Recurrent Implantation Failure-update overview on etiology, diagnosis, treatment and future directions

The term ‘implantation failure’ can be used to describe both patients who have never shown quantifiable signs of implantation such as increased levels of hCG, and those who have increased hCG production without later ultrasound evidence of a gestational sac [4]. Implantation failure can apply to patients undergoing assisted reproductive technology (ART) and patients trying to conceive without any fertility treatment.

Recurrent implantation failure (RIF) is only applicable to patients undergoing ART. Although there is no accepted formal definition for recurrent implantation failure, Orvieto suggests that it is after three failed in vitro fertilization-embryo transfer (IVF-ET) cycles with good quality embryos transferred [5]. Zeyneloglu et al. agree that it is after three unsuccessful cycles of IVF specifically with two embryos of high quality [6]. Simon and Laufer add that the embryo and the endometrium can both play an active role in RIF [7]. It is also important to consider maternal age in the definition, and whether the embryos were transferred at the cleavage-stage or as blastocysts [8]. Coughlan et al. suggest a more complete working definition taking into account maternal age, number of embryos transferred, and number of cycles completed. They define RIF as the failure of clinical pregnancy after 4 good quality embryo transfers, with at least three fresh or frozen IVF cycles, and in women under the age of 40 [4].

Biochemical pregnancy is defined in similar terms by a variety of authors. Maesawa uses the International Committee Monitoring Assisted Reproductive Technologies and World Health Organization definition for biochemical pregnancy as the detection of hCG in blood or urine without subsequent clinical signs of pregnancy [9]. Coulam defines biochemical pregnancy when two or more increased values of hCG can be measured, yet there can be no evidence of gestational sac detected on transvaginal ultrasound 2 weeks later. Therefore, only a chemical measure indicates that a pregnancy occurred. By previous definition, biochemical pregnancy falls into the category of implantation failure [10]. This differs from a clinical pregnancy which is determined, as defined by the American Society of Reproductive Medicine (ASRM), by ultrasound examination with evidence of a gestational sac or by histopathological examination [1].

Despite the fact that many authors agree on a similar general definition, the parameters used to measure the hCG level differ substantially between > 5 to > 25 mIU/ml [11‐13].

Due to variations in definitions for recurrent implantation failure and biochemical pregnancy, there is scarce data that accurately represents the incidence or prevalence. Biochemical pregnancy is actually not uncommon, and its reported incidence varies from 8 to 33% in the general population, including those who spontaneously conceived [9]. However, it is not clear how accurate these figures are since most patients who recognized that they had a biochemical pregnancy were undergoing ART. Therefore, it is likely that these patients are measuring their hCG levels earlier than those who are conceiving spontaneously and waiting until they a miss a period before taking a pregnancy test [14]. In spontaneous conception it is estimated that 30% of pregnancies are lost before implantation and 10% are clinical pregnancy losses [15]. It is also important to note that spontaneous pregnancy is only achieved in around 30% of normal fertile couples on the first try, and many succeed on subsequent efforts [16]. Moreover, it may be worth considering whether or not biochemical pregnancy is a pathological process.

Maternal age plays a crucial role in the quality of the embryos that are used for IVF [4]. It has long been known that as maternal age increases so does the frequency of aneuploidy [6]. As a result, many authors have age cut offs in their studies. Pregnancy rates also have been found to be decreased as maternal age increases [11]. In particular, Salumets et al. found that the major predictive factor contributing to pregnancy outcome in frozen embryo transfer, specifically with Intracytoplasmic Sperm Injection (ICSI) technique, was maternal age. The patient’s age at oocyte collection and freezing was the sole determining factor for biochemical pregnancy outcome. Starting around age 39, there was a significantly higher rate of occurrence of biochemical pregnancy. Delivery rate is also affected by both maternal age and embryo quality [17]. Shapiro et al. found that there are higher rates of embryo-endometrial asynchrony with increasing maternal age. 50% of transfers were asynchronous in women < 35 years old in comparison with 68.1% which were asynchronous in women > 35 years old. In addition, implantation rates, which were calculated for each blastocyst transfer and therefore treated as a numeric value, were significantly lower in IVF cycles in women of age ≥ 35 (mean 24.5 ± 36.8) compared with women < 35 (mean 41.1 ± 42.1), and biochemical pregnancy rates were significantly higher in women ≥35 compared to < 35 (28.1% vs. 14.9% respectively). Live birth rate was significantly higher in women < 35 (50.7%) than in women > 35 (28.5%) Patients > 35 years old also had reduced oocyte yield, blastocyst formation, and endometrial thickness [18]. The United States Center for Disease Control and Prevention reports overall ART success rates each year, with the most recent statistics currently from 2015. In agreement with the literature, implantation rates in both fresh and frozen embryo transfer in patients < 35 (41.3% and 47.1%) were dramatically higher than those rates in women > 44 (1.9% and 16.2%). It was interesting to note that the implantation rate when using donor oocytes, representing all age categories, was 53.6% in fresh embryo transfers and 40.2% in frozen embryo transfers [19]. This further supports the idea that embryo quality including genetic characteristics, declines with increasing maternal age.

