Researchers in the Nuffield Department of Primary Care Health Sciences at the University of Oxford have today unveiled a ground-breaking tool that could revolutionise the early detection of oesophageal cancer – the long tube that carries food from the throat to the stomach. This is the 8th most common cancer in the world. Using vast patient databases and cutting-edge computational techniques, the team has developed a prediction algorithm called ‘CanPredict (oesophageal)’ that identifies individuals at high risk of this cancer and could potentially save countless lives through targeted screening and early intervention.
Published today in ‘The Lancet Regional Health – Europe’ the team of researchers from the Universities of Oxford, Cambridge, and Nottingham developed this innovative tool to predict the 10-year risk of oesophageal cancer and to identify high-risk patients for further screening, potentially leading to earlier detection and improved patient outcomes. While there are methods available for detecting oesophageal cancer, such as endoscopy, they are often reserved for patients showing symptoms or those already known to be at high risk.
Professor Julia Hippisley-Cox, a practising GP and lead researcher from the Nuffield Department of Primary Care Health Sciences at the University of Oxford, emphasised the potential impact of the CanPredict tool:
'With no widespread screening program currently in place in the NHS, developing a new strategy to enable earlier detection remains paramount. CanPredict offers a tailored approach, concentrating on those most in need, and identifying patients at risk of oesophageal cancer. This has the potential to make diagnoses of cancer earlier when there are likely to be more treatment options.'
To put it into perspective, when using CanPredict to monitor only the top 20% of high-risk individuals, we can catch more than 3 in 4 cases (76%) of expected oesophageal cancer diagnoses in the coming decade.
Oesophageal cancer, a significant health concern worldwide, often remains undetected until its advanced stages, making early identification crucial. This new algorithm has the potential to revolutionise the way primary care practitioners – and healthcare systems more broadly – approach the disease. It could, for example, be something that a GP practice runs a few times a year to identify high-risk patients, without them having to come in for consultations.
The team developed the new tool by analysing the anonymised medical records from over 12 million patients from GP practices contributing to the QResearch database across England and identified over 16,000 cases of oesophageal cancer. The researchers incorporated key factors like age, lifestyle habits, medical history, and medication use into the CanPredict algorithm.
Once developed, CanPredict was checked by testing it in a separate set of QResearch practices (over 4 million patients) and the Clinical Practice Research Database (over 2.5 million patients). In testing, CanPredict accurately predicted an individual’s risk of oesophageal cancer within the next decade. It outperformed existing models for estimating oesophageal cancer risk.
Winnie Mei, co-author and Research Fellow in Medical Statistics and Epidemiology at the University of Oxford’s Nuffield Department of Primary Care Health Sciences, said:
'Our study bridges a significant gap in primary care. By identifying high-risk patients earlier, we can potentially offer them life-saving interventions. This tool is a testament to the power of combining technology with medical research.'
The study also highlighted the importance of factors such as age, body mass index, smoking, alcohol consumption, and previous medical conditions in determining the risk of developing oesophageal cancer. The algorithm’s ability to integrate these factors offers a comprehensive and personalised risk assessment for patients and can also help the NHS optimise the use of its resources by targeting those at highest risk who are most likely to benefit from screening.
Professor Rebecca Fitzgerald, OBE, FMedSci, co-author and Professor of Cancer Prevention at the University of Cambridge, said:
'While our findings are promising, it’s essential to approach them with cautious optimism. Our next steps to realising the potential of CanPredict involve assessing the cost-effectiveness of this tool and exploring its integration into national clinical computer systems.'
Professor Julia Hippisley-Cox said:
'We thank the many thousands of GPs who share anonymised data with QResearch without whom this research would not be possible.'