Breasts typically age more quickly than the rest of the female body. So suggests a system that may be the most accurate way yet of identifying a person’s age from a blood or tissue sample.

As we age, the pattern of chemical markings on our DNA changes. Each gene becomes more or less methylated, that is, they have methyl chemical groups added or removed. This generally increases or decreases gene expression. The whole process is known as epigenetics.

Steve Horvath at the University of California, Los Angeles, and his colleagues have used these changes to estimate a person’s age. To do so, they first performed a detailed statistical analysis of methylation patterns in 7844 healthy tissue samples from 51 different types of tissue. The tissue covered a range of ages – from fetuses to people 101 years old.

Universal ageing

The analysis allowed the team to weed out methylation patterns that varied between tissues, leaving just those that are common to all tissues. This enabled them to identify a subset of 353 specific regions of the genome that became either more or less methylated with age in almost all types of tissue.

By measuring the total amount of methylation in these regions, the team was able to create an algorithm that identified the age of the tissue.

The team validated the algorithm against thousands more samples of known age. Horvath says the method is twice as accurate as the next best method of ageing tissue, which is based on the length of telomeres – tips of chromosomes that “burn down” with age like candle wicks. He says that his method has a 96 per cent chance of accurately identifying someone’s age to within 3.6 years compared with around 53 per cent for telomeres.

“What’s unique about this study is the idea that there’s a signature of ageing common across tissues in spite of the significant tissue specificity of DNA methylation patterns,” comments Moshe Szyf, who studies methylation at McGill University in Montreal, Canada. “The data point to the possibility that DNA methylation signatures could be used as robust markers of biological ageing.”

Young at heart

Horvath says that, remarkably, their analysis shows that some parts of the body age at different rates. When they used their algorithm on healthy breast tissue from a group of women of average age 46, for example, it churned out a result that was on average two to three years older than the woman’s actual age. Whereas in two groups aged 55 and 60 across both sexes, heart tissue appeared nine years younger than true age.

If it is known where the sample comes from, it is still possible to accurately predict age after some straightforward adjustment, says Horvath. However, in general, the algorithm is most accurate for samples from people under 30 years of age. “The older one gets, the less accurate it becomes,” he says.

Horvath thinks that breast tissue ages more quickly because of its constant exposure to hormones. Heart tissue may remain younger, by contrast, because it is constantly regenerated by stem cells.

Cancerous tissue also appeared to age prematurely, coming out at 36 years older than the person’s actual age on average across 20 cancers from 20 different organs.

Because ageing is a risk factor for all cancers, Horvath suggests that the premature ageing of breast tissue might explain why it is the most common cancer in women. “It could be so prevalent because that part of the female body is older,” he says.

Blood work

Because the method also works on blood it might have the potential to be used forensically, to reveal the age of a murder suspect, suggests Horvath. It might also be used to diagnose cancer, by revealing accelerated ageing in tissue biopsies.

“The data raises questions about whether these DNA methylation changes play a causal role in ageing and, if so, whether epigenetic interventions could reverse these and therefore slow down ageing,” says Szyf. “The chemical robustness of DNA methylation and the ability to accurately measure it make it a very attractive tool to study ageing, which could well be superior to measuring telomere length, which is the current practice.”

Horvath says that further studies comparing telomere and epigenetic ageingcould be useful, and hopes the two can be complementary. He also says that the software for his algorithm is openly available so that other researchers can try validating it on their own tissue samples.

Journal reference: Genome Biology, DOI: 10.1186/gb-2013-14-10r115