Use of percentage of free prostate-specific antigen to identify men at high risk of prostate cancer when PSA levels are 2.51 to 4 ng/mL and digital rectal examination is not suspicious for prostate cancer: An alternative model

Objectives. Currently, many clinicians do not recommend prostate biopsy for men with digital rectal examination (DRE) results that are not suspicious for cancer and prostate-specific antigen (PSA) values between 2.51 and 4 ng /mL. We propose a new model for the detection of prostate cancer using the percentage of free PSA (%FPSA) in the limited range of PSA values between 2 .51 and 4 ng/mL that maximizes clinical specificity (ie, minimizes false-po sitive results). This model identifies higher risk patients in this relativ ely low-risk population. Methods, Three hundred sixty-eight archived serum samples from men evaluate d and treated at two academic institutions were reviewed. All men had a his tologic diagnosis, findings not suspicious for cancer on DRE, and PSA level s between 2.51 and 4 ng/mL. Samples were tested in Hybritech's Tandem-R PSA and Tandem-R free PSA (FPSA) assays in the same laboratory at each institu tion. Results. Various models for cancer detection using %FPSA when PSA is 2.51 t o 4 ng/mL and DRE is not; suspicious for cancer are proposed. These models recommend biopsy for only 10% to 36% of the men in this population and woul d identify as many as 30% to 54% of the detectable cancers. There is eviden ce that the cancers that would be detected are the most aggressive cancers in this population. Conclusions. Our models identified men with a higher risk of prostate cance r in a relatively low-risk population that currently does not routinely und ergo biopsy. This may allow for a more cost-effective way to increase cance r detection when PSA values are between 2.51 and 4 ng/mL and DRE is not sus picious for cancer. This model has the potential to detect a greater number of clinically important and potentially curable cancers than would be dete cted with current practice. (C) 1999, Elsevier Science Inc.