By CAMERON HENDERSON FOR THE MAIL ON SUNDAY
PUBLISHED: 17:00 EDT, 23 July 2022 | UPDATED: 17:12 EDT, 23 July 2022
Artificial intelligence can dramatically improve the chances of picking up the early-warning signs of oesophageal cancer, a pioneering NHS programme has shown.
The computer technology works by analysing pictures taken during a procedure known as an endoscopy, when a tiny camera on the end of a flexible tube is put down the throat – acting as ‘an extra pair of eyes’ to help doctors to identify pre-cancerous cells in the gullet.
At present, one in five cases are missed by doctors during a regular endoscopy – with sometimes severe consequences. But new trial results of the artificial intelligence software show it can accurately spot signs of pre-cancer during endoscopies in 92 per cent of patients.
At present, one in five cases are missed by doctors during a regular endoscopy – with sometimes severe consequences. But new trial results of the artificial intelligence software show it can accurately spot signs of pre-cancer during endoscopies in 92 percent of patients
About 9,000 Britons are diagnosed with oesophageal cancer each year. Patients with the most common type – adenocarcinoma – live for an average of just a year after diagnosis, making it one of the most deadly forms of the disease
Professor Rehan Haidry, a consultant endoscopist at University College London Hospitals NHS Trust, who offers the procedure, said: ‘If we pick up oesophageal cancer early, we can treat it with minimally invasive surgery and patients don’t need surgery or rounds of chemo and radiotherapy.’
About 9,000 Britons are diagnosed with oesophageal cancer each year. Patients with the most common type – adenocarcinoma – live for an average of just a year after diagnosis, making it one of the most deadly forms of the disease.
Common symptoms include a loss of appetite, difficulty swallowing and acid reflux. In up to 13 percent of patients, the disease is preceded by a condition called Barrett’s oesophagus – when cells lining the oesophagus change and mutate. Acid reflux is thought to be the main cause – as the stomach acid rising up into the oesophagus damages these cells. One in ten patients with acid reflux develop Barrett’s.
In roughly one in 100 cases each year, the cell changes can become pre-cancerous – a condition called dysplasia – and later turn into full blown cancer.
Pre-cancerous cells can be surgically removed during the same endoscopy procedure, using a fine wire passed down the throat to scrape away mutated cells. Studies show the procedure is highly effective with 95 per cent of patients still cancer free after ten years.
In the case of full-blown oesophageal cancer, treatment involves removing a portion of the gullet. During the operation, incisions are made into the neck, stomach or abdomen of the patient to remove part or all of the gullet and replace it with a section of the stomach or bowel. But for more than half of these patients, the cancer returns within two to three years.
Advanced disease is treated with a combination of chemotherapy and radiotherapy. This may be followed by immunotherapy if other treatments stop working.
However, at this point the cancer is usually incurable – and just 15 per cent of patients survive for five years or more.
The artificial intelligence software, called CADU, increases the likelihood of detecting oesophageal cancer early by highlighting areas of concern in the patient’s throat invisible to the naked eye. As the doctor feeds a camera down the patient’s throat, the software analyses pictures the doctor is seeing in real time and produces warnings that flash up on-screen, directing the surgeon to potentially abnormal cells.
The technology is approved by UK health regulators and has been in use at London’s UCH for the past year. Experts believe it will be reviewed for wider use in NHS hospitals within the next year.
Professor Haidry says: ‘One man in his 60s came in recently suffering with symptoms of Barrett’s oesophagus. Looking at the screen during the endoscopy, I couldn’t see any signs of pre-cancer. But the AI system picked it up straight away and highlighted the area.
‘After scraping away the cells and sending them off to the lab, the results came back within a week – and the computer had been right. The cells were precancerous.’
The patient is now, hopefully, no longer at risk of cancer developing, and Prof Haidry says: ‘Compare this to treatment for actual oesophageal cancer – patients have a massive operation to remove part of the oesophagus.
‘This involves spending weeks in hospital, and it can take up to six months to recover.
‘With the AI procedure, the risk involved is minimal and patients can go home the following day.’
Leave a Reply