Thread by @RFitzgerald_lab

Here is our Tweetorial from our Nature Medicine paper published today. (1/8)
nature.com/articles/s4159…

DNA from patients with Barrett’s oesophagus can show who is most likely at risk to develop oesophageal cancer up to *8 years* before cancer is diagnosed, improving detection and decreasing overdiagnosis. (2/8)

Using a low resolution whole genome sequencing method, we looked for signals of copy number changes in the DNA. This is common in tumours and important in oesophageal cancer. Samples from patients with early cancer showed more changes than samples from patients who had not. (3/8)

We used these signals to train a statistical model to classify a single sample from a patient as being ‘high’, ‘low’, or ‘moderate’ risk of cancer. We compared these risks across histopathological grades and found that the pathology of the sample did not matter. (4/8)

Equally important the risks were consistent for a single patient over time. A high-risk patient was often high-risk from the earliest sample. This meant we could detect high risk patients at a rate of nearly 80% two years before diagnosis. (5/8)

Finally, we suggest it is possible to improve diagnosis and treatment with this model as half of high-risk patients could have been treated earlier and half of the low-risk patients would benefit from fewer unnecessary treatments and less invasive monitoring. (6/8)

Ultimately this study shows that it is possible to use genomic information for early detection of cancer. This was a great collaborative effort between @RFitzgerald_lab at the University of Cambridge and Cancer Data Science @MoritzGerstung at EMBL-EBI. (7/8)

Thanks to all our funding providers, collaborators and the patients who agree to provide samples for important studies like these.
@MRC_CU @Cambridge_Uni #BarrettsOesophagus #OesophagealCancer #EsophagealCancer #EarlyDetection
(8/8)

@threadreaderapp unroll please