1IVF and Infertility Unit, Department of Obstetrics and Gynecology, Assaf Harofeh Medical .... and/or pattern assessed by ultrasonography during natural or.
Human Reproduction Update 1996, Vol. 2, No. 4 pp. 323–335
European Society for Human Reproduction and Embryology
The role of ultrasonography in the evaluation of endometrial receptivity following assisted reproductive treatments: a critical review Shevach Friedler1,3, Joseph G.Schenker2, Arie Herman1 and Aby Lewin2 1IVF and Infertility Unit, Department of Obstetrics and Gynecology, Assaf Harofeh 2Hadassah Hebrew University Medical School, Ein-Kerem, Jerusalem, Israel
TABLE OF CONTENTS Introduction Uterine predictors for implantation measurable by high–resolution ultrasonography Clinical context in which ultrasonographic predictors were used Exact timing of endometrial ultrasonographic evaluation to predict implantation Anatomical ultrasonographic parameters predicting uterine receptivity: endometrial thickness and pattern Physiological ultrasonographic parameters predicting uterine receptivity: transvaginal colour Doppler sonography Estimate of the predictive value of the various sonographic parameters Possible confounding factors that may influence evaluation of the results Options for treatment Conclusion References
323 324 325 326
327
329 330 332 333 333 334
We set out to estimate the value of ultrasonographic parameters as prognostic indicators of implantation following in-vitro fertilization (IVF) and embryo transfer. Our survey included 414 natural cycles, 3558 cycles following ovarian stimulation for IVF and embryo transfer, and 411 cycles with hormone replacement therapy for oocyte donation, reviewing 27 reports identified in a computerized literature research. The ultrasonographic prognostic indicators for implantation evaluated included peri-ovulatory endometrial thickness and pattern and Doppler measurements of uterine artery blood flow. Topics include: definitions of 3To
whom correspondence should be addressed
Medical Center, Zerifin 70300 and
the ultrasonographic parameters proposed to evaluate uterine receptivity; the clinical context in which they were used; the proposed optimal timing for sonographic evaluation; and, finally, their actual correlation with pregnancy rate following assisted reproductive technologies. For various sonographic parameters, negative predictive value, positive predictive value, sensitivity and specificity were calculated, based on published data. Sonographic parameters had a high negative predictive value and sensitivity, but a limited positive predictive value and low specificity. Several confounding factors may influence the interpretation of reports, and the statistical evaluation sometimes lacks calculation of the positive and negative predictive values of the parameters examined. Although ultrasonographic parameters of endometrial receptivity have a strong negative value in setting some minimum criteria, their value as prognostic indicators for implantation following embryo transfer has yet to be proved. Key words: endometrial pattern/endometrial receptivity/ endometrial thickness/uterine Doppler/uterine sonography Introduction There is no controversy regarding the importance of precise and specific endometrial maturational development in allowing implantation following assisted reproduction treatment. Adequate proliferation and differentiation during the proliferative phase must be followed by timely secretory changes during the luteal phase with stromal decidualization. These changes are influenced, and may be altered, by the hormonal environment. However, our knowledge concerning many biological mechanisms, including factors such as uterine blood flow and endometrial protein secretions that regulate and control implantation, is
324
S.Friedler et al.
