Objectives: The ability of percent free prostate-specific antigen (PSA) to distinguish benign from malignant prostate disease has been established within the 4.0 to 20.0 ng/mL total PSA range, but its utility within the less than 4.0 ng/mL total PSA range has not been clearly defined. We undertook this study to determine the lower limit for the percent free PSA reflex range.
Methods: Four hundred seventy-nine men (mean age [+/-SD] 63.2 +/- 9.68 years) met the following criteria: (1) a measurable total serum PSA level of 4.0 ng/mL or less (mean 2.64 +/- 0.050); (2) concurrently measured free PSA and percent free PSA calculated (mean 19.3% +/- 0.59%); (3) a sextant biopsy diagnosed benign (B) (n = 376) or malignant (M) (n = 103), at one institution, within 90 days of serum collection; and (4) no prior history of prostate cancer. We defined the lower limit to be the lowest total PSA value at which (1) percent free PSA distributions differed between benign and malignant cases; and (2) percent free PSA could predict malignant disease. We compared age, total PSA, and percent free PSA differences with the Mann-Whitney U test and analysis of variance, and used univariate logistic regression to determine each variable's predictive value. Other statistical analysis was performed with contingency tables, Fisher's exact test, and linear regression.
Results: The lowest total PSA value at which percent free PSA both differed between benign and malignant cases and predicted prostate cancer was 4.0 ng/mL.
Conclusions: The lower limit for the percent free PSA reflex range should be 4.0 ng/mL.