Comparing PSA Screening Patterns and Their Role as Predictor of Prostate Cancer Diagnosis: Analysis of a Contemporary North American Cohort.

Abstract

Introduction: PSA screening remains a pivotal tool for early prostate cancer (PCa) detection. International guidelines rely on evidence from three major randomized clinical trials: ERSPC, PLCO, and CAP. We aim to examine the percentage of patients in real-world practice who get PSA screening as defined by each of the aforementioned trials. Moreover, we seek to evaluate if the different PSA screening patterns have a different impact on PCa incidence and its features at diagnosis.

Materials and methods: Our institutional database was queried to identify men aged 55-69 who received at least one PSA test, did not develop PCa or die within 6 years of the initial test, had follow-up within our system at least 6 years after the initial test, and did not have a previous PCa diagnosis. A total of 28,612 patients met our selection criteria. We categorized patients into three distinct PSA screening patterns based on testing frequency (PLCO: 1 PSA test per year for 6 years; ERSPC: 2 or 3 PSA tests over 6 years; CAP: 1 PSA test over 6 years). Our primary outcomes were any PCa incidence and clinically significant PCa (csPCa, defined as ISUP ≥ 3) incidence. Secondary outcome was the rate of cM1 disease. Competing risks cumulative incidence curves were used to depict any PCa and csPCa diagnosis with death before a diagnosis considered a competing risk. Multivariable competing risks regression (CRR) was used to assess the impact of the different screening patterns on any PCa and csPCa incidence, after adjusting for confounding factors.

Results: The most prevalent PSA screening pattern was ERSPC, including 15,530 patients (54.3%), followed by the CAP with 9003 patients (31.5%), and the PLCO with only 4079 patients (14.2%). The median (IQR) follow-up time was 4.8 (1.7-10.8) years. At 10 years, any PCa incidence was 7.4% versus 5.6% versus 2.5% for PLCO versus ERSPC versus CAP, respectively, while for csPCa, the rates were 2.5% versus 2.5% versus 1.2% (both p < 0.001). On multivariable analyses, PLCO and ERSPC patterns were associated with 2.92-fold and 2.31-fold higher risks from 1 year to the next of any PCa diagnosis, respectively, compared to CAP pattern (both p < 0.001). Similarly, patients with PLCO and ERSPC patterns had 2.07-fold and 2.31-fold higher risks, respectively, of csPCa diagnosis compared to CAP pattern (both p < 0.001). In men with PCa diagnosis, the rates of cM1 disease were respectively 1.7% vs 5.6% vs 10.8% for PLCO versus ERSPC versus CAP, respectively (p = 0.0009).

Conclusion: We observed that the most common screening pattern in "real-world" clinical practice is close to what ERSPC recommend, and this pattern seems to achieve a reasonable reduction in the risk of advanced PCa, while limiting overdiagnosis.