Comparison of prostate specific antigen concentration versus prostate specific antigen density in the early detection of prostate cancer: receiver operating characteristic curves

J Urol. 1994 Dec;152(6 Pt 1):2031-6. doi: 10.1016/s0022-5347(17)32299-1.


We present the results of a prospective multicenter clinical trial of nearly 5,000 men in which prostate specific antigen (PSA) density was compared to the serum PSA concentration alone for early detection of prostate cancer. All men were evaluated with PSA and digital rectal examination. If PSA was elevated (greater than 4 ng./ml., Hybritech Tandem assay) or digital rectal examination was suspicious, transrectal ultrasound guided biopsies were recommended. Prostate volume was estimated by transrectal ultrasound measurements using a prolate ellipse volume calculation and PSA density was calculated by dividing serum PSA concentration by gland volume. Using a PSA density cutoff of 0.15 as recommended in the literature enhanced specificity but at the cost of missing half of the tumors. Of the organ confined neoplasms 47% were detected by a PSA of greater than 4.0 ng./ml. but they were missed by a PSA density of more than 0.15. PSA density may not be predictive for cancer because accurate estimation of transrectal ultrasound volume is difficult (r = 0.61 for estimated transrectal ultrasound volume versus pathological prostate weight). However, a relationship does exist among transrectal ultrasound volume, PSA and positive predictive value for cancer. PSA concentrations of less than 4.0 ng./ml. did not indicate a need for biopsy (positive predictive value 12 to 17%) unless the digital rectal examination findings were suspicious for cancer. A high percentage of patients with a PSA of more than 10 ng./ml. had cancer (30 to 75%), regardless of gland size. Patients with intermediate PSA concentrations (4.1 to 9.9 ng./ml.) and a gland size of 50 cc or less had a 35 to 51% positive predictive value, while those with intermediate PSA concentrations and a large gland (more than 50 cc) had a 15% positive predictive value. We conclude that in men with a PSA level of 4.1 to 9.9 ng./ml., and normal digital rectal examination and transrectal ultrasound findings, the use of a PSA density cutoff of more than 0.15 for biopsy results in half of the tumors being missed. Thus, we recommend that men in this group undergo biopsy based upon serum PSA concentration rather than PSA density.

Publication types

  • Clinical Trial
  • Comparative Study
  • Multicenter Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Aged, 80 and over
  • Biopsy
  • Humans
  • Male
  • Middle Aged
  • Neoplasm Staging
  • Predictive Value of Tests
  • Prospective Studies
  • Prostate / pathology
  • Prostate-Specific Antigen / blood*
  • Prostatic Neoplasms / blood
  • Prostatic Neoplasms / diagnosis*
  • ROC Curve
  • Sensitivity and Specificity
  • Time Factors


  • Prostate-Specific Antigen