Here we go again with PGT-A: The next promised revolution, PGS 4.0, but only if you believe it

Pursuing evidence that examining embryos prior to transfer for chromosomal abnormalities and excluding from transfer those demonstrated to be chromosomal abnormal has now been going on for over 20 years. Yet, in over two decades, not a single properly conducted study has been able to demonstrate any significant outcome benefit for IVF from pursuing this hypothesis. In the process, the procedure has been completely revamped three times (PGS 1.0 through PGS 3.0), the last time in July of 2016, at which time it also changed names from preimplantation genetic screening (PGS) to preimplantation genetic testing for aneuploidy (PGT-A).

A brief history of PGS/PGT-A

Image courtesy of Markus Spiske via Unsplash

The two changes from the original version of PGS 1.0 were not “voluntary” in the sense of a testing procedure being improved in an attempt to improve its accuracy. Instead, changes were radical, having little in common with what was done before but the intent of testing embryos for chromosomal abnormalities prior to transfer. They were introduced by the “testing industry” because they had run out of excuses why the promised benefits of the original test were not being observed: The switch form PGS 1.0 to PGS 2.0 occurred in 2007 when Dutch colleagues published a prospectively randomized study demonstrating that PGS 1.0 not only did not improve pregnancy and live birth rates but, actually, in older women reduced those and, a year later in 2008 the American Society for Assisted Reproductive Medicine (ASRM) published a first formal opinion, declaring the procedure as ineffective in reaching the goals of improving pregnancy and live birth rates and reducing miscarriages.

The two changes from the original version of PGS 1.0 were not “voluntary” in the sense of a testing procedure being improved in an attempt to improve its accuracy. Instead, changes were radical, having little in common with what was done before but the intent of testing embryos for chromosomal abnormalities prior to transfer. They were introduced by the “testing industry” because they had run out of excuses why the promised benefits of the original test were not being observed: The switch form PGS 1.0 to PGS 2.0 occurred in 2007 when Dutch colleagues published a prospectively randomized study demonstrating that PGS 1.0 not only did not improve pregnancy and live birth rates but, actually, in older women reduced those and, a year later in 2008 the American Society for Assisted Reproductive Medicine (ASRM) published a first formal opinion, declaring the procedure as ineffective in reaching the goals of improving pregnancy and live birth rates and reducing miscarriages.

This first switch moved the procedure for biopsying embryos on day 3 after fertilization (cleavage stage) to biopsy days 5 or 6 (blastocyst stage). It was accompanied by availability of better testing platforms that allowed for testing of all 46 chromosomes, while PGS 1.0 allowed only testing of a limited number of chromosomes, known to be the most frequent aneuploid (chromosomally abnormal) embryos. These obvious technical improvements in diagnostic tools allowed the testing industry to continue representations that the procedure, finally, would prove its ability to improve IVF outcomes.

That these claims for PGS 2.0 were just as false as prior claims made for PGS 1.0 became increasingly obvious, as in 2015 CHR investigators and others started reporting normal births after transfer of embryos that had been falsely diagnosed as chromosomally “abnormal.” These normal births demonstrated indisputably that PGS 2.0 resulted in significant false positives in diagnoses of embryos, leading to the discarding of a large number of embryos with high pregnancy potential. With skepticism about the procedure reaching again new heights, the testing industry in July of 2016 announced the most radical changes up to that point in testing of human embryos for chromosomal abnormalities, thereby creating PGS 3.0, which represents the testing procedure currently in use worldwide.

PGS 3.0 with NGS

The switch to PGS 3.0 affected every aspect of the procedure, from how the testing was done (only next generation sequencing platforms, NGS, were considered adequate since none of the other platforms used up to that point, the public was suddenly informed, were able to detect more than one cell line in a biopsy sample, thereby allowing for the diagnosis of “mosaicism” in a biopsy), how it was reported (instead of reporting embryos as “euploid-normal” or “aneuploid-abnormal,” they now were reportred as “euploid-normal,” “mosaic,” or “aneuploid-abnormal.” The switch to PGS 3.0, and the formal guidelines published in 2016 by the Preimplantation Genetic Diagnosis International Society (PGDIS) (clearly an interest group of the genetic testing industry) even offered clinical recommendations to physicians how to treat their patients based on the reported results.

Once again, the genetic testing industry claimed that PGS 3.0 would, finally, be able to prove that the procedure was, indeed, capable of improving IVF outcomes, as originally proposed by the PGS hypothesis, but this, again, did not happen. Continuing research for the first time, indeed, established what some opponents of the procedure, with CHR’s investigators leading the charge, had claimed for some time: PGS/PGT-A, for biological reasons, simply, could not work.

