Redistribution of age-specific prostate cancer stage at diagnosis

Recent reports indicate reductions in the incidence of prostate cancer that may be age or stage specific. Because age affects staging intensity in prostate cancer, one cannot fully understand the interaction of stage with calendar time without adjusting for the effects of age. This study was designed to define the interactions of calendar time with stage at diagnosis after adjusting for age. We used the 1973-1992 public-use files of the Surveillance Epidemiology and End Results Program to model the relationship between age and calendar time with age-specific incidence at different stages at diagnosis. The incidence of age-specific prostate cancer increased for all age groups over most years of the study, and the proportion of patients diagnosed at specific disease stages varied with age and calendar time. Bivariate generalized additive models of age-specific incidence showed highly significant nonlinear relationships with age and calendar time for all disease stages. Age-specific incidence increased with age at all disease stages. The age-specific incidence of all stages increased in the mid-1970s, but this increase disappeared or was substantially reduced from the late 1970s to the mid-1980s. The age-specific incidence of localized and regional disease increased again in the late 1980s, while the age-specific incidence of distant disease, which had peaked in the late 1980s, decreased. Our analysis demonstrates an age-adjusted redistribution of prostate cancer stage at diagnosis over calendar time. One possible explanation for stage redistribution is the introduction of the prostate-specific antigen test in the late 1980s. However, the trends from the period before the late 1980s demonstrate that additional factors change the prostate cancer stage over calendar time. Without identification of the factors that underlie stage redistribution, we will not be able to maximally reduce the stage at diagnosis.