Epigenetic 'mortality test' uses cheek cells

Researchers have developed a tool they say can be used to predict mortality risk using DNA methylation data, from cheek cells rather than from blood.

Previously developed epigenetic 'clocks' such as GrimAge use data on methylation of CG sites on the genome of certain blood plasma cells to train deep learning systems to predict the 'biological age' of an individual, alongside other predictors such as smoking status. The authors of this latest paper published in Frontiers in Ageing developed a mortality risk test, called CheekAge, using the methylation data from DNA taken from cells removed via a swab on the inside of the cheek. They then checked the accuracy of this data against methylation data taken from blood cells.

Professor Dusko Ilic, professor of stem cell sciences at King's College London - who was not involved in the research - said: 'While the development of CheekAge, a next-generation epigenetic buccal clock, represents a significant advancement in non-invasive ageing biomarkers, the use of the term "predictive of mortality" raises ethical concerns. I believe this phrasing can be misleading, as it implies a deterministic ability to forecast death, which is not scientifically or ethically appropriate in this particular case.'

Epigenetic 'clocks' use data from large cohorts, often looking at the methylation of DNA specific sites of the genome to use deep learning to develop a predictor of 'biological age' (see BioNews 1212). These sites can often be in the genes involved in cancer risk, or genes involved in the immune system. Methylation status is thought to be influenced by environmental and lifestyle factors, and could be one way in which these factors influence risk of death and disease.

Researchers from the company Tally Health in New York 'trained' CheekAge on data from 200,000 DNA methylation sites across more than 8000 samples of adult cheek cells, to develop a score for predicting the risk of mortality of an individual. They then validated their findings using blood cells obtained from the Lothian Cohort Study of people born between 1921 and 1936.

They found that CheekAge is associated with mortality risk in older adults, as an increase by a single standard deviation in CheekAge is linked to an increase of 21 percent in all-cause mortality risk.

Dr Maxim Shokhirev, the study's first author and head of computational biology and data science at Tally Health, said: 'The fact that our epigenetic clock trained on cheek cells predicts mortality when measuring the methylome in blood cells suggests there are common mortality signals across tissues.'