Increased BMI (> 25 kg/m²) has also been shown to impact implantation rate [20]. In patients undergoing IVF, Class I, II, and III obese patients (BMI > 30 kg/m²) had the highest chance of implantation failure demonstrated by respective odds ratios, 0.69 (0.53–0.90), 0.52 (0.36–0.74), and 0.58 (0.35–0.96), when compared with patients of normal weight (BMI 18.5–24.99 kg/m²). Though there have been no reported differences across different BMI groups for biochemical pregnancy specifically, the Class III Obesity Patients (BMI > 40 kg/m²) had the highest overall rates of miscarriage (including biochemical pregnancy) [21]. In addition, overweight and obese women (BMI > 25 kg/m²) undergoing IVF who had fewer oocytes collected had higher risks of implantation failure and miscarriage than women of healthy weight with the same number of collected oocytes [20]. When more oocytes are collected, there is a higher chance of increased quality of embryo for transfer, and with fewer oocytes comes a higher likelihood of negative pregnancy outcome due to fewer good quality embryos. Obese women required more gonadotropin stimulation cycles, yet they had statistically fewer oocytes for collection (average of 8 vs. 10 in non-obese women, P = .03). This suggests that the oocyte quality and follicular development might be affected by obesity [22]. Though recurrent pregnancy loss is a different entity than recurrent implantation failure, elevated BMI is considered to be the most important risk factor, after increasing maternal age, in miscarriage for patients suffering from recurrent pregnancy loss [15].

It can be particularly difficult to find valid information on smoking’s impact on fertility due to the bias of self-reported smoking during pregnancy. This is largely due to the fact that women might be concealing their smoking since there is a negative attitude toward smoking in pregnancy [23]. Smoking has been shown to lead to a significantly increased risk of miscarriage (time unspecified) for each pregnancy in comparison with non-smoking patients undergoing ART [24]. In women undergoing IVF, smoking patients were found to have lower estradiol levels during ovarian stimulation. Cigarette toxins might play a role in disrupting corpus luteum formation and implantation of the embryo [23]. Fuentes et al. also demonstrated that female smokers with higher serum cotinine levels (a nicotine metabolite) had significantly fewer ova retrieved during IVF cycles, though cotinine levels had no significant impact on rates of implantation in IVF cycles [25]. In addition, smoking patients had a decreased live birth rate suggesting that smoking has an overall negative impact on pregnancy outcome [24]. Maternal smoking was more commonly linked to spontaneous miscarriage with normal fetal karyotype than abnormal karyotype, suggesting that the toxic effects of carbon monoxide and nicotine might be the principle factors causing harm. Carbon monoxide can cause a depletion of oxygen to the fetus, and nicotine can lead to vasoconstriction and decreased nutrients to the fetus due to maternal appetite suppression [26]. Pregnancy rates have been shown to be lower overall among smokers when compared with non-smokers, yet there are minimal differences in conception rates between the two groups [23]. It is also necessary to consider the effect of smoking on male fertility. Kunzle et al. found that male smokers had a significantly decreased sperm count (229.4 ± 251.5 × 10⁶ cells vs. 278.1 ± 264.2 × 10⁶ cells, P = .0001), higher percentage of abnormal morphology (21.2 ± 14.6% normal forms vs. 23.7 ± 15.5% normal forms, P = .0007) decreased motility (105.6 ± 132.7 × 10⁶ cells vs. 126.6 ± 136.8 × 10⁶ cells, P = .0016) and increased pH level measured by citrate concentration (86.7 ± 57.3 vs. 111.7 ± 303.1, P = .0072) [27].

