smooth muscle in the maternal resistance circulation during pregnancy in the rat. Am. J. Obstet. Gynecol., 169, 15101516. Miles, R.D., Menke, J.A., Bashiru, M. and Colliver, J.A. (1987) Relationships of five Doppler measures with flow in an in vitro model and clinical findings in newborn infants. J. Ultrasound Med., 6, 597599. Miller, V.M. and Vanhoutte, P.M. (1990) 17-Estradiol augments endothelium-dependent contractions to arachidonic acid in rabbit aorta. Am. J. Physiol., 258, 15021507. Molnar, M., Suto, T., Toth, T. and Hertelendy, F. (1994) Prolonged blockade of nitric oxide synthesis in gravid rats produces sustained hypertension, proteinuria, thrombocytopenia, and intrauterine growth retardation. Am. J. Obstet. Gynecol., 170, 14581466. Moncada, S., Palmer, R.M.J. and Higgs, E.A. (1991) Nitric oxide: physiology, pathophysiology and pharmacology. Pharmacol. Rev., 43, 109142. Montan, S., Liedholm, H., Lingman, G. et al. (1987) Fetal and uteroplacental haemodynamics during short term atenolol treatment of hypertension in pregnancy. Br. J. Obstet. Gynaecol., 94, 312317. Montan, S., Anandakumar, C., Arulkumaran, S. et al. (1993) Effects of methyldopa on uteroplacental and fetal hemodynamics in pregnancy induced hypertension. Am. J. Obstet. Gynecol., 168, 152156. Morris, N., Osborn, S.B., Wright, H.P. and Hart, A. (1956) Effective uterine blood flow during exercise in normal and preeclamptic pregnancies. Lancet, ii, 481484. Morris, N.H., Eaton, B.M. and Dekker, G. (1996a) Nitric oxide, the endothelium, pregnancy, and preeclampsia. Br. J. Obstet. Gynaecol., 103, 415. Morris, J.M., Fay, R.A., Ellwood, D.A. et al. (1996b) A randomized controlled trial of aspirin in patients with abnormal uterine artery blood flow. Obstet. Gynecol., 87, 7478. Murakoshi, T., Sekizuka, N., Takakuwa, H. et al. (1996) Uterine and spiral artery flow velocity waveforms in pregnancy-induced hypertension

