[wahern]

> natural system of transcription, translation, and protein synthesis at the core of DNA still makes mistakes.

Exactly, which means the system is resilient, which means it's more forgiving of random mutations introduced by gene editing, which is perhaps why most scientists are less concerned than you think they ought to be.

Imagine if we took the same approach to surgery--overly worried that scar tissue of any kind at any point would completely disrupt the functioning of an organ. We'd be a hundred years or more behind where we are now. Scientists and doctors understand that the body is both much more complex than a simple machine, but also much more resilient.

Such criticism cuts both ways. Likewise for GMOs. Demanding perfection is unreasonable and unnecessary.

[ak217]

There's a huge difference between somatic and germline editing. Surgery is like somatic editing. It's elective, based on medical need, and its effects are confined to one individual. Germline editing is more like forcing an entire population to undergo surgery that is claimed to enhance their bodies.

Nobody is particularly concerned about somatic editing where a medical need exists. Everybody is rightly freaking out about germline editing because it's non-elective for the babies, its consequences are permanent and can be severe, and we aren't good enough at it yet (not even the somatic variety) to claim anything about its safety.

[wahern]

IIRC, when I studied biology in secondary school and as a freshman in college, it was believed that gametes were completely isolated and untouchable by viral infections. Turns out this is not true. But if you believe that gametes are incredibly fragile, I can see how it would be easy to believe this as otherwise you'd expect to see mutants everywhere.

I realize the stakes are higher for germline editing, but rationally speaking higher stakes alone don't change the calculus.

[ak217]

I don't follow. Gametes are isolated from many viral infections, but even assuming that a gamete undergoes viral mutagenesis, that's not evolutionarily novel (viruses follow relatively predictable insertion patterns, and transposons are basically degenerate/grounded viruses that are almost certain to generate mutation events in any given zygote, unlike viral mutagenesis). It's also outside our control, unlike germline editing, which is not random in completely novel ways.

It is precisely because the stakes are so much higher that the calculus changes. We have to be reasonably certain that we can safely edit germline by experimenting with somatic editing and germline in lab animals before we can do something as consequential as deploying it clinically.

[wahern]

Sure, they are more isolated. And in some ways they are more fragile. But my point is that they're not nearly as fragile as we believed. And more importantly a faulty premise--that they're extremely fragile--gave way to a faulty scientific conclusion--that gametes were impervious to common environmental stressors like the many infections that ravage our bodies.

We can't draw simple, categorical conclusions about gene editing, nor even germline editing in particular, because there are no simple, categorical distinctions. Suffice it to say, it's complex. Whenever we try to be reductivist about such things we end up drawing erroneous and even dangerous conclusions; e.g. gamete fragility -> viral imperviousness as solution to observed lack of harm -> underestimation of viral stressors and risks, and overestimation of gene editing risks.

Regarding stakes, what I had in mind was classic economic behavioral experiments where they show that a change in the magnitude of a bet changes choices in an irrational manner even though the expected payoff is exactly the same; even when you take into account marginal utility effects.

Yes, germline edits don't just effect one person, they theoretically could infect all of humanity. But so what? Remember when they were firing up the LHC and people were freaking out at the possible creation of blackholes. Given the error bars in known physics, there was a non-zero chance running the experiments could have destroyed the world.[1] Because nothing we could possibly learn would compensate for losing everything, does that mean we should never have turned it on? No. The calculus didn't change. That germline edits propagate doesn't mean our harm+benefit calculus changes; it's just that one of the risk factors in the equation changes from a 1 to something larger. Other factors, like confidence, may or may not need to be changed.

Regarding the argument that by intervening scientifically we're categorically more culpable than if we didn't do anything, that touches upon the is/ought and naturalistic fallacies. From a utilitarian perspective intervention vs non-intervention is irrelevant. I'm not a Utilitarian (capital U), but I don't see how we can have a constructive, scientific debate outside a utilitarian framework. Such lines of reasoning are more relevant to political and religious contexts.

[1] Well, maybe. Actually, perhaps many physicists would have said that the consensus science would have put the chance at 0. But the best argument was made by people pointing out that the Earth's atmosphere was constantly bombarded by particles far more energetic than what the LHC would create. Which is exactly analogous to germline editing. The fact is, the germline undergoes far more genetic stressors than we once believed. It must follow that it's more resilient than we believed, genetically, developmentally, and from an evolutionary perspective.

[ak217]

The concerns about the LHC were a hell of a lot more hypothetical than about germline editing, and that did change the calculus. We know that current genome editing techniques have off-target effects.

You're arguing from some abstract philosophical perspective, but the practical situation is much simpler. Nobody is drawing categorical conclusions and saying that we should never edit the germline, and at the same time the opinion that we should do germline editing right now is fringe. The tools, while much better than ten years ago, still suck. Outside a few well-characterized alleles in Mendelian diseases, nobody knows what to edit, what side effects edits will have or why. It's likely that in a few years we will know, given that we're quickly improving both the molecular techniques and the genome knowledge bases necessary to understand the consequences of the edits. But until then, it's dangerous and unethical to experiment on babies without their consent or pressing medical need, and scientists are right to freak out about it.

[wahern]

If you believe that germline gene editing is so risky and potentially costly as to be absolutely unwarranted, then you shouldn't support somatic gene therapy as there remains a very real and non-negligible risk of germline integration. Some vectors might be safer than others, but proving the impossibility of gene integration into the germline seems extremely costly and possibly unprovable. I mean, heck, there's at least one scientifically proven case of a virgin birth.[1] Where does that leave us?