still quite limited. Therefore the ideal method to predict uterine receptivity has yet to be established. The standard method of endometrial dating is the histological evaluation of an endometrial biopsy specimen (Noyes et al., 1950). Indeed, this technique has allowed the demonstration of a possible asynchrony in endometrial development in the course of cycles with ovarian stimulation for in-vitro fertilization (IVF) when embryo transfer had to be cancelled (Frydman et al., 1982; Cohen et al., 1984; Garcia et al., 1984). Obviously, the invasiveness of endometrial biopsy is not acceptable in the clinical context of assisted reproduction treatment cycles. Serum oestradiol and progesterone concentrations may reflect endometrial status. Oestradiol stimulates both increases in the size and number of myometrial and endometrial cells and changes in endometrial thickness. Indeed, endometrial thickness has been shown to be correlated with serum oestradiol/progesterone concentrations during the natural ovulatory cycle (Hall et al., 1979; Hackeloer, 1984) or ovarian stimulation cycles (Fleischer et al., 1984; Smith et al., 1984; Glissant et al., 1985). However, serum oestradiol concentration per se, expressing mainly the activity of the granulosa cells, is not a reliable indicator of endometrial maturity. On examining natural cycles, cycles treated with anti-oestrogens and artificial cycles, serum oestradiol concentrations or oestradiol:progesterone ratios were found to be inaccurate in predicting endometrial development assessed histologically, because normal endometrial development occurred over a wide range of serum hormonal concentrations (Li et al., 1993). Furthermore, Ben Nun et al. (1992) reported a lack of a relationship between serum oestradiol concentrations and histological endometrial maturation in oocyte donation cycles prepared artificially. Indeed, several authors have noted no correlation between serum oestradiol concentrations and endometrial thickness and/or pattern assessed by ultrasonography during natural or ovarian stimulation cycles (Glissant et al., 1985; Fleischer et al., 1986; Rabinowitz et al., 1986; Gonen et al., 1989; Gonen and Casper, 1990; Lentz and Lindberg, 1990; Davies et al., 1991; Ueno et al., 1991; Dickey et al., 1992; Khalifa et al., 1992). Even the positive correlation reported by Dickey et al. (1992), between increasing maturity of the ultrasonographic endometrial pattern and serum oestradiol concentrations, was not corroborated by Check et al. (1993a) following ovarian stimulation for IVF or by Alam et al. (1993) in patients receiving hormonal replacement therapy (HRT) for oocyte donation. Therefore one must conclude that estimates of the hormonal milieu failed to express endometrial receptivity. The reason for this asynchrony is not known. Basically one might postulate that the influence of oestrogens in the
circulation on the endometrium is dependent on the status of the endometrial oestrogen receptors in terms of quantity and function. The status of these receptors may depend on a variety of factors. As endometrial biopsy is invasive and hormonal milieu assessment inaccurate, the need to evaluate endometrial development encouraged the use of high-resolution ultrasonography as an alternative non-invasive method of assessment of uterine receptivity. Endometrial physical characteristics may be measured and expressed in terms of the tissue’s properties of sound-wave reflection (Forrest et al., 1988; Randall et al., 1989). Obviously, the ability to identify a receptive uterus prospectively by a non-invasive method would have an invaluable impact on treatment efficiency and success rates following assisted reproduction. Magnetic resonance imaging (MRI) has been proposed as an alternative non-invasive technique, demonstrating significant differences in the relative MRI signal intensities of the myometrium between conception and non-conception cycles (Turnbull et al., 1994); however, incorporation of this technique into routine assisted reproduction treatment seems hard to visualize yet because of practical obstacles such as availability and cost. The aim of this study was to review existing data concerning sonographic uterine predictors of implantation and clinical pregnancy, limited to the clinical context of assisted reproduction treatment. Our survey included 414 natural cycles, 3558 cycles following ovarian stimulation for IVF and embryo transfer and 411 cycles with HRT for oocyte donation, reviewing 27 reports identified in a computerized literature research (Medline). The topics covered include: definitions of the ultrasonographic parameters proposed to evaluate uterine receptivity; the clinical context in which they were used; the proposed optimal timing of sonographic evaluation; and, finally, the relationship between sonographic parameters and pregnancy rate following assisted reproduction treatment. To estimate the predictive value of the various sonographic parameters, their negative predictive value (NPV), positive predictive value (PPV), sensitivity and specificity were calculated, based on published data.
Uterine predictors for implantation measurable by high-resolution ultrasonography Anatomical parameters
Two anatomical parameters were suggested to evaluate uterine endometrium by ultrasound: endometrial thickness and endometrial pattern.
Uterine sonography for assisted reproductive technology
Figure 1. Longitudinal axis of the uterine cavity demonstrating a day 14 endometrium, 10.5 mm thick, with a triple-layer pattern.
325
Figure 2. Typical uterine artery blood flow pattern on the day of embryo transfer, with a pulsatility index (PI) of 2.07 and a resistance index (RI) of 0.83.