Doppler ultrasound study of uterine and ovarian blood flow

Schiff, E., Tamarkin, M., Ben-Baruch, G. et al. (1991) Prediction of pregnancy-induced hypertension by a combination of rollover test and Doppler flow velocity measurement. J. Perinatal Med., 19, 245250. Scholtes, M.C.W., Wladimiroff, J.W., van Rijen, H.J.M. and Hop, W.C.J. (1989) Uterine and ovarian flow velocity waveforms in the normal menstrual cycle: a transvaginal study. Fertil. Steril., 52, 981985. Seligman, S.P., Buyon, J.P., Clancy, R.M. et al. (1994) The role of nitric oxide in the pathogenesis of preeclampsia. Am. J. Obstet. Gynecol., 171, 944948. Serafini, P., Batzofin, J., Nelson, J. and Olive, D. (1994) Sonographic uterine predictors of pregnancy in women undergoing ovulation induction for assisted reproductive treatments. Fertil. Steril., 62, 815822. Sladkevicius, P., Valentin, L. and Marsal, K. (1993) Blood flow velocity in the uterine and ovarian arteries during the normal menstrual cycle. Ultrasound Obstet. Gynecol., 3, 199208. Spencer, J.A.D., Giussani, D.A., Moore, P.D. and Hanson, M.A. (1991) In vitro validation of Doppler indices using blood and water. J. Ultrasound Med., 10, 305308. Stabile, I., Grudzinskas, J. and Campbell, S. (1990). Doppler ultrasonographic evaluation of abnormal pregnancies in the first trimester. J. Clin. Ultrasound, 18, 497501. Steer, C.V., Campbell, S., Pampiglione, J. et al. (1990) Transvaginal colour flow imaging of the uterine arteries during the ovarian and menstrual cycles. Hum. Reprod., 5, 391395. Steer, C.V., Campbell, S., Tan, S.L. et al. (1992) The use of transvaginal color flow imaging after in vitro fertilization to identify optimum uterine conditions before embryo transfer. Fertil. Steril. 57, 372376. Steer, C.V., Tan, S.L., Mason, B.A. and Campbell, S. (1994) Midluteal phase vaginal color Doppler assessment of uterine artery impedance in a subfertile population. Fertil. Steril., 61, 5358. Steer, C.V., Williams, J., Zaidi, J. et al. (1995a) Intraobserver, interobserver, interultrasound transducer and intercycle variation in colour Doppler assessment of uterine artery impedance. Hum. Reprod., 10, 479481. Steer, C.V., Tan, S.L., Dillon, D. et al. (1995b) Vaginal color Doppler assessment of uterine artery impedance correlates with immunohistochemical markers of endometrial receptivity required for the implantation of an embryo. Fertil. Steril., 63, 101108. Sterzik, K., Hutter, W., Grab, D. et al. (1989) Doppler sonographic findings and their correlation with implantation in an in vitro fertilization program. Fertil. Steril., 52, 825828. Striginc, F., Scida, P., Parri, C. et al. (1995) Modifications in uterine and intraovarian artery impedance in cycles of treatment with exogenous gonadotropins: effects of luteal phase support. Fertil. Steril., 64, 7680. Strohmer, H., Herczeg, C., Plockinger, B. et al. (1991) Prognostic appraisal of success and failure in an in vitro fertilization program by transvaginal Doppler ultrasound at the time of ovulation induction. Ultrasound Obstet. Gynecol., 1, 272274. Suonio, S., Simpanen, A.L., Olkkonen, H. and Haring, P. (1976) Effect of the left lateral recumbent position compared with the supine and upright positions on placental blood flow in normal late pregnancy. Ann. Clin. Res., 8, 2226. Tan, S.L., Zaidi, J., Campbell, S. et al. (1996) Blood flow changes in the ovarian and uterine arteries during the normal menstrual cycle. Am. J. Obstet. Gynecol., 175, 625631. Taylor, K.J.M., Burns, P.N., Wells, P.N.T. et al. (1985) Ultrasound Doppler flow studies of the ovarian and uterine arteries. Br. J. Obstet. Gynecol., 92, 240246. Tekay, A. and Jouppila, P. (1996) Intraobserver reproducibility of transvaginal Doppler measurements in uterine and intraovarian arteries in regularly menstruating women. Ultrasound Obstet. Gynecol., 7, 129134. Tekay, A., Martikainen, H. and Jouppila, P. (1995a) Blood flow changes in uterine and ovarian vasculature, and predictive value of transvaginal pulsed colour Doppler ultrasonography in an in-vitro fertilization programme. Hum. Reprod., 10, 688693. Tekay, A., Martikainen, H. and Jouppila, P. (1995b) Doppler parameters of the ovarian and uterine blood circulation in ovarian hyperstimulation syndrome. Ultrasound Obstet. Gynecol., 6, 5053.

