Effectiveness of in vitro fertilization with preimplantation genetic screening A reanalysis of United States assisted reproductive technology data 2011 2012

Authors:
Vitaly A. Kushnir, M.D., Sarah K. Darmon, Ph.D., David F. Albertini, Ph.D., David H. Barad, M.D., Norbert Gleicher, M.D.

Abstract:

Objective:
To assess effectiveness of preimplantation genetic screening (PGS) in fresh IVF cycles.

Design:
Reanalysis of retrospective US national data.

Setting:
Not applicable.

Patient(s):
A total of 5,471 fresh autologous IVF cycles with PGS and 97,069 cycles without PGS, reported in 2011–2012 to the Centers for Disease Control and Prevention.

Intervention(s):
Not applicable.

Main Outcome Measure(s):
Cycles that reached ET, miscarriage rates, live birth rates per cycle and per transfer.

Result(s):
More PGS than non-PGS cycles reached ET (64.2% vs. 62.3%), suggesting favorable patient selection bias for patients using PGS. Nevertheless, live births rates per cycle start (25.2% vs. 28.8%) and per ET (39.3% vs. 46.2%) were significantly better in non-PGS cycles, whereas miscarriage rates were similar (13.7% vs. 13.9%). With a maternal age >37 years significantly more cycles in the PGS group reached ET (53.1% vs. 41.9%), suggesting a significant selection bias for more favorable patients in the PGS population. This bias rather than the PGS procedure may partially explain the observed improved live birth rate per cycle (17.7% vs. 12.7%) and lower miscarriage rate (16.8% vs. 26.0%) in the older PGS group.

Conclusion(s):
Overall, PGS decreased chances of live birth in association with IVF. National improvements in live birth and miscarriage rates reported with PGS in older women are likely the consequence of favorable patient selection biases.

  • Vitaly Kushnir

    N. Gleicher, D.H. Barad, V.A. Kushnir

    We sincerely appreciate the repeated interest of Grifo and associates in our PGS work but repeating incorrect facts over and over (and referencing their own writings in support) does not change the reality that in over 10 years of two PGS generations, contrary to their representation, not a single properly designed study of PGS demonstrated any IVF outcome benefit. The alleged prospectively randomized studies they quote as demonstrating efficacy of PGS are, as we in detail documented Gleicher et al (2014), so badly flawed in design, that it is difficult to understand that Grifo and associates still quote these studies in support of their positions. Scott et al (2012) and Forman et al (2013) employ inadequate patient selection and outcome assessments with reference point embryo transfer, thus excluding all patients who do not reach embryo transfer, and Yang et al (2012), to their credit, themselves note the preliminary nature of their study because of small patient numbers and selection of only good prognosis patients.

    It also does not change the fact that our manuscript very clearly demonstrates that the initial analysis of PGS effects by CDC colleagues was mistaken. Grifo et al criticize our concentration on intent to treat analysis and question the method of our analysis without pointing out where our calculation went wrong. Yet, is there really anybody who does not believe that IVF outcomes should be presented by intent to treat (i.e., with reference point cycle start)?

    It is also important to point out that data are currently in the publications pipeline which, using national IVF registry outcomes, offer evidence that even in donor/recipient cycles (i.e. best outcome cycles) PGS actually significantly negatively affects IVF outcomes.

    We, however, want to direct the interested reader in this subject to a very important and very elegant recent publication in Nature Communications (Bolton et al., Mouse model of chromosome mosaicism reveals lineage-specific depletion of aneuploid cells and normal developmental potential. 2016;7:11165). Since aneuploid cells highly preferentially accumulate in the trophectoderm rather than the inner cell mass of embryos, their study demonstrates why PGS for biological reasons simply cannot work. A single biopsy of trophectoderm cannot accurately determine whether any single biopsied cell island in the trophectoderm really reflects the chromosomal make up of the inner cell mass. In other words, because of much higher mosaicism in trophectoderm than has been previously reported, it is impossible to say whether a single biopsy reflects the embryo or segregated chromosomally abnormal cells. If segregated cells are biopsied, the biopsy will give a false positive result and biologically normal embryos will be discarded.