Endometrial thickness
Endometrial thickness is defined as the minimal distance between the echogenic interfaces of the myometrium and endometrium measured in the plane through the central longitudinal axis of the uterine body (Figure 1). Endometrial pattern
Endometrial pattern (texture, reflectivity) is defined as the type of relative echogenicity of the endometrium and adjacent myometrium, as demonstrated on a longitudinal uterine ultrasonic section (Figure 2). In principle, the central echogenic line represents the uterine cavity; the outer lines represent the basal layer of the endometrium or the interface between the endometrium and the myometrium. The relatively hypo-echogenic regions between the two outer lines and the central line may represent the functional layer of the endometrium (Forrest et al., 1988). Endometrial pattern, as demonstrated by high-resolution ultrasonography, varies during the cycle, as described in detail elsewhere (Fleischer et al., 1984; Smith et al., 1984; Glissant et al., 1985; Forrest et al., 1988). Several classifications exist, as presented in Table I. At first, four endometrial patterns were suggested (Smith et al., 1984). This classification was then simplified to three patterns (Gonen and Casper, 1990). Finally, two patterns were used by several authors (Sher et al., 1991).
third of the myometrium (Fleischer et al., 1991). The impedance of blood flow through the uterine arteries may be expressed as the pulsatility index (PI), unitless and angle independent, and the resistance index (RI), unitless and angle independent. The PI is measured from the flow velocity waveforms as the systolic peak velocity minus enddiastolic velocity divided by the mean. It may be classified as low, medium and high in the ranges 0.00–1.99, 2.00–2.99 and ≥3.0 respectively (Steer et al., 1992). The RI is calculated as the ratio of peak systolic flow velocity minus end-diastolic velocity divided by peak systolic velocity, ranging from 0.0 to 1.0. Diastolic blood flow may be categorized as full or continuous and discontinuous, i.e. reduced or absent flow velocity (Fleischer, 1991). It is also possible to measure endometrial blood flow concentrating on endometrial vessels located within 10 mm of the lateral endometrial border (Achiron et al., 1995) (Figure 3).
Clinical context in which ultrasonographic predictors were used Ultrasonographic predictors for uterine receptivity following embryo transfer were used in several clinical contexts with different regimens of endometrial preparation.
Uterine blood flow: Doppler of the uterine arteries
Uterine blood flow, as measured by colour Doppler, was suggested as a physiological parameter to assess receptivity. Colour Doppler signals are measured at the uterine arteries and their ascending branches located in the outer
Natural cycles
Embryo transfer was reported following natural cycles without ovulation induction, usually used in cryopreserved–thawed embryo replacement or natural IVF cycles.
326
S.Friedler et al.
Table I. Classifications of ultrasonographic endometrial patterns used in the evaluation of assisted reproduction treatment cycles Grade of endometrium
Definition
Reference
Presence of a halo, endometrial reflectivity increased compared with myometrium (i.e. brighter) Presence of a halo between the myometrium and endometrium and comparable reflectivity between myometrium and endometrium Reduced reflectivity, darker area surrounding the midline echo Echogenic black region surrounding the midline echo
Smith et al. (1984)
Entirely homogeneous, hyperechogenic pattern, without a central echogenic line Intermediate iso-echogenic pattern, with the same reflectivity as the surrounding myometrium and a non-prominent or absent central echogenic line A multilayered ‘triple-line’ endometrium consisting of a prominent outer and central hyperechogenic line and inner hypo-echogenic or black region
Gonen and Casper (1990)
Homogeneous hyperechogenic or iso-echogenic endometrium compared with the myometrium Triple-line multilayered pattern, ‘Halo pattern’ = outer peripheral layer of denser echogenicity and a central sonolucent area
Sher et al. (1991)
Four grades Grade A Grade B Grade C Grade D Three grades Type A Type B Type C Two grades Non-multilayered Multilayered
Ovarian stimulation
Ovarian stimulation was routinely performed during an IVF cycle, leading to the transfer of fresh embryos. Artificial preparation of the endometrium
Artificial preparation of the endometrium by the exogenous administration of oestrogen and progesterone is used for embryo transfer following oocyte donation or for cryopreserved–thawed embryo transfers.
Exact timing of endometrial ultrasonographic evaluation to predict implantation The window of implantation in the human is quite wide. Endometrial thickness is maximal at around ovulation, but no standardization exists as to the exact timing of the ultrasonographic examination that best predicts pregnancy to occur. Examinations performed during the luteal phase of the transfer cycle were related to conception (Rabinowitz et al., 1986; Imoedemhe et al., 1987) but, being performed post-transfer, the information is not relevant to the decisionmaking process prior to embryo transfer. In practice, one would need prognostic indicators that could be used, at the latest, on the day of human chorionic gonadotrophin (HCG) administration, allowing maximal time for necessary modifications to the stimulation protocol, oocyte retrieval scheduling or preparation for embryo cryopreservation should embryo transfer be postponed. Indeed, as presented in Table II, most reports have evaluated the role of ultrasound examinations performed during the late follicular phase, on the day of HCG administration
Figure 3. Spiral artery blood flow pattern on the day of embryo transfer, with a pulsatility index (PI) of 1.42 and a resistance index (RI) of 0.77.
