Successful embryo implantation can take place only in a receptive uterus. In human beings, the uterus becomes receptive during the mid-secretory phase (days 19–23) of the menstrual cycle, a functional period widely referred to as the window of implantation (WOI)1–3. During this transient WOI, the endometrium undergoes profound molecular changes that enable a developmentally competent blastocyst to contact the luminal epithelium and then invade the underlying stroma4. Importantly, the signals that induce this receptive phenotype in the luminal epithelium are derived from underlying stromal cells5. In fact, the WOI is characterized by profound changes in this stromal compartment, which include recruitment of specialized natural killer cells, vascular remodeling and—perhaps most striking—transformation of stromal fibroblasts into specialized rounded decidual cells4, 6, 7.
Decidual transformation of the stromal compartment occurs only in those mammalian species where implantation involves breaching of maternal tissues by the embryo8, 9. Furthermore, a correlation exists between the extent of the decidual process and the depth of trophoblast invasion among various species7. Human placenta formation, however, involves not only trophoblast invasion of the maternal decidua but also of the inner layer of the myometrium, a highly specialized uterine structure known as the junctional zone (JZ). Importantly, trophoblast invasion of the decidua and JZ, which forms the placental bed, is not only interstitial but also intravascular9–11. In fact, this process of intravascular trophoblast invasion critically controls the transformation of high-resistance, low-capacity spiral arteries into low-resistance, high-capacity vessels of pregnancy9, 12. In other words, the decidua and JZ form a functional unit that largely determines the likelihood of successful pregnancy outcome13.
The current model of fetal–maternal interactions views the trophoblast as the active invader of the passive decidua. However, this paradigm has been challenged by recent observations demonstrating that decidual cells acquire an invasive phenotype upon contact with the trophoblast14. Rather than being passively invaded, decidual cells may in fact actively encapsulate embryos that breach the luminal epithelium. Furthermore, emerging evidence suggests that decidual cells are programmed to respond to embryos of limited developmental potential, thus serving as biosensors that enable the mother to limit investment into failing pregnancies15. Furthermore, it has been suggested that impaired decidualization is associated with a prolonged WOI and lack of embryo quality control, thus facilitating implantation of developmentally compromised embryos15. This concept is supported by the landmark study of Wilcox and colleagues, demonstrating that implantation delayed beyond the conventional WOI is strongly associated with early pregnancy loss16.
These observations suggest that prospective assessment of the quality of decidualization response in the endometrium may be an important tool for predicting the likelihood of successful implantation and pregnancy outcome. Since its introduction into the clinic, ultrasound has been used widely to assess uterine features such as endometrial thickness, endometrial pattern and uterine blood flow that may be predictive of pregnancy, especially in the context of assisted reproductive technology. Furthermore, the development of three-dimensional (3D) ultrasound and 3D Doppler studies has enabled a much more detailed examination of uterine morphology, including visualization of the JZ.
In this review, we critically assess the role of various ultrasound techniques and markers in predicting the likelihood of conception and subsequent pregnancy outcome. In addition, we argue that, with the introduction of high-resolution ultrasound technologies, imaging of extremely early implantation events may yield novel markers predictive of pregnancy outcome.
Several early studies assessed the value of endometrial thickness measurements by ultrasonography in predicting the likelihood of pregnancy. The data were conflicting. While the endometrium was reported to be thicker in conception cycles than in non-conception cycles17, this was not confirmed by another study18. In an in-vitro fertilization (IVF) population, endometrial thickness on the day after embryo transfer was reported to be higher in patients who subsequently conceived19. In contrast, from a review of the early literature, it was concluded that endometrial thickness is comparable between successful and unsuccessful IVF treatment cycles (range, 8.6–11.8 and 8.6–11.9 mm, respectively)20. The concept that the endometrium must measure at least 6 mm to sustain a pregnancy in natural cycles was first established two decades ago21. This concept was subsequently refined by the recommendation that endometrial thickness should be ≥ 7 mm on the day of human chorionic gonadotropin (hCG) administration and ≥ 8 mm on the day of embryo transfer22. These recommendations were supported by the observation that endometrial thickness of < 6 mm has a high negative predictive value (NPV) for pregnancy20. Thus, while ‘normal’ endometrial thickness does not necessarily predict pregnancy, a thin endometrium means that implantation is highly unlikely to occur.
