Defining recurrent implantation failure: a profusion of confusion or simply an illusion?

To define implantation failure, it is necessary to first define successful implantation and implantation rate. Currently, there are several different ways to determine the implantation rate. An obvious approach is to use serum hCG as a surrogate for trophoblast invasion. Data reported to SART measures implantation rate as number of gestational sacs/number of embryos transferred, as this allows a definition based on the number that implanted per number transferred. However, the SART definition cannot account for embryos that were implanted but lost before ultrasound. An additional definition used in some studies is the sustained implantation rate, which is the number of embryos with cardiac activity per number of embryos transferred. The sustained implantation rate is useful clinically as it tracks closely with live birth. With each of these definitions, the absence of the appropriate finding (hCG, sac, cardiac motion) represents a failed treatment cycle, though it remains unclear whether failure at each of these points have the same or similar causes. For example, biochemical losses seem to occur with similar frequency with aneuploid and euploid embryos, while failure to achieve detectable hCG is about half as likely for aneuploid embryos ( 7 ). It would seem useful to track all three definitions, at least in research, and report on all three until a universal definition is applied. It is important to note that a distinction, or lack thereof, between miscarriage and implantation failure is needed. Clearly from a reproductive outcome and treatment perspective, but also given the adverse perinatal outcomes associated with recurrent miscarriage ( 27 ). Additionally, there may be long term impact of recurrent loss, such as increased risk of myocardial infarction, that may warrant specific screening in patients at high risk ( 28 ).

Once a definition of implantation is determined, the definition of its failure is obvious. Yet, how we define recurrent remains unclear. The aim of a definition of RIF should be to identify those women who have an abnormally low chance of pregnancy per embryo, in order to provide prognostic data and allow interventions that may improve implantation in subsequent transfers. Currently, there is heterogeneity in criteria deemed indicative of RIF between types of providers, clinics, and geographic location. A recent international survey of 735 clinicians highlights the heterogeneity of currently-used clinical criteria [27]. In this survey, 84% of clinicians defined RIF based on the number of embryos transferred, with the majority (45%) defining RIF as failure of three fresh or frozen embryo transfers. Interestingly, factors such as location of the clinic (European versus non-European) and private versus public were correlated with the definition used ( 29 ). Individual authors have proposed specific definitions. Tan et. al suggested a definition of failure to achieve a pregnancy after 3 completed IVF cycles ( 30 ). Two additional studies defined RIF as 3 unsuccessful cycles of IVF with at least 2 embryos of high quality or failure of clinic pregnancy after 4 good quality embryo transfers with at least 3 fresh or frozen IVF cycles in women under the age of 40 ( 31 , 32 ). There is also variation in how professional societies define RIF. The preimplantation genetic diagnosis consortium of the European Society of Human Reproduction and Embryology PGD Consortium ( 33 ) has defined RIF as >3 failed embryo transfers with high quality embryos, or the failed transfer of ≥10 embryos in multiple transfers ( 3 , 10 , 16 , 34 ). ASRM has not published specific criteria. A critical problem with these definitions is that there are many important factors influencing implantation success are not accounted for, including oocyte and uterine age, length of infertility, euploidy (and how this was determined), systemic diseases, lifestyle issues (e.g. obesity or smoking), uterine structural abnormalities, chronic endometritis, and presence of endometriosis. Some of these are routinely screened for and others are not, but might be if RIF is diagnosed. Obviously, the lack of implantation of a high-quality blastocyst in a 28 year old woman with absent tubes as her only infertility factor would be much less likely than a 39 year old with transfer of untested, cleavage-stage embryos. Thus, recently, authors have suggested personalized definitions of implantation failure. Embryo aneuploidy (and therefore oocyte age) is arguably the most important contributing factor to failure of ART. For this reason, Ata et al. proposed a new definition of RIF that accounts for anticipated euploidy rate on the basis of age, using a statistical model ( 34 ). The statistical simulation found that no age category was associated with a 95% probably of successful implantation with six embryos transferred! Rather, a 95% probability of success was not reached until seven blastocysts were transferred in women < 35 years old. The needed number of embryos increased with age; at 38 years of age, 10 blastocysts were needed, and at age 42 there was no practical number that allowed a 95% probability of implantation. Of course, the number needed would be reduced, if euploid embryos are transferred. Rozen et al. suggested using a theoretical implantation rate (TIR) to create a personalized diagnosis of RIF and to account for many of the aforementioned factors, but the authors did not provide a way to calculate the TIR, and large scale data taking all of these factors into account is lacking ( 35 ).

In conclusion, RIF does not have a universal definition. Clinical experience tells us that some women have a greatly reduced chance of embryo implantation. These women, if they possess an adequate supply of euploid oocytes, often are able to conceive with further IVF attempts, since the chances are seldom 0%. However, the emotional and financial burdens of these choices are high and the chance of success is certainly not optimal. Therefore, a practical definition of RIF is needed to inform both research and clinical practice. We propose that large scale data be applied to allow personalization of the diagnosis by modeling multiple factors. Until we have the ability to more fully personalize, definitions should at minimum account for the risk of aneuploidy as a significant factor governing implantation.

The implantation of an embryo is fundamental to a successful pregnancy. It is contingent on the presence of a specific intrauterine environment, a biologically intact embryo, and a complex series of interactions between them. This process, necessary for species survival, must be strongly influenced by natural selection but is curiously inefficient; most intercourse does not result in a fertilized egg, and only a fraction of fertilized eggs can become a baby. Even with an apparently euploid embryo, a large determinant of successful conception, embryo transfer results in ongoing pregnancy less than 60% of the time ( 1 ). Early abnormalities in implantation contribute to infertility by causing biochemical and first trimester losses, while later abnormalities are associated with pregnancy complications, such as miscarriage and preeclampsia ( 2 ). Recurrent implantation failure (RIF) is a poorly defined but devastating clinical scenario where pregnancy is not achieved after multiple embryo transfers. However, there is a profound lack of agreement about the definition of RIF, and even some heterogeneity for the clinical determination of implantation and its failure. A standard definition of RIF would benefit the field by improving research study design and allowing the synthesis of independent studies. The data, thus created, would enhance our ability to determine specific causes, methods of diagnosis, and methods of treating and preventing RIF. The goal of this review is to make progress toward a universally accepted definition of RIF by exploring what is known about implantation and its failure.