(day 0) or during the early luteal phase (day +1). Some studies have reported findings on the day of oocyte retrieval (day +2) or on the day of embryo transfer (day +4 or +5). No study has addressed systematically the question of the relative efficiency of sonographic evaluation as a prognostic indicator for conception according to its exact day of performance. However, ultrasound evaluation may be performed efficiently prior to the day of embryo transfer. Khalifa et al. (1992) found no statistically significant differences in the endometrial thickness or pattern on the day of HCG administration or on the day of embryo transfer between conception and non-conception cycles. No significant difference in the PI measured on the day of oocyte retrieval or embryo transfer was reported by Coulam et al. (1995).
Uterine sonography for assisted reproductive technology
327
Table II. Differences in ultrasonographic measurement of endometrial thickness and endometrial pattern in conception (C+) versus non-conception (C–) cycles, during 4256 assisted reproduction treatment cycles identified in a literature survey, according to the day of examination Treatment and ovarian stimulation protocol
Natural cycle (cryopreserved) Natural cycle (IVF)
No. of cycles in the study
References by examination day Day
0a
(unless otherwise indicated)
Significance of difference in Day
4b
endometrium, C+ versus C– Thickness
Pattern
Coulam et al. (1994)
NS
NS
77
Al-Shawaf et al. (1993)
NS
NS
18
Ueno et al. (1991) (day ±1)
NS
P < 0.01
100
Ovarian stimulation for IVF and embryo transfer CC+HMG
Strohmer et al. (1994)
76
NS
Long GnRHa
107
Coulam et al. (1994)
NS
NS
Long GnRHa
102
Serafini et al. (1994)
NS
P < 0.002
Long GnRHa
1302
NS
NS
CC+HMG
139
Sher et al. (1993)
P < 0.001
Oliveira et al. (1993)
Long GnRHa
273
Check et al. (1993a)
CC/HMG/short GnRHa
148
Eichler et al. (1993)
74
Khalifa et al. (1992)
200
Dickey et al. (1992)
P < 0.05
P < 0.05
Ueno et al. (1991) (day +1)
NS
P < 0.05
Long GnRHa CC/HMG/short GnRHa Long GnRHa
38
Long GnRHa
320
Long GnRHa
85
P = 0.01 Khalifa et al. (1992)
NS
NS
NS
NS
Sher et al. (1991)
P < 0.01
P < 0.01
Check et al. (1991)
P = 0.06
P = 0.02 P < 0.01
CC+HMG
123
Gonen and Casper (1990) (day +1)
P < 0.01
CC+HMG
108
Gonen et al. (1989) (day +1)
P < 0.05
CC+HMG
190
HMG only
47
Rabinowitz et al. (1986)c
CC+HMG
30
Fleischer et al. (1986)c
NS
CC+HMG
196
Glissant et al. (1985)c
P < 0.01
HRT cycle (cryopreserved)
72
Al-Shawaf et al. (1993)
NS
NS
HRT cycle (oocyte donation)
88
Bustillo et al. (1995)
NS
P < 0.00001
99
Coulam et al. (1994)
NS
P = 0.001
59
Abdalla et al. (1994)
P = 0.03
58
Check et al. (1993b)
P < 0.01
NS
12
Shapiro et al. (1993) P < 0.05
NS
111
Welker et al. (1989)
Alam et al. (1993)
NS Rabinowitz et al. (1986)c
P < 0.05
NS –
P < 0.05
C+ versus C– = conception versus non-conception cycles, including clinical pregnancies only as conception. HRT = hormone replacement therapy; CC = clomiphene citrate; GnRHa = gonadotrophin-releasing hormone analogue; HMG = human menopausal gonadotrophin; IVF = in-vitro fertilization; NS = not significant. aDay 0 = luteinizing hormone surge/human chorionic gonadotrophin/progesterone start. bDay 4 = day of embryo transfer. cAbdominal ultrasound; when not noted, vaginal ultrasound.