I return to my original point: Demanding perfection [and omniscience] is unreasonable and unnecessary. All of these other concerns are typical of any medical procedure: you attempt as best you can to integrate known risks as well as unknowns (known unknowns and unknown unknowns) into a cost and you compare it to the benefit, and if the benefit outweighs the cost then go for it. For germline editing in particular the costs will likely outweigh the potential benefits in most cases for some time, but we still need to make that determination regularly, honestly, and in context (actual proposed cases), without our fingers on the scale.

Issues like consent are ancillary. And they exist regardless of gene therapy. People don't consent to be born. Or consent to be "identified" through family members choosing to publish their genetic information. Social engineering experiments have lead to holocausts, even when they began innocently; if you go back far enough in the causal chain, they're all innocent and completely unintended. These problems, high-stakes consequences, and paradoxes already exist; we already struggle with them. Gene therapy, not even germline therapy, create fundamentally de novo issues. That's the real hubris, the delusion that we're not already playing with fire.

At the end of the day what the Chinese researcher did was reprehensible, but mostly for very particular reasons. I'd wager big money that a significant plurality of medical scientists, if not a majority, are today already prepared to approve germline editing given a good candidate therapy--patient, vector, payload, etc. As for medical ethicists, as scholars they tend to splinter into radical advocates or skeptics because that's how you get tenure and attention; and unlike doctors they don't get fired (or "disappeared") when they're wrong.

[1] "Oral conception. Impregnation via the proximal gastrointestinal tract in a patient with an aplastic distal vagina. Case report", http://img2.timg.co.il/CommunaFiles/21227065.pdf

[coolguy132435]

Is what you describe the basis of Human Endogenous Retroviruses?

[wahern]

I've assumed so, but I'm not a scientist by trade and never looked deeply into it. I just know that over the years I've crossed paths with several papers that showed infection of gametes (usually sperm) by common viruses and often gave them a cursory look. These papers always catch my attention because what I was taught in school never made sense to me. I mean, I know gametes are in some ways more fragile than most cells, and they are clearly isolated to an extent (for obvious reasons), but the imperviousness I was taught seemed unlikely and unsupportable. And because of that I've always been especially skeptical of arguments premised, directly or indirectly, on the extreme mutagenic fragility of gametes and germlines.

[deusofnull]

I agree it is resilient, but that resilience has been tested by a slow, small increment evolution over millions of years. And even given that, genetic disorders with novel expressions exist that confound modern science. if we start making very fast changes, the resilience of the genetic system shouldn't be expected to hold up against a far faster change rate. And more to my point about the complexity of it all, there are huge regions of our genome that we thought were "unrelated" to human phenotypes (physical expression of genetic code) until very recently when we discovered some of this "junk" DNA actually is critical for certain types of RNA transcription / repair.

GMOs are a different issue. My underlying concern about the influx of new genes introduce to the wild quickly being potentially dangerous aside, the genetic modification have mostly been to make plants resilient to various herbicides/fungicides/pesticides to allow large scale monoculture farming. That's one solution, but not the only. And we have seen invasive forms of GMO crops spread out of containment, thus entering the wild population, and we wont know the consequence of that for a long time, if any.

takeaway being, its probably worth being conservative about even moderate scale genetic manipulation of any species we rely on, and especially our own genome. Not saying dont do any genetic modification, just that it really has to be air tight, you might say NASA level standards of engineering. Now, you tell me if you think Monsanto's engineering standards are at NASA level, or that the Chinese Governments ethical standards are without problems.

[pythonaut_16]

To play devil's advocate: Scar tissue doesn't get passed to offspring or propagated throughout the population though.

[TeMPOraL]

To play the angel's advocate (or whatever is the next-step reverse of devil's advocate): natural selection helps eliminate problematic genes in populations. A good chunk of harmful mutations will prevent people from procreating, either by killing them first, or making it unlikely for them to find a mate.

[deusofnull]

>A good chunk of harmful mutations will prevent people from procreating, either by killing them first, or making it unlikely for them to find a mate.

well, what about a recessive allele that doesnt cause problems in the first generation but does in subsequent? What about co-dependent deadly / devastating gene expressions between newly added genes that pop up after 2 people with these co-active genes expressing like a hundred years from now? look, given what i studied and why, i would never say we should forever abstain from gene editing technology. BUT, we have to hold those who conduct genetic modification experiments to the absolute highest ethical and engineering standards. It is akin to the level of power given by atomic science, maybe even more. one can be sure that the powers of a God would include the power of the atom and the power of the gene near the top of the list.

[sudosysgen]

Not nearly as much in 2019's modern societies.

[mattkrause]

Less now, for sure, but you're overlooking a major category.

Many of the really bad mutations are embryonic-lethal, which is why a surprisingly high proportion of pregnancies "don't take" or end in miscarriage.

[ak217]

Yes, and many more cause carrier mutations that won't surface until the next generation, or late-onset metabolic or neural disorders that will cripple the individual later in life, or increase mutational load resulting in much higher chance of cancer. The fact that some mutations are benign and others fatal really doesn't change the issue that we don't yet know how to safely edit the human germline.

[mattkrause]

Totally agreed--I think this was a horrific and stupid thing to do.

I just wanted to push back a bit on the idea that medicine is badly "breaking" evolution by keeping people alive.

[imtringued]

>Exactly, which means the system is resilient, which means it's more forgiving of random mutations introduced by gene editing, which is perhaps why most scientists are less concerned than you think they ought to be.

The argument should go the other way. Because these DNA mistakes are predictable our genes have designed safeguards against them. If you now randomly introduce a "mistake" that is beyond the capability of the system then it won't help you.

[VectorLock]

The human body, incredibly robust and frustratingly fragile at the same time.