503

Tekay, A., Martikainen, H. and Jouppila, P. (1996a) The clinical value of Doppler ultrasound. Hum. Reprod., 11, 15891593. Tekay, A., Martikainen, H. and Jouppila, P. (1996b) Comparison of uterine blood flow characteristics between spontaneous and stimulated cycles before embryo transfer. Hum. Reprod., 11, 364368. Thaler, I., Manor, D., Itskovitz, J. et al. (1990) Changes in uterine blood flow during human pregnancy. Am. J. Obstet. Gynecol., 162, 121125. Thompson, R.S., Trudinger, B.J. and Cook, C.M. (1988) Doppler ultrasound waveform indices: A/B ratio, pulsatility index, and Pourcelot ratio. Br. J. Obstet. Gynaecol., 95, 581588. Trudinger, B.J. (1991) Obstetric Doppler applications. In Fleischer, A.C., Romero, R., Manning, F.A. et al. (eds), The Principles and Practice of Ultrasonography, Chapter 12. Appleton and Lange, CT, pp. 173192. Trudinger, B.J., Giles, W.B., Cook, C.M. et al. (1985) Fetal umbilical artery flow waveform and placental resistance: clinical significance. Br. J. Obstet. Gynaecol., 92, 2330. Usan M, Haddad B, Breart G, Uzan, n. (1994) Uteroplacental Doppler and aspirin therapy in the prediction and prevention of pregnancy complications. Ultrasound Obstet. Gynecol., 4, 342349. Valentin, L., Sladkevicius, P., Laurini, R. et al. (1996) Uteroplacental and luteal circulation in normal first trimester pregnancies: Doppler ultrasonographic and morphologic study. Am. J. Obstet. Gynecol., 174, 768775. Vallance, P. and Collier, J. (1994) Biology and clinical relevance of nitric oxide. Br. Med. J., 309, 453457. Van Buren, G.A., Yang, D. and Clark, K.E. (1992) Estrogen-induced uterine vasodilatation is antagonized by L-nitroarginine methyl ester, an inhibitor of nitric oxide synthesis. Am. J. Obstet. Gynecol., 167, 828833. Van Zalen-Sprock, M.M., Van Vugt, J.M.G., Colenbrander, G.J. and Van Geijin, H.P. (1994) First trimester uteroplacental and fetal blood flow velocity waveforms in normally developing fetuses: a longitudinal study. Ultrasound Obstet. Gynecol., 4, 284288. Wada, I., Hsu, C.C., Williams, G. et al. (1994) The benefits of low-dose aspirin therapy in women with poor uterine perfusion. Abstract of the 2nd International Meeting of the British Fertility Society, Glasgow, pp. 35 and 61. Weiner, Z., Thaler, I., Levron, J. et al. (1993) Assessment of ovarian and uterine blood flow by transvaginal color Doppler in ovarian-stimulated women: correlation with the number of follicles and steroid hormone levels. Fertil. Steril., 59, 743749. Weiner, C.P., Knowles, R.G. and Moncada, S. (1994) Induction of nitric oxide synthases early in pregnancy. Am. J. Obstet. Gynecol., 171, 838843. Wilken, P. (1958) Morphogenese. In Snoeck, J. (ed.), Le Placenta Humain. Masson, Paris, p. 23. Zaidi, J., Campbell, S., Pittrof, R. et al. (1995a) Ovarian stromal blood flow in women with polycystic ovariesa possible new marker for diagnosis. Hum. Reprod., 10, 19921996. Zaidi, J., Jurkovic, D., Campbell, S. et al. (1995b) Luteinized unruptured follicles: morphology, endocrine function and blood flow changes during the menstrual cycle. Hum. Reprod., 10, 4449. Zaidi, J., Jurkovic, D., Campbell, S. et al. (1995c) Description of circadian rhythm in uterine artery blood flow during the periovulatory period. Hum. Reprod., 10, 16421646. Zaidi, J., Barber, J., Kyei-Mensah, A. et al. (1996a) Relationship of ovarian stromal blood flow at the baseline ultrasound scan to subsequent follicular response in an in vitro fertilization program. Obstet. Gynecol., 88, 779784. Zaidi, J., Collins, W., Campbell, S. et al. (1996b) Blood flow changes in the intraovarian arteries during the periovulatory period: relationship to the time of day. Ultrasound Obstet. Gynecol., 7, 135140. Zaidi, J., Pittrof, R., Shaker, A. et al. (1996c) Assessment of uterine artery blood flow on the day of human chorionic gonadotropin administration by transvaginal color Doppler ultrasound in an in vitro fertilization program. Fertil. Steril., 65, 377381.

Received on December 30, 1996; accepted on August 18, 1997