Anatomical ultrasonographic parameters predicting uterine receptivity: endometrial thickness and pattern Endometrial thickness
Endometrial thickness is an easily measurable ultrasonographic parameter, expressing endometrial growth during the cycle. It is a distinct parameter, unrelated to endometrial pattern on the day of HCG (Fleischer et al., 1986; Gonen et al., 1989; Dickey et al., 1992; Li et al., 1993). Measured on day 0, in natural cycles the mean endometrial
thickness was found to be significantly thinner compared with cycles with ovarian stimulation (8.9 ± 8.0 versus 10.6 ± 2.5, P = 0.01; Ueno et al., 1991). The first question to be addressed is whether the mean endometrial thickness in conceptional cycles is significantly greater than in non-conception cycles. Using abdominal ultrasound, Glissant et al. (1985) reported significantly thicker endometrium in conception cycles compared with non-conception cycles; however, several reports using abdominal sonography (Fleischer et al., 1986; Rabinowitz et al., 1986; Welker et al., 1989) gave
328
S.Friedler et al.
contradictory findings. Endometrial thickness as a single parameter had no predictive value on the occurrence of pregnancy (Glissant et al., 1985). Li et al. (1992) reported no correlation between endometrial thickness measured by abdominal ultrasound and histological dating of the endometrium. In the clinical context of assisted reproduction treatment, endometrial growth, as assessed by the changes in endometrial thickness during IVF cycles stimulated by human menopausal gonadotrophin (HMG)/HCG, was well described by Rabinowitz et al. (1986). They noted from day –3 to day +2 (day of oocyte retrieval) a daily growth of 0.5 mm. Continuing measurement through the luteal phase, a slower linear growth rate of 0.1 mm/day was noted, until day +11. Conception cycles were characterized by an accelerated growth compared with non-conception cycles, arriving at day +17 to a significant difference in endometrial thickness. One must note, however, that prior to day +11, endometrial thickness had no predictive value on pregnancy occurrence. Imoedemhe et al. (1987) have also found a positive correlation between endometrial thickness in the luteal phase and conception rates following IVF and embryo transfer. Several reports advocated the use of endometrial thickness measurement as a clinical tool to predict implantation following ovarian stimulation for IVF and embryo transfer,
having found a significantly higher mean endometrial thickness measurement in conception compared with nonconception cycles. Gonen et al. (1989), using the more accurate technology of vaginal ultrasound, raised hopes for a useful and easily obtainable predictor of conception. They reported, in 108 patients undergoing IVF and embryo transfer with a clomiphene citrate + HMG ovulation induction protocol, a significantly greater endometrial growth during ovarian stimulation and greater endometrial thickness on day +1 in conception compared with non-conception cycles. Table II lists the reported differences comparing the mean endometrial thickness between conception and non-conception cycles, identified by a literature survey, showing the controversy regarding this subject. At present, insufficient data exist describing a linear correlation between endometrial thickness and the probability of conception. Is there an ideal range for endometrial thickness? Based on the literature survey, data exist regarding the mean endometrial thickness in 1605 assisted reproductive technology (ART) cycles, including 514 conception and 1110 nonconception cycles (Table III). Interestingly, the ranges of mean endometrial thickness for conception and nonconception cycles in the reports are virtually the same, being 8.6–11.8 and 8.6–11.9 respectively.