Prediction of PGS 4/.0

CHR investigators, for some time, had predicted that the genetic testing industry, which found itself once again threatened with loss of hundreds of millions of dollars in revenue worldwide, would resort to the by now only too familiar method of, simply, inventing yet another radical change in performing embryo diagnosis and creating yet another radically new method of embryo testing (“PGS 4.0”). They, indeed, hypothesized that the next step in the evolution of PGS/PGT-A would be the so-called non-invasive testing of embryos. Avoiding embryo biopsy at blastocyst stage, testing would be done on the spent media in which embryos were cultured for the so-called cell-free DNA, which the dying cells, undergoing apoptosis, secrete into the spent media.

CHR’s investigators expected this PGS 4.0 for good reasons: One important reason was a paper by a past president of ASRM and professor at the University of Southern California at Los Angeles (USCLA) who in 2017 suggested that a big reason for the high number of false positive diagnoses with PGS/PGT-A was the damage caused by the embryo biopsy itself, which adversely affected pregnancy chances of embryos. Considering how sensitive human embryos are to any form of manipulation, this is a logical conclusion; however, CHR’s investigators never believed it to be of as much importance as suggested by the colleague’s paper. Instead, they since ca. 2005 had been convinced that the major reason for the failure of PGS/PGT-A were not technical but biological realities of early-stage embryos.

That this, indeed, is the likely dominant reason for the more than 20 years of failures of PGS/PGT-A is coming increasingly into focus: CHR investigators reported two years ago that a single embryo biopsy of 5-7 trophectoderm cells cannot mathematically define the ploidy of a whole embryo. Most importantly, however, it is increasingly becoming obvious that day 5/6 blastocyst stage embryos represent the peak of embryo aneuploidy. At this stage, many embryos have, mostly mitotic, aneuploidies which are self-corrected downstream from days 5/6 onward. If a significant portion of aneuploid embryos self-correct post-blastocyst stage, what then would be the purpose of diagnosing embryos’ ploidy at blastocyst stage?

This is the main reason why it is so surprising that a very prominent scientific journal just very recently accepted and published a study, claiming better accuracy of embryo diagnosis by using cell-free DNA in spent media than from embryo biopsy. A small number of papers, reporting on the use of cell-free DNA, had been published already over the last year, and were the second reason why CHR investigators expected this methodology to become the “next big thing” of the genetic testing industry. These papers reported mixed results, though one prominent IVF center in Colorado already announced in a press release a number of months ago that it was offering this technology for clinical use. The following section describes the recently published paper in more detail and why its conclusions are faulty.

A proposal for PGS 4.0: Huang et al., Proc Natl Accad Science doi.10.1073.pnas.1907472116

We noted above that embryo diagnosis by measuring the so-called cell-free DNA at blastocyst stage in place of embryo biopsy has been proposed before, even though results of a small number of studies diverged in very significant ways. Now one of the world’s most prestigious science journals, the Proceedings of the National Academy of Sciences (PNAS), accepted and published a study from Boston with the senior author being Catherine Racowsky, a Harvard PhD slated to be the next year’s President of the ASRM. In reality, most of the scientific work was, however, done outside Harvard University at different labs at Peking University in Beijing and by a Chinese start-up company, Yikon Genomics Company, Ltd, in Shanghai, China. This manuscript, therefore, must be understood as what it very obviously is: yet another effort by the genetic testing industry to reshape PGS/PGT-A.

In the paper, the investigators studied 52 human blastocyst-stage embryos that, previously, had been diagnosed as “chromosomally abnormal” and, therefore, became available for research purposes because they had been rejected from transfer. Those embryos were thawed out and cultured for 24 hours. The spent media from this 24-hour culture was then used to test for cell-free aneuploid DNA under the assumption that an embryo with chromosomally abnormal cells, whether in trophectoderm (TE, which becomes the placenta) or inner cell mass (ICM which becomes the fetus), “leaks” cell-free DNA into spent media. The study then compared the “chromosomally abnormal” results of the first embryo assessment by trophectoderm biopsy (TEB), which had led to the exclusion of embryos from transfer, to the results obtained from analyzing the above-noted 24-hour spent media, using an “adjusting” formula the investigators used, based on assumed differences in “leakage” between TE and ICM. Both diagnostic methods were compared to genomic analysis of the whole embryo.

The final conclusion of the study was that the non-invasive cell-free DNA testing was more frequently compatible with the genomic analysis of the whole embryo than TEB. Unfortunately, this conclusion is not supported by the data they present in their manuscript. The authors acknowledge that the study involves only a relatively small number of embryos (n=52), but that is not the principal problem. Those involve technical as well as biological oversights and/or transgressions, starting with the design of the study, unproven assumptions and unvalidated adjustments to data analyses. Here is some detail:

The authors assume that cell-free DNA from both TE and ICM “leaks” into the media in which embryos are cultured. This, however, has not been established and, indeed, is somewhat questionable since TE, the outer layer of cells of the blastocyst stage embryo is, of course, in direct contact with culture medium; the ICM, located inside the blastocyst, however, is not. Moreover, at blastocyst stage, TE consists of in excess of 200 cells, while the ICM represents only approximately 20-25 cells. It, therefore, appears reasonable to assume that TE cells will “leak” more cell-free DNA than ICM cells. In calculating “aneuploid” cell-free DNA as a percentage of total cell-free DNA, the authors, however, assumed more profound “leakage” from cells of the ICM, based on the observation that in the mouse the ICM has been demonstrated to undergo more apoptosis than cells of the TE (an important point we will come back to, since apoptosis is the process that removes the debris of degenerating cells, an important step when chromosomally abnormal embryos self-correct). The assumptions made by the authors to calculate the results of their study are, therefore, at best unvalidated and, more likely, outright false.

More importantly, however, while using the above-noted mouse data, which has since also been confirmed in human embryos (though still unpublished), the authors ignore the main observation of the original mouse study that led to the discovery that apoptosis in ICM was significantly more profound than in TE. That crucially important finding was that aneuploidy within the ICM in a very high percentage self-corrects via apoptosis of chromosomally abnormal cells downstream from blastocyst-stage. In other words, cells of the ICM have much more ability to self-correct than cells of the TE, a finding in still unpublished data now also confirmed in human embryos and artificial embryos made from stem cells.

These biological realities in early stage embryos, left unmentioned by the authors of the study, raise two obvious questions: First, as noted above, why would anybody assume that these biological realities favor ICM leakage into spent media and, even more obviously, why would anybody suggest any test of embryos at blastocyst stage, whether by TEB or cell-free DNA from spent media, if a high percentage of those embryos still self-correct downstream from blastocyst stage? Even technically accurate diagnoses of aneuploidy at blastocyst stage in such cases would become false positive diagnoses. It appears curious that the authors of this manuscript specifically referred to the mouse study as crucial for their statistical evaluation of results but completely ignored the mouse study’s main conclusion, which for the first time demonstrated how profound self-correction really is in embryos post-blastocyst stage.

Beyond these conceptual and biological main problems, this study also demonstrates additional shortcomings:

The study utilized as diagnostic platform next generation sequencing (NGS) because it is currently the only platform capable of diagnosing more than one cell line (i.e., euploid as well as aneuploid cell lines in a single specimen). However, even current NGS is only able to detect a second aneuploid DNA at concentrations over 20%. In other words, low-grade mosaicism (<20%) will be missed, a fact not acknowledged by the authors.
The 52 embryos investigated in this study were found to be chromosomally “abnormal” by various diagnostic platforms, many using technologies other than NGS. That, likely, means highly unreliable diagnoses and should prohibit any reference to those original diagnoses (unless performed by NGS).

All 52 embryos were thawed and again cultured for a full 24 hours. The spent media from these 24 hours of culture were used to measure cell-free DNA. Since embryos were day-5/6 embryos after fertilization when cryopreserved, they were day-6/7 embryos when cultured in the investigated media (i.e., they were a full day older than blastocysts usually are at time of diagnoses).

If embryos are cultured beyond days 5/6, embryos may already hatch, which may significantly affect leakage.

Single cell DNAsic studies have demonstrated that aneuploidy in embryos peaks on days 5/6 after fertilization and then very quickly declines because of the above-noted self-correction. A delay in embryo diagnosis by one day may, therefore, end up assessing embryos at different levels of aneuploidy. Comparisons between day -5/6 and day-6/7 embryos may, therefore, not be appropriate, nor may be the finding in spent media of embryos that are a day older.

In summary, the recent PNAS paper by Huang et al may very well indicate the worldwide initiation of PGS 4.0. Anybody who, however, one more time falls for the concept of PGS/PGT-A and believes that any test on a blastocyst-stage embryo can accurately determine whether an embryo can be transferred or should be disposed of, definitely should seek further advice before spending the money and making the required efforts to have embryos tested for chromosomal abnormalities.

This is a part of the July 2019 issue of the CHR VOICE.

Norbert Gleicher, MD, leads CHR’s clinical and research efforts as Medical Director and Chief Scientist. A world-renowned reproductive endocrinologist, Dr. Gleicher has published hundreds of peer-reviewed papers and lectured globally while keeping an active clinical career focused on ovarian aging, immunological issues and other difficult cases of infertility.