Am Fam Physician. 2005;72(6):1091-1092
What is the risk of prostate cancer in a patient who is referred for biopsy?
Considerable uncertainty surrounds the screening, diagnosis, and management of prostate cancer. Although prostate cancer is one of the most commonly diagnosed malignancies, only a minority of patients die from the condition or suffer major disease-related morbidity. Because it is not uncommon for treatment to cause major morbidity and even mortality, patients need objective and accurate prognostic advice from their physicians. Validated clinical decision rules can help guide physicians. This Point-of-Care Guide addresses the likelihood of prostate cancer in patients who are referred for biopsy; a future Point-of-Care Guide will address the likelihood of recurrence following definitive treatment of patients with prostate cancer.
A recent study1 presented two validated clinical decision rules. The first rule was developed based on 4,193 men in Montreal, who were referred for prostate biopsy because of an abnormal digital rectal examination (DRE) and/or prostate-specific antigen (PSA). The rule was validated in a group of 1,762 men from Hamburg, Germany. The second rule, which was developed based on the Hamburg group, also incorporated free PSA and was validated in a group of 514 men from Montreal. All participants had a PSA between 0 and 50 ng per mL, with an ultrasonography-guided prostate biopsy providing the reference standard. The two groups used to validate the decision rules were similar, with a mean age of 64 years, mean PSA of 8.5 ng per mL, PSA range from 0 to 50 ng per mL, and a mean free PSA of 18 percent. The DRE was abnormal in 16.7 percent of patients in the Hamburg group and in 30.4 percent of patients in the second Montreal group. Prostate cancer was found in 41.9 percent of patients in the Hamburg group and in 36.8 percent of patients in the second Montreal group. Both rules (with and without free PSA) are shown as nomograms in the accompanying figure.1 It is also available online as a free handheld computer application athttp://www.auq.org/an/index.html
These clinical decision rules performed similarly in the development and validation groups, suggesting that they will apply well to other groups of patients. The accuracy of the rules was expressed as the area under the receiver operating characteristic curve (AUROCC), which was 0.69 for the nomogram without free PSA and 0.77 for the nomogram with this biomarker. A nomogram with an AUROCC of 0.77 will correctly classify two randomly selected patients, one with cancer and one without, 77 percent of the time. The AUROCC is a number between 0 and 1 where 1 is a perfectly accurate test and 0.5 is no better than random guessing. A review of the literature did not find a more accurate rule for patients presenting with this broad range of PSA values. Although a rule by Eastham and colleagues had an AUROCC of 0.75, it was limited to men with an abnormal DRE and a PSA value less than 4.0 ng per mL.2
Applying the Evidence
A 55-year-old man has an abnormal DRE at his annual physical examination. His physician orders a PSA test, which is in the normal range at 2.9 ng per mL, with a free PSA of 20 percent. The patient is hesitant to undergo a biopsy because his “prostate number” was normal. What is the likelihood that a prostate biopsy will reveal prostate cancer?
Answer: His physician uses his handheld computer to estimate the likelihood of prostate cancer using the electronic version of the nomogram shown in the accompanying figure. The likelihood of prostate cancer is 73 percent (95% confidence interval ±4%). The patient is encouraged to undergo a biopsy.