Prostate cancer is the most common cancer in men in the UK, with over 40,000 men diagnosed each year. A key challenge is finding a way to distinguish men with harmful prostate cancers (that require urgent treatment) from men with less harmful prostate cancers (that do not require invasive treatment) and men with non-cancerous prostate conditions.
Most men with prostate cancer are identified after visiting their GP and being tested for Prostate Specific Antigen (PSA); a blood test which is used to help decide whether prostate cancer might be present. We believe that by using PSA together with other blood tests and medical information in an electronic tool, GPs will be better able to catch harmful prostate cancer early and reduce unnecessary tests and treatments for patients.
In this study, we will use anonymised information from the GP records (CPRD) of over 220,000 patients, in addition to data from the cancer registry (NCRAS), office on national statistics (ONS) and Hospital Episodic Statics (HES), to determine how good PSA is at finding prostate cancer in men who visit their GP and have the test. We will develop and test a new tool which will include information, such as blood test results and the patient’s body mass index, alongside PSA. This tool will ultimately be part of the GP practice’s computer system and linked to a patient’s records, making it easy to use, and provide useful information to aid decisions about whether further tests for prostate cancer are needed.
Background
Prostate cancer is the most common cancer type in UK men. Approximately 15% are diagnosed with indolent disease which seldom causes harm and does not require radical treatment. A key challenge is identifying men with clinically significant prostate cancer early, while avoiding overdiagnosis of indolent disease and over-investigation of those with benign disease. Most men with prostate cancer are diagnosed after visiting their GP and having a prostate specific antigen (PSA) test. However, the diagnostic performance of PSA for the detection of clinically significant prostate cancer (or all prostate cancer) in primary care is unknown.
Aim
To determine the diagnostic accuracy of PSA for the detection of a) clinically significant prostate cancer and b) all prostate cancer, and develop multivariable models incorporating PSA to improve prostate cancer detection.
Methods
This retrospective cohort study will use routine primary care (CPRD) and linked cancer registry (NCRAS), hospital (HES) and mortality (ONS) data from over 220,000 men who underwent PSA testing in primary care in England (2010-2015). The diagnostic accuracy of PSA for the detection of clinically significant prostate cancer and all prostate cancer will be determined, applying existing National Institute for Health and Care Excellence (NICE) thresholds. Multivariate logistic regression will be used to develop diagnostic prediction models containing the most predictive variables within GP records. These models will estimate the probability of undiagnosed clinically significant prostate cancer and all prostate cancer in individual patients. The sensitivity and specificity of models, at a range of risk thresholds, will be compared to PSA at thresholds used in current practice.
Expected public health benefit
This study will improve understanding of how well PSA performs in primary care. The models developed are intended to be integrated within GP IT systems to aid informed decision making about the need for further investigation and referral.
A new diagnosis of clinically significant prostate cancer (Gleason Score of >7 and/or Stage>T3) recorded in NCRAS data within the 24 months following index PSA testing; a new diagnosis of any prostate cancer recorded within NCRAS in the 24 months after the index PSA test.
Sam Merriel - Chief Investigator - University of Exeter
Sam Merriel - Corresponding Applicant - University of Exeter
Fiona Walter - Collaborator - Queen Mary University of London
Garth Funston - Collaborator - University of Cambridge
Gary Abel - Collaborator - University of Exeter
Peter Buttle - Collaborator - University of Exeter
HES Admitted Patient Care;NCRAS Cancer Registration Data;No additional NCRAS data required;ONS Death Registration Data;Patient Level Townsend Index