Nature News has an intriguing article
on the next three decades of reproductive medicine: essentially a
series of short musings from scientists working in the field about the
issues we will be facing in 30 year’s time.
It’s worth reading through in full, but this statement from Susannah Baruch at Johns Hopkins caught my eye:
speculation that people will have designer babies, but I don’t think
the data are there to support that. The spectre of people wanting the
perfect child is based on a false premise. No single gene predicts
blondness or thinness or height or whatever the ‘perfect baby’ looks
like. You might find genetic contributors but there are so many
environmental factors too.
More likely is that you’ll have a set of embryos and you’ll know every single thing about every gene in every embryo.
For example, one embryo will have three genes associated with tallness,
two for weakness, three for poor vision and some for disease; and the
second embryo will have some other set. They’re very complicated data.
This is not creating a baby from scratch. None of us is a perfect
specimen and none of our embryos will be either. [my emphasis]
unsure how selecting amongst these embryos doesn’t count as making
“designer babies” (it’s still a choice, even if it’s between a set of
imperfect options), but I think Baruch’s second paragraph is spot-on.
It’s clear from recent genome-wide association studies (GWAS) for
height and weight that many (if not most) traits are hideously complex at the genetic level
– a mess of common and rare genetic variants scattered throughout the
genome, each contributing only the tiniest proportion of the overall
Height is a great example: GWAS results from more
than 30,000 individuals have uncovered dozens of contributing variants,
which together predict less than 5% of the variation in height.
There’s no doubt we’ll uncover much of the remaining variation with
emerging technologies (analysis of structural variation, and
large-scale sequencing) and larger cohorts, by which time we’ll likely
have hundreds of contributing
variants. The same will hold more or less true for other complex
traits, including susceptibility to common diseases like diabetes or
The point is not that we will never understand the
genetic basis of complex traits – we will, at least to a pretty good
approximation, given advanced tools and sufficiently large cohorts. The
point is that even once we understand
the genetics of complex traits perfectly, that won’t be enough to
generate a “perfect baby” through embryo screening alone.
illustrate this, imagine – ten or fifteen years from now – a couple who
have just had IVF to generate perhaps two dozen embryos, and want to
use genetic testing to decide which one(s) to implant. There won’t be a
single, stand-out embryo, perfect and disease-free, because generating
a “perfect” embryo – one with the “desirable” variant at every single
position in the genome – runs up against a pretty serious probabilistic
challenge. Let’s say there are only 5,000 DNA variants that negatively
affected human health (an under-estimate) each with a frequency of just
1%: that means you would get a “perfect” embryo around once in every 1022 attempts (that’s a 1 followed by 22 zeroes, a stupidly large number).
likely that methods to generate large numbers of embryos will be
developed, particularly once stem cell technology enables sperm and egg
cells to be created from adult tissue, but generating and screening 1022
embryos is no mean feat: at the rate of one every second, this would
take you about 200,000,000,000,000 years, ten thousand times longer
than the current age of the universe.
So it’s safe to say that there will be no perfect baby.
Instead, the prospective parents will face a tough choice between
embryo A, who will likely be tall, slim, smart and cancer-free but have
a higher-than-average chance of bipolar, early-onset dementia, and
infertility; embryo B, who will be a little shorter, dark-haired,
probably fairly gregarious, resistant to coronary artery disease,
susceptible to bowel cancer, hypertension and early deafness; embryo C,
who will be of average intelligence, unlikely to suffer premature
baldness, prone to mild obesity and diabetes, but not at a high risk of
any of the other major common diseases; and embryos D-N, who present a
similar panel of competing probabilities.
Of course, a few
embryos may carry mutations with known, severe health outcomes – those
associated with rare diseases like muscular dystrophy, for instance.
Embryo screening will have a very real impact on the frequencies of
these conditions, just as we are already seeing
following pre-natal diagnosis of conditions such as Down syndrome. But
for the complex, common diseases there will be no easy answers; just a
set of trade-offs.
The parents-to-be will sit down together with
dossiers listing a huge set of statistical predictions for each of
their potential children, and make a decision as to which (if any) of
these abstract collections of traits and risks they wish to bring into
this world. Decisions don’t get much more emotionally traumatic than
this: not only will they be making a decision that will shape their own
lives and that of their future offspring, parents will carry a new, extra burden of responsibility for the fate of their children.
If they decide on embryo A, and their child goes on to develop severe
bipolar disease, they will carry the guilt of that decision in addition
to the trauma of the disease itself.
That’s not to say that
embryo selection is unworkable – in fact, I think it’s inevitable – but
rather that this process is likely to require a degree of agonising
trade-offs on the part of parents-to-be that is seldom fully