    That this happening has been demonstrated by our group in collaboration with other centers from the International PGS Consortium in initially reporting 3 healthy births after transfer of allegedly aneuploid embryos in 5 cases (Gleicher et al, ASRM 2015). By now this number has increased to 5/8. Shortly after our initial report Greco et al (NEJM, 2016) reported 6 chromosomally normal births in 18 attempts of embryo transfer with “mosaic” embryos in Italy. Finally, we also want to point out that a recent manuscript by Orvieto et al (2016) demonstrated significant discrepancies between trophectoderm and inner cell mass biopsies of human embryos, and we are aware of at least one additional manuscript from the International PGS Consortium that demonstrates an approximately 50% divergence between multiple trophectoderm biopsies in same embryos.

    In summary, we strongly suggest that Grifo and associates and other promoters of PGS carefully consider the adverse effects PGS appears to have on their patients’ IVF cycle outcomes. These effects may be particularly severe in poor prognosis patients who have few embryos, and where PGS may deprive them of their last pregnancy chances by discarding potentially viable embryos. Considering the growing doubts about the utility of PGS, one, indeed, has to wonder why the FDA has not yet focused on the increasing utilization of the various PGS techniques in IVF.

  • Jamie Grifo MD PhD

    Grifo J D, Munné S, McCullough D, Wells, D

    When setting the context for the discussion of new data it is
    important to consider information from recent studies, using current
    technologies. Consequently, we were surprised to see that the paper by Kushnir
    et al, concerning the effectiveness of IVF with preimplantation genetic
    screening, neglected to mention that there are three peer-reviewed clinical
    randomized studies showing improved ongoing pregnancy rates using modern PGS
    techniques (Scott et al. 2012, Yang et al. 2012, Forman et al. 2013).
    Additionally, there are metananalyses and systematic reviews that also point to
    potential benefits of PGS employing current methods (Lee et al. 2014, Chen et
    al. 2015, Dahdouh et al. 2015). Ignoring these publications, the authors instead
    cite a single, highly criticized study (Mastenbroek et al. 2007) as evidence
    that PGS is ineffective. The study in question, as with all studies that have
    failed to show improved outcomes using PGS, employed suboptimal day-3 biopsy
    and inferior FISH techniques (Munné et al. 2007a,b, Cohen and Grifo 2007,
    Simpson et al. 2008). Indeed, so poor was the cleavage stage biopsy technique
    in the quoted paper (Mastenbroek et al. 2007) that the damage induced is
    believed to have led to implantation rates being reduced by ~50% (Munné et al.
    2007a,b). Current PGS methods, considered standard throughout the world,
    utilize blastocyst biopsy, which has been shown not to be detrimental (Scott et
    al. 2013). By the unfortunate choice of citations, the authors give the
    impression that modern PGS strategies, using blastocyst biopsy and
    comprehensive chromosome screening methods, are no different from techniques considered obsolete for the best part of a decade. They state that ‘The recently published study from the CDC appeared to be counterintuitive and contradictory to the
    published literature’. However, the reality is that the findings are not
    contradictory to any data from well controlled studies published in the last
    five years. On the contrary, it mirrors expectations based upon published
    randomized trials using modern PGS technologies.