The appearance of the endometrium on ultrasound changes in a cycle-dependent manner. Consequently, endometrial morphology has been studied widely in an attempt to predict the likelihood of pregnancy. The results are similar to those of endometrial thickness in that a normal trilaminar appearance of the endometrium (multilayered or presence of midline echo) has a low positive predictive value (PPV) for pregnancy (33.1%), whereas the absence of a multilayered pattern does not exclude conception but renders it unlikely (NPV, 85.7%)20, 23.
Uterine Blood Flow
Uterine artery blood flow can be expressed usefully by impedance indices, the pulsatility index (PI) and resistance index (RI) (Table 1). An early study in 198824 reported that a poor uterine artery blood flow response to treatment with exogenous hormones (estradiol and norgestrel) was associated with a low pregnancy rate in IVF cycles. The authors hypothesized that a decreased uterine perfusion response is a contributing factor to infertility. In addition, they reported greater pregnancy rates after hormone therapy and improvement in uterine perfusion. This was in agreement with their previous study the same year, which showed increasing uterine perfusion with rising levels of plasma estradiol and progesterone25. Taking this work a step further in 1992, Steer et al. reported higher pregnancy rates and implantation rates and more multiple pregnancies with lower uterine blood flow impedance before embryo transfer26. They concluded that uterine artery blood flow is a useful method for assessing uterine receptivity in assisted conception programs27. In agreement with these findings, other reports demonstrated that low impedance uterine artery blood flow in the early and mid luteal phase is associated with higher conception rates in assisted reproduction cycles23, while high uterine artery impedance (PI > 3.3) predicts treatment failure with relatively high sensitivity but low specificity (88% and 26%, respectively)28. In a review of the literature in 1996, Friedler et al. reported a high NPV and sensitivity (range, 88–100% and 96–100%, respectively) and a relatively higher PPV and specificity (range, 44–56% and 13–35%, respectively) with Doppler assessment of uterine artery blood flow using an upper limit for PI of 3 or 3.3, compared with the other ultrasonic parameters20. These data suggest that uterine vascularity, or the factors that affect it, are important for implantation and for the subsequent pregnancy to be successful.
Table 1. Summary of published data on the role of uterine artery Doppler and endometrial and subendometrial vascularity in predicting outcome in in-vitro fertilization (IVF) and recurrent pregnancy loss
A recent study examined the optimal timing of B-mode and Doppler ultrasound evaluation of uterine receptivity in IVF populations. Different variables were studied, including endometrial thickness, endometrial morphology, uterine artery PI, the presence or absence of a protodiastolic notch, the presence of end-diastolic blood flow, and endometrial and subendometrial blood flow distribution patterns. The most effective combination for evaluating uterine receptivity was the presence or absence of end-diastolic blood flow in the uterine arteries, endometrial morphology and endometrial thickness. The best sensitivity and specificity were obtained on the day of hCG administration (81.1% and 81.3%, respectively)29.
Three-dimensional Endometrial Vascularity
3D ultrasound has been shown to have a low intra- and interobserver variability in assessing endometrial volume, with Virtual Organ Computer-aided AnaLysis being the most reproducible technique. Endometrial vascularity can be assessed also using 3D power Doppler30.
