|Home | About | Journals | Submit | Contact Us | Français|
We read with interest the commentary by Griesinger recently published in Human Reproduction detailing the utilization of implantation rate as an outcome marker in studies of human reproduction (Griesinger, 2016). In it he suggests that utilization of implantation rate distorts outcomes data in assisted reproductive technology (ART) and should be entirely abandoned as a marker in favor of live birth rate. We propose that, as is often the case when extreme stances are taken in debates, truth lies in a more moderate stance. Indeed, it is likely that no single parameter in ART will work for everything. Both live birth on an intent-to-treat basis and sustained implantation rate, which in point of fact is representative of live birth per embryo transfer, are of great importance.
This can be clearly observed when analyzing the effectiveness of implementing new technology of any type in ART. Live birth on an intent-to-treat analysis must of course be utilized in order to determine the impact on overall live birth rate. This end-point incorporates the impact which any diagnostic tool or clinical intervention might have that would cause embryos to arrest or be discarded prior to transfer or cryopreservation. Given that every technique has some risk, delivery rates on an intent-to-treat basis provide the best estimate of the overall impact of the technique.
Does that then provide all the information needed to critically evaluate a new technology or treatment paradigm? Not even close. It is important to note that live birth rate is easily manipulated by simply changing transfer order. This motivation to increase transfer order to ‘optimize’ delivery rates is inevitably accompanied by an increase in multiple gestation rates and increased maternal and neonatal morbidity. The sustained implantation rate through delivery, not just the presence of a gestational sac (which is not a helpful end-point for most studies) is a parameter that allows greater insight into what is in essence live birth per embryo transferred. This allows for a more critical analysis of outcomes while controlling for the number of embryos transferred. If universal practice was single embryo transfer this debate would not exist; however, we are far from this ideal case in the field in general.
This drive toward improvement in the field which entails decreased transfer order, and ideally single embryo transfer mandates the use of more than a single end-point to fully evaluate critical class I data. An example is the study by Forman et al. which was referenced and discussed by Griesinger. Several points deserve clarification before further discussion. In this particular randomized controlled trial, elective single embryo transfer (eSET) combined with preimplantation genetic screening-comprehensive chromosome screening (PGS-CCS) was compared with an untested double embryo transfer (DET) (Forman et al., 2013). The primary end-point was live birth rate based on intent-to-treat. It is important to note that the study design as indicated in the manuscript was a non-inferiority trial comparing the live birth rate after transferring a single euploid embryo versus two untested embryos. The primary aim was to investigate if high pregnancy rates could be maintained while dramatically reducing the risk of multiple gestations. Sustained implantation rate was only one of several important secondary outcomes reported. As noted above, when interventions occur in the pre-transfer setting which alter the transfer order of embryos, it is important to analyze the sustained implantation rate as a secondary outcome to gain a more complete understanding of the impact the intervention might have.
Clearly investigators have been seeking a means to attain the same outcomes following SET that are enjoyed following DET. Five prior trials of SET versus DET and a subsequent Cochrane Database Review (Pandian, 2005) detailed unsuccessful attempts. The study by Forman et al. was the first adequately powered trial to show that SET could be performed while maintaining similar success rates as DET by utilizing PGS. This was a meaningful study specifically because it was able to demonstrate equivalent outcomes by using a new paradigm which decreased transfer order and enhance safety by the near elimination of multiple gestation—all while maintaining delivery rates previously attained through routine but higher risk treatment models.
It is also important to note that the mathematical example provided by Griesinger on implantation rate with one versus two embryos being transferred requires clarification. The core of the divergence of pregnancy rates and implantation rates in SET versus DET cases lies in simple probability calculations and algebra. Figure 1 demonstrates the resulting calculated delivery rates in SET (DRSET) and DET (DRDET) when sustained implantation rate is held equal. When you examine the curves, you see that if they are evaluated anywhere other than the extremes of sustained implantation rates (0 or 100%), the curves have a fairly significant difference (up to 25% when sustained implantation rate is 50%). The divergence in outcomes is simply because of transfer order. It is not due to the denominator issue of patients that have failed implantation as implied in the example provided by Griesinger.
Finally, sustained implantation rate is a helpful consideration when thinking about cumulative pregnancy rates per oocyte retrieval. In the hypothetical example provided by Griesinger, the DET group would have had twice as many embryos transferred and arguably fewer supernumerary embryos available for cryopreservation. Thus, those patients who failed to conceive after DET may have less opportunity to conceive on a subsequent frozen embryo transfer. Couples who desire to have more than one singleton sequentially may have less of an opportunity to do so and a large number would be exposed to the adverse risks of a twin pregnancy. Would it also be a distortion of data to analyze the health of deliveries (for example the proportion of full term singletons)? We think not. As clinical and laboratory procedures have improved our attention has moved beyond maximizing pregnancy rates to optimizing the safety of IVF outcomes. Enhancing embryo selection, and thus implantation rates, to the point that routine eSET is a viable option is a great step forward in improving the safety of ART.
In conclusion, we agree that, as was indicated in the manuscript, patients are interested in babies and not implantations. The sustained implantation rate is about babies. The argument that we should entirely abandon sustained implantation rate as an outcome parameter runs the risk of throwing the baby out with the bathwater. Further, the debate of which single outcome parameter should be utilized is ultimately based upon a faulty premise. Improvements in ART are complex and it is unlikely that a single outcome parameter will provide clinicians and investigators with the information needed to evaluate and implement interventions which will improve the rate of healthy offspring for our patients.
Affiliation 1 has been corrected.