My eyes boggled earlier this week as I leafed through The Guardian newspaper and came across the story of an Indian woman who had just had her IVF baby boy safely delivered. The woman, Daljinder Kaur, was reported to be 72 years old – her husband, Mohinder Singh Gill, 79. It was their first successful pregnancy in 46 years of marriage. The eggs had been donated of course – nobody’s ovaries perform at that great age – but this septuagenarian uterus had managed to implant an embryo and its owner carry the subsequent fetus successfully to term. And it wasn’t the first geriatric pregnancy achieved by the head of the Test Tube Baby Center in Hisar, Anurag Bishnoi.

Stories like this invite us to believe that there is almost limitless potential for human reproduction, especially when it is assisted by reproductive technologies like IVF. And it is probably not for us to judge on the advisability of two seventy year-olds embarking on child rearing for the first time even if we grimly suspect that endless hours of night nursing and nappy-changing, at that age, will inevitably take their toll. Indeed, a brief overview of our apparently crowded planet – 7.4 billion people and counting – also gives us an inflated picture of human fecundity. But the truth is that we are one of the least fecund species on the planet – as the 46 fruitless years spent by our Indian couple show – and a number of reproductive biologists, informed by evolution, are beginning to unravel the inside story of human fertility to explain why. It involves a staggering amount of genetic abnormality in human early embryos, the huge costs, particularly in humans, of maternal investment in offspring, and a challenge to the tenets of assisted reproduction technology.

It only involved 23 good-quality human embryos, and was rather discreetly published as a technical report in Nature Medicine, say reproductive scientists Jan Brosens and Birgit Gellersen, but it may have changed our understanding of human reproduction for ever. Evelyne Vanneste and her colleagues, from the Catholic University of Leuven, had set out to try and understand why no amount of pre-implantation genetic screening of embryos seems able to improve the success rate of IVF, which languishes between 20% and 30%, even though that screening results in a large percentage rejection of embryos because they appear to be genetically abnormal. Such embryos are traditionally thought likely to be indolent and biologically inert. They developed two new array-based methods to explore the genomes of embryo cells in minute detail, looking for any tell-tale departures from the normal diploid condition, and determined a baseline frequency for abnormality from 23 embryos donated by perfectly healthy young women who were undergoing IVF for genetic reasons not related to fertility. To their astonishment they uncovered an extraordinary amount of genetic abnormality among these embryos. Only 2 embryos were found to be genetically normal. All the rest – over 90% – contained some or all embryo cells (blastomeres) with significant forms of genetic abnormality, including loss of chromosome arms or whole chromosomes, uniparental disomy (where the genetic contribution from one parent is entirely lacking), to a whole mishmash of chromosomal deletions, amplifications, fragmentations and duplications. Half the embryos had no normal diploid cells at all. In the absence of in vivo early embryos and embryos subject to early spontaneous abortion, these results, says Vanneste, are the best evidence we have for what constitutes normal human embryogenesis. The only intriguing clue Vanneste planted in her discussion was her observation that while IVF success rate in her clinic was comparable to centers worldwide at 20% per embryo transferred, the number of mosaic embryos, containing a mix of genetically normal and abnormal blastomeres, was much higher at about 40%. So it appeared that some mosaic embryos could go on to make normal babies.

This discrepancy was seized upon by Jan Brosens, head of reproductive health at the University of Warwick, and he expressed it in a slightly different way. It is known, he says, that pregnancy loss in humans runs at about 70% – due to implantation failure, spontaneous early miscarriage and later clinical miscarriage. But, although this is extremely high, it falls far short of the percentage of genetically chaotic embryos, which is, according to Vanneste, approximately 90%. So, again, the number of perfectly healthy live births greatly exceeds the number of perfectly normal embryos. There is not very much comparative mammalian research on embryo genetic quality but it is known, for instance, that the mouse has very low levels of abnormality while the human level is very high. This head-scratcher immediately sets up a number of questions:

  1. Why are mice and humans so very different in this respect?
  2. If some genetically aberrant embryos can go on to make healthy babies why do they go through a period of genetic instability in the first place?
  3. How do genetically abnormal embryos develop into healthy babies? Do they correct these genetic mistakes within some of their blastomeres or do they simply jettison abnormal blastomeres, after implantation, leaving only normal blastomeres to go on to make the baby?

The answer to question 1 probably lies in profound differences between mouse and human reproduction. As Brosens points out, mouse reproductive success is based on quantity, characterized by rapid breeding cycles, multiple synchronous implantations, large litter size, and, crucially, huge natural selection among offspring. Mouse offspring “quality” is arrived at mainly through sibling rivalry after birth, whereas in humans, typified by serial singleton pregnancies, long gestation times, and very deep and invasive placentation, “quality” is more likely achieved at outset through evolved forms of cryptic female choice over which sperm will fertilize an egg, and which embryo will be allowed to implant. This is in order to protect maternal investment against perceived low quality or otherwise incompatible embryos.