It has been shown that elevated levels of cortisol, also known as “the stress hormone,” lead to a 2.7 times greater chance (95% CI = 1.2–6.2) of miscarriage within the first 3 weeks after conception in comparison with women with low cortisol levels. Cortisol production in the body rises in response to psychological, immunological, and other stressors, suggesting that it serves as a marker signaling the female body that it is not in its best state for reproduction [28]. This suggests that preventing or decreasing maternal stressors may have a positive outcome on pregnancy. In contrast to this study, Pasch et al. found that psychological stress such as clinical anxiety or depression does not have a significant affect on IVF outcome in women undergoing a first time fertility treatment. However, it is IVF failure that may lead to higher rates of both anxiety and depression in the immediate period after a negative IVF outcome. There were higher rates of post-IVF depression in women with IVF failure than in women who achieved successful pregnancy (44% vs. 30% P < .001). IVF failure was also associated with higher rates of post-IVF anxiety in comparison with women who were able to become pregnant successfully (60% vs. 50% P < .001) [29]. It is important to consider that this study addressed only those undergoing first time fertility treatment, and that it was not necessarily those who would go on to suffer from recurrent implantation failure.

Many women who have experienced RIF have also been determined to have chronic endometritis (CE) from bacterial colonization, often with minimal or no clinical signs of infection [44]. Kushnir et al. found that among a sample of infertile patients, 45% had CE, particularly those with RIF [45]. CE is a uterine pathology that has traditionally been diagnosed on histological examination, on visualization with hysteroscopy, and by bacterial culture. Immunohistochemistry stain for syndecan-1(CD 138), (Fig. 1) a plasma cell marker, can be used to provide a more accurate diagnosis, though agreement on a standard number of plasma cells present is not yet established. Bouet et al. confirmed a prevalence of 14% of CE using histological evaluation in RIF patients. In addition, hysteroscopy had a 40% sensitivity for detection of CE, visualizing mucosal edema, endometrial hyperemia, and the presence of micropolyps (all part of diagnostic criteria for CE) [46, 47]. Cicinelli et al. found that 66% of women were diagnosed with CE via hysteroscopy, 57.5% with histology, and 45% were also culture positive. Though culture is the least reliable method of detecting CE, it did allow for specific pathogens to be detected for targeted therapy. Most of the pathogens found consist of common bacteria like Group B Streptococcus, Escheria Coli, and Enterococcus Faecalis, or Mycoplasma. In some cases, sexually transmitted infections such as Chlamydia can be responsible [44]. A molecular method of diagnosis of chronic endometritis has shown promising results as presented by Moreno et al. in a new study published in June 2018. Real time polymerase chain reaction (RT-PCR) can identify bacterial DNA with an agreement of 76.92% when samples showed concordance by hysteroscopy, histology, and culture. RT-PCR showed 75% sensitivity and 100% specificity compared with the concordant results of the other three tests. It allows for detection of culturable and unculturable bacteria colonizing the endometrium even without histological signs of infection. This molecular test shows promise as a faster and more reliable tool for a streamlined diagnosis of chronic endometritis [48]. The bacteria present in the endometrium lead to abnormal lymphocyte counts and as a result, an environment that interrupts normal endometrial receptivity. Implantation rate of those cured of infection was 37%, compared with 17% in those that were not, however these rates did not reach statistical significance. (See treatment section for details on antibiotic protocols) However, live birth rates with next cycle of IVF after infection was cured was significant, with a rate of 61% in comparison with 13% in those that could not be cured with antibiotics [44]. In some cases, it is not necessarily chronic infection that leads to decreased implantation rates, but rather the constituents of bacterial flora present in the endometrium. While it was once thought that the endometrium was a sterile environment, it has now been accepted that Lactobacillus colonize this region in addition to the vagina. In a recent study, Moreno et al. demonstrated that women with Lactobacillus dominated endometrium undergoing IVF have been shown to achieve higher rates of successful implantation (60.7% vs. 23.1%) and live birth (58.8% vs. 6.75%) rates compared to those with non-Lactobacillus dominated endometrium (Gardnerella, Streptococcus, and other organisms present) (Fig. 2) [49].

There are several types of uterine pathologies including polyps, myomas, and adhesions that can impact implantation rates in patients undergoing IVF. Most of the time the patients are asymptomatic and sometimes these problems can even go unnoticed on transvaginal ultrasound suggesting that another method of uterine cavity assesssment such as hysteroscopy be required (see treatment section for more details) [58]. Myomas can cause a distortion of the endometrial cavity and adhesions which often develop following surgery or infection can prevent attachment of the embryo to the luminal surface [4]. The ASRM has reported that hydrosalpinges can also have a negative impact on implantation in women undergoing IVF. Though not well understood, several mechanisms have been considered such as deprivation of proper embryo development, due to the fluid’s lack of nutrients and energy [59]. In addition, the fluid can negatively impact endometrial receptivity, and can also physically flush the embryo out preventing implantation [60]. Mullerian abnormalities such as septate uterus and bicornuate uterus should also be taken into consideration in patients with RIF. Septate uterus might contribute to RIF as those with untreated septa have poor outcomes after IVF in comparison with those who underwent hysteroscopic metroplasty. However, women with a bicornuate uterus usually have normal implantation, but these patients have a higher risk of mid trimester pregnancy loss [4].

