Abstract
Objective: To establish the value of serum prostate‐specific antigen (PSA) and prostatespecific antigen per unit volume of prostate gland (PSAD) in detecting prostate carcinoma (CaP) in a hypothetical screening algorithm, a meta‐analysis of the sensitivities, specificities, predictive values and likelihood ratios were combined from the published data.
Data Sources: Journal articles identified by a MEDLINE database search from 1988 to October 1992, using prostate‐specific antigen as a key word were used to calculate the distribution of PSA in healthy men, men with benign prostatic hyperplasia (BPH) and men with prostate carcinoma (CaP)
Study Selection: Only studies that contained the specified serum PSA values and patient outcomes were included.
Data Extraction: The distributions of the serum PSA were plotted versus serum PSA for healthy men (2567), men with BPH (798) and men with CaP (835) from the abstracted data. Prostate volume distributions were estimated from the published transrectal ultrasound (TRUS) calculations.
Data Synthesis: Hypothetical cohorts of 1,000 men between the ages of 60 and 70 years were screened using three different screening decision algorithms. Using a serum PSA cutoff of 3.0 ng/ml for referral for transrectal biopsy, 59 of 80 (74%) CaP would be detected and 21 (26%) would be missed. 209 transrectal biopsies would be performed, and 150 (72%) of them would be negative for CaP. Using a serum PSA cutoff of 4.0 ng/ml, 52 of 80 (65%) CaP would be detected and 28 (35%) would be missed. 146 transrectal biopsies would be performed, and 94 (64%) of them would be unnecessary. Using a cutoff of 2.0 ng/ml for serum PSA and 0.1 ng/ml/cc for PSAD, 55 of 80 (69%) of the cancers would be detected and 25 (31%) would be missed. Only 84 transrectal biopsies would be performed, and 29 (35%) of them would be negative for cancer.
Conclusion: This algorithm maximizes the number of cancers detected (true‐positive cases) and at the same time reduces the number of false‐positive cases, minimizing the number of patients who would have to receive an unnecessary transrectal biopsy, compared to using a serum PSA cutoff of 3.0 or 4.0 ng/ml. © 1993 Wiley‐Liss, Inc.