In discussion of my recent book review of The Past is a Future Country: The Coming Conservative Demographic Revolution, J.O.A. Rayner-Hilles questioned whether embryo selection could really become powerful enough to combat the on-going dysgenics. Well, that depends on certain key facts:
- How fast will the technology increase in uptake?
- How good will selection be with regards to intelligence and other precious phenotypes?
The rise of assisted reproductive technology (ART)
In 2020 I made a plot showing the growth of ART, as a proportion of all babies born in various European countries. The reason this can be done is that ESHRE (European Society of Human Reproduction and Embryology) collects and publishes all the data (here’s their papers). In my 2020 plot, the data series ended in 2014 because the academics are 6 years slow. However, they have since caught up a bit, and the newest data series now ends in 2018 (data file here, I only bothered to compile the ART% in recent years since I don’t use the other data and they are somewhat inconsistent across reports). Actually, there are abstracts about conference talks given about the 2019 and 2020 data, but those aren’t fully published from what I can tell, so we can’t use them yet. Here’s how it looks like now, from 1997 to 2018:
The data are a mess. Countries routinely fail to report sufficient data to generate the ART%, even a top country like Norway didn’t report data for 2015-2018. Still, we see a general upwards trend. A statistical model says that this increase is 0.12%points per year across countries (p < .001). Clearly, some countries are rising faster than others, especially Spain (current fertility 1.17, always look at Birth Gauge on Twitter for up to date fertility trends). A growth of 0.12%points/year is not impressive, but it is steady progress (1.2%points per decade). Of course, one can say that this represents a sort of anti-process in that people wouldn’t bother to use ART if they didn’t have to. True enough, but the use of ART is an indicator of how many people are willing to at least use some technology to help with fertility. The uptake of this technology will pave the way for more advanced technologies that can make a real difference.
How good is the current selection?
It’s terrible. Less than half of current ART is IVF (in intro fertilization), and of that, only some of IVF uses genetic testing (PGT, pre-implantation genetic testing) to chose the right embryo. And of those that use PGT, most of it is looking for crude chromosome errors (PGT-A, aneuploidy, think of Down syndrome etc.). Some minority of PGT involves testing for major genetic disorders (PGT-M, Mendelian disorders). And some smaller proportion yet uses PGT-P (polygenic), which can be used for any complex genetic phenotype, including height, obesity, schizophrenia, and intelligence. It is safe to say that currently, ART itself is having just about zero net impact on the selection because it’s relatively rare and doesn’t have much effect. However, since older mothers tend to be more well-educated and smarter, the self-selection into using ART is probably slightly eugenic, boosting the fertility of the right-tail of the social status and intelligence distribution a little.
What is the future?
Based on this, one might say that ART is not likely to make a big difference for the dysgenic problem. That is true for now, but the situation can rapidly change. The main problem with using ART is that to do serious selection — PGT-P (polygenic selection) — one needs to do IVF to retrieve eggs in order to make embryos. This is expensive, time consuming and unpleasant for women. However, this is about to change. Instead of retrieving eggs from women, one can collect other cells that are easier to collect (e.g. skin or blood cells). Using epigenetic tricks, these can be converted into egg cells in large quantities, a process called IVG (in vitro gametogenesis). This method has been done in animals a few times, leading to live births without any apparent problems. Here’s some studies:
The female germ line undergoes a unique sequence of differentiation processes that confers totipotency to the egg1,2. The reconstitution of these events in vitro using pluripotent stem cells is a key achievement in reproductive biology and regenerative medicine. Here we report successful reconstitution in vitro of the entire process of oogenesis from mouse pluripotent stem cells. Fully potent mature oocytes were generated in culture from embryonic stem cells and from induced pluripotent stem cells derived from both embryonic fibroblasts and adult tail tip fibroblasts. Moreover, pluripotent stem cell lines were re-derived from the eggs that were generated in vitro, thereby reconstituting the full female germline cycle in a dish. This culture system will provide a platform for elucidating the molecular mechanisms underlying totipotency and the production of oocytes of other mammalian species in culture.
