The use of multiphase nonlinear mixed models to define and quantify long-term changes in serum prostate-specific antigen: data from the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial

Abstract

PurposeTo test the hypothesis that the pattern of prostate-specific antigen (PSA) change in men diagnosed with high-risk prostate cancer (PrCA) differs from the pattern evident in men diagnosed with low-risk PrCA or those with no evidence of PrCA. MethodsA retrospective cohort study from which PSA measures were taken before PrCA diagnosis from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Data were fitted using a nonlinear regression model to estimate the adjusted absolute and relative (%) change of PSA. ResultsData on 20,888 men with an average age of 61.61 years were included in the analysis. Of these, the 324 (1.55%) diagnosed with high-risk PrCA had a steeper and earlier transition into an exponential pattern of PSA change than the 1368 men diagnosed with low-risk cancer. At 1 year before diagnosis and/or exit, the average absolute PSA rates were 0.05 ng/mL/year (0.05–0.05), 0.59 (0.52–0.66), and 2.60 (2.11–3.09) for men with no evidence of PrCA, men with low-risk PrCA and those with high-risk PrCA, respectively. ConclusionsThe pattern of PSA change with time was significantly different for men who develop high-risk PrCA from those diagnosed with low-risk PrCA. Further research is required to validate this method and its utilization in PrCA screening.