Using B-mode and color Doppler imaging, the cyclical changes in uterine size, echogenicity and vascularity have been studied throughout the menstrual cycle in relation to a positive urinary luteinizing hormone test and in relation to the first day of menses. Endometrial thickness was found to increase up to days 3 and 4 of the cycle, after which it remains relatively constant. This was associated with a gradual decrease in the uterine arterial PI throughout the cycle, which significantly increased at the time of next menses31. These cyclical changes have been further studied using 3D ultrasonography. Endometrial volume increases in the follicular phase and plateaus in the luteal phase32. Vascularity of the endometrial and subendometrial regions increases from the mid-follicular phase and peaks 3 days prior to ovulation before decreasing again over the next 5 days and then increasing until the next cycle33, 34. Raine-Fenning et al. further showed that endometrial and subendometrial vascularity indices were significantly lower in women with unexplained subfertility during the mid and late follicular phase, irrespective of estradiol and progesterone levels35. The subendometrial region was defined as a 5-mm shell around the defined endometrial contour35, 36. As would be expected from endometrial thickness data, 3D measurements of endometrial volume have showed a high NPV for implantation failure. However, there have been conflicting results among investigators regarding differences in endometrial volume, as well as endometrial and subendometrial vascularity between conception and non-conception cycles30, 37. Several factors have been cited as possible explanations for these conflicting results, e.g. different ultrasound examination and analysis techniques, variations in the resolution of equipment and different treatment protocols, especially in the fertility groups20. However, endometrial and subendometrial 3D power Doppler indices have been shown to have acceptable reproducibility in evaluating physiological and pathological changes of the endometrium38. While Doppler studies of the uterine arteries have suggested a mechanism whereby uterine vascularity impacts on implantation, 3D Doppler studies of the endometrium have not supported this concept. Currently we do not know whether compromising endometrial vascularity leads to implantation failure. To illustrate this point, Ng et al.39 examined endometrial vascularity in 451 IVF treatment cycles that resulted in 94 clinical pregnancies. No significant differences in blood flow were seen in pregnant versus non-pregnant cycles. The authors concluded that measurements on endometrial and subendometrial blood flow are not good predictors of pregnancy. The same authors subsequently examined a further 161 patients and found increased vascularity in cycles destined to lead to a live birth compared to a miscarriage40.
A few studies have tested the hypothesis that changes in endometrial vascularity play an essential role in recurrent miscarriage (Table 1). Higher uterine artery PI was reported in women with recurrent pregnancy loss (RPL), especially in those with antinuclear antibodies, than in controls41. Findings were similar in women with unexplained RPL or uterine congenital abnormalities and in women with antiphospholipid antibodies syndrome42. Most recently the endometrial and subendometrial vascularity in cases of unexplained RPL were reported to be lower in women with RPL than in controls. This assessment took place 7 days post ovulation43. It is of interest to observe the same patterns of endometrial vascularity in patients with RPL and failed assisted reproduction, suggesting that endometrial vascularity does play a role in endometrial receptivity and pregnancy maintenance.
Uterine Junctional Zone
The JZ, or endometrial–myometrial junction (EMJ), is the transitional zone, sandwiched between the endometrium and the outer myometrium. Unlike most human tissues with a mucosa, the endometrium does not contain a submucosal layer. This layer usually exists to protect against mucosal invasion into adjacent tissue44. Brosens et al. postulated in 1995 that the JZ differs from the outer myometrium not only structurally but also functionally. Furthermore, they proposed that irregular thickening of the JZ is the magnetic resonance (MR) criterion for diagnosis of diffuse adenomyosis13. An increased diameter of the posterior JZ of the uterus on MR imaging correlated with invasion of the basal endometrium into the inner myometrium from as early as the third decade of life45.
The JZ is thought to play an important role in regulating uterine function and hence fertility46. Video-vaginosonography studies have shown that propagated myometrial contractions in the non-pregnant uterus originate only from the JZ and that the frequency and orientation of these contraction waves are dependent on the phase of the menstrual cycle47. These inner myometrial contractions vary in orientation, amplitude and frequency throughout the menstrual cycle, and are thought to be influenced by estradiol and progesterone. In the follicular phase of the cycle, these contractions are from the cervix towards the fundus and their amplitude and frequency increase significantly as ovulation approaches. Following ovulation a decrease in contractility under the influence of progesterone is observed44. There is evidence that this pattern of contractions facilitates sperm transport48, aids implantation of the developing blastocyst, improves the supply of oxygen and nutrients to the decidua49 and, in addition, contributes to menstrual shedding50, 51. On the other hand, alteration of the JZ interface, and hence contractility, is proposed to have an integral role in diverse reproductive disorders50. Indeed, high-frequency uterine contractility in women undergoing IVF treatment on the day of embryo transfer has been shown to negatively affect the outcome, possibly by expelling embryos from the uterine cavity52. It has been suggested that aberrant uterine peristaltic activity at the EMJ may cause local microtraumas that enable invasion of endometrial glands and stroma53, 54. While this proposed model of adenomyosis remains speculative, electron microscopy studies demonstrated that smooth muscle cells from uteri with adenomyosis are ultrastructurally different from myocytes of disease-free uteri55, 56.