Reconstituting gametogenesis in vitro is a key goal for reproductive biology and regenerative medicine. Successful in vitro reconstitution of primordial germ cells and spermatogenesis has recently had a significant effect in the field. However, recapitulation of oogenesis in vitro remains unachieved. Here we demonstrate the first reconstitution, to our knowledge, of the entire process of mammalian oogenesis in vitro from primordial germ cells, using an estrogen-receptor antagonist that promotes normal follicle formation, which in turn is crucial for supporting oocyte growth. The fundamental events in oogenesis (i.e., meiosis, oocyte growth, and genomic imprinting) were reproduced in the culture system. The most rigorous evidence of the recapitulation of oogenesis was the birth of fertile offspring, with a maximum of seven pups obtained from a cultured gonad. Moreover, cryopreserved gonads yielded functional oocytes and offspring in this culture system. Thus, our in vitro system will enable both innovative approaches for a deeper understanding of oogenesis and a new avenue to create and preserve female germ cells.
There are several companies working on making this work in humans too, e.g. 2023 news:
This technology is easy to sell to egalitarians because it actually enables homosexuals to have biological children. A recent study was able to generate live offspring from two male rats:
Sex chromosome disorders severely compromise gametogenesis in both males and females. In oogenesis, the presence of an additional Y chromosome or the loss of an X chromosome disturbs the robust production of oocytes1,2,3,4,5. Here we efficiently converted the XY chromosome set to XX without an additional Y chromosome in mouse pluripotent stem (PS) cells. In addition, this chromosomal alteration successfully eradicated trisomy 16, a model of Down’s syndrome, in PS cells. Artificially produced euploid XX PS cells differentiated into mature oocytes in culture with similar efficiency to native XX PS cells. Using this method, we differentiated induced pluripotent stem cells from the tail of a sexually mature male mouse into fully potent oocytes, which gave rise to offspring after fertilization. This study provides insights that could ameliorate infertility caused by sex chromosome or autosomal disorders, and opens the possibility of bipaternal reproduction.
This technology is very important because if we can side-step the IVF procedure, the entire process becomes far easier and cheaper. No need for cycles to retrieve a few eggs at a time, just take 10k skin cells, convert them to 1000 eggs, fertilize the eggs, and get 100 embryos (numbers made up). Repeat as necessary. Do the genetic testing of the embryos, choose the best one. The genetic testing itself will not be so expensive because the cost of sequencing looks like this:
As the embryos are all siblings, and are genetically related also to the parents, one doesn’t have to sequence the embryos at depth. One can sequence them lightly and impute the remaining data from the parents. I imagine sequencing costs will fall to <10 USD per embryo, enabling relatively cheap sequencing of even 1000 embryos.
The accuracy of genetic predictions
Similarly, there is a long-running increase in the genetic prediction models’ accuracy. Here’s the plots from one of the recent education GWASs (EA4):
It is a poor way of showing the progress because we care about the correlation metric, not the R squared. You just have to take the square root to get it from the chart. Thus, EA1 had r = .17, EA2 had r = .26, EA3 had r = .33 and EA4 had r = .40 (Add Health dataset). These values are somewhat inflated because of bias in the training data (shared environment confounding, indirect genetic effects etc.), but my friends tell me that current state of the art sibling accuracy is about r = .30 for IQ. As such, if we select the best embryo out of a 100 for IQ using a predictor with r = .30 (R2 = .11), one would roughly get an improvement of 0.59 standard deviations in the PGS for IQ, that is, 8.8 genetic IQ (using the Shai Carmi R simulator). This is easily enough to surpass the slight dysgenic fertility relationship (negative 1-2 IQ per generation) at the individual level.
Thus, the question with regards to the dysgenic doom scenario is whether we can develop this technology in time, and whether it can be popular enough to have a large enough impact. I don’t see any technological or monetary problems, so mainly this comes down to a difficult forecast about the reproductive politics of tomorrow. Some countries will pursue this technology even subsidize it, others will attempt to ban it. If you are seriously concerned with dysgenics and don’t want to wait for religious conservatives to maybe fix the problem, you should probably promote this technology. In fact, if you are concerned about immigration’s effects on the state budget, you should also pursue this technology. Our research shows that the main reason for their poor performance in Western societies is their relatively lower average intelligence. But this is something that one can improve too.