The endometrium itself can also be the source of implantation failure. Previous endometrial trauma, prolonged use of birth control pills, and impaired uterine blood flow are all potential etiologies of a thin endometrium [61]. Evaluating the blood flow in the uterine radial arteries is useful in assessing the degree of blood flow to the endometrium. When there is a decrease in blood flow, epithelial cell growth and vascular endothelial growth factor (VEGF) production can be reduced, which ultimately deprives the endometrium of necessary angiogenesis and growth factors to reach proper thickness for successful implantation. Miwa et al. conducted a transvaginal Doppler ultrasonography study that found that uterine radial artery resistance index was significantly higher in patients with thin endometrium < 8 mm than with normal thickness endometrium > 8 mm (0.852 (0.826–0.955) vs. 0.751 (0.549–0.840), P < .05) [62] The thickness of the endometrium has been determined to have an effect on implantation rates. Though an exact threshold for proper endometrial thickness has not been unanimously agreed upon, it is suggested that about 8 mm is the lower limit for which ART can still usually be successful. From a thickness of 9 mm to 16 mm chance of pregnancy increases from 53 to 77% showing that a significant difference does exist in implantation rates [63].

Chromosomal abnormalities including translocations, mosaicism, inversions, and deletions are more frequent in RIF patients than the general population, and yet the prevalence is only about 2%. The most common abnormality is translocation [64, 65]. Voullaire et al. looked at the difference between women with aneuploidy in one or two chromosomes, and women with complex abnormalities, defined as three or more chromosomes with aneuploidy. Though this is a rare occurrence, it was shown to be significantly more likely to be in women who had RIF, whereas aneuploidy in only one or two chromosomes was not considered to be a significant cause of RIF. It is accepted that this complex abnormality is mitotically derived as it is observed more commonly in the embryos than in the oocytes upon collection, though it is unclear what exactly causes this problem. [66] It is recommended that parental karyotyping is involved in cases of women suffering from RIF, but only in specific subgroups such as nulliparous women with history of miscarriage, as it would be very expensive to make it a universal test, and it would only protect a small group of patients. In addition, men with severe infertility are also recommended for karyotyping [64]. In 2016 Koot et al. published a study which found an endometrial gene signature made up of 303 genes which was found to be predictive of RIF. RIF patients had down regulation of genes that were involved in cell regulation and division and cytoskeleton and cilia formation. It is interesting to note that there was one group of RIF patients who were consistently incorrectly classified with a > 90% error rate, suggesting that this is not random, but that the underlying etiology in these patients is not due to the expression of this group of endometrial genes, further confirming the complexity of the approach to evaluation and treatment of RIF [67].

Heparin has been evaluated for use in RIF patients, though there is not yet evidence to recommend its use for improved pregnancy outcomes in these patients. A group of RIF patients treated with low molecular weight heparin had almost identical implantation, clinical pregnancy, and live birth rate outcomes when compared with controls [81].

Several different immunological therapies used to increase rates of implantation have been studied such as Tacrolimus, intravenous immunoglobulin (IVIG), peripheral blood mononuclear cells, and granulocyte colony stimulating factor. An elevated Th1/Th2 ratio has been shown to negatively impact implantation rates. However, IFNg is a Th1 specific cytokine that is necessary in some degree in arterial development for implantation, and yet levels beyond a certain threshold are implicated in implantation failure.

In patients who have had CE diagnosed via hysteroscopy and culture, antibiotic therapy has been shown to be an effective intervention to cure most infections, leading to more successful implantation rates in future IVF cycles. Women infected with common gram positive bacteria, Enterococcus and Strep Agalactiae, were given Amoxicillin and Clavulanate twice a day for 8 days, and gram negative bacteria such as Escheria Coli were given ciprofloxacin twice a day for 10 days. Women with Mycoplasma and Ureaplasma were treated with 1g Josamycin twice a day for 12 days with the addition of Minocycline in persistent cases. When infections were cured, the implantation rate was found to be higher in the next cycle at 37%, though not statistically significant in comparison with a rate of 17% in those who had persistent infection even after three antibiotic treatments. The clinical pregnancy rate in those with CE who cleared their infection with antibiotics was 65.2% in comparison with 33% in those with persistent infection (P = 0.039). The live birth rate in those who had cleared their CE with antibiotics was 60.8%, significantly higher than the 13.3% in those who had not cleared the infection (P = 0.02) [44]. There is currently a clinical trial at the University of Valencia in Spain recruiting participants aiming to find a new non-invasive diagnostic test for CE, using bacterial DNA retrieved from endometrial fluid. The antibiotics given to patients who test positive for CE will be determined based on bacteria detected on sequencing and quantitative PCR. The new sequencing technique will be compared with patients who have either biopsy or culture to determine if the new technology has greater efficacy [90]. There is another trial being conducted at Fu Xing hospital in Beijing China which is evaluating whether the use of amoxicillin and clavulanate will cure the CE and ultimately lead to greater live birth rates. This trial will evaluate the presence of CE solely using hysteroscopy and immunohistochemical staining of CD 138 antibody (Syndecan 1) [91].