This leads us on nicely to question 2. Brosens, and his collaborator Nick Macklon, from the University of Southampton, equate genetic abnormality with invasiveness. They view implanting embryos as the reproductive equivalent of invasive, malignant cancer, where it is well known that genetic instability is a prime mover in cancer evolution and the progression to more malignant states. Embryos are inherently invasive – as the occasional occurrence of ectopic pregnancy demonstrates – they don’t always need a uterus at all. In the face of maternal resistance to the indiscriminate acceptance of embryos for implantation the embryo at least temporarily transitions to genetic chaos. We know this happens during the first cell divisions of the blastocyst because Vanneste has satisfied herself that the genetic instability in embryos arrives with mitotic divisions post-fertilization, rather than meiotic divisions associated with the production of gametes. In fact, Brosens and his many collaborators have produced an elegant theory to account for the chemical, immunological and anatomical changes that occur in the uterine wall about 5 days after ovulation – called decidualization – which prepares the cells of the uterus to surround implanting embryos and interrogate them for viability, quality and compatibility. It is an evolved rejoinder to invasive embryos, and the mother invests in decidualization whether or not an embryo is present. If a poor quality embryo is rejected, or none arrive, the uterine wall breaks down in the painful and messy process 50% of readers will instantly recognise – menstruation. It is the price women pay for evolved mechanisms aimed at protecting them against wasting precious resources gestating unwanted embryos, says Brosens.

But what about question 3? How do genetically aberrant embryos give rise to normal, genetically sound, babies? This is precisely the question that began exercising Magdalena Zernicka-Goetz, a head of research at the Gurdon Institute in Cambridge, UK, when she became pregnant for the second time at the ripe old age of 44. As a formidable developmental biologist she was acutely aware of the increased risk of genetic disorders like Down syndrome in babies born to older mothers. As a precaution, she had a chorionic villus sampling test done at about 12 weeks. This doesn’t risk harming the baby because it takes a few cells from the placenta and analyses them. In Magdalena’s case it was found that 25% of the cells in the placenta were genetically abnormal. Could this mean her baby was similarly compromised? In fact, baby Simon was born perfectly healthy – and has remained so. But, in asking around a number of geneticist colleagues, she discovered an alarming ignorance about the effects on the development of embryos which contained defective cells, and what happened to these abnormal cells as the foetus developed. She decided to investigate.

Restrictions on the use of human embryos for research meant that Zernicka-Goetz had to use a mouse model. But, as said earlier, mice and women are very different in that mouse embryos suffer from very little genetic abnormality and humans suffer from very high levels. So the research team had to induce aneuploidy (departures from the normal diploid condition) in the mouse embryos in order to mimic the human condition. They treated the early embryos with reversine, a chemical that interferes with the fidelity of mitotic cell division, at the four to eight-cell stage. During mitosis, individual chromosomes divide to form identical sisters which line up on a structure that appears in the center of the nucleus called the spindle. Each sister is then pulled apart down the spindle to opposite poles which form the nuclei of two daughter cells. By inhibiting this spindle assembly reversine allowed them to create a number of genetically abnormal, aneuploid cells, where chromosomes, or parts of chromosomes had gone missing. After a short burst of reversine treatment the embryos were transferred to an inhibitor-free medium and normal cell division resumed. They ended up with mosaic embryos composed of a mixture of 50% normal cells and 50% aneuploid cells and tracked the development of these mosaic embryos to determine the fate of the aneuploid cells depending on which part of the embryo they were situated in.

As the cells of the blastocyst (implanting embryo) divide they organise into different lineages. One lineage goes on to form the fetus, the other forms the placenta. The researchers found that aneuploid cells in the fetal lineage were rapidly destroyed by apoptosis – programmed cell death. The aberrant genetics were removed by clonal depletion rather than some hypothetical mechanism that corrected genetic mistakes inside cell nuclei. On the other hand, although aneuploid cells in the placental lineage did decline in number over time, many remained and appeared to be tolerated such that the resulting placentas were mosaic.

They took matters further by transferring the mosaic embryos into foster mothers and discovered that, provided the proportion of aneuploid cells did not exceed 50%, implantation rates and viability equalled that of normal embryo controls. They concluded that the mosaic mouse embryos had an extraordinary ability to self-correct by attrition of defective cells and Zernicka-Goetz strongly believes the same will be found to be true of human embryos – which could explain why baby Simon was born genetically normal while his placenta was a complex mosaic of normal and aneuploid cells.

Why is aneuploidy tolerated in the placenta but not in the fetus? It could be that some mechanism more sternly interrogates the genetics of cells in the fetus because of the danger of corruption in the ensuing individual which could give rise to a plethora of genetic diseases. It could be that this hypothetical stern interrogation can afford to be more relaxed in the placenta. Or could there be an adaptive reason why the placenta tolerates a substantial burden of aneuploid cells? To the extent that these cells are invasive, by virtue of their wonky genetics, those talents can be put to great use in an organ whose job it is to burrow deeply into the uterine wall. Humans, along with the great apes, have the most invasive placentas in the animal kingdom, the so-called hemochorial placentation. A successful placenta so remodels the spiral arteries of the uterine wall that it becomes impossible for a human mother to withhold blood and nutrients from her fetus without literally starving herself to death first. If this scenario is true then the human embryo is fiendishly evolved to further its selfish interests. It resorts to genetic chaos to inveigle the uterine wall to allow implantation, later correcting the damage in the fetus, while maintaining it in the placenta to allow the fetus an unfettered nine month long smash-and-grab raid on its mother’s resources.

