Limits imposed by the experimental design of a large prospective non-inferiority study on PGT-A invalidate many of the conclusions

Abstract

The New England Journal of Medicine recently published a large study addressing the efficacy of preimplantation genetic testing for aneuploidy (PGT-A). The 14-centre randomized control non-inferiority trial used cumulative live birth rate (CLBR) as a clinical endpoint to examine the value of PGT-A and concluded that conventional IVF was not inferior to IVF with PGT-A. Unfortunately, the experimental design was highly flawed; and in fact, the data generated in the study do not support the major conclusions presented in the publication. The embryos in each patient's three-embryo pool, which were available for transfer, were selected solely by morphology. The investigators then randomized patients to either the PGT-A group or the control group. It is important to note that PGT-A screening in the study group was done only after the embryos were selected. PGT-A was not really used in a meaningful way, which would have been for the PGT-A results to help in selecting which embryos would be in the three-embryo group. Thus, the outcomes were wholly determined prior to the study intervention. The ultimate delivery rate for each group of three embryos was determined when they were selected by morphology. The randomization, which occurred after embryo selection, would assure equal distribution of those cohorts destined to deliver and those destined to fail to the two study groups, the PGT-A and control groups. Thus, there was no potential for PGT-A to enhance selection and thus no possible way to improve the cumulative outcomes. Since there was no possible way for the control group to be inferior, the experimental design precluded any chance of evaluating the primary endpoint of the study. The primary question of the study was never evaluated. Another serious flaw was that the study was initiated prior to knowing how to interpret the data provided in the PGT-A analytical result. Specifically, the design excluded mosaic embryos from transfer despite the literature demonstrating the significant reproductive potential for these embryos. When accounting for the lost deliveries induced by this non-evidence-based decision, the expected delivery rates in the two groups become virtually identical. That is an important issue because the data from the study actually demonstrate the safety of PGT-A without diminution in outcomes from the impact of trophectoderm biopsy or the discarding of competent embryos which had wrongfully been considered aneuploid. A final serious flaw in the experimental design and interpretation of the data surrounding the issue of the miscarriage rate. The investigators noted that the miscarriage rate was lower in the PGT-A group but stated that its impact was insufficient to alter the CLBR. Of course, by design, the CLBRs were limited to being equivalent. There was no potential for enhanced outcomes in the PGT-A group and thus no possibility that the lower risk of miscarriage in the PGT-A group would raise the CLBR. The benefit of a lower miscarriage rate is real and significant. Its relevance should not be diminished based on the lack of a change in the CLBR since that was never possible in this study. The investigators of the study concluded that the CLBR with conventional ART is equivalent to that with PGT-A, but a simple review of the experiment reassigns their genuine findings to those of a safety study. Significantly, the data in the study demonstrate that the intervention of PGT-A is safe. This study neither supports nor refutes the efficacy of clinical PGT-A.

Keywords: IVF; cumulative live birth rate; embryo diagnostic; non-inferiority trial; preimplantation genetic testing for aneuploidy; sustained implantation rate.