In addition to maternal factors playing a role in RIF, male factor, particularly spermatozoal morphology also can play a part. Spermatozoa have to have smooth nuclei with normal chromatin content and normal head shape to function properly. Intracytoplasmic morphologically selected sperm injection (IMSI) requires examination of spermatozoa under ultra-magnification X3600 before injection into the oocyte. While implantation rate (19.2% vs. 7.8%, P = 0.042), clinical pregnancy rate (43.1% vs. 10.5%, P = 0.02) and live birth rate (34.7% vs. 0%, P = 0.003) were reported to be significantly higher among those undergoing sperm selection via IMSI procedure than those undergoing simple intracytoplasmic sperm injection (ICSI) in a retrospective study conducted by Shalom-Paz et al. in 2015 [109], other studies could not demonstrate any benefit of using IMSI. There is still a lack of specific microscopic criteria for the assessment of sperm morphology, and therefore more studies are required to confirm the advantages of IMSI before a standardized clinical protocol can be created for this particular procedure [110]. A Cochrane review in 2013 called for further trials to be conducted since higher quality evidence needs to be established in order to recommend this technique for clinical practice [111]. Table 2 summarizes the results of treatment options with regard to implantation, clinical pregnancy, and live birth rates.

In November 2017 The European Society of Human Reproduction and Embryology released a new set of guidelines which included recommendations for lifestyle modifications for RPL patients. Although RPL is a distinctly different disorder, there are similarities and many overlapping risk factors with RIF. Recommendations for RPL patients include smoking cessation because of its possible negative impact on live birth, and acheiving a healthy range BMI since obesity is associated with obstetric complications and lower chances of live birth. Stress has also been shown to be associated with RPL, but there is no evidence of a cause and effect relationship [2]. Evidence suggests that smoking, obesity, and high cortisol levels also play a role in implantation failure. Though more research needs to be conducted regarding the specific physiology behind these issues, lifestyle interventions such as assistance in quitting smoking, healthier diet and regular exercise, and emphasis on taking care of mental health may positively impact those suffering from RIF. These modifications require less invasive medical assistance and seem like an optimal first step in trying to change future implantation outcome in IVF treatment in couples struggling to get pregnant and have a successful delivery outcome. In addition, other behavioral changes may also be relevant to improve outcomes in these patients.

A groundbreaking study published by Tamar Ben-Shaanan et al. in 2016 has led us to consider a completely new direction for RIF treatment intervention in addition to the previously mentioned therapies. It has long been suggested that the placebo effect has an association with activation of the reward system, yet the mechanism has not been known. This group demonstrated in a mouse model that the ventral tegmental area, one of the major components of the brain’s reward system and a dynamic player in the placebo response, when activated by positive experience, led to activation of the immune system. Both the innate and adaptive immune system were activated in response to exposure and re-exposure to Escherichia coli Bacteria. There was increased macrophage activity, decreased bacterial load, and greater lymphocyte response [112]. Since many etiologies of RIF are related to the immune system’s function, it seems that this could very well be applied to RIF patients. It may be a great first step in a new protocol for IVF patients to focus on creating an environment that would lead to activation of this reward system in couples suffering from RIF. Encouraging the patients to engage in activities that create enjoyment, and establishing a positive environment could be the first step in protocol for treating RIF patients. This intervention is relatively simple in that it does not involve medications or any invasive procedures. It can be done in the comfort of the patient’s familial and social circle, and on the patient’s own time. For each patient a positive activity or setting might mean something different, like spending 3 days meeting with friends each week, eating out in a nice restaurant once a week, making specific time for creative activity, or watching a comedy film a few times a week. This type of therapy can easily be tailored to each patient depending on personal hobbies and interests. This concept might be a great starting point for future studies in RIF treatment, and even now might be a great first step in protocol for evaluating and treating patients suffering from RIF.