Objective: To assess the risk of over-diagnosing and over-treating prostate cancer if population-based screening with serum prostate-specific antigen (PSA) is instituted.
Patients and methods: From a serum bank stored in 1980, PSA was analysed in 658 men with no previously known prostate cancer from a well-defined cohort from Göteborg, Sweden (men born in 1913); the incidence of clinical prostate cancer was registered until 1995. From the same area, and with the same selection criteria, another cohort of 710 men born in 1930-31, who in 1995 accepted an invitation for PSA screening, was also analysed.
Results: Of men born in 1913, 18 (2.7%) had died from prostate cancer and the cumulative probability of being diagnosed with clinical prostate cancer was 11.1% (5.0% in those with a PSA level of < 3 ng/mL vs 32.9% in those with a PSA level of > 3 ng/mL, P < 0.01). The mean lead-time from increased PSA (> 3 ng/mL) to clinical diagnosis was 7 years. The prostate cancer detection rate in men born in 1930-31 was 4.4% (22% among those with increased PSA levels) and 30 of 31 detected cancers were clinically localized.
Conclusions: Screening and sextant biopsies resulted in a lower detection rate (22%) than the cumulative risk of having clinical prostate cancer (33%) in men with increased PSA levels, indicating that under-diagnosis rather than over-diagnosis is the case at least with 'one-time' screening. Even if the stage distribution in screening-detected cancers seems promising (and thus may result in reduced mortality) it is notable that screening 67-year-old men will result in treatment a mean of 7 years before clinical symptoms occur and only one in four men anticipated to develop prostate cancer will die from the disease within 15 years. Large randomized screening trials seem mandatory to further explore the benefits and hazards of PSA screening.