Table III. Variations in the reported mean endometrial thickness measured during 1605 assisted reproduction treatments, in 514 conception and 1110 non-conception cycles, and in the minimal value of endometrial thickness associated with clinical pregnancy, according to the ovarian stimulation protocol Assisted reproduction treatment
No. of cycles
and ovarian stimulation protocol
in the study
Natural cycle (cryopreserved)
100 77 219
References
Minimum endometrial
Mean endometrial thickness (mm)
thickness (mm)
Conception (n)
Non-conception (n)
Coulam et al. (1994)
6
11.9 ± 2.7 (34)
11.5 ± 2.8 (66)
Al-Shawaf et al. (1993) Friedler et al. (1994)
8 6
9.30 ± 0.36 (20) 11.2 ± 2.5 (37)
10.00 ± 0.24 (57) 10.2 ± 1.6 (182)
(Total 91)
(Total 305)
7 6 6
11.7 ± 2.6 (20) 8.0 ± 1.7 (35) 8.7 ± 0.4 (24)
11.8 ± 2.8 (56) 8.6 ± 2.0 (104) 7.5 ± 0.2 (61)
6
8.6 ± 0.3 (36)
7.1 ± 0.3 (87)
6 6
10.4 ± 2.0 (40) 11.5 ± 2.7 (47)
10.2 ± 2.5 (108) 11.2 ± 2.4 (60)
11.0 ± 0.3 (42)a (Total 244)
11.0 ± 0.4 (50) (Total 526)
(Total 396) Ovarian stimulation for in-vitro fertilization and embryo transfer CC+HMG CC+HMG CC+HMG
76 139 85
CC+HMG
123
CC/HMG/short GnRHa Long GnRHa
148 107
Long GnRHa
102 (Total 780)
HRT cycle (cryopreserved) HRT cycle (oocyte donation)
72 88 99 59 111
Strohmer et al. (1994) Oliveira et al. (1993) Gonen and Casper (1990) (day +1) Gonen et al. (1989) (day +1) Eichler et al. (1993) Coulam et al. (1994) Serafini et al. (1994) Al-Shawaf et al. (1993) Bustillo et al. (1995) Coulam et al. (1994) Abdalla et al. (1994) Alam et al. (1993)
6
9.0 ± 0.19 (18) 10.2 ± 2.3 (51) 10.3 ± 2.4 (40)
8.6 ± 0.17 (54) 9.7 ± 2.7 (66) 9.8 ± 2.8 (59)
5 7
10.24 ± 2.63 (19) 10.5 ± 3.5 (51)
8.62 ± 3.49 (40) 9.6 ± 4.2 (60)
8
(Total 429)
Values in parentheses are numbers of cycles. For abbreviations, see Table II. aData for term pregnancies only.
(Total 179)
(Total 279)
Uterine sonography for assisted reproductive technology
329
Dickey et al. (1992) observed a higher rate of biochemical pregnancies when endometrial thickness was 13 mm. The latter observation may be explained by mechanical damage to the hypertrophic endometrium at embryo transfer. In the context of pregnancies following ovulation induction (not IVF), contrary to the report by Dickey et al. (1993), Kroupl and Feichtinger (1993) found no correlation between endometrial thickness and the incidence of biochemical pregnancy. Finally, is there a minimal endometrial thickness required to establish a clinical pregnancy following ART? Since Gonen et al. (1991) observed a minimal endometrial thickness of 6 mm to achieve a pregnancy in natural artificial insemination by donor cycles, the concept that a minimal endometrial thickness is required to establish a clinical pregnancy became widely accepted. As for assisted reproduction treatment, several authors have reported a minimal endometrial thickness below which no pregnancies are observed. These minimal endometrial thicknesses vary between 5 and 8 mm, measured during the late proliferative to early luteal phase (Table III). Khalifa et al. (1992) reported a minimal endometrial thickness associated with conception as being 7 mm on the day of HCG and 8 mm on the day of embryo transfer (day +2).
younger ones (25 versus 5%), as well as in patients with uterine pathology in contrast to women with healthy uteri (87 versus 11%). These cases may represent a group of women with a reduced potential for endometrial proliferation. A triple-line pattern may be the sonographic parameter that mostly reflects endometrial receptivity (Smith et al., 1984; Welker et al., 1989; Fleischer et al., 1991; Dickey et al., 1992; Serafini et al., 1994), because it was associated more frequently, but not exclusively, with conception cycles. As shown by Serafini et al. (1994), after prospectively examining 102 cycles of assisted reproduction treatment and performing a logistic regression analysis, the endometrial pattern (predominantly the triple-line pattern) and diastolic blood flow were found to be the only predictive markers of clinical and term pregnancies. However, iso-echogenic or intermediate patterns did not preclude the possibility of pregnancy (Fleischer, 1991; Dickey et al., 1992; Serafini et al., 1994), although a lower pregnancy rate was observed if the endometrial pattern was more advanced than intermediate (Khalifa et al., 1992). Al-Shawaf et al. (1993), using intermediate classifications of endometrial pattern, did not significantly improve their ability to predict conception following assisted reproduction treatment.
