You are prohibited from using or uploading content you accessed through this website into external applications, bots, software, or websites, including those using artificial intelligence technologies and infrastructure, including deep learning, machine learning and large language models and generative AI.
You have accessJournal of UrologyProstate Cancer: Detection & Screening III (MP30)1 Sep 2021


    View All Author Information


    Prostate specific antigen (PSA) based screening is limited in its specificity for detecting clinically significant prostate cancer. While more fluctuation in PSA measurements may be associated with benign processes, there are no established methods for measuring variation in PSA values over time. The wealth of longitudinal clinical data in electronic health records provides an opportunity to evaluate longitudinal variations in PSA measurements. We sought to evaluate whether variation in pre-diagnosis PSA levels was a biomarker associated with clinically significant prostate cancer.


    We identified patients diagnosed with localized prostate cancer in the Veterans Health Administration from 2010-2018. We abstracted all available PSA measurements in the 5 years prior to a prostate cancer diagnosis. We defined clinically significant prostate cancer as intermediate or higher risk prostate cancer. We fit logistic regression models to determine the odds of receiving a diagnosis of clinically significant prostate cancer. We measured the variability between PSA values using multiple models in several ways, measuring the standard deviation, average real variability (ARV), ARV with time correction, and variance of the residuals from linear regression.


    We identified 48,623 patients with a PSA measurement from 4-10 mg/dL diagnosed with localized prostate cancer. The average age at diagnosis was 66 years (IQR 62-70), and the average PSA at diagnosis was 5.9 md/dL (4.9-7.3). The ARV (average change between subsequent PSA measurements) was 0.98 mg/dL (IQR 0.71-1.41). Multiple analysis of PSA kinetics (PSA velocity) and variation (ARV, variance of residuals) were associated with the detection of clinically significant prostate cancer after adjusting for PSA at diagnosis. In fully adjusted models accounting for age, comorbidity, and PSA at diagnosis, the ARV was inversely associated with a diagnosis of clinically significant prostate cancer (OR 0.96, 95%CI 0.94-0.98). These results were consistent when examining patients with ≥5 pre-diagnosis PSA measurements (OR 0.91, 95%CI 0.89-0.94).


    Using longitudinal data from the largest integrated health care system, we demonstrate that variation in routine PSA measurements is significantly associated with the diagnosis of clinically significant prostate cancer, adjusting for age and charlson score, for patients with PSA values between 4-10 mg/dL. ARV is a new and accessible way to evaluate PSA variability in patients being screened for prostate cancer.

    Source of Funding: