Scientists have discovered new genes that cause oesophageal cancer, using a unique AI approach. Their findings, published in Nature Communications, shed light on possible therapeutic targets for the disease, which remains hard to treat.
Each year, over 8,000 people die due to oesophagus cancer in the UK. Incidence of the disease has risen substantially in the western world since the 1970s and this is thought to be due to lifestyle changes.
Oesophageal cancer does not normally cause any symptoms until a late stage when the cancer has spread, so there is an urgent need to find an effective treatment.
Current treatment methods are limited due to difficulties identifying the cancer-causing genes. Unlike most other forms of cancer, there are many different genes which can lead to patients developing oesophagus cancer, some of which are patient specific. This makes finding treatments for large cohorts of people, with different genetic drivers, especially challenging.
To better understand the disease, researchers in Francesca Ciccarelli’s lab at the Crick and King’s College London, developed new machine learning algorithms which characterise the properties of cancer-causing genes among all faulty genes in individual patients.
The team applied their algorithms to a cohort of 261 oesophageal cancer patients from the OCCAMS consortium, a national network that aims to collect and analyse oesophageal cancers from all over the UK.
They identified 952 novel genes that, together with known cancer-causing genes, help promote cancer development. The large majority of these newly predicted cancer ‘helper’ genes are rare or patient-specific.
By looking at the ways in which these genes perturbed cancer-related pathways, the team stratified the cancer patients into six groups, which showed distinct molecular and clinical features, suggesting that these different patient groups could benefit from different types of treatment.
“We are now working towards extending our approach to study the early phases of oesophageal cancer and identify the genetic drivers that initiate the disease,” says Francesca Ciccarelli, lead author of the study. “This way we can hopefully improve early detection and suggest patient-specific intervention strategies.”