Endometrial pattern
Physiological ultrasonographic parameters predicting uterine receptivity: transvaginal colour Doppler sonography
Welker et al. (1989) suggested that the endometrial pattern, as assessed by ultrasound, may be a predictor of implantation following IVF and embryo transfer. The methodology of endometrial preparation had no specific influence upon the observed endometrial pattern because no significant differences were noted between conception and non-conception natural cycles and HRT cycles (Al-Shawaf et al., 1993) or between natural IVF and ovarian stimulated IVF cycles (Lentz and Lindberg, 1990; Ueno et al., 1991). However, Coulam et al. (1994) reported that significant differences in endometrial pattern were noted only among HRT oocyte donation cycles, and not following natural cryopreserved–thawed embryo transfer or ovarian stimulated IVF conception and non-conception cycles. Table II lists the reported differences comparing endometrial patterns between conception and non-conception cycles according to the method of treatment. An endometrial pattern of low grade, characterized by an entirely homogeneous, hyperechogenic pattern without a central echogenic line, was frequently associated with nonconception cycles (Check et al., 1991, 1992, 1993a,b; Sher et al., 1991, 1993; Ueno et al., 1991), although its presence did not altogether preclude the chance of implantation (Gonen and Casper, 1990; Grunfeld et al., 1991; Serafini et al., 1994). Sher et al. (1991) observed a higher prevalence of poor endometrial grade in women aged >40 years than in
Doppler instrumentation enabled the measurement of uterine blood flow and an assessment of uterine perfusion (Taylor et al., 1985). Goswamy et al. (1988) and Goswamy and Steptoe (1988), after suggesting that decreased uterine perfusion plays a role in infertility, assessed its possible role in assisted reproduction treatment by measuring the blood flow through the exterior uterine vessels using transabdominal Doppler sonography. High frequency transducers (6.5–7.5 MHz) installed in the transvaginal ultrasound probes allowed not only an improvement of the resolution and imaging of the uterine structures, but also the combination of pulsed-wave Doppler techniques (Steer et al., 1990; Fleischer, 1991; Tekay et al., 1995). Colour flow imaging and pulsed-wave form analysis enabled an evaluation of flow impedance in the uterine arteries and the measurement of uterine perfusion (Steer et al., 1990). Thus, sonographic imaging gained a physiological dimension to add to the anatomical one. Good uterine perfusion, as shown by full diastolic blood flow with low resistance during the early and mid-secretory phases and expressed by a low RI, was correlated with conception following assisted reproduction treatment (Sterzik et al., 1989; Serafini et al., 1994). In a prospective study (Serafini et al., 1994), diastolic uterine blood flow measured
330
S.Friedler et al.
at the first branch of the uterine artery was found to be a valuable predictor of clinical pregnancy and term delivery following IVF and embryo transfer, along with the triple-line uterine pattern. Diastolic uterine blood flow was also found to be a significant predictor of term pregnancy, perhaps suggesting that insufficient diastolic flow impairs uterine receptivity (Serafini et al., 1994). Several studies have shown that an increase in uterine vascular resistance with concomitant diminished uterine blood flow significantly reduced the likelihood of pregnancy during an IVF treatment cycle (Sterzik et al., 1989; Steer et al., 1992). Increased resistance to blood flow may also be expressed by an increased PI. Steer et al. (1992) measured uterine artery impedance by PI on the day of embryo transfer, showing that in 82 women undergoing ovarian stimulation for IVF, a PI of >3.0 predicted a 35% implantation failure rate. Furthermore, the PI may even be assessed accurately on the day of oocyte retrieval (Coulam et al., 1995) or before cryopreserved embryo thawing (Steer et al., 1995), allowing earlier planning. In a study including not only ovarian stimulated IVF cycles but also spontaneous cryopreserved–thawed embryo transfer cycles and oocyte donation cycles, the blood flow impedance of the uterine arteries was studied using colour Doppler flow measurement of the right and left ascending branches of the arteries (Coulam et al., 1994). The upper limit of the PI was set to 3.3, being the 95th percentile of the PI among conception cycles. On the day of HCG in IVF, on the day of the luteinizing hormone (LH) surge in spontaneous cryopreserved–thawed embryo transfer cycles and on the day of progesterone administration in oocyte donation cycles, a PI >3.3 predicted 88% non-conception cycles, with 