Professor Tom Kirkwood: There are limits to our powers of prediction

Ageing is not caused by any single genetic 'clock'. Rather, there are many, many genes that play a part

There is good evidence that long life runs in families and the search is heating up to pin down exactly which genes might be involved.

The study by Thomas Perls and his colleagues claims to have found genetic "signatures" in human centenarians that take us a big step towards this goal. I am unconvinced. What Perls and his team have done is to develop and apply some intriguing and potentially valuable new techniques to analyse the data.

The problem is, however, that the sample size, even with 801 centenarians and a roughly similar numbers of younger "controls", is very small to justify confidence about the results.

Ageing is not caused by any single genetic "clock". Rather, there are many, many genes that play a part. Over recent years, the field of human genetics has been littered with early claims to have found the genes responsible for complex characteristics (and in terms of complexity longevity is surely the grand-daddy of them all), which have not stood the crucial test of repeatability in independent studies by others.

Having been often bitten, geneticists now rely on building a lot of evidence from so-called replication studies.

Time will tell whether the signatures claimed by the Boston group are robust, and thankfully we should not have too long to wait. Bigger studies are nearing completion, including the large European Genetics of Healthy Ageing Study based on almost 3,000 families from across Europe in which two or more family members (brothers and sisters) have lived past 90.

The report from Perls et al states that their signatures could "predict exceptional longevity with 77 per cent accuracy". This particular claim is open to serious misunderstanding.

They are not suggesting that they can screen the genes of you and me, for example, and tell us the chance we will live to 100. This would be a tall order indeed, given that only a quarter of what determines the length of human life is genetic.

What they mean is that in a fresh sample, which included a lot of people who were 100 already, the signatures could pick, with reasonable reliability, which of the individuals they analysed actually belonged to the centenarian group.

This is a different matter altogether, for which we can perhaps breathe a collective sigh of relief. From what we know already, it is rather unlikely that genetic screens will ever be able to forecast how long an individual will live.

For me, and I suspect many others, an element of uncertainty in one's future keeps the excitement alive.

Tom Kirkwood is director of the Institute for Ageing and Health at Newcastle University. Together with Francois Schächter and Daniel Cohen, he wrote the first scientific article mapping the prospects for the genetics of human longevity, which was published in the journal "Human Genetics" in 1993