A new algorithm used to detect bowel cancer risk based on blood test results could improve the accuracy of colorectal cancer screening, according to new research.
Researchers believe that the algorithm, developed by a team from Israel, could improve the accuracy of cancer screening in primary care and increase early detection.
The algorithm uses previously stored data from FBC tests to identify patients who may be at risk of developing bowel cancer by comparing their FBCs to those of a large patient dataset, under the assumption that patients who go on to develop bowel cancer may have subtle changes in their blood counts before developing overt symptoms.
The researchers used the algorithm on a large UK dataset of just over 2,900,000 patients with at least one FBC result present. A subset of patients was given a risk score based on their blood results and then followed up to determine how accurate the algorithm was at predicting their risk of colorectal cancer.
They found that the algorithm had a positive predictive value of just under 9%, meaning that there is a 9% probability that patients who score positively for colorectal cancer risk under the algorithm are truly at risk. NICE currently recommends that if there is a positive predictive value of at least 3% for suspected cancer, then patients should be referred for further investigation.
They also found that the area under the receiver operating characteristic curve (AUROC) value for the test was just under 0.8, meaning the test had good accuracy when separating people who were and were not at risk of getting colorectal cancer.
The researchers concluded: ‘These findings may provide a role for the risk score as a tool to assist case finding in a range of settings. The score could be applied to primary care electronic health records and be automatically updated each time a FBC result appeared in the record. Patients with high (or increasing) values could be identified and considered for investigation.’
Professor Julietta Patnick, a visiting professor at Oxford University and one of the authors, said: ‘The beauty of this algorithm is that it is a system which can be applied to GP surgery data, running quickly and efficiently to give a more accurate level of risk. It works by analysing demographic information and results from blood tests and blood test markers in a patient’s medical record.’