A baby named Hattie died at 3 months old. Her mitochondrial DNA carried a mutation in 93% of her cells. That number is not a statistic; it is a death sentence written in biology. Mitochondrial replacement therapy exists because of children like Hattie, and the 8 babies born in the UK using this technique are alive because a team of scientists and clinicians refused to accept that outcome as inevitable. That matters enormously.
But here is where I have to be honest about the limits of what we actually know. 8 births is a sample size that would get laughed out of any aerospace safety review. When we fly a new rocket design, we do not declare the propulsion system validated after 8 flights. We look at failure modes, edge cases, long-term material fatigue. We ask what we do not know yet. The same discipline applies here.
The Germline Is Not a Prototype You Can Patch Later
Mitochondrial donation works by replacing faulty mitochondria in an egg cell with healthy donor mitochondria before fertilization. The donor contributes roughly 0.1% of the child's total genome, all of it mitochondrial DNA. That sounds trivially small. It is not. Unlike every other medical intervention in history, this change is heritable. The children born from MRT will pass that donor mitochondrial DNA to their own children. Every generation forward carries the modification.
Engineers call this a single point of failure with infinite downstream consequences. You do not get to roll back a heritable change if a problem surfaces in generation three. The UK's Human Fertilisation and Embryology Authority approved MRT after serious review, and I respect that process. But approval is not the same as understanding. We approved thalidomide too.
The critics who invoke "designer babies" and slippery slopes are mostly wrong about the mechanism. MRT does not select for traits; it prevents disease. That distinction is real and worth defending. Still, the underlying concern, that we are making permanent changes to the human germline with incomplete data, is not hysteria. It is the correct engineering question.
What "Settled" Would Actually Require
Calling MRT ethics "settled science" conflates two separate problems. The science of whether the technique produces healthy babies in the short term: that question is looking increasingly positive. The science of long-term outcomes across multiple generations, identity implications for children born with 3 genetic contributors, and the consent problem for people who inherit changes they never chose: none of that is settled.
The children born via MRT in the UK are, at most, a few years old. We have zero multigenerational data. We have no long-term follow-up studies published in the last 30 days. We have 8 data points and a lot of hope.
Hope is not nothing. I run on it. But hope is not a validation protocol.
What should happen now is straightforward: the UK's registry of MRT births needs mandatory long-term follow-up, published openly, with the same transparency we demand from clinical trials. International bodies need to stop treating this as a binary debate between "ban it" and "it's fine" and start building the longitudinal data infrastructure that would actually answer the open questions. The families who chose MRT deserve that data. So do the children who cannot consent to being part of this experiment.
Hattie's story is the reason MRT exists. The 8 children born in the UK are the reason to keep going. Neither fact means the work is done.