    The main theme of the paper by Kushnir
    and colleagues is a re-analysis of the data from the Centers for Disease
    Control, recently presented by Chang et al (2016) that concluded that the use
    of PGS led to reductions in the risk of miscarriage and increased odds of
    clinical pregnancy and live birth in specific patient groups, particularly
    those with a female age >37. We certainly agree that the Chang et al. 2016
    paper is imperfect and has some ascertainment biases, discussed below, and concur
    that conclusions drawn from RCTs are preferable whenever it is appropriate and
    feasible to carry out such studies. However, we would argue that the biases in
    the analysis are even more accentuated in the Kushnir et al. article, that
    there have been some fundamental misinterpretations of the data and a misunderstanding of the way in which the information was collected. Together these have led the authors to erroneous conclusions.
    The authors suggest that false
    positives caused by mosaicism may result in potentially viable embryos being
    excluded from transfer. This is certainly a risk if PGS is performed at the
    cleavage stage, a time when the proportion of embryos displaying mosaicism is
    at its highest and when cytogenetic testing is limited to a single cell. The
    risk is lower if testing is carried out at the blastocyst stage, when mosaicism
    is less prevalent, and falls lower still if high-resolution next-generation
    sequencing (hr-NGS) is utilized. High–resolution NGS provides a means for
    embryos with mosaicism in the trophectoderm biopsy specimen to be detected. We have proposed that mosaic embryos should be classified as such and could be
    considered for transfer in some cases (Munné et al. 2016). Embryos with mosaic
    trophectoderm biopsies are associated lower implantation potential than euploid
    embryos and a higher risk of miscarriage. Consequently, it is advisable to give
    them a lower priority for transfer, but if no euploid embryo is available they
    can be replaced with patient consent following thorough genetic counseling.

    Kushnir and colleagues agree that review of the CDC data reveals
    evidence for an improved birth rate per cycle and a lower miscarriage rate in
    the older PGS group. However, they cannot bring themselves to give PGS the
    credit for this and instead insist that it must be due to patient selection. They
    point out that the PGS patients in the CDC dataset were more likely to be
    multiparous, have more prior deliveries, have undergone more IVF cycles, have
    more miscarriages, and produce more oocytes. They suggest that this is evidence
    of a selection bias. Indeed, in the final paragraph of the paper they assert that
    the PGS group contained a disproportionate number of good prognosis patients as
    though this were an irrefutable fact. Once again their reasoning is flawed. That
    PGS patients would have had more previous pregnancy losses and more IVF cycles is
    entirely in-line with expectation, as recurrent miscarriage and multiple
    implantation failures are common reasons for referral for PGS. Furthermore, if
    the patients had more IVF cycles, they will have had more opportunities for a
    pregnancy and a delivery, so it is not surprising to see increases in these
    areas. In many cases, PGS patients will have had chromosome screening in a
    previous cycle too, which might also contribute to higher numbers of births in
    this patient group. The authors noted that ’Patients using PGS also produced
    marginally more oocytes‘. Again, this is not surprising. Patients receiving PGS
    are often poorer prognosis, with high levels of aneuploidy anticipated in their
    embryos. Knowing that there is an increased likelihood that some embryos will
    be discarded due to adverse genetic results, many clinics stimulate their PGS
    patients more aggressively in the hope of producing at least one euploid
    embryo.

    Interestingly, the authors do not entertain the possibility of ‘patient
    selection bias’ when concluding that outcomes for young PGS patients (<35
    years of age) are poorer than non-PGS patients, but instead accept this at face
    value. If any patient selection bias exists it is in this younger group and
    would actually favor the non-PGS group rather than the cycles involving PGS.
    Although in recent years it has become increasingly common to apply PGS to good
    prognosis patients as a means of enhancing embryo selection, at the time that
    the CDC data was collected it was extremely rare for PGS to be applied to such
    patients. In fact, PGS was largely reserved for patients considered to have a low
    chances of achieving a pregnancy, often being of advanced reproductive age,
    combined with a history of previous miscarriages and/or failed IVF cycles.
    Young patients offered PGS in 2011-2012 tended to fall into two groups. Most
    were offered PGS because previous treatment cycles had been challenging and/or
    they were considered to have an unusually poor prognosis for their age. A second
    group of patients were those having PGS because they wanted to determine the
    sex of their child. In the latter case, half of the viable embryos are usually excluded
    as they are not the sex preferred by the patients. Inevitably, this will have a
    negative impact on the likelihood of a pregnancy. Furthermore, patients having
    sex selection in 2011-2012 often only had a very limited number of chromosomes
    tested (often only 21, X and Y), providing very little useful information in
    terms of assessing aneuploidy. Consequently, we agree that it is very likely that
    differences exist between the groups of patient featured in the CDC data, but it
    does not favor the PGS group as insisted by Kushnir and colleagues, on the
    contrary.

