Of the main types of cancer, pancreatic cancer has a particularly poor prognosis, with a median survival time of four to six months. Key reasons for this are non-specific disease symptoms and a lack of effective screening tools, causing most patients to be diagnosed when the disease is too advanced for surgery to be an option. Consequently, a simple blood test that could detect pancreatic cancer in asymptomatic or mildly symptomatic individuals could have a significant impact in enabling earlier diagnosis and improved survival.
Designing a new cancer blood test requires identifying molecules that are present in the blood of people who either have cancer currently or are at risk of developing it, but are not found in people without cancer. Some potential biomarkers have previously been found for pancreatic cancer, but these are not able to discriminate sufficiently between cancer and non-cancer for them to be applied in the clinic, especially for people who have no cancer symptoms.
In a new study, researchers from the Medical Research Council Population Health Research Unit at the University of Oxford (MRC PHRU) used data from the China Kadoorie Biobank study to search for proteins in the blood that were associated with a subsequent diagnosis of pancreatic cancer, and could be used as biomarkers. The China Kadoorie Biobank study comprises over 500,000 Chinese adults, who donated blood samples at recruitment, and were then tracked through linkage to health insurance records and death registers. The findings have been published today in the International Journal of Epidemiology.
During nine years of follow-up, 700 individuals went on to develop pancreatic cancer. The levels of 92 different proteins were compared between the group that developed pancreatic cancer and a group of 700 individuals that did not develop pancreatic cancer (a case-subcohort approach).
Fifteen proteins were associated with pancreatic cancer risk and, for most of these, the risk increased with higher levels of the protein. Some of these associations varied with time prior to a diagnosis, meaning it may be possible to use these proteins to predict short-term risk.
Adding these proteins to a risk prediction model with established risk factors for pancreatic cancer (such as smoking, alcohol and family history of cancer) slightly increased the discriminatory ability of the model, especially in the short term.
Dr Christiana Kartsonaki, senior scientist at MRC PHRU and lead author of the study, said: ‘Earlier detection, even if only by a few months, could benefit patients and may allow the use of treatments such as surgery. The protein biomarkers we have identified here now need to be replicated in other studies. Future studies could also look at whether these proteins can be used to monitor high-risk individuals over time, or used as part of diagnostic testing in symptomatic individuals.’