Age-specific prostate-specific antigen: a reassessment.

Abstract

Our objective was to compare expected survival benefits when screening for prostate cancer with PSA, using an age-specific bound relative to a cutoff of 4.0 ng/ml. We used a decision analysis modeling the cancer yield in a cohort screened by both screening tests, and the survival of cancer cases given screen detection and in the absence of screening. Expected cancer yields and positive predictive values were from an ultrasound-guided biopsy series. Stage distributions of screen-detected cases were obtained from the literature. For localized causes, survival given screen detection was assumed to be equal to normal life expectancy for the population. For these cases, survival in the absence of screening was modeled as time from clinical diagnosis to death added to time remaining after time of screen and before clinical diagnosis was made (lead time). For nonlocalized cases at screen detection, survival given screen detection was assumed to be equal to survival in the absence of screening. The average difference between expected survival with and without screening as calculated for age-specific PSA and for PSA > 4.0 ng/ml and compared. Average years of life saved per subject screened using PSA > 4.0 ng/ml were comparable to those using the age-specific bound. Average years of life saved per cancer case, however, appeared to be potentially greater for PSA > 4.0 ng/ml than for age-specific. PSA. PSA > 4.0 ng/ml detected markedly more prostate cancer cases than age-specific PSA. Using a bound of 4.0 ng/ml for all ages appears to be more efficient in identifying men with cancer in a screening cohort, which translates into a greater expected survival benefit per cancer case.