    The CDC data clearly shows that miscarriage rate is essentially
    flat with age in the patient group receiving PGS. This is concordant with other
    studies examining the effect of transferring euploid embryos following PGS
    (Harton et al. 2013). Conversely, the non-PGS group shows an accelerating
    increase in miscarriage rate with age (more than doubling from the 37 age groups). This is as expected for a population receiving no form
    of chromosome screening. Kushnir et al suggest that PGS might prevent one
    miscarriage for every 11 fresh IVF cycles and brush this off as though it were
    nothing. Their calculation is debatable and contradicts recent data showing that
    pregnancy loss rates reduce to 30% in patients over 40 without PGS). Nonetheless, even if they were
    correct, avoiding this number of miscarriages would still seem to be a benefit
    worth having.

    Probably the biggest flaw in the
    assumptions made by Kushnir and co-workers is their analysis of the data on an ‘intention
    to treat’ basis. The data used was not collected in a manner that would permit such
    an analysis to be undertaken. To get around this problem, the authors make
    assumptions so that they can try and back calculate the intention to treat.
    However, no proof is provided that this assumption is correct. That is
    unjustifiable. It is unclear how many cycles they assumed no euploid with no embryo transfer under the guise of freeze all, thus delaying transfer of euploid embryos. [DW2] The authors assume that no transfer meant no normal embryos, but show
    no data to prove it. The paper rest on this data being correct and yet it is
    not verified. Indeed, as the assumptions are clearly flawed, all of the
    conclusions of the paper become highly questionable. Errors in data analysis may
    be one of the reasons that the analysis undertaken by Kushnir and colleagues
    fails to corroborate recent RCT data.

    Another point of note, many centers
    using PGS at the time of the CDC data collection, including those centers
    involved in the three successful RCTs, performed blastocyst biopsy and
    vitrification and are not included in the dataset. That was also the practice
    of many of the most experienced PGS centers at that time. This highlights a hazard
    of using a dataset not designed to analyze the question addressed by the RCT
    studies. Eliminating cycles involving blastocyst biopsy and vitrification has
    the effect of enriching the dataset for cycles in which PGS was undertaken
    using inferior cleavage stage biopsy methods. In 2011-2012, at least 60% of all PGS tests were performed in conjunction with day-3 (blastomere) biopsy. Contrast this to approximately 95% of all PGS tests are undertaken at the blastocyst stage today. Furthermore, about one-quarter of PGS analyses conducted in 2011-2012 still involved a limited chromosomal analysis using FISH, whereas virtually all PGS carried out
    today involves comprehensive chromosome screening, associated with much higher
    accuracy and revealing aneuploidy affecting any chromosome. The randomized
    controlled trials demonstrating the benefits of PGS have all focused on
    blastocyst biopsy and comprehensive chromosome analysis.

    In summary, it is our opinion that few of the conclusions of the
    Kushnir et al paper are valid. This is due to errors in the way the data was
    treated, misunderstanding of the way in which PGS was offered in 2011-2012 and
    to whom and various other errors and biases outlined above. When analyzing data
    not specifically collected for the purpose of answering a particular question,
    scientists need to exercise extreme care and, above all, keep an open mind.
    Randomised controlled trials remain the best way for avoiding bias, accidental
    or deliberate. To date, only a limited number of RCTs using modern PGS methods
    have been published, but so far all results have been positive. It is hoped
    that further RCTs, currently nearing completion, will finally provide data on
    the efficacy of PGS that everyone can agree upon.

    References:

    Chang J,
    Boulet SL, Jeng G, et al. (2016). Outcomes of in vitro fertilization with
    preimplantation genetic diagnosis: an analysis of the United States Assisted
    Reproductive Technology Surveillance Data, 2011-2012. Fertil Steril.
    105(2):394-400

    Chen M, Wei
    S, Hu J, Quan S (2015) Can Comprehensive Chromosome Screening Technology
    Improve IVF/ICSI Outcomes? A Meta-Analysis. PLoS One 10:(10)

    Cohen J,
    Grifo J (2007) Multicenter trial of preimplantation genetic screening reported
    in the New England Journal of Medicine: an in-depth look at the findings.
    Reproductive Biomed Online 15: 365-366

    Dahdouh EM,
    Balayla J, García-Velasco JA (2015) Impact of blastocyst biopsy and
    comprehensive chromosome screening technology on preimplantation genetic
    screening: a systematic review of randomized controlled trials. Reprod Biomed
    Online. 30:281-9

    Forman EJ, Hong
    KH, Ferry KM, Tao X, Taylor D, Levy B, Treff NR, Scott RT (2013) In vitro
    fertilization with single euploid blastocyst transfer: a randomized controlled
    trial, Fertil Steril, 100:100-107

    Harton G,
    Munné S, Surrey M, Grifo J, Kaplan B, Griffin DK, Wells D, and PGD
    practitioners group (2013) Diminished effect of maternal age on implantation
    after Preimplantation Genetic Diagnosis with array comparative genomic
    hybridization. Fertil Steril, 100:1695-1703

    Kushnir VA,
    Darmon SK, Albertini DF, Barad DH, Gleicher N (2016) Effectiveness of in vitro
    fertilization with preimplantation genetic screening: a reanalysis of United
    States assisted reproductive technology data 2011–2012. Fertil Steril, in press

    Lee E,
    Illingworth P, Wilton L, Chambers GM. (2014). ‘The clinical effectiveness of
    preimplantation genetic diagnosis for aneuploidy in all 24 chromosomes (PGD-A):
    systematic review’. Hum Reprod. 30(2):473-83.

    Mastenbroek
    S, Twisk M, van der Veen F, Repping S. Preimplantation genetic screening: a
    systematic review and meta-analysis of RCTs. Hum Reprod Update 2011;17:454–66.

    Munné S,
    Cohen J, Simpson JL (2007) In vitro fertilization with preimplantation genetic
    screening. N Engl J Med 357:1769-70

    Munné S,
    Gianaroli L, Tur-Kaspa I, Magli C, Sandalinas M, Grifo J, Cram D, Kahraman S,
    Verlinsky Y, Simpson JL (2007) Sub-standard application of PGS may interfere
    with its clinical success. Fertil Steril, 88:781-784

    Munné S,
    Grifo J, Wells D (2016) Mosaicism: “survival of the fittest” versus “no embryo
    left behind” Fertil Steril in press
    (ttp://dx.doi.org/10.1016/j.fertnstert.2016.01.016)

    Schoolcraft
    WB, Fragouli E, Stevens J, Munné S, Katz-Jaffe MG, Wells D. Clinical
    application of comprehensive chromosomal screening at the blastocyst stage.
    Fertil Steril 2012;94:1700–6

    Scott RT
    Jr., Ferry K, Su J, Tao X, Scott K, Treff NRT. Comprehensive chromosome
    screening is highly predictive of the reproductive potential of human embryos:
    a prospective, blinded, nonselection study. Fertil Steril, 2012, 97:870-875

    Scott RT,
    Upham KM, Forman EJ, Zhao T, Treff NR. Cleavage-stage biopsy significantly
    impairs human embryonic implantation potential while blasto- cyst biopsy does
    not: a randomized and paired clinical trial. Fertil Steril 2013; 100:624–30.

    Simpson
    (2008) The Randomized Clinical Trial in Assessing Preimplantation Genetic
    Screening (PGS): Necessary but not Sufficient. Human Reprod 23: 2179–2181

    Yang Z, Liu
    J, Collins GS, Salem SA, Liu X, Lyle SS, Peck AC, Sills ES, Salem RD. Selection
    of single blastocysts for fresh transfer via standard morphology assessment
    alone and with array CGH for good prognosis IVF patients: results from a randomized
    pilot study. Mol Cytogenet 2012;5:24

    [

Translate »