96% sensitivity and 26% specificity, in all groups examined (see Table VI). Comparing conception with non-conception cycles by treatment, a significant difference in the mean PI was found after natural cryopreserved–thawed embryo transfer cycles (2.79 versus 3.50; P = 0.03) but not after ovarian stimulation or HRT cycles. Comparing all groups, significantly (P < 0.001) more non-conception cycles had a PI >3.3 than conception cycles. Inter- and intra-operator variabilities of the PI were both 11%. The RI was found to be a less predictive parameter than diastolic uterine blood flow. When an evaluation was performed in a dichotomous way, the threshold value of 0.8 was without power to predict pregnancy and term delivery (Serafini et al., 1994). Uterine artery impedance, measured by the mean PI of left and right uterine arteries, was shown to have a significant correlation with biochemical markers of uterine receptivity, including 24 kDa protein, uterine oestradiol receptor and endometrial histology dating (Steer et al., 1995). In a study based on 76 HRT cycles undergoing cryopreserved–thawed embryo transfer, the PI on day 14 was significantly higher in
non-conception than in conception cycles (4.15, range 2.1–6.8, versus 2.85, range 1.4–3.6; P = 0.0053; Steer et al., 1995). The PI was not correlated with endometrial thickness. A PI >3.6 predicted 33% of the non-conception cycles. Using PI and endometrial pattern, on day 14, 39% (22/57) of the non-conception cycles could have been predicted. PI was shown as repeatable in a preliminary mock and subsequent real HRT cycle (Steer et al., 1995). Goswamy et al. (1988) postulated that uterine blood flow is a function of oestrogen concentration and the duration of steroidal exposure: as the cycle progressed, uterine blood flow increased with rising and decreased with falling oestrogen concentrations, and increased again in the luteal phase with the elevation of progesterone and oestrogen concentrations. Using transvaginal sonography, Serafini et al. (1994) could not corroborate this finding because all their patients had a similar length of exposure to gonadotrophins, achieved similar oestradiol concentrations and still had dissimilar uterine blood flows. Whereas most of the studies concentrated on the main external uterine arteries, recently Achiron et al. (1995) used Doppler flow measurement of the small endometrial vessels to assess endometrial perfusion. During endometrial preparation by HRT for oocyte donation candidates, they succeeded in demonstrating a significant change in endometrial flow, with PI changes correlated with the hormonal profile. An increase in endometrial diastolic blood flow during the proliferative phase was noted because of reduced impedance. At maximal oestrogen concentrations, minimal impedance to flow was noted. During the secretory phase, an increasing RI indicated a reduced endometrial blood flow because of increased diastolic impedance. This observation corroborated previous ones (Goswamy and Steptoe, 1988; Scholtes et al., 1989; Steer et al., 1992). Hypo-oestrogenic premature ovarian failure patients had high impedance to blood flow, corrected to normal values by exogenous oestrogen administration. Estimate of the predictive value of the various sonographic parameters Endometrial thickness
Finally, the practical question we aimed to clarify was whether it is possible to use the mean endometrial thickness measured during the late follicular or early luteal phase to predict the likelihood of subsequent conception in patients undergoing ovarian stimulation for IVF and embryo transfer. In fact, in our literature survey concerning endometrial thickness, which included 25 reports comprising 2665 assisted reproduction treatment cycles as listed in Table II, eight reports (1203 cycles) found that the difference in the mean endometrial thickness of conception and non-conception cycles was statistically significant, while 17 reports (1462
Uterine sonography for assisted reproductive technology
cycles) found no such significant difference. It seems that an endometrial thickness 6 mm, validated by ultrasonography, was strongly correlated with conception even more so than an endometrial biopsy dating performed in a previous mock cycle (Shapiro et al., 1993).
Table IV. Predictive values of sonographic parameters, according to the different discriminatory endometrial thickness value suggested by the different reports Endometrial thickness (mm)
Additional sonographic criteria
PPV (%)
NPV (%)
≥6 mm
Multilayer, PI < 3.1
45
87
95
≥8 mm
Multilayer
26
100
100
≥9 mm
Multilayer
34.9
≥10 mm
Multilayer
21.5
90.7
100
Sensitivity (%)
98.1
100
Specificity (%)
Pregnancy rate (study versus comparison groups) (%)
Comparison group criteria
Reference/data
22
b
