Incidence of prostate cancer after termination of screening in a population-based randomised screening trial.

BACKGROUND: In a previous publication from the Göteborg randomised screening trial from 2010, biennial prostate-specific antigen (PSA) screening for men ≤69 yr of age was shown to lower prostate cancer (PCa) mortality by 44%. The evidence of the optimal age to stop screening, however, is limited.

OBJECTIVE: To examine the risk of PCa after the discontinuation of screening.

DESIGN, SETTING, AND PARTICIPANTS: In December 1994, 20 000 men in Göteborg, Sweden, between the ages of 50 and 65 yr were randomised to a screening arm (invited biennially to PSA testing) and a control arm (not invited). At the upper age limit (average: 69 yr), a total of 13 423 men (6449 and 6974 in the screening and control arms, respectively) were still alive without PCa. The incidence of PCa hereafter was established by matching with the Western Swedish Cancer Register. Participants were followed until a diagnosis of PCa, death, or final follow-up on June 30, 2012, or for a maximum of 12 yr after the last invitation.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Incidence rates and disease-free survival were calculated with life table models and Kaplan-Meier estimates. A competing risk model was also applied.

RESULTS AND LIMITATIONS: Postscreening, 173 cases of PCa were diagnosed in the screening arm (median follow-up: 4.8 yr) and 371 in the control arm (median follow-up: 4.9 yr). Up to 9 yr postscreening, all risk groups were more commonly diagnosed in the control arm, but after 9 yr the rates in the screening arm caught up, other than those for the low-risk group. PCa mortality also caught up after 9 yr.

CONCLUSIONS: Nine years after the termination of PSA testing, the incidence of potentially lethal cancers equals that of nonscreened men. Considering the high PCa mortality rate in men >80 yr of age, a general age of 70 yr to discontinue screening might be too low. Instead, a flexible age to discontinue based on individual risk stratification should be recommended.