The biological danger in germ-line gene therapy lies in the randomness of available means of gene insertion, with its capacity to disrupt other genes and its capacity to overdose the embryo.

To be sure, there are critics of reproductive technology who attach moral importance to germ-line DNA generally and whose objections would persist no matter how safe oocyte repair might be demonstrated to be. But in any case ooplasmic transfer from healthy donated eggs does not involve altering genes.

But it does mean transferring some 13 genes that are contained within the small "self-charging batteries" we call mitochondria, and of which there are about a million or so in an ovulated egg.

Whether intentionally (for serious mitochondrial disease) or unintentionally (when for some other reason an egg otherwise won't start to develop properly), cytoplasmic transfer means transferring mitochondria - and that means transferring mitochondrial DNA.

For context, we should begin with the harms associated with the much more commonly practiced alternative to egg cytoplasm donation, namely the donation of a whole egg (see the Box, Commerce in eggs and pregnancy after the menopause).

The mitochondrial DNA -- the mitochondrial genome -- is unlike the nuclear genome in several ways. To begin with, just 13 proteins are coded for, all concerned with a fundamental metabolic process called oxidative phosphorylation within the mitochondrion, with little or no genetic difference from one healthy person to the next.

Second, instead of the two alleles of each nuclear-encoded gene per cell, mitochondrial alleles in a cell number in the hundreds or the thousands (hundreds of thousands, in the case of eggs), so any advertent or inadvertent "altering" of the germ-line is at most a dilution effect, not a replacement.

Third, the mtDNA from the donor detectable in some of the offspring after cytoplasmic transfer has little or nothing to do with mitochondrial genes. It's in a small area of the mtDNA molecule called the D-loop that does not code for anything (it was once called "junk DNA"; it is variable from family to family, even more so from racial group to racial group, etc; and because you inherit mitochondria and mtDNA only from your mother, anthropologists use it to trace matrilineal descent).

There are, therefore, no novel or altered genes introduced by egg cytoplasmic transfer. So it is not gene therapy.

Those of us who are students of mitochondrial genetics are still cautious, though. There is no experimental animal to serve as a model for sterility from ooplasmic failure, because physiological sterility from dodgy egg function years before you run out of eggs is, as far as we know, a peculiarly human phenomenon.

Some people argue that the cytoplasmic transfer of defective mtDNA might enable propagation, in the rest of the child's tissues years later, of deleted or otherwise mutated mtDNA, to result in mitochondrial disorders of a kind known clinically to be associated with progressive, untreatable and ultimately fatal metabolic diseases.

But to invoke problems of this kind you first have to suppose that the donated egg cytoplasm has a higher mutant mtDNA load than the recipient egg, otherwise the situation will only be helped.

Obviously egg cytoplasm donors should in future be screened for mitochondrial disease and, ideally, for mtDNA genomic integrity. But meanwhile, at least as judged by the donor's previous reproductive performance, you would have more confidence in the oocyte mitochondria of the donors than in that of the recipients.

For rare families with mitochondrial DNA mutations there's no other hope

Important that it is to improve the reproductive window for women in modern western society, there is more to consider medically, namely families that suffer from mitochondrial genetic diseases.

Unlike conventional genetic disease, where a quarter to a half of all embryos will be geneticaly normal (depending on whether the condition is recessive or dominant), with mitochondrial mutations up to 100% of offspring are affected, and the inherited diseases that result are among the most miserable infantile diseases imaginable.

An example is Leigh disease, with progressive and hopeless deterioration of muscle and brain function from soon after birth until death, months or years into infancy or childhood.

"Conventional treatment" for affected families consists of egg donation: a new set of mtDNA, to be sure, but also a new 32,000-plus nuclear alleles that are not faulty in the affected woman.

There is a compassionate case here for nuclear transfer from the affected egg to a donated egg, in effect deliberately replacing the mitochondrial genome we should all have in common, while preserving the nuclear inheritance that characterizes our differences.

I argue that we should proceed cautiously with clinical trials of cytoplasmic transfer for age-related female sterility and of nuclear transfer for families with mitochondrial disease. The research should take place in centers equipped for adequate mitochondrial analyses of donors and the follow-up of infants and children.