This research has repercussions for IVF technology. If abnormal embryos can go on to make perfect babies, pre-implantation genetic diagnosis (PGD) – as it stands at the moment – might well be a waste of time. Millions of embryos might be being erroneously disposed of and millions more erroneously selected for implantation. If Zernicka-Goetz is right the search for the perfect embryo might be over.

All this highlights the need for the input of evolutionary thought to medical technology, to question the assumptions about biological mechanisms that are made in its absence. IVF, for instance, can be viewed as a gigantic biological experiment similar in scale and ramifications to the birth pill, and several evolutionists have recently weighed in with thoughtful and cautionary remarks concerning the possible un-considered evolutionary effects of widespread use of IVF technology.

Pascal Gagneux, at the University of California, San Diego, points out that 250,000 babies a year are now born via assisted reproduction. Because it is to a large extent a commercial business, he says, clinics offering IVF don’t need to reveal the enzymes, culture media and storage buffers they use, which all represent novel environments for sperm, eggs and embryos and are thus capable of exerting novel selection pressures on reproductive biology. In the early days of IVF, millions of sperm were introduced to the egg on the basis of let the best man win, and although the modern technique of intra-cytoplasmic sperm injection (ICSI) has reduced this to one, it is a lab technician who selects the sperm to be used, not the mother’s cervix, uterus and fallopian tubes. There is evidence, he says, that the epithelium of the fallopian tube is heavily ciliated and can hold the sperm while it is immunologically scrutinised. All this is by-passed by IVF. It is also known, says Gagneux, that gene expression and placental growth rates are different in IVF pregnancies compared to normal in vivo pregnancy. IVF technology might be changing the imprinting of genes which may account for all the imprinting disorders, malformations, cancer, increased occurrence of twinning, preterm birth, Caesarian section and pre-eclampsia that have been documented. Do all those IVF children face a time-bomb of poor health and short lives, worries Gagneux. The trouble is – we shall have to wait and see.

Hans Ivar Hanevik and his colleagues from the University of Oslo paint a wider picture of concern about IVF. Assisted reproduction technologies intervene, they point out, at a point in the human life cycle where natural selection operates at its strongest. They replace selective forces at large during oocyte maturation in the ovary, in selection among sperm for candidates to penetrate the zona pellucida of the egg and fertilise it, and factors governing embryo development and implantation, with the selective forces represented by harvesting, screening and selection of embryos, the freezing and thawing associated with embryo storage, and the laboratory hardware in which all this is done. Furthermore, IVF is typically carried out for infertile couples where the causes of their infertility are heritable. These range from endometriosis and polycystic ovary syndrome in women to oligospermia and hypospadia in men. This malformation of the penis is more common in boys born of ART. Hanevik worries that IVF thus propagates the genetically inheritable traits of sub-fertile couples and that studies done over several generations might reveal an increased risk of sub-fertility in the descendants of IVF couples and the population at large. He points out that 5.9% of all births in Denmark are now achieved with assisted reproductive technology and that figure is 3% if you pool all the Nordic countries. Furthermore, because of the commercial nature of much assisted reproduction, there is selection among the sub-fertile population for wealth. Even in countries like Norway, where there is a high rate of publicly funded IVF, women with low body mass indices will be selected for treatment and obese women denied – so there is selection for weight, and by implication, health in general. Overall, he says, IVF favours healthy sub-fertile couples in stable relationships who live in high-income societies. Thinking like this is bound to be controversial, there’s more than a faint whiff of eugenics about it, and Hanevik is careful to state that the question of whether these traits have anything to do with heredity, genetics and evolution is therefore bound to raise hackles. So, to point out that IVF may favor disease-prone individuals or lead to reduced fitness over generations is surely provocative – but he believes they are questions worth asking. IVF, he concludes, is a classic example of how the human species – human evolution – is becoming not only culturally but also biologically dependent upon our own technology.


  1. Chromosome instability is common in human cleavage-stage embryos. Vanneste et al, Nature Medicine, Vol. 15, Number 5, 2009, pp. 577 – 583.
  2. Body by Darwin. Jeremy Taylor. University of Chicago Press, 2015.
  3. Something new about early pregnancy: decidual biosensoring and natural embryo selection. J.J. Brosens and B. Gellersen, Ultrasound Obstet. Gynecol. 2010; 36; pp. 1-5.
  4. Mouse model of chromosome mosaicism reveals lineage-specific depletion of aneuploid cells and normal developmental potential. Helen Bolton et al, Nature Communications, 29. March 2016.
  5. Can IVF influence human evolution? Hans Ivar Hanevik et al, Human Reproduction, doi:10.1093/humrep/dew089