Diagnostic accuracy of Prostate Specific Antigen (PSA) for prostate cancer detection in primary care: the ProsDetect study

Conference:

SAPC ASM 2024 - Bristol

Talk Code:

12A.1

Presenter:

Dr Samuel W D Merriel

Twitter:

@SamMerriel

Co-authors:

Dr Janice Hoang, Mr Petter Buttle, Mr John McGrath, Prof Fiona M Walter, Prof Gary Abel, Dr Garth Funston

Author institutions:

University of Manchester, University of Oxford, Royal Devon University Healthcare NHS Foundation Trust, Queen Mary University of London, University of Exeter

Problem

Prostate cancer is the most commonly diagnosed cancer type in the UK, affecting over 50,000 males each year. Over 80% of patients with prostate cancer start their diagnostic journey in primary care; either through asymptomatic, opportunistic PSA screening or symptomatic presentation with lower urinary tract symptoms (LUTS). A recent systematic review demonstrated the virtual absence of any primary care evidence for the accuracy of PSA for prostate cancer detection in this setting.

Approach

A retrospective cohort study was undertaken including all males who underwent a PSA test between 01/01/2010 and 31/12/2016 within the Clinical Practice Research Datalink (CPRD) Aurum, with data linkage to the Office for National Statistics, National Cancer Registration and Analysis Service, and Hospital Episode Statistics. Performance characteristics of PSA were determined for any prostate cancer and clinically significant prostate cancer (Gleason score ≥ 7) against nationally recommended screening (PSA ≥ 3ng/mL) and age-adjusted thresholds (as per NICE guideline 12). The investigator team included a PPI co-investigator who chaired regular meetings with a PPI panel (two white males, one black male, one white female) to inform all aspects of the research from development to dissemination.

Findings

620,990 males aged between 18-109 (mean age 63, SD 12.4 years) were included in this study. 89.5% of participants were identified as having white ethnicity. Index of multiple deprivation ranged from 13.93% (86,408) in the least deprived decile to 7.96% (49,372) in the most deprived decile. Mean PSA level was 4.73ng/mL (range 0-11,883). 3.61% (22,394) men were diagnosed with prostate cancer within 24 months of a PSA test, 69.4% (15,539) of whom had clinically significant prostate cancer. PSA demonstrated high levels of accuracy for clinically significant prostate cancer (AUC 0.96) overall, with sensitivity rising with increasing age. No significant difference in accuracy was found between males with recorded symptoms vs presumed asymptomatic screening, or different ethnic groups.

Consequences

This study represents the first to evaluate the performance of PSA for the detection of prostate cancer as it is currently used in UK primary care. PSA appears to be very accurate for identifying clinically significant prostate cancer across all age groups and ethnicities. Partial verification bias likely affects the estimation of false negative PSA test results, as males with a ‘normal’ PSA result would seldom have undergone further investigation for prostate cancer. Incorporating data on stage at diagnosis and treatment outcomes could inform a more targeted use of PSA in primary care to avoid worsening overdiagnosis and overtreatment in males who would be unlikely to benefit. The use of PSA for prostate cancer detection in the future will also need to take into account recent changes to the prostate cancer diagnostic pathway, including new MRI techniques.

Submitted by:

Sam Merriel

Funding acknowledgement:

This project is funded by the NIHR Research for Patient Benefit programme (NIHR202734). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. This research is linked to the CanTest Collaborative, which was funded by Cancer Research UK [C8640/A23385], of which SWDM and GF were Clinical Research Fellows, FMW was Director, and GA was Co-investigator.