Strategies combining total and percent free prostate specific antigen for detecting prostate cancer: A prospective evaluation

Purpose: Determining total prostate specific antigen (PSA) in plasma can often identify men who are subsequently diagnosed with prostate cancer. However, excess false-positives create large financial and psychological burdens in prostate cancer screening. The percent free PSA is lower when prostate cancer is present, although to our knowledge no large prospective analyses to date have evaluated whether adding testing for free PSA may decrease false-positives, while maintaining or perhaps improving the detection of potentially curable tumors. Materials and Methods: We measured total and percent free PSA in banked plasma samples from the Physicians' Health Study in 430 men who were later diagnosed with prostate cancer and 1,642 age matched controls who were not diagnosed with prostate cancer during a 12-year observation period. We calculated the number of cancers detected and the number of false-positives for various strategies of combined free and total PSA levels, and compared them to the use of total PSA alone. Results: Total PSA with a cutoff of 4 ng./ml. detected 149 cases but also yielded 144 false-positives. A strategy that applied percent free PSA to men with total PSA 4 to 10 ng./ml. detected 133 to 140 cancers and decreased false-positives to 83 of 117 depending on the percent free PSA cutoff used. As the percent free PSA cutoff was lowered from 25% to 20%, additional undetected cancers did not occur until year 9 of followup and the 20% cutoff decreased false-positives and, thus, potential negative biopsies, by 42%. Percent free PSA was superior to total PSA for discriminating cases from controls within the total PSA range of 4 to 10 or 3 to 10 ng./ml. (p <0.0001). A percent free PSA cutoff of 20% in men with total PSA 3 to 10 ng./ml. detected 10% more cancers with 12.5% fewer false-positives than the conventional strategy of total PSA greater than 4 ng./ml. Cancers missed by combined total and free PSA testing had longer intervals between blood sampling and diagnosis, and a greater likelihood of later diagnosis at an organ confined stage. Conclusions: These results demonstrate that in a prospective setting with long-term followup free PSA strategies can be identified that decrease unnecessary biopsies, while preserving or even improving cancer detection. Thus, total and free PSA can be combined without the need to weigh subjectively the trade-offs and relative costs of false-negative and false-positive results.