High-resolution MRI has been used to monitor changes in endometrial and JZ morphometry during the normal menstrual cycle. This analysis demonstrated that both endometrial and JZ volumes increase significantly towards ovulation. While the endometrial volume decreases significantly post ovulation, the JZ appears to be less regular during the luteal phase of the cycle57. These findings support the hypothesis that the JZ and endometrium constitute a functional unit13.
Changes in the JZ during pregnancy were first observed incidentally on MR scans carried out 7 days post ovulation in what turned out to be a conception cycle. Even this early in the pregnancy, a low-signal intensity mass at the site of implantation has been visualized, with changes in the regularity of the adjacent JZ58. These uterine changes are not observed in cases of ectopic pregnancy59. It was therefore postulated that these MRI features are the result of a change in local blood flow in the area underlying the implantation site, in response to the presence of implantation factors60, 61. The impact of the JZ structure on the likelihood of pregnancy after IVF treatment was recently evaluated in a prospective study. The uterus was imaged prior to IVF treatment and the average and maximal JZ thickness values were measured on T2-weighted sequences. Strikingly, the implantation failure rate was as high as 95.8% in women with an average or maximal JZ of 7 and 10 mm, respectively, compared to 37.5% in other patients (P < 0.0001). This strong association between an abnormally thickened JZ and IVF failure was independent of the cause of infertility or the age of the patient62.
3D transvaginal ultrasonography now enables reliable assessment and visualization of the JZ. As on T2-weighted MR images (Figure 1), the JZ on ultrasound appears as a hypoechogenic zone underlying the endometrium (Figures 2 and 3). 3D reconstruction of coronal sections of the uterine cavity has made it possible to assess minor changes in the lateral and fundal aspects of the JZ, which are impossible to delineate using standard 2D ultrasound. In addition, processing modalities such as volume contrast imaging further enhance visualization of the hypoechoic JZ in comparison to that using 2D imaging44, 63. Thus, 3D technology has made it possible to accurately assess and grade changes in the JZ architecture such as thickening, disruption and protrusion of the endometrium into the inner myometrium.
Ultrasound Assessment of Peri-implantation Events
Although there is considerable interest in visualizing the JZ for the diagnosis of adenomyosis44, no studies have been reported to date on potential changes in JZ structure or morphology associated with early implantation events. Based on the observations outlined above, it appears reasonable to speculate that the JZ is exquisitely sensitive to pregnancy-associated hormonal and embryonic signals. If so, failure of the JZ to remodel in early pregnancy may be predictive of subsequent failure or even obstetrical complications such as preterm labor11.
Unlike MRI, ultrasonography is not contraindicated in early pregnancy. As outlined above, 3D ultrasonography appears to be more accurate than is conventional 2D imaging in characterizing changes in the JZ. Increased ultrasound resolution also means that it has become possible to visualize the probable site of implantation. Early implantation sites are usually characterized by the presence of a hyperechoic ring around the conceptus that protrudes into the endometrial lumen (Figure 4). This appearance on ultrasound fits well with the emerging concept that the implanting embryo is rapidly and actively encapsulated by migratory decidual cells64. It remains to be seen whether the absence of decidual protrusion or encapsulation detected on ultrasound could serve as an indirect marker of inadequate decidualization, and thus predict subsequent pregnancy loss.
There is a growing body of evidence supporting the notion that defects in the implantation process create an adverse ripple effect during the subsequent course of pregnancy, ultimately culminating in either pregnancy loss or obstetrical disorders associated with impaired placentation such as pre-eclampsia, placental abruption and fetal growth restriction9, 65, 66. As outlined in this review, high-resolution ultrasonography has the potential to provide new insights into the implantation process and to functionally assess the quality of the maternal decidual response and associated, but as yet ill-defined, changes in the JZ. If this notion is correct, it should in turn be possible to predict pregnancy complications at the implantation stage and to design tailored interventions that would genuinely prevent these subsequent complications.