False positives and mosaicism in PGT
The true rate of false-positive diagnosis, when a chromosomally normal embryo is mistakenly identified as abnormal, is unknown. It can be as high as 30% according to some published reports. The rate of so-called mosaicism in a properly run PGT testing laboratory must be below 3%.
Thanks to reliable embryo culture to the blastocyst stage, laser-assisted biopsy, vitrification, and advances in testing, PGT is becoming a standard of care. Still, it is important to appreciate that PGT is not diagnostic testing and is not regulated. Diagnostic testing standards and regulations are only applicable to specimens taken from humans. PGT is performed on a biopsy sample taken from a preimplantation embryo at a stage when by law is not considered a human being. Therefore, the diagnostic testing standards do not apply to PGT. It is merely a non-conclusive screening of embryos for numerical chromosomal errors with a higher chance to be wrong than diagnostic testing. On the face of it, the PGT concept seems straightforward: most of the failures to establish a pregnancy are due to chromosomal errors in embryos. It stands to reason, that weeding out chromosomally abnormal embryos can improve the chance of pregnancy.
Yet, the specificity of PGT has been called into question by an apparently high rate of false-positive results, threatening to undermine the entire premise of chromosomal screening of preimplantation embryos. The argument can be made that the conventional use of the term “false-positive” is not applicable to PGS. This is because within the conventional laboratory testing framework, “false-positive” implies an error in testing the sample. It is argued that with PGS, there are virtually no errors in the samples testing, instead, the sample simply does not represent the entire embryo.
However, the purpose of PGT is not to determine whether the sample collected for testing is normal, but to identify viable embryos. Therefore, whether the error is due to the properties of the embryos (i.e. so-called mosaicism), embryo biopsy, or to problems in the analytical phase, it has to be referred to as “false-positive” whenever a viable embryo is incorrectly identified as non-viable.
Embryonic mosaicism
To date, false-positive errors have been almost solely attributed to the so-called mosaicism, which in its own turn has been attributed to the intrinsic properties of the embryos, inadequate stimulation, or IVF laboratory protocols. This has been inferred from several studies of blastocysts diagnosed as mosaics and recently a large degree of discrepancy in the level of “mosaicism” between different IVF clinics. However, the reported rate of mosaicism is not reconcilable with the rate of mosaicism found in embryos post-implantation. Furthermore, a review by Cabalbo and co-authors that the rate of true (consequential for development) mosaicism in preimplantation embryos is not a major contributing factor to false-positive results.
In a well-run PGT testing laboratory, the rate of mosaicism must be below 3% and the rate of trisomies and monosomies has to be about the same. The prevalence of monosomies indicates a problem at some stage of the testing.
Below is an example, demonstrating a cumulative rate of mosaicism of 2.41% from Progenesis. This is what would be expected from a reputable PGT laboratory (I have no conflict of interest to declare here).
Embryo biopsy
Small number of cells collected during embryo biopsy has always been a limiting factor in PGS specificity. Moreover, it is usually not-known how many nucleated cells have been provided for testing. This makes it very difficult to interpret the significance of the level of mosaicism seen after amplification and can therefore become a source for false-positive result. Since no standards of training for embryo biopsy exists, it is very likely that the number of cells provided for testing varies widely from one IVF laboratory to another and may be responsible for observed variation in embryo mosaicism between IVF laboratories as well as for a false impression of a high preponderancy of “mosaicism” in general.
Analytical phase of PGS
Since a very small number of embryos diagnosed as abnormal is retested or transferred, the direct assessment of errors in analytical phase is impractical. However, an indirect assessment can be performed by comparing the ratio of trisomy to monosomy between PGS reference laboratories. Such assessment is completely independent of mosaicism, which would not produce a preference for a chromosome loss or gains. Furthermore, since each PGS laboratory has a large number of IVF laboratories supplying the samples, this cancels out any differences in patients population, stimulation, culture or biopsy techniques.
Another consideration used in EPC’s assessment’s design was that the probability of errors is independent for each chromosomal pair and therefore the trisomy / monosomy ratio must be analyzed separately for each individual chromosomal pair. Based on the available data the largest discrepancy was found for chromosomes 22 (most common in preimplantation embryos) and chromosome 9.
Gain or loss of chromosome 22 found during PGS testing
PGS testing laboratory 1 | PGS testing laboratory 2 | PGS testing laboratory 3 | |
Gain | 8% | 2.7% | 6% |
Loss | 3.6% | 4.6% | 7% |
Ratio | 2.2 | 0.6 | 0.85 |
p | < 0.0001 | <0.0001 | N/S |
Gain or loss of chromosome 9 found during PGS testing
PGS testing laboratory 1 | PGS testing laboratory 2 | PGS testing laboratory 3 | |
Gain | 0.76% | 1.2% | 1.4% |
Loss | 1.1% | 0.6% | 0.9% |
Ratio | 0.69 | 2.0 | 1.6 |
p | 0.002 | 0.002 | 0.002 |
The data above indicate that the trisomies to monosomies ratio is reversed for the laboratories in opposite directions for different chromosomal pairs. The trends are small for each individual chromosomal pair, but they add up to a non-negligible level, when combined together. Thus, our analysis points to the analytical phase of PGS as a potential source of significant errors.
Conclusion
The sensitivity of PGS is not in doubts. This means that the chance that an abnormal embryo will be reported as normal is very low. It is a concern that the lack of standards in embryo biopsy and some type of analytical error are contributing to false-positive results. In other words, a non-negligible number of normal embryos are reported as abnormal, thereby reducing value